{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.9994754098360654,
  "eval_steps": 500,
  "global_step": 22872,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00017486338797814208,
      "grad_norm": 4.858227827524086,
      "learning_rate": 2.9112081513828244e-07,
      "loss": 35.4685,
      "step": 1
    },
    {
      "epoch": 0.00034972677595628416,
      "grad_norm": 4.801647595594541,
      "learning_rate": 5.822416302765649e-07,
      "loss": 35.4762,
      "step": 2
    },
    {
      "epoch": 0.0005245901639344263,
      "grad_norm": 4.658281354615583,
      "learning_rate": 8.733624454148472e-07,
      "loss": 35.4743,
      "step": 3
    },
    {
      "epoch": 0.0006994535519125683,
      "grad_norm": 4.602554427931584,
      "learning_rate": 1.1644832605531297e-06,
      "loss": 35.4696,
      "step": 4
    },
    {
      "epoch": 0.0008743169398907104,
      "grad_norm": 4.395797374099342,
      "learning_rate": 1.455604075691412e-06,
      "loss": 35.4689,
      "step": 5
    },
    {
      "epoch": 0.0010491803278688525,
      "grad_norm": 4.793119669483129,
      "learning_rate": 1.7467248908296944e-06,
      "loss": 35.4527,
      "step": 6
    },
    {
      "epoch": 0.0012240437158469945,
      "grad_norm": 4.689138451023335,
      "learning_rate": 2.037845705967977e-06,
      "loss": 35.4498,
      "step": 7
    },
    {
      "epoch": 0.0013989071038251367,
      "grad_norm": 4.957555591408782,
      "learning_rate": 2.3289665211062595e-06,
      "loss": 35.4587,
      "step": 8
    },
    {
      "epoch": 0.0015737704918032786,
      "grad_norm": 4.43091983152915,
      "learning_rate": 2.6200873362445413e-06,
      "loss": 35.4652,
      "step": 9
    },
    {
      "epoch": 0.0017486338797814208,
      "grad_norm": 4.719134838783565,
      "learning_rate": 2.911208151382824e-06,
      "loss": 35.4375,
      "step": 10
    },
    {
      "epoch": 0.001923497267759563,
      "grad_norm": 4.613442320628435,
      "learning_rate": 3.2023289665211066e-06,
      "loss": 35.4535,
      "step": 11
    },
    {
      "epoch": 0.002098360655737705,
      "grad_norm": 4.5597143388472645,
      "learning_rate": 3.493449781659389e-06,
      "loss": 35.4529,
      "step": 12
    },
    {
      "epoch": 0.002273224043715847,
      "grad_norm": 4.334022626715464,
      "learning_rate": 3.7845705967976715e-06,
      "loss": 35.4001,
      "step": 13
    },
    {
      "epoch": 0.002448087431693989,
      "grad_norm": 4.854454373398284,
      "learning_rate": 4.075691411935954e-06,
      "loss": 35.4003,
      "step": 14
    },
    {
      "epoch": 0.002622950819672131,
      "grad_norm": 4.858204778281603,
      "learning_rate": 4.3668122270742355e-06,
      "loss": 35.3838,
      "step": 15
    },
    {
      "epoch": 0.0027978142076502733,
      "grad_norm": 4.799402797673639,
      "learning_rate": 4.657933042212519e-06,
      "loss": 35.3757,
      "step": 16
    },
    {
      "epoch": 0.0029726775956284155,
      "grad_norm": 4.590828136854875,
      "learning_rate": 4.949053857350801e-06,
      "loss": 35.3106,
      "step": 17
    },
    {
      "epoch": 0.003147540983606557,
      "grad_norm": 4.5831422537878925,
      "learning_rate": 5.240174672489083e-06,
      "loss": 35.2828,
      "step": 18
    },
    {
      "epoch": 0.0033224043715846994,
      "grad_norm": 4.426187621985402,
      "learning_rate": 5.531295487627366e-06,
      "loss": 35.2436,
      "step": 19
    },
    {
      "epoch": 0.0034972677595628415,
      "grad_norm": 4.899956707978409,
      "learning_rate": 5.822416302765648e-06,
      "loss": 35.1976,
      "step": 20
    },
    {
      "epoch": 0.0036721311475409837,
      "grad_norm": 4.523777019394532,
      "learning_rate": 6.11353711790393e-06,
      "loss": 35.1597,
      "step": 21
    },
    {
      "epoch": 0.003846994535519126,
      "grad_norm": 4.864827183382703,
      "learning_rate": 6.404657933042213e-06,
      "loss": 35.1295,
      "step": 22
    },
    {
      "epoch": 0.004021857923497268,
      "grad_norm": 4.83475735291328,
      "learning_rate": 6.695778748180495e-06,
      "loss": 34.9726,
      "step": 23
    },
    {
      "epoch": 0.00419672131147541,
      "grad_norm": 4.626747580043113,
      "learning_rate": 6.986899563318778e-06,
      "loss": 34.8663,
      "step": 24
    },
    {
      "epoch": 0.004371584699453552,
      "grad_norm": 5.020351763956844,
      "learning_rate": 7.2780203784570594e-06,
      "loss": 34.7855,
      "step": 25
    },
    {
      "epoch": 0.004546448087431694,
      "grad_norm": 5.340557073808984,
      "learning_rate": 7.569141193595343e-06,
      "loss": 34.61,
      "step": 26
    },
    {
      "epoch": 0.004721311475409836,
      "grad_norm": 4.941400651136073,
      "learning_rate": 7.860262008733626e-06,
      "loss": 34.4378,
      "step": 27
    },
    {
      "epoch": 0.004896174863387978,
      "grad_norm": 4.996652579476298,
      "learning_rate": 8.151382823871907e-06,
      "loss": 34.2486,
      "step": 28
    },
    {
      "epoch": 0.0050710382513661206,
      "grad_norm": 5.026365229810972,
      "learning_rate": 8.44250363901019e-06,
      "loss": 34.0479,
      "step": 29
    },
    {
      "epoch": 0.005245901639344262,
      "grad_norm": 4.639946788924763,
      "learning_rate": 8.733624454148471e-06,
      "loss": 33.8573,
      "step": 30
    },
    {
      "epoch": 0.005420765027322404,
      "grad_norm": 4.787401869629746,
      "learning_rate": 9.024745269286754e-06,
      "loss": 33.5362,
      "step": 31
    },
    {
      "epoch": 0.005595628415300547,
      "grad_norm": 4.709066499104172,
      "learning_rate": 9.315866084425038e-06,
      "loss": 33.2677,
      "step": 32
    },
    {
      "epoch": 0.005770491803278688,
      "grad_norm": 5.0546101734638125,
      "learning_rate": 9.60698689956332e-06,
      "loss": 32.9384,
      "step": 33
    },
    {
      "epoch": 0.005945355191256831,
      "grad_norm": 4.511784285136623,
      "learning_rate": 9.898107714701602e-06,
      "loss": 32.5265,
      "step": 34
    },
    {
      "epoch": 0.006120218579234973,
      "grad_norm": 4.021027649449401,
      "learning_rate": 1.0189228529839883e-05,
      "loss": 32.2188,
      "step": 35
    },
    {
      "epoch": 0.006295081967213114,
      "grad_norm": 4.008547935872386,
      "learning_rate": 1.0480349344978165e-05,
      "loss": 31.9469,
      "step": 36
    },
    {
      "epoch": 0.006469945355191257,
      "grad_norm": 3.5321797028332123,
      "learning_rate": 1.0771470160116449e-05,
      "loss": 31.6246,
      "step": 37
    },
    {
      "epoch": 0.006644808743169399,
      "grad_norm": 3.444831748055979,
      "learning_rate": 1.1062590975254732e-05,
      "loss": 31.3527,
      "step": 38
    },
    {
      "epoch": 0.006819672131147541,
      "grad_norm": 3.1577710671566757,
      "learning_rate": 1.1353711790393014e-05,
      "loss": 31.0588,
      "step": 39
    },
    {
      "epoch": 0.006994535519125683,
      "grad_norm": 2.7644321191757637,
      "learning_rate": 1.1644832605531296e-05,
      "loss": 30.7962,
      "step": 40
    },
    {
      "epoch": 0.007169398907103825,
      "grad_norm": 2.7750808587830504,
      "learning_rate": 1.1935953420669578e-05,
      "loss": 30.5999,
      "step": 41
    },
    {
      "epoch": 0.007344262295081967,
      "grad_norm": 2.354807471551902,
      "learning_rate": 1.222707423580786e-05,
      "loss": 30.3392,
      "step": 42
    },
    {
      "epoch": 0.007519125683060109,
      "grad_norm": 2.2750269136777708,
      "learning_rate": 1.2518195050946145e-05,
      "loss": 30.1607,
      "step": 43
    },
    {
      "epoch": 0.007693989071038252,
      "grad_norm": 2.179303531299442,
      "learning_rate": 1.2809315866084426e-05,
      "loss": 30.0291,
      "step": 44
    },
    {
      "epoch": 0.007868852459016393,
      "grad_norm": 2.142226989739817,
      "learning_rate": 1.3100436681222708e-05,
      "loss": 29.8922,
      "step": 45
    },
    {
      "epoch": 0.008043715846994535,
      "grad_norm": 1.991184494485336,
      "learning_rate": 1.339155749636099e-05,
      "loss": 29.7463,
      "step": 46
    },
    {
      "epoch": 0.008218579234972677,
      "grad_norm": 1.7455051705674154,
      "learning_rate": 1.3682678311499272e-05,
      "loss": 29.5483,
      "step": 47
    },
    {
      "epoch": 0.00839344262295082,
      "grad_norm": 1.6949320793374287,
      "learning_rate": 1.3973799126637555e-05,
      "loss": 29.5007,
      "step": 48
    },
    {
      "epoch": 0.008568306010928962,
      "grad_norm": 1.5928693248236963,
      "learning_rate": 1.4264919941775837e-05,
      "loss": 29.3745,
      "step": 49
    },
    {
      "epoch": 0.008743169398907104,
      "grad_norm": 1.5742491771905276,
      "learning_rate": 1.4556040756914119e-05,
      "loss": 29.1881,
      "step": 50
    },
    {
      "epoch": 0.008918032786885246,
      "grad_norm": 1.3854995743065592,
      "learning_rate": 1.4847161572052404e-05,
      "loss": 29.1953,
      "step": 51
    },
    {
      "epoch": 0.009092896174863387,
      "grad_norm": 1.4045876996905653,
      "learning_rate": 1.5138282387190686e-05,
      "loss": 29.0487,
      "step": 52
    },
    {
      "epoch": 0.00926775956284153,
      "grad_norm": 1.2725173295453058,
      "learning_rate": 1.5429403202328968e-05,
      "loss": 28.9743,
      "step": 53
    },
    {
      "epoch": 0.009442622950819672,
      "grad_norm": 1.1710639268755112,
      "learning_rate": 1.572052401746725e-05,
      "loss": 28.8587,
      "step": 54
    },
    {
      "epoch": 0.009617486338797814,
      "grad_norm": 1.1624331311753608,
      "learning_rate": 1.601164483260553e-05,
      "loss": 28.7838,
      "step": 55
    },
    {
      "epoch": 0.009792349726775956,
      "grad_norm": 1.1514356594414488,
      "learning_rate": 1.6302765647743815e-05,
      "loss": 28.7686,
      "step": 56
    },
    {
      "epoch": 0.009967213114754098,
      "grad_norm": 1.1111917249834578,
      "learning_rate": 1.6593886462882098e-05,
      "loss": 28.7136,
      "step": 57
    },
    {
      "epoch": 0.010142076502732241,
      "grad_norm": 1.0709493408428137,
      "learning_rate": 1.688500727802038e-05,
      "loss": 28.6078,
      "step": 58
    },
    {
      "epoch": 0.010316939890710383,
      "grad_norm": 1.0025446076271842,
      "learning_rate": 1.7176128093158662e-05,
      "loss": 28.5875,
      "step": 59
    },
    {
      "epoch": 0.010491803278688525,
      "grad_norm": 0.9404645305196282,
      "learning_rate": 1.7467248908296942e-05,
      "loss": 28.515,
      "step": 60
    },
    {
      "epoch": 0.010666666666666666,
      "grad_norm": 0.9575992256041006,
      "learning_rate": 1.7758369723435225e-05,
      "loss": 28.4825,
      "step": 61
    },
    {
      "epoch": 0.010841530054644808,
      "grad_norm": 0.9987959921982974,
      "learning_rate": 1.804949053857351e-05,
      "loss": 28.4431,
      "step": 62
    },
    {
      "epoch": 0.011016393442622952,
      "grad_norm": 0.907553662525052,
      "learning_rate": 1.8340611353711792e-05,
      "loss": 28.3703,
      "step": 63
    },
    {
      "epoch": 0.011191256830601093,
      "grad_norm": 0.863473977494591,
      "learning_rate": 1.8631732168850076e-05,
      "loss": 28.3926,
      "step": 64
    },
    {
      "epoch": 0.011366120218579235,
      "grad_norm": 0.907288040081988,
      "learning_rate": 1.8922852983988356e-05,
      "loss": 28.384,
      "step": 65
    },
    {
      "epoch": 0.011540983606557377,
      "grad_norm": 0.808357087523523,
      "learning_rate": 1.921397379912664e-05,
      "loss": 28.3287,
      "step": 66
    },
    {
      "epoch": 0.011715846994535518,
      "grad_norm": 0.8834394795464178,
      "learning_rate": 1.950509461426492e-05,
      "loss": 28.3359,
      "step": 67
    },
    {
      "epoch": 0.011890710382513662,
      "grad_norm": 0.9166304348146431,
      "learning_rate": 1.9796215429403203e-05,
      "loss": 28.3174,
      "step": 68
    },
    {
      "epoch": 0.012065573770491804,
      "grad_norm": 0.8787353807380196,
      "learning_rate": 2.0087336244541487e-05,
      "loss": 28.3359,
      "step": 69
    },
    {
      "epoch": 0.012240437158469945,
      "grad_norm": 0.8412778666293005,
      "learning_rate": 2.0378457059679767e-05,
      "loss": 28.2465,
      "step": 70
    },
    {
      "epoch": 0.012415300546448087,
      "grad_norm": 0.8319302161991783,
      "learning_rate": 2.066957787481805e-05,
      "loss": 28.256,
      "step": 71
    },
    {
      "epoch": 0.012590163934426229,
      "grad_norm": 0.7703861034721007,
      "learning_rate": 2.096069868995633e-05,
      "loss": 28.2831,
      "step": 72
    },
    {
      "epoch": 0.012765027322404372,
      "grad_norm": 0.7827538341150281,
      "learning_rate": 2.1251819505094614e-05,
      "loss": 28.2754,
      "step": 73
    },
    {
      "epoch": 0.012939890710382514,
      "grad_norm": 0.755334409152163,
      "learning_rate": 2.1542940320232897e-05,
      "loss": 28.2308,
      "step": 74
    },
    {
      "epoch": 0.013114754098360656,
      "grad_norm": 0.8155200839423544,
      "learning_rate": 2.183406113537118e-05,
      "loss": 28.2191,
      "step": 75
    },
    {
      "epoch": 0.013289617486338797,
      "grad_norm": 0.7270763282425197,
      "learning_rate": 2.2125181950509464e-05,
      "loss": 28.266,
      "step": 76
    },
    {
      "epoch": 0.01346448087431694,
      "grad_norm": 0.7859956085724417,
      "learning_rate": 2.2416302765647744e-05,
      "loss": 28.131,
      "step": 77
    },
    {
      "epoch": 0.013639344262295083,
      "grad_norm": 0.7629243109026331,
      "learning_rate": 2.2707423580786028e-05,
      "loss": 28.2468,
      "step": 78
    },
    {
      "epoch": 0.013814207650273224,
      "grad_norm": 0.7708892205499209,
      "learning_rate": 2.2998544395924308e-05,
      "loss": 28.2614,
      "step": 79
    },
    {
      "epoch": 0.013989071038251366,
      "grad_norm": 0.6830867016821024,
      "learning_rate": 2.328966521106259e-05,
      "loss": 28.2187,
      "step": 80
    },
    {
      "epoch": 0.014163934426229508,
      "grad_norm": 0.7225204392629097,
      "learning_rate": 2.3580786026200875e-05,
      "loss": 28.1995,
      "step": 81
    },
    {
      "epoch": 0.01433879781420765,
      "grad_norm": 0.7727340034875037,
      "learning_rate": 2.3871906841339155e-05,
      "loss": 28.178,
      "step": 82
    },
    {
      "epoch": 0.014513661202185793,
      "grad_norm": 0.7561934667903626,
      "learning_rate": 2.416302765647744e-05,
      "loss": 28.1877,
      "step": 83
    },
    {
      "epoch": 0.014688524590163935,
      "grad_norm": 0.7336204365263571,
      "learning_rate": 2.445414847161572e-05,
      "loss": 28.2271,
      "step": 84
    },
    {
      "epoch": 0.014863387978142076,
      "grad_norm": 0.7613616291901817,
      "learning_rate": 2.4745269286754006e-05,
      "loss": 28.0973,
      "step": 85
    },
    {
      "epoch": 0.015038251366120218,
      "grad_norm": 0.6869487500707409,
      "learning_rate": 2.503639010189229e-05,
      "loss": 28.1714,
      "step": 86
    },
    {
      "epoch": 0.01521311475409836,
      "grad_norm": 0.7385604659791998,
      "learning_rate": 2.5327510917030566e-05,
      "loss": 28.0808,
      "step": 87
    },
    {
      "epoch": 0.015387978142076503,
      "grad_norm": 0.7172503153519855,
      "learning_rate": 2.5618631732168853e-05,
      "loss": 28.1445,
      "step": 88
    },
    {
      "epoch": 0.015562841530054645,
      "grad_norm": 0.7238458403112161,
      "learning_rate": 2.5909752547307133e-05,
      "loss": 28.1326,
      "step": 89
    },
    {
      "epoch": 0.015737704918032787,
      "grad_norm": 0.7302276365530735,
      "learning_rate": 2.6200873362445416e-05,
      "loss": 28.1393,
      "step": 90
    },
    {
      "epoch": 0.01591256830601093,
      "grad_norm": 0.7224246259903495,
      "learning_rate": 2.64919941775837e-05,
      "loss": 28.1479,
      "step": 91
    },
    {
      "epoch": 0.01608743169398907,
      "grad_norm": 0.7064990039215003,
      "learning_rate": 2.678311499272198e-05,
      "loss": 28.0959,
      "step": 92
    },
    {
      "epoch": 0.016262295081967214,
      "grad_norm": 0.8488375652626062,
      "learning_rate": 2.7074235807860267e-05,
      "loss": 27.9862,
      "step": 93
    },
    {
      "epoch": 0.016437158469945354,
      "grad_norm": 0.7161923184394462,
      "learning_rate": 2.7365356622998544e-05,
      "loss": 28.1135,
      "step": 94
    },
    {
      "epoch": 0.016612021857923497,
      "grad_norm": 0.7176265168769299,
      "learning_rate": 2.765647743813683e-05,
      "loss": 28.0787,
      "step": 95
    },
    {
      "epoch": 0.01678688524590164,
      "grad_norm": 0.7609800817652234,
      "learning_rate": 2.794759825327511e-05,
      "loss": 28.0539,
      "step": 96
    },
    {
      "epoch": 0.01696174863387978,
      "grad_norm": 0.7292517722860066,
      "learning_rate": 2.8238719068413394e-05,
      "loss": 28.0362,
      "step": 97
    },
    {
      "epoch": 0.017136612021857924,
      "grad_norm": 0.8352733256915932,
      "learning_rate": 2.8529839883551674e-05,
      "loss": 28.0754,
      "step": 98
    },
    {
      "epoch": 0.017311475409836064,
      "grad_norm": 0.7842613673249885,
      "learning_rate": 2.8820960698689958e-05,
      "loss": 28.0063,
      "step": 99
    },
    {
      "epoch": 0.017486338797814208,
      "grad_norm": 0.7977047100649938,
      "learning_rate": 2.9112081513828238e-05,
      "loss": 28.0549,
      "step": 100
    },
    {
      "epoch": 0.01766120218579235,
      "grad_norm": 0.7352262249813267,
      "learning_rate": 2.940320232896652e-05,
      "loss": 28.1024,
      "step": 101
    },
    {
      "epoch": 0.01783606557377049,
      "grad_norm": 0.8413644837981467,
      "learning_rate": 2.9694323144104808e-05,
      "loss": 27.9918,
      "step": 102
    },
    {
      "epoch": 0.018010928961748635,
      "grad_norm": 0.7490569279565699,
      "learning_rate": 2.9985443959243088e-05,
      "loss": 28.0378,
      "step": 103
    },
    {
      "epoch": 0.018185792349726775,
      "grad_norm": 0.7375675134023838,
      "learning_rate": 3.0276564774381372e-05,
      "loss": 28.0381,
      "step": 104
    },
    {
      "epoch": 0.018360655737704918,
      "grad_norm": 0.6902314840243771,
      "learning_rate": 3.056768558951965e-05,
      "loss": 28.0028,
      "step": 105
    },
    {
      "epoch": 0.01853551912568306,
      "grad_norm": 0.7693875782573195,
      "learning_rate": 3.0858806404657935e-05,
      "loss": 27.9179,
      "step": 106
    },
    {
      "epoch": 0.0187103825136612,
      "grad_norm": 0.7710189941140785,
      "learning_rate": 3.114992721979622e-05,
      "loss": 27.9299,
      "step": 107
    },
    {
      "epoch": 0.018885245901639345,
      "grad_norm": 0.735663708687888,
      "learning_rate": 3.14410480349345e-05,
      "loss": 27.974,
      "step": 108
    },
    {
      "epoch": 0.019060109289617485,
      "grad_norm": 0.7516170468736362,
      "learning_rate": 3.173216885007278e-05,
      "loss": 27.9379,
      "step": 109
    },
    {
      "epoch": 0.01923497267759563,
      "grad_norm": 0.7454387514837676,
      "learning_rate": 3.202328966521106e-05,
      "loss": 27.894,
      "step": 110
    },
    {
      "epoch": 0.019409836065573772,
      "grad_norm": 0.7375331142150675,
      "learning_rate": 3.2314410480349346e-05,
      "loss": 27.8571,
      "step": 111
    },
    {
      "epoch": 0.019584699453551912,
      "grad_norm": 0.797024661824084,
      "learning_rate": 3.260553129548763e-05,
      "loss": 27.9281,
      "step": 112
    },
    {
      "epoch": 0.019759562841530055,
      "grad_norm": 0.7660456399276548,
      "learning_rate": 3.289665211062591e-05,
      "loss": 27.854,
      "step": 113
    },
    {
      "epoch": 0.019934426229508195,
      "grad_norm": 0.7767315923142535,
      "learning_rate": 3.3187772925764197e-05,
      "loss": 27.9189,
      "step": 114
    },
    {
      "epoch": 0.02010928961748634,
      "grad_norm": 0.8379655949045451,
      "learning_rate": 3.347889374090248e-05,
      "loss": 27.98,
      "step": 115
    },
    {
      "epoch": 0.020284153005464482,
      "grad_norm": 0.7775421979142323,
      "learning_rate": 3.377001455604076e-05,
      "loss": 27.853,
      "step": 116
    },
    {
      "epoch": 0.020459016393442622,
      "grad_norm": 0.7921790445204386,
      "learning_rate": 3.406113537117904e-05,
      "loss": 27.8188,
      "step": 117
    },
    {
      "epoch": 0.020633879781420766,
      "grad_norm": 0.7569832311017485,
      "learning_rate": 3.4352256186317324e-05,
      "loss": 27.8411,
      "step": 118
    },
    {
      "epoch": 0.020808743169398906,
      "grad_norm": 0.8060693425909563,
      "learning_rate": 3.464337700145561e-05,
      "loss": 27.8287,
      "step": 119
    },
    {
      "epoch": 0.02098360655737705,
      "grad_norm": 0.7619832843190928,
      "learning_rate": 3.4934497816593884e-05,
      "loss": 27.8225,
      "step": 120
    },
    {
      "epoch": 0.021158469945355193,
      "grad_norm": 0.8157861708325441,
      "learning_rate": 3.5225618631732174e-05,
      "loss": 27.9128,
      "step": 121
    },
    {
      "epoch": 0.021333333333333333,
      "grad_norm": 0.7518610386719049,
      "learning_rate": 3.551673944687045e-05,
      "loss": 27.8537,
      "step": 122
    },
    {
      "epoch": 0.021508196721311476,
      "grad_norm": 0.7262519445121263,
      "learning_rate": 3.5807860262008734e-05,
      "loss": 27.8517,
      "step": 123
    },
    {
      "epoch": 0.021683060109289616,
      "grad_norm": 0.7969019000772365,
      "learning_rate": 3.609898107714702e-05,
      "loss": 27.8104,
      "step": 124
    },
    {
      "epoch": 0.02185792349726776,
      "grad_norm": 0.8226886695484136,
      "learning_rate": 3.63901018922853e-05,
      "loss": 27.6992,
      "step": 125
    },
    {
      "epoch": 0.022032786885245903,
      "grad_norm": 0.7573128768328627,
      "learning_rate": 3.6681222707423585e-05,
      "loss": 27.7366,
      "step": 126
    },
    {
      "epoch": 0.022207650273224043,
      "grad_norm": 0.830080657385566,
      "learning_rate": 3.697234352256186e-05,
      "loss": 27.7963,
      "step": 127
    },
    {
      "epoch": 0.022382513661202186,
      "grad_norm": 0.7741770941301985,
      "learning_rate": 3.726346433770015e-05,
      "loss": 27.7001,
      "step": 128
    },
    {
      "epoch": 0.022557377049180326,
      "grad_norm": 0.7752952921767421,
      "learning_rate": 3.755458515283843e-05,
      "loss": 27.7253,
      "step": 129
    },
    {
      "epoch": 0.02273224043715847,
      "grad_norm": 0.8402940794488526,
      "learning_rate": 3.784570596797671e-05,
      "loss": 27.7235,
      "step": 130
    },
    {
      "epoch": 0.022907103825136613,
      "grad_norm": 0.7631581212871928,
      "learning_rate": 3.8136826783114996e-05,
      "loss": 27.7487,
      "step": 131
    },
    {
      "epoch": 0.023081967213114753,
      "grad_norm": 0.8585310191131846,
      "learning_rate": 3.842794759825328e-05,
      "loss": 27.704,
      "step": 132
    },
    {
      "epoch": 0.023256830601092897,
      "grad_norm": 0.9018100827478776,
      "learning_rate": 3.8719068413391556e-05,
      "loss": 27.7118,
      "step": 133
    },
    {
      "epoch": 0.023431693989071037,
      "grad_norm": 0.8590299675769272,
      "learning_rate": 3.901018922852984e-05,
      "loss": 27.6358,
      "step": 134
    },
    {
      "epoch": 0.02360655737704918,
      "grad_norm": 0.8580963136255146,
      "learning_rate": 3.930131004366812e-05,
      "loss": 27.7141,
      "step": 135
    },
    {
      "epoch": 0.023781420765027324,
      "grad_norm": 0.8618723360070963,
      "learning_rate": 3.9592430858806406e-05,
      "loss": 27.5508,
      "step": 136
    },
    {
      "epoch": 0.023956284153005464,
      "grad_norm": 0.7830752111233122,
      "learning_rate": 3.988355167394469e-05,
      "loss": 27.6374,
      "step": 137
    },
    {
      "epoch": 0.024131147540983607,
      "grad_norm": 0.9094931510532055,
      "learning_rate": 4.017467248908297e-05,
      "loss": 27.6821,
      "step": 138
    },
    {
      "epoch": 0.024306010928961747,
      "grad_norm": 0.9143243086279058,
      "learning_rate": 4.046579330422126e-05,
      "loss": 27.6104,
      "step": 139
    },
    {
      "epoch": 0.02448087431693989,
      "grad_norm": 0.7967976319500599,
      "learning_rate": 4.0756914119359534e-05,
      "loss": 27.7039,
      "step": 140
    },
    {
      "epoch": 0.024655737704918034,
      "grad_norm": 0.845620800806745,
      "learning_rate": 4.104803493449782e-05,
      "loss": 27.6392,
      "step": 141
    },
    {
      "epoch": 0.024830601092896174,
      "grad_norm": 0.7382241472777337,
      "learning_rate": 4.13391557496361e-05,
      "loss": 27.5491,
      "step": 142
    },
    {
      "epoch": 0.025005464480874318,
      "grad_norm": 0.8669498484762337,
      "learning_rate": 4.1630276564774384e-05,
      "loss": 27.5517,
      "step": 143
    },
    {
      "epoch": 0.025180327868852458,
      "grad_norm": 0.825284693612297,
      "learning_rate": 4.192139737991266e-05,
      "loss": 27.4516,
      "step": 144
    },
    {
      "epoch": 0.0253551912568306,
      "grad_norm": 0.811499158140326,
      "learning_rate": 4.221251819505095e-05,
      "loss": 27.5386,
      "step": 145
    },
    {
      "epoch": 0.025530054644808745,
      "grad_norm": 0.8790549722359482,
      "learning_rate": 4.250363901018923e-05,
      "loss": 27.5134,
      "step": 146
    },
    {
      "epoch": 0.025704918032786885,
      "grad_norm": 0.8804342190883664,
      "learning_rate": 4.279475982532751e-05,
      "loss": 27.4554,
      "step": 147
    },
    {
      "epoch": 0.025879781420765028,
      "grad_norm": 0.8439028477719986,
      "learning_rate": 4.3085880640465795e-05,
      "loss": 27.5564,
      "step": 148
    },
    {
      "epoch": 0.026054644808743168,
      "grad_norm": 0.8926297667108473,
      "learning_rate": 4.337700145560408e-05,
      "loss": 27.5131,
      "step": 149
    },
    {
      "epoch": 0.02622950819672131,
      "grad_norm": 0.8341363538351917,
      "learning_rate": 4.366812227074236e-05,
      "loss": 27.4962,
      "step": 150
    },
    {
      "epoch": 0.026404371584699455,
      "grad_norm": 0.8700651244991656,
      "learning_rate": 4.395924308588064e-05,
      "loss": 27.4942,
      "step": 151
    },
    {
      "epoch": 0.026579234972677595,
      "grad_norm": 0.8757258218882743,
      "learning_rate": 4.425036390101893e-05,
      "loss": 27.4849,
      "step": 152
    },
    {
      "epoch": 0.02675409836065574,
      "grad_norm": 0.8454041893722523,
      "learning_rate": 4.4541484716157205e-05,
      "loss": 27.4698,
      "step": 153
    },
    {
      "epoch": 0.02692896174863388,
      "grad_norm": 0.9937364700381729,
      "learning_rate": 4.483260553129549e-05,
      "loss": 27.3711,
      "step": 154
    },
    {
      "epoch": 0.027103825136612022,
      "grad_norm": 0.9716399225902266,
      "learning_rate": 4.512372634643377e-05,
      "loss": 27.4338,
      "step": 155
    },
    {
      "epoch": 0.027278688524590165,
      "grad_norm": 1.010176252437027,
      "learning_rate": 4.5414847161572056e-05,
      "loss": 27.4526,
      "step": 156
    },
    {
      "epoch": 0.027453551912568305,
      "grad_norm": 0.9959740392687261,
      "learning_rate": 4.570596797671033e-05,
      "loss": 27.4454,
      "step": 157
    },
    {
      "epoch": 0.02762841530054645,
      "grad_norm": 0.9481359662157296,
      "learning_rate": 4.5997088791848616e-05,
      "loss": 27.3652,
      "step": 158
    },
    {
      "epoch": 0.02780327868852459,
      "grad_norm": 1.6516440911844725,
      "learning_rate": 4.6288209606986906e-05,
      "loss": 27.4036,
      "step": 159
    },
    {
      "epoch": 0.027978142076502732,
      "grad_norm": 1.113401355405252,
      "learning_rate": 4.657933042212518e-05,
      "loss": 27.3874,
      "step": 160
    },
    {
      "epoch": 0.028153005464480876,
      "grad_norm": 1.2288137364866547,
      "learning_rate": 4.687045123726347e-05,
      "loss": 27.4323,
      "step": 161
    },
    {
      "epoch": 0.028327868852459016,
      "grad_norm": 1.5149757812730573,
      "learning_rate": 4.716157205240175e-05,
      "loss": 27.2317,
      "step": 162
    },
    {
      "epoch": 0.02850273224043716,
      "grad_norm": 0.9837944778649729,
      "learning_rate": 4.7452692867540034e-05,
      "loss": 27.2512,
      "step": 163
    },
    {
      "epoch": 0.0286775956284153,
      "grad_norm": 1.698585403157032,
      "learning_rate": 4.774381368267831e-05,
      "loss": 27.4224,
      "step": 164
    },
    {
      "epoch": 0.028852459016393443,
      "grad_norm": 1.1405733250348584,
      "learning_rate": 4.8034934497816594e-05,
      "loss": 27.3893,
      "step": 165
    },
    {
      "epoch": 0.029027322404371586,
      "grad_norm": 1.1816783369208228,
      "learning_rate": 4.832605531295488e-05,
      "loss": 27.3156,
      "step": 166
    },
    {
      "epoch": 0.029202185792349726,
      "grad_norm": 1.2134201945382115,
      "learning_rate": 4.861717612809316e-05,
      "loss": 27.1875,
      "step": 167
    },
    {
      "epoch": 0.02937704918032787,
      "grad_norm": 1.1563321580892374,
      "learning_rate": 4.890829694323144e-05,
      "loss": 27.2056,
      "step": 168
    },
    {
      "epoch": 0.02955191256830601,
      "grad_norm": 1.5430221773251895,
      "learning_rate": 4.919941775836973e-05,
      "loss": 27.1854,
      "step": 169
    },
    {
      "epoch": 0.029726775956284153,
      "grad_norm": 0.9353092811853465,
      "learning_rate": 4.949053857350801e-05,
      "loss": 27.1968,
      "step": 170
    },
    {
      "epoch": 0.029901639344262296,
      "grad_norm": 1.539137422549989,
      "learning_rate": 4.978165938864629e-05,
      "loss": 27.2731,
      "step": 171
    },
    {
      "epoch": 0.030076502732240436,
      "grad_norm": 0.8933755285740038,
      "learning_rate": 5.007278020378458e-05,
      "loss": 27.193,
      "step": 172
    },
    {
      "epoch": 0.03025136612021858,
      "grad_norm": 1.5379755238348432,
      "learning_rate": 5.0363901018922855e-05,
      "loss": 27.2524,
      "step": 173
    },
    {
      "epoch": 0.03042622950819672,
      "grad_norm": 1.2632026741284703,
      "learning_rate": 5.065502183406113e-05,
      "loss": 27.0718,
      "step": 174
    },
    {
      "epoch": 0.030601092896174863,
      "grad_norm": 1.3765797026922875,
      "learning_rate": 5.094614264919942e-05,
      "loss": 27.1541,
      "step": 175
    },
    {
      "epoch": 0.030775956284153007,
      "grad_norm": 1.573851298670216,
      "learning_rate": 5.1237263464337706e-05,
      "loss": 27.1578,
      "step": 176
    },
    {
      "epoch": 0.030950819672131147,
      "grad_norm": 1.0087007183640675,
      "learning_rate": 5.152838427947598e-05,
      "loss": 27.0613,
      "step": 177
    },
    {
      "epoch": 0.03112568306010929,
      "grad_norm": 1.3718824035585646,
      "learning_rate": 5.1819505094614266e-05,
      "loss": 27.2194,
      "step": 178
    },
    {
      "epoch": 0.031300546448087434,
      "grad_norm": 0.9821517207261886,
      "learning_rate": 5.2110625909752556e-05,
      "loss": 27.0628,
      "step": 179
    },
    {
      "epoch": 0.031475409836065574,
      "grad_norm": 1.434947652763146,
      "learning_rate": 5.240174672489083e-05,
      "loss": 27.0351,
      "step": 180
    },
    {
      "epoch": 0.031650273224043714,
      "grad_norm": 1.0676815958943728,
      "learning_rate": 5.269286754002911e-05,
      "loss": 27.0567,
      "step": 181
    },
    {
      "epoch": 0.03182513661202186,
      "grad_norm": 1.2998013621451086,
      "learning_rate": 5.29839883551674e-05,
      "loss": 27.145,
      "step": 182
    },
    {
      "epoch": 0.032,
      "grad_norm": 1.1835769038417956,
      "learning_rate": 5.327510917030568e-05,
      "loss": 26.9679,
      "step": 183
    },
    {
      "epoch": 0.03217486338797814,
      "grad_norm": 1.0638697025836061,
      "learning_rate": 5.356622998544396e-05,
      "loss": 27.0695,
      "step": 184
    },
    {
      "epoch": 0.03234972677595629,
      "grad_norm": 1.0635364569946943,
      "learning_rate": 5.3857350800582243e-05,
      "loss": 26.9618,
      "step": 185
    },
    {
      "epoch": 0.03252459016393443,
      "grad_norm": 1.0846005406456318,
      "learning_rate": 5.4148471615720534e-05,
      "loss": 26.9004,
      "step": 186
    },
    {
      "epoch": 0.03269945355191257,
      "grad_norm": 1.2100189004960713,
      "learning_rate": 5.443959243085881e-05,
      "loss": 26.8839,
      "step": 187
    },
    {
      "epoch": 0.03287431693989071,
      "grad_norm": 1.0655684576443292,
      "learning_rate": 5.473071324599709e-05,
      "loss": 26.965,
      "step": 188
    },
    {
      "epoch": 0.033049180327868855,
      "grad_norm": 1.1906080296110308,
      "learning_rate": 5.502183406113537e-05,
      "loss": 26.8019,
      "step": 189
    },
    {
      "epoch": 0.033224043715846995,
      "grad_norm": 1.163572962705198,
      "learning_rate": 5.531295487627366e-05,
      "loss": 26.7386,
      "step": 190
    },
    {
      "epoch": 0.033398907103825135,
      "grad_norm": 1.283704210525223,
      "learning_rate": 5.560407569141194e-05,
      "loss": 26.9112,
      "step": 191
    },
    {
      "epoch": 0.03357377049180328,
      "grad_norm": 1.072050672347309,
      "learning_rate": 5.589519650655022e-05,
      "loss": 26.8555,
      "step": 192
    },
    {
      "epoch": 0.03374863387978142,
      "grad_norm": 1.2800841976031785,
      "learning_rate": 5.618631732168851e-05,
      "loss": 26.7828,
      "step": 193
    },
    {
      "epoch": 0.03392349726775956,
      "grad_norm": 1.2038292999039304,
      "learning_rate": 5.647743813682679e-05,
      "loss": 26.6075,
      "step": 194
    },
    {
      "epoch": 0.03409836065573771,
      "grad_norm": 1.185359317071209,
      "learning_rate": 5.6768558951965065e-05,
      "loss": 26.7736,
      "step": 195
    },
    {
      "epoch": 0.03427322404371585,
      "grad_norm": 1.5197621641192542,
      "learning_rate": 5.705967976710335e-05,
      "loss": 26.6822,
      "step": 196
    },
    {
      "epoch": 0.03444808743169399,
      "grad_norm": 1.4362900516619068,
      "learning_rate": 5.735080058224164e-05,
      "loss": 26.7824,
      "step": 197
    },
    {
      "epoch": 0.03462295081967213,
      "grad_norm": 1.166851262795451,
      "learning_rate": 5.7641921397379915e-05,
      "loss": 26.8014,
      "step": 198
    },
    {
      "epoch": 0.034797814207650275,
      "grad_norm": 1.2406925826190267,
      "learning_rate": 5.79330422125182e-05,
      "loss": 26.7722,
      "step": 199
    },
    {
      "epoch": 0.034972677595628415,
      "grad_norm": 1.9677596271856832,
      "learning_rate": 5.8224163027656476e-05,
      "loss": 26.6667,
      "step": 200
    },
    {
      "epoch": 0.035147540983606555,
      "grad_norm": 1.62582721738291,
      "learning_rate": 5.8515283842794766e-05,
      "loss": 26.6975,
      "step": 201
    },
    {
      "epoch": 0.0353224043715847,
      "grad_norm": 1.2003857017055128,
      "learning_rate": 5.880640465793304e-05,
      "loss": 26.6629,
      "step": 202
    },
    {
      "epoch": 0.03549726775956284,
      "grad_norm": 1.5518318191935114,
      "learning_rate": 5.9097525473071326e-05,
      "loss": 26.5217,
      "step": 203
    },
    {
      "epoch": 0.03567213114754098,
      "grad_norm": 1.607914419916819,
      "learning_rate": 5.9388646288209616e-05,
      "loss": 26.4687,
      "step": 204
    },
    {
      "epoch": 0.03584699453551913,
      "grad_norm": 1.2981424118749079,
      "learning_rate": 5.967976710334789e-05,
      "loss": 26.4628,
      "step": 205
    },
    {
      "epoch": 0.03602185792349727,
      "grad_norm": 1.339146409509743,
      "learning_rate": 5.9970887918486177e-05,
      "loss": 26.406,
      "step": 206
    },
    {
      "epoch": 0.03619672131147541,
      "grad_norm": 1.219396668909946,
      "learning_rate": 6.026200873362445e-05,
      "loss": 26.5024,
      "step": 207
    },
    {
      "epoch": 0.03637158469945355,
      "grad_norm": 1.2131797490843124,
      "learning_rate": 6.0553129548762744e-05,
      "loss": 26.3981,
      "step": 208
    },
    {
      "epoch": 0.036546448087431696,
      "grad_norm": 1.2027993950155582,
      "learning_rate": 6.084425036390102e-05,
      "loss": 26.5549,
      "step": 209
    },
    {
      "epoch": 0.036721311475409836,
      "grad_norm": 1.2184255486361413,
      "learning_rate": 6.11353711790393e-05,
      "loss": 26.4823,
      "step": 210
    },
    {
      "epoch": 0.036896174863387976,
      "grad_norm": 1.3245899913721868,
      "learning_rate": 6.142649199417758e-05,
      "loss": 26.4097,
      "step": 211
    },
    {
      "epoch": 0.03707103825136612,
      "grad_norm": 1.3831958966699174,
      "learning_rate": 6.171761280931587e-05,
      "loss": 26.1267,
      "step": 212
    },
    {
      "epoch": 0.03724590163934426,
      "grad_norm": 1.268597950885905,
      "learning_rate": 6.200873362445415e-05,
      "loss": 26.4753,
      "step": 213
    },
    {
      "epoch": 0.0374207650273224,
      "grad_norm": 1.6101528095695532,
      "learning_rate": 6.229985443959244e-05,
      "loss": 26.4279,
      "step": 214
    },
    {
      "epoch": 0.03759562841530055,
      "grad_norm": 2.5357399665599343,
      "learning_rate": 6.259097525473073e-05,
      "loss": 26.1701,
      "step": 215
    },
    {
      "epoch": 0.03777049180327869,
      "grad_norm": 2.705997155486437,
      "learning_rate": 6.2882096069869e-05,
      "loss": 26.4284,
      "step": 216
    },
    {
      "epoch": 0.03794535519125683,
      "grad_norm": 1.4341152283182867,
      "learning_rate": 6.317321688500728e-05,
      "loss": 26.4271,
      "step": 217
    },
    {
      "epoch": 0.03812021857923497,
      "grad_norm": 2.209359769898925,
      "learning_rate": 6.346433770014556e-05,
      "loss": 26.3408,
      "step": 218
    },
    {
      "epoch": 0.03829508196721312,
      "grad_norm": 2.1986823489058245,
      "learning_rate": 6.375545851528385e-05,
      "loss": 26.0622,
      "step": 219
    },
    {
      "epoch": 0.03846994535519126,
      "grad_norm": 1.5463276277358817,
      "learning_rate": 6.404657933042213e-05,
      "loss": 26.329,
      "step": 220
    },
    {
      "epoch": 0.0386448087431694,
      "grad_norm": 2.3498520065660924,
      "learning_rate": 6.433770014556042e-05,
      "loss": 26.0753,
      "step": 221
    },
    {
      "epoch": 0.038819672131147544,
      "grad_norm": 1.8170504164488395,
      "learning_rate": 6.462882096069869e-05,
      "loss": 25.9207,
      "step": 222
    },
    {
      "epoch": 0.038994535519125684,
      "grad_norm": 2.1267555527348305,
      "learning_rate": 6.491994177583698e-05,
      "loss": 26.1286,
      "step": 223
    },
    {
      "epoch": 0.039169398907103824,
      "grad_norm": 1.510259103010917,
      "learning_rate": 6.521106259097526e-05,
      "loss": 26.046,
      "step": 224
    },
    {
      "epoch": 0.03934426229508197,
      "grad_norm": 2.5155035592802264,
      "learning_rate": 6.550218340611354e-05,
      "loss": 25.9292,
      "step": 225
    },
    {
      "epoch": 0.03951912568306011,
      "grad_norm": 1.947811182933867,
      "learning_rate": 6.579330422125183e-05,
      "loss": 26.2108,
      "step": 226
    },
    {
      "epoch": 0.03969398907103825,
      "grad_norm": 1.7766399500660492,
      "learning_rate": 6.60844250363901e-05,
      "loss": 25.8123,
      "step": 227
    },
    {
      "epoch": 0.03986885245901639,
      "grad_norm": 1.7951678838560963,
      "learning_rate": 6.637554585152839e-05,
      "loss": 26.1061,
      "step": 228
    },
    {
      "epoch": 0.04004371584699454,
      "grad_norm": 2.086486179839398,
      "learning_rate": 6.666666666666667e-05,
      "loss": 25.7166,
      "step": 229
    },
    {
      "epoch": 0.04021857923497268,
      "grad_norm": 1.7831147871348605,
      "learning_rate": 6.695778748180496e-05,
      "loss": 26.1088,
      "step": 230
    },
    {
      "epoch": 0.04039344262295082,
      "grad_norm": 1.6455452453370347,
      "learning_rate": 6.724890829694324e-05,
      "loss": 26.0308,
      "step": 231
    },
    {
      "epoch": 0.040568306010928964,
      "grad_norm": 1.984885146977419,
      "learning_rate": 6.754002911208151e-05,
      "loss": 25.9063,
      "step": 232
    },
    {
      "epoch": 0.040743169398907104,
      "grad_norm": 1.5175732120111274,
      "learning_rate": 6.783114992721979e-05,
      "loss": 25.9336,
      "step": 233
    },
    {
      "epoch": 0.040918032786885244,
      "grad_norm": 1.8143450974119226,
      "learning_rate": 6.812227074235808e-05,
      "loss": 25.9265,
      "step": 234
    },
    {
      "epoch": 0.04109289617486339,
      "grad_norm": 2.0738312676238277,
      "learning_rate": 6.841339155749637e-05,
      "loss": 25.9459,
      "step": 235
    },
    {
      "epoch": 0.04126775956284153,
      "grad_norm": 1.5013351650539357,
      "learning_rate": 6.870451237263465e-05,
      "loss": 25.822,
      "step": 236
    },
    {
      "epoch": 0.04144262295081967,
      "grad_norm": 1.4973363354770857,
      "learning_rate": 6.899563318777294e-05,
      "loss": 25.7051,
      "step": 237
    },
    {
      "epoch": 0.04161748633879781,
      "grad_norm": 1.5569349600233906,
      "learning_rate": 6.928675400291121e-05,
      "loss": 25.6541,
      "step": 238
    },
    {
      "epoch": 0.04179234972677596,
      "grad_norm": 1.4966107115242193,
      "learning_rate": 6.957787481804949e-05,
      "loss": 25.6869,
      "step": 239
    },
    {
      "epoch": 0.0419672131147541,
      "grad_norm": 1.621737244665646,
      "learning_rate": 6.986899563318777e-05,
      "loss": 25.7983,
      "step": 240
    },
    {
      "epoch": 0.04214207650273224,
      "grad_norm": 1.4489131489534333,
      "learning_rate": 7.016011644832606e-05,
      "loss": 25.7189,
      "step": 241
    },
    {
      "epoch": 0.042316939890710385,
      "grad_norm": 2.123211650835742,
      "learning_rate": 7.045123726346435e-05,
      "loss": 25.7123,
      "step": 242
    },
    {
      "epoch": 0.042491803278688525,
      "grad_norm": 2.062974216213105,
      "learning_rate": 7.074235807860263e-05,
      "loss": 25.4817,
      "step": 243
    },
    {
      "epoch": 0.042666666666666665,
      "grad_norm": 1.662994795281594,
      "learning_rate": 7.10334788937409e-05,
      "loss": 25.7265,
      "step": 244
    },
    {
      "epoch": 0.04284153005464481,
      "grad_norm": 1.9280872472443746,
      "learning_rate": 7.132459970887919e-05,
      "loss": 25.5856,
      "step": 245
    },
    {
      "epoch": 0.04301639344262295,
      "grad_norm": 1.9446757992150758,
      "learning_rate": 7.161572052401747e-05,
      "loss": 25.6793,
      "step": 246
    },
    {
      "epoch": 0.04319125683060109,
      "grad_norm": 1.591131789781628,
      "learning_rate": 7.190684133915575e-05,
      "loss": 25.6371,
      "step": 247
    },
    {
      "epoch": 0.04336612021857923,
      "grad_norm": 1.9780398514259774,
      "learning_rate": 7.219796215429404e-05,
      "loss": 25.4816,
      "step": 248
    },
    {
      "epoch": 0.04354098360655738,
      "grad_norm": 1.883295315838173,
      "learning_rate": 7.248908296943233e-05,
      "loss": 25.56,
      "step": 249
    },
    {
      "epoch": 0.04371584699453552,
      "grad_norm": 2.1007774905342167,
      "learning_rate": 7.27802037845706e-05,
      "loss": 25.5883,
      "step": 250
    },
    {
      "epoch": 0.04389071038251366,
      "grad_norm": 1.79223338982794,
      "learning_rate": 7.307132459970888e-05,
      "loss": 25.5479,
      "step": 251
    },
    {
      "epoch": 0.044065573770491806,
      "grad_norm": 1.6643341358701862,
      "learning_rate": 7.336244541484717e-05,
      "loss": 25.29,
      "step": 252
    },
    {
      "epoch": 0.044240437158469946,
      "grad_norm": 1.4197331786785994,
      "learning_rate": 7.365356622998545e-05,
      "loss": 25.4078,
      "step": 253
    },
    {
      "epoch": 0.044415300546448086,
      "grad_norm": 1.5876977010638336,
      "learning_rate": 7.394468704512372e-05,
      "loss": 25.4815,
      "step": 254
    },
    {
      "epoch": 0.04459016393442623,
      "grad_norm": 1.9760993090597638,
      "learning_rate": 7.423580786026201e-05,
      "loss": 25.2131,
      "step": 255
    },
    {
      "epoch": 0.04476502732240437,
      "grad_norm": 1.805835566313163,
      "learning_rate": 7.45269286754003e-05,
      "loss": 25.2173,
      "step": 256
    },
    {
      "epoch": 0.04493989071038251,
      "grad_norm": 1.8704476802454293,
      "learning_rate": 7.481804949053858e-05,
      "loss": 25.4722,
      "step": 257
    },
    {
      "epoch": 0.04511475409836065,
      "grad_norm": 1.4083685334290639,
      "learning_rate": 7.510917030567686e-05,
      "loss": 25.2182,
      "step": 258
    },
    {
      "epoch": 0.0452896174863388,
      "grad_norm": 1.4002775882968206,
      "learning_rate": 7.540029112081513e-05,
      "loss": 25.0156,
      "step": 259
    },
    {
      "epoch": 0.04546448087431694,
      "grad_norm": 1.4269088443128282,
      "learning_rate": 7.569141193595342e-05,
      "loss": 25.2994,
      "step": 260
    },
    {
      "epoch": 0.04563934426229508,
      "grad_norm": 1.6559053513140938,
      "learning_rate": 7.59825327510917e-05,
      "loss": 25.0617,
      "step": 261
    },
    {
      "epoch": 0.04581420765027323,
      "grad_norm": 1.4126745967429315,
      "learning_rate": 7.627365356622999e-05,
      "loss": 25.3572,
      "step": 262
    },
    {
      "epoch": 0.04598907103825137,
      "grad_norm": 1.5826433286352937,
      "learning_rate": 7.656477438136828e-05,
      "loss": 24.9536,
      "step": 263
    },
    {
      "epoch": 0.04616393442622951,
      "grad_norm": 1.8039926241525248,
      "learning_rate": 7.685589519650656e-05,
      "loss": 24.9893,
      "step": 264
    },
    {
      "epoch": 0.046338797814207654,
      "grad_norm": 1.7813262835757864,
      "learning_rate": 7.714701601164484e-05,
      "loss": 25.1392,
      "step": 265
    },
    {
      "epoch": 0.046513661202185794,
      "grad_norm": 1.7022125449451142,
      "learning_rate": 7.743813682678311e-05,
      "loss": 24.7403,
      "step": 266
    },
    {
      "epoch": 0.046688524590163934,
      "grad_norm": 1.2785879795967467,
      "learning_rate": 7.77292576419214e-05,
      "loss": 25.1456,
      "step": 267
    },
    {
      "epoch": 0.046863387978142074,
      "grad_norm": 1.4026221813738287,
      "learning_rate": 7.802037845705968e-05,
      "loss": 25.0238,
      "step": 268
    },
    {
      "epoch": 0.04703825136612022,
      "grad_norm": 1.4847050647493034,
      "learning_rate": 7.831149927219797e-05,
      "loss": 25.05,
      "step": 269
    },
    {
      "epoch": 0.04721311475409836,
      "grad_norm": 1.5973951592957962,
      "learning_rate": 7.860262008733625e-05,
      "loss": 25.0366,
      "step": 270
    },
    {
      "epoch": 0.0473879781420765,
      "grad_norm": 1.8676823348120772,
      "learning_rate": 7.889374090247454e-05,
      "loss": 25.13,
      "step": 271
    },
    {
      "epoch": 0.04756284153005465,
      "grad_norm": 2.237158588737557,
      "learning_rate": 7.918486171761281e-05,
      "loss": 25.0945,
      "step": 272
    },
    {
      "epoch": 0.04773770491803279,
      "grad_norm": 1.8193672213196743,
      "learning_rate": 7.947598253275109e-05,
      "loss": 25.0728,
      "step": 273
    },
    {
      "epoch": 0.04791256830601093,
      "grad_norm": 1.2823614857658412,
      "learning_rate": 7.976710334788938e-05,
      "loss": 24.9924,
      "step": 274
    },
    {
      "epoch": 0.048087431693989074,
      "grad_norm": 1.7136978107003056,
      "learning_rate": 8.005822416302766e-05,
      "loss": 24.9639,
      "step": 275
    },
    {
      "epoch": 0.048262295081967214,
      "grad_norm": 1.4718963383338464,
      "learning_rate": 8.034934497816595e-05,
      "loss": 25.0084,
      "step": 276
    },
    {
      "epoch": 0.048437158469945354,
      "grad_norm": 1.590326550025717,
      "learning_rate": 8.064046579330422e-05,
      "loss": 24.9821,
      "step": 277
    },
    {
      "epoch": 0.048612021857923494,
      "grad_norm": 1.6555902636451443,
      "learning_rate": 8.093158660844251e-05,
      "loss": 25.0593,
      "step": 278
    },
    {
      "epoch": 0.04878688524590164,
      "grad_norm": 1.6020044209350943,
      "learning_rate": 8.122270742358079e-05,
      "loss": 24.816,
      "step": 279
    },
    {
      "epoch": 0.04896174863387978,
      "grad_norm": 1.5677950069883146,
      "learning_rate": 8.151382823871907e-05,
      "loss": 24.6626,
      "step": 280
    },
    {
      "epoch": 0.04913661202185792,
      "grad_norm": 1.4810434480036576,
      "learning_rate": 8.180494905385734e-05,
      "loss": 24.4926,
      "step": 281
    },
    {
      "epoch": 0.04931147540983607,
      "grad_norm": 1.502493084163773,
      "learning_rate": 8.209606986899563e-05,
      "loss": 24.7021,
      "step": 282
    },
    {
      "epoch": 0.04948633879781421,
      "grad_norm": 1.4051191930180396,
      "learning_rate": 8.238719068413392e-05,
      "loss": 24.7164,
      "step": 283
    },
    {
      "epoch": 0.04966120218579235,
      "grad_norm": 1.612147536286588,
      "learning_rate": 8.26783114992722e-05,
      "loss": 24.8932,
      "step": 284
    },
    {
      "epoch": 0.049836065573770495,
      "grad_norm": 1.797976152809971,
      "learning_rate": 8.296943231441049e-05,
      "loss": 24.6564,
      "step": 285
    },
    {
      "epoch": 0.050010928961748635,
      "grad_norm": 2.1065953046046855,
      "learning_rate": 8.326055312954877e-05,
      "loss": 24.578,
      "step": 286
    },
    {
      "epoch": 0.050185792349726775,
      "grad_norm": 2.1737246588053565,
      "learning_rate": 8.355167394468704e-05,
      "loss": 24.8123,
      "step": 287
    },
    {
      "epoch": 0.050360655737704915,
      "grad_norm": 1.8532217768531927,
      "learning_rate": 8.384279475982532e-05,
      "loss": 24.7567,
      "step": 288
    },
    {
      "epoch": 0.05053551912568306,
      "grad_norm": 1.5388204244811574,
      "learning_rate": 8.413391557496361e-05,
      "loss": 24.6791,
      "step": 289
    },
    {
      "epoch": 0.0507103825136612,
      "grad_norm": 1.7066392046740728,
      "learning_rate": 8.44250363901019e-05,
      "loss": 24.7939,
      "step": 290
    },
    {
      "epoch": 0.05088524590163934,
      "grad_norm": 1.4438406407923643,
      "learning_rate": 8.471615720524018e-05,
      "loss": 24.8109,
      "step": 291
    },
    {
      "epoch": 0.05106010928961749,
      "grad_norm": 1.650374947724586,
      "learning_rate": 8.500727802037846e-05,
      "loss": 24.7218,
      "step": 292
    },
    {
      "epoch": 0.05123497267759563,
      "grad_norm": 1.4737869362948661,
      "learning_rate": 8.529839883551675e-05,
      "loss": 24.7123,
      "step": 293
    },
    {
      "epoch": 0.05140983606557377,
      "grad_norm": 1.7594820347650046,
      "learning_rate": 8.558951965065502e-05,
      "loss": 24.5317,
      "step": 294
    },
    {
      "epoch": 0.051584699453551916,
      "grad_norm": 1.8589271090305002,
      "learning_rate": 8.58806404657933e-05,
      "loss": 24.1516,
      "step": 295
    },
    {
      "epoch": 0.051759562841530056,
      "grad_norm": 1.76251454448185,
      "learning_rate": 8.617176128093159e-05,
      "loss": 24.7636,
      "step": 296
    },
    {
      "epoch": 0.051934426229508196,
      "grad_norm": 1.4890102985794584,
      "learning_rate": 8.646288209606988e-05,
      "loss": 24.7124,
      "step": 297
    },
    {
      "epoch": 0.052109289617486336,
      "grad_norm": 1.3629057841622985,
      "learning_rate": 8.675400291120816e-05,
      "loss": 24.5357,
      "step": 298
    },
    {
      "epoch": 0.05228415300546448,
      "grad_norm": 1.5729512024891854,
      "learning_rate": 8.704512372634643e-05,
      "loss": 24.4104,
      "step": 299
    },
    {
      "epoch": 0.05245901639344262,
      "grad_norm": 1.4881651086361458,
      "learning_rate": 8.733624454148472e-05,
      "loss": 24.5997,
      "step": 300
    },
    {
      "epoch": 0.05263387978142076,
      "grad_norm": 1.520744315669692,
      "learning_rate": 8.7627365356623e-05,
      "loss": 24.1185,
      "step": 301
    },
    {
      "epoch": 0.05280874316939891,
      "grad_norm": 1.5085966779779123,
      "learning_rate": 8.791848617176128e-05,
      "loss": 24.7374,
      "step": 302
    },
    {
      "epoch": 0.05298360655737705,
      "grad_norm": 1.3640422054555936,
      "learning_rate": 8.820960698689957e-05,
      "loss": 24.44,
      "step": 303
    },
    {
      "epoch": 0.05315846994535519,
      "grad_norm": 1.5271062719585973,
      "learning_rate": 8.850072780203786e-05,
      "loss": 24.3415,
      "step": 304
    },
    {
      "epoch": 0.05333333333333334,
      "grad_norm": 1.5919714463815677,
      "learning_rate": 8.879184861717613e-05,
      "loss": 24.3522,
      "step": 305
    },
    {
      "epoch": 0.05350819672131148,
      "grad_norm": 1.6233566536623074,
      "learning_rate": 8.908296943231441e-05,
      "loss": 24.4637,
      "step": 306
    },
    {
      "epoch": 0.05368306010928962,
      "grad_norm": 1.5412487825022165,
      "learning_rate": 8.93740902474527e-05,
      "loss": 24.4463,
      "step": 307
    },
    {
      "epoch": 0.05385792349726776,
      "grad_norm": 1.6157793757782746,
      "learning_rate": 8.966521106259098e-05,
      "loss": 24.2687,
      "step": 308
    },
    {
      "epoch": 0.054032786885245904,
      "grad_norm": 1.8298527993106863,
      "learning_rate": 8.995633187772925e-05,
      "loss": 24.5764,
      "step": 309
    },
    {
      "epoch": 0.054207650273224044,
      "grad_norm": 1.9313871536350464,
      "learning_rate": 9.024745269286754e-05,
      "loss": 24.6032,
      "step": 310
    },
    {
      "epoch": 0.054382513661202184,
      "grad_norm": 1.6459765902680061,
      "learning_rate": 9.053857350800584e-05,
      "loss": 24.5166,
      "step": 311
    },
    {
      "epoch": 0.05455737704918033,
      "grad_norm": 1.584838771712912,
      "learning_rate": 9.082969432314411e-05,
      "loss": 24.4847,
      "step": 312
    },
    {
      "epoch": 0.05473224043715847,
      "grad_norm": 1.5398283144620688,
      "learning_rate": 9.112081513828239e-05,
      "loss": 24.4558,
      "step": 313
    },
    {
      "epoch": 0.05490710382513661,
      "grad_norm": 1.6684106118012627,
      "learning_rate": 9.141193595342067e-05,
      "loss": 24.1756,
      "step": 314
    },
    {
      "epoch": 0.05508196721311476,
      "grad_norm": 1.795799444108209,
      "learning_rate": 9.170305676855896e-05,
      "loss": 24.2043,
      "step": 315
    },
    {
      "epoch": 0.0552568306010929,
      "grad_norm": 1.7181886350678617,
      "learning_rate": 9.199417758369723e-05,
      "loss": 24.3879,
      "step": 316
    },
    {
      "epoch": 0.05543169398907104,
      "grad_norm": 1.7372045845954183,
      "learning_rate": 9.228529839883552e-05,
      "loss": 24.3597,
      "step": 317
    },
    {
      "epoch": 0.05560655737704918,
      "grad_norm": 1.6676066341143148,
      "learning_rate": 9.257641921397381e-05,
      "loss": 24.3777,
      "step": 318
    },
    {
      "epoch": 0.055781420765027324,
      "grad_norm": 1.4540890420893315,
      "learning_rate": 9.286754002911209e-05,
      "loss": 24.2898,
      "step": 319
    },
    {
      "epoch": 0.055956284153005464,
      "grad_norm": 1.4927549160150082,
      "learning_rate": 9.315866084425037e-05,
      "loss": 24.3403,
      "step": 320
    },
    {
      "epoch": 0.056131147540983604,
      "grad_norm": 1.607259397696805,
      "learning_rate": 9.344978165938864e-05,
      "loss": 24.2566,
      "step": 321
    },
    {
      "epoch": 0.05630601092896175,
      "grad_norm": 1.5843590143261463,
      "learning_rate": 9.374090247452693e-05,
      "loss": 24.0865,
      "step": 322
    },
    {
      "epoch": 0.05648087431693989,
      "grad_norm": 1.637643809427229,
      "learning_rate": 9.403202328966521e-05,
      "loss": 24.2962,
      "step": 323
    },
    {
      "epoch": 0.05665573770491803,
      "grad_norm": 2.110144708222153,
      "learning_rate": 9.43231441048035e-05,
      "loss": 24.3536,
      "step": 324
    },
    {
      "epoch": 0.05683060109289618,
      "grad_norm": 2.629089512015341,
      "learning_rate": 9.461426491994178e-05,
      "loss": 24.1337,
      "step": 325
    },
    {
      "epoch": 0.05700546448087432,
      "grad_norm": 2.2965427989084457,
      "learning_rate": 9.490538573508007e-05,
      "loss": 24.3286,
      "step": 326
    },
    {
      "epoch": 0.05718032786885246,
      "grad_norm": 1.795802678667208,
      "learning_rate": 9.519650655021834e-05,
      "loss": 24.143,
      "step": 327
    },
    {
      "epoch": 0.0573551912568306,
      "grad_norm": 1.7697278948941118,
      "learning_rate": 9.548762736535662e-05,
      "loss": 24.0436,
      "step": 328
    },
    {
      "epoch": 0.057530054644808745,
      "grad_norm": 1.771193083591786,
      "learning_rate": 9.577874818049491e-05,
      "loss": 24.2306,
      "step": 329
    },
    {
      "epoch": 0.057704918032786885,
      "grad_norm": 1.4567836863430175,
      "learning_rate": 9.606986899563319e-05,
      "loss": 24.1842,
      "step": 330
    },
    {
      "epoch": 0.057879781420765025,
      "grad_norm": 1.89155834849994,
      "learning_rate": 9.636098981077148e-05,
      "loss": 23.9633,
      "step": 331
    },
    {
      "epoch": 0.05805464480874317,
      "grad_norm": 1.5660450595323832,
      "learning_rate": 9.665211062590975e-05,
      "loss": 24.118,
      "step": 332
    },
    {
      "epoch": 0.05822950819672131,
      "grad_norm": 1.5566791086757217,
      "learning_rate": 9.694323144104805e-05,
      "loss": 24.2498,
      "step": 333
    },
    {
      "epoch": 0.05840437158469945,
      "grad_norm": 1.5720407550946218,
      "learning_rate": 9.723435225618632e-05,
      "loss": 24.0363,
      "step": 334
    },
    {
      "epoch": 0.0585792349726776,
      "grad_norm": 1.3624431265377177,
      "learning_rate": 9.75254730713246e-05,
      "loss": 24.0775,
      "step": 335
    },
    {
      "epoch": 0.05875409836065574,
      "grad_norm": 1.6101528986560303,
      "learning_rate": 9.781659388646288e-05,
      "loss": 23.7333,
      "step": 336
    },
    {
      "epoch": 0.05892896174863388,
      "grad_norm": 1.5567336482391283,
      "learning_rate": 9.810771470160117e-05,
      "loss": 24.2285,
      "step": 337
    },
    {
      "epoch": 0.05910382513661202,
      "grad_norm": 1.5594302516487635,
      "learning_rate": 9.839883551673946e-05,
      "loss": 24.1747,
      "step": 338
    },
    {
      "epoch": 0.059278688524590166,
      "grad_norm": 1.6803754863827143,
      "learning_rate": 9.868995633187773e-05,
      "loss": 23.885,
      "step": 339
    },
    {
      "epoch": 0.059453551912568306,
      "grad_norm": 1.5252941983686166,
      "learning_rate": 9.898107714701602e-05,
      "loss": 23.7884,
      "step": 340
    },
    {
      "epoch": 0.059628415300546446,
      "grad_norm": 1.5544602648895525,
      "learning_rate": 9.92721979621543e-05,
      "loss": 24.0603,
      "step": 341
    },
    {
      "epoch": 0.05980327868852459,
      "grad_norm": 1.5326752788567934,
      "learning_rate": 9.956331877729258e-05,
      "loss": 23.9152,
      "step": 342
    },
    {
      "epoch": 0.05997814207650273,
      "grad_norm": 1.8493340450754099,
      "learning_rate": 9.985443959243085e-05,
      "loss": 23.3931,
      "step": 343
    },
    {
      "epoch": 0.06015300546448087,
      "grad_norm": 1.8954966367646715,
      "learning_rate": 0.00010014556040756916,
      "loss": 23.9974,
      "step": 344
    },
    {
      "epoch": 0.06032786885245902,
      "grad_norm": 1.8043808027332617,
      "learning_rate": 0.00010043668122270742,
      "loss": 23.8874,
      "step": 345
    },
    {
      "epoch": 0.06050273224043716,
      "grad_norm": 2.0527961602907534,
      "learning_rate": 0.00010072780203784571,
      "loss": 23.5479,
      "step": 346
    },
    {
      "epoch": 0.0606775956284153,
      "grad_norm": 2.2223220086727378,
      "learning_rate": 0.000101018922852984,
      "loss": 23.9523,
      "step": 347
    },
    {
      "epoch": 0.06085245901639344,
      "grad_norm": 2.1300511548676977,
      "learning_rate": 0.00010131004366812226,
      "loss": 24.1174,
      "step": 348
    },
    {
      "epoch": 0.06102732240437159,
      "grad_norm": 1.633269166606116,
      "learning_rate": 0.00010160116448326055,
      "loss": 24.0107,
      "step": 349
    },
    {
      "epoch": 0.06120218579234973,
      "grad_norm": 1.5110586231021181,
      "learning_rate": 0.00010189228529839884,
      "loss": 23.9964,
      "step": 350
    },
    {
      "epoch": 0.06137704918032787,
      "grad_norm": 1.7023746306332708,
      "learning_rate": 0.00010218340611353712,
      "loss": 23.9922,
      "step": 351
    },
    {
      "epoch": 0.061551912568306014,
      "grad_norm": 1.4607545666042512,
      "learning_rate": 0.00010247452692867541,
      "loss": 23.6991,
      "step": 352
    },
    {
      "epoch": 0.061726775956284154,
      "grad_norm": 1.5038967593400225,
      "learning_rate": 0.0001027656477438137,
      "loss": 23.8144,
      "step": 353
    },
    {
      "epoch": 0.061901639344262294,
      "grad_norm": 1.4576678810200119,
      "learning_rate": 0.00010305676855895196,
      "loss": 23.8565,
      "step": 354
    },
    {
      "epoch": 0.06207650273224044,
      "grad_norm": 1.635477555400767,
      "learning_rate": 0.00010334788937409025,
      "loss": 23.77,
      "step": 355
    },
    {
      "epoch": 0.06225136612021858,
      "grad_norm": 1.4778936630869124,
      "learning_rate": 0.00010363901018922853,
      "loss": 23.6985,
      "step": 356
    },
    {
      "epoch": 0.06242622950819672,
      "grad_norm": 1.4862666578445007,
      "learning_rate": 0.00010393013100436682,
      "loss": 23.8497,
      "step": 357
    },
    {
      "epoch": 0.06260109289617487,
      "grad_norm": 1.5889889107785695,
      "learning_rate": 0.00010422125181950511,
      "loss": 23.8538,
      "step": 358
    },
    {
      "epoch": 0.062775956284153,
      "grad_norm": 1.6761480860955447,
      "learning_rate": 0.00010451237263464338,
      "loss": 23.6655,
      "step": 359
    },
    {
      "epoch": 0.06295081967213115,
      "grad_norm": 1.8971422789245758,
      "learning_rate": 0.00010480349344978167,
      "loss": 23.6797,
      "step": 360
    },
    {
      "epoch": 0.0631256830601093,
      "grad_norm": 1.6027527986713557,
      "learning_rate": 0.00010509461426491996,
      "loss": 23.7689,
      "step": 361
    },
    {
      "epoch": 0.06330054644808743,
      "grad_norm": 1.6637658277492975,
      "learning_rate": 0.00010538573508005822,
      "loss": 23.7113,
      "step": 362
    },
    {
      "epoch": 0.06347540983606557,
      "grad_norm": 1.6040905021737304,
      "learning_rate": 0.00010567685589519651,
      "loss": 23.4954,
      "step": 363
    },
    {
      "epoch": 0.06365027322404372,
      "grad_norm": 1.4905587656526835,
      "learning_rate": 0.0001059679767103348,
      "loss": 23.5572,
      "step": 364
    },
    {
      "epoch": 0.06382513661202185,
      "grad_norm": 1.7476407328840693,
      "learning_rate": 0.00010625909752547308,
      "loss": 23.673,
      "step": 365
    },
    {
      "epoch": 0.064,
      "grad_norm": 1.6880374609524296,
      "learning_rate": 0.00010655021834061137,
      "loss": 23.4894,
      "step": 366
    },
    {
      "epoch": 0.06417486338797815,
      "grad_norm": 1.4261624982621444,
      "learning_rate": 0.00010684133915574963,
      "loss": 23.7792,
      "step": 367
    },
    {
      "epoch": 0.06434972677595628,
      "grad_norm": 1.4872810804573036,
      "learning_rate": 0.00010713245997088792,
      "loss": 23.5914,
      "step": 368
    },
    {
      "epoch": 0.06452459016393443,
      "grad_norm": 1.847560942133172,
      "learning_rate": 0.00010742358078602621,
      "loss": 23.3203,
      "step": 369
    },
    {
      "epoch": 0.06469945355191258,
      "grad_norm": 1.8361577026966331,
      "learning_rate": 0.00010771470160116449,
      "loss": 23.5731,
      "step": 370
    },
    {
      "epoch": 0.06487431693989071,
      "grad_norm": 1.5088348403002405,
      "learning_rate": 0.00010800582241630278,
      "loss": 23.397,
      "step": 371
    },
    {
      "epoch": 0.06504918032786886,
      "grad_norm": 1.5210757157628552,
      "learning_rate": 0.00010829694323144107,
      "loss": 23.4875,
      "step": 372
    },
    {
      "epoch": 0.06522404371584699,
      "grad_norm": 1.494687523461371,
      "learning_rate": 0.00010858806404657933,
      "loss": 23.2187,
      "step": 373
    },
    {
      "epoch": 0.06539890710382514,
      "grad_norm": 1.5156636500146325,
      "learning_rate": 0.00010887918486171762,
      "loss": 23.3875,
      "step": 374
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 1.8072463055780572,
      "learning_rate": 0.00010917030567685591,
      "loss": 23.5285,
      "step": 375
    },
    {
      "epoch": 0.06574863387978142,
      "grad_norm": 1.8795986482641827,
      "learning_rate": 0.00010946142649199417,
      "loss": 23.8167,
      "step": 376
    },
    {
      "epoch": 0.06592349726775956,
      "grad_norm": 2.0793954075916665,
      "learning_rate": 0.00010975254730713246,
      "loss": 23.4038,
      "step": 377
    },
    {
      "epoch": 0.06609836065573771,
      "grad_norm": 1.9887618133233158,
      "learning_rate": 0.00011004366812227074,
      "loss": 23.1114,
      "step": 378
    },
    {
      "epoch": 0.06627322404371584,
      "grad_norm": 1.7306994131067632,
      "learning_rate": 0.00011033478893740903,
      "loss": 23.3788,
      "step": 379
    },
    {
      "epoch": 0.06644808743169399,
      "grad_norm": 1.4065226847812573,
      "learning_rate": 0.00011062590975254732,
      "loss": 23.3755,
      "step": 380
    },
    {
      "epoch": 0.06662295081967214,
      "grad_norm": 1.7324546077946927,
      "learning_rate": 0.00011091703056768559,
      "loss": 23.7026,
      "step": 381
    },
    {
      "epoch": 0.06679781420765027,
      "grad_norm": 1.555961491854603,
      "learning_rate": 0.00011120815138282388,
      "loss": 23.4618,
      "step": 382
    },
    {
      "epoch": 0.06697267759562842,
      "grad_norm": 1.660588306799803,
      "learning_rate": 0.00011149927219796217,
      "loss": 23.2671,
      "step": 383
    },
    {
      "epoch": 0.06714754098360656,
      "grad_norm": 1.4334543103808455,
      "learning_rate": 0.00011179039301310044,
      "loss": 23.4299,
      "step": 384
    },
    {
      "epoch": 0.0673224043715847,
      "grad_norm": 1.451309412939816,
      "learning_rate": 0.00011208151382823873,
      "loss": 23.3057,
      "step": 385
    },
    {
      "epoch": 0.06749726775956284,
      "grad_norm": 1.6425900372741782,
      "learning_rate": 0.00011237263464337702,
      "loss": 23.2509,
      "step": 386
    },
    {
      "epoch": 0.06767213114754099,
      "grad_norm": 1.6609530520958562,
      "learning_rate": 0.00011266375545851529,
      "loss": 23.3019,
      "step": 387
    },
    {
      "epoch": 0.06784699453551912,
      "grad_norm": 1.6278792087089404,
      "learning_rate": 0.00011295487627365358,
      "loss": 22.9281,
      "step": 388
    },
    {
      "epoch": 0.06802185792349727,
      "grad_norm": 1.4562436056695214,
      "learning_rate": 0.00011324599708879184,
      "loss": 23.4387,
      "step": 389
    },
    {
      "epoch": 0.06819672131147542,
      "grad_norm": 1.702304276844428,
      "learning_rate": 0.00011353711790393013,
      "loss": 23.2949,
      "step": 390
    },
    {
      "epoch": 0.06837158469945355,
      "grad_norm": 1.5570558856592212,
      "learning_rate": 0.00011382823871906842,
      "loss": 23.4653,
      "step": 391
    },
    {
      "epoch": 0.0685464480874317,
      "grad_norm": 1.4828643621026556,
      "learning_rate": 0.0001141193595342067,
      "loss": 23.2007,
      "step": 392
    },
    {
      "epoch": 0.06872131147540983,
      "grad_norm": 1.4753161011605864,
      "learning_rate": 0.00011441048034934499,
      "loss": 23.2303,
      "step": 393
    },
    {
      "epoch": 0.06889617486338798,
      "grad_norm": 1.6179191886278683,
      "learning_rate": 0.00011470160116448328,
      "loss": 23.0926,
      "step": 394
    },
    {
      "epoch": 0.06907103825136612,
      "grad_norm": 1.407502039965593,
      "learning_rate": 0.00011499272197962154,
      "loss": 23.2189,
      "step": 395
    },
    {
      "epoch": 0.06924590163934426,
      "grad_norm": 1.5782610106328678,
      "learning_rate": 0.00011528384279475983,
      "loss": 23.3387,
      "step": 396
    },
    {
      "epoch": 0.0694207650273224,
      "grad_norm": 1.7417642665848734,
      "learning_rate": 0.00011557496360989812,
      "loss": 23.1119,
      "step": 397
    },
    {
      "epoch": 0.06959562841530055,
      "grad_norm": 1.6000346124467806,
      "learning_rate": 0.0001158660844250364,
      "loss": 23.3362,
      "step": 398
    },
    {
      "epoch": 0.06977049180327868,
      "grad_norm": 1.3545297120686592,
      "learning_rate": 0.00011615720524017469,
      "loss": 23.5147,
      "step": 399
    },
    {
      "epoch": 0.06994535519125683,
      "grad_norm": 1.4990677267392138,
      "learning_rate": 0.00011644832605531295,
      "loss": 22.8708,
      "step": 400
    },
    {
      "epoch": 0.07012021857923498,
      "grad_norm": 1.4166278645023833,
      "learning_rate": 0.00011673944687045124,
      "loss": 22.965,
      "step": 401
    },
    {
      "epoch": 0.07029508196721311,
      "grad_norm": 1.5343575010874186,
      "learning_rate": 0.00011703056768558953,
      "loss": 23.1915,
      "step": 402
    },
    {
      "epoch": 0.07046994535519126,
      "grad_norm": 1.6272514340240438,
      "learning_rate": 0.0001173216885007278,
      "loss": 23.1593,
      "step": 403
    },
    {
      "epoch": 0.0706448087431694,
      "grad_norm": 1.7659525105342464,
      "learning_rate": 0.00011761280931586609,
      "loss": 23.512,
      "step": 404
    },
    {
      "epoch": 0.07081967213114754,
      "grad_norm": 1.5093294774776185,
      "learning_rate": 0.00011790393013100438,
      "loss": 23.1701,
      "step": 405
    },
    {
      "epoch": 0.07099453551912568,
      "grad_norm": 1.5888639934320266,
      "learning_rate": 0.00011819505094614265,
      "loss": 23.0159,
      "step": 406
    },
    {
      "epoch": 0.07116939890710383,
      "grad_norm": 1.767817548196274,
      "learning_rate": 0.00011848617176128094,
      "loss": 23.0024,
      "step": 407
    },
    {
      "epoch": 0.07134426229508196,
      "grad_norm": 1.4673989778560106,
      "learning_rate": 0.00011877729257641923,
      "loss": 22.9293,
      "step": 408
    },
    {
      "epoch": 0.07151912568306011,
      "grad_norm": 1.6253302011746638,
      "learning_rate": 0.0001190684133915575,
      "loss": 23.2095,
      "step": 409
    },
    {
      "epoch": 0.07169398907103826,
      "grad_norm": 1.5720608139613514,
      "learning_rate": 0.00011935953420669579,
      "loss": 23.0159,
      "step": 410
    },
    {
      "epoch": 0.07186885245901639,
      "grad_norm": 1.4877955702903833,
      "learning_rate": 0.00011965065502183406,
      "loss": 23.0208,
      "step": 411
    },
    {
      "epoch": 0.07204371584699454,
      "grad_norm": 1.4984481373634677,
      "learning_rate": 0.00011994177583697235,
      "loss": 23.3145,
      "step": 412
    },
    {
      "epoch": 0.07221857923497267,
      "grad_norm": 1.5757387326930987,
      "learning_rate": 0.00012023289665211064,
      "loss": 23.2608,
      "step": 413
    },
    {
      "epoch": 0.07239344262295082,
      "grad_norm": 1.7678481178831238,
      "learning_rate": 0.0001205240174672489,
      "loss": 23.1327,
      "step": 414
    },
    {
      "epoch": 0.07256830601092897,
      "grad_norm": 1.8708612331527676,
      "learning_rate": 0.0001208151382823872,
      "loss": 22.994,
      "step": 415
    },
    {
      "epoch": 0.0727431693989071,
      "grad_norm": 1.9133936282471877,
      "learning_rate": 0.00012110625909752549,
      "loss": 23.1843,
      "step": 416
    },
    {
      "epoch": 0.07291803278688525,
      "grad_norm": 1.6016096555666448,
      "learning_rate": 0.00012139737991266375,
      "loss": 22.9557,
      "step": 417
    },
    {
      "epoch": 0.07309289617486339,
      "grad_norm": 1.8252430780494104,
      "learning_rate": 0.00012168850072780204,
      "loss": 23.1131,
      "step": 418
    },
    {
      "epoch": 0.07326775956284153,
      "grad_norm": 1.7012852281890778,
      "learning_rate": 0.00012197962154294033,
      "loss": 22.9941,
      "step": 419
    },
    {
      "epoch": 0.07344262295081967,
      "grad_norm": 1.5444141903609288,
      "learning_rate": 0.0001222707423580786,
      "loss": 23.0729,
      "step": 420
    },
    {
      "epoch": 0.07361748633879782,
      "grad_norm": 1.5613326968777352,
      "learning_rate": 0.00012256186317321688,
      "loss": 22.839,
      "step": 421
    },
    {
      "epoch": 0.07379234972677595,
      "grad_norm": 1.5723883380478998,
      "learning_rate": 0.00012285298398835516,
      "loss": 23.2044,
      "step": 422
    },
    {
      "epoch": 0.0739672131147541,
      "grad_norm": 1.7202003716403056,
      "learning_rate": 0.00012314410480349346,
      "loss": 22.9493,
      "step": 423
    },
    {
      "epoch": 0.07414207650273225,
      "grad_norm": 1.449964906973359,
      "learning_rate": 0.00012343522561863174,
      "loss": 23.0971,
      "step": 424
    },
    {
      "epoch": 0.07431693989071038,
      "grad_norm": 1.5910606050661162,
      "learning_rate": 0.00012372634643377002,
      "loss": 22.7028,
      "step": 425
    },
    {
      "epoch": 0.07449180327868853,
      "grad_norm": 1.7526327077703918,
      "learning_rate": 0.0001240174672489083,
      "loss": 23.1262,
      "step": 426
    },
    {
      "epoch": 0.07466666666666667,
      "grad_norm": 1.70665805528873,
      "learning_rate": 0.0001243085880640466,
      "loss": 23.2169,
      "step": 427
    },
    {
      "epoch": 0.0748415300546448,
      "grad_norm": 1.5722919495697307,
      "learning_rate": 0.00012459970887918488,
      "loss": 22.9104,
      "step": 428
    },
    {
      "epoch": 0.07501639344262295,
      "grad_norm": 1.6316238497104445,
      "learning_rate": 0.00012489082969432315,
      "loss": 23.2203,
      "step": 429
    },
    {
      "epoch": 0.0751912568306011,
      "grad_norm": 1.6687383052367324,
      "learning_rate": 0.00012518195050946146,
      "loss": 22.8183,
      "step": 430
    },
    {
      "epoch": 0.07536612021857923,
      "grad_norm": 1.4666474183576386,
      "learning_rate": 0.0001254730713245997,
      "loss": 22.7024,
      "step": 431
    },
    {
      "epoch": 0.07554098360655738,
      "grad_norm": 1.7579013997935369,
      "learning_rate": 0.000125764192139738,
      "loss": 22.9963,
      "step": 432
    },
    {
      "epoch": 0.07571584699453551,
      "grad_norm": 1.5509535318791368,
      "learning_rate": 0.00012605531295487626,
      "loss": 23.0768,
      "step": 433
    },
    {
      "epoch": 0.07589071038251366,
      "grad_norm": 1.5671793288951497,
      "learning_rate": 0.00012634643377001456,
      "loss": 22.9309,
      "step": 434
    },
    {
      "epoch": 0.0760655737704918,
      "grad_norm": 1.7535590839256112,
      "learning_rate": 0.00012663755458515284,
      "loss": 22.9734,
      "step": 435
    },
    {
      "epoch": 0.07624043715846994,
      "grad_norm": 1.7003989292114854,
      "learning_rate": 0.00012692867540029112,
      "loss": 22.5731,
      "step": 436
    },
    {
      "epoch": 0.07641530054644809,
      "grad_norm": 1.7056731894120893,
      "learning_rate": 0.00012721979621542942,
      "loss": 22.9391,
      "step": 437
    },
    {
      "epoch": 0.07659016393442623,
      "grad_norm": 1.6327547063624632,
      "learning_rate": 0.0001275109170305677,
      "loss": 22.9388,
      "step": 438
    },
    {
      "epoch": 0.07676502732240437,
      "grad_norm": 1.5004680414388534,
      "learning_rate": 0.00012780203784570597,
      "loss": 22.9921,
      "step": 439
    },
    {
      "epoch": 0.07693989071038251,
      "grad_norm": 1.6759074090075843,
      "learning_rate": 0.00012809315866084425,
      "loss": 22.7257,
      "step": 440
    },
    {
      "epoch": 0.07711475409836066,
      "grad_norm": 1.4576413052729287,
      "learning_rate": 0.00012838427947598255,
      "loss": 22.8815,
      "step": 441
    },
    {
      "epoch": 0.0772896174863388,
      "grad_norm": 1.5238191151802614,
      "learning_rate": 0.00012867540029112083,
      "loss": 22.8157,
      "step": 442
    },
    {
      "epoch": 0.07746448087431694,
      "grad_norm": 1.4673325753864486,
      "learning_rate": 0.0001289665211062591,
      "loss": 22.8471,
      "step": 443
    },
    {
      "epoch": 0.07763934426229509,
      "grad_norm": 1.2808229945814997,
      "learning_rate": 0.00012925764192139738,
      "loss": 23.039,
      "step": 444
    },
    {
      "epoch": 0.07781420765027322,
      "grad_norm": 1.3969266156664553,
      "learning_rate": 0.00012954876273653566,
      "loss": 22.8692,
      "step": 445
    },
    {
      "epoch": 0.07798907103825137,
      "grad_norm": 1.463845681874112,
      "learning_rate": 0.00012983988355167396,
      "loss": 22.8521,
      "step": 446
    },
    {
      "epoch": 0.07816393442622951,
      "grad_norm": 1.4586950150932854,
      "learning_rate": 0.00013013100436681221,
      "loss": 23.0092,
      "step": 447
    },
    {
      "epoch": 0.07833879781420765,
      "grad_norm": 1.4220709289281153,
      "learning_rate": 0.00013042212518195052,
      "loss": 22.8936,
      "step": 448
    },
    {
      "epoch": 0.0785136612021858,
      "grad_norm": 1.3993112050628034,
      "learning_rate": 0.0001307132459970888,
      "loss": 22.7261,
      "step": 449
    },
    {
      "epoch": 0.07868852459016394,
      "grad_norm": 1.2721082173766716,
      "learning_rate": 0.00013100436681222707,
      "loss": 23.015,
      "step": 450
    },
    {
      "epoch": 0.07886338797814207,
      "grad_norm": 1.4237062193174685,
      "learning_rate": 0.00013129548762736538,
      "loss": 22.9981,
      "step": 451
    },
    {
      "epoch": 0.07903825136612022,
      "grad_norm": 1.5350199394801547,
      "learning_rate": 0.00013158660844250365,
      "loss": 22.7321,
      "step": 452
    },
    {
      "epoch": 0.07921311475409835,
      "grad_norm": 1.7422623217566935,
      "learning_rate": 0.00013187772925764193,
      "loss": 22.8742,
      "step": 453
    },
    {
      "epoch": 0.0793879781420765,
      "grad_norm": 1.5254557337702308,
      "learning_rate": 0.0001321688500727802,
      "loss": 22.8924,
      "step": 454
    },
    {
      "epoch": 0.07956284153005465,
      "grad_norm": 1.3990083768951251,
      "learning_rate": 0.00013245997088791848,
      "loss": 22.6424,
      "step": 455
    },
    {
      "epoch": 0.07973770491803278,
      "grad_norm": 1.4553282787832615,
      "learning_rate": 0.00013275109170305679,
      "loss": 22.2907,
      "step": 456
    },
    {
      "epoch": 0.07991256830601093,
      "grad_norm": 1.4582354154570176,
      "learning_rate": 0.00013304221251819506,
      "loss": 22.6866,
      "step": 457
    },
    {
      "epoch": 0.08008743169398908,
      "grad_norm": 1.318334262572395,
      "learning_rate": 0.00013333333333333334,
      "loss": 22.6531,
      "step": 458
    },
    {
      "epoch": 0.08026229508196721,
      "grad_norm": 1.6829405781674536,
      "learning_rate": 0.00013362445414847162,
      "loss": 22.6634,
      "step": 459
    },
    {
      "epoch": 0.08043715846994536,
      "grad_norm": 1.685515622698371,
      "learning_rate": 0.00013391557496360992,
      "loss": 22.8268,
      "step": 460
    },
    {
      "epoch": 0.0806120218579235,
      "grad_norm": 1.5799528712864717,
      "learning_rate": 0.00013420669577874817,
      "loss": 22.3475,
      "step": 461
    },
    {
      "epoch": 0.08078688524590164,
      "grad_norm": 1.438581752254662,
      "learning_rate": 0.00013449781659388647,
      "loss": 22.5351,
      "step": 462
    },
    {
      "epoch": 0.08096174863387978,
      "grad_norm": 1.4782926101471001,
      "learning_rate": 0.00013478893740902475,
      "loss": 22.7822,
      "step": 463
    },
    {
      "epoch": 0.08113661202185793,
      "grad_norm": 1.3876596551813574,
      "learning_rate": 0.00013508005822416303,
      "loss": 22.514,
      "step": 464
    },
    {
      "epoch": 0.08131147540983606,
      "grad_norm": 1.4453530872168874,
      "learning_rate": 0.00013537117903930133,
      "loss": 22.804,
      "step": 465
    },
    {
      "epoch": 0.08148633879781421,
      "grad_norm": 1.4465012200020597,
      "learning_rate": 0.00013566229985443958,
      "loss": 22.7162,
      "step": 466
    },
    {
      "epoch": 0.08166120218579236,
      "grad_norm": 1.478938541467581,
      "learning_rate": 0.00013595342066957788,
      "loss": 22.349,
      "step": 467
    },
    {
      "epoch": 0.08183606557377049,
      "grad_norm": 1.3590285263780253,
      "learning_rate": 0.00013624454148471616,
      "loss": 22.6118,
      "step": 468
    },
    {
      "epoch": 0.08201092896174864,
      "grad_norm": 1.581211681204892,
      "learning_rate": 0.00013653566229985444,
      "loss": 22.7725,
      "step": 469
    },
    {
      "epoch": 0.08218579234972678,
      "grad_norm": 1.4505170066343975,
      "learning_rate": 0.00013682678311499274,
      "loss": 22.6525,
      "step": 470
    },
    {
      "epoch": 0.08236065573770492,
      "grad_norm": 1.518122176269526,
      "learning_rate": 0.00013711790393013102,
      "loss": 22.611,
      "step": 471
    },
    {
      "epoch": 0.08253551912568306,
      "grad_norm": 1.6320440019803404,
      "learning_rate": 0.0001374090247452693,
      "loss": 22.6619,
      "step": 472
    },
    {
      "epoch": 0.0827103825136612,
      "grad_norm": 1.5239886716001014,
      "learning_rate": 0.00013770014556040757,
      "loss": 22.5572,
      "step": 473
    },
    {
      "epoch": 0.08288524590163934,
      "grad_norm": 1.5518146213648156,
      "learning_rate": 0.00013799126637554588,
      "loss": 22.6816,
      "step": 474
    },
    {
      "epoch": 0.08306010928961749,
      "grad_norm": 1.4845141607251249,
      "learning_rate": 0.00013828238719068413,
      "loss": 22.7092,
      "step": 475
    },
    {
      "epoch": 0.08323497267759562,
      "grad_norm": 1.56897724501392,
      "learning_rate": 0.00013857350800582243,
      "loss": 22.5792,
      "step": 476
    },
    {
      "epoch": 0.08340983606557377,
      "grad_norm": 1.4514842246583897,
      "learning_rate": 0.0001388646288209607,
      "loss": 22.6063,
      "step": 477
    },
    {
      "epoch": 0.08358469945355192,
      "grad_norm": 1.4520367623479358,
      "learning_rate": 0.00013915574963609898,
      "loss": 22.1147,
      "step": 478
    },
    {
      "epoch": 0.08375956284153005,
      "grad_norm": 1.446432748820845,
      "learning_rate": 0.00013944687045123729,
      "loss": 22.7715,
      "step": 479
    },
    {
      "epoch": 0.0839344262295082,
      "grad_norm": 1.4383460591411132,
      "learning_rate": 0.00013973799126637554,
      "loss": 22.4846,
      "step": 480
    },
    {
      "epoch": 0.08410928961748634,
      "grad_norm": 1.4633115167330337,
      "learning_rate": 0.00014002911208151384,
      "loss": 22.2121,
      "step": 481
    },
    {
      "epoch": 0.08428415300546448,
      "grad_norm": 1.716734170412671,
      "learning_rate": 0.00014032023289665212,
      "loss": 22.0777,
      "step": 482
    },
    {
      "epoch": 0.08445901639344262,
      "grad_norm": 1.6848183309281108,
      "learning_rate": 0.0001406113537117904,
      "loss": 22.3715,
      "step": 483
    },
    {
      "epoch": 0.08463387978142077,
      "grad_norm": 1.6264596561432727,
      "learning_rate": 0.0001409024745269287,
      "loss": 22.4402,
      "step": 484
    },
    {
      "epoch": 0.0848087431693989,
      "grad_norm": 1.5371173047994933,
      "learning_rate": 0.00014119359534206697,
      "loss": 22.5413,
      "step": 485
    },
    {
      "epoch": 0.08498360655737705,
      "grad_norm": 1.4083117788808575,
      "learning_rate": 0.00014148471615720525,
      "loss": 22.3886,
      "step": 486
    },
    {
      "epoch": 0.0851584699453552,
      "grad_norm": 1.3588446907807274,
      "learning_rate": 0.00014177583697234353,
      "loss": 22.5361,
      "step": 487
    },
    {
      "epoch": 0.08533333333333333,
      "grad_norm": 1.3368550200411702,
      "learning_rate": 0.0001420669577874818,
      "loss": 22.6155,
      "step": 488
    },
    {
      "epoch": 0.08550819672131148,
      "grad_norm": 1.4312039917464432,
      "learning_rate": 0.00014235807860262008,
      "loss": 22.4005,
      "step": 489
    },
    {
      "epoch": 0.08568306010928962,
      "grad_norm": 1.4744227088945228,
      "learning_rate": 0.00014264919941775838,
      "loss": 22.2673,
      "step": 490
    },
    {
      "epoch": 0.08585792349726776,
      "grad_norm": 1.3969404698604655,
      "learning_rate": 0.00014294032023289666,
      "loss": 22.4473,
      "step": 491
    },
    {
      "epoch": 0.0860327868852459,
      "grad_norm": 1.5932381400105993,
      "learning_rate": 0.00014323144104803494,
      "loss": 22.0964,
      "step": 492
    },
    {
      "epoch": 0.08620765027322404,
      "grad_norm": 1.4292217771087823,
      "learning_rate": 0.00014352256186317324,
      "loss": 22.5321,
      "step": 493
    },
    {
      "epoch": 0.08638251366120218,
      "grad_norm": 1.3507717787136788,
      "learning_rate": 0.0001438136826783115,
      "loss": 22.528,
      "step": 494
    },
    {
      "epoch": 0.08655737704918033,
      "grad_norm": 1.4572532049869362,
      "learning_rate": 0.0001441048034934498,
      "loss": 21.9636,
      "step": 495
    },
    {
      "epoch": 0.08673224043715846,
      "grad_norm": 1.5040216870467895,
      "learning_rate": 0.00014439592430858807,
      "loss": 22.1324,
      "step": 496
    },
    {
      "epoch": 0.08690710382513661,
      "grad_norm": 1.5196775602457315,
      "learning_rate": 0.00014468704512372635,
      "loss": 22.2759,
      "step": 497
    },
    {
      "epoch": 0.08708196721311476,
      "grad_norm": 1.5043995068861449,
      "learning_rate": 0.00014497816593886465,
      "loss": 22.273,
      "step": 498
    },
    {
      "epoch": 0.08725683060109289,
      "grad_norm": 1.4134075172126126,
      "learning_rate": 0.0001452692867540029,
      "loss": 22.1873,
      "step": 499
    },
    {
      "epoch": 0.08743169398907104,
      "grad_norm": 1.3908874282262231,
      "learning_rate": 0.0001455604075691412,
      "loss": 22.3984,
      "step": 500
    },
    {
      "epoch": 0.08760655737704919,
      "grad_norm": 1.3403051096903207,
      "learning_rate": 0.00014585152838427948,
      "loss": 22.4374,
      "step": 501
    },
    {
      "epoch": 0.08778142076502732,
      "grad_norm": 1.4045429039684316,
      "learning_rate": 0.00014614264919941776,
      "loss": 22.3605,
      "step": 502
    },
    {
      "epoch": 0.08795628415300547,
      "grad_norm": 1.4467104947291989,
      "learning_rate": 0.00014643377001455604,
      "loss": 22.5123,
      "step": 503
    },
    {
      "epoch": 0.08813114754098361,
      "grad_norm": 1.4551466681969414,
      "learning_rate": 0.00014672489082969434,
      "loss": 22.2153,
      "step": 504
    },
    {
      "epoch": 0.08830601092896175,
      "grad_norm": 1.286924417706148,
      "learning_rate": 0.00014701601164483262,
      "loss": 22.2911,
      "step": 505
    },
    {
      "epoch": 0.08848087431693989,
      "grad_norm": 1.3972315970638476,
      "learning_rate": 0.0001473071324599709,
      "loss": 22.452,
      "step": 506
    },
    {
      "epoch": 0.08865573770491804,
      "grad_norm": 1.4700128562068457,
      "learning_rate": 0.0001475982532751092,
      "loss": 22.482,
      "step": 507
    },
    {
      "epoch": 0.08883060109289617,
      "grad_norm": 1.421977695715378,
      "learning_rate": 0.00014788937409024745,
      "loss": 22.198,
      "step": 508
    },
    {
      "epoch": 0.08900546448087432,
      "grad_norm": 1.452104478670894,
      "learning_rate": 0.00014818049490538575,
      "loss": 21.7293,
      "step": 509
    },
    {
      "epoch": 0.08918032786885247,
      "grad_norm": 1.4697217115235794,
      "learning_rate": 0.00014847161572052403,
      "loss": 22.3597,
      "step": 510
    },
    {
      "epoch": 0.0893551912568306,
      "grad_norm": 1.4122076762801012,
      "learning_rate": 0.0001487627365356623,
      "loss": 22.4375,
      "step": 511
    },
    {
      "epoch": 0.08953005464480875,
      "grad_norm": 1.370530437339351,
      "learning_rate": 0.0001490538573508006,
      "loss": 22.0387,
      "step": 512
    },
    {
      "epoch": 0.08970491803278688,
      "grad_norm": 1.375552579633963,
      "learning_rate": 0.00014934497816593886,
      "loss": 22.2982,
      "step": 513
    },
    {
      "epoch": 0.08987978142076503,
      "grad_norm": 1.3429248156388356,
      "learning_rate": 0.00014963609898107716,
      "loss": 22.2496,
      "step": 514
    },
    {
      "epoch": 0.09005464480874317,
      "grad_norm": 1.5699616431785262,
      "learning_rate": 0.00014992721979621544,
      "loss": 22.0034,
      "step": 515
    },
    {
      "epoch": 0.0902295081967213,
      "grad_norm": 1.6080712543419378,
      "learning_rate": 0.00015021834061135371,
      "loss": 22.3073,
      "step": 516
    },
    {
      "epoch": 0.09040437158469945,
      "grad_norm": 1.599558082016222,
      "learning_rate": 0.000150509461426492,
      "loss": 22.0225,
      "step": 517
    },
    {
      "epoch": 0.0905792349726776,
      "grad_norm": 1.3346708044652396,
      "learning_rate": 0.00015080058224163027,
      "loss": 21.9341,
      "step": 518
    },
    {
      "epoch": 0.09075409836065573,
      "grad_norm": 1.5671260208201225,
      "learning_rate": 0.00015109170305676857,
      "loss": 21.8986,
      "step": 519
    },
    {
      "epoch": 0.09092896174863388,
      "grad_norm": 1.5864355212860883,
      "learning_rate": 0.00015138282387190685,
      "loss": 21.8989,
      "step": 520
    },
    {
      "epoch": 0.09110382513661203,
      "grad_norm": 1.509458350514079,
      "learning_rate": 0.00015167394468704513,
      "loss": 22.3104,
      "step": 521
    },
    {
      "epoch": 0.09127868852459016,
      "grad_norm": 1.5367360368201364,
      "learning_rate": 0.0001519650655021834,
      "loss": 22.2397,
      "step": 522
    },
    {
      "epoch": 0.0914535519125683,
      "grad_norm": 1.6970694761859506,
      "learning_rate": 0.0001522561863173217,
      "loss": 21.9828,
      "step": 523
    },
    {
      "epoch": 0.09162841530054645,
      "grad_norm": 1.5276838666355452,
      "learning_rate": 0.00015254730713245998,
      "loss": 22.1673,
      "step": 524
    },
    {
      "epoch": 0.09180327868852459,
      "grad_norm": 1.4983257637223737,
      "learning_rate": 0.00015283842794759826,
      "loss": 22.4933,
      "step": 525
    },
    {
      "epoch": 0.09197814207650273,
      "grad_norm": 1.562914128096435,
      "learning_rate": 0.00015312954876273656,
      "loss": 22.4053,
      "step": 526
    },
    {
      "epoch": 0.09215300546448088,
      "grad_norm": 1.4631652767944567,
      "learning_rate": 0.0001534206695778748,
      "loss": 21.9225,
      "step": 527
    },
    {
      "epoch": 0.09232786885245901,
      "grad_norm": 1.5055019574056663,
      "learning_rate": 0.00015371179039301312,
      "loss": 22.0535,
      "step": 528
    },
    {
      "epoch": 0.09250273224043716,
      "grad_norm": 1.479238830495733,
      "learning_rate": 0.00015400291120815137,
      "loss": 22.2302,
      "step": 529
    },
    {
      "epoch": 0.09267759562841531,
      "grad_norm": 1.597265904682234,
      "learning_rate": 0.00015429403202328967,
      "loss": 22.3833,
      "step": 530
    },
    {
      "epoch": 0.09285245901639344,
      "grad_norm": 1.5398852655131015,
      "learning_rate": 0.00015458515283842795,
      "loss": 22.2323,
      "step": 531
    },
    {
      "epoch": 0.09302732240437159,
      "grad_norm": 1.4796253955369119,
      "learning_rate": 0.00015487627365356622,
      "loss": 21.8137,
      "step": 532
    },
    {
      "epoch": 0.09320218579234972,
      "grad_norm": 1.3831431374589278,
      "learning_rate": 0.00015516739446870453,
      "loss": 22.273,
      "step": 533
    },
    {
      "epoch": 0.09337704918032787,
      "grad_norm": 1.3882398332491057,
      "learning_rate": 0.0001554585152838428,
      "loss": 22.2597,
      "step": 534
    },
    {
      "epoch": 0.09355191256830601,
      "grad_norm": 1.44914137896571,
      "learning_rate": 0.00015574963609898108,
      "loss": 22.0332,
      "step": 535
    },
    {
      "epoch": 0.09372677595628415,
      "grad_norm": 1.2992646586248215,
      "learning_rate": 0.00015604075691411936,
      "loss": 22.3205,
      "step": 536
    },
    {
      "epoch": 0.0939016393442623,
      "grad_norm": 1.5095619951096872,
      "learning_rate": 0.00015633187772925766,
      "loss": 22.3049,
      "step": 537
    },
    {
      "epoch": 0.09407650273224044,
      "grad_norm": 1.4141797662941924,
      "learning_rate": 0.00015662299854439594,
      "loss": 22.4709,
      "step": 538
    },
    {
      "epoch": 0.09425136612021857,
      "grad_norm": 1.450944408371475,
      "learning_rate": 0.00015691411935953421,
      "loss": 22.0584,
      "step": 539
    },
    {
      "epoch": 0.09442622950819672,
      "grad_norm": 1.3864968136391909,
      "learning_rate": 0.0001572052401746725,
      "loss": 22.2214,
      "step": 540
    },
    {
      "epoch": 0.09460109289617487,
      "grad_norm": 1.4507920047411949,
      "learning_rate": 0.00015749636098981077,
      "loss": 21.8695,
      "step": 541
    },
    {
      "epoch": 0.094775956284153,
      "grad_norm": 1.3024795962518598,
      "learning_rate": 0.00015778748180494907,
      "loss": 21.8023,
      "step": 542
    },
    {
      "epoch": 0.09495081967213115,
      "grad_norm": 1.3736802123460847,
      "learning_rate": 0.00015807860262008732,
      "loss": 21.6333,
      "step": 543
    },
    {
      "epoch": 0.0951256830601093,
      "grad_norm": 1.2361520490317528,
      "learning_rate": 0.00015836972343522563,
      "loss": 21.842,
      "step": 544
    },
    {
      "epoch": 0.09530054644808743,
      "grad_norm": 1.3365014173646153,
      "learning_rate": 0.00015866084425036393,
      "loss": 22.3086,
      "step": 545
    },
    {
      "epoch": 0.09547540983606558,
      "grad_norm": 1.4363637253258235,
      "learning_rate": 0.00015895196506550218,
      "loss": 22.1006,
      "step": 546
    },
    {
      "epoch": 0.09565027322404372,
      "grad_norm": 1.401724142096061,
      "learning_rate": 0.00015924308588064048,
      "loss": 22.047,
      "step": 547
    },
    {
      "epoch": 0.09582513661202186,
      "grad_norm": 1.3889925389020148,
      "learning_rate": 0.00015953420669577876,
      "loss": 21.9344,
      "step": 548
    },
    {
      "epoch": 0.096,
      "grad_norm": 1.2340031486432443,
      "learning_rate": 0.00015982532751091704,
      "loss": 22.3023,
      "step": 549
    },
    {
      "epoch": 0.09617486338797815,
      "grad_norm": 1.334001010609382,
      "learning_rate": 0.0001601164483260553,
      "loss": 22.1635,
      "step": 550
    },
    {
      "epoch": 0.09634972677595628,
      "grad_norm": 1.389372850675959,
      "learning_rate": 0.0001604075691411936,
      "loss": 22.1411,
      "step": 551
    },
    {
      "epoch": 0.09652459016393443,
      "grad_norm": 1.3316901317141636,
      "learning_rate": 0.0001606986899563319,
      "loss": 22.0842,
      "step": 552
    },
    {
      "epoch": 0.09669945355191256,
      "grad_norm": 1.2649477682165418,
      "learning_rate": 0.00016098981077147017,
      "loss": 21.9649,
      "step": 553
    },
    {
      "epoch": 0.09687431693989071,
      "grad_norm": 1.3143071840167686,
      "learning_rate": 0.00016128093158660845,
      "loss": 21.9495,
      "step": 554
    },
    {
      "epoch": 0.09704918032786886,
      "grad_norm": 1.4356123825783207,
      "learning_rate": 0.00016157205240174672,
      "loss": 21.9288,
      "step": 555
    },
    {
      "epoch": 0.09722404371584699,
      "grad_norm": 1.593493260404579,
      "learning_rate": 0.00016186317321688503,
      "loss": 21.8751,
      "step": 556
    },
    {
      "epoch": 0.09739890710382514,
      "grad_norm": 1.4694204458567413,
      "learning_rate": 0.00016215429403202328,
      "loss": 22.2,
      "step": 557
    },
    {
      "epoch": 0.09757377049180328,
      "grad_norm": 1.627405870880949,
      "learning_rate": 0.00016244541484716158,
      "loss": 21.4713,
      "step": 558
    },
    {
      "epoch": 0.09774863387978142,
      "grad_norm": 1.3963772425911007,
      "learning_rate": 0.00016273653566229988,
      "loss": 21.5944,
      "step": 559
    },
    {
      "epoch": 0.09792349726775956,
      "grad_norm": 1.2881118676862169,
      "learning_rate": 0.00016302765647743813,
      "loss": 22.0721,
      "step": 560
    },
    {
      "epoch": 0.09809836065573771,
      "grad_norm": 1.3794239515783528,
      "learning_rate": 0.00016331877729257644,
      "loss": 21.9058,
      "step": 561
    },
    {
      "epoch": 0.09827322404371584,
      "grad_norm": 1.3423625930480436,
      "learning_rate": 0.0001636098981077147,
      "loss": 21.9508,
      "step": 562
    },
    {
      "epoch": 0.09844808743169399,
      "grad_norm": 1.3274088303566092,
      "learning_rate": 0.000163901018922853,
      "loss": 21.9788,
      "step": 563
    },
    {
      "epoch": 0.09862295081967214,
      "grad_norm": 1.507267379862649,
      "learning_rate": 0.00016419213973799127,
      "loss": 22.0353,
      "step": 564
    },
    {
      "epoch": 0.09879781420765027,
      "grad_norm": 1.4578830287332831,
      "learning_rate": 0.00016448326055312955,
      "loss": 21.984,
      "step": 565
    },
    {
      "epoch": 0.09897267759562842,
      "grad_norm": 1.529987170295893,
      "learning_rate": 0.00016477438136826785,
      "loss": 21.9447,
      "step": 566
    },
    {
      "epoch": 0.09914754098360656,
      "grad_norm": 1.3047649115094935,
      "learning_rate": 0.00016506550218340613,
      "loss": 21.8526,
      "step": 567
    },
    {
      "epoch": 0.0993224043715847,
      "grad_norm": 1.4275832354522628,
      "learning_rate": 0.0001653566229985444,
      "loss": 22.1092,
      "step": 568
    },
    {
      "epoch": 0.09949726775956284,
      "grad_norm": 1.4094101389261473,
      "learning_rate": 0.00016564774381368268,
      "loss": 21.7813,
      "step": 569
    },
    {
      "epoch": 0.09967213114754099,
      "grad_norm": 1.4123434506696892,
      "learning_rate": 0.00016593886462882098,
      "loss": 22.1264,
      "step": 570
    },
    {
      "epoch": 0.09984699453551912,
      "grad_norm": 1.3244227289445312,
      "learning_rate": 0.00016622998544395926,
      "loss": 21.8615,
      "step": 571
    },
    {
      "epoch": 0.10002185792349727,
      "grad_norm": 1.4127877984615547,
      "learning_rate": 0.00016652110625909754,
      "loss": 22.0976,
      "step": 572
    },
    {
      "epoch": 0.1001967213114754,
      "grad_norm": 1.492996523948267,
      "learning_rate": 0.0001668122270742358,
      "loss": 22.1721,
      "step": 573
    },
    {
      "epoch": 0.10037158469945355,
      "grad_norm": 1.5488295377527597,
      "learning_rate": 0.0001671033478893741,
      "loss": 21.6584,
      "step": 574
    },
    {
      "epoch": 0.1005464480874317,
      "grad_norm": 1.5690601251143899,
      "learning_rate": 0.0001673944687045124,
      "loss": 21.7101,
      "step": 575
    },
    {
      "epoch": 0.10072131147540983,
      "grad_norm": 1.6718152132564577,
      "learning_rate": 0.00016768558951965064,
      "loss": 22.0258,
      "step": 576
    },
    {
      "epoch": 0.10089617486338798,
      "grad_norm": 1.4855615072051465,
      "learning_rate": 0.00016797671033478895,
      "loss": 22.011,
      "step": 577
    },
    {
      "epoch": 0.10107103825136612,
      "grad_norm": 1.691893907890245,
      "learning_rate": 0.00016826783114992722,
      "loss": 22.0605,
      "step": 578
    },
    {
      "epoch": 0.10124590163934426,
      "grad_norm": 1.4627405858241282,
      "learning_rate": 0.0001685589519650655,
      "loss": 21.6257,
      "step": 579
    },
    {
      "epoch": 0.1014207650273224,
      "grad_norm": 1.3481884215857318,
      "learning_rate": 0.0001688500727802038,
      "loss": 21.9238,
      "step": 580
    },
    {
      "epoch": 0.10159562841530055,
      "grad_norm": 1.4454545306826156,
      "learning_rate": 0.00016914119359534208,
      "loss": 22.0198,
      "step": 581
    },
    {
      "epoch": 0.10177049180327868,
      "grad_norm": 1.3237021834268017,
      "learning_rate": 0.00016943231441048036,
      "loss": 21.5658,
      "step": 582
    },
    {
      "epoch": 0.10194535519125683,
      "grad_norm": 1.375646388833536,
      "learning_rate": 0.00016972343522561863,
      "loss": 21.5363,
      "step": 583
    },
    {
      "epoch": 0.10212021857923498,
      "grad_norm": 1.39694610711503,
      "learning_rate": 0.0001700145560407569,
      "loss": 21.5553,
      "step": 584
    },
    {
      "epoch": 0.10229508196721311,
      "grad_norm": 1.457008981049635,
      "learning_rate": 0.00017030567685589521,
      "loss": 21.6849,
      "step": 585
    },
    {
      "epoch": 0.10246994535519126,
      "grad_norm": 1.409675986569437,
      "learning_rate": 0.0001705967976710335,
      "loss": 21.9126,
      "step": 586
    },
    {
      "epoch": 0.1026448087431694,
      "grad_norm": 1.4167465599323392,
      "learning_rate": 0.00017088791848617177,
      "loss": 21.768,
      "step": 587
    },
    {
      "epoch": 0.10281967213114754,
      "grad_norm": 1.3603954653901718,
      "learning_rate": 0.00017117903930131005,
      "loss": 21.5731,
      "step": 588
    },
    {
      "epoch": 0.10299453551912569,
      "grad_norm": 1.3660542694018107,
      "learning_rate": 0.00017147016011644835,
      "loss": 21.6579,
      "step": 589
    },
    {
      "epoch": 0.10316939890710383,
      "grad_norm": 1.4195895064915969,
      "learning_rate": 0.0001717612809315866,
      "loss": 21.8757,
      "step": 590
    },
    {
      "epoch": 0.10334426229508197,
      "grad_norm": 1.362208689373076,
      "learning_rate": 0.0001720524017467249,
      "loss": 21.7715,
      "step": 591
    },
    {
      "epoch": 0.10351912568306011,
      "grad_norm": 1.266035084602043,
      "learning_rate": 0.00017234352256186318,
      "loss": 21.9304,
      "step": 592
    },
    {
      "epoch": 0.10369398907103825,
      "grad_norm": 1.2358019411598884,
      "learning_rate": 0.00017263464337700146,
      "loss": 21.9384,
      "step": 593
    },
    {
      "epoch": 0.10386885245901639,
      "grad_norm": 1.2743352478067393,
      "learning_rate": 0.00017292576419213976,
      "loss": 21.968,
      "step": 594
    },
    {
      "epoch": 0.10404371584699454,
      "grad_norm": 1.246500874788049,
      "learning_rate": 0.000173216885007278,
      "loss": 22.0048,
      "step": 595
    },
    {
      "epoch": 0.10421857923497267,
      "grad_norm": 1.2464230480318714,
      "learning_rate": 0.0001735080058224163,
      "loss": 21.7034,
      "step": 596
    },
    {
      "epoch": 0.10439344262295082,
      "grad_norm": 1.2329045122137428,
      "learning_rate": 0.0001737991266375546,
      "loss": 21.8193,
      "step": 597
    },
    {
      "epoch": 0.10456830601092897,
      "grad_norm": 1.424769037119886,
      "learning_rate": 0.00017409024745269287,
      "loss": 21.7963,
      "step": 598
    },
    {
      "epoch": 0.1047431693989071,
      "grad_norm": 1.2154921692285876,
      "learning_rate": 0.00017438136826783117,
      "loss": 21.9275,
      "step": 599
    },
    {
      "epoch": 0.10491803278688525,
      "grad_norm": 1.3460768849693758,
      "learning_rate": 0.00017467248908296945,
      "loss": 21.7539,
      "step": 600
    },
    {
      "epoch": 0.10509289617486339,
      "grad_norm": 1.4194572288519696,
      "learning_rate": 0.00017496360989810772,
      "loss": 21.5784,
      "step": 601
    },
    {
      "epoch": 0.10526775956284153,
      "grad_norm": 1.4946945254203137,
      "learning_rate": 0.000175254730713246,
      "loss": 21.1626,
      "step": 602
    },
    {
      "epoch": 0.10544262295081967,
      "grad_norm": 1.373281172885359,
      "learning_rate": 0.0001755458515283843,
      "loss": 21.4467,
      "step": 603
    },
    {
      "epoch": 0.10561748633879782,
      "grad_norm": 1.6018749089646847,
      "learning_rate": 0.00017583697234352255,
      "loss": 21.4513,
      "step": 604
    },
    {
      "epoch": 0.10579234972677595,
      "grad_norm": 1.7595575614921646,
      "learning_rate": 0.00017612809315866086,
      "loss": 21.5402,
      "step": 605
    },
    {
      "epoch": 0.1059672131147541,
      "grad_norm": 1.7261106772228527,
      "learning_rate": 0.00017641921397379913,
      "loss": 21.8483,
      "step": 606
    },
    {
      "epoch": 0.10614207650273225,
      "grad_norm": 1.5004891590798635,
      "learning_rate": 0.0001767103347889374,
      "loss": 21.6288,
      "step": 607
    },
    {
      "epoch": 0.10631693989071038,
      "grad_norm": 1.746550444742766,
      "learning_rate": 0.00017700145560407572,
      "loss": 21.5064,
      "step": 608
    },
    {
      "epoch": 0.10649180327868853,
      "grad_norm": 1.4407676731521317,
      "learning_rate": 0.00017729257641921396,
      "loss": 21.481,
      "step": 609
    },
    {
      "epoch": 0.10666666666666667,
      "grad_norm": 1.374131815928321,
      "learning_rate": 0.00017758369723435227,
      "loss": 21.7729,
      "step": 610
    },
    {
      "epoch": 0.1068415300546448,
      "grad_norm": 1.406425636076739,
      "learning_rate": 0.00017787481804949055,
      "loss": 21.9492,
      "step": 611
    },
    {
      "epoch": 0.10701639344262295,
      "grad_norm": 1.389335571498122,
      "learning_rate": 0.00017816593886462882,
      "loss": 22.1293,
      "step": 612
    },
    {
      "epoch": 0.10719125683060109,
      "grad_norm": 1.4966718290226284,
      "learning_rate": 0.00017845705967976713,
      "loss": 21.639,
      "step": 613
    },
    {
      "epoch": 0.10736612021857923,
      "grad_norm": 1.4143708043326926,
      "learning_rate": 0.0001787481804949054,
      "loss": 21.3856,
      "step": 614
    },
    {
      "epoch": 0.10754098360655738,
      "grad_norm": 1.4078307936933814,
      "learning_rate": 0.00017903930131004368,
      "loss": 21.273,
      "step": 615
    },
    {
      "epoch": 0.10771584699453551,
      "grad_norm": 1.3120217061571497,
      "learning_rate": 0.00017933042212518196,
      "loss": 21.9017,
      "step": 616
    },
    {
      "epoch": 0.10789071038251366,
      "grad_norm": 1.407008884015569,
      "learning_rate": 0.00017962154294032023,
      "loss": 21.578,
      "step": 617
    },
    {
      "epoch": 0.10806557377049181,
      "grad_norm": 1.40128861287518,
      "learning_rate": 0.0001799126637554585,
      "loss": 21.1404,
      "step": 618
    },
    {
      "epoch": 0.10824043715846994,
      "grad_norm": 1.384785157709237,
      "learning_rate": 0.0001802037845705968,
      "loss": 21.8182,
      "step": 619
    },
    {
      "epoch": 0.10841530054644809,
      "grad_norm": 1.362162332802793,
      "learning_rate": 0.0001804949053857351,
      "loss": 21.5367,
      "step": 620
    },
    {
      "epoch": 0.10859016393442623,
      "grad_norm": 1.4053907582102796,
      "learning_rate": 0.00018078602620087337,
      "loss": 21.3859,
      "step": 621
    },
    {
      "epoch": 0.10876502732240437,
      "grad_norm": 1.4421536580952798,
      "learning_rate": 0.00018107714701601167,
      "loss": 21.7444,
      "step": 622
    },
    {
      "epoch": 0.10893989071038251,
      "grad_norm": 1.2950506914110667,
      "learning_rate": 0.00018136826783114992,
      "loss": 21.8832,
      "step": 623
    },
    {
      "epoch": 0.10911475409836066,
      "grad_norm": 1.337751232883318,
      "learning_rate": 0.00018165938864628822,
      "loss": 21.6633,
      "step": 624
    },
    {
      "epoch": 0.1092896174863388,
      "grad_norm": 1.443785605454688,
      "learning_rate": 0.0001819505094614265,
      "loss": 21.7408,
      "step": 625
    },
    {
      "epoch": 0.10946448087431694,
      "grad_norm": 1.39933564780641,
      "learning_rate": 0.00018224163027656478,
      "loss": 21.7031,
      "step": 626
    },
    {
      "epoch": 0.10963934426229509,
      "grad_norm": 1.3631719260298274,
      "learning_rate": 0.00018253275109170308,
      "loss": 21.7808,
      "step": 627
    },
    {
      "epoch": 0.10981420765027322,
      "grad_norm": 1.4400429512666104,
      "learning_rate": 0.00018282387190684133,
      "loss": 21.4784,
      "step": 628
    },
    {
      "epoch": 0.10998907103825137,
      "grad_norm": 1.4290801367986647,
      "learning_rate": 0.00018311499272197963,
      "loss": 21.1977,
      "step": 629
    },
    {
      "epoch": 0.11016393442622952,
      "grad_norm": 1.4503441021431465,
      "learning_rate": 0.0001834061135371179,
      "loss": 21.8117,
      "step": 630
    },
    {
      "epoch": 0.11033879781420765,
      "grad_norm": 1.4628299258857562,
      "learning_rate": 0.0001836972343522562,
      "loss": 21.3241,
      "step": 631
    },
    {
      "epoch": 0.1105136612021858,
      "grad_norm": 1.271560440125128,
      "learning_rate": 0.00018398835516739446,
      "loss": 21.2958,
      "step": 632
    },
    {
      "epoch": 0.11068852459016393,
      "grad_norm": 1.358261631752825,
      "learning_rate": 0.00018427947598253277,
      "loss": 21.5342,
      "step": 633
    },
    {
      "epoch": 0.11086338797814208,
      "grad_norm": 1.4298137454347326,
      "learning_rate": 0.00018457059679767105,
      "loss": 20.8222,
      "step": 634
    },
    {
      "epoch": 0.11103825136612022,
      "grad_norm": 1.303235402517678,
      "learning_rate": 0.00018486171761280932,
      "loss": 21.3385,
      "step": 635
    },
    {
      "epoch": 0.11121311475409836,
      "grad_norm": 1.3241753810391426,
      "learning_rate": 0.00018515283842794763,
      "loss": 21.5962,
      "step": 636
    },
    {
      "epoch": 0.1113879781420765,
      "grad_norm": 1.2249677321923924,
      "learning_rate": 0.00018544395924308588,
      "loss": 21.6908,
      "step": 637
    },
    {
      "epoch": 0.11156284153005465,
      "grad_norm": 1.2644967826208733,
      "learning_rate": 0.00018573508005822418,
      "loss": 21.5706,
      "step": 638
    },
    {
      "epoch": 0.11173770491803278,
      "grad_norm": 1.408204182904665,
      "learning_rate": 0.00018602620087336246,
      "loss": 21.6293,
      "step": 639
    },
    {
      "epoch": 0.11191256830601093,
      "grad_norm": 1.5048060734688002,
      "learning_rate": 0.00018631732168850073,
      "loss": 21.4851,
      "step": 640
    },
    {
      "epoch": 0.11208743169398908,
      "grad_norm": 1.28680340062844,
      "learning_rate": 0.00018660844250363904,
      "loss": 21.5735,
      "step": 641
    },
    {
      "epoch": 0.11226229508196721,
      "grad_norm": 1.2222234847861604,
      "learning_rate": 0.00018689956331877729,
      "loss": 21.3579,
      "step": 642
    },
    {
      "epoch": 0.11243715846994536,
      "grad_norm": 1.3867732545401024,
      "learning_rate": 0.0001871906841339156,
      "loss": 21.364,
      "step": 643
    },
    {
      "epoch": 0.1126120218579235,
      "grad_norm": 1.457089274709064,
      "learning_rate": 0.00018748180494905387,
      "loss": 21.3702,
      "step": 644
    },
    {
      "epoch": 0.11278688524590164,
      "grad_norm": 1.2539171389958024,
      "learning_rate": 0.00018777292576419214,
      "loss": 21.5108,
      "step": 645
    },
    {
      "epoch": 0.11296174863387978,
      "grad_norm": 1.2706367810309376,
      "learning_rate": 0.00018806404657933042,
      "loss": 21.57,
      "step": 646
    },
    {
      "epoch": 0.11313661202185793,
      "grad_norm": 1.2888957288759606,
      "learning_rate": 0.00018835516739446872,
      "loss": 21.2841,
      "step": 647
    },
    {
      "epoch": 0.11331147540983606,
      "grad_norm": 1.2950278483586692,
      "learning_rate": 0.000188646288209607,
      "loss": 21.7643,
      "step": 648
    },
    {
      "epoch": 0.11348633879781421,
      "grad_norm": 1.3339637335404704,
      "learning_rate": 0.00018893740902474528,
      "loss": 21.3407,
      "step": 649
    },
    {
      "epoch": 0.11366120218579236,
      "grad_norm": 1.3088860757925103,
      "learning_rate": 0.00018922852983988355,
      "loss": 21.7132,
      "step": 650
    },
    {
      "epoch": 0.11383606557377049,
      "grad_norm": 1.3102508848177488,
      "learning_rate": 0.00018951965065502183,
      "loss": 21.3275,
      "step": 651
    },
    {
      "epoch": 0.11401092896174864,
      "grad_norm": 1.316057927588679,
      "learning_rate": 0.00018981077147016013,
      "loss": 21.7666,
      "step": 652
    },
    {
      "epoch": 0.11418579234972677,
      "grad_norm": 1.406560266878359,
      "learning_rate": 0.0001901018922852984,
      "loss": 21.1598,
      "step": 653
    },
    {
      "epoch": 0.11436065573770492,
      "grad_norm": 1.23835112426514,
      "learning_rate": 0.0001903930131004367,
      "loss": 21.274,
      "step": 654
    },
    {
      "epoch": 0.11453551912568306,
      "grad_norm": 1.4350798636654338,
      "learning_rate": 0.000190684133915575,
      "loss": 21.2106,
      "step": 655
    },
    {
      "epoch": 0.1147103825136612,
      "grad_norm": 1.384719072573009,
      "learning_rate": 0.00019097525473071324,
      "loss": 21.5876,
      "step": 656
    },
    {
      "epoch": 0.11488524590163934,
      "grad_norm": 1.4377312902076536,
      "learning_rate": 0.00019126637554585155,
      "loss": 21.3774,
      "step": 657
    },
    {
      "epoch": 0.11506010928961749,
      "grad_norm": 1.650348802960177,
      "learning_rate": 0.00019155749636098982,
      "loss": 21.3125,
      "step": 658
    },
    {
      "epoch": 0.11523497267759562,
      "grad_norm": 1.530066886600272,
      "learning_rate": 0.0001918486171761281,
      "loss": 21.5736,
      "step": 659
    },
    {
      "epoch": 0.11540983606557377,
      "grad_norm": 1.2836130523033733,
      "learning_rate": 0.00019213973799126638,
      "loss": 21.3604,
      "step": 660
    },
    {
      "epoch": 0.11558469945355192,
      "grad_norm": 1.3777582540898112,
      "learning_rate": 0.00019243085880640465,
      "loss": 21.4882,
      "step": 661
    },
    {
      "epoch": 0.11575956284153005,
      "grad_norm": 1.3597579826359423,
      "learning_rate": 0.00019272197962154296,
      "loss": 21.3784,
      "step": 662
    },
    {
      "epoch": 0.1159344262295082,
      "grad_norm": 1.3252133603245067,
      "learning_rate": 0.00019301310043668123,
      "loss": 20.862,
      "step": 663
    },
    {
      "epoch": 0.11610928961748634,
      "grad_norm": 1.277924882914437,
      "learning_rate": 0.0001933042212518195,
      "loss": 21.5051,
      "step": 664
    },
    {
      "epoch": 0.11628415300546448,
      "grad_norm": 1.263424453758219,
      "learning_rate": 0.00019359534206695779,
      "loss": 21.3973,
      "step": 665
    },
    {
      "epoch": 0.11645901639344262,
      "grad_norm": 1.2028999871305226,
      "learning_rate": 0.0001938864628820961,
      "loss": 21.2383,
      "step": 666
    },
    {
      "epoch": 0.11663387978142077,
      "grad_norm": 1.2340423925266857,
      "learning_rate": 0.00019417758369723437,
      "loss": 21.3394,
      "step": 667
    },
    {
      "epoch": 0.1168087431693989,
      "grad_norm": 1.1473558339459906,
      "learning_rate": 0.00019446870451237264,
      "loss": 21.2379,
      "step": 668
    },
    {
      "epoch": 0.11698360655737705,
      "grad_norm": 1.2754821305057509,
      "learning_rate": 0.00019475982532751095,
      "loss": 21.1828,
      "step": 669
    },
    {
      "epoch": 0.1171584699453552,
      "grad_norm": 1.192620726432035,
      "learning_rate": 0.0001950509461426492,
      "loss": 21.207,
      "step": 670
    },
    {
      "epoch": 0.11733333333333333,
      "grad_norm": 1.1947939607773086,
      "learning_rate": 0.0001953420669577875,
      "loss": 21.3577,
      "step": 671
    },
    {
      "epoch": 0.11750819672131148,
      "grad_norm": 1.299697456540166,
      "learning_rate": 0.00019563318777292575,
      "loss": 21.3456,
      "step": 672
    },
    {
      "epoch": 0.11768306010928961,
      "grad_norm": 1.2307101098110667,
      "learning_rate": 0.00019592430858806405,
      "loss": 21.5448,
      "step": 673
    },
    {
      "epoch": 0.11785792349726776,
      "grad_norm": 1.2681991123673637,
      "learning_rate": 0.00019621542940320233,
      "loss": 21.453,
      "step": 674
    },
    {
      "epoch": 0.1180327868852459,
      "grad_norm": 1.2826144157038366,
      "learning_rate": 0.0001965065502183406,
      "loss": 21.241,
      "step": 675
    },
    {
      "epoch": 0.11820765027322404,
      "grad_norm": 1.3295557892629974,
      "learning_rate": 0.0001967976710334789,
      "loss": 21.3696,
      "step": 676
    },
    {
      "epoch": 0.11838251366120218,
      "grad_norm": 1.2185367268697351,
      "learning_rate": 0.0001970887918486172,
      "loss": 21.3411,
      "step": 677
    },
    {
      "epoch": 0.11855737704918033,
      "grad_norm": 1.2774297241800645,
      "learning_rate": 0.00019737991266375546,
      "loss": 21.2517,
      "step": 678
    },
    {
      "epoch": 0.11873224043715846,
      "grad_norm": 1.2419414244227434,
      "learning_rate": 0.00019767103347889374,
      "loss": 21.2824,
      "step": 679
    },
    {
      "epoch": 0.11890710382513661,
      "grad_norm": 1.2513834234643353,
      "learning_rate": 0.00019796215429403205,
      "loss": 21.126,
      "step": 680
    },
    {
      "epoch": 0.11908196721311476,
      "grad_norm": 1.30178248389453,
      "learning_rate": 0.00019825327510917032,
      "loss": 21.1708,
      "step": 681
    },
    {
      "epoch": 0.11925683060109289,
      "grad_norm": 1.3017649734644328,
      "learning_rate": 0.0001985443959243086,
      "loss": 21.4295,
      "step": 682
    },
    {
      "epoch": 0.11943169398907104,
      "grad_norm": 1.2492378744333437,
      "learning_rate": 0.00019883551673944688,
      "loss": 21.2674,
      "step": 683
    },
    {
      "epoch": 0.11960655737704919,
      "grad_norm": 1.2725928365933679,
      "learning_rate": 0.00019912663755458515,
      "loss": 21.0584,
      "step": 684
    },
    {
      "epoch": 0.11978142076502732,
      "grad_norm": 1.2323354185538045,
      "learning_rate": 0.00019941775836972346,
      "loss": 21.4929,
      "step": 685
    },
    {
      "epoch": 0.11995628415300547,
      "grad_norm": 1.257547650426803,
      "learning_rate": 0.0001997088791848617,
      "loss": 21.4419,
      "step": 686
    },
    {
      "epoch": 0.12013114754098361,
      "grad_norm": 1.2452913251125688,
      "learning_rate": 0.0002,
      "loss": 21.1449,
      "step": 687
    },
    {
      "epoch": 0.12030601092896175,
      "grad_norm": 1.2672489121909472,
      "learning_rate": 0.00019999999899734648,
      "loss": 21.3557,
      "step": 688
    },
    {
      "epoch": 0.12048087431693989,
      "grad_norm": 1.2069112362674919,
      "learning_rate": 0.00019999999598938596,
      "loss": 21.3366,
      "step": 689
    },
    {
      "epoch": 0.12065573770491804,
      "grad_norm": 1.24716738602866,
      "learning_rate": 0.00019999999097611846,
      "loss": 21.0427,
      "step": 690
    },
    {
      "epoch": 0.12083060109289617,
      "grad_norm": 1.1937316754983451,
      "learning_rate": 0.0001999999839575441,
      "loss": 20.9424,
      "step": 691
    },
    {
      "epoch": 0.12100546448087432,
      "grad_norm": 1.2831256119797594,
      "learning_rate": 0.000199999974933663,
      "loss": 21.0718,
      "step": 692
    },
    {
      "epoch": 0.12118032786885245,
      "grad_norm": 1.2554371600890715,
      "learning_rate": 0.00019999996390447538,
      "loss": 20.7998,
      "step": 693
    },
    {
      "epoch": 0.1213551912568306,
      "grad_norm": 1.3844780106760781,
      "learning_rate": 0.00019999995086998146,
      "loss": 21.2253,
      "step": 694
    },
    {
      "epoch": 0.12153005464480875,
      "grad_norm": 1.399658815262009,
      "learning_rate": 0.0001999999358301815,
      "loss": 21.2803,
      "step": 695
    },
    {
      "epoch": 0.12170491803278688,
      "grad_norm": 1.411613691692199,
      "learning_rate": 0.00019999991878507574,
      "loss": 20.763,
      "step": 696
    },
    {
      "epoch": 0.12187978142076503,
      "grad_norm": 1.1652369927043826,
      "learning_rate": 0.0001999998997346646,
      "loss": 21.2622,
      "step": 697
    },
    {
      "epoch": 0.12205464480874317,
      "grad_norm": 1.2751960673586091,
      "learning_rate": 0.00019999987867894842,
      "loss": 21.1909,
      "step": 698
    },
    {
      "epoch": 0.1222295081967213,
      "grad_norm": 1.3014339922654075,
      "learning_rate": 0.00019999985561792763,
      "loss": 21.2352,
      "step": 699
    },
    {
      "epoch": 0.12240437158469945,
      "grad_norm": 1.2526622798009772,
      "learning_rate": 0.00019999983055160267,
      "loss": 21.0226,
      "step": 700
    },
    {
      "epoch": 0.1225792349726776,
      "grad_norm": 1.2300750399462768,
      "learning_rate": 0.00019999980347997412,
      "loss": 21.4469,
      "step": 701
    },
    {
      "epoch": 0.12275409836065573,
      "grad_norm": 1.2175610083509778,
      "learning_rate": 0.00019999977440304242,
      "loss": 21.4393,
      "step": 702
    },
    {
      "epoch": 0.12292896174863388,
      "grad_norm": 1.2517918312789524,
      "learning_rate": 0.00019999974332080825,
      "loss": 21.4138,
      "step": 703
    },
    {
      "epoch": 0.12310382513661203,
      "grad_norm": 1.2409342147531581,
      "learning_rate": 0.00019999971023327214,
      "loss": 20.8849,
      "step": 704
    },
    {
      "epoch": 0.12327868852459016,
      "grad_norm": 1.124418330477856,
      "learning_rate": 0.00019999967514043482,
      "loss": 21.4207,
      "step": 705
    },
    {
      "epoch": 0.12345355191256831,
      "grad_norm": 1.3104749170223764,
      "learning_rate": 0.00019999963804229697,
      "loss": 21.0704,
      "step": 706
    },
    {
      "epoch": 0.12362841530054645,
      "grad_norm": 1.272129458174149,
      "learning_rate": 0.00019999959893885936,
      "loss": 21.2648,
      "step": 707
    },
    {
      "epoch": 0.12380327868852459,
      "grad_norm": 1.1535664351675998,
      "learning_rate": 0.00019999955783012275,
      "loss": 21.0479,
      "step": 708
    },
    {
      "epoch": 0.12397814207650273,
      "grad_norm": 1.27879183345176,
      "learning_rate": 0.00019999951471608794,
      "loss": 21.0698,
      "step": 709
    },
    {
      "epoch": 0.12415300546448088,
      "grad_norm": 1.272097058108315,
      "learning_rate": 0.00019999946959675584,
      "loss": 21.367,
      "step": 710
    },
    {
      "epoch": 0.12432786885245901,
      "grad_norm": 1.2930545482826359,
      "learning_rate": 0.00019999942247212732,
      "loss": 21.2177,
      "step": 711
    },
    {
      "epoch": 0.12450273224043716,
      "grad_norm": 1.1887856988947612,
      "learning_rate": 0.00019999937334220336,
      "loss": 21.2692,
      "step": 712
    },
    {
      "epoch": 0.1246775956284153,
      "grad_norm": 1.2997634521740702,
      "learning_rate": 0.0001999993222069849,
      "loss": 21.5065,
      "step": 713
    },
    {
      "epoch": 0.12485245901639344,
      "grad_norm": 1.2091154521108385,
      "learning_rate": 0.000199999269066473,
      "loss": 21.1229,
      "step": 714
    },
    {
      "epoch": 0.1250273224043716,
      "grad_norm": 1.3339806546044042,
      "learning_rate": 0.00019999921392066877,
      "loss": 21.0368,
      "step": 715
    },
    {
      "epoch": 0.12520218579234973,
      "grad_norm": 1.266595236471338,
      "learning_rate": 0.0001999991567695732,
      "loss": 20.7383,
      "step": 716
    },
    {
      "epoch": 0.12537704918032788,
      "grad_norm": 1.2118182336692997,
      "learning_rate": 0.00019999909761318755,
      "loss": 20.8513,
      "step": 717
    },
    {
      "epoch": 0.125551912568306,
      "grad_norm": 1.3442902818287474,
      "learning_rate": 0.0001999990364515129,
      "loss": 21.006,
      "step": 718
    },
    {
      "epoch": 0.12572677595628415,
      "grad_norm": 1.199342496895941,
      "learning_rate": 0.00019999897328455056,
      "loss": 21.1644,
      "step": 719
    },
    {
      "epoch": 0.1259016393442623,
      "grad_norm": 1.325366428045287,
      "learning_rate": 0.00019999890811230175,
      "loss": 21.3503,
      "step": 720
    },
    {
      "epoch": 0.12607650273224044,
      "grad_norm": 1.1833238975201374,
      "learning_rate": 0.00019999884093476782,
      "loss": 21.1509,
      "step": 721
    },
    {
      "epoch": 0.1262513661202186,
      "grad_norm": 1.3157833764641684,
      "learning_rate": 0.00019999877175195008,
      "loss": 21.0314,
      "step": 722
    },
    {
      "epoch": 0.1264262295081967,
      "grad_norm": 1.3196826509660144,
      "learning_rate": 0.00019999870056384994,
      "loss": 21.2302,
      "step": 723
    },
    {
      "epoch": 0.12660109289617485,
      "grad_norm": 1.2535367270191395,
      "learning_rate": 0.0001999986273704688,
      "loss": 20.7077,
      "step": 724
    },
    {
      "epoch": 0.126775956284153,
      "grad_norm": 1.1950425391110204,
      "learning_rate": 0.00019999855217180817,
      "loss": 21.0853,
      "step": 725
    },
    {
      "epoch": 0.12695081967213115,
      "grad_norm": 1.3629293841892345,
      "learning_rate": 0.0001999984749678695,
      "loss": 21.2403,
      "step": 726
    },
    {
      "epoch": 0.1271256830601093,
      "grad_norm": 1.3706178347354803,
      "learning_rate": 0.0001999983957586544,
      "loss": 20.8483,
      "step": 727
    },
    {
      "epoch": 0.12730054644808744,
      "grad_norm": 1.2273060680751238,
      "learning_rate": 0.00019999831454416443,
      "loss": 21.289,
      "step": 728
    },
    {
      "epoch": 0.12747540983606556,
      "grad_norm": 1.334264325672355,
      "learning_rate": 0.00019999823132440118,
      "loss": 21.0566,
      "step": 729
    },
    {
      "epoch": 0.1276502732240437,
      "grad_norm": 1.2452754269682471,
      "learning_rate": 0.0001999981460993664,
      "loss": 20.8925,
      "step": 730
    },
    {
      "epoch": 0.12782513661202186,
      "grad_norm": 1.3798399875704277,
      "learning_rate": 0.00019999805886906176,
      "loss": 20.7262,
      "step": 731
    },
    {
      "epoch": 0.128,
      "grad_norm": 1.1970883510240893,
      "learning_rate": 0.00019999796963348897,
      "loss": 21.1814,
      "step": 732
    },
    {
      "epoch": 0.12817486338797815,
      "grad_norm": 1.3572096987861348,
      "learning_rate": 0.00019999787839264987,
      "loss": 21.0124,
      "step": 733
    },
    {
      "epoch": 0.1283497267759563,
      "grad_norm": 1.278688772708804,
      "learning_rate": 0.0001999977851465463,
      "loss": 21.1609,
      "step": 734
    },
    {
      "epoch": 0.12852459016393442,
      "grad_norm": 1.3071874363059746,
      "learning_rate": 0.00019999768989518004,
      "loss": 21.0788,
      "step": 735
    },
    {
      "epoch": 0.12869945355191256,
      "grad_norm": 1.16793461832381,
      "learning_rate": 0.0001999975926385531,
      "loss": 21.1025,
      "step": 736
    },
    {
      "epoch": 0.1288743169398907,
      "grad_norm": 1.1975503669770076,
      "learning_rate": 0.00019999749337666742,
      "loss": 20.9532,
      "step": 737
    },
    {
      "epoch": 0.12904918032786886,
      "grad_norm": 1.286772249270567,
      "learning_rate": 0.00019999739210952493,
      "loss": 21.0763,
      "step": 738
    },
    {
      "epoch": 0.129224043715847,
      "grad_norm": 1.2621332972000072,
      "learning_rate": 0.0001999972888371277,
      "loss": 21.1802,
      "step": 739
    },
    {
      "epoch": 0.12939890710382515,
      "grad_norm": 1.1906021592021707,
      "learning_rate": 0.00019999718355947783,
      "loss": 20.8837,
      "step": 740
    },
    {
      "epoch": 0.12957377049180327,
      "grad_norm": 1.1927147964475513,
      "learning_rate": 0.00019999707627657736,
      "loss": 20.9095,
      "step": 741
    },
    {
      "epoch": 0.12974863387978142,
      "grad_norm": 1.2399978013809312,
      "learning_rate": 0.0001999969669884285,
      "loss": 20.7463,
      "step": 742
    },
    {
      "epoch": 0.12992349726775956,
      "grad_norm": 1.3024543136290205,
      "learning_rate": 0.0001999968556950334,
      "loss": 20.9965,
      "step": 743
    },
    {
      "epoch": 0.1300983606557377,
      "grad_norm": 1.2392361028777008,
      "learning_rate": 0.00019999674239639433,
      "loss": 20.9954,
      "step": 744
    },
    {
      "epoch": 0.13027322404371586,
      "grad_norm": 1.3233551355277269,
      "learning_rate": 0.00019999662709251355,
      "loss": 20.8485,
      "step": 745
    },
    {
      "epoch": 0.13044808743169398,
      "grad_norm": 1.2375066438224753,
      "learning_rate": 0.00019999650978339334,
      "loss": 20.4842,
      "step": 746
    },
    {
      "epoch": 0.13062295081967212,
      "grad_norm": 1.226014771377986,
      "learning_rate": 0.00019999639046903612,
      "loss": 21.064,
      "step": 747
    },
    {
      "epoch": 0.13079781420765027,
      "grad_norm": 1.1520941184623457,
      "learning_rate": 0.00019999626914944421,
      "loss": 20.9458,
      "step": 748
    },
    {
      "epoch": 0.13097267759562842,
      "grad_norm": 1.2731830249462042,
      "learning_rate": 0.0001999961458246201,
      "loss": 20.9778,
      "step": 749
    },
    {
      "epoch": 0.13114754098360656,
      "grad_norm": 1.2553221375563024,
      "learning_rate": 0.0001999960204945662,
      "loss": 21.0677,
      "step": 750
    },
    {
      "epoch": 0.1313224043715847,
      "grad_norm": 1.2271505783256444,
      "learning_rate": 0.0001999958931592851,
      "loss": 21.0791,
      "step": 751
    },
    {
      "epoch": 0.13149726775956283,
      "grad_norm": 1.1929767503137658,
      "learning_rate": 0.00019999576381877932,
      "loss": 21.1864,
      "step": 752
    },
    {
      "epoch": 0.13167213114754098,
      "grad_norm": 1.2775253638927375,
      "learning_rate": 0.00019999563247305143,
      "loss": 21.3264,
      "step": 753
    },
    {
      "epoch": 0.13184699453551912,
      "grad_norm": 1.388102945838071,
      "learning_rate": 0.00019999549912210404,
      "loss": 20.6109,
      "step": 754
    },
    {
      "epoch": 0.13202185792349727,
      "grad_norm": 1.2924986505413323,
      "learning_rate": 0.00019999536376593994,
      "loss": 20.6238,
      "step": 755
    },
    {
      "epoch": 0.13219672131147542,
      "grad_norm": 1.2752204486804317,
      "learning_rate": 0.0001999952264045617,
      "loss": 20.6596,
      "step": 756
    },
    {
      "epoch": 0.13237158469945356,
      "grad_norm": 1.162206731556745,
      "learning_rate": 0.0001999950870379722,
      "loss": 20.7815,
      "step": 757
    },
    {
      "epoch": 0.13254644808743168,
      "grad_norm": 1.3094594020991734,
      "learning_rate": 0.00019999494566617414,
      "loss": 20.8947,
      "step": 758
    },
    {
      "epoch": 0.13272131147540983,
      "grad_norm": 1.2979140771131883,
      "learning_rate": 0.0001999948022891704,
      "loss": 21.1339,
      "step": 759
    },
    {
      "epoch": 0.13289617486338798,
      "grad_norm": 1.1553302776600594,
      "learning_rate": 0.00019999465690696388,
      "loss": 20.71,
      "step": 760
    },
    {
      "epoch": 0.13307103825136612,
      "grad_norm": 1.2785546579403633,
      "learning_rate": 0.00019999450951955746,
      "loss": 20.9763,
      "step": 761
    },
    {
      "epoch": 0.13324590163934427,
      "grad_norm": 1.2484379450102037,
      "learning_rate": 0.00019999436012695408,
      "loss": 20.9039,
      "step": 762
    },
    {
      "epoch": 0.1334207650273224,
      "grad_norm": 1.1116335081971727,
      "learning_rate": 0.00019999420872915679,
      "loss": 20.8573,
      "step": 763
    },
    {
      "epoch": 0.13359562841530054,
      "grad_norm": 1.2930995745123741,
      "learning_rate": 0.00019999405532616856,
      "loss": 21.1692,
      "step": 764
    },
    {
      "epoch": 0.13377049180327868,
      "grad_norm": 1.2144268481389386,
      "learning_rate": 0.00019999389991799248,
      "loss": 21.061,
      "step": 765
    },
    {
      "epoch": 0.13394535519125683,
      "grad_norm": 1.1120871160444203,
      "learning_rate": 0.0001999937425046317,
      "loss": 20.979,
      "step": 766
    },
    {
      "epoch": 0.13412021857923498,
      "grad_norm": 1.250866094942296,
      "learning_rate": 0.00019999358308608938,
      "loss": 20.6954,
      "step": 767
    },
    {
      "epoch": 0.13429508196721313,
      "grad_norm": 1.1816679807039885,
      "learning_rate": 0.00019999342166236868,
      "loss": 21.0526,
      "step": 768
    },
    {
      "epoch": 0.13446994535519124,
      "grad_norm": 1.2049292654629589,
      "learning_rate": 0.0001999932582334729,
      "loss": 20.8368,
      "step": 769
    },
    {
      "epoch": 0.1346448087431694,
      "grad_norm": 1.182417124458382,
      "learning_rate": 0.00019999309279940524,
      "loss": 20.6583,
      "step": 770
    },
    {
      "epoch": 0.13481967213114754,
      "grad_norm": 1.354775411945175,
      "learning_rate": 0.00019999292536016902,
      "loss": 20.9707,
      "step": 771
    },
    {
      "epoch": 0.13499453551912569,
      "grad_norm": 1.1062845890155526,
      "learning_rate": 0.00019999275591576766,
      "loss": 20.8864,
      "step": 772
    },
    {
      "epoch": 0.13516939890710383,
      "grad_norm": 1.2100625497746211,
      "learning_rate": 0.00019999258446620454,
      "loss": 20.9669,
      "step": 773
    },
    {
      "epoch": 0.13534426229508198,
      "grad_norm": 1.2279503817800446,
      "learning_rate": 0.00019999241101148306,
      "loss": 20.6101,
      "step": 774
    },
    {
      "epoch": 0.1355191256830601,
      "grad_norm": 1.172857746046995,
      "learning_rate": 0.00019999223555160672,
      "loss": 20.2331,
      "step": 775
    },
    {
      "epoch": 0.13569398907103825,
      "grad_norm": 1.171666529707958,
      "learning_rate": 0.00019999205808657905,
      "loss": 20.5655,
      "step": 776
    },
    {
      "epoch": 0.1358688524590164,
      "grad_norm": 1.1550711976494052,
      "learning_rate": 0.00019999187861640362,
      "loss": 20.3847,
      "step": 777
    },
    {
      "epoch": 0.13604371584699454,
      "grad_norm": 1.1514960368977676,
      "learning_rate": 0.00019999169714108399,
      "loss": 20.5533,
      "step": 778
    },
    {
      "epoch": 0.1362185792349727,
      "grad_norm": 1.1309742256262072,
      "learning_rate": 0.0001999915136606238,
      "loss": 20.714,
      "step": 779
    },
    {
      "epoch": 0.13639344262295083,
      "grad_norm": 1.1260902161683668,
      "learning_rate": 0.00019999132817502676,
      "loss": 20.8176,
      "step": 780
    },
    {
      "epoch": 0.13656830601092895,
      "grad_norm": 1.1496690304251236,
      "learning_rate": 0.0001999911406842966,
      "loss": 20.4335,
      "step": 781
    },
    {
      "epoch": 0.1367431693989071,
      "grad_norm": 1.2126983843865309,
      "learning_rate": 0.00019999095118843703,
      "loss": 20.622,
      "step": 782
    },
    {
      "epoch": 0.13691803278688525,
      "grad_norm": 1.185237225952874,
      "learning_rate": 0.00019999075968745188,
      "loss": 20.8474,
      "step": 783
    },
    {
      "epoch": 0.1370928961748634,
      "grad_norm": 1.1960629010519879,
      "learning_rate": 0.00019999056618134499,
      "loss": 20.7931,
      "step": 784
    },
    {
      "epoch": 0.13726775956284154,
      "grad_norm": 1.2429424709454742,
      "learning_rate": 0.00019999037067012024,
      "loss": 20.4781,
      "step": 785
    },
    {
      "epoch": 0.13744262295081966,
      "grad_norm": 1.1926958197655626,
      "learning_rate": 0.00019999017315378153,
      "loss": 20.9557,
      "step": 786
    },
    {
      "epoch": 0.1376174863387978,
      "grad_norm": 1.2482107711444779,
      "learning_rate": 0.00019998997363233285,
      "loss": 20.8116,
      "step": 787
    },
    {
      "epoch": 0.13779234972677595,
      "grad_norm": 1.2281269355762665,
      "learning_rate": 0.00019998977210577818,
      "loss": 21.0576,
      "step": 788
    },
    {
      "epoch": 0.1379672131147541,
      "grad_norm": 1.2620594010542976,
      "learning_rate": 0.0001999895685741216,
      "loss": 20.9445,
      "step": 789
    },
    {
      "epoch": 0.13814207650273225,
      "grad_norm": 1.341058783756389,
      "learning_rate": 0.00019998936303736712,
      "loss": 20.8814,
      "step": 790
    },
    {
      "epoch": 0.1383169398907104,
      "grad_norm": 1.1615920738571979,
      "learning_rate": 0.00019998915549551894,
      "loss": 20.7068,
      "step": 791
    },
    {
      "epoch": 0.1384918032786885,
      "grad_norm": 1.0974161261287017,
      "learning_rate": 0.00019998894594858117,
      "loss": 20.9505,
      "step": 792
    },
    {
      "epoch": 0.13866666666666666,
      "grad_norm": 1.198302420477859,
      "learning_rate": 0.00019998873439655804,
      "loss": 20.8043,
      "step": 793
    },
    {
      "epoch": 0.1388415300546448,
      "grad_norm": 1.1770521879729234,
      "learning_rate": 0.00019998852083945374,
      "loss": 20.6752,
      "step": 794
    },
    {
      "epoch": 0.13901639344262295,
      "grad_norm": 1.1974791214583917,
      "learning_rate": 0.00019998830527727263,
      "loss": 20.8775,
      "step": 795
    },
    {
      "epoch": 0.1391912568306011,
      "grad_norm": 1.2198646818220114,
      "learning_rate": 0.00019998808771001897,
      "loss": 20.7361,
      "step": 796
    },
    {
      "epoch": 0.13936612021857925,
      "grad_norm": 1.310174603124577,
      "learning_rate": 0.00019998786813769717,
      "loss": 20.3399,
      "step": 797
    },
    {
      "epoch": 0.13954098360655737,
      "grad_norm": 1.247085434400403,
      "learning_rate": 0.0001999876465603116,
      "loss": 20.6578,
      "step": 798
    },
    {
      "epoch": 0.13971584699453551,
      "grad_norm": 1.1866304694072158,
      "learning_rate": 0.0001999874229778667,
      "loss": 21.0758,
      "step": 799
    },
    {
      "epoch": 0.13989071038251366,
      "grad_norm": 1.3038864146476685,
      "learning_rate": 0.00019998719739036699,
      "loss": 20.8201,
      "step": 800
    },
    {
      "epoch": 0.1400655737704918,
      "grad_norm": 1.4379994362558501,
      "learning_rate": 0.00019998696979781698,
      "loss": 20.5928,
      "step": 801
    },
    {
      "epoch": 0.14024043715846995,
      "grad_norm": 1.277816937368116,
      "learning_rate": 0.00019998674020022122,
      "loss": 20.5641,
      "step": 802
    },
    {
      "epoch": 0.14041530054644807,
      "grad_norm": 1.2089200415494932,
      "learning_rate": 0.0001999865085975843,
      "loss": 20.6527,
      "step": 803
    },
    {
      "epoch": 0.14059016393442622,
      "grad_norm": 1.2085945433263448,
      "learning_rate": 0.0001999862749899109,
      "loss": 20.9075,
      "step": 804
    },
    {
      "epoch": 0.14076502732240437,
      "grad_norm": 1.1854156290155573,
      "learning_rate": 0.00019998603937720569,
      "loss": 20.6369,
      "step": 805
    },
    {
      "epoch": 0.14093989071038251,
      "grad_norm": 1.3361321894254456,
      "learning_rate": 0.00019998580175947336,
      "loss": 20.6066,
      "step": 806
    },
    {
      "epoch": 0.14111475409836066,
      "grad_norm": 1.2239647820297301,
      "learning_rate": 0.00019998556213671873,
      "loss": 20.795,
      "step": 807
    },
    {
      "epoch": 0.1412896174863388,
      "grad_norm": 1.1886043268288946,
      "learning_rate": 0.00019998532050894658,
      "loss": 20.6364,
      "step": 808
    },
    {
      "epoch": 0.14146448087431693,
      "grad_norm": 1.22798284790972,
      "learning_rate": 0.00019998507687616179,
      "loss": 20.9643,
      "step": 809
    },
    {
      "epoch": 0.14163934426229507,
      "grad_norm": 1.1714578478696838,
      "learning_rate": 0.00019998483123836918,
      "loss": 20.6376,
      "step": 810
    },
    {
      "epoch": 0.14181420765027322,
      "grad_norm": 1.1208243837783844,
      "learning_rate": 0.00019998458359557372,
      "loss": 20.4645,
      "step": 811
    },
    {
      "epoch": 0.14198907103825137,
      "grad_norm": 1.2399542760608138,
      "learning_rate": 0.00019998433394778035,
      "loss": 20.6629,
      "step": 812
    },
    {
      "epoch": 0.14216393442622952,
      "grad_norm": 1.1870445592810857,
      "learning_rate": 0.0001999840822949941,
      "loss": 20.8663,
      "step": 813
    },
    {
      "epoch": 0.14233879781420766,
      "grad_norm": 1.2206975013393524,
      "learning_rate": 0.00019998382863722002,
      "loss": 20.2193,
      "step": 814
    },
    {
      "epoch": 0.14251366120218578,
      "grad_norm": 1.1468230655645648,
      "learning_rate": 0.0001999835729744632,
      "loss": 20.2957,
      "step": 815
    },
    {
      "epoch": 0.14268852459016393,
      "grad_norm": 1.168142523298871,
      "learning_rate": 0.0001999833153067287,
      "loss": 20.4482,
      "step": 816
    },
    {
      "epoch": 0.14286338797814208,
      "grad_norm": 1.1981626267610659,
      "learning_rate": 0.0001999830556340218,
      "loss": 20.4856,
      "step": 817
    },
    {
      "epoch": 0.14303825136612022,
      "grad_norm": 1.2301279571986727,
      "learning_rate": 0.0001999827939563476,
      "loss": 20.5634,
      "step": 818
    },
    {
      "epoch": 0.14321311475409837,
      "grad_norm": 1.303975982982482,
      "learning_rate": 0.0001999825302737114,
      "loss": 20.2195,
      "step": 819
    },
    {
      "epoch": 0.14338797814207652,
      "grad_norm": 1.156022240141936,
      "learning_rate": 0.0001999822645861185,
      "loss": 21.0107,
      "step": 820
    },
    {
      "epoch": 0.14356284153005464,
      "grad_norm": 1.1406101023843813,
      "learning_rate": 0.0001999819968935742,
      "loss": 20.2041,
      "step": 821
    },
    {
      "epoch": 0.14373770491803278,
      "grad_norm": 1.1442670419503067,
      "learning_rate": 0.0001999817271960839,
      "loss": 20.6615,
      "step": 822
    },
    {
      "epoch": 0.14391256830601093,
      "grad_norm": 1.0906664627953904,
      "learning_rate": 0.00019998145549365295,
      "loss": 20.443,
      "step": 823
    },
    {
      "epoch": 0.14408743169398908,
      "grad_norm": 1.083046976750993,
      "learning_rate": 0.00019998118178628685,
      "loss": 21.0208,
      "step": 824
    },
    {
      "epoch": 0.14426229508196722,
      "grad_norm": 1.1847320763262443,
      "learning_rate": 0.00019998090607399106,
      "loss": 20.1614,
      "step": 825
    },
    {
      "epoch": 0.14443715846994534,
      "grad_norm": 1.2131755354524114,
      "learning_rate": 0.00019998062835677115,
      "loss": 20.5491,
      "step": 826
    },
    {
      "epoch": 0.1446120218579235,
      "grad_norm": 1.2082373410352019,
      "learning_rate": 0.00019998034863463263,
      "loss": 20.0382,
      "step": 827
    },
    {
      "epoch": 0.14478688524590164,
      "grad_norm": 1.1684350381563307,
      "learning_rate": 0.00019998006690758117,
      "loss": 20.3413,
      "step": 828
    },
    {
      "epoch": 0.14496174863387978,
      "grad_norm": 1.1811786284608214,
      "learning_rate": 0.00019997978317562237,
      "loss": 20.3672,
      "step": 829
    },
    {
      "epoch": 0.14513661202185793,
      "grad_norm": 1.0987762338877582,
      "learning_rate": 0.00019997949743876193,
      "loss": 20.9086,
      "step": 830
    },
    {
      "epoch": 0.14531147540983608,
      "grad_norm": 1.0973025052524588,
      "learning_rate": 0.0001999792096970056,
      "loss": 20.4612,
      "step": 831
    },
    {
      "epoch": 0.1454863387978142,
      "grad_norm": 1.1724777801131605,
      "learning_rate": 0.00019997891995035912,
      "loss": 20.7044,
      "step": 832
    },
    {
      "epoch": 0.14566120218579234,
      "grad_norm": 1.2081669520192053,
      "learning_rate": 0.00019997862819882837,
      "loss": 20.8273,
      "step": 833
    },
    {
      "epoch": 0.1458360655737705,
      "grad_norm": 1.1717499245912764,
      "learning_rate": 0.00019997833444241911,
      "loss": 20.4573,
      "step": 834
    },
    {
      "epoch": 0.14601092896174864,
      "grad_norm": 1.2418781461130703,
      "learning_rate": 0.00019997803868113727,
      "loss": 20.7081,
      "step": 835
    },
    {
      "epoch": 0.14618579234972678,
      "grad_norm": 1.1474266422290047,
      "learning_rate": 0.00019997774091498879,
      "loss": 20.6359,
      "step": 836
    },
    {
      "epoch": 0.14636065573770493,
      "grad_norm": 1.1731753808033187,
      "learning_rate": 0.00019997744114397964,
      "loss": 20.4207,
      "step": 837
    },
    {
      "epoch": 0.14653551912568305,
      "grad_norm": 1.2711685878143018,
      "learning_rate": 0.00019997713936811582,
      "loss": 20.3896,
      "step": 838
    },
    {
      "epoch": 0.1467103825136612,
      "grad_norm": 1.3735233331202292,
      "learning_rate": 0.0001999768355874034,
      "loss": 20.4517,
      "step": 839
    },
    {
      "epoch": 0.14688524590163934,
      "grad_norm": 1.3153800160954672,
      "learning_rate": 0.00019997652980184843,
      "loss": 20.6396,
      "step": 840
    },
    {
      "epoch": 0.1470601092896175,
      "grad_norm": 1.1227244676497143,
      "learning_rate": 0.00019997622201145708,
      "loss": 20.6592,
      "step": 841
    },
    {
      "epoch": 0.14723497267759564,
      "grad_norm": 1.1856963224648707,
      "learning_rate": 0.00019997591221623551,
      "loss": 20.5246,
      "step": 842
    },
    {
      "epoch": 0.14740983606557376,
      "grad_norm": 1.3039968858021926,
      "learning_rate": 0.00019997560041618993,
      "loss": 20.6975,
      "step": 843
    },
    {
      "epoch": 0.1475846994535519,
      "grad_norm": 1.2117590726403504,
      "learning_rate": 0.0001999752866113266,
      "loss": 20.5504,
      "step": 844
    },
    {
      "epoch": 0.14775956284153005,
      "grad_norm": 1.1101439385413892,
      "learning_rate": 0.0001999749708016518,
      "loss": 20.5789,
      "step": 845
    },
    {
      "epoch": 0.1479344262295082,
      "grad_norm": 1.1547324305300988,
      "learning_rate": 0.00019997465298717188,
      "loss": 20.5973,
      "step": 846
    },
    {
      "epoch": 0.14810928961748634,
      "grad_norm": 1.1621972063868249,
      "learning_rate": 0.0001999743331678932,
      "loss": 20.365,
      "step": 847
    },
    {
      "epoch": 0.1482841530054645,
      "grad_norm": 1.1243650648827996,
      "learning_rate": 0.0001999740113438222,
      "loss": 20.429,
      "step": 848
    },
    {
      "epoch": 0.1484590163934426,
      "grad_norm": 1.1838262971207,
      "learning_rate": 0.00019997368751496528,
      "loss": 20.5273,
      "step": 849
    },
    {
      "epoch": 0.14863387978142076,
      "grad_norm": 1.219793892525656,
      "learning_rate": 0.00019997336168132897,
      "loss": 20.3593,
      "step": 850
    },
    {
      "epoch": 0.1488087431693989,
      "grad_norm": 1.096136458667286,
      "learning_rate": 0.00019997303384291986,
      "loss": 20.494,
      "step": 851
    },
    {
      "epoch": 0.14898360655737705,
      "grad_norm": 1.1072847194953714,
      "learning_rate": 0.00019997270399974438,
      "loss": 20.5106,
      "step": 852
    },
    {
      "epoch": 0.1491584699453552,
      "grad_norm": 1.1568785945086908,
      "learning_rate": 0.0001999723721518093,
      "loss": 20.248,
      "step": 853
    },
    {
      "epoch": 0.14933333333333335,
      "grad_norm": 1.1327890858031724,
      "learning_rate": 0.00019997203829912117,
      "loss": 20.906,
      "step": 854
    },
    {
      "epoch": 0.14950819672131146,
      "grad_norm": 1.101866490667469,
      "learning_rate": 0.00019997170244168667,
      "loss": 20.7887,
      "step": 855
    },
    {
      "epoch": 0.1496830601092896,
      "grad_norm": 1.16220466845584,
      "learning_rate": 0.00019997136457951267,
      "loss": 20.517,
      "step": 856
    },
    {
      "epoch": 0.14985792349726776,
      "grad_norm": 1.1405378580851482,
      "learning_rate": 0.0001999710247126058,
      "loss": 20.507,
      "step": 857
    },
    {
      "epoch": 0.1500327868852459,
      "grad_norm": 1.128100515737731,
      "learning_rate": 0.00019997068284097295,
      "loss": 20.6307,
      "step": 858
    },
    {
      "epoch": 0.15020765027322405,
      "grad_norm": 1.133218336996158,
      "learning_rate": 0.00019997033896462094,
      "loss": 20.4824,
      "step": 859
    },
    {
      "epoch": 0.1503825136612022,
      "grad_norm": 1.1506208286838726,
      "learning_rate": 0.0001999699930835567,
      "loss": 20.6349,
      "step": 860
    },
    {
      "epoch": 0.15055737704918032,
      "grad_norm": 1.1569486467907346,
      "learning_rate": 0.0001999696451977872,
      "loss": 20.7112,
      "step": 861
    },
    {
      "epoch": 0.15073224043715847,
      "grad_norm": 1.1354629687557583,
      "learning_rate": 0.0001999692953073193,
      "loss": 20.3681,
      "step": 862
    },
    {
      "epoch": 0.1509071038251366,
      "grad_norm": 1.2169267494619502,
      "learning_rate": 0.0001999689434121601,
      "loss": 20.3204,
      "step": 863
    },
    {
      "epoch": 0.15108196721311476,
      "grad_norm": 1.1420943104516428,
      "learning_rate": 0.00019996858951231664,
      "loss": 20.6892,
      "step": 864
    },
    {
      "epoch": 0.1512568306010929,
      "grad_norm": 1.1586845569599944,
      "learning_rate": 0.00019996823360779602,
      "loss": 20.2772,
      "step": 865
    },
    {
      "epoch": 0.15143169398907103,
      "grad_norm": 1.1008087442434542,
      "learning_rate": 0.00019996787569860538,
      "loss": 20.6909,
      "step": 866
    },
    {
      "epoch": 0.15160655737704917,
      "grad_norm": 1.2039186646046627,
      "learning_rate": 0.00019996751578475186,
      "loss": 20.1026,
      "step": 867
    },
    {
      "epoch": 0.15178142076502732,
      "grad_norm": 1.2443750809392427,
      "learning_rate": 0.00019996715386624275,
      "loss": 20.6535,
      "step": 868
    },
    {
      "epoch": 0.15195628415300547,
      "grad_norm": 1.2230704114877686,
      "learning_rate": 0.00019996678994308522,
      "loss": 20.7272,
      "step": 869
    },
    {
      "epoch": 0.1521311475409836,
      "grad_norm": 1.3763934366255393,
      "learning_rate": 0.00019996642401528666,
      "loss": 20.328,
      "step": 870
    },
    {
      "epoch": 0.15230601092896176,
      "grad_norm": 1.3537850566329461,
      "learning_rate": 0.00019996605608285433,
      "loss": 20.3743,
      "step": 871
    },
    {
      "epoch": 0.15248087431693988,
      "grad_norm": 1.2117339685511677,
      "learning_rate": 0.00019996568614579564,
      "loss": 19.9703,
      "step": 872
    },
    {
      "epoch": 0.15265573770491803,
      "grad_norm": 1.3467822451770526,
      "learning_rate": 0.00019996531420411802,
      "loss": 20.6395,
      "step": 873
    },
    {
      "epoch": 0.15283060109289617,
      "grad_norm": 1.2584267129228768,
      "learning_rate": 0.00019996494025782894,
      "loss": 20.1374,
      "step": 874
    },
    {
      "epoch": 0.15300546448087432,
      "grad_norm": 1.25679283903781,
      "learning_rate": 0.00019996456430693585,
      "loss": 20.1532,
      "step": 875
    },
    {
      "epoch": 0.15318032786885247,
      "grad_norm": 1.22505927775054,
      "learning_rate": 0.0001999641863514463,
      "loss": 19.9953,
      "step": 876
    },
    {
      "epoch": 0.15335519125683061,
      "grad_norm": 1.3520813158663458,
      "learning_rate": 0.0001999638063913679,
      "loss": 20.5257,
      "step": 877
    },
    {
      "epoch": 0.15353005464480873,
      "grad_norm": 1.1923310646715068,
      "learning_rate": 0.00019996342442670827,
      "loss": 20.8686,
      "step": 878
    },
    {
      "epoch": 0.15370491803278688,
      "grad_norm": 1.2524129085602949,
      "learning_rate": 0.000199963040457475,
      "loss": 20.486,
      "step": 879
    },
    {
      "epoch": 0.15387978142076503,
      "grad_norm": 1.1063771527174753,
      "learning_rate": 0.00019996265448367587,
      "loss": 20.2987,
      "step": 880
    },
    {
      "epoch": 0.15405464480874317,
      "grad_norm": 1.1692655159696501,
      "learning_rate": 0.0001999622665053186,
      "loss": 20.2652,
      "step": 881
    },
    {
      "epoch": 0.15422950819672132,
      "grad_norm": 1.2189137844266904,
      "learning_rate": 0.00019996187652241096,
      "loss": 20.4024,
      "step": 882
    },
    {
      "epoch": 0.15440437158469944,
      "grad_norm": 1.1959741071879255,
      "learning_rate": 0.00019996148453496076,
      "loss": 20.3893,
      "step": 883
    },
    {
      "epoch": 0.1545792349726776,
      "grad_norm": 1.1613339510062641,
      "learning_rate": 0.00019996109054297589,
      "loss": 20.6367,
      "step": 884
    },
    {
      "epoch": 0.15475409836065573,
      "grad_norm": 1.0757410227962092,
      "learning_rate": 0.0001999606945464642,
      "loss": 20.1419,
      "step": 885
    },
    {
      "epoch": 0.15492896174863388,
      "grad_norm": 1.2730274081910118,
      "learning_rate": 0.00019996029654543375,
      "loss": 20.2736,
      "step": 886
    },
    {
      "epoch": 0.15510382513661203,
      "grad_norm": 1.140764034113491,
      "learning_rate": 0.00019995989653989233,
      "loss": 20.1782,
      "step": 887
    },
    {
      "epoch": 0.15527868852459017,
      "grad_norm": 1.6265875468663638,
      "learning_rate": 0.00019995949452984813,
      "loss": 20.4056,
      "step": 888
    },
    {
      "epoch": 0.1554535519125683,
      "grad_norm": 1.219388424091354,
      "learning_rate": 0.00019995909051530913,
      "loss": 20.3878,
      "step": 889
    },
    {
      "epoch": 0.15562841530054644,
      "grad_norm": 4.115344835357046,
      "learning_rate": 0.00019995868449628346,
      "loss": 20.2097,
      "step": 890
    },
    {
      "epoch": 0.1558032786885246,
      "grad_norm": 3.098059593013793,
      "learning_rate": 0.00019995827647277924,
      "loss": 20.8368,
      "step": 891
    },
    {
      "epoch": 0.15597814207650273,
      "grad_norm": 1.5012788335483116,
      "learning_rate": 0.00019995786644480467,
      "loss": 20.1027,
      "step": 892
    },
    {
      "epoch": 0.15615300546448088,
      "grad_norm": 1.6492358676144767,
      "learning_rate": 0.00019995745441236796,
      "loss": 20.4549,
      "step": 893
    },
    {
      "epoch": 0.15632786885245903,
      "grad_norm": 1.3432006043244378,
      "learning_rate": 0.0001999570403754774,
      "loss": 20.4448,
      "step": 894
    },
    {
      "epoch": 0.15650273224043715,
      "grad_norm": 1.2781352321903294,
      "learning_rate": 0.00019995662433414127,
      "loss": 20.4475,
      "step": 895
    },
    {
      "epoch": 0.1566775956284153,
      "grad_norm": 1.4381919639084557,
      "learning_rate": 0.0001999562062883679,
      "loss": 20.3973,
      "step": 896
    },
    {
      "epoch": 0.15685245901639344,
      "grad_norm": 1.3177925262231946,
      "learning_rate": 0.00019995578623816567,
      "loss": 20.489,
      "step": 897
    },
    {
      "epoch": 0.1570273224043716,
      "grad_norm": 1.1592456703121692,
      "learning_rate": 0.00019995536418354303,
      "loss": 20.4219,
      "step": 898
    },
    {
      "epoch": 0.15720218579234974,
      "grad_norm": 1.2414028927132759,
      "learning_rate": 0.00019995494012450848,
      "loss": 20.5641,
      "step": 899
    },
    {
      "epoch": 0.15737704918032788,
      "grad_norm": 1.241158915134143,
      "learning_rate": 0.00019995451406107045,
      "loss": 20.0185,
      "step": 900
    },
    {
      "epoch": 0.157551912568306,
      "grad_norm": 1.2307859921600048,
      "learning_rate": 0.00019995408599323748,
      "loss": 20.3554,
      "step": 901
    },
    {
      "epoch": 0.15772677595628415,
      "grad_norm": 1.226747286155619,
      "learning_rate": 0.0001999536559210182,
      "loss": 20.455,
      "step": 902
    },
    {
      "epoch": 0.1579016393442623,
      "grad_norm": 1.220364620360482,
      "learning_rate": 0.00019995322384442127,
      "loss": 20.7222,
      "step": 903
    },
    {
      "epoch": 0.15807650273224044,
      "grad_norm": 1.1589881189821716,
      "learning_rate": 0.00019995278976345525,
      "loss": 20.5553,
      "step": 904
    },
    {
      "epoch": 0.1582513661202186,
      "grad_norm": 1.1801383161691918,
      "learning_rate": 0.00019995235367812892,
      "loss": 20.6794,
      "step": 905
    },
    {
      "epoch": 0.1584262295081967,
      "grad_norm": 1.1443536173009263,
      "learning_rate": 0.000199951915588451,
      "loss": 20.447,
      "step": 906
    },
    {
      "epoch": 0.15860109289617486,
      "grad_norm": 1.163177888513321,
      "learning_rate": 0.0001999514754944303,
      "loss": 20.1682,
      "step": 907
    },
    {
      "epoch": 0.158775956284153,
      "grad_norm": 1.1008612861083096,
      "learning_rate": 0.00019995103339607562,
      "loss": 20.127,
      "step": 908
    },
    {
      "epoch": 0.15895081967213115,
      "grad_norm": 1.1478223938766616,
      "learning_rate": 0.0001999505892933958,
      "loss": 20.3866,
      "step": 909
    },
    {
      "epoch": 0.1591256830601093,
      "grad_norm": 1.153191839511669,
      "learning_rate": 0.0001999501431863998,
      "loss": 20.2292,
      "step": 910
    },
    {
      "epoch": 0.15930054644808744,
      "grad_norm": 1.207724722890054,
      "learning_rate": 0.00019994969507509657,
      "loss": 20.5437,
      "step": 911
    },
    {
      "epoch": 0.15947540983606556,
      "grad_norm": 1.2198341751902386,
      "learning_rate": 0.00019994924495949504,
      "loss": 20.1894,
      "step": 912
    },
    {
      "epoch": 0.1596502732240437,
      "grad_norm": 1.0960666333462543,
      "learning_rate": 0.00019994879283960425,
      "loss": 20.1885,
      "step": 913
    },
    {
      "epoch": 0.15982513661202186,
      "grad_norm": 1.1125798400401608,
      "learning_rate": 0.00019994833871543328,
      "loss": 20.2079,
      "step": 914
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0754537980256291,
      "learning_rate": 0.00019994788258699125,
      "loss": 19.9287,
      "step": 915
    },
    {
      "epoch": 0.16017486338797815,
      "grad_norm": 1.1776035907529772,
      "learning_rate": 0.0001999474244542873,
      "loss": 20.4767,
      "step": 916
    },
    {
      "epoch": 0.1603497267759563,
      "grad_norm": 1.1767066845436929,
      "learning_rate": 0.0001999469643173306,
      "loss": 20.3868,
      "step": 917
    },
    {
      "epoch": 0.16052459016393442,
      "grad_norm": 1.1041286209200132,
      "learning_rate": 0.00019994650217613038,
      "loss": 20.0613,
      "step": 918
    },
    {
      "epoch": 0.16069945355191256,
      "grad_norm": 1.1643201186043257,
      "learning_rate": 0.00019994603803069594,
      "loss": 20.2459,
      "step": 919
    },
    {
      "epoch": 0.1608743169398907,
      "grad_norm": 1.180307022706722,
      "learning_rate": 0.00019994557188103654,
      "loss": 20.294,
      "step": 920
    },
    {
      "epoch": 0.16104918032786886,
      "grad_norm": 1.182689609666058,
      "learning_rate": 0.00019994510372716158,
      "loss": 20.2165,
      "step": 921
    },
    {
      "epoch": 0.161224043715847,
      "grad_norm": 1.2421841102969842,
      "learning_rate": 0.0001999446335690804,
      "loss": 20.4606,
      "step": 922
    },
    {
      "epoch": 0.16139890710382512,
      "grad_norm": 1.185672595337847,
      "learning_rate": 0.00019994416140680242,
      "loss": 19.9548,
      "step": 923
    },
    {
      "epoch": 0.16157377049180327,
      "grad_norm": 1.2649378441355699,
      "learning_rate": 0.00019994368724033715,
      "loss": 20.0461,
      "step": 924
    },
    {
      "epoch": 0.16174863387978142,
      "grad_norm": 1.1912336304817628,
      "learning_rate": 0.00019994321106969412,
      "loss": 20.4432,
      "step": 925
    },
    {
      "epoch": 0.16192349726775956,
      "grad_norm": 1.1075694442955983,
      "learning_rate": 0.00019994273289488282,
      "loss": 20.2407,
      "step": 926
    },
    {
      "epoch": 0.1620983606557377,
      "grad_norm": 1.1799781656323498,
      "learning_rate": 0.00019994225271591286,
      "loss": 19.8587,
      "step": 927
    },
    {
      "epoch": 0.16227322404371586,
      "grad_norm": 1.179028574081262,
      "learning_rate": 0.00019994177053279385,
      "loss": 20.0197,
      "step": 928
    },
    {
      "epoch": 0.16244808743169398,
      "grad_norm": 1.1084277139693868,
      "learning_rate": 0.00019994128634553547,
      "loss": 20.356,
      "step": 929
    },
    {
      "epoch": 0.16262295081967212,
      "grad_norm": 1.1084323280323154,
      "learning_rate": 0.0001999408001541475,
      "loss": 20.3689,
      "step": 930
    },
    {
      "epoch": 0.16279781420765027,
      "grad_norm": 1.1062709953709768,
      "learning_rate": 0.00019994031195863957,
      "loss": 19.94,
      "step": 931
    },
    {
      "epoch": 0.16297267759562842,
      "grad_norm": 1.2217549822655656,
      "learning_rate": 0.00019993982175902156,
      "loss": 19.9671,
      "step": 932
    },
    {
      "epoch": 0.16314754098360656,
      "grad_norm": 1.1974087689909596,
      "learning_rate": 0.00019993932955530326,
      "loss": 20.4846,
      "step": 933
    },
    {
      "epoch": 0.1633224043715847,
      "grad_norm": 1.096794790708185,
      "learning_rate": 0.00019993883534749455,
      "loss": 20.5132,
      "step": 934
    },
    {
      "epoch": 0.16349726775956283,
      "grad_norm": 1.2539160260689184,
      "learning_rate": 0.00019993833913560532,
      "loss": 20.1072,
      "step": 935
    },
    {
      "epoch": 0.16367213114754098,
      "grad_norm": 1.2063934075268554,
      "learning_rate": 0.00019993784091964557,
      "loss": 20.0174,
      "step": 936
    },
    {
      "epoch": 0.16384699453551912,
      "grad_norm": 1.142556027133535,
      "learning_rate": 0.00019993734069962524,
      "loss": 20.2933,
      "step": 937
    },
    {
      "epoch": 0.16402185792349727,
      "grad_norm": 1.1866043251079206,
      "learning_rate": 0.0001999368384755544,
      "loss": 20.3275,
      "step": 938
    },
    {
      "epoch": 0.16419672131147542,
      "grad_norm": 1.289759666320743,
      "learning_rate": 0.00019993633424744307,
      "loss": 20.2263,
      "step": 939
    },
    {
      "epoch": 0.16437158469945357,
      "grad_norm": 1.291577234661986,
      "learning_rate": 0.00019993582801530141,
      "loss": 20.2732,
      "step": 940
    },
    {
      "epoch": 0.16454644808743168,
      "grad_norm": 1.0785138854397796,
      "learning_rate": 0.00019993531977913954,
      "loss": 20.2085,
      "step": 941
    },
    {
      "epoch": 0.16472131147540983,
      "grad_norm": 1.172998887826983,
      "learning_rate": 0.0001999348095389677,
      "loss": 20.6441,
      "step": 942
    },
    {
      "epoch": 0.16489617486338798,
      "grad_norm": 1.1275195216467675,
      "learning_rate": 0.0001999342972947961,
      "loss": 20.2569,
      "step": 943
    },
    {
      "epoch": 0.16507103825136613,
      "grad_norm": 1.2304500003525545,
      "learning_rate": 0.00019993378304663495,
      "loss": 20.1409,
      "step": 944
    },
    {
      "epoch": 0.16524590163934427,
      "grad_norm": 1.1460719048702748,
      "learning_rate": 0.00019993326679449466,
      "loss": 19.9369,
      "step": 945
    },
    {
      "epoch": 0.1654207650273224,
      "grad_norm": 1.2183447527462807,
      "learning_rate": 0.00019993274853838551,
      "loss": 20.0512,
      "step": 946
    },
    {
      "epoch": 0.16559562841530054,
      "grad_norm": 1.1726588660603268,
      "learning_rate": 0.0001999322282783179,
      "loss": 20.4235,
      "step": 947
    },
    {
      "epoch": 0.16577049180327869,
      "grad_norm": 1.186104442402955,
      "learning_rate": 0.0001999317060143023,
      "loss": 20.059,
      "step": 948
    },
    {
      "epoch": 0.16594535519125683,
      "grad_norm": 1.1386906446543563,
      "learning_rate": 0.00019993118174634918,
      "loss": 20.1688,
      "step": 949
    },
    {
      "epoch": 0.16612021857923498,
      "grad_norm": 1.1698263182944173,
      "learning_rate": 0.00019993065547446902,
      "loss": 20.0738,
      "step": 950
    },
    {
      "epoch": 0.16629508196721313,
      "grad_norm": 1.1547755652932663,
      "learning_rate": 0.00019993012719867238,
      "loss": 20.3351,
      "step": 951
    },
    {
      "epoch": 0.16646994535519125,
      "grad_norm": 1.1554728396243306,
      "learning_rate": 0.00019992959691896988,
      "loss": 20.2557,
      "step": 952
    },
    {
      "epoch": 0.1666448087431694,
      "grad_norm": 1.1490690205012857,
      "learning_rate": 0.00019992906463537212,
      "loss": 19.9878,
      "step": 953
    },
    {
      "epoch": 0.16681967213114754,
      "grad_norm": 1.1982987168205987,
      "learning_rate": 0.0001999285303478898,
      "loss": 19.9687,
      "step": 954
    },
    {
      "epoch": 0.1669945355191257,
      "grad_norm": 1.1675197991998942,
      "learning_rate": 0.00019992799405653365,
      "loss": 19.9986,
      "step": 955
    },
    {
      "epoch": 0.16716939890710383,
      "grad_norm": 1.1342371717190083,
      "learning_rate": 0.00019992745576131435,
      "loss": 20.1421,
      "step": 956
    },
    {
      "epoch": 0.16734426229508198,
      "grad_norm": 1.2033432065404768,
      "learning_rate": 0.00019992691546224278,
      "loss": 20.0178,
      "step": 957
    },
    {
      "epoch": 0.1675191256830601,
      "grad_norm": 1.1076342721987193,
      "learning_rate": 0.00019992637315932973,
      "loss": 19.8328,
      "step": 958
    },
    {
      "epoch": 0.16769398907103825,
      "grad_norm": 1.164326207478792,
      "learning_rate": 0.0001999258288525861,
      "loss": 19.9503,
      "step": 959
    },
    {
      "epoch": 0.1678688524590164,
      "grad_norm": 1.129747780444674,
      "learning_rate": 0.00019992528254202276,
      "loss": 20.058,
      "step": 960
    },
    {
      "epoch": 0.16804371584699454,
      "grad_norm": 1.1709380318164213,
      "learning_rate": 0.00019992473422765072,
      "loss": 19.7679,
      "step": 961
    },
    {
      "epoch": 0.1682185792349727,
      "grad_norm": 1.1800078022585427,
      "learning_rate": 0.00019992418390948094,
      "loss": 19.8085,
      "step": 962
    },
    {
      "epoch": 0.1683934426229508,
      "grad_norm": 1.2441739663942575,
      "learning_rate": 0.00019992363158752446,
      "loss": 19.9235,
      "step": 963
    },
    {
      "epoch": 0.16856830601092895,
      "grad_norm": 1.102192876811978,
      "learning_rate": 0.00019992307726179239,
      "loss": 20.1165,
      "step": 964
    },
    {
      "epoch": 0.1687431693989071,
      "grad_norm": 1.1748850828980537,
      "learning_rate": 0.00019992252093229578,
      "loss": 20.0796,
      "step": 965
    },
    {
      "epoch": 0.16891803278688525,
      "grad_norm": 1.1262773898359262,
      "learning_rate": 0.00019992196259904584,
      "loss": 20.4954,
      "step": 966
    },
    {
      "epoch": 0.1690928961748634,
      "grad_norm": 1.1672845204066251,
      "learning_rate": 0.00019992140226205373,
      "loss": 19.8036,
      "step": 967
    },
    {
      "epoch": 0.16926775956284154,
      "grad_norm": 1.0759996052961989,
      "learning_rate": 0.00019992083992133073,
      "loss": 20.3033,
      "step": 968
    },
    {
      "epoch": 0.16944262295081966,
      "grad_norm": 1.1459908734992068,
      "learning_rate": 0.00019992027557688807,
      "loss": 20.0917,
      "step": 969
    },
    {
      "epoch": 0.1696174863387978,
      "grad_norm": 1.1965906326005173,
      "learning_rate": 0.0001999197092287371,
      "loss": 20.2545,
      "step": 970
    },
    {
      "epoch": 0.16979234972677595,
      "grad_norm": 1.1868939177887525,
      "learning_rate": 0.00019991914087688918,
      "loss": 19.9493,
      "step": 971
    },
    {
      "epoch": 0.1699672131147541,
      "grad_norm": 1.1673312045489024,
      "learning_rate": 0.00019991857052135568,
      "loss": 20.1219,
      "step": 972
    },
    {
      "epoch": 0.17014207650273225,
      "grad_norm": 1.1235285653222282,
      "learning_rate": 0.00019991799816214805,
      "loss": 20.1555,
      "step": 973
    },
    {
      "epoch": 0.1703169398907104,
      "grad_norm": 1.1069394765937701,
      "learning_rate": 0.00019991742379927777,
      "loss": 19.7365,
      "step": 974
    },
    {
      "epoch": 0.17049180327868851,
      "grad_norm": 1.151556640453905,
      "learning_rate": 0.00019991684743275636,
      "loss": 19.723,
      "step": 975
    },
    {
      "epoch": 0.17066666666666666,
      "grad_norm": 1.1553433132218378,
      "learning_rate": 0.00019991626906259535,
      "loss": 19.7851,
      "step": 976
    },
    {
      "epoch": 0.1708415300546448,
      "grad_norm": 1.1246546993518263,
      "learning_rate": 0.0001999156886888064,
      "loss": 19.7819,
      "step": 977
    },
    {
      "epoch": 0.17101639344262295,
      "grad_norm": 1.0803224241559586,
      "learning_rate": 0.0001999151063114011,
      "loss": 20.1077,
      "step": 978
    },
    {
      "epoch": 0.1711912568306011,
      "grad_norm": 1.1377626621504113,
      "learning_rate": 0.00019991452193039112,
      "loss": 19.8148,
      "step": 979
    },
    {
      "epoch": 0.17136612021857925,
      "grad_norm": 1.1242303353175351,
      "learning_rate": 0.0001999139355457882,
      "loss": 19.9692,
      "step": 980
    },
    {
      "epoch": 0.17154098360655737,
      "grad_norm": 1.1554989254379227,
      "learning_rate": 0.0001999133471576041,
      "loss": 20.2511,
      "step": 981
    },
    {
      "epoch": 0.17171584699453551,
      "grad_norm": 1.075134602588708,
      "learning_rate": 0.00019991275676585062,
      "loss": 19.6947,
      "step": 982
    },
    {
      "epoch": 0.17189071038251366,
      "grad_norm": 1.0540639863238987,
      "learning_rate": 0.0001999121643705396,
      "loss": 20.1618,
      "step": 983
    },
    {
      "epoch": 0.1720655737704918,
      "grad_norm": 1.2021063643822423,
      "learning_rate": 0.00019991156997168293,
      "loss": 20.1272,
      "step": 984
    },
    {
      "epoch": 0.17224043715846996,
      "grad_norm": 1.1637090541847974,
      "learning_rate": 0.00019991097356929246,
      "loss": 19.8793,
      "step": 985
    },
    {
      "epoch": 0.17241530054644807,
      "grad_norm": 1.0995996072982834,
      "learning_rate": 0.00019991037516338024,
      "loss": 19.8358,
      "step": 986
    },
    {
      "epoch": 0.17259016393442622,
      "grad_norm": 1.186572305161176,
      "learning_rate": 0.0001999097747539582,
      "loss": 19.9433,
      "step": 987
    },
    {
      "epoch": 0.17276502732240437,
      "grad_norm": 1.2147950340509788,
      "learning_rate": 0.00019990917234103848,
      "loss": 19.8446,
      "step": 988
    },
    {
      "epoch": 0.17293989071038252,
      "grad_norm": 1.141789734107336,
      "learning_rate": 0.00019990856792463305,
      "loss": 19.7494,
      "step": 989
    },
    {
      "epoch": 0.17311475409836066,
      "grad_norm": 1.1413237363524735,
      "learning_rate": 0.00019990796150475405,
      "loss": 19.6178,
      "step": 990
    },
    {
      "epoch": 0.1732896174863388,
      "grad_norm": 1.1808059132617226,
      "learning_rate": 0.00019990735308141372,
      "loss": 20.0813,
      "step": 991
    },
    {
      "epoch": 0.17346448087431693,
      "grad_norm": 1.167360496278687,
      "learning_rate": 0.00019990674265462418,
      "loss": 20.0559,
      "step": 992
    },
    {
      "epoch": 0.17363934426229508,
      "grad_norm": 1.1449828994809284,
      "learning_rate": 0.0001999061302243977,
      "loss": 19.6851,
      "step": 993
    },
    {
      "epoch": 0.17381420765027322,
      "grad_norm": 1.1943760020827703,
      "learning_rate": 0.00019990551579074652,
      "loss": 19.8585,
      "step": 994
    },
    {
      "epoch": 0.17398907103825137,
      "grad_norm": 1.123064695664274,
      "learning_rate": 0.00019990489935368304,
      "loss": 20.1157,
      "step": 995
    },
    {
      "epoch": 0.17416393442622952,
      "grad_norm": 1.2875204739416846,
      "learning_rate": 0.00019990428091321957,
      "loss": 19.7159,
      "step": 996
    },
    {
      "epoch": 0.17433879781420766,
      "grad_norm": 1.326803978623737,
      "learning_rate": 0.00019990366046936851,
      "loss": 19.9101,
      "step": 997
    },
    {
      "epoch": 0.17451366120218578,
      "grad_norm": 1.22149536704579,
      "learning_rate": 0.00019990303802214232,
      "loss": 19.951,
      "step": 998
    },
    {
      "epoch": 0.17468852459016393,
      "grad_norm": 1.0699935065022694,
      "learning_rate": 0.00019990241357155348,
      "loss": 20.1762,
      "step": 999
    },
    {
      "epoch": 0.17486338797814208,
      "grad_norm": 1.2719569145420992,
      "learning_rate": 0.0001999017871176145,
      "loss": 20.0314,
      "step": 1000
    },
    {
      "epoch": 0.17503825136612022,
      "grad_norm": 1.2295560019169895,
      "learning_rate": 0.00019990115866033796,
      "loss": 20.0919,
      "step": 1001
    },
    {
      "epoch": 0.17521311475409837,
      "grad_norm": 1.238065300459919,
      "learning_rate": 0.00019990052819973642,
      "loss": 19.631,
      "step": 1002
    },
    {
      "epoch": 0.1753879781420765,
      "grad_norm": 1.1367486266582882,
      "learning_rate": 0.00019989989573582258,
      "loss": 19.9963,
      "step": 1003
    },
    {
      "epoch": 0.17556284153005464,
      "grad_norm": 1.3453960389928026,
      "learning_rate": 0.00019989926126860912,
      "loss": 19.6054,
      "step": 1004
    },
    {
      "epoch": 0.17573770491803278,
      "grad_norm": 1.3666583532674228,
      "learning_rate": 0.0001998986247981087,
      "loss": 19.9609,
      "step": 1005
    },
    {
      "epoch": 0.17591256830601093,
      "grad_norm": 1.228433439906144,
      "learning_rate": 0.00019989798632433415,
      "loss": 20.0835,
      "step": 1006
    },
    {
      "epoch": 0.17608743169398908,
      "grad_norm": 1.1640511555106214,
      "learning_rate": 0.00019989734584729822,
      "loss": 20.1507,
      "step": 1007
    },
    {
      "epoch": 0.17626229508196722,
      "grad_norm": 1.1791547168599357,
      "learning_rate": 0.00019989670336701383,
      "loss": 19.9704,
      "step": 1008
    },
    {
      "epoch": 0.17643715846994534,
      "grad_norm": 1.2387319899093845,
      "learning_rate": 0.00019989605888349376,
      "loss": 19.948,
      "step": 1009
    },
    {
      "epoch": 0.1766120218579235,
      "grad_norm": 1.142555522502723,
      "learning_rate": 0.000199895412396751,
      "loss": 20.2427,
      "step": 1010
    },
    {
      "epoch": 0.17678688524590164,
      "grad_norm": 1.1190865867223208,
      "learning_rate": 0.00019989476390679854,
      "loss": 20.2577,
      "step": 1011
    },
    {
      "epoch": 0.17696174863387978,
      "grad_norm": 1.2146670295353175,
      "learning_rate": 0.0001998941134136493,
      "loss": 19.7205,
      "step": 1012
    },
    {
      "epoch": 0.17713661202185793,
      "grad_norm": 1.2220881047391459,
      "learning_rate": 0.0001998934609173164,
      "loss": 19.9641,
      "step": 1013
    },
    {
      "epoch": 0.17731147540983608,
      "grad_norm": 1.1152874263478156,
      "learning_rate": 0.0001998928064178129,
      "loss": 19.9733,
      "step": 1014
    },
    {
      "epoch": 0.1774863387978142,
      "grad_norm": 1.1341400816443308,
      "learning_rate": 0.0001998921499151519,
      "loss": 19.9546,
      "step": 1015
    },
    {
      "epoch": 0.17766120218579234,
      "grad_norm": 1.2584939437697873,
      "learning_rate": 0.0001998914914093466,
      "loss": 19.8885,
      "step": 1016
    },
    {
      "epoch": 0.1778360655737705,
      "grad_norm": 1.2020124160353796,
      "learning_rate": 0.00019989083090041017,
      "loss": 19.8467,
      "step": 1017
    },
    {
      "epoch": 0.17801092896174864,
      "grad_norm": 1.1804644950101044,
      "learning_rate": 0.00019989016838835587,
      "loss": 19.8539,
      "step": 1018
    },
    {
      "epoch": 0.17818579234972678,
      "grad_norm": 1.2181090977397007,
      "learning_rate": 0.00019988950387319701,
      "loss": 19.5595,
      "step": 1019
    },
    {
      "epoch": 0.17836065573770493,
      "grad_norm": 1.2961070321624584,
      "learning_rate": 0.0001998888373549469,
      "loss": 19.7027,
      "step": 1020
    },
    {
      "epoch": 0.17853551912568305,
      "grad_norm": 1.318327761090375,
      "learning_rate": 0.0001998881688336189,
      "loss": 20.0953,
      "step": 1021
    },
    {
      "epoch": 0.1787103825136612,
      "grad_norm": 1.382550390459152,
      "learning_rate": 0.0001998874983092264,
      "loss": 20.0295,
      "step": 1022
    },
    {
      "epoch": 0.17888524590163934,
      "grad_norm": 1.201144574187912,
      "learning_rate": 0.00019988682578178288,
      "loss": 19.866,
      "step": 1023
    },
    {
      "epoch": 0.1790601092896175,
      "grad_norm": 1.3183490982713186,
      "learning_rate": 0.00019988615125130182,
      "loss": 19.9394,
      "step": 1024
    },
    {
      "epoch": 0.17923497267759564,
      "grad_norm": 1.2942332442186464,
      "learning_rate": 0.00019988547471779672,
      "loss": 19.8535,
      "step": 1025
    },
    {
      "epoch": 0.17940983606557376,
      "grad_norm": 1.2388961659139954,
      "learning_rate": 0.00019988479618128117,
      "loss": 20.0865,
      "step": 1026
    },
    {
      "epoch": 0.1795846994535519,
      "grad_norm": 1.2391009001851334,
      "learning_rate": 0.00019988411564176876,
      "loss": 19.9712,
      "step": 1027
    },
    {
      "epoch": 0.17975956284153005,
      "grad_norm": 1.1455901926822212,
      "learning_rate": 0.00019988343309927314,
      "loss": 20.1749,
      "step": 1028
    },
    {
      "epoch": 0.1799344262295082,
      "grad_norm": 1.2363609447876118,
      "learning_rate": 0.00019988274855380804,
      "loss": 19.9605,
      "step": 1029
    },
    {
      "epoch": 0.18010928961748635,
      "grad_norm": 1.2899734468845627,
      "learning_rate": 0.00019988206200538712,
      "loss": 19.7994,
      "step": 1030
    },
    {
      "epoch": 0.1802841530054645,
      "grad_norm": 1.2790147015553956,
      "learning_rate": 0.00019988137345402417,
      "loss": 19.8291,
      "step": 1031
    },
    {
      "epoch": 0.1804590163934426,
      "grad_norm": 1.1795386594901733,
      "learning_rate": 0.00019988068289973306,
      "loss": 20.0279,
      "step": 1032
    },
    {
      "epoch": 0.18063387978142076,
      "grad_norm": 1.2522155196003895,
      "learning_rate": 0.00019987999034252752,
      "loss": 20.1047,
      "step": 1033
    },
    {
      "epoch": 0.1808087431693989,
      "grad_norm": 1.2176255623431762,
      "learning_rate": 0.00019987929578242156,
      "loss": 19.2694,
      "step": 1034
    },
    {
      "epoch": 0.18098360655737705,
      "grad_norm": 1.1748427851462961,
      "learning_rate": 0.00019987859921942903,
      "loss": 19.7805,
      "step": 1035
    },
    {
      "epoch": 0.1811584699453552,
      "grad_norm": 1.213298949589247,
      "learning_rate": 0.00019987790065356385,
      "loss": 20.1237,
      "step": 1036
    },
    {
      "epoch": 0.18133333333333335,
      "grad_norm": 1.2503338224956664,
      "learning_rate": 0.0001998772000848402,
      "loss": 19.7769,
      "step": 1037
    },
    {
      "epoch": 0.18150819672131147,
      "grad_norm": 1.3091546479193876,
      "learning_rate": 0.00019987649751327196,
      "loss": 19.8886,
      "step": 1038
    },
    {
      "epoch": 0.1816830601092896,
      "grad_norm": 1.333465270057225,
      "learning_rate": 0.0001998757929388733,
      "loss": 19.7106,
      "step": 1039
    },
    {
      "epoch": 0.18185792349726776,
      "grad_norm": 1.2227856637886774,
      "learning_rate": 0.00019987508636165835,
      "loss": 19.9109,
      "step": 1040
    },
    {
      "epoch": 0.1820327868852459,
      "grad_norm": 1.2084623910373642,
      "learning_rate": 0.00019987437778164123,
      "loss": 19.7074,
      "step": 1041
    },
    {
      "epoch": 0.18220765027322405,
      "grad_norm": 1.2514955580068692,
      "learning_rate": 0.0001998736671988362,
      "loss": 19.3862,
      "step": 1042
    },
    {
      "epoch": 0.18238251366120217,
      "grad_norm": 1.1781934555210494,
      "learning_rate": 0.0001998729546132575,
      "loss": 19.6758,
      "step": 1043
    },
    {
      "epoch": 0.18255737704918032,
      "grad_norm": 1.2133342782964363,
      "learning_rate": 0.0001998722400249194,
      "loss": 19.6454,
      "step": 1044
    },
    {
      "epoch": 0.18273224043715847,
      "grad_norm": 1.183762781031916,
      "learning_rate": 0.00019987152343383624,
      "loss": 19.7782,
      "step": 1045
    },
    {
      "epoch": 0.1829071038251366,
      "grad_norm": 1.2144104381932932,
      "learning_rate": 0.00019987080484002237,
      "loss": 19.7426,
      "step": 1046
    },
    {
      "epoch": 0.18308196721311476,
      "grad_norm": 1.2170234715428914,
      "learning_rate": 0.00019987008424349225,
      "loss": 19.8575,
      "step": 1047
    },
    {
      "epoch": 0.1832568306010929,
      "grad_norm": 1.2848991813970367,
      "learning_rate": 0.00019986936164426025,
      "loss": 19.3252,
      "step": 1048
    },
    {
      "epoch": 0.18343169398907103,
      "grad_norm": 1.2882332368965819,
      "learning_rate": 0.000199868637042341,
      "loss": 19.5533,
      "step": 1049
    },
    {
      "epoch": 0.18360655737704917,
      "grad_norm": 1.2032203207067222,
      "learning_rate": 0.00019986791043774888,
      "loss": 19.9087,
      "step": 1050
    },
    {
      "epoch": 0.18378142076502732,
      "grad_norm": 1.1472867999927339,
      "learning_rate": 0.0001998671818304985,
      "loss": 19.6404,
      "step": 1051
    },
    {
      "epoch": 0.18395628415300547,
      "grad_norm": 1.3357062183503907,
      "learning_rate": 0.00019986645122060451,
      "loss": 19.5959,
      "step": 1052
    },
    {
      "epoch": 0.18413114754098361,
      "grad_norm": 1.4446660986910416,
      "learning_rate": 0.00019986571860808157,
      "loss": 19.7479,
      "step": 1053
    },
    {
      "epoch": 0.18430601092896176,
      "grad_norm": 1.2513152776907368,
      "learning_rate": 0.00019986498399294435,
      "loss": 19.7416,
      "step": 1054
    },
    {
      "epoch": 0.18448087431693988,
      "grad_norm": 1.157068715132537,
      "learning_rate": 0.0001998642473752075,
      "loss": 19.7205,
      "step": 1055
    },
    {
      "epoch": 0.18465573770491803,
      "grad_norm": 1.3286448273521287,
      "learning_rate": 0.00019986350875488595,
      "loss": 19.7106,
      "step": 1056
    },
    {
      "epoch": 0.18483060109289617,
      "grad_norm": 1.2721178144659686,
      "learning_rate": 0.00019986276813199438,
      "loss": 19.5673,
      "step": 1057
    },
    {
      "epoch": 0.18500546448087432,
      "grad_norm": 1.243996562584831,
      "learning_rate": 0.00019986202550654773,
      "loss": 19.8324,
      "step": 1058
    },
    {
      "epoch": 0.18518032786885247,
      "grad_norm": 1.1436627356383338,
      "learning_rate": 0.0001998612808785608,
      "loss": 19.8835,
      "step": 1059
    },
    {
      "epoch": 0.18535519125683061,
      "grad_norm": 1.2840773200328772,
      "learning_rate": 0.0001998605342480486,
      "loss": 19.8233,
      "step": 1060
    },
    {
      "epoch": 0.18553005464480873,
      "grad_norm": 1.2261209360483256,
      "learning_rate": 0.00019985978561502606,
      "loss": 19.5534,
      "step": 1061
    },
    {
      "epoch": 0.18570491803278688,
      "grad_norm": 1.3340656679246712,
      "learning_rate": 0.00019985903497950823,
      "loss": 19.6372,
      "step": 1062
    },
    {
      "epoch": 0.18587978142076503,
      "grad_norm": 1.2617093997379982,
      "learning_rate": 0.00019985828234151017,
      "loss": 19.3933,
      "step": 1063
    },
    {
      "epoch": 0.18605464480874317,
      "grad_norm": 1.2657138335497389,
      "learning_rate": 0.0001998575277010469,
      "loss": 19.4275,
      "step": 1064
    },
    {
      "epoch": 0.18622950819672132,
      "grad_norm": 1.107500674190652,
      "learning_rate": 0.00019985677105813362,
      "loss": 19.8533,
      "step": 1065
    },
    {
      "epoch": 0.18640437158469944,
      "grad_norm": 1.2964454503564615,
      "learning_rate": 0.00019985601241278547,
      "loss": 19.4697,
      "step": 1066
    },
    {
      "epoch": 0.1865792349726776,
      "grad_norm": 1.2431852297698436,
      "learning_rate": 0.00019985525176501764,
      "loss": 19.7901,
      "step": 1067
    },
    {
      "epoch": 0.18675409836065573,
      "grad_norm": 1.2074134516912622,
      "learning_rate": 0.00019985448911484546,
      "loss": 19.7868,
      "step": 1068
    },
    {
      "epoch": 0.18692896174863388,
      "grad_norm": 1.2199208423854755,
      "learning_rate": 0.00019985372446228416,
      "loss": 19.8265,
      "step": 1069
    },
    {
      "epoch": 0.18710382513661203,
      "grad_norm": 1.2098436752609751,
      "learning_rate": 0.0001998529578073491,
      "loss": 19.4768,
      "step": 1070
    },
    {
      "epoch": 0.18727868852459018,
      "grad_norm": 1.256914971654573,
      "learning_rate": 0.00019985218915005563,
      "loss": 19.4988,
      "step": 1071
    },
    {
      "epoch": 0.1874535519125683,
      "grad_norm": 1.2837621202744474,
      "learning_rate": 0.0001998514184904192,
      "loss": 19.5879,
      "step": 1072
    },
    {
      "epoch": 0.18762841530054644,
      "grad_norm": 1.3285834399133396,
      "learning_rate": 0.00019985064582845526,
      "loss": 19.2766,
      "step": 1073
    },
    {
      "epoch": 0.1878032786885246,
      "grad_norm": 1.2870460580020187,
      "learning_rate": 0.00019984987116417928,
      "loss": 19.744,
      "step": 1074
    },
    {
      "epoch": 0.18797814207650274,
      "grad_norm": 1.1216753486806452,
      "learning_rate": 0.0001998490944976068,
      "loss": 19.7202,
      "step": 1075
    },
    {
      "epoch": 0.18815300546448088,
      "grad_norm": 1.1938222916660803,
      "learning_rate": 0.0001998483158287534,
      "loss": 19.0789,
      "step": 1076
    },
    {
      "epoch": 0.18832786885245903,
      "grad_norm": 1.162116110193304,
      "learning_rate": 0.0001998475351576347,
      "loss": 19.8614,
      "step": 1077
    },
    {
      "epoch": 0.18850273224043715,
      "grad_norm": 1.133583167091585,
      "learning_rate": 0.0001998467524842663,
      "loss": 19.0632,
      "step": 1078
    },
    {
      "epoch": 0.1886775956284153,
      "grad_norm": 1.2521902013531792,
      "learning_rate": 0.00019984596780866397,
      "loss": 19.6123,
      "step": 1079
    },
    {
      "epoch": 0.18885245901639344,
      "grad_norm": 1.1828390301142793,
      "learning_rate": 0.00019984518113084346,
      "loss": 19.7575,
      "step": 1080
    },
    {
      "epoch": 0.1890273224043716,
      "grad_norm": 1.2778341684171686,
      "learning_rate": 0.00019984439245082045,
      "loss": 19.7224,
      "step": 1081
    },
    {
      "epoch": 0.18920218579234974,
      "grad_norm": 1.4375978651153978,
      "learning_rate": 0.00019984360176861083,
      "loss": 19.3491,
      "step": 1082
    },
    {
      "epoch": 0.18937704918032786,
      "grad_norm": 1.3621548092370634,
      "learning_rate": 0.00019984280908423043,
      "loss": 19.5733,
      "step": 1083
    },
    {
      "epoch": 0.189551912568306,
      "grad_norm": 1.231212123445474,
      "learning_rate": 0.00019984201439769516,
      "loss": 19.8113,
      "step": 1084
    },
    {
      "epoch": 0.18972677595628415,
      "grad_norm": 1.1953418929948523,
      "learning_rate": 0.00019984121770902092,
      "loss": 19.514,
      "step": 1085
    },
    {
      "epoch": 0.1899016393442623,
      "grad_norm": 1.214674655383464,
      "learning_rate": 0.00019984041901822375,
      "loss": 19.8074,
      "step": 1086
    },
    {
      "epoch": 0.19007650273224044,
      "grad_norm": 1.2346713635225972,
      "learning_rate": 0.0001998396183253196,
      "loss": 19.2558,
      "step": 1087
    },
    {
      "epoch": 0.1902513661202186,
      "grad_norm": 1.1828372623010048,
      "learning_rate": 0.00019983881563032457,
      "loss": 19.571,
      "step": 1088
    },
    {
      "epoch": 0.1904262295081967,
      "grad_norm": 1.2022095711808132,
      "learning_rate": 0.00019983801093325472,
      "loss": 19.7332,
      "step": 1089
    },
    {
      "epoch": 0.19060109289617486,
      "grad_norm": 1.266817807208223,
      "learning_rate": 0.00019983720423412624,
      "loss": 19.5523,
      "step": 1090
    },
    {
      "epoch": 0.190775956284153,
      "grad_norm": 1.2299221058581014,
      "learning_rate": 0.00019983639553295527,
      "loss": 19.7993,
      "step": 1091
    },
    {
      "epoch": 0.19095081967213115,
      "grad_norm": 1.4123240415821794,
      "learning_rate": 0.00019983558482975799,
      "loss": 19.381,
      "step": 1092
    },
    {
      "epoch": 0.1911256830601093,
      "grad_norm": 1.2217465107533667,
      "learning_rate": 0.00019983477212455074,
      "loss": 19.7275,
      "step": 1093
    },
    {
      "epoch": 0.19130054644808744,
      "grad_norm": 1.3079727901633809,
      "learning_rate": 0.00019983395741734976,
      "loss": 19.4376,
      "step": 1094
    },
    {
      "epoch": 0.19147540983606556,
      "grad_norm": 1.2012898144911488,
      "learning_rate": 0.00019983314070817141,
      "loss": 19.948,
      "step": 1095
    },
    {
      "epoch": 0.1916502732240437,
      "grad_norm": 1.651205999245776,
      "learning_rate": 0.00019983232199703206,
      "loss": 19.577,
      "step": 1096
    },
    {
      "epoch": 0.19182513661202186,
      "grad_norm": 1.4995464210000282,
      "learning_rate": 0.0001998315012839481,
      "loss": 19.0168,
      "step": 1097
    },
    {
      "epoch": 0.192,
      "grad_norm": 1.4537817265835158,
      "learning_rate": 0.00019983067856893606,
      "loss": 19.2421,
      "step": 1098
    },
    {
      "epoch": 0.19217486338797815,
      "grad_norm": 1.275953935515983,
      "learning_rate": 0.00019982985385201237,
      "loss": 19.8255,
      "step": 1099
    },
    {
      "epoch": 0.1923497267759563,
      "grad_norm": 1.5058719980877766,
      "learning_rate": 0.00019982902713319361,
      "loss": 19.5886,
      "step": 1100
    },
    {
      "epoch": 0.19252459016393442,
      "grad_norm": 1.4036561853372114,
      "learning_rate": 0.00019982819841249632,
      "loss": 19.7262,
      "step": 1101
    },
    {
      "epoch": 0.19269945355191256,
      "grad_norm": 1.6664447194353265,
      "learning_rate": 0.00019982736768993714,
      "loss": 19.9258,
      "step": 1102
    },
    {
      "epoch": 0.1928743169398907,
      "grad_norm": 1.4437504556107965,
      "learning_rate": 0.00019982653496553276,
      "loss": 19.127,
      "step": 1103
    },
    {
      "epoch": 0.19304918032786886,
      "grad_norm": 1.3239957303585588,
      "learning_rate": 0.0001998257002392998,
      "loss": 19.2583,
      "step": 1104
    },
    {
      "epoch": 0.193224043715847,
      "grad_norm": 1.1741154491180112,
      "learning_rate": 0.0001998248635112551,
      "loss": 19.679,
      "step": 1105
    },
    {
      "epoch": 0.19339890710382512,
      "grad_norm": 1.2317910672480414,
      "learning_rate": 0.00019982402478141534,
      "loss": 19.5524,
      "step": 1106
    },
    {
      "epoch": 0.19357377049180327,
      "grad_norm": 1.2539799844489594,
      "learning_rate": 0.00019982318404979742,
      "loss": 19.6415,
      "step": 1107
    },
    {
      "epoch": 0.19374863387978142,
      "grad_norm": 1.2551392235008234,
      "learning_rate": 0.0001998223413164181,
      "loss": 19.952,
      "step": 1108
    },
    {
      "epoch": 0.19392349726775956,
      "grad_norm": 1.2423064914522626,
      "learning_rate": 0.00019982149658129438,
      "loss": 19.5192,
      "step": 1109
    },
    {
      "epoch": 0.1940983606557377,
      "grad_norm": 1.1233589407317093,
      "learning_rate": 0.00019982064984444317,
      "loss": 19.8155,
      "step": 1110
    },
    {
      "epoch": 0.19427322404371586,
      "grad_norm": 1.2472054569080966,
      "learning_rate": 0.00019981980110588143,
      "loss": 19.2046,
      "step": 1111
    },
    {
      "epoch": 0.19444808743169398,
      "grad_norm": 1.158540112699403,
      "learning_rate": 0.00019981895036562622,
      "loss": 19.6753,
      "step": 1112
    },
    {
      "epoch": 0.19462295081967212,
      "grad_norm": 1.4607422626041087,
      "learning_rate": 0.00019981809762369452,
      "loss": 19.404,
      "step": 1113
    },
    {
      "epoch": 0.19479781420765027,
      "grad_norm": 1.812170881975119,
      "learning_rate": 0.0001998172428801035,
      "loss": 19.6754,
      "step": 1114
    },
    {
      "epoch": 0.19497267759562842,
      "grad_norm": 1.7128439784137908,
      "learning_rate": 0.0001998163861348703,
      "loss": 19.8966,
      "step": 1115
    },
    {
      "epoch": 0.19514754098360657,
      "grad_norm": 1.1415430157350845,
      "learning_rate": 0.00019981552738801206,
      "loss": 19.562,
      "step": 1116
    },
    {
      "epoch": 0.1953224043715847,
      "grad_norm": 1.3936804169589179,
      "learning_rate": 0.000199814666639546,
      "loss": 19.0767,
      "step": 1117
    },
    {
      "epoch": 0.19549726775956283,
      "grad_norm": 1.4073718938548696,
      "learning_rate": 0.00019981380388948945,
      "loss": 19.6506,
      "step": 1118
    },
    {
      "epoch": 0.19567213114754098,
      "grad_norm": 1.4489468831311556,
      "learning_rate": 0.00019981293913785963,
      "loss": 19.402,
      "step": 1119
    },
    {
      "epoch": 0.19584699453551913,
      "grad_norm": 1.1896048879387235,
      "learning_rate": 0.00019981207238467394,
      "loss": 19.5605,
      "step": 1120
    },
    {
      "epoch": 0.19602185792349727,
      "grad_norm": 1.2752609443065164,
      "learning_rate": 0.00019981120362994972,
      "loss": 19.7298,
      "step": 1121
    },
    {
      "epoch": 0.19619672131147542,
      "grad_norm": 1.2951962115081852,
      "learning_rate": 0.00019981033287370443,
      "loss": 19.6057,
      "step": 1122
    },
    {
      "epoch": 0.19637158469945354,
      "grad_norm": 1.430728517859216,
      "learning_rate": 0.00019980946011595547,
      "loss": 19.356,
      "step": 1123
    },
    {
      "epoch": 0.19654644808743169,
      "grad_norm": 1.4847004902815957,
      "learning_rate": 0.0001998085853567204,
      "loss": 19.2275,
      "step": 1124
    },
    {
      "epoch": 0.19672131147540983,
      "grad_norm": 1.3176651565898667,
      "learning_rate": 0.00019980770859601675,
      "loss": 19.5733,
      "step": 1125
    },
    {
      "epoch": 0.19689617486338798,
      "grad_norm": 1.2668206888403448,
      "learning_rate": 0.00019980682983386207,
      "loss": 19.3653,
      "step": 1126
    },
    {
      "epoch": 0.19707103825136613,
      "grad_norm": 1.2607351213167546,
      "learning_rate": 0.00019980594907027403,
      "loss": 19.208,
      "step": 1127
    },
    {
      "epoch": 0.19724590163934427,
      "grad_norm": 1.3062518543953086,
      "learning_rate": 0.00019980506630527022,
      "loss": 19.3826,
      "step": 1128
    },
    {
      "epoch": 0.1974207650273224,
      "grad_norm": 1.6282819623461982,
      "learning_rate": 0.00019980418153886845,
      "loss": 19.3088,
      "step": 1129
    },
    {
      "epoch": 0.19759562841530054,
      "grad_norm": 1.4497704436080718,
      "learning_rate": 0.00019980329477108636,
      "loss": 19.4497,
      "step": 1130
    },
    {
      "epoch": 0.1977704918032787,
      "grad_norm": 1.3586462724213713,
      "learning_rate": 0.0001998024060019418,
      "loss": 19.3329,
      "step": 1131
    },
    {
      "epoch": 0.19794535519125683,
      "grad_norm": 1.2973806948134066,
      "learning_rate": 0.00019980151523145255,
      "loss": 19.414,
      "step": 1132
    },
    {
      "epoch": 0.19812021857923498,
      "grad_norm": 1.356374328184987,
      "learning_rate": 0.0001998006224596365,
      "loss": 19.1384,
      "step": 1133
    },
    {
      "epoch": 0.19829508196721313,
      "grad_norm": 1.3161938111510776,
      "learning_rate": 0.00019979972768651151,
      "loss": 18.9664,
      "step": 1134
    },
    {
      "epoch": 0.19846994535519125,
      "grad_norm": 1.4980018097385037,
      "learning_rate": 0.00019979883091209559,
      "loss": 18.4223,
      "step": 1135
    },
    {
      "epoch": 0.1986448087431694,
      "grad_norm": 1.3200038404279868,
      "learning_rate": 0.00019979793213640665,
      "loss": 19.3463,
      "step": 1136
    },
    {
      "epoch": 0.19881967213114754,
      "grad_norm": 1.2904733146772662,
      "learning_rate": 0.00019979703135946278,
      "loss": 19.4268,
      "step": 1137
    },
    {
      "epoch": 0.1989945355191257,
      "grad_norm": 1.4405056416466069,
      "learning_rate": 0.00019979612858128204,
      "loss": 19.3527,
      "step": 1138
    },
    {
      "epoch": 0.19916939890710383,
      "grad_norm": 1.6012625850627917,
      "learning_rate": 0.00019979522380188244,
      "loss": 19.2393,
      "step": 1139
    },
    {
      "epoch": 0.19934426229508198,
      "grad_norm": 1.4004760519288635,
      "learning_rate": 0.00019979431702128224,
      "loss": 19.3067,
      "step": 1140
    },
    {
      "epoch": 0.1995191256830601,
      "grad_norm": 1.2980994954779332,
      "learning_rate": 0.00019979340823949958,
      "loss": 19.5482,
      "step": 1141
    },
    {
      "epoch": 0.19969398907103825,
      "grad_norm": 1.434184025405644,
      "learning_rate": 0.00019979249745655264,
      "loss": 19.0846,
      "step": 1142
    },
    {
      "epoch": 0.1998688524590164,
      "grad_norm": 1.3867414861656766,
      "learning_rate": 0.00019979158467245975,
      "loss": 19.306,
      "step": 1143
    },
    {
      "epoch": 0.20004371584699454,
      "grad_norm": 1.633680944747411,
      "learning_rate": 0.0001997906698872392,
      "loss": 19.1142,
      "step": 1144
    },
    {
      "epoch": 0.2002185792349727,
      "grad_norm": 1.2446773363182233,
      "learning_rate": 0.0001997897531009093,
      "loss": 19.3802,
      "step": 1145
    },
    {
      "epoch": 0.2003934426229508,
      "grad_norm": 1.4213527850374903,
      "learning_rate": 0.00019978883431348845,
      "loss": 18.9061,
      "step": 1146
    },
    {
      "epoch": 0.20056830601092895,
      "grad_norm": 1.3098337770135602,
      "learning_rate": 0.0001997879135249951,
      "loss": 18.9905,
      "step": 1147
    },
    {
      "epoch": 0.2007431693989071,
      "grad_norm": 1.5400242082104931,
      "learning_rate": 0.00019978699073544768,
      "loss": 19.195,
      "step": 1148
    },
    {
      "epoch": 0.20091803278688525,
      "grad_norm": 1.4693635398978386,
      "learning_rate": 0.0001997860659448647,
      "loss": 19.0946,
      "step": 1149
    },
    {
      "epoch": 0.2010928961748634,
      "grad_norm": 1.7255569256049321,
      "learning_rate": 0.00019978513915326472,
      "loss": 19.3281,
      "step": 1150
    },
    {
      "epoch": 0.20126775956284154,
      "grad_norm": 1.2850074981554709,
      "learning_rate": 0.00019978421036066633,
      "loss": 18.8806,
      "step": 1151
    },
    {
      "epoch": 0.20144262295081966,
      "grad_norm": 1.3634900112981367,
      "learning_rate": 0.00019978327956708816,
      "loss": 19.4069,
      "step": 1152
    },
    {
      "epoch": 0.2016174863387978,
      "grad_norm": 1.3762238338225923,
      "learning_rate": 0.00019978234677254882,
      "loss": 19.2857,
      "step": 1153
    },
    {
      "epoch": 0.20179234972677595,
      "grad_norm": 1.369805994467164,
      "learning_rate": 0.00019978141197706707,
      "loss": 19.4787,
      "step": 1154
    },
    {
      "epoch": 0.2019672131147541,
      "grad_norm": 1.6800486975022955,
      "learning_rate": 0.00019978047518066165,
      "loss": 19.1987,
      "step": 1155
    },
    {
      "epoch": 0.20214207650273225,
      "grad_norm": 1.7354734060901142,
      "learning_rate": 0.0001997795363833513,
      "loss": 19.3728,
      "step": 1156
    },
    {
      "epoch": 0.2023169398907104,
      "grad_norm": 1.4460831346905336,
      "learning_rate": 0.00019977859558515494,
      "loss": 18.9131,
      "step": 1157
    },
    {
      "epoch": 0.20249180327868851,
      "grad_norm": 1.8907816013206171,
      "learning_rate": 0.00019977765278609137,
      "loss": 19.2342,
      "step": 1158
    },
    {
      "epoch": 0.20266666666666666,
      "grad_norm": 1.5407943264804012,
      "learning_rate": 0.00019977670798617945,
      "loss": 18.6079,
      "step": 1159
    },
    {
      "epoch": 0.2028415300546448,
      "grad_norm": 1.6176238728983716,
      "learning_rate": 0.00019977576118543824,
      "loss": 18.997,
      "step": 1160
    },
    {
      "epoch": 0.20301639344262296,
      "grad_norm": 1.4824776797074126,
      "learning_rate": 0.00019977481238388664,
      "loss": 19.2823,
      "step": 1161
    },
    {
      "epoch": 0.2031912568306011,
      "grad_norm": 1.5994637186385923,
      "learning_rate": 0.0001997738615815437,
      "loss": 19.0843,
      "step": 1162
    },
    {
      "epoch": 0.20336612021857922,
      "grad_norm": 1.2672709191138487,
      "learning_rate": 0.00019977290877842848,
      "loss": 19.2636,
      "step": 1163
    },
    {
      "epoch": 0.20354098360655737,
      "grad_norm": 1.4079702993843715,
      "learning_rate": 0.0001997719539745601,
      "loss": 19.1611,
      "step": 1164
    },
    {
      "epoch": 0.20371584699453552,
      "grad_norm": 1.3328869759940154,
      "learning_rate": 0.00019977099716995776,
      "loss": 18.9119,
      "step": 1165
    },
    {
      "epoch": 0.20389071038251366,
      "grad_norm": 1.3750752380066562,
      "learning_rate": 0.00019977003836464053,
      "loss": 18.8559,
      "step": 1166
    },
    {
      "epoch": 0.2040655737704918,
      "grad_norm": 1.498454794789553,
      "learning_rate": 0.0001997690775586277,
      "loss": 19.2411,
      "step": 1167
    },
    {
      "epoch": 0.20424043715846996,
      "grad_norm": 1.606801350172861,
      "learning_rate": 0.00019976811475193857,
      "loss": 19.1986,
      "step": 1168
    },
    {
      "epoch": 0.20441530054644808,
      "grad_norm": 1.5972962522696752,
      "learning_rate": 0.0001997671499445924,
      "loss": 19.1521,
      "step": 1169
    },
    {
      "epoch": 0.20459016393442622,
      "grad_norm": 1.6019116114245793,
      "learning_rate": 0.0001997661831366085,
      "loss": 19.1882,
      "step": 1170
    },
    {
      "epoch": 0.20476502732240437,
      "grad_norm": 1.5696267070850312,
      "learning_rate": 0.00019976521432800635,
      "loss": 18.869,
      "step": 1171
    },
    {
      "epoch": 0.20493989071038252,
      "grad_norm": 1.4764180858993705,
      "learning_rate": 0.0001997642435188053,
      "loss": 19.0508,
      "step": 1172
    },
    {
      "epoch": 0.20511475409836066,
      "grad_norm": 1.6972254800697144,
      "learning_rate": 0.0001997632707090249,
      "loss": 19.2651,
      "step": 1173
    },
    {
      "epoch": 0.2052896174863388,
      "grad_norm": 1.4111211246790611,
      "learning_rate": 0.00019976229589868457,
      "loss": 18.9528,
      "step": 1174
    },
    {
      "epoch": 0.20546448087431693,
      "grad_norm": 1.3323946551187549,
      "learning_rate": 0.0001997613190878039,
      "loss": 18.9111,
      "step": 1175
    },
    {
      "epoch": 0.20563934426229508,
      "grad_norm": 1.5743366210255239,
      "learning_rate": 0.00019976034027640249,
      "loss": 18.9334,
      "step": 1176
    },
    {
      "epoch": 0.20581420765027322,
      "grad_norm": 1.4193603540721436,
      "learning_rate": 0.00019975935946449992,
      "loss": 18.9803,
      "step": 1177
    },
    {
      "epoch": 0.20598907103825137,
      "grad_norm": 1.4817552454472063,
      "learning_rate": 0.00019975837665211593,
      "loss": 19.1929,
      "step": 1178
    },
    {
      "epoch": 0.20616393442622952,
      "grad_norm": 1.6319358201156133,
      "learning_rate": 0.0001997573918392702,
      "loss": 18.4559,
      "step": 1179
    },
    {
      "epoch": 0.20633879781420766,
      "grad_norm": 1.5324618676223967,
      "learning_rate": 0.00019975640502598244,
      "loss": 18.7896,
      "step": 1180
    },
    {
      "epoch": 0.20651366120218578,
      "grad_norm": 1.4553857923908278,
      "learning_rate": 0.00019975541621227247,
      "loss": 19.0452,
      "step": 1181
    },
    {
      "epoch": 0.20668852459016393,
      "grad_norm": 1.4139720287450612,
      "learning_rate": 0.00019975442539816012,
      "loss": 19.0383,
      "step": 1182
    },
    {
      "epoch": 0.20686338797814208,
      "grad_norm": 1.4406113946439811,
      "learning_rate": 0.00019975343258366523,
      "loss": 18.8892,
      "step": 1183
    },
    {
      "epoch": 0.20703825136612022,
      "grad_norm": 1.49847166587984,
      "learning_rate": 0.00019975243776880776,
      "loss": 18.8392,
      "step": 1184
    },
    {
      "epoch": 0.20721311475409837,
      "grad_norm": 1.451834392359958,
      "learning_rate": 0.00019975144095360766,
      "loss": 18.5782,
      "step": 1185
    },
    {
      "epoch": 0.2073879781420765,
      "grad_norm": 1.5848668410190943,
      "learning_rate": 0.00019975044213808483,
      "loss": 18.537,
      "step": 1186
    },
    {
      "epoch": 0.20756284153005464,
      "grad_norm": 1.6952389122298004,
      "learning_rate": 0.0001997494413222594,
      "loss": 18.8536,
      "step": 1187
    },
    {
      "epoch": 0.20773770491803278,
      "grad_norm": 1.4656959386461832,
      "learning_rate": 0.0001997484385061514,
      "loss": 18.9252,
      "step": 1188
    },
    {
      "epoch": 0.20791256830601093,
      "grad_norm": 1.5549271293919198,
      "learning_rate": 0.00019974743368978093,
      "loss": 18.7648,
      "step": 1189
    },
    {
      "epoch": 0.20808743169398908,
      "grad_norm": 1.623767039805432,
      "learning_rate": 0.00019974642687316813,
      "loss": 18.8595,
      "step": 1190
    },
    {
      "epoch": 0.20826229508196722,
      "grad_norm": 1.3500399768492317,
      "learning_rate": 0.00019974541805633324,
      "loss": 18.5881,
      "step": 1191
    },
    {
      "epoch": 0.20843715846994534,
      "grad_norm": 1.5831911711409519,
      "learning_rate": 0.00019974440723929645,
      "loss": 18.4978,
      "step": 1192
    },
    {
      "epoch": 0.2086120218579235,
      "grad_norm": 1.6790469912837134,
      "learning_rate": 0.00019974339442207803,
      "loss": 18.7143,
      "step": 1193
    },
    {
      "epoch": 0.20878688524590164,
      "grad_norm": 1.5241024165741224,
      "learning_rate": 0.0001997423796046983,
      "loss": 18.2884,
      "step": 1194
    },
    {
      "epoch": 0.20896174863387978,
      "grad_norm": 1.2997781835481208,
      "learning_rate": 0.00019974136278717763,
      "loss": 18.8656,
      "step": 1195
    },
    {
      "epoch": 0.20913661202185793,
      "grad_norm": 1.4156785450202574,
      "learning_rate": 0.00019974034396953637,
      "loss": 18.2964,
      "step": 1196
    },
    {
      "epoch": 0.20931147540983608,
      "grad_norm": 1.4786829057727557,
      "learning_rate": 0.000199739323151795,
      "loss": 18.4603,
      "step": 1197
    },
    {
      "epoch": 0.2094863387978142,
      "grad_norm": 1.5194331670776264,
      "learning_rate": 0.00019973830033397395,
      "loss": 18.5435,
      "step": 1198
    },
    {
      "epoch": 0.20966120218579234,
      "grad_norm": 1.480620262240517,
      "learning_rate": 0.0001997372755160937,
      "loss": 18.8647,
      "step": 1199
    },
    {
      "epoch": 0.2098360655737705,
      "grad_norm": 1.5155767274904588,
      "learning_rate": 0.0001997362486981749,
      "loss": 18.9648,
      "step": 1200
    },
    {
      "epoch": 0.21001092896174864,
      "grad_norm": 1.4502492183799505,
      "learning_rate": 0.00019973521988023805,
      "loss": 18.4829,
      "step": 1201
    },
    {
      "epoch": 0.21018579234972679,
      "grad_norm": 1.3354078251930883,
      "learning_rate": 0.0001997341890623038,
      "loss": 18.8582,
      "step": 1202
    },
    {
      "epoch": 0.2103606557377049,
      "grad_norm": 1.9248615591568814,
      "learning_rate": 0.00019973315624439286,
      "loss": 18.434,
      "step": 1203
    },
    {
      "epoch": 0.21053551912568305,
      "grad_norm": 2.235577943233893,
      "learning_rate": 0.00019973212142652594,
      "loss": 18.7399,
      "step": 1204
    },
    {
      "epoch": 0.2107103825136612,
      "grad_norm": 2.19590411340284,
      "learning_rate": 0.00019973108460872373,
      "loss": 18.9173,
      "step": 1205
    },
    {
      "epoch": 0.21088524590163935,
      "grad_norm": 1.8489004819736288,
      "learning_rate": 0.00019973004579100704,
      "loss": 18.6325,
      "step": 1206
    },
    {
      "epoch": 0.2110601092896175,
      "grad_norm": 1.4223875083746464,
      "learning_rate": 0.00019972900497339677,
      "loss": 18.6395,
      "step": 1207
    },
    {
      "epoch": 0.21123497267759564,
      "grad_norm": 1.6704317325874,
      "learning_rate": 0.0001997279621559137,
      "loss": 18.1645,
      "step": 1208
    },
    {
      "epoch": 0.21140983606557376,
      "grad_norm": 1.4623612973991087,
      "learning_rate": 0.00019972691733857883,
      "loss": 18.4806,
      "step": 1209
    },
    {
      "epoch": 0.2115846994535519,
      "grad_norm": 1.6394412227627881,
      "learning_rate": 0.00019972587052141303,
      "loss": 18.7789,
      "step": 1210
    },
    {
      "epoch": 0.21175956284153005,
      "grad_norm": 1.5406000855721909,
      "learning_rate": 0.00019972482170443732,
      "loss": 18.6543,
      "step": 1211
    },
    {
      "epoch": 0.2119344262295082,
      "grad_norm": 1.499202263225831,
      "learning_rate": 0.00019972377088767277,
      "loss": 18.5972,
      "step": 1212
    },
    {
      "epoch": 0.21210928961748635,
      "grad_norm": 1.5419550683934915,
      "learning_rate": 0.00019972271807114041,
      "loss": 18.6173,
      "step": 1213
    },
    {
      "epoch": 0.2122841530054645,
      "grad_norm": 1.4998646660908883,
      "learning_rate": 0.00019972166325486136,
      "loss": 18.5063,
      "step": 1214
    },
    {
      "epoch": 0.2124590163934426,
      "grad_norm": 1.9169005822387668,
      "learning_rate": 0.00019972060643885678,
      "loss": 18.5705,
      "step": 1215
    },
    {
      "epoch": 0.21263387978142076,
      "grad_norm": 1.7557261131166781,
      "learning_rate": 0.00019971954762314784,
      "loss": 18.1199,
      "step": 1216
    },
    {
      "epoch": 0.2128087431693989,
      "grad_norm": 1.3733933940155296,
      "learning_rate": 0.00019971848680775582,
      "loss": 18.271,
      "step": 1217
    },
    {
      "epoch": 0.21298360655737705,
      "grad_norm": 1.524525544390028,
      "learning_rate": 0.00019971742399270195,
      "loss": 18.2978,
      "step": 1218
    },
    {
      "epoch": 0.2131584699453552,
      "grad_norm": 1.5001944629731219,
      "learning_rate": 0.00019971635917800755,
      "loss": 18.5731,
      "step": 1219
    },
    {
      "epoch": 0.21333333333333335,
      "grad_norm": 1.4021659804557505,
      "learning_rate": 0.00019971529236369399,
      "loss": 18.3885,
      "step": 1220
    },
    {
      "epoch": 0.21350819672131147,
      "grad_norm": 1.4431230966988913,
      "learning_rate": 0.00019971422354978264,
      "loss": 18.6463,
      "step": 1221
    },
    {
      "epoch": 0.2136830601092896,
      "grad_norm": 1.3630707565327798,
      "learning_rate": 0.00019971315273629496,
      "loss": 18.2629,
      "step": 1222
    },
    {
      "epoch": 0.21385792349726776,
      "grad_norm": 1.427191954766837,
      "learning_rate": 0.0001997120799232524,
      "loss": 18.5385,
      "step": 1223
    },
    {
      "epoch": 0.2140327868852459,
      "grad_norm": 1.2277331799182356,
      "learning_rate": 0.0001997110051106765,
      "loss": 17.969,
      "step": 1224
    },
    {
      "epoch": 0.21420765027322405,
      "grad_norm": 1.5587932110034348,
      "learning_rate": 0.00019970992829858878,
      "loss": 18.2267,
      "step": 1225
    },
    {
      "epoch": 0.21438251366120217,
      "grad_norm": 1.4600585711584868,
      "learning_rate": 0.00019970884948701084,
      "loss": 18.5284,
      "step": 1226
    },
    {
      "epoch": 0.21455737704918032,
      "grad_norm": 1.447782615974436,
      "learning_rate": 0.0001997077686759643,
      "loss": 18.4844,
      "step": 1227
    },
    {
      "epoch": 0.21473224043715847,
      "grad_norm": 1.542476129959631,
      "learning_rate": 0.0001997066858654709,
      "loss": 18.5393,
      "step": 1228
    },
    {
      "epoch": 0.21490710382513661,
      "grad_norm": 1.6935290564540757,
      "learning_rate": 0.0001997056010555523,
      "loss": 18.2429,
      "step": 1229
    },
    {
      "epoch": 0.21508196721311476,
      "grad_norm": 1.5917208741786213,
      "learning_rate": 0.00019970451424623025,
      "loss": 18.0018,
      "step": 1230
    },
    {
      "epoch": 0.2152568306010929,
      "grad_norm": 1.4974146129905292,
      "learning_rate": 0.00019970342543752654,
      "loss": 18.6601,
      "step": 1231
    },
    {
      "epoch": 0.21543169398907103,
      "grad_norm": 1.681299655077825,
      "learning_rate": 0.00019970233462946303,
      "loss": 18.4797,
      "step": 1232
    },
    {
      "epoch": 0.21560655737704917,
      "grad_norm": 1.9131503894174526,
      "learning_rate": 0.0001997012418220616,
      "loss": 18.1164,
      "step": 1233
    },
    {
      "epoch": 0.21578142076502732,
      "grad_norm": 1.7636108724054402,
      "learning_rate": 0.00019970014701534413,
      "loss": 18.1013,
      "step": 1234
    },
    {
      "epoch": 0.21595628415300547,
      "grad_norm": 1.473694485951664,
      "learning_rate": 0.0001996990502093326,
      "loss": 18.3227,
      "step": 1235
    },
    {
      "epoch": 0.21613114754098361,
      "grad_norm": 1.393079968259773,
      "learning_rate": 0.000199697951404049,
      "loss": 18.091,
      "step": 1236
    },
    {
      "epoch": 0.21630601092896176,
      "grad_norm": 1.448109605064782,
      "learning_rate": 0.00019969685059951538,
      "loss": 18.0404,
      "step": 1237
    },
    {
      "epoch": 0.21648087431693988,
      "grad_norm": 1.5384899041029603,
      "learning_rate": 0.00019969574779575376,
      "loss": 18.1744,
      "step": 1238
    },
    {
      "epoch": 0.21665573770491803,
      "grad_norm": 1.780669431842745,
      "learning_rate": 0.00019969464299278634,
      "loss": 18.2271,
      "step": 1239
    },
    {
      "epoch": 0.21683060109289617,
      "grad_norm": 2.130498891778064,
      "learning_rate": 0.0001996935361906352,
      "loss": 18.3546,
      "step": 1240
    },
    {
      "epoch": 0.21700546448087432,
      "grad_norm": 1.6963646049559724,
      "learning_rate": 0.00019969242738932254,
      "loss": 17.78,
      "step": 1241
    },
    {
      "epoch": 0.21718032786885247,
      "grad_norm": 1.6222947638899172,
      "learning_rate": 0.00019969131658887062,
      "loss": 18.6278,
      "step": 1242
    },
    {
      "epoch": 0.2173551912568306,
      "grad_norm": 1.8808094999329035,
      "learning_rate": 0.0001996902037893017,
      "loss": 17.9585,
      "step": 1243
    },
    {
      "epoch": 0.21753005464480873,
      "grad_norm": 1.5961499912471295,
      "learning_rate": 0.00019968908899063815,
      "loss": 18.1102,
      "step": 1244
    },
    {
      "epoch": 0.21770491803278688,
      "grad_norm": 1.6002007985171625,
      "learning_rate": 0.00019968797219290226,
      "loss": 17.7449,
      "step": 1245
    },
    {
      "epoch": 0.21787978142076503,
      "grad_norm": 1.6372163978747887,
      "learning_rate": 0.00019968685339611644,
      "loss": 18.0406,
      "step": 1246
    },
    {
      "epoch": 0.21805464480874318,
      "grad_norm": 1.625844719776831,
      "learning_rate": 0.00019968573260030313,
      "loss": 18.0464,
      "step": 1247
    },
    {
      "epoch": 0.21822950819672132,
      "grad_norm": 1.5501578555263436,
      "learning_rate": 0.0001996846098054848,
      "loss": 18.1609,
      "step": 1248
    },
    {
      "epoch": 0.21840437158469944,
      "grad_norm": 1.4255748290808514,
      "learning_rate": 0.00019968348501168398,
      "loss": 17.9732,
      "step": 1249
    },
    {
      "epoch": 0.2185792349726776,
      "grad_norm": 1.5011485706041976,
      "learning_rate": 0.0001996823582189232,
      "loss": 18.1465,
      "step": 1250
    },
    {
      "epoch": 0.21875409836065574,
      "grad_norm": 1.5392909193688102,
      "learning_rate": 0.0001996812294272251,
      "loss": 17.5414,
      "step": 1251
    },
    {
      "epoch": 0.21892896174863388,
      "grad_norm": 1.4904901097349845,
      "learning_rate": 0.00019968009863661226,
      "loss": 18.0853,
      "step": 1252
    },
    {
      "epoch": 0.21910382513661203,
      "grad_norm": 1.4610435940034996,
      "learning_rate": 0.00019967896584710738,
      "loss": 17.9514,
      "step": 1253
    },
    {
      "epoch": 0.21927868852459018,
      "grad_norm": 1.4433305280378594,
      "learning_rate": 0.00019967783105873324,
      "loss": 17.9884,
      "step": 1254
    },
    {
      "epoch": 0.2194535519125683,
      "grad_norm": 1.4016723089037297,
      "learning_rate": 0.0001996766942715125,
      "loss": 17.7051,
      "step": 1255
    },
    {
      "epoch": 0.21962841530054644,
      "grad_norm": 1.4683704158807684,
      "learning_rate": 0.00019967555548546797,
      "loss": 17.8512,
      "step": 1256
    },
    {
      "epoch": 0.2198032786885246,
      "grad_norm": 1.6434725447324665,
      "learning_rate": 0.0001996744147006225,
      "loss": 18.2098,
      "step": 1257
    },
    {
      "epoch": 0.21997814207650274,
      "grad_norm": 1.3429536322409776,
      "learning_rate": 0.00019967327191699902,
      "loss": 18.107,
      "step": 1258
    },
    {
      "epoch": 0.22015300546448088,
      "grad_norm": 2.0936522742882646,
      "learning_rate": 0.00019967212713462037,
      "loss": 18.0581,
      "step": 1259
    },
    {
      "epoch": 0.22032786885245903,
      "grad_norm": 2.1975754100317686,
      "learning_rate": 0.00019967098035350954,
      "loss": 17.8859,
      "step": 1260
    },
    {
      "epoch": 0.22050273224043715,
      "grad_norm": 1.7728395591719854,
      "learning_rate": 0.00019966983157368958,
      "loss": 17.8041,
      "step": 1261
    },
    {
      "epoch": 0.2206775956284153,
      "grad_norm": 1.6763982859028652,
      "learning_rate": 0.0001996686807951834,
      "loss": 18.0079,
      "step": 1262
    },
    {
      "epoch": 0.22085245901639344,
      "grad_norm": 1.497441382472963,
      "learning_rate": 0.00019966752801801416,
      "loss": 17.9161,
      "step": 1263
    },
    {
      "epoch": 0.2210273224043716,
      "grad_norm": 1.6813692317222766,
      "learning_rate": 0.00019966637324220497,
      "loss": 18.0851,
      "step": 1264
    },
    {
      "epoch": 0.22120218579234974,
      "grad_norm": 1.667966601059015,
      "learning_rate": 0.00019966521646777897,
      "loss": 17.8451,
      "step": 1265
    },
    {
      "epoch": 0.22137704918032786,
      "grad_norm": 1.5361185448922825,
      "learning_rate": 0.0001996640576947594,
      "loss": 17.7248,
      "step": 1266
    },
    {
      "epoch": 0.221551912568306,
      "grad_norm": 1.6039727242085138,
      "learning_rate": 0.00019966289692316944,
      "loss": 17.7845,
      "step": 1267
    },
    {
      "epoch": 0.22172677595628415,
      "grad_norm": 1.6821504518055224,
      "learning_rate": 0.00019966173415303243,
      "loss": 17.5514,
      "step": 1268
    },
    {
      "epoch": 0.2219016393442623,
      "grad_norm": 1.5176710604950692,
      "learning_rate": 0.0001996605693843716,
      "loss": 17.7387,
      "step": 1269
    },
    {
      "epoch": 0.22207650273224044,
      "grad_norm": 1.725073131811503,
      "learning_rate": 0.00019965940261721036,
      "loss": 17.7923,
      "step": 1270
    },
    {
      "epoch": 0.2222513661202186,
      "grad_norm": 1.585279300216969,
      "learning_rate": 0.00019965823385157213,
      "loss": 17.5691,
      "step": 1271
    },
    {
      "epoch": 0.2224262295081967,
      "grad_norm": 1.4410561615988242,
      "learning_rate": 0.00019965706308748028,
      "loss": 17.0933,
      "step": 1272
    },
    {
      "epoch": 0.22260109289617486,
      "grad_norm": 1.5514079301174952,
      "learning_rate": 0.00019965589032495837,
      "loss": 17.9288,
      "step": 1273
    },
    {
      "epoch": 0.222775956284153,
      "grad_norm": 1.531853523370924,
      "learning_rate": 0.00019965471556402986,
      "loss": 17.7372,
      "step": 1274
    },
    {
      "epoch": 0.22295081967213115,
      "grad_norm": 1.725724156944193,
      "learning_rate": 0.0001996535388047183,
      "loss": 17.8046,
      "step": 1275
    },
    {
      "epoch": 0.2231256830601093,
      "grad_norm": 1.596554337010921,
      "learning_rate": 0.00019965236004704736,
      "loss": 17.2044,
      "step": 1276
    },
    {
      "epoch": 0.22330054644808744,
      "grad_norm": 1.5454538846757053,
      "learning_rate": 0.00019965117929104058,
      "loss": 17.8825,
      "step": 1277
    },
    {
      "epoch": 0.22347540983606556,
      "grad_norm": 1.4683080586972335,
      "learning_rate": 0.00019964999653672173,
      "loss": 17.4892,
      "step": 1278
    },
    {
      "epoch": 0.2236502732240437,
      "grad_norm": 1.6137296161435295,
      "learning_rate": 0.00019964881178411443,
      "loss": 17.9316,
      "step": 1279
    },
    {
      "epoch": 0.22382513661202186,
      "grad_norm": 1.4971214295145792,
      "learning_rate": 0.00019964762503324254,
      "loss": 17.827,
      "step": 1280
    },
    {
      "epoch": 0.224,
      "grad_norm": 1.513251759716374,
      "learning_rate": 0.0001996464362841298,
      "loss": 17.5643,
      "step": 1281
    },
    {
      "epoch": 0.22417486338797815,
      "grad_norm": 1.4181782025763916,
      "learning_rate": 0.00019964524553680004,
      "loss": 17.3549,
      "step": 1282
    },
    {
      "epoch": 0.22434972677595627,
      "grad_norm": 1.563357190865229,
      "learning_rate": 0.00019964405279127717,
      "loss": 17.6498,
      "step": 1283
    },
    {
      "epoch": 0.22452459016393442,
      "grad_norm": 1.5145275199221904,
      "learning_rate": 0.0001996428580475851,
      "loss": 17.3173,
      "step": 1284
    },
    {
      "epoch": 0.22469945355191256,
      "grad_norm": 1.6029822891479353,
      "learning_rate": 0.00019964166130574776,
      "loss": 17.7069,
      "step": 1285
    },
    {
      "epoch": 0.2248743169398907,
      "grad_norm": 1.516717943619749,
      "learning_rate": 0.00019964046256578917,
      "loss": 17.4642,
      "step": 1286
    },
    {
      "epoch": 0.22504918032786886,
      "grad_norm": 1.521348993875161,
      "learning_rate": 0.0001996392618277334,
      "loss": 17.5636,
      "step": 1287
    },
    {
      "epoch": 0.225224043715847,
      "grad_norm": 1.4926963770794843,
      "learning_rate": 0.00019963805909160444,
      "loss": 17.4448,
      "step": 1288
    },
    {
      "epoch": 0.22539890710382512,
      "grad_norm": 1.4106160987057728,
      "learning_rate": 0.00019963685435742652,
      "loss": 17.477,
      "step": 1289
    },
    {
      "epoch": 0.22557377049180327,
      "grad_norm": 1.5886902110972747,
      "learning_rate": 0.00019963564762522372,
      "loss": 17.49,
      "step": 1290
    },
    {
      "epoch": 0.22574863387978142,
      "grad_norm": 1.8106424807293813,
      "learning_rate": 0.00019963443889502028,
      "loss": 17.6615,
      "step": 1291
    },
    {
      "epoch": 0.22592349726775957,
      "grad_norm": 1.7740294568552262,
      "learning_rate": 0.00019963322816684043,
      "loss": 17.3393,
      "step": 1292
    },
    {
      "epoch": 0.2260983606557377,
      "grad_norm": 1.4403200557123494,
      "learning_rate": 0.0001996320154407084,
      "loss": 17.7242,
      "step": 1293
    },
    {
      "epoch": 0.22627322404371586,
      "grad_norm": 1.5162961316981471,
      "learning_rate": 0.0001996308007166486,
      "loss": 17.5068,
      "step": 1294
    },
    {
      "epoch": 0.22644808743169398,
      "grad_norm": 1.5583918847205738,
      "learning_rate": 0.0001996295839946853,
      "loss": 17.5229,
      "step": 1295
    },
    {
      "epoch": 0.22662295081967213,
      "grad_norm": 1.5471029471913433,
      "learning_rate": 0.00019962836527484296,
      "loss": 17.504,
      "step": 1296
    },
    {
      "epoch": 0.22679781420765027,
      "grad_norm": 1.5492434368683523,
      "learning_rate": 0.000199627144557146,
      "loss": 17.6199,
      "step": 1297
    },
    {
      "epoch": 0.22697267759562842,
      "grad_norm": 1.6166266586956815,
      "learning_rate": 0.00019962592184161887,
      "loss": 17.3257,
      "step": 1298
    },
    {
      "epoch": 0.22714754098360657,
      "grad_norm": 1.6884299567108247,
      "learning_rate": 0.00019962469712828614,
      "loss": 17.8271,
      "step": 1299
    },
    {
      "epoch": 0.2273224043715847,
      "grad_norm": 1.5065935035612412,
      "learning_rate": 0.00019962347041717232,
      "loss": 17.2096,
      "step": 1300
    },
    {
      "epoch": 0.22749726775956283,
      "grad_norm": 1.760346874663249,
      "learning_rate": 0.00019962224170830205,
      "loss": 17.556,
      "step": 1301
    },
    {
      "epoch": 0.22767213114754098,
      "grad_norm": 1.535391264604409,
      "learning_rate": 0.00019962101100169996,
      "loss": 17.6556,
      "step": 1302
    },
    {
      "epoch": 0.22784699453551913,
      "grad_norm": 1.5954680120783211,
      "learning_rate": 0.0001996197782973907,
      "loss": 17.3028,
      "step": 1303
    },
    {
      "epoch": 0.22802185792349727,
      "grad_norm": 1.4703203110300038,
      "learning_rate": 0.00019961854359539904,
      "loss": 17.2546,
      "step": 1304
    },
    {
      "epoch": 0.22819672131147542,
      "grad_norm": 1.6084903599162494,
      "learning_rate": 0.00019961730689574972,
      "loss": 17.6425,
      "step": 1305
    },
    {
      "epoch": 0.22837158469945354,
      "grad_norm": 1.47113866861294,
      "learning_rate": 0.0001996160681984675,
      "loss": 17.4958,
      "step": 1306
    },
    {
      "epoch": 0.2285464480874317,
      "grad_norm": 1.490961283851412,
      "learning_rate": 0.00019961482750357727,
      "loss": 17.2941,
      "step": 1307
    },
    {
      "epoch": 0.22872131147540983,
      "grad_norm": 1.4715537976591955,
      "learning_rate": 0.00019961358481110385,
      "loss": 17.4837,
      "step": 1308
    },
    {
      "epoch": 0.22889617486338798,
      "grad_norm": 1.6686007977804125,
      "learning_rate": 0.00019961234012107226,
      "loss": 17.4796,
      "step": 1309
    },
    {
      "epoch": 0.22907103825136613,
      "grad_norm": 1.4561090782334218,
      "learning_rate": 0.00019961109343350736,
      "loss": 17.6405,
      "step": 1310
    },
    {
      "epoch": 0.22924590163934427,
      "grad_norm": 1.5380027608509084,
      "learning_rate": 0.00019960984474843421,
      "loss": 17.2195,
      "step": 1311
    },
    {
      "epoch": 0.2294207650273224,
      "grad_norm": 1.4949066927855852,
      "learning_rate": 0.00019960859406587783,
      "loss": 17.369,
      "step": 1312
    },
    {
      "epoch": 0.22959562841530054,
      "grad_norm": 1.532716334772002,
      "learning_rate": 0.0001996073413858633,
      "loss": 17.1494,
      "step": 1313
    },
    {
      "epoch": 0.2297704918032787,
      "grad_norm": 1.4595830756909556,
      "learning_rate": 0.00019960608670841574,
      "loss": 17.3196,
      "step": 1314
    },
    {
      "epoch": 0.22994535519125683,
      "grad_norm": 1.333962222162595,
      "learning_rate": 0.00019960483003356028,
      "loss": 17.5105,
      "step": 1315
    },
    {
      "epoch": 0.23012021857923498,
      "grad_norm": 1.485629059851527,
      "learning_rate": 0.0001996035713613222,
      "loss": 17.31,
      "step": 1316
    },
    {
      "epoch": 0.23029508196721313,
      "grad_norm": 1.4292521288959719,
      "learning_rate": 0.0001996023106917267,
      "loss": 17.2308,
      "step": 1317
    },
    {
      "epoch": 0.23046994535519125,
      "grad_norm": 1.5091232494661873,
      "learning_rate": 0.000199601048024799,
      "loss": 17.3949,
      "step": 1318
    },
    {
      "epoch": 0.2306448087431694,
      "grad_norm": 1.5659035545024542,
      "learning_rate": 0.00019959978336056447,
      "loss": 17.338,
      "step": 1319
    },
    {
      "epoch": 0.23081967213114754,
      "grad_norm": 1.5148477091630672,
      "learning_rate": 0.0001995985166990485,
      "loss": 17.2083,
      "step": 1320
    },
    {
      "epoch": 0.2309945355191257,
      "grad_norm": 1.6247796794376257,
      "learning_rate": 0.00019959724804027648,
      "loss": 17.3326,
      "step": 1321
    },
    {
      "epoch": 0.23116939890710383,
      "grad_norm": 1.8028706489705364,
      "learning_rate": 0.0001995959773842738,
      "loss": 17.5329,
      "step": 1322
    },
    {
      "epoch": 0.23134426229508195,
      "grad_norm": 1.96442028458043,
      "learning_rate": 0.00019959470473106597,
      "loss": 17.1913,
      "step": 1323
    },
    {
      "epoch": 0.2315191256830601,
      "grad_norm": 1.9397384150022512,
      "learning_rate": 0.00019959343008067853,
      "loss": 17.1748,
      "step": 1324
    },
    {
      "epoch": 0.23169398907103825,
      "grad_norm": 1.543905071296274,
      "learning_rate": 0.00019959215343313703,
      "loss": 17.3592,
      "step": 1325
    },
    {
      "epoch": 0.2318688524590164,
      "grad_norm": 1.7125679757325758,
      "learning_rate": 0.00019959087478846707,
      "loss": 16.8106,
      "step": 1326
    },
    {
      "epoch": 0.23204371584699454,
      "grad_norm": 1.8448770151989788,
      "learning_rate": 0.00019958959414669424,
      "loss": 17.3811,
      "step": 1327
    },
    {
      "epoch": 0.2322185792349727,
      "grad_norm": 1.6624065738923515,
      "learning_rate": 0.00019958831150784431,
      "loss": 17.2738,
      "step": 1328
    },
    {
      "epoch": 0.2323934426229508,
      "grad_norm": 1.4981301383859775,
      "learning_rate": 0.00019958702687194294,
      "loss": 17.1319,
      "step": 1329
    },
    {
      "epoch": 0.23256830601092895,
      "grad_norm": 1.4427870120484432,
      "learning_rate": 0.00019958574023901592,
      "loss": 17.2908,
      "step": 1330
    },
    {
      "epoch": 0.2327431693989071,
      "grad_norm": 1.4078176559006874,
      "learning_rate": 0.00019958445160908904,
      "loss": 17.25,
      "step": 1331
    },
    {
      "epoch": 0.23291803278688525,
      "grad_norm": 1.4126107738978808,
      "learning_rate": 0.00019958316098218814,
      "loss": 17.0122,
      "step": 1332
    },
    {
      "epoch": 0.2330928961748634,
      "grad_norm": 1.4430636682970606,
      "learning_rate": 0.00019958186835833912,
      "loss": 16.7539,
      "step": 1333
    },
    {
      "epoch": 0.23326775956284154,
      "grad_norm": 1.486885123868685,
      "learning_rate": 0.00019958057373756785,
      "loss": 17.1035,
      "step": 1334
    },
    {
      "epoch": 0.23344262295081966,
      "grad_norm": 1.4909836865272228,
      "learning_rate": 0.00019957927711990035,
      "loss": 16.9643,
      "step": 1335
    },
    {
      "epoch": 0.2336174863387978,
      "grad_norm": 1.4361088055023525,
      "learning_rate": 0.00019957797850536258,
      "loss": 17.1784,
      "step": 1336
    },
    {
      "epoch": 0.23379234972677596,
      "grad_norm": 1.5721317937397017,
      "learning_rate": 0.0001995766778939806,
      "loss": 17.3093,
      "step": 1337
    },
    {
      "epoch": 0.2339672131147541,
      "grad_norm": 1.443099855791318,
      "learning_rate": 0.00019957537528578046,
      "loss": 16.8148,
      "step": 1338
    },
    {
      "epoch": 0.23414207650273225,
      "grad_norm": 1.415687427156258,
      "learning_rate": 0.00019957407068078834,
      "loss": 16.3041,
      "step": 1339
    },
    {
      "epoch": 0.2343169398907104,
      "grad_norm": 1.5269961663666891,
      "learning_rate": 0.00019957276407903037,
      "loss": 17.1088,
      "step": 1340
    },
    {
      "epoch": 0.23449180327868852,
      "grad_norm": 1.3998874474404526,
      "learning_rate": 0.00019957145548053274,
      "loss": 16.8605,
      "step": 1341
    },
    {
      "epoch": 0.23466666666666666,
      "grad_norm": 1.4396974750741027,
      "learning_rate": 0.00019957014488532172,
      "loss": 17.1838,
      "step": 1342
    },
    {
      "epoch": 0.2348415300546448,
      "grad_norm": 1.4615582702968322,
      "learning_rate": 0.0001995688322934236,
      "loss": 17.2527,
      "step": 1343
    },
    {
      "epoch": 0.23501639344262296,
      "grad_norm": 1.6709799012093718,
      "learning_rate": 0.00019956751770486462,
      "loss": 17.0287,
      "step": 1344
    },
    {
      "epoch": 0.2351912568306011,
      "grad_norm": 1.8678473162765776,
      "learning_rate": 0.00019956620111967122,
      "loss": 16.8805,
      "step": 1345
    },
    {
      "epoch": 0.23536612021857922,
      "grad_norm": 1.417141057340589,
      "learning_rate": 0.00019956488253786974,
      "loss": 17.1538,
      "step": 1346
    },
    {
      "epoch": 0.23554098360655737,
      "grad_norm": 1.6760824191417314,
      "learning_rate": 0.00019956356195948674,
      "loss": 16.948,
      "step": 1347
    },
    {
      "epoch": 0.23571584699453552,
      "grad_norm": 1.430420456197103,
      "learning_rate": 0.00019956223938454856,
      "loss": 16.9487,
      "step": 1348
    },
    {
      "epoch": 0.23589071038251366,
      "grad_norm": 1.5772214612477906,
      "learning_rate": 0.00019956091481308183,
      "loss": 17.427,
      "step": 1349
    },
    {
      "epoch": 0.2360655737704918,
      "grad_norm": 1.4280290427821376,
      "learning_rate": 0.00019955958824511303,
      "loss": 17.037,
      "step": 1350
    },
    {
      "epoch": 0.23624043715846996,
      "grad_norm": 1.5623524228244545,
      "learning_rate": 0.0001995582596806688,
      "loss": 17.0458,
      "step": 1351
    },
    {
      "epoch": 0.23641530054644808,
      "grad_norm": 1.6529516105313657,
      "learning_rate": 0.0001995569291197758,
      "loss": 17.3866,
      "step": 1352
    },
    {
      "epoch": 0.23659016393442622,
      "grad_norm": 1.4005693667094417,
      "learning_rate": 0.00019955559656246067,
      "loss": 16.8811,
      "step": 1353
    },
    {
      "epoch": 0.23676502732240437,
      "grad_norm": 1.535756455881919,
      "learning_rate": 0.00019955426200875018,
      "loss": 16.988,
      "step": 1354
    },
    {
      "epoch": 0.23693989071038252,
      "grad_norm": 1.3979744100243925,
      "learning_rate": 0.00019955292545867104,
      "loss": 16.8203,
      "step": 1355
    },
    {
      "epoch": 0.23711475409836066,
      "grad_norm": 1.446603198346712,
      "learning_rate": 0.0001995515869122501,
      "loss": 17.0074,
      "step": 1356
    },
    {
      "epoch": 0.2372896174863388,
      "grad_norm": 1.516462735927915,
      "learning_rate": 0.00019955024636951417,
      "loss": 16.9041,
      "step": 1357
    },
    {
      "epoch": 0.23746448087431693,
      "grad_norm": 1.4580627147317802,
      "learning_rate": 0.00019954890383049013,
      "loss": 16.8958,
      "step": 1358
    },
    {
      "epoch": 0.23763934426229508,
      "grad_norm": 1.5320480635510993,
      "learning_rate": 0.0001995475592952049,
      "loss": 16.876,
      "step": 1359
    },
    {
      "epoch": 0.23781420765027322,
      "grad_norm": 1.4619055842630784,
      "learning_rate": 0.00019954621276368548,
      "loss": 16.827,
      "step": 1360
    },
    {
      "epoch": 0.23798907103825137,
      "grad_norm": 1.5412224688520022,
      "learning_rate": 0.00019954486423595885,
      "loss": 16.7633,
      "step": 1361
    },
    {
      "epoch": 0.23816393442622952,
      "grad_norm": 1.5660088428670724,
      "learning_rate": 0.00019954351371205203,
      "loss": 16.7356,
      "step": 1362
    },
    {
      "epoch": 0.23833879781420764,
      "grad_norm": 1.490374258211001,
      "learning_rate": 0.0001995421611919921,
      "loss": 16.5865,
      "step": 1363
    },
    {
      "epoch": 0.23851366120218578,
      "grad_norm": 1.6186688100419346,
      "learning_rate": 0.00019954080667580626,
      "loss": 17.0455,
      "step": 1364
    },
    {
      "epoch": 0.23868852459016393,
      "grad_norm": 1.459891338017439,
      "learning_rate": 0.00019953945016352157,
      "loss": 16.557,
      "step": 1365
    },
    {
      "epoch": 0.23886338797814208,
      "grad_norm": 1.5314108002906377,
      "learning_rate": 0.0001995380916551653,
      "loss": 17.0864,
      "step": 1366
    },
    {
      "epoch": 0.23903825136612022,
      "grad_norm": 1.5257072464289017,
      "learning_rate": 0.00019953673115076468,
      "loss": 16.9139,
      "step": 1367
    },
    {
      "epoch": 0.23921311475409837,
      "grad_norm": 1.5146212693888934,
      "learning_rate": 0.00019953536865034696,
      "loss": 16.8872,
      "step": 1368
    },
    {
      "epoch": 0.2393879781420765,
      "grad_norm": 1.5783103382072963,
      "learning_rate": 0.0001995340041539395,
      "loss": 17.1613,
      "step": 1369
    },
    {
      "epoch": 0.23956284153005464,
      "grad_norm": 1.4314882118821417,
      "learning_rate": 0.00019953263766156964,
      "loss": 16.2099,
      "step": 1370
    },
    {
      "epoch": 0.23973770491803278,
      "grad_norm": 1.4709944399329216,
      "learning_rate": 0.00019953126917326478,
      "loss": 16.7962,
      "step": 1371
    },
    {
      "epoch": 0.23991256830601093,
      "grad_norm": 1.4807522157571542,
      "learning_rate": 0.0001995298986890524,
      "loss": 16.5572,
      "step": 1372
    },
    {
      "epoch": 0.24008743169398908,
      "grad_norm": 1.4827083893014115,
      "learning_rate": 0.00019952852620895996,
      "loss": 16.6574,
      "step": 1373
    },
    {
      "epoch": 0.24026229508196723,
      "grad_norm": 1.5116794277079648,
      "learning_rate": 0.00019952715173301492,
      "loss": 16.9768,
      "step": 1374
    },
    {
      "epoch": 0.24043715846994534,
      "grad_norm": 1.5886214848109932,
      "learning_rate": 0.00019952577526124494,
      "loss": 16.7657,
      "step": 1375
    },
    {
      "epoch": 0.2406120218579235,
      "grad_norm": 1.5036056752573626,
      "learning_rate": 0.00019952439679367758,
      "loss": 16.5799,
      "step": 1376
    },
    {
      "epoch": 0.24078688524590164,
      "grad_norm": 1.397939600969703,
      "learning_rate": 0.0001995230163303405,
      "loss": 16.6247,
      "step": 1377
    },
    {
      "epoch": 0.24096174863387979,
      "grad_norm": 1.3985231164118785,
      "learning_rate": 0.00019952163387126134,
      "loss": 17.0167,
      "step": 1378
    },
    {
      "epoch": 0.24113661202185793,
      "grad_norm": 1.539891580721835,
      "learning_rate": 0.00019952024941646785,
      "loss": 16.7053,
      "step": 1379
    },
    {
      "epoch": 0.24131147540983608,
      "grad_norm": 1.4253788823490199,
      "learning_rate": 0.0001995188629659878,
      "loss": 16.6225,
      "step": 1380
    },
    {
      "epoch": 0.2414863387978142,
      "grad_norm": 1.4658431270840049,
      "learning_rate": 0.00019951747451984897,
      "loss": 16.655,
      "step": 1381
    },
    {
      "epoch": 0.24166120218579235,
      "grad_norm": 1.4710032955432832,
      "learning_rate": 0.00019951608407807925,
      "loss": 16.763,
      "step": 1382
    },
    {
      "epoch": 0.2418360655737705,
      "grad_norm": 1.5275843632146995,
      "learning_rate": 0.00019951469164070646,
      "loss": 16.8514,
      "step": 1383
    },
    {
      "epoch": 0.24201092896174864,
      "grad_norm": 1.3957857959903677,
      "learning_rate": 0.00019951329720775855,
      "loss": 17.0068,
      "step": 1384
    },
    {
      "epoch": 0.2421857923497268,
      "grad_norm": 1.4047549023257189,
      "learning_rate": 0.0001995119007792635,
      "loss": 16.6511,
      "step": 1385
    },
    {
      "epoch": 0.2423606557377049,
      "grad_norm": 1.4035547756448619,
      "learning_rate": 0.0001995105023552493,
      "loss": 16.815,
      "step": 1386
    },
    {
      "epoch": 0.24253551912568305,
      "grad_norm": 1.3745830689416267,
      "learning_rate": 0.00019950910193574397,
      "loss": 16.4641,
      "step": 1387
    },
    {
      "epoch": 0.2427103825136612,
      "grad_norm": 1.4022753328578113,
      "learning_rate": 0.00019950769952077564,
      "loss": 16.5989,
      "step": 1388
    },
    {
      "epoch": 0.24288524590163935,
      "grad_norm": 1.3738725940047025,
      "learning_rate": 0.00019950629511037237,
      "loss": 16.9377,
      "step": 1389
    },
    {
      "epoch": 0.2430601092896175,
      "grad_norm": 1.6213267190910325,
      "learning_rate": 0.0001995048887045624,
      "loss": 16.6295,
      "step": 1390
    },
    {
      "epoch": 0.24323497267759564,
      "grad_norm": 1.442074198872693,
      "learning_rate": 0.00019950348030337386,
      "loss": 16.5912,
      "step": 1391
    },
    {
      "epoch": 0.24340983606557376,
      "grad_norm": 1.5968806511134406,
      "learning_rate": 0.00019950206990683507,
      "loss": 16.7561,
      "step": 1392
    },
    {
      "epoch": 0.2435846994535519,
      "grad_norm": 1.5039760188969662,
      "learning_rate": 0.00019950065751497421,
      "loss": 16.6633,
      "step": 1393
    },
    {
      "epoch": 0.24375956284153005,
      "grad_norm": 1.4298418593968016,
      "learning_rate": 0.0001994992431278197,
      "loss": 16.8661,
      "step": 1394
    },
    {
      "epoch": 0.2439344262295082,
      "grad_norm": 1.4336516535893324,
      "learning_rate": 0.00019949782674539982,
      "loss": 16.7622,
      "step": 1395
    },
    {
      "epoch": 0.24410928961748635,
      "grad_norm": 1.3923757873943996,
      "learning_rate": 0.00019949640836774307,
      "loss": 16.4289,
      "step": 1396
    },
    {
      "epoch": 0.2442841530054645,
      "grad_norm": 1.446985839276883,
      "learning_rate": 0.0001994949879948778,
      "loss": 16.3673,
      "step": 1397
    },
    {
      "epoch": 0.2444590163934426,
      "grad_norm": 1.406193735291883,
      "learning_rate": 0.00019949356562683256,
      "loss": 16.6166,
      "step": 1398
    },
    {
      "epoch": 0.24463387978142076,
      "grad_norm": 1.4955703127454136,
      "learning_rate": 0.00019949214126363583,
      "loss": 16.6198,
      "step": 1399
    },
    {
      "epoch": 0.2448087431693989,
      "grad_norm": 1.3923317951400072,
      "learning_rate": 0.00019949071490531616,
      "loss": 16.6196,
      "step": 1400
    },
    {
      "epoch": 0.24498360655737705,
      "grad_norm": 1.4214659013501212,
      "learning_rate": 0.0001994892865519022,
      "loss": 16.3978,
      "step": 1401
    },
    {
      "epoch": 0.2451584699453552,
      "grad_norm": 1.4243915685055428,
      "learning_rate": 0.0001994878562034226,
      "loss": 16.5371,
      "step": 1402
    },
    {
      "epoch": 0.24533333333333332,
      "grad_norm": 1.3515087258346936,
      "learning_rate": 0.00019948642385990596,
      "loss": 16.5282,
      "step": 1403
    },
    {
      "epoch": 0.24550819672131147,
      "grad_norm": 1.3499138511633448,
      "learning_rate": 0.0001994849895213811,
      "loss": 16.6236,
      "step": 1404
    },
    {
      "epoch": 0.2456830601092896,
      "grad_norm": 1.4508050393426823,
      "learning_rate": 0.0001994835531878767,
      "loss": 16.2087,
      "step": 1405
    },
    {
      "epoch": 0.24585792349726776,
      "grad_norm": 1.5794867695207997,
      "learning_rate": 0.00019948211485942166,
      "loss": 16.9677,
      "step": 1406
    },
    {
      "epoch": 0.2460327868852459,
      "grad_norm": 1.3962977340987932,
      "learning_rate": 0.00019948067453604473,
      "loss": 16.5138,
      "step": 1407
    },
    {
      "epoch": 0.24620765027322405,
      "grad_norm": 1.411061745115642,
      "learning_rate": 0.00019947923221777484,
      "loss": 16.3915,
      "step": 1408
    },
    {
      "epoch": 0.24638251366120217,
      "grad_norm": 1.4252958627960042,
      "learning_rate": 0.0001994777879046409,
      "loss": 16.5835,
      "step": 1409
    },
    {
      "epoch": 0.24655737704918032,
      "grad_norm": 1.4537849855261207,
      "learning_rate": 0.00019947634159667187,
      "loss": 16.4444,
      "step": 1410
    },
    {
      "epoch": 0.24673224043715847,
      "grad_norm": 1.5035430143572446,
      "learning_rate": 0.00019947489329389677,
      "loss": 16.4453,
      "step": 1411
    },
    {
      "epoch": 0.24690710382513661,
      "grad_norm": 1.452773540740262,
      "learning_rate": 0.00019947344299634464,
      "loss": 16.6061,
      "step": 1412
    },
    {
      "epoch": 0.24708196721311476,
      "grad_norm": 1.6977579218866305,
      "learning_rate": 0.00019947199070404455,
      "loss": 16.4171,
      "step": 1413
    },
    {
      "epoch": 0.2472568306010929,
      "grad_norm": 1.5040773726223995,
      "learning_rate": 0.0001994705364170256,
      "loss": 16.597,
      "step": 1414
    },
    {
      "epoch": 0.24743169398907103,
      "grad_norm": 1.4547884845835044,
      "learning_rate": 0.000199469080135317,
      "loss": 16.5814,
      "step": 1415
    },
    {
      "epoch": 0.24760655737704917,
      "grad_norm": 1.5412071182767848,
      "learning_rate": 0.00019946762185894793,
      "loss": 16.5002,
      "step": 1416
    },
    {
      "epoch": 0.24778142076502732,
      "grad_norm": 1.4273980171564526,
      "learning_rate": 0.00019946616158794764,
      "loss": 16.7242,
      "step": 1417
    },
    {
      "epoch": 0.24795628415300547,
      "grad_norm": 1.583978862433937,
      "learning_rate": 0.00019946469932234542,
      "loss": 16.5936,
      "step": 1418
    },
    {
      "epoch": 0.24813114754098362,
      "grad_norm": 1.450570796130311,
      "learning_rate": 0.00019946323506217056,
      "loss": 16.651,
      "step": 1419
    },
    {
      "epoch": 0.24830601092896176,
      "grad_norm": 1.5753634890710009,
      "learning_rate": 0.00019946176880745244,
      "loss": 16.4396,
      "step": 1420
    },
    {
      "epoch": 0.24848087431693988,
      "grad_norm": 1.6257625135372968,
      "learning_rate": 0.0001994603005582205,
      "loss": 16.6132,
      "step": 1421
    },
    {
      "epoch": 0.24865573770491803,
      "grad_norm": 1.6281697126611518,
      "learning_rate": 0.00019945883031450412,
      "loss": 16.5417,
      "step": 1422
    },
    {
      "epoch": 0.24883060109289618,
      "grad_norm": 1.7408747496459687,
      "learning_rate": 0.00019945735807633282,
      "loss": 16.6427,
      "step": 1423
    },
    {
      "epoch": 0.24900546448087432,
      "grad_norm": 1.7176444581801578,
      "learning_rate": 0.00019945588384373612,
      "loss": 16.2699,
      "step": 1424
    },
    {
      "epoch": 0.24918032786885247,
      "grad_norm": 1.5116750584266565,
      "learning_rate": 0.00019945440761674359,
      "loss": 16.5916,
      "step": 1425
    },
    {
      "epoch": 0.2493551912568306,
      "grad_norm": 1.5398947663619245,
      "learning_rate": 0.0001994529293953848,
      "loss": 16.6969,
      "step": 1426
    },
    {
      "epoch": 0.24953005464480874,
      "grad_norm": 1.361323658391138,
      "learning_rate": 0.00019945144917968944,
      "loss": 16.8073,
      "step": 1427
    },
    {
      "epoch": 0.24970491803278688,
      "grad_norm": 1.5750874980995826,
      "learning_rate": 0.00019944996696968716,
      "loss": 16.2587,
      "step": 1428
    },
    {
      "epoch": 0.24987978142076503,
      "grad_norm": 1.3998113724618504,
      "learning_rate": 0.0001994484827654077,
      "loss": 16.3463,
      "step": 1429
    },
    {
      "epoch": 0.2500546448087432,
      "grad_norm": 1.39007310993594,
      "learning_rate": 0.00019944699656688076,
      "loss": 16.1514,
      "step": 1430
    },
    {
      "epoch": 0.2502295081967213,
      "grad_norm": 1.4119254875542757,
      "learning_rate": 0.00019944550837413627,
      "loss": 16.7489,
      "step": 1431
    },
    {
      "epoch": 0.25040437158469947,
      "grad_norm": 1.7721603122663623,
      "learning_rate": 0.00019944401818720395,
      "loss": 16.6073,
      "step": 1432
    },
    {
      "epoch": 0.2505792349726776,
      "grad_norm": 1.4017927898541294,
      "learning_rate": 0.00019944252600611375,
      "loss": 16.5039,
      "step": 1433
    },
    {
      "epoch": 0.25075409836065576,
      "grad_norm": 1.5402583578132987,
      "learning_rate": 0.00019944103183089558,
      "loss": 16.2575,
      "step": 1434
    },
    {
      "epoch": 0.2509289617486339,
      "grad_norm": 1.4322747091913215,
      "learning_rate": 0.00019943953566157937,
      "loss": 16.4137,
      "step": 1435
    },
    {
      "epoch": 0.251103825136612,
      "grad_norm": 1.603753402713574,
      "learning_rate": 0.0001994380374981952,
      "loss": 16.3978,
      "step": 1436
    },
    {
      "epoch": 0.2512786885245902,
      "grad_norm": 1.36091679882279,
      "learning_rate": 0.00019943653734077303,
      "loss": 16.3547,
      "step": 1437
    },
    {
      "epoch": 0.2514535519125683,
      "grad_norm": 1.4808478517557324,
      "learning_rate": 0.00019943503518934298,
      "loss": 16.1956,
      "step": 1438
    },
    {
      "epoch": 0.25162841530054647,
      "grad_norm": 1.4886827485653622,
      "learning_rate": 0.00019943353104393517,
      "loss": 16.2719,
      "step": 1439
    },
    {
      "epoch": 0.2518032786885246,
      "grad_norm": 1.378473810353744,
      "learning_rate": 0.00019943202490457974,
      "loss": 16.2545,
      "step": 1440
    },
    {
      "epoch": 0.2519781420765027,
      "grad_norm": 1.4345341302878465,
      "learning_rate": 0.00019943051677130696,
      "loss": 15.9395,
      "step": 1441
    },
    {
      "epoch": 0.2521530054644809,
      "grad_norm": 1.3401014118953045,
      "learning_rate": 0.00019942900664414697,
      "loss": 16.3178,
      "step": 1442
    },
    {
      "epoch": 0.252327868852459,
      "grad_norm": 1.4636254446227561,
      "learning_rate": 0.00019942749452313017,
      "loss": 16.4572,
      "step": 1443
    },
    {
      "epoch": 0.2525027322404372,
      "grad_norm": 1.4064522494588934,
      "learning_rate": 0.0001994259804082868,
      "loss": 16.3431,
      "step": 1444
    },
    {
      "epoch": 0.2526775956284153,
      "grad_norm": 1.398696692256123,
      "learning_rate": 0.00019942446429964723,
      "loss": 16.0907,
      "step": 1445
    },
    {
      "epoch": 0.2528524590163934,
      "grad_norm": 1.4080415386255787,
      "learning_rate": 0.00019942294619724188,
      "loss": 16.6402,
      "step": 1446
    },
    {
      "epoch": 0.2530273224043716,
      "grad_norm": 1.4441812553621867,
      "learning_rate": 0.00019942142610110118,
      "loss": 16.5037,
      "step": 1447
    },
    {
      "epoch": 0.2532021857923497,
      "grad_norm": 1.4011383122542316,
      "learning_rate": 0.00019941990401125566,
      "loss": 16.1223,
      "step": 1448
    },
    {
      "epoch": 0.2533770491803279,
      "grad_norm": 1.4673824074939845,
      "learning_rate": 0.00019941837992773576,
      "loss": 16.2197,
      "step": 1449
    },
    {
      "epoch": 0.253551912568306,
      "grad_norm": 1.423161502374066,
      "learning_rate": 0.00019941685385057212,
      "loss": 16.2086,
      "step": 1450
    },
    {
      "epoch": 0.2537267759562842,
      "grad_norm": 1.448723176045208,
      "learning_rate": 0.00019941532577979533,
      "loss": 16.3017,
      "step": 1451
    },
    {
      "epoch": 0.2539016393442623,
      "grad_norm": 1.5617117010503279,
      "learning_rate": 0.00019941379571543596,
      "loss": 16.344,
      "step": 1452
    },
    {
      "epoch": 0.2540765027322404,
      "grad_norm": 1.390536753057732,
      "learning_rate": 0.00019941226365752476,
      "loss": 16.3247,
      "step": 1453
    },
    {
      "epoch": 0.2542513661202186,
      "grad_norm": 1.4307317265237303,
      "learning_rate": 0.0001994107296060925,
      "loss": 16.1741,
      "step": 1454
    },
    {
      "epoch": 0.2544262295081967,
      "grad_norm": 1.4514199551907223,
      "learning_rate": 0.0001994091935611698,
      "loss": 16.1164,
      "step": 1455
    },
    {
      "epoch": 0.2546010928961749,
      "grad_norm": 1.399041159791189,
      "learning_rate": 0.00019940765552278758,
      "loss": 16.1415,
      "step": 1456
    },
    {
      "epoch": 0.254775956284153,
      "grad_norm": 1.605114333472007,
      "learning_rate": 0.00019940611549097663,
      "loss": 16.2288,
      "step": 1457
    },
    {
      "epoch": 0.2549508196721311,
      "grad_norm": 1.3177085543194407,
      "learning_rate": 0.00019940457346576785,
      "loss": 16.1418,
      "step": 1458
    },
    {
      "epoch": 0.2551256830601093,
      "grad_norm": 1.4280048085241506,
      "learning_rate": 0.00019940302944719215,
      "loss": 16.1602,
      "step": 1459
    },
    {
      "epoch": 0.2553005464480874,
      "grad_norm": 1.4872348662276549,
      "learning_rate": 0.00019940148343528056,
      "loss": 16.3747,
      "step": 1460
    },
    {
      "epoch": 0.2554754098360656,
      "grad_norm": 1.4082153642020885,
      "learning_rate": 0.00019939993543006395,
      "loss": 15.9495,
      "step": 1461
    },
    {
      "epoch": 0.2556502732240437,
      "grad_norm": 1.3192131332313677,
      "learning_rate": 0.00019939838543157348,
      "loss": 16.1491,
      "step": 1462
    },
    {
      "epoch": 0.25582513661202183,
      "grad_norm": 1.4535355199256195,
      "learning_rate": 0.00019939683343984016,
      "loss": 16.2176,
      "step": 1463
    },
    {
      "epoch": 0.256,
      "grad_norm": 1.3396863140071873,
      "learning_rate": 0.00019939527945489518,
      "loss": 16.4396,
      "step": 1464
    },
    {
      "epoch": 0.2561748633879781,
      "grad_norm": 1.348573124469627,
      "learning_rate": 0.00019939372347676964,
      "loss": 16.4669,
      "step": 1465
    },
    {
      "epoch": 0.2563497267759563,
      "grad_norm": 1.3749175482812361,
      "learning_rate": 0.00019939216550549476,
      "loss": 16.4904,
      "step": 1466
    },
    {
      "epoch": 0.2565245901639344,
      "grad_norm": 1.3234354029057405,
      "learning_rate": 0.0001993906055411018,
      "loss": 15.87,
      "step": 1467
    },
    {
      "epoch": 0.2566994535519126,
      "grad_norm": 1.3147738612068538,
      "learning_rate": 0.00019938904358362202,
      "loss": 16.1693,
      "step": 1468
    },
    {
      "epoch": 0.2568743169398907,
      "grad_norm": 1.300110291761299,
      "learning_rate": 0.00019938747963308678,
      "loss": 16.5135,
      "step": 1469
    },
    {
      "epoch": 0.25704918032786883,
      "grad_norm": 1.3572207230022295,
      "learning_rate": 0.0001993859136895274,
      "loss": 15.9788,
      "step": 1470
    },
    {
      "epoch": 0.257224043715847,
      "grad_norm": 1.4290309028853396,
      "learning_rate": 0.00019938434575297526,
      "loss": 16.0725,
      "step": 1471
    },
    {
      "epoch": 0.2573989071038251,
      "grad_norm": 1.2521539991028086,
      "learning_rate": 0.00019938277582346188,
      "loss": 15.9043,
      "step": 1472
    },
    {
      "epoch": 0.2575737704918033,
      "grad_norm": 1.317954551734004,
      "learning_rate": 0.00019938120390101868,
      "loss": 15.8965,
      "step": 1473
    },
    {
      "epoch": 0.2577486338797814,
      "grad_norm": 1.238954542956187,
      "learning_rate": 0.0001993796299856772,
      "loss": 16.1898,
      "step": 1474
    },
    {
      "epoch": 0.25792349726775954,
      "grad_norm": 1.360518794392187,
      "learning_rate": 0.000199378054077469,
      "loss": 16.0008,
      "step": 1475
    },
    {
      "epoch": 0.2580983606557377,
      "grad_norm": 1.3858567590679574,
      "learning_rate": 0.0001993764761764257,
      "loss": 16.1224,
      "step": 1476
    },
    {
      "epoch": 0.25827322404371583,
      "grad_norm": 1.2997503084491024,
      "learning_rate": 0.0001993748962825789,
      "loss": 16.0867,
      "step": 1477
    },
    {
      "epoch": 0.258448087431694,
      "grad_norm": 1.6635983245001524,
      "learning_rate": 0.00019937331439596033,
      "loss": 16.0371,
      "step": 1478
    },
    {
      "epoch": 0.2586229508196721,
      "grad_norm": 1.437325660385664,
      "learning_rate": 0.00019937173051660172,
      "loss": 15.9941,
      "step": 1479
    },
    {
      "epoch": 0.2587978142076503,
      "grad_norm": 1.5715012284754932,
      "learning_rate": 0.00019937014464453475,
      "loss": 16.1067,
      "step": 1480
    },
    {
      "epoch": 0.2589726775956284,
      "grad_norm": 1.5340996927225319,
      "learning_rate": 0.0001993685567797913,
      "loss": 15.8096,
      "step": 1481
    },
    {
      "epoch": 0.25914754098360654,
      "grad_norm": 1.4788719621096202,
      "learning_rate": 0.00019936696692240318,
      "loss": 16.3087,
      "step": 1482
    },
    {
      "epoch": 0.2593224043715847,
      "grad_norm": 1.3135199983770116,
      "learning_rate": 0.00019936537507240227,
      "loss": 16.1511,
      "step": 1483
    },
    {
      "epoch": 0.25949726775956283,
      "grad_norm": 1.376645449461234,
      "learning_rate": 0.00019936378122982048,
      "loss": 16.2403,
      "step": 1484
    },
    {
      "epoch": 0.259672131147541,
      "grad_norm": 1.6292639961854478,
      "learning_rate": 0.00019936218539468983,
      "loss": 16.2273,
      "step": 1485
    },
    {
      "epoch": 0.2598469945355191,
      "grad_norm": 1.4607756270919152,
      "learning_rate": 0.00019936058756704226,
      "loss": 16.1736,
      "step": 1486
    },
    {
      "epoch": 0.26002185792349725,
      "grad_norm": 1.395463433907819,
      "learning_rate": 0.00019935898774690983,
      "loss": 15.9778,
      "step": 1487
    },
    {
      "epoch": 0.2601967213114754,
      "grad_norm": 1.3549512694013497,
      "learning_rate": 0.00019935738593432464,
      "loss": 16.2326,
      "step": 1488
    },
    {
      "epoch": 0.26037158469945354,
      "grad_norm": 1.4366352956515245,
      "learning_rate": 0.00019935578212931877,
      "loss": 15.9159,
      "step": 1489
    },
    {
      "epoch": 0.2605464480874317,
      "grad_norm": 1.3859825920898734,
      "learning_rate": 0.00019935417633192443,
      "loss": 16.164,
      "step": 1490
    },
    {
      "epoch": 0.26072131147540983,
      "grad_norm": 1.4207040487522933,
      "learning_rate": 0.00019935256854217381,
      "loss": 15.997,
      "step": 1491
    },
    {
      "epoch": 0.26089617486338795,
      "grad_norm": 1.4563535972445354,
      "learning_rate": 0.0001993509587600991,
      "loss": 16.1048,
      "step": 1492
    },
    {
      "epoch": 0.2610710382513661,
      "grad_norm": 1.4985579885451739,
      "learning_rate": 0.00019934934698573263,
      "loss": 16.038,
      "step": 1493
    },
    {
      "epoch": 0.26124590163934425,
      "grad_norm": 1.4820770709011784,
      "learning_rate": 0.00019934773321910671,
      "loss": 16.2351,
      "step": 1494
    },
    {
      "epoch": 0.2614207650273224,
      "grad_norm": 1.3631234880307348,
      "learning_rate": 0.00019934611746025372,
      "loss": 15.9503,
      "step": 1495
    },
    {
      "epoch": 0.26159562841530054,
      "grad_norm": 1.4305835327286514,
      "learning_rate": 0.00019934449970920604,
      "loss": 16.1211,
      "step": 1496
    },
    {
      "epoch": 0.2617704918032787,
      "grad_norm": 1.4318510838002276,
      "learning_rate": 0.00019934287996599607,
      "loss": 15.9177,
      "step": 1497
    },
    {
      "epoch": 0.26194535519125683,
      "grad_norm": 1.3500140026936427,
      "learning_rate": 0.00019934125823065638,
      "loss": 16.0183,
      "step": 1498
    },
    {
      "epoch": 0.26212021857923495,
      "grad_norm": 1.3751339950632093,
      "learning_rate": 0.00019933963450321945,
      "loss": 15.9105,
      "step": 1499
    },
    {
      "epoch": 0.26229508196721313,
      "grad_norm": 1.3729185788833598,
      "learning_rate": 0.00019933800878371783,
      "loss": 16.1053,
      "step": 1500
    },
    {
      "epoch": 0.26246994535519125,
      "grad_norm": 1.3497145329942677,
      "learning_rate": 0.00019933638107218408,
      "loss": 16.4379,
      "step": 1501
    },
    {
      "epoch": 0.2626448087431694,
      "grad_norm": 1.2869798564147354,
      "learning_rate": 0.00019933475136865093,
      "loss": 15.9753,
      "step": 1502
    },
    {
      "epoch": 0.26281967213114754,
      "grad_norm": 1.3739094776909255,
      "learning_rate": 0.000199333119673151,
      "loss": 16.216,
      "step": 1503
    },
    {
      "epoch": 0.26299453551912566,
      "grad_norm": 1.4436668334132958,
      "learning_rate": 0.00019933148598571702,
      "loss": 16.326,
      "step": 1504
    },
    {
      "epoch": 0.26316939890710384,
      "grad_norm": 1.2686064598012732,
      "learning_rate": 0.0001993298503063818,
      "loss": 15.8449,
      "step": 1505
    },
    {
      "epoch": 0.26334426229508195,
      "grad_norm": 1.2378752182232646,
      "learning_rate": 0.00019932821263517805,
      "loss": 15.6542,
      "step": 1506
    },
    {
      "epoch": 0.26351912568306013,
      "grad_norm": 1.3258510379884365,
      "learning_rate": 0.0001993265729721387,
      "loss": 16.0338,
      "step": 1507
    },
    {
      "epoch": 0.26369398907103825,
      "grad_norm": 1.3496559739319018,
      "learning_rate": 0.00019932493131729655,
      "loss": 15.7092,
      "step": 1508
    },
    {
      "epoch": 0.26386885245901637,
      "grad_norm": 1.2882558621466313,
      "learning_rate": 0.00019932328767068458,
      "loss": 16.0605,
      "step": 1509
    },
    {
      "epoch": 0.26404371584699454,
      "grad_norm": 1.234351500590412,
      "learning_rate": 0.00019932164203233574,
      "loss": 16.0146,
      "step": 1510
    },
    {
      "epoch": 0.26421857923497266,
      "grad_norm": 1.3455710485125176,
      "learning_rate": 0.000199319994402283,
      "loss": 15.9075,
      "step": 1511
    },
    {
      "epoch": 0.26439344262295084,
      "grad_norm": 1.4476067260961911,
      "learning_rate": 0.00019931834478055944,
      "loss": 15.9923,
      "step": 1512
    },
    {
      "epoch": 0.26456830601092896,
      "grad_norm": 1.4185332809941917,
      "learning_rate": 0.0001993166931671981,
      "loss": 16.1178,
      "step": 1513
    },
    {
      "epoch": 0.26474316939890713,
      "grad_norm": 1.417698975854757,
      "learning_rate": 0.0001993150395622321,
      "loss": 15.4684,
      "step": 1514
    },
    {
      "epoch": 0.26491803278688525,
      "grad_norm": 1.310916149776771,
      "learning_rate": 0.00019931338396569467,
      "loss": 16.1556,
      "step": 1515
    },
    {
      "epoch": 0.26509289617486337,
      "grad_norm": 1.396582858460126,
      "learning_rate": 0.00019931172637761895,
      "loss": 15.8932,
      "step": 1516
    },
    {
      "epoch": 0.26526775956284154,
      "grad_norm": 1.5096242148504846,
      "learning_rate": 0.0001993100667980382,
      "loss": 16.0344,
      "step": 1517
    },
    {
      "epoch": 0.26544262295081966,
      "grad_norm": 1.488688347034736,
      "learning_rate": 0.00019930840522698563,
      "loss": 16.0147,
      "step": 1518
    },
    {
      "epoch": 0.26561748633879784,
      "grad_norm": 1.5173132575409076,
      "learning_rate": 0.00019930674166449466,
      "loss": 15.6314,
      "step": 1519
    },
    {
      "epoch": 0.26579234972677596,
      "grad_norm": 1.31656371052008,
      "learning_rate": 0.0001993050761105986,
      "loss": 15.8136,
      "step": 1520
    },
    {
      "epoch": 0.2659672131147541,
      "grad_norm": 1.5394458505551598,
      "learning_rate": 0.00019930340856533089,
      "loss": 15.8439,
      "step": 1521
    },
    {
      "epoch": 0.26614207650273225,
      "grad_norm": 1.2441773684391508,
      "learning_rate": 0.0001993017390287249,
      "loss": 15.6818,
      "step": 1522
    },
    {
      "epoch": 0.26631693989071037,
      "grad_norm": 1.524414815419542,
      "learning_rate": 0.00019930006750081415,
      "loss": 15.7396,
      "step": 1523
    },
    {
      "epoch": 0.26649180327868854,
      "grad_norm": 1.4416118279800503,
      "learning_rate": 0.00019929839398163214,
      "loss": 16.1356,
      "step": 1524
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 1.4331890130023355,
      "learning_rate": 0.00019929671847121244,
      "loss": 15.81,
      "step": 1525
    },
    {
      "epoch": 0.2668415300546448,
      "grad_norm": 1.4130811063530302,
      "learning_rate": 0.0001992950409695887,
      "loss": 15.8198,
      "step": 1526
    },
    {
      "epoch": 0.26701639344262296,
      "grad_norm": 1.2724754115875112,
      "learning_rate": 0.00019929336147679448,
      "loss": 15.8929,
      "step": 1527
    },
    {
      "epoch": 0.2671912568306011,
      "grad_norm": 1.3071179290977548,
      "learning_rate": 0.0001992916799928635,
      "loss": 16.0062,
      "step": 1528
    },
    {
      "epoch": 0.26736612021857925,
      "grad_norm": 1.3126479120037544,
      "learning_rate": 0.00019928999651782947,
      "loss": 15.3902,
      "step": 1529
    },
    {
      "epoch": 0.26754098360655737,
      "grad_norm": 1.388286748000811,
      "learning_rate": 0.00019928831105172614,
      "loss": 15.7528,
      "step": 1530
    },
    {
      "epoch": 0.26771584699453554,
      "grad_norm": 1.256065215652202,
      "learning_rate": 0.00019928662359458733,
      "loss": 15.7813,
      "step": 1531
    },
    {
      "epoch": 0.26789071038251366,
      "grad_norm": 1.3257765380326862,
      "learning_rate": 0.00019928493414644687,
      "loss": 15.4948,
      "step": 1532
    },
    {
      "epoch": 0.2680655737704918,
      "grad_norm": 1.2642765535310212,
      "learning_rate": 0.00019928324270733862,
      "loss": 15.8031,
      "step": 1533
    },
    {
      "epoch": 0.26824043715846996,
      "grad_norm": 1.5049292655547437,
      "learning_rate": 0.00019928154927729654,
      "loss": 16.0271,
      "step": 1534
    },
    {
      "epoch": 0.2684153005464481,
      "grad_norm": 1.3715123905355033,
      "learning_rate": 0.00019927985385635456,
      "loss": 15.7102,
      "step": 1535
    },
    {
      "epoch": 0.26859016393442625,
      "grad_norm": 1.37862288630126,
      "learning_rate": 0.00019927815644454663,
      "loss": 15.8336,
      "step": 1536
    },
    {
      "epoch": 0.26876502732240437,
      "grad_norm": 1.3901151331853427,
      "learning_rate": 0.0001992764570419069,
      "loss": 15.6622,
      "step": 1537
    },
    {
      "epoch": 0.2689398907103825,
      "grad_norm": 1.3987957734087966,
      "learning_rate": 0.00019927475564846935,
      "loss": 16.1251,
      "step": 1538
    },
    {
      "epoch": 0.26911475409836066,
      "grad_norm": 1.2802634438329414,
      "learning_rate": 0.00019927305226426816,
      "loss": 15.8015,
      "step": 1539
    },
    {
      "epoch": 0.2692896174863388,
      "grad_norm": 1.335018293007372,
      "learning_rate": 0.00019927134688933746,
      "loss": 15.6013,
      "step": 1540
    },
    {
      "epoch": 0.26946448087431696,
      "grad_norm": 1.3195724110624223,
      "learning_rate": 0.00019926963952371143,
      "loss": 15.7907,
      "step": 1541
    },
    {
      "epoch": 0.2696393442622951,
      "grad_norm": 1.3110564991379756,
      "learning_rate": 0.00019926793016742435,
      "loss": 15.4235,
      "step": 1542
    },
    {
      "epoch": 0.2698142076502732,
      "grad_norm": 1.3055205487630137,
      "learning_rate": 0.00019926621882051046,
      "loss": 15.7717,
      "step": 1543
    },
    {
      "epoch": 0.26998907103825137,
      "grad_norm": 1.262955471593066,
      "learning_rate": 0.00019926450548300414,
      "loss": 15.58,
      "step": 1544
    },
    {
      "epoch": 0.2701639344262295,
      "grad_norm": 1.2368435103440254,
      "learning_rate": 0.00019926279015493965,
      "loss": 15.5439,
      "step": 1545
    },
    {
      "epoch": 0.27033879781420767,
      "grad_norm": 1.2610178162125965,
      "learning_rate": 0.00019926107283635148,
      "loss": 15.7255,
      "step": 1546
    },
    {
      "epoch": 0.2705136612021858,
      "grad_norm": 1.3159097412987129,
      "learning_rate": 0.000199259353527274,
      "loss": 15.7485,
      "step": 1547
    },
    {
      "epoch": 0.27068852459016396,
      "grad_norm": 1.4887325300017353,
      "learning_rate": 0.00019925763222774172,
      "loss": 15.6448,
      "step": 1548
    },
    {
      "epoch": 0.2708633879781421,
      "grad_norm": 1.3378545844535248,
      "learning_rate": 0.00019925590893778917,
      "loss": 15.8563,
      "step": 1549
    },
    {
      "epoch": 0.2710382513661202,
      "grad_norm": 1.2980084904088727,
      "learning_rate": 0.00019925418365745085,
      "loss": 15.6834,
      "step": 1550
    },
    {
      "epoch": 0.27121311475409837,
      "grad_norm": 1.421096570713836,
      "learning_rate": 0.00019925245638676144,
      "loss": 15.8023,
      "step": 1551
    },
    {
      "epoch": 0.2713879781420765,
      "grad_norm": 1.2941133269787555,
      "learning_rate": 0.0001992507271257555,
      "loss": 15.7768,
      "step": 1552
    },
    {
      "epoch": 0.27156284153005467,
      "grad_norm": 1.3808861901923597,
      "learning_rate": 0.00019924899587446778,
      "loss": 15.5966,
      "step": 1553
    },
    {
      "epoch": 0.2717377049180328,
      "grad_norm": 1.3154089856802813,
      "learning_rate": 0.00019924726263293295,
      "loss": 15.6293,
      "step": 1554
    },
    {
      "epoch": 0.2719125683060109,
      "grad_norm": 1.434030738688493,
      "learning_rate": 0.00019924552740118576,
      "loss": 15.8053,
      "step": 1555
    },
    {
      "epoch": 0.2720874316939891,
      "grad_norm": 1.5758368477732954,
      "learning_rate": 0.00019924379017926102,
      "loss": 15.9638,
      "step": 1556
    },
    {
      "epoch": 0.2722622950819672,
      "grad_norm": 1.2870135460081988,
      "learning_rate": 0.0001992420509671936,
      "loss": 15.9845,
      "step": 1557
    },
    {
      "epoch": 0.2724371584699454,
      "grad_norm": 1.4552936975931379,
      "learning_rate": 0.00019924030976501832,
      "loss": 15.6322,
      "step": 1558
    },
    {
      "epoch": 0.2726120218579235,
      "grad_norm": 1.5228871261288157,
      "learning_rate": 0.0001992385665727701,
      "loss": 15.5188,
      "step": 1559
    },
    {
      "epoch": 0.27278688524590167,
      "grad_norm": 1.2889715173624918,
      "learning_rate": 0.00019923682139048396,
      "loss": 16.1338,
      "step": 1560
    },
    {
      "epoch": 0.2729617486338798,
      "grad_norm": 1.3206500491946878,
      "learning_rate": 0.0001992350742181948,
      "loss": 15.6206,
      "step": 1561
    },
    {
      "epoch": 0.2731366120218579,
      "grad_norm": 1.273722424815893,
      "learning_rate": 0.00019923332505593778,
      "loss": 15.8033,
      "step": 1562
    },
    {
      "epoch": 0.2733114754098361,
      "grad_norm": 1.320983601547608,
      "learning_rate": 0.00019923157390374784,
      "loss": 15.4539,
      "step": 1563
    },
    {
      "epoch": 0.2734863387978142,
      "grad_norm": 1.3549923800002073,
      "learning_rate": 0.0001992298207616602,
      "loss": 15.8718,
      "step": 1564
    },
    {
      "epoch": 0.2736612021857924,
      "grad_norm": 1.2658728584772654,
      "learning_rate": 0.00019922806562971,
      "loss": 15.6401,
      "step": 1565
    },
    {
      "epoch": 0.2738360655737705,
      "grad_norm": 1.4000943545577156,
      "learning_rate": 0.00019922630850793236,
      "loss": 15.7019,
      "step": 1566
    },
    {
      "epoch": 0.2740109289617486,
      "grad_norm": 1.3076654811044344,
      "learning_rate": 0.00019922454939636262,
      "loss": 15.7693,
      "step": 1567
    },
    {
      "epoch": 0.2741857923497268,
      "grad_norm": 1.28034741206925,
      "learning_rate": 0.00019922278829503595,
      "loss": 15.5877,
      "step": 1568
    },
    {
      "epoch": 0.2743606557377049,
      "grad_norm": 1.3869627515222023,
      "learning_rate": 0.00019922102520398776,
      "loss": 15.4847,
      "step": 1569
    },
    {
      "epoch": 0.2745355191256831,
      "grad_norm": 1.2607470122633075,
      "learning_rate": 0.00019921926012325333,
      "loss": 15.4084,
      "step": 1570
    },
    {
      "epoch": 0.2747103825136612,
      "grad_norm": 1.3098375105112179,
      "learning_rate": 0.00019921749305286813,
      "loss": 15.7848,
      "step": 1571
    },
    {
      "epoch": 0.2748852459016393,
      "grad_norm": 1.27943176466393,
      "learning_rate": 0.00019921572399286754,
      "loss": 15.8228,
      "step": 1572
    },
    {
      "epoch": 0.2750601092896175,
      "grad_norm": 1.260192070594701,
      "learning_rate": 0.00019921395294328708,
      "loss": 15.5371,
      "step": 1573
    },
    {
      "epoch": 0.2752349726775956,
      "grad_norm": 1.3545149980592688,
      "learning_rate": 0.0001992121799041622,
      "loss": 15.5827,
      "step": 1574
    },
    {
      "epoch": 0.2754098360655738,
      "grad_norm": 1.4450481811972147,
      "learning_rate": 0.00019921040487552851,
      "loss": 15.7558,
      "step": 1575
    },
    {
      "epoch": 0.2755846994535519,
      "grad_norm": 1.392323791701436,
      "learning_rate": 0.00019920862785742158,
      "loss": 15.6521,
      "step": 1576
    },
    {
      "epoch": 0.2757595628415301,
      "grad_norm": 1.344755998806779,
      "learning_rate": 0.00019920684884987706,
      "loss": 15.4207,
      "step": 1577
    },
    {
      "epoch": 0.2759344262295082,
      "grad_norm": 1.2598075534719946,
      "learning_rate": 0.0001992050678529306,
      "loss": 15.384,
      "step": 1578
    },
    {
      "epoch": 0.2761092896174863,
      "grad_norm": 1.3004230332283466,
      "learning_rate": 0.00019920328486661793,
      "loss": 15.6913,
      "step": 1579
    },
    {
      "epoch": 0.2762841530054645,
      "grad_norm": 1.4230393207878662,
      "learning_rate": 0.0001992014998909748,
      "loss": 15.5848,
      "step": 1580
    },
    {
      "epoch": 0.2764590163934426,
      "grad_norm": 1.3223435820071203,
      "learning_rate": 0.00019919971292603703,
      "loss": 15.4017,
      "step": 1581
    },
    {
      "epoch": 0.2766338797814208,
      "grad_norm": 1.4434206089468167,
      "learning_rate": 0.00019919792397184041,
      "loss": 15.5286,
      "step": 1582
    },
    {
      "epoch": 0.2768087431693989,
      "grad_norm": 1.3531616225517973,
      "learning_rate": 0.00019919613302842086,
      "loss": 15.5415,
      "step": 1583
    },
    {
      "epoch": 0.276983606557377,
      "grad_norm": 1.4063394047810736,
      "learning_rate": 0.00019919434009581428,
      "loss": 15.6966,
      "step": 1584
    },
    {
      "epoch": 0.2771584699453552,
      "grad_norm": 1.4962547021383883,
      "learning_rate": 0.00019919254517405656,
      "loss": 15.4604,
      "step": 1585
    },
    {
      "epoch": 0.2773333333333333,
      "grad_norm": 1.3631492405051415,
      "learning_rate": 0.0001991907482631838,
      "loss": 15.3381,
      "step": 1586
    },
    {
      "epoch": 0.2775081967213115,
      "grad_norm": 1.3812239918649052,
      "learning_rate": 0.00019918894936323195,
      "loss": 15.6307,
      "step": 1587
    },
    {
      "epoch": 0.2776830601092896,
      "grad_norm": 1.4174290691723614,
      "learning_rate": 0.00019918714847423714,
      "loss": 15.5005,
      "step": 1588
    },
    {
      "epoch": 0.27785792349726773,
      "grad_norm": 1.3102450871543876,
      "learning_rate": 0.00019918534559623544,
      "loss": 15.56,
      "step": 1589
    },
    {
      "epoch": 0.2780327868852459,
      "grad_norm": 1.3635406084289985,
      "learning_rate": 0.00019918354072926304,
      "loss": 15.7613,
      "step": 1590
    },
    {
      "epoch": 0.278207650273224,
      "grad_norm": 1.3016737566311176,
      "learning_rate": 0.0001991817338733561,
      "loss": 15.6387,
      "step": 1591
    },
    {
      "epoch": 0.2783825136612022,
      "grad_norm": 1.347182859153052,
      "learning_rate": 0.00019917992502855087,
      "loss": 15.6698,
      "step": 1592
    },
    {
      "epoch": 0.2785573770491803,
      "grad_norm": 1.2692307245702645,
      "learning_rate": 0.00019917811419488363,
      "loss": 15.506,
      "step": 1593
    },
    {
      "epoch": 0.2787322404371585,
      "grad_norm": 1.5146946042943867,
      "learning_rate": 0.00019917630137239068,
      "loss": 15.6257,
      "step": 1594
    },
    {
      "epoch": 0.2789071038251366,
      "grad_norm": 1.2658199100026337,
      "learning_rate": 0.00019917448656110837,
      "loss": 15.4306,
      "step": 1595
    },
    {
      "epoch": 0.27908196721311473,
      "grad_norm": 1.560126809605502,
      "learning_rate": 0.00019917266976107308,
      "loss": 15.3115,
      "step": 1596
    },
    {
      "epoch": 0.2792568306010929,
      "grad_norm": 1.2975777310647665,
      "learning_rate": 0.00019917085097232131,
      "loss": 15.7686,
      "step": 1597
    },
    {
      "epoch": 0.27943169398907103,
      "grad_norm": 1.4709482602666892,
      "learning_rate": 0.00019916903019488947,
      "loss": 15.7993,
      "step": 1598
    },
    {
      "epoch": 0.2796065573770492,
      "grad_norm": 1.2711723264452857,
      "learning_rate": 0.00019916720742881405,
      "loss": 15.5373,
      "step": 1599
    },
    {
      "epoch": 0.2797814207650273,
      "grad_norm": 1.3162771489025995,
      "learning_rate": 0.00019916538267413168,
      "loss": 15.679,
      "step": 1600
    },
    {
      "epoch": 0.27995628415300544,
      "grad_norm": 1.3596308969889892,
      "learning_rate": 0.00019916355593087886,
      "loss": 15.4965,
      "step": 1601
    },
    {
      "epoch": 0.2801311475409836,
      "grad_norm": 1.3821112722964646,
      "learning_rate": 0.0001991617271990923,
      "loss": 15.3995,
      "step": 1602
    },
    {
      "epoch": 0.28030601092896174,
      "grad_norm": 1.4292389673447619,
      "learning_rate": 0.00019915989647880866,
      "loss": 15.5563,
      "step": 1603
    },
    {
      "epoch": 0.2804808743169399,
      "grad_norm": 1.262583373650931,
      "learning_rate": 0.00019915806377006462,
      "loss": 15.5781,
      "step": 1604
    },
    {
      "epoch": 0.28065573770491803,
      "grad_norm": 1.3484360305886192,
      "learning_rate": 0.00019915622907289694,
      "loss": 15.7471,
      "step": 1605
    },
    {
      "epoch": 0.28083060109289615,
      "grad_norm": 1.3263610792485354,
      "learning_rate": 0.00019915439238734242,
      "loss": 15.5978,
      "step": 1606
    },
    {
      "epoch": 0.2810054644808743,
      "grad_norm": 1.319927124539139,
      "learning_rate": 0.00019915255371343791,
      "loss": 15.3482,
      "step": 1607
    },
    {
      "epoch": 0.28118032786885244,
      "grad_norm": 1.313792867405522,
      "learning_rate": 0.00019915071305122024,
      "loss": 15.61,
      "step": 1608
    },
    {
      "epoch": 0.2813551912568306,
      "grad_norm": 1.282027935046418,
      "learning_rate": 0.00019914887040072633,
      "loss": 15.5506,
      "step": 1609
    },
    {
      "epoch": 0.28153005464480874,
      "grad_norm": 1.361850643713849,
      "learning_rate": 0.00019914702576199315,
      "loss": 15.3016,
      "step": 1610
    },
    {
      "epoch": 0.2817049180327869,
      "grad_norm": 1.2317174142758123,
      "learning_rate": 0.0001991451791350577,
      "loss": 15.661,
      "step": 1611
    },
    {
      "epoch": 0.28187978142076503,
      "grad_norm": 1.317169404436471,
      "learning_rate": 0.00019914333051995699,
      "loss": 15.3098,
      "step": 1612
    },
    {
      "epoch": 0.28205464480874315,
      "grad_norm": 1.519317656775145,
      "learning_rate": 0.00019914147991672806,
      "loss": 15.2557,
      "step": 1613
    },
    {
      "epoch": 0.2822295081967213,
      "grad_norm": 1.3049180913294185,
      "learning_rate": 0.00019913962732540807,
      "loss": 15.49,
      "step": 1614
    },
    {
      "epoch": 0.28240437158469944,
      "grad_norm": 1.311000278015808,
      "learning_rate": 0.00019913777274603418,
      "loss": 15.4332,
      "step": 1615
    },
    {
      "epoch": 0.2825792349726776,
      "grad_norm": 1.4309351359348226,
      "learning_rate": 0.0001991359161786435,
      "loss": 15.5873,
      "step": 1616
    },
    {
      "epoch": 0.28275409836065574,
      "grad_norm": 1.2489479215193935,
      "learning_rate": 0.00019913405762327338,
      "loss": 15.2718,
      "step": 1617
    },
    {
      "epoch": 0.28292896174863386,
      "grad_norm": 1.2421476794156114,
      "learning_rate": 0.00019913219707996097,
      "loss": 15.5424,
      "step": 1618
    },
    {
      "epoch": 0.28310382513661203,
      "grad_norm": 1.4292012443014566,
      "learning_rate": 0.00019913033454874366,
      "loss": 15.4145,
      "step": 1619
    },
    {
      "epoch": 0.28327868852459015,
      "grad_norm": 1.3017356187451048,
      "learning_rate": 0.00019912847002965874,
      "loss": 15.4027,
      "step": 1620
    },
    {
      "epoch": 0.2834535519125683,
      "grad_norm": 1.546292257688183,
      "learning_rate": 0.00019912660352274365,
      "loss": 15.6031,
      "step": 1621
    },
    {
      "epoch": 0.28362841530054644,
      "grad_norm": 1.2639934115734022,
      "learning_rate": 0.00019912473502803582,
      "loss": 15.3595,
      "step": 1622
    },
    {
      "epoch": 0.28380327868852456,
      "grad_norm": 1.4220258448072984,
      "learning_rate": 0.00019912286454557267,
      "loss": 15.5398,
      "step": 1623
    },
    {
      "epoch": 0.28397814207650274,
      "grad_norm": 1.3215677140732258,
      "learning_rate": 0.00019912099207539176,
      "loss": 15.3496,
      "step": 1624
    },
    {
      "epoch": 0.28415300546448086,
      "grad_norm": 1.2280329124847553,
      "learning_rate": 0.0001991191176175306,
      "loss": 15.704,
      "step": 1625
    },
    {
      "epoch": 0.28432786885245903,
      "grad_norm": 1.3817631702007227,
      "learning_rate": 0.00019911724117202683,
      "loss": 15.7469,
      "step": 1626
    },
    {
      "epoch": 0.28450273224043715,
      "grad_norm": 1.2398331918406336,
      "learning_rate": 0.000199115362738918,
      "loss": 15.6471,
      "step": 1627
    },
    {
      "epoch": 0.2846775956284153,
      "grad_norm": 1.2587922790460768,
      "learning_rate": 0.00019911348231824188,
      "loss": 15.5558,
      "step": 1628
    },
    {
      "epoch": 0.28485245901639344,
      "grad_norm": 1.2695106424560842,
      "learning_rate": 0.00019911159991003606,
      "loss": 15.5373,
      "step": 1629
    },
    {
      "epoch": 0.28502732240437156,
      "grad_norm": 1.2911110908030095,
      "learning_rate": 0.0001991097155143384,
      "loss": 15.3308,
      "step": 1630
    },
    {
      "epoch": 0.28520218579234974,
      "grad_norm": 1.3414799515298517,
      "learning_rate": 0.0001991078291311866,
      "loss": 15.0247,
      "step": 1631
    },
    {
      "epoch": 0.28537704918032786,
      "grad_norm": 1.3400849263227608,
      "learning_rate": 0.00019910594076061853,
      "loss": 15.505,
      "step": 1632
    },
    {
      "epoch": 0.28555191256830603,
      "grad_norm": 1.3428019885896139,
      "learning_rate": 0.00019910405040267208,
      "loss": 15.3743,
      "step": 1633
    },
    {
      "epoch": 0.28572677595628415,
      "grad_norm": 1.3659551183312733,
      "learning_rate": 0.0001991021580573851,
      "loss": 15.1704,
      "step": 1634
    },
    {
      "epoch": 0.28590163934426227,
      "grad_norm": 1.3380937834171052,
      "learning_rate": 0.00019910026372479557,
      "loss": 15.7186,
      "step": 1635
    },
    {
      "epoch": 0.28607650273224045,
      "grad_norm": 1.3098001757686957,
      "learning_rate": 0.0001990983674049415,
      "loss": 15.3332,
      "step": 1636
    },
    {
      "epoch": 0.28625136612021856,
      "grad_norm": 1.1551027260115259,
      "learning_rate": 0.00019909646909786085,
      "loss": 15.3549,
      "step": 1637
    },
    {
      "epoch": 0.28642622950819674,
      "grad_norm": 1.3555882641285912,
      "learning_rate": 0.00019909456880359177,
      "loss": 15.3138,
      "step": 1638
    },
    {
      "epoch": 0.28660109289617486,
      "grad_norm": 1.3630813310732224,
      "learning_rate": 0.0001990926665221723,
      "loss": 15.2917,
      "step": 1639
    },
    {
      "epoch": 0.28677595628415303,
      "grad_norm": 1.3099544691973908,
      "learning_rate": 0.0001990907622536406,
      "loss": 15.117,
      "step": 1640
    },
    {
      "epoch": 0.28695081967213115,
      "grad_norm": 1.298508561193531,
      "learning_rate": 0.0001990888559980349,
      "loss": 15.0646,
      "step": 1641
    },
    {
      "epoch": 0.28712568306010927,
      "grad_norm": 1.2689353495188778,
      "learning_rate": 0.00019908694775539338,
      "loss": 15.3613,
      "step": 1642
    },
    {
      "epoch": 0.28730054644808745,
      "grad_norm": 1.2945890008873346,
      "learning_rate": 0.00019908503752575434,
      "loss": 15.2375,
      "step": 1643
    },
    {
      "epoch": 0.28747540983606557,
      "grad_norm": 1.2614124233028552,
      "learning_rate": 0.00019908312530915603,
      "loss": 15.3492,
      "step": 1644
    },
    {
      "epoch": 0.28765027322404374,
      "grad_norm": 1.3527914712307822,
      "learning_rate": 0.00019908121110563686,
      "loss": 15.0967,
      "step": 1645
    },
    {
      "epoch": 0.28782513661202186,
      "grad_norm": 1.3436682449993032,
      "learning_rate": 0.00019907929491523517,
      "loss": 15.7473,
      "step": 1646
    },
    {
      "epoch": 0.288,
      "grad_norm": 1.282402911934903,
      "learning_rate": 0.00019907737673798944,
      "loss": 15.379,
      "step": 1647
    },
    {
      "epoch": 0.28817486338797815,
      "grad_norm": 1.2984314384912197,
      "learning_rate": 0.00019907545657393805,
      "loss": 14.9956,
      "step": 1648
    },
    {
      "epoch": 0.28834972677595627,
      "grad_norm": 1.2808238311324944,
      "learning_rate": 0.00019907353442311958,
      "loss": 14.8806,
      "step": 1649
    },
    {
      "epoch": 0.28852459016393445,
      "grad_norm": 1.282651001748413,
      "learning_rate": 0.00019907161028557253,
      "loss": 15.4869,
      "step": 1650
    },
    {
      "epoch": 0.28869945355191257,
      "grad_norm": 1.2682846870556181,
      "learning_rate": 0.00019906968416133554,
      "loss": 15.4927,
      "step": 1651
    },
    {
      "epoch": 0.2888743169398907,
      "grad_norm": 1.3541528268654057,
      "learning_rate": 0.00019906775605044717,
      "loss": 15.5148,
      "step": 1652
    },
    {
      "epoch": 0.28904918032786886,
      "grad_norm": 1.2408480788085092,
      "learning_rate": 0.00019906582595294613,
      "loss": 15.5282,
      "step": 1653
    },
    {
      "epoch": 0.289224043715847,
      "grad_norm": 1.3112051367220865,
      "learning_rate": 0.0001990638938688711,
      "loss": 15.1561,
      "step": 1654
    },
    {
      "epoch": 0.28939890710382515,
      "grad_norm": 1.3327790502441712,
      "learning_rate": 0.0001990619597982608,
      "loss": 15.323,
      "step": 1655
    },
    {
      "epoch": 0.2895737704918033,
      "grad_norm": 1.3000732039425076,
      "learning_rate": 0.0001990600237411541,
      "loss": 15.368,
      "step": 1656
    },
    {
      "epoch": 0.28974863387978145,
      "grad_norm": 1.1786294628941267,
      "learning_rate": 0.00019905808569758973,
      "loss": 15.0586,
      "step": 1657
    },
    {
      "epoch": 0.28992349726775957,
      "grad_norm": 1.2897530227418208,
      "learning_rate": 0.0001990561456676066,
      "loss": 15.3926,
      "step": 1658
    },
    {
      "epoch": 0.2900983606557377,
      "grad_norm": 1.280976006685602,
      "learning_rate": 0.00019905420365124362,
      "loss": 15.6183,
      "step": 1659
    },
    {
      "epoch": 0.29027322404371586,
      "grad_norm": 1.19879174477667,
      "learning_rate": 0.00019905225964853972,
      "loss": 15.1666,
      "step": 1660
    },
    {
      "epoch": 0.290448087431694,
      "grad_norm": 1.3180351004459676,
      "learning_rate": 0.0001990503136595339,
      "loss": 14.9473,
      "step": 1661
    },
    {
      "epoch": 0.29062295081967215,
      "grad_norm": 1.2185361740291534,
      "learning_rate": 0.00019904836568426513,
      "loss": 15.219,
      "step": 1662
    },
    {
      "epoch": 0.2907978142076503,
      "grad_norm": 1.2425081913262597,
      "learning_rate": 0.00019904641572277254,
      "loss": 15.0453,
      "step": 1663
    },
    {
      "epoch": 0.2909726775956284,
      "grad_norm": 1.4027261839999843,
      "learning_rate": 0.00019904446377509518,
      "loss": 15.4633,
      "step": 1664
    },
    {
      "epoch": 0.29114754098360657,
      "grad_norm": 1.3322757021980836,
      "learning_rate": 0.00019904250984127222,
      "loss": 15.6252,
      "step": 1665
    },
    {
      "epoch": 0.2913224043715847,
      "grad_norm": 1.2448524262549143,
      "learning_rate": 0.00019904055392134284,
      "loss": 15.3246,
      "step": 1666
    },
    {
      "epoch": 0.29149726775956286,
      "grad_norm": 1.3145492780721553,
      "learning_rate": 0.00019903859601534628,
      "loss": 15.4843,
      "step": 1667
    },
    {
      "epoch": 0.291672131147541,
      "grad_norm": 1.3721543702511019,
      "learning_rate": 0.00019903663612332175,
      "loss": 15.4718,
      "step": 1668
    },
    {
      "epoch": 0.2918469945355191,
      "grad_norm": 1.3431052113432473,
      "learning_rate": 0.0001990346742453086,
      "loss": 15.0768,
      "step": 1669
    },
    {
      "epoch": 0.2920218579234973,
      "grad_norm": 1.1856506277425687,
      "learning_rate": 0.00019903271038134614,
      "loss": 15.368,
      "step": 1670
    },
    {
      "epoch": 0.2921967213114754,
      "grad_norm": 1.2097199865285624,
      "learning_rate": 0.00019903074453147377,
      "loss": 15.0861,
      "step": 1671
    },
    {
      "epoch": 0.29237158469945357,
      "grad_norm": 1.350561845013831,
      "learning_rate": 0.00019902877669573092,
      "loss": 15.3054,
      "step": 1672
    },
    {
      "epoch": 0.2925464480874317,
      "grad_norm": 1.2925944670503835,
      "learning_rate": 0.00019902680687415705,
      "loss": 15.2326,
      "step": 1673
    },
    {
      "epoch": 0.29272131147540986,
      "grad_norm": 1.292926367084816,
      "learning_rate": 0.00019902483506679163,
      "loss": 15.2025,
      "step": 1674
    },
    {
      "epoch": 0.292896174863388,
      "grad_norm": 1.302772150427414,
      "learning_rate": 0.0001990228612736742,
      "loss": 15.1231,
      "step": 1675
    },
    {
      "epoch": 0.2930710382513661,
      "grad_norm": 1.357147331416414,
      "learning_rate": 0.00019902088549484442,
      "loss": 14.9938,
      "step": 1676
    },
    {
      "epoch": 0.2932459016393443,
      "grad_norm": 1.236021248824734,
      "learning_rate": 0.0001990189077303418,
      "loss": 15.3995,
      "step": 1677
    },
    {
      "epoch": 0.2934207650273224,
      "grad_norm": 1.2111633637690307,
      "learning_rate": 0.00019901692798020604,
      "loss": 15.367,
      "step": 1678
    },
    {
      "epoch": 0.29359562841530057,
      "grad_norm": 1.2917165359371146,
      "learning_rate": 0.0001990149462444769,
      "loss": 15.1937,
      "step": 1679
    },
    {
      "epoch": 0.2937704918032787,
      "grad_norm": 1.2066767526766173,
      "learning_rate": 0.00019901296252319405,
      "loss": 14.9791,
      "step": 1680
    },
    {
      "epoch": 0.2939453551912568,
      "grad_norm": 1.304235480516952,
      "learning_rate": 0.0001990109768163973,
      "loss": 15.5771,
      "step": 1681
    },
    {
      "epoch": 0.294120218579235,
      "grad_norm": 1.2975072511674335,
      "learning_rate": 0.00019900898912412647,
      "loss": 15.2739,
      "step": 1682
    },
    {
      "epoch": 0.2942950819672131,
      "grad_norm": 1.3951988918581852,
      "learning_rate": 0.00019900699944642137,
      "loss": 15.3904,
      "step": 1683
    },
    {
      "epoch": 0.2944699453551913,
      "grad_norm": 1.287647196883572,
      "learning_rate": 0.000199005007783322,
      "loss": 15.3816,
      "step": 1684
    },
    {
      "epoch": 0.2946448087431694,
      "grad_norm": 1.2335776651707753,
      "learning_rate": 0.0001990030141348682,
      "loss": 15.2553,
      "step": 1685
    },
    {
      "epoch": 0.2948196721311475,
      "grad_norm": 1.2074420025724446,
      "learning_rate": 0.00019900101850109999,
      "loss": 15.0037,
      "step": 1686
    },
    {
      "epoch": 0.2949945355191257,
      "grad_norm": 1.3140559237898006,
      "learning_rate": 0.0001989990208820574,
      "loss": 14.9572,
      "step": 1687
    },
    {
      "epoch": 0.2951693989071038,
      "grad_norm": 1.2558298141216249,
      "learning_rate": 0.00019899702127778045,
      "loss": 15.3162,
      "step": 1688
    },
    {
      "epoch": 0.295344262295082,
      "grad_norm": 1.2410111970208733,
      "learning_rate": 0.0001989950196883093,
      "loss": 15.1947,
      "step": 1689
    },
    {
      "epoch": 0.2955191256830601,
      "grad_norm": 1.2823569236737085,
      "learning_rate": 0.00019899301611368402,
      "loss": 15.3924,
      "step": 1690
    },
    {
      "epoch": 0.2956939890710383,
      "grad_norm": 1.4052429413597223,
      "learning_rate": 0.0001989910105539448,
      "loss": 15.4355,
      "step": 1691
    },
    {
      "epoch": 0.2958688524590164,
      "grad_norm": 1.30044150161245,
      "learning_rate": 0.00019898900300913193,
      "loss": 14.916,
      "step": 1692
    },
    {
      "epoch": 0.2960437158469945,
      "grad_norm": 1.225181346079453,
      "learning_rate": 0.00019898699347928558,
      "loss": 15.154,
      "step": 1693
    },
    {
      "epoch": 0.2962185792349727,
      "grad_norm": 1.2600360198123444,
      "learning_rate": 0.0001989849819644461,
      "loss": 15.1815,
      "step": 1694
    },
    {
      "epoch": 0.2963934426229508,
      "grad_norm": 1.2442994936748781,
      "learning_rate": 0.00019898296846465377,
      "loss": 15.0278,
      "step": 1695
    },
    {
      "epoch": 0.296568306010929,
      "grad_norm": 1.2773466735110115,
      "learning_rate": 0.00019898095297994902,
      "loss": 15.2568,
      "step": 1696
    },
    {
      "epoch": 0.2967431693989071,
      "grad_norm": 1.2652479354465187,
      "learning_rate": 0.00019897893551037226,
      "loss": 15.2907,
      "step": 1697
    },
    {
      "epoch": 0.2969180327868852,
      "grad_norm": 1.4119903442834911,
      "learning_rate": 0.00019897691605596393,
      "loss": 15.3683,
      "step": 1698
    },
    {
      "epoch": 0.2970928961748634,
      "grad_norm": 1.2818995055469182,
      "learning_rate": 0.00019897489461676452,
      "loss": 15.1854,
      "step": 1699
    },
    {
      "epoch": 0.2972677595628415,
      "grad_norm": 1.3100531840079535,
      "learning_rate": 0.00019897287119281462,
      "loss": 15.1874,
      "step": 1700
    },
    {
      "epoch": 0.2974426229508197,
      "grad_norm": 1.2312959527130205,
      "learning_rate": 0.00019897084578415476,
      "loss": 15.2855,
      "step": 1701
    },
    {
      "epoch": 0.2976174863387978,
      "grad_norm": 1.3901945994936213,
      "learning_rate": 0.00019896881839082556,
      "loss": 15.5879,
      "step": 1702
    },
    {
      "epoch": 0.29779234972677593,
      "grad_norm": 1.2692506243958348,
      "learning_rate": 0.00019896678901286765,
      "loss": 15.4812,
      "step": 1703
    },
    {
      "epoch": 0.2979672131147541,
      "grad_norm": 1.196156313277884,
      "learning_rate": 0.00019896475765032175,
      "loss": 15.1549,
      "step": 1704
    },
    {
      "epoch": 0.2981420765027322,
      "grad_norm": 1.2835417919742091,
      "learning_rate": 0.00019896272430322865,
      "loss": 15.066,
      "step": 1705
    },
    {
      "epoch": 0.2983169398907104,
      "grad_norm": 1.196090368843067,
      "learning_rate": 0.00019896068897162902,
      "loss": 15.255,
      "step": 1706
    },
    {
      "epoch": 0.2984918032786885,
      "grad_norm": 1.2982803032399892,
      "learning_rate": 0.00019895865165556377,
      "loss": 15.1913,
      "step": 1707
    },
    {
      "epoch": 0.2986666666666667,
      "grad_norm": 1.2786372109336825,
      "learning_rate": 0.00019895661235507365,
      "loss": 15.1508,
      "step": 1708
    },
    {
      "epoch": 0.2988415300546448,
      "grad_norm": 1.2917015739834696,
      "learning_rate": 0.00019895457107019967,
      "loss": 15.1534,
      "step": 1709
    },
    {
      "epoch": 0.29901639344262293,
      "grad_norm": 1.252760555563385,
      "learning_rate": 0.00019895252780098268,
      "loss": 15.1043,
      "step": 1710
    },
    {
      "epoch": 0.2991912568306011,
      "grad_norm": 1.355776209936991,
      "learning_rate": 0.0001989504825474637,
      "loss": 15.1729,
      "step": 1711
    },
    {
      "epoch": 0.2993661202185792,
      "grad_norm": 1.346709122528613,
      "learning_rate": 0.00019894843530968373,
      "loss": 15.5374,
      "step": 1712
    },
    {
      "epoch": 0.2995409836065574,
      "grad_norm": 1.1645169572475094,
      "learning_rate": 0.0001989463860876838,
      "loss": 15.1013,
      "step": 1713
    },
    {
      "epoch": 0.2997158469945355,
      "grad_norm": 1.2390384634567009,
      "learning_rate": 0.00019894433488150504,
      "loss": 15.1841,
      "step": 1714
    },
    {
      "epoch": 0.29989071038251364,
      "grad_norm": 1.200142551366145,
      "learning_rate": 0.00019894228169118857,
      "loss": 14.8654,
      "step": 1715
    },
    {
      "epoch": 0.3000655737704918,
      "grad_norm": 1.1747902158481611,
      "learning_rate": 0.00019894022651677554,
      "loss": 15.1118,
      "step": 1716
    },
    {
      "epoch": 0.30024043715846993,
      "grad_norm": 1.273738203421593,
      "learning_rate": 0.00019893816935830718,
      "loss": 15.0865,
      "step": 1717
    },
    {
      "epoch": 0.3004153005464481,
      "grad_norm": 1.2103067398448513,
      "learning_rate": 0.00019893611021582476,
      "loss": 14.9977,
      "step": 1718
    },
    {
      "epoch": 0.3005901639344262,
      "grad_norm": 1.2685836075647379,
      "learning_rate": 0.00019893404908936954,
      "loss": 15.0637,
      "step": 1719
    },
    {
      "epoch": 0.3007650273224044,
      "grad_norm": 1.2487302615664617,
      "learning_rate": 0.0001989319859789829,
      "loss": 15.2148,
      "step": 1720
    },
    {
      "epoch": 0.3009398907103825,
      "grad_norm": 1.20996641597275,
      "learning_rate": 0.00019892992088470617,
      "loss": 15.2235,
      "step": 1721
    },
    {
      "epoch": 0.30111475409836064,
      "grad_norm": 1.2141663343669835,
      "learning_rate": 0.00019892785380658078,
      "loss": 15.3029,
      "step": 1722
    },
    {
      "epoch": 0.3012896174863388,
      "grad_norm": 1.1235661611292167,
      "learning_rate": 0.00019892578474464814,
      "loss": 15.0022,
      "step": 1723
    },
    {
      "epoch": 0.30146448087431693,
      "grad_norm": 1.2654026661325124,
      "learning_rate": 0.00019892371369894979,
      "loss": 15.2047,
      "step": 1724
    },
    {
      "epoch": 0.3016393442622951,
      "grad_norm": 1.2030570624495718,
      "learning_rate": 0.00019892164066952727,
      "loss": 14.9972,
      "step": 1725
    },
    {
      "epoch": 0.3018142076502732,
      "grad_norm": 1.1932302632637382,
      "learning_rate": 0.00019891956565642208,
      "loss": 15.1276,
      "step": 1726
    },
    {
      "epoch": 0.30198907103825134,
      "grad_norm": 1.3112665235240044,
      "learning_rate": 0.0001989174886596759,
      "loss": 15.0714,
      "step": 1727
    },
    {
      "epoch": 0.3021639344262295,
      "grad_norm": 1.299181612653098,
      "learning_rate": 0.00019891540967933034,
      "loss": 15.1674,
      "step": 1728
    },
    {
      "epoch": 0.30233879781420764,
      "grad_norm": 1.1633604700153208,
      "learning_rate": 0.0001989133287154271,
      "loss": 15.2072,
      "step": 1729
    },
    {
      "epoch": 0.3025136612021858,
      "grad_norm": 1.1943984886536043,
      "learning_rate": 0.00019891124576800793,
      "loss": 15.0891,
      "step": 1730
    },
    {
      "epoch": 0.30268852459016393,
      "grad_norm": 1.2633488995520674,
      "learning_rate": 0.0001989091608371146,
      "loss": 14.985,
      "step": 1731
    },
    {
      "epoch": 0.30286338797814205,
      "grad_norm": 1.2320286775811506,
      "learning_rate": 0.0001989070739227889,
      "loss": 15.2444,
      "step": 1732
    },
    {
      "epoch": 0.3030382513661202,
      "grad_norm": 1.307850292071702,
      "learning_rate": 0.00019890498502507268,
      "loss": 15.1745,
      "step": 1733
    },
    {
      "epoch": 0.30321311475409835,
      "grad_norm": 1.2110553081795523,
      "learning_rate": 0.00019890289414400782,
      "loss": 15.3517,
      "step": 1734
    },
    {
      "epoch": 0.3033879781420765,
      "grad_norm": 1.2481318628395124,
      "learning_rate": 0.0001989008012796363,
      "loss": 14.9925,
      "step": 1735
    },
    {
      "epoch": 0.30356284153005464,
      "grad_norm": 1.290984895189028,
      "learning_rate": 0.00019889870643199998,
      "loss": 15.2308,
      "step": 1736
    },
    {
      "epoch": 0.3037377049180328,
      "grad_norm": 1.1964619142344273,
      "learning_rate": 0.00019889660960114097,
      "loss": 15.2128,
      "step": 1737
    },
    {
      "epoch": 0.30391256830601093,
      "grad_norm": 1.2084937775504918,
      "learning_rate": 0.00019889451078710128,
      "loss": 15.075,
      "step": 1738
    },
    {
      "epoch": 0.30408743169398905,
      "grad_norm": 1.3082390920242966,
      "learning_rate": 0.00019889240998992303,
      "loss": 15.1332,
      "step": 1739
    },
    {
      "epoch": 0.3042622950819672,
      "grad_norm": 1.2708382965027027,
      "learning_rate": 0.0001988903072096483,
      "loss": 14.8696,
      "step": 1740
    },
    {
      "epoch": 0.30443715846994535,
      "grad_norm": 1.4042825218354658,
      "learning_rate": 0.00019888820244631927,
      "loss": 15.2846,
      "step": 1741
    },
    {
      "epoch": 0.3046120218579235,
      "grad_norm": 1.2504513388128835,
      "learning_rate": 0.00019888609569997817,
      "loss": 15.2589,
      "step": 1742
    },
    {
      "epoch": 0.30478688524590164,
      "grad_norm": 1.2004187073445025,
      "learning_rate": 0.00019888398697066724,
      "loss": 15.1016,
      "step": 1743
    },
    {
      "epoch": 0.30496174863387976,
      "grad_norm": 1.3490881980717897,
      "learning_rate": 0.00019888187625842873,
      "loss": 15.1061,
      "step": 1744
    },
    {
      "epoch": 0.30513661202185793,
      "grad_norm": 1.2444788286840263,
      "learning_rate": 0.000198879763563305,
      "loss": 14.8526,
      "step": 1745
    },
    {
      "epoch": 0.30531147540983605,
      "grad_norm": 1.3642911856039048,
      "learning_rate": 0.00019887764888533842,
      "loss": 15.1049,
      "step": 1746
    },
    {
      "epoch": 0.3054863387978142,
      "grad_norm": 1.4281137548724585,
      "learning_rate": 0.0001988755322245714,
      "loss": 15.2663,
      "step": 1747
    },
    {
      "epoch": 0.30566120218579235,
      "grad_norm": 1.4225942957840507,
      "learning_rate": 0.00019887341358104635,
      "loss": 14.915,
      "step": 1748
    },
    {
      "epoch": 0.30583606557377047,
      "grad_norm": 1.2572418645072392,
      "learning_rate": 0.00019887129295480577,
      "loss": 15.1046,
      "step": 1749
    },
    {
      "epoch": 0.30601092896174864,
      "grad_norm": 1.3575288413160893,
      "learning_rate": 0.00019886917034589221,
      "loss": 15.126,
      "step": 1750
    },
    {
      "epoch": 0.30618579234972676,
      "grad_norm": 1.2925546200337994,
      "learning_rate": 0.00019886704575434818,
      "loss": 14.8943,
      "step": 1751
    },
    {
      "epoch": 0.30636065573770493,
      "grad_norm": 1.233223066750762,
      "learning_rate": 0.00019886491918021638,
      "loss": 15.2434,
      "step": 1752
    },
    {
      "epoch": 0.30653551912568305,
      "grad_norm": 1.2580235922955496,
      "learning_rate": 0.00019886279062353936,
      "loss": 14.8609,
      "step": 1753
    },
    {
      "epoch": 0.30671038251366123,
      "grad_norm": 1.3607753566114205,
      "learning_rate": 0.00019886066008435984,
      "loss": 15.282,
      "step": 1754
    },
    {
      "epoch": 0.30688524590163935,
      "grad_norm": 1.2160268949739512,
      "learning_rate": 0.00019885852756272053,
      "loss": 15.2385,
      "step": 1755
    },
    {
      "epoch": 0.30706010928961747,
      "grad_norm": 1.263403088591983,
      "learning_rate": 0.00019885639305866422,
      "loss": 15.2146,
      "step": 1756
    },
    {
      "epoch": 0.30723497267759564,
      "grad_norm": 1.2126245257939285,
      "learning_rate": 0.00019885425657223372,
      "loss": 15.0419,
      "step": 1757
    },
    {
      "epoch": 0.30740983606557376,
      "grad_norm": 1.2169373222097535,
      "learning_rate": 0.00019885211810347184,
      "loss": 15.0746,
      "step": 1758
    },
    {
      "epoch": 0.30758469945355194,
      "grad_norm": 1.3320277424860072,
      "learning_rate": 0.00019884997765242147,
      "loss": 14.9191,
      "step": 1759
    },
    {
      "epoch": 0.30775956284153005,
      "grad_norm": 1.214312085706158,
      "learning_rate": 0.00019884783521912554,
      "loss": 15.2151,
      "step": 1760
    },
    {
      "epoch": 0.3079344262295082,
      "grad_norm": 1.3199621172227647,
      "learning_rate": 0.000198845690803627,
      "loss": 14.9663,
      "step": 1761
    },
    {
      "epoch": 0.30810928961748635,
      "grad_norm": 1.3071421248300543,
      "learning_rate": 0.00019884354440596885,
      "loss": 15.0718,
      "step": 1762
    },
    {
      "epoch": 0.30828415300546447,
      "grad_norm": 1.1834161511984487,
      "learning_rate": 0.0001988413960261942,
      "loss": 14.9526,
      "step": 1763
    },
    {
      "epoch": 0.30845901639344264,
      "grad_norm": 1.3568884107483483,
      "learning_rate": 0.00019883924566434604,
      "loss": 15.1173,
      "step": 1764
    },
    {
      "epoch": 0.30863387978142076,
      "grad_norm": 1.223500053703096,
      "learning_rate": 0.00019883709332046753,
      "loss": 14.9538,
      "step": 1765
    },
    {
      "epoch": 0.3088087431693989,
      "grad_norm": 1.2624896249732398,
      "learning_rate": 0.00019883493899460183,
      "loss": 15.1908,
      "step": 1766
    },
    {
      "epoch": 0.30898360655737706,
      "grad_norm": 1.2553217635499523,
      "learning_rate": 0.00019883278268679216,
      "loss": 14.7031,
      "step": 1767
    },
    {
      "epoch": 0.3091584699453552,
      "grad_norm": 1.214250379358453,
      "learning_rate": 0.00019883062439708172,
      "loss": 15.0654,
      "step": 1768
    },
    {
      "epoch": 0.30933333333333335,
      "grad_norm": 1.2706952836056697,
      "learning_rate": 0.0001988284641255138,
      "loss": 14.991,
      "step": 1769
    },
    {
      "epoch": 0.30950819672131147,
      "grad_norm": 1.4178968899191995,
      "learning_rate": 0.0001988263018721318,
      "loss": 15.0575,
      "step": 1770
    },
    {
      "epoch": 0.30968306010928964,
      "grad_norm": 1.2248753170819218,
      "learning_rate": 0.00019882413763697891,
      "loss": 14.984,
      "step": 1771
    },
    {
      "epoch": 0.30985792349726776,
      "grad_norm": 1.1627811238037906,
      "learning_rate": 0.0001988219714200987,
      "loss": 14.8636,
      "step": 1772
    },
    {
      "epoch": 0.3100327868852459,
      "grad_norm": 1.1970810386120825,
      "learning_rate": 0.0001988198032215345,
      "loss": 14.9553,
      "step": 1773
    },
    {
      "epoch": 0.31020765027322406,
      "grad_norm": 1.2498883474697138,
      "learning_rate": 0.00019881763304132986,
      "loss": 15.0265,
      "step": 1774
    },
    {
      "epoch": 0.3103825136612022,
      "grad_norm": 1.3067548601466343,
      "learning_rate": 0.00019881546087952825,
      "loss": 15.0244,
      "step": 1775
    },
    {
      "epoch": 0.31055737704918035,
      "grad_norm": 1.2064524583262262,
      "learning_rate": 0.00019881328673617327,
      "loss": 14.8399,
      "step": 1776
    },
    {
      "epoch": 0.31073224043715847,
      "grad_norm": 1.1773787846876818,
      "learning_rate": 0.00019881111061130846,
      "loss": 15.1112,
      "step": 1777
    },
    {
      "epoch": 0.3109071038251366,
      "grad_norm": 1.2688018582704748,
      "learning_rate": 0.00019880893250497753,
      "loss": 15.0193,
      "step": 1778
    },
    {
      "epoch": 0.31108196721311476,
      "grad_norm": 1.272635975743096,
      "learning_rate": 0.0001988067524172241,
      "loss": 14.7808,
      "step": 1779
    },
    {
      "epoch": 0.3112568306010929,
      "grad_norm": 1.2073866437644458,
      "learning_rate": 0.00019880457034809188,
      "loss": 14.9209,
      "step": 1780
    },
    {
      "epoch": 0.31143169398907106,
      "grad_norm": 1.2114668818777086,
      "learning_rate": 0.00019880238629762468,
      "loss": 14.7976,
      "step": 1781
    },
    {
      "epoch": 0.3116065573770492,
      "grad_norm": 1.2253708817174598,
      "learning_rate": 0.00019880020026586626,
      "loss": 15.0122,
      "step": 1782
    },
    {
      "epoch": 0.3117814207650273,
      "grad_norm": 1.2342463658971021,
      "learning_rate": 0.00019879801225286049,
      "loss": 14.7764,
      "step": 1783
    },
    {
      "epoch": 0.31195628415300547,
      "grad_norm": 1.235060711493219,
      "learning_rate": 0.0001987958222586512,
      "loss": 15.0373,
      "step": 1784
    },
    {
      "epoch": 0.3121311475409836,
      "grad_norm": 1.1820340196111239,
      "learning_rate": 0.00019879363028328236,
      "loss": 14.9796,
      "step": 1785
    },
    {
      "epoch": 0.31230601092896176,
      "grad_norm": 1.2524468858541435,
      "learning_rate": 0.00019879143632679788,
      "loss": 14.8786,
      "step": 1786
    },
    {
      "epoch": 0.3124808743169399,
      "grad_norm": 1.1483197803775702,
      "learning_rate": 0.00019878924038924177,
      "loss": 15.0709,
      "step": 1787
    },
    {
      "epoch": 0.31265573770491806,
      "grad_norm": 1.1606290104526598,
      "learning_rate": 0.00019878704247065806,
      "loss": 14.6293,
      "step": 1788
    },
    {
      "epoch": 0.3128306010928962,
      "grad_norm": 1.196703223838402,
      "learning_rate": 0.00019878484257109083,
      "loss": 14.8883,
      "step": 1789
    },
    {
      "epoch": 0.3130054644808743,
      "grad_norm": 1.2115891058118924,
      "learning_rate": 0.0001987826406905842,
      "loss": 14.9655,
      "step": 1790
    },
    {
      "epoch": 0.31318032786885247,
      "grad_norm": 1.2112724109744712,
      "learning_rate": 0.00019878043682918235,
      "loss": 14.9263,
      "step": 1791
    },
    {
      "epoch": 0.3133551912568306,
      "grad_norm": 1.266582356851692,
      "learning_rate": 0.00019877823098692944,
      "loss": 14.8254,
      "step": 1792
    },
    {
      "epoch": 0.31353005464480876,
      "grad_norm": 1.2963044675262427,
      "learning_rate": 0.0001987760231638697,
      "loss": 14.9089,
      "step": 1793
    },
    {
      "epoch": 0.3137049180327869,
      "grad_norm": 1.1596563845803751,
      "learning_rate": 0.0001987738133600474,
      "loss": 14.7805,
      "step": 1794
    },
    {
      "epoch": 0.313879781420765,
      "grad_norm": 1.1547462747790787,
      "learning_rate": 0.00019877160157550687,
      "loss": 14.7587,
      "step": 1795
    },
    {
      "epoch": 0.3140546448087432,
      "grad_norm": 1.1789574661515536,
      "learning_rate": 0.00019876938781029249,
      "loss": 14.8824,
      "step": 1796
    },
    {
      "epoch": 0.3142295081967213,
      "grad_norm": 1.2552893904799296,
      "learning_rate": 0.00019876717206444858,
      "loss": 15.0807,
      "step": 1797
    },
    {
      "epoch": 0.31440437158469947,
      "grad_norm": 1.2276523355184823,
      "learning_rate": 0.00019876495433801962,
      "loss": 15.0469,
      "step": 1798
    },
    {
      "epoch": 0.3145792349726776,
      "grad_norm": 1.336599620516299,
      "learning_rate": 0.00019876273463105012,
      "loss": 14.6332,
      "step": 1799
    },
    {
      "epoch": 0.31475409836065577,
      "grad_norm": 1.306181159793415,
      "learning_rate": 0.0001987605129435845,
      "loss": 14.8868,
      "step": 1800
    },
    {
      "epoch": 0.3149289617486339,
      "grad_norm": 1.303650610998474,
      "learning_rate": 0.0001987582892756674,
      "loss": 15.0466,
      "step": 1801
    },
    {
      "epoch": 0.315103825136612,
      "grad_norm": 1.3093302804251574,
      "learning_rate": 0.00019875606362734335,
      "loss": 15.0917,
      "step": 1802
    },
    {
      "epoch": 0.3152786885245902,
      "grad_norm": 1.3428779486195572,
      "learning_rate": 0.000198753835998657,
      "loss": 14.8781,
      "step": 1803
    },
    {
      "epoch": 0.3154535519125683,
      "grad_norm": 1.259601461629174,
      "learning_rate": 0.00019875160638965303,
      "loss": 14.9582,
      "step": 1804
    },
    {
      "epoch": 0.31562841530054647,
      "grad_norm": 1.1914582741969595,
      "learning_rate": 0.00019874937480037614,
      "loss": 15.0426,
      "step": 1805
    },
    {
      "epoch": 0.3158032786885246,
      "grad_norm": 1.343889786215702,
      "learning_rate": 0.00019874714123087111,
      "loss": 14.8336,
      "step": 1806
    },
    {
      "epoch": 0.3159781420765027,
      "grad_norm": 1.1880390164369583,
      "learning_rate": 0.00019874490568118266,
      "loss": 14.8225,
      "step": 1807
    },
    {
      "epoch": 0.3161530054644809,
      "grad_norm": 1.2602629162864787,
      "learning_rate": 0.00019874266815135568,
      "loss": 14.4834,
      "step": 1808
    },
    {
      "epoch": 0.316327868852459,
      "grad_norm": 1.4201086081229082,
      "learning_rate": 0.00019874042864143507,
      "loss": 14.9128,
      "step": 1809
    },
    {
      "epoch": 0.3165027322404372,
      "grad_norm": 1.2691537932111148,
      "learning_rate": 0.00019873818715146566,
      "loss": 14.9501,
      "step": 1810
    },
    {
      "epoch": 0.3166775956284153,
      "grad_norm": 1.265255690873543,
      "learning_rate": 0.00019873594368149242,
      "loss": 14.5538,
      "step": 1811
    },
    {
      "epoch": 0.3168524590163934,
      "grad_norm": 1.4843612852346841,
      "learning_rate": 0.00019873369823156036,
      "loss": 14.6291,
      "step": 1812
    },
    {
      "epoch": 0.3170273224043716,
      "grad_norm": 1.1703323576177935,
      "learning_rate": 0.00019873145080171449,
      "loss": 14.7944,
      "step": 1813
    },
    {
      "epoch": 0.3172021857923497,
      "grad_norm": 1.3483476125568963,
      "learning_rate": 0.00019872920139199992,
      "loss": 15.1139,
      "step": 1814
    },
    {
      "epoch": 0.3173770491803279,
      "grad_norm": 1.3218437352446675,
      "learning_rate": 0.00019872695000246166,
      "loss": 14.9334,
      "step": 1815
    },
    {
      "epoch": 0.317551912568306,
      "grad_norm": 1.2037186945721585,
      "learning_rate": 0.00019872469663314497,
      "loss": 14.9298,
      "step": 1816
    },
    {
      "epoch": 0.3177267759562842,
      "grad_norm": 1.250305431719327,
      "learning_rate": 0.00019872244128409498,
      "loss": 14.9078,
      "step": 1817
    },
    {
      "epoch": 0.3179016393442623,
      "grad_norm": 1.297243335292834,
      "learning_rate": 0.0001987201839553569,
      "loss": 14.9041,
      "step": 1818
    },
    {
      "epoch": 0.3180765027322404,
      "grad_norm": 1.3253890424455963,
      "learning_rate": 0.00019871792464697604,
      "loss": 15.1107,
      "step": 1819
    },
    {
      "epoch": 0.3182513661202186,
      "grad_norm": 1.1942241110084926,
      "learning_rate": 0.0001987156633589977,
      "loss": 14.9892,
      "step": 1820
    },
    {
      "epoch": 0.3184262295081967,
      "grad_norm": 1.424125933300536,
      "learning_rate": 0.0001987134000914672,
      "loss": 15.1649,
      "step": 1821
    },
    {
      "epoch": 0.3186010928961749,
      "grad_norm": 1.3587246093606136,
      "learning_rate": 0.00019871113484442994,
      "loss": 14.9242,
      "step": 1822
    },
    {
      "epoch": 0.318775956284153,
      "grad_norm": 1.3026711164045657,
      "learning_rate": 0.00019870886761793135,
      "loss": 14.9067,
      "step": 1823
    },
    {
      "epoch": 0.3189508196721311,
      "grad_norm": 1.2491967603810492,
      "learning_rate": 0.00019870659841201687,
      "loss": 15.0322,
      "step": 1824
    },
    {
      "epoch": 0.3191256830601093,
      "grad_norm": 1.2258051267426773,
      "learning_rate": 0.00019870432722673203,
      "loss": 14.9485,
      "step": 1825
    },
    {
      "epoch": 0.3193005464480874,
      "grad_norm": 1.2691370354506446,
      "learning_rate": 0.00019870205406212236,
      "loss": 14.7922,
      "step": 1826
    },
    {
      "epoch": 0.3194754098360656,
      "grad_norm": 1.250106220380363,
      "learning_rate": 0.00019869977891823348,
      "loss": 14.9249,
      "step": 1827
    },
    {
      "epoch": 0.3196502732240437,
      "grad_norm": 1.2877513129564944,
      "learning_rate": 0.00019869750179511095,
      "loss": 15.0863,
      "step": 1828
    },
    {
      "epoch": 0.31982513661202183,
      "grad_norm": 1.3455302555893953,
      "learning_rate": 0.0001986952226928005,
      "loss": 14.7635,
      "step": 1829
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.178009399549713,
      "learning_rate": 0.00019869294161134777,
      "loss": 14.9364,
      "step": 1830
    },
    {
      "epoch": 0.3201748633879781,
      "grad_norm": 1.2664552457850444,
      "learning_rate": 0.00019869065855079857,
      "loss": 14.9241,
      "step": 1831
    },
    {
      "epoch": 0.3203497267759563,
      "grad_norm": 1.291111663965794,
      "learning_rate": 0.0001986883735111986,
      "loss": 14.7381,
      "step": 1832
    },
    {
      "epoch": 0.3205245901639344,
      "grad_norm": 1.3509945899957636,
      "learning_rate": 0.00019868608649259377,
      "loss": 15.0445,
      "step": 1833
    },
    {
      "epoch": 0.3206994535519126,
      "grad_norm": 2.129631949240313,
      "learning_rate": 0.00019868379749502987,
      "loss": 14.8044,
      "step": 1834
    },
    {
      "epoch": 0.3208743169398907,
      "grad_norm": 1.197083501457805,
      "learning_rate": 0.00019868150651855285,
      "loss": 14.9252,
      "step": 1835
    },
    {
      "epoch": 0.32104918032786883,
      "grad_norm": 1.1588676460756453,
      "learning_rate": 0.0001986792135632086,
      "loss": 14.7405,
      "step": 1836
    },
    {
      "epoch": 0.321224043715847,
      "grad_norm": 1.2212200887768234,
      "learning_rate": 0.00019867691862904317,
      "loss": 14.6717,
      "step": 1837
    },
    {
      "epoch": 0.3213989071038251,
      "grad_norm": 1.2394479078372993,
      "learning_rate": 0.00019867462171610254,
      "loss": 14.4881,
      "step": 1838
    },
    {
      "epoch": 0.3215737704918033,
      "grad_norm": 1.1399464527834597,
      "learning_rate": 0.00019867232282443277,
      "loss": 14.8982,
      "step": 1839
    },
    {
      "epoch": 0.3217486338797814,
      "grad_norm": 1.2707514618182498,
      "learning_rate": 0.00019867002195407993,
      "loss": 14.6681,
      "step": 1840
    },
    {
      "epoch": 0.32192349726775954,
      "grad_norm": 1.2422819465150463,
      "learning_rate": 0.00019866771910509023,
      "loss": 14.6561,
      "step": 1841
    },
    {
      "epoch": 0.3220983606557377,
      "grad_norm": 1.2629323216396504,
      "learning_rate": 0.0001986654142775098,
      "loss": 14.6617,
      "step": 1842
    },
    {
      "epoch": 0.32227322404371583,
      "grad_norm": 1.3308929354231613,
      "learning_rate": 0.0001986631074713849,
      "loss": 14.678,
      "step": 1843
    },
    {
      "epoch": 0.322448087431694,
      "grad_norm": 1.254003409617093,
      "learning_rate": 0.00019866079868676172,
      "loss": 14.7963,
      "step": 1844
    },
    {
      "epoch": 0.3226229508196721,
      "grad_norm": 1.4261990766204966,
      "learning_rate": 0.0001986584879236866,
      "loss": 14.6868,
      "step": 1845
    },
    {
      "epoch": 0.32279781420765025,
      "grad_norm": 1.223757023493585,
      "learning_rate": 0.0001986561751822059,
      "loss": 15.2272,
      "step": 1846
    },
    {
      "epoch": 0.3229726775956284,
      "grad_norm": 1.3269212942112407,
      "learning_rate": 0.00019865386046236596,
      "loss": 14.5531,
      "step": 1847
    },
    {
      "epoch": 0.32314754098360654,
      "grad_norm": 1.2565103131047992,
      "learning_rate": 0.00019865154376421323,
      "loss": 14.905,
      "step": 1848
    },
    {
      "epoch": 0.3233224043715847,
      "grad_norm": 1.4544021228471624,
      "learning_rate": 0.00019864922508779412,
      "loss": 14.8906,
      "step": 1849
    },
    {
      "epoch": 0.32349726775956283,
      "grad_norm": 1.402382326406243,
      "learning_rate": 0.00019864690443315515,
      "loss": 14.7287,
      "step": 1850
    },
    {
      "epoch": 0.323672131147541,
      "grad_norm": 1.345999297951487,
      "learning_rate": 0.0001986445818003429,
      "loss": 15.0431,
      "step": 1851
    },
    {
      "epoch": 0.32384699453551913,
      "grad_norm": 1.3071939961249515,
      "learning_rate": 0.00019864225718940385,
      "loss": 14.5099,
      "step": 1852
    },
    {
      "epoch": 0.32402185792349725,
      "grad_norm": 1.285425209281917,
      "learning_rate": 0.00019863993060038472,
      "loss": 14.7983,
      "step": 1853
    },
    {
      "epoch": 0.3241967213114754,
      "grad_norm": 1.2521631841872596,
      "learning_rate": 0.00019863760203333208,
      "loss": 14.9411,
      "step": 1854
    },
    {
      "epoch": 0.32437158469945354,
      "grad_norm": 1.3753514018652413,
      "learning_rate": 0.0001986352714882927,
      "loss": 14.6725,
      "step": 1855
    },
    {
      "epoch": 0.3245464480874317,
      "grad_norm": 1.2791747477842885,
      "learning_rate": 0.00019863293896531326,
      "loss": 14.9728,
      "step": 1856
    },
    {
      "epoch": 0.32472131147540984,
      "grad_norm": 1.3429814257410133,
      "learning_rate": 0.00019863060446444054,
      "loss": 15.1375,
      "step": 1857
    },
    {
      "epoch": 0.32489617486338795,
      "grad_norm": 1.2757447443682726,
      "learning_rate": 0.00019862826798572138,
      "loss": 14.7633,
      "step": 1858
    },
    {
      "epoch": 0.32507103825136613,
      "grad_norm": 1.2542929709204342,
      "learning_rate": 0.0001986259295292026,
      "loss": 14.8694,
      "step": 1859
    },
    {
      "epoch": 0.32524590163934425,
      "grad_norm": 1.2310774879843125,
      "learning_rate": 0.00019862358909493113,
      "loss": 14.7921,
      "step": 1860
    },
    {
      "epoch": 0.3254207650273224,
      "grad_norm": 1.157816759689114,
      "learning_rate": 0.00019862124668295388,
      "loss": 14.7705,
      "step": 1861
    },
    {
      "epoch": 0.32559562841530054,
      "grad_norm": 1.2015216676527423,
      "learning_rate": 0.00019861890229331782,
      "loss": 14.5214,
      "step": 1862
    },
    {
      "epoch": 0.32577049180327866,
      "grad_norm": 1.140397455330967,
      "learning_rate": 0.00019861655592606997,
      "loss": 14.8069,
      "step": 1863
    },
    {
      "epoch": 0.32594535519125684,
      "grad_norm": 1.2311442562575006,
      "learning_rate": 0.0001986142075812574,
      "loss": 14.9389,
      "step": 1864
    },
    {
      "epoch": 0.32612021857923496,
      "grad_norm": 1.1745421008816146,
      "learning_rate": 0.00019861185725892713,
      "loss": 14.9728,
      "step": 1865
    },
    {
      "epoch": 0.32629508196721313,
      "grad_norm": 1.2359177279887754,
      "learning_rate": 0.00019860950495912642,
      "loss": 14.9059,
      "step": 1866
    },
    {
      "epoch": 0.32646994535519125,
      "grad_norm": 1.1830468556134714,
      "learning_rate": 0.00019860715068190232,
      "loss": 14.6809,
      "step": 1867
    },
    {
      "epoch": 0.3266448087431694,
      "grad_norm": 1.2391638083727659,
      "learning_rate": 0.00019860479442730208,
      "loss": 14.6553,
      "step": 1868
    },
    {
      "epoch": 0.32681967213114754,
      "grad_norm": 1.1975170982785948,
      "learning_rate": 0.000198602436195373,
      "loss": 14.736,
      "step": 1869
    },
    {
      "epoch": 0.32699453551912566,
      "grad_norm": 1.2834062048593349,
      "learning_rate": 0.00019860007598616227,
      "loss": 14.8551,
      "step": 1870
    },
    {
      "epoch": 0.32716939890710384,
      "grad_norm": 1.2587389108940408,
      "learning_rate": 0.0001985977137997173,
      "loss": 14.4298,
      "step": 1871
    },
    {
      "epoch": 0.32734426229508196,
      "grad_norm": 1.1639264830846634,
      "learning_rate": 0.00019859534963608542,
      "loss": 14.583,
      "step": 1872
    },
    {
      "epoch": 0.32751912568306013,
      "grad_norm": 1.2413689655927351,
      "learning_rate": 0.00019859298349531406,
      "loss": 14.6361,
      "step": 1873
    },
    {
      "epoch": 0.32769398907103825,
      "grad_norm": 1.226670078463559,
      "learning_rate": 0.00019859061537745066,
      "loss": 14.9937,
      "step": 1874
    },
    {
      "epoch": 0.32786885245901637,
      "grad_norm": 1.1914022269147924,
      "learning_rate": 0.0001985882452825427,
      "loss": 14.3465,
      "step": 1875
    },
    {
      "epoch": 0.32804371584699454,
      "grad_norm": 1.1113345069406773,
      "learning_rate": 0.00019858587321063776,
      "loss": 14.6836,
      "step": 1876
    },
    {
      "epoch": 0.32821857923497266,
      "grad_norm": 1.1456465688166932,
      "learning_rate": 0.00019858349916178334,
      "loss": 14.6813,
      "step": 1877
    },
    {
      "epoch": 0.32839344262295084,
      "grad_norm": 1.1956261886135504,
      "learning_rate": 0.00019858112313602707,
      "loss": 14.7937,
      "step": 1878
    },
    {
      "epoch": 0.32856830601092896,
      "grad_norm": 1.1943654694716335,
      "learning_rate": 0.00019857874513341657,
      "loss": 14.6547,
      "step": 1879
    },
    {
      "epoch": 0.32874316939890713,
      "grad_norm": 1.137484566875946,
      "learning_rate": 0.00019857636515399957,
      "loss": 14.8963,
      "step": 1880
    },
    {
      "epoch": 0.32891803278688525,
      "grad_norm": 1.1604069303950746,
      "learning_rate": 0.00019857398319782376,
      "loss": 14.844,
      "step": 1881
    },
    {
      "epoch": 0.32909289617486337,
      "grad_norm": 1.181843516825482,
      "learning_rate": 0.00019857159926493697,
      "loss": 14.745,
      "step": 1882
    },
    {
      "epoch": 0.32926775956284154,
      "grad_norm": 1.277317972878874,
      "learning_rate": 0.00019856921335538694,
      "loss": 14.7238,
      "step": 1883
    },
    {
      "epoch": 0.32944262295081966,
      "grad_norm": 1.1435161254111972,
      "learning_rate": 0.00019856682546922155,
      "loss": 14.6624,
      "step": 1884
    },
    {
      "epoch": 0.32961748633879784,
      "grad_norm": 1.3243262068540136,
      "learning_rate": 0.00019856443560648863,
      "loss": 15.0109,
      "step": 1885
    },
    {
      "epoch": 0.32979234972677596,
      "grad_norm": 1.1513335533100855,
      "learning_rate": 0.00019856204376723617,
      "loss": 14.7447,
      "step": 1886
    },
    {
      "epoch": 0.3299672131147541,
      "grad_norm": 1.237369327477382,
      "learning_rate": 0.0001985596499515121,
      "loss": 14.7572,
      "step": 1887
    },
    {
      "epoch": 0.33014207650273225,
      "grad_norm": 1.285463857540582,
      "learning_rate": 0.00019855725415936443,
      "loss": 14.7873,
      "step": 1888
    },
    {
      "epoch": 0.33031693989071037,
      "grad_norm": 1.1805978201078446,
      "learning_rate": 0.00019855485639084123,
      "loss": 14.7956,
      "step": 1889
    },
    {
      "epoch": 0.33049180327868855,
      "grad_norm": 1.1803995355175803,
      "learning_rate": 0.0001985524566459905,
      "loss": 14.5523,
      "step": 1890
    },
    {
      "epoch": 0.33066666666666666,
      "grad_norm": 1.1303275335555618,
      "learning_rate": 0.00019855005492486045,
      "loss": 14.7151,
      "step": 1891
    },
    {
      "epoch": 0.3308415300546448,
      "grad_norm": 1.238677753424216,
      "learning_rate": 0.00019854765122749924,
      "loss": 14.7309,
      "step": 1892
    },
    {
      "epoch": 0.33101639344262296,
      "grad_norm": 1.2452169961748132,
      "learning_rate": 0.00019854524555395502,
      "loss": 14.5531,
      "step": 1893
    },
    {
      "epoch": 0.3311912568306011,
      "grad_norm": 1.2011293359955206,
      "learning_rate": 0.00019854283790427603,
      "loss": 14.7028,
      "step": 1894
    },
    {
      "epoch": 0.33136612021857925,
      "grad_norm": 1.2070369854573955,
      "learning_rate": 0.00019854042827851062,
      "loss": 14.6526,
      "step": 1895
    },
    {
      "epoch": 0.33154098360655737,
      "grad_norm": 1.1745601329344735,
      "learning_rate": 0.00019853801667670703,
      "loss": 14.8588,
      "step": 1896
    },
    {
      "epoch": 0.33171584699453555,
      "grad_norm": 1.1865496920429115,
      "learning_rate": 0.00019853560309891365,
      "loss": 14.4447,
      "step": 1897
    },
    {
      "epoch": 0.33189071038251367,
      "grad_norm": 1.213335594139242,
      "learning_rate": 0.00019853318754517888,
      "loss": 14.715,
      "step": 1898
    },
    {
      "epoch": 0.3320655737704918,
      "grad_norm": 1.1185873469372847,
      "learning_rate": 0.00019853077001555117,
      "loss": 14.4343,
      "step": 1899
    },
    {
      "epoch": 0.33224043715846996,
      "grad_norm": 1.184387618138646,
      "learning_rate": 0.000198528350510079,
      "loss": 14.6366,
      "step": 1900
    },
    {
      "epoch": 0.3324153005464481,
      "grad_norm": 1.259128809604082,
      "learning_rate": 0.0001985259290288109,
      "loss": 14.4785,
      "step": 1901
    },
    {
      "epoch": 0.33259016393442625,
      "grad_norm": 1.1423174687501672,
      "learning_rate": 0.00019852350557179538,
      "loss": 14.3936,
      "step": 1902
    },
    {
      "epoch": 0.33276502732240437,
      "grad_norm": 1.2771914259203019,
      "learning_rate": 0.00019852108013908106,
      "loss": 14.6488,
      "step": 1903
    },
    {
      "epoch": 0.3329398907103825,
      "grad_norm": 1.1882028088418697,
      "learning_rate": 0.0001985186527307166,
      "loss": 14.6762,
      "step": 1904
    },
    {
      "epoch": 0.33311475409836067,
      "grad_norm": 1.179913299820119,
      "learning_rate": 0.00019851622334675066,
      "loss": 14.8597,
      "step": 1905
    },
    {
      "epoch": 0.3332896174863388,
      "grad_norm": 1.1989570090577832,
      "learning_rate": 0.00019851379198723197,
      "loss": 14.6514,
      "step": 1906
    },
    {
      "epoch": 0.33346448087431696,
      "grad_norm": 1.1853830576482427,
      "learning_rate": 0.00019851135865220923,
      "loss": 14.8423,
      "step": 1907
    },
    {
      "epoch": 0.3336393442622951,
      "grad_norm": 1.1919037343120773,
      "learning_rate": 0.00019850892334173132,
      "loss": 14.4729,
      "step": 1908
    },
    {
      "epoch": 0.3338142076502732,
      "grad_norm": 1.1690116219192492,
      "learning_rate": 0.00019850648605584703,
      "loss": 14.6972,
      "step": 1909
    },
    {
      "epoch": 0.3339890710382514,
      "grad_norm": 1.2769663771411728,
      "learning_rate": 0.00019850404679460523,
      "loss": 14.5921,
      "step": 1910
    },
    {
      "epoch": 0.3341639344262295,
      "grad_norm": 1.1757359798783926,
      "learning_rate": 0.00019850160555805486,
      "loss": 14.6607,
      "step": 1911
    },
    {
      "epoch": 0.33433879781420767,
      "grad_norm": 1.1827546768062398,
      "learning_rate": 0.00019849916234624483,
      "loss": 14.6568,
      "step": 1912
    },
    {
      "epoch": 0.3345136612021858,
      "grad_norm": 1.22355201238856,
      "learning_rate": 0.0001984967171592242,
      "loss": 14.6678,
      "step": 1913
    },
    {
      "epoch": 0.33468852459016396,
      "grad_norm": 1.1841032039635624,
      "learning_rate": 0.00019849426999704194,
      "loss": 14.469,
      "step": 1914
    },
    {
      "epoch": 0.3348633879781421,
      "grad_norm": 1.1376311134675339,
      "learning_rate": 0.00019849182085974714,
      "loss": 14.4137,
      "step": 1915
    },
    {
      "epoch": 0.3350382513661202,
      "grad_norm": 1.1344711294947634,
      "learning_rate": 0.00019848936974738895,
      "loss": 14.7513,
      "step": 1916
    },
    {
      "epoch": 0.3352131147540984,
      "grad_norm": 1.2748473234669684,
      "learning_rate": 0.00019848691666001647,
      "loss": 14.731,
      "step": 1917
    },
    {
      "epoch": 0.3353879781420765,
      "grad_norm": 1.0934954683502423,
      "learning_rate": 0.00019848446159767896,
      "loss": 14.2695,
      "step": 1918
    },
    {
      "epoch": 0.33556284153005467,
      "grad_norm": 1.1723159920526938,
      "learning_rate": 0.00019848200456042557,
      "loss": 14.4128,
      "step": 1919
    },
    {
      "epoch": 0.3357377049180328,
      "grad_norm": 1.2329993494531204,
      "learning_rate": 0.00019847954554830558,
      "loss": 14.7569,
      "step": 1920
    },
    {
      "epoch": 0.3359125683060109,
      "grad_norm": 1.1885856325440385,
      "learning_rate": 0.00019847708456136837,
      "loss": 14.4921,
      "step": 1921
    },
    {
      "epoch": 0.3360874316939891,
      "grad_norm": 1.2083887095066677,
      "learning_rate": 0.00019847462159966326,
      "loss": 14.5807,
      "step": 1922
    },
    {
      "epoch": 0.3362622950819672,
      "grad_norm": 1.2320035768911477,
      "learning_rate": 0.0001984721566632396,
      "loss": 14.5938,
      "step": 1923
    },
    {
      "epoch": 0.3364371584699454,
      "grad_norm": 1.1732139851469165,
      "learning_rate": 0.0001984696897521469,
      "loss": 14.5697,
      "step": 1924
    },
    {
      "epoch": 0.3366120218579235,
      "grad_norm": 1.359745219775113,
      "learning_rate": 0.0001984672208664345,
      "loss": 14.8715,
      "step": 1925
    },
    {
      "epoch": 0.3367868852459016,
      "grad_norm": 1.1980984518503504,
      "learning_rate": 0.00019846475000615205,
      "loss": 14.7032,
      "step": 1926
    },
    {
      "epoch": 0.3369617486338798,
      "grad_norm": 1.1040579394358827,
      "learning_rate": 0.00019846227717134902,
      "loss": 14.1584,
      "step": 1927
    },
    {
      "epoch": 0.3371366120218579,
      "grad_norm": 1.1856503205872149,
      "learning_rate": 0.00019845980236207503,
      "loss": 14.7754,
      "step": 1928
    },
    {
      "epoch": 0.3373114754098361,
      "grad_norm": 1.1588823291739117,
      "learning_rate": 0.00019845732557837966,
      "loss": 14.2169,
      "step": 1929
    },
    {
      "epoch": 0.3374863387978142,
      "grad_norm": 1.2108922562633588,
      "learning_rate": 0.00019845484682031264,
      "loss": 14.4658,
      "step": 1930
    },
    {
      "epoch": 0.3376612021857924,
      "grad_norm": 1.1302570863679369,
      "learning_rate": 0.00019845236608792362,
      "loss": 14.6631,
      "step": 1931
    },
    {
      "epoch": 0.3378360655737705,
      "grad_norm": 1.194182074387355,
      "learning_rate": 0.00019844988338126238,
      "loss": 14.8016,
      "step": 1932
    },
    {
      "epoch": 0.3380109289617486,
      "grad_norm": 1.1932963964517003,
      "learning_rate": 0.00019844739870037868,
      "loss": 14.5749,
      "step": 1933
    },
    {
      "epoch": 0.3381857923497268,
      "grad_norm": 1.210063828258073,
      "learning_rate": 0.00019844491204532236,
      "loss": 14.6613,
      "step": 1934
    },
    {
      "epoch": 0.3383606557377049,
      "grad_norm": 1.136621002101936,
      "learning_rate": 0.00019844242341614333,
      "loss": 14.6717,
      "step": 1935
    },
    {
      "epoch": 0.3385355191256831,
      "grad_norm": 1.198606085973608,
      "learning_rate": 0.00019843993281289143,
      "loss": 14.4972,
      "step": 1936
    },
    {
      "epoch": 0.3387103825136612,
      "grad_norm": 1.1742521621115898,
      "learning_rate": 0.0001984374402356166,
      "loss": 14.8703,
      "step": 1937
    },
    {
      "epoch": 0.3388852459016393,
      "grad_norm": 1.1656844410549443,
      "learning_rate": 0.0001984349456843689,
      "loss": 14.8035,
      "step": 1938
    },
    {
      "epoch": 0.3390601092896175,
      "grad_norm": 1.1537823194783074,
      "learning_rate": 0.00019843244915919827,
      "loss": 14.4693,
      "step": 1939
    },
    {
      "epoch": 0.3392349726775956,
      "grad_norm": 1.0328615521691973,
      "learning_rate": 0.0001984299506601548,
      "loss": 14.5735,
      "step": 1940
    },
    {
      "epoch": 0.3394098360655738,
      "grad_norm": 1.1530780600907633,
      "learning_rate": 0.00019842745018728867,
      "loss": 14.3576,
      "step": 1941
    },
    {
      "epoch": 0.3395846994535519,
      "grad_norm": 1.1184881624853211,
      "learning_rate": 0.00019842494774064987,
      "loss": 14.9143,
      "step": 1942
    },
    {
      "epoch": 0.33975956284153,
      "grad_norm": 1.1308775699765268,
      "learning_rate": 0.00019842244332028867,
      "loss": 14.7243,
      "step": 1943
    },
    {
      "epoch": 0.3399344262295082,
      "grad_norm": 1.278292234408509,
      "learning_rate": 0.00019841993692625533,
      "loss": 14.413,
      "step": 1944
    },
    {
      "epoch": 0.3401092896174863,
      "grad_norm": 1.1058861400454307,
      "learning_rate": 0.00019841742855860003,
      "loss": 14.4538,
      "step": 1945
    },
    {
      "epoch": 0.3402841530054645,
      "grad_norm": 1.2682502630901047,
      "learning_rate": 0.00019841491821737312,
      "loss": 14.6058,
      "step": 1946
    },
    {
      "epoch": 0.3404590163934426,
      "grad_norm": 1.3100449093543247,
      "learning_rate": 0.00019841240590262493,
      "loss": 14.4953,
      "step": 1947
    },
    {
      "epoch": 0.3406338797814208,
      "grad_norm": 1.0884096577336249,
      "learning_rate": 0.00019840989161440585,
      "loss": 14.493,
      "step": 1948
    },
    {
      "epoch": 0.3408087431693989,
      "grad_norm": 1.3477530439816745,
      "learning_rate": 0.00019840737535276626,
      "loss": 14.3688,
      "step": 1949
    },
    {
      "epoch": 0.34098360655737703,
      "grad_norm": 1.095904681187809,
      "learning_rate": 0.00019840485711775663,
      "loss": 14.5171,
      "step": 1950
    },
    {
      "epoch": 0.3411584699453552,
      "grad_norm": 1.3144221713153574,
      "learning_rate": 0.0001984023369094275,
      "loss": 14.6185,
      "step": 1951
    },
    {
      "epoch": 0.3413333333333333,
      "grad_norm": 1.2446948260403272,
      "learning_rate": 0.00019839981472782937,
      "loss": 14.6376,
      "step": 1952
    },
    {
      "epoch": 0.3415081967213115,
      "grad_norm": 1.1061954516142283,
      "learning_rate": 0.00019839729057301284,
      "loss": 14.416,
      "step": 1953
    },
    {
      "epoch": 0.3416830601092896,
      "grad_norm": 1.1283289818293867,
      "learning_rate": 0.0001983947644450285,
      "loss": 14.5254,
      "step": 1954
    },
    {
      "epoch": 0.34185792349726773,
      "grad_norm": 1.1305938487171743,
      "learning_rate": 0.00019839223634392702,
      "loss": 14.7145,
      "step": 1955
    },
    {
      "epoch": 0.3420327868852459,
      "grad_norm": 1.0898322805339002,
      "learning_rate": 0.0001983897062697591,
      "loss": 14.3998,
      "step": 1956
    },
    {
      "epoch": 0.34220765027322403,
      "grad_norm": 1.164960270672159,
      "learning_rate": 0.00019838717422257548,
      "loss": 14.6289,
      "step": 1957
    },
    {
      "epoch": 0.3423825136612022,
      "grad_norm": 1.1432573247409656,
      "learning_rate": 0.00019838464020242692,
      "loss": 14.6783,
      "step": 1958
    },
    {
      "epoch": 0.3425573770491803,
      "grad_norm": 1.205921681647014,
      "learning_rate": 0.00019838210420936427,
      "loss": 14.5782,
      "step": 1959
    },
    {
      "epoch": 0.3427322404371585,
      "grad_norm": 1.1652100626285122,
      "learning_rate": 0.0001983795662434383,
      "loss": 14.689,
      "step": 1960
    },
    {
      "epoch": 0.3429071038251366,
      "grad_norm": 1.2425549633458248,
      "learning_rate": 0.0001983770263047,
      "loss": 14.5822,
      "step": 1961
    },
    {
      "epoch": 0.34308196721311474,
      "grad_norm": 1.2787962116012288,
      "learning_rate": 0.00019837448439320027,
      "loss": 14.5334,
      "step": 1962
    },
    {
      "epoch": 0.3432568306010929,
      "grad_norm": 1.0921164717525256,
      "learning_rate": 0.0001983719405089901,
      "loss": 14.4191,
      "step": 1963
    },
    {
      "epoch": 0.34343169398907103,
      "grad_norm": 1.177440807123896,
      "learning_rate": 0.00019836939465212044,
      "loss": 14.5905,
      "step": 1964
    },
    {
      "epoch": 0.3436065573770492,
      "grad_norm": 1.2529589581319336,
      "learning_rate": 0.00019836684682264242,
      "loss": 14.5859,
      "step": 1965
    },
    {
      "epoch": 0.3437814207650273,
      "grad_norm": 1.3043534812574837,
      "learning_rate": 0.00019836429702060703,
      "loss": 14.8607,
      "step": 1966
    },
    {
      "epoch": 0.34395628415300544,
      "grad_norm": 1.373424372306472,
      "learning_rate": 0.00019836174524606555,
      "loss": 14.2885,
      "step": 1967
    },
    {
      "epoch": 0.3441311475409836,
      "grad_norm": 1.1925585128683067,
      "learning_rate": 0.00019835919149906902,
      "loss": 14.6236,
      "step": 1968
    },
    {
      "epoch": 0.34430601092896174,
      "grad_norm": 1.2642788751576814,
      "learning_rate": 0.0001983566357796687,
      "loss": 14.5353,
      "step": 1969
    },
    {
      "epoch": 0.3444808743169399,
      "grad_norm": 1.2278142344991387,
      "learning_rate": 0.00019835407808791583,
      "loss": 14.7265,
      "step": 1970
    },
    {
      "epoch": 0.34465573770491803,
      "grad_norm": 1.1459467657869988,
      "learning_rate": 0.0001983515184238617,
      "loss": 14.3382,
      "step": 1971
    },
    {
      "epoch": 0.34483060109289615,
      "grad_norm": 1.2616848846094157,
      "learning_rate": 0.0001983489567875577,
      "loss": 14.8732,
      "step": 1972
    },
    {
      "epoch": 0.3450054644808743,
      "grad_norm": 1.2340729926217762,
      "learning_rate": 0.0001983463931790551,
      "loss": 14.8117,
      "step": 1973
    },
    {
      "epoch": 0.34518032786885244,
      "grad_norm": 1.171842693270691,
      "learning_rate": 0.00019834382759840538,
      "loss": 14.2266,
      "step": 1974
    },
    {
      "epoch": 0.3453551912568306,
      "grad_norm": 1.186922988768751,
      "learning_rate": 0.00019834126004565994,
      "loss": 14.6692,
      "step": 1975
    },
    {
      "epoch": 0.34553005464480874,
      "grad_norm": 1.1757039039072585,
      "learning_rate": 0.0001983386905208703,
      "loss": 14.857,
      "step": 1976
    },
    {
      "epoch": 0.3457049180327869,
      "grad_norm": 1.150721818526667,
      "learning_rate": 0.00019833611902408794,
      "loss": 14.4662,
      "step": 1977
    },
    {
      "epoch": 0.34587978142076503,
      "grad_norm": 1.1715753858106959,
      "learning_rate": 0.00019833354555536453,
      "loss": 14.6327,
      "step": 1978
    },
    {
      "epoch": 0.34605464480874315,
      "grad_norm": 1.1499856210821564,
      "learning_rate": 0.00019833097011475156,
      "loss": 14.4725,
      "step": 1979
    },
    {
      "epoch": 0.3462295081967213,
      "grad_norm": 1.176268102103095,
      "learning_rate": 0.00019832839270230075,
      "loss": 14.4117,
      "step": 1980
    },
    {
      "epoch": 0.34640437158469944,
      "grad_norm": 1.0571678748728683,
      "learning_rate": 0.00019832581331806376,
      "loss": 14.1751,
      "step": 1981
    },
    {
      "epoch": 0.3465792349726776,
      "grad_norm": 1.114322569668311,
      "learning_rate": 0.00019832323196209228,
      "loss": 14.4234,
      "step": 1982
    },
    {
      "epoch": 0.34675409836065574,
      "grad_norm": 1.1630039208754057,
      "learning_rate": 0.0001983206486344381,
      "loss": 14.6061,
      "step": 1983
    },
    {
      "epoch": 0.34692896174863386,
      "grad_norm": 1.2523002307330204,
      "learning_rate": 0.00019831806333515307,
      "loss": 14.5919,
      "step": 1984
    },
    {
      "epoch": 0.34710382513661203,
      "grad_norm": 1.1632324918816839,
      "learning_rate": 0.000198315476064289,
      "loss": 14.6565,
      "step": 1985
    },
    {
      "epoch": 0.34727868852459015,
      "grad_norm": 1.129661566361788,
      "learning_rate": 0.00019831288682189773,
      "loss": 14.4779,
      "step": 1986
    },
    {
      "epoch": 0.3474535519125683,
      "grad_norm": 1.0836039504478945,
      "learning_rate": 0.00019831029560803123,
      "loss": 14.5182,
      "step": 1987
    },
    {
      "epoch": 0.34762841530054645,
      "grad_norm": 1.0747127251662845,
      "learning_rate": 0.00019830770242274147,
      "loss": 14.7169,
      "step": 1988
    },
    {
      "epoch": 0.34780327868852456,
      "grad_norm": 1.2342548745756625,
      "learning_rate": 0.0001983051072660804,
      "loss": 14.6145,
      "step": 1989
    },
    {
      "epoch": 0.34797814207650274,
      "grad_norm": 1.1826573091253993,
      "learning_rate": 0.0001983025101381001,
      "loss": 14.6803,
      "step": 1990
    },
    {
      "epoch": 0.34815300546448086,
      "grad_norm": 1.0389510916300377,
      "learning_rate": 0.00019829991103885268,
      "loss": 14.6331,
      "step": 1991
    },
    {
      "epoch": 0.34832786885245903,
      "grad_norm": 1.159999623880298,
      "learning_rate": 0.0001982973099683902,
      "loss": 14.3747,
      "step": 1992
    },
    {
      "epoch": 0.34850273224043715,
      "grad_norm": 1.1413321023283325,
      "learning_rate": 0.00019829470692676483,
      "loss": 14.1945,
      "step": 1993
    },
    {
      "epoch": 0.3486775956284153,
      "grad_norm": 1.1979054483525884,
      "learning_rate": 0.0001982921019140288,
      "loss": 14.6207,
      "step": 1994
    },
    {
      "epoch": 0.34885245901639345,
      "grad_norm": 1.1316273271132258,
      "learning_rate": 0.00019828949493023429,
      "loss": 14.5419,
      "step": 1995
    },
    {
      "epoch": 0.34902732240437156,
      "grad_norm": 1.213842421272067,
      "learning_rate": 0.0001982868859754337,
      "loss": 14.7934,
      "step": 1996
    },
    {
      "epoch": 0.34920218579234974,
      "grad_norm": 1.1955217597056442,
      "learning_rate": 0.0001982842750496792,
      "loss": 14.4467,
      "step": 1997
    },
    {
      "epoch": 0.34937704918032786,
      "grad_norm": 1.1585799113926976,
      "learning_rate": 0.00019828166215302322,
      "loss": 14.6249,
      "step": 1998
    },
    {
      "epoch": 0.34955191256830603,
      "grad_norm": 1.1650052986945594,
      "learning_rate": 0.00019827904728551816,
      "loss": 14.4032,
      "step": 1999
    },
    {
      "epoch": 0.34972677595628415,
      "grad_norm": 1.496819603183941,
      "learning_rate": 0.00019827643044721644,
      "loss": 14.9382,
      "step": 2000
    },
    {
      "epoch": 0.34990163934426227,
      "grad_norm": 1.1287369255605217,
      "learning_rate": 0.00019827381163817055,
      "loss": 14.6407,
      "step": 2001
    },
    {
      "epoch": 0.35007650273224045,
      "grad_norm": 1.1050248915769243,
      "learning_rate": 0.00019827119085843298,
      "loss": 14.3575,
      "step": 2002
    },
    {
      "epoch": 0.35025136612021857,
      "grad_norm": 1.2122879545273872,
      "learning_rate": 0.00019826856810805632,
      "loss": 14.5225,
      "step": 2003
    },
    {
      "epoch": 0.35042622950819674,
      "grad_norm": 1.1001823675684952,
      "learning_rate": 0.00019826594338709315,
      "loss": 14.4176,
      "step": 2004
    },
    {
      "epoch": 0.35060109289617486,
      "grad_norm": 1.1225874704480754,
      "learning_rate": 0.00019826331669559608,
      "loss": 14.6701,
      "step": 2005
    },
    {
      "epoch": 0.350775956284153,
      "grad_norm": 1.1698724038729733,
      "learning_rate": 0.00019826068803361778,
      "loss": 14.6032,
      "step": 2006
    },
    {
      "epoch": 0.35095081967213115,
      "grad_norm": 1.1787990225764706,
      "learning_rate": 0.00019825805740121105,
      "loss": 14.2671,
      "step": 2007
    },
    {
      "epoch": 0.3511256830601093,
      "grad_norm": 1.126467866237879,
      "learning_rate": 0.00019825542479842855,
      "loss": 14.3922,
      "step": 2008
    },
    {
      "epoch": 0.35130054644808745,
      "grad_norm": 1.1801566187541106,
      "learning_rate": 0.00019825279022532306,
      "loss": 14.8742,
      "step": 2009
    },
    {
      "epoch": 0.35147540983606557,
      "grad_norm": 1.1416901437071498,
      "learning_rate": 0.00019825015368194748,
      "loss": 14.6395,
      "step": 2010
    },
    {
      "epoch": 0.35165027322404374,
      "grad_norm": 1.1958989633827881,
      "learning_rate": 0.00019824751516835467,
      "loss": 14.6624,
      "step": 2011
    },
    {
      "epoch": 0.35182513661202186,
      "grad_norm": 1.220281106854509,
      "learning_rate": 0.00019824487468459752,
      "loss": 13.8363,
      "step": 2012
    },
    {
      "epoch": 0.352,
      "grad_norm": 1.1520341511105143,
      "learning_rate": 0.00019824223223072895,
      "loss": 14.3811,
      "step": 2013
    },
    {
      "epoch": 0.35217486338797815,
      "grad_norm": 1.0907501455380912,
      "learning_rate": 0.000198239587806802,
      "loss": 14.435,
      "step": 2014
    },
    {
      "epoch": 0.3523497267759563,
      "grad_norm": 1.175127491266462,
      "learning_rate": 0.00019823694141286966,
      "loss": 14.743,
      "step": 2015
    },
    {
      "epoch": 0.35252459016393445,
      "grad_norm": 1.103363828677532,
      "learning_rate": 0.00019823429304898505,
      "loss": 14.3331,
      "step": 2016
    },
    {
      "epoch": 0.35269945355191257,
      "grad_norm": 1.0957783954896272,
      "learning_rate": 0.0001982316427152012,
      "loss": 14.2694,
      "step": 2017
    },
    {
      "epoch": 0.3528743169398907,
      "grad_norm": 1.0667330713205052,
      "learning_rate": 0.00019822899041157136,
      "loss": 14.4814,
      "step": 2018
    },
    {
      "epoch": 0.35304918032786886,
      "grad_norm": 1.1224344818511514,
      "learning_rate": 0.00019822633613814862,
      "loss": 14.6571,
      "step": 2019
    },
    {
      "epoch": 0.353224043715847,
      "grad_norm": 1.2190276864258756,
      "learning_rate": 0.00019822367989498626,
      "loss": 14.4442,
      "step": 2020
    },
    {
      "epoch": 0.35339890710382516,
      "grad_norm": 1.1738926560502352,
      "learning_rate": 0.00019822102168213753,
      "loss": 14.1634,
      "step": 2021
    },
    {
      "epoch": 0.3535737704918033,
      "grad_norm": 1.1823350633539755,
      "learning_rate": 0.00019821836149965574,
      "loss": 14.53,
      "step": 2022
    },
    {
      "epoch": 0.3537486338797814,
      "grad_norm": 1.166786643243264,
      "learning_rate": 0.00019821569934759422,
      "loss": 14.5136,
      "step": 2023
    },
    {
      "epoch": 0.35392349726775957,
      "grad_norm": 1.1398361143627447,
      "learning_rate": 0.00019821303522600636,
      "loss": 14.5644,
      "step": 2024
    },
    {
      "epoch": 0.3540983606557377,
      "grad_norm": 1.1116873699375183,
      "learning_rate": 0.0001982103691349456,
      "loss": 14.5127,
      "step": 2025
    },
    {
      "epoch": 0.35427322404371586,
      "grad_norm": 1.199443237662345,
      "learning_rate": 0.0001982077010744654,
      "loss": 14.4538,
      "step": 2026
    },
    {
      "epoch": 0.354448087431694,
      "grad_norm": 1.1064674756174773,
      "learning_rate": 0.00019820503104461926,
      "loss": 14.3571,
      "step": 2027
    },
    {
      "epoch": 0.35462295081967216,
      "grad_norm": 1.2094383084643792,
      "learning_rate": 0.0001982023590454607,
      "loss": 14.357,
      "step": 2028
    },
    {
      "epoch": 0.3547978142076503,
      "grad_norm": 1.0907819051162233,
      "learning_rate": 0.00019819968507704333,
      "loss": 14.4781,
      "step": 2029
    },
    {
      "epoch": 0.3549726775956284,
      "grad_norm": 1.155155314883403,
      "learning_rate": 0.00019819700913942077,
      "loss": 14.3284,
      "step": 2030
    },
    {
      "epoch": 0.35514754098360657,
      "grad_norm": 1.2068301047492624,
      "learning_rate": 0.00019819433123264668,
      "loss": 14.5868,
      "step": 2031
    },
    {
      "epoch": 0.3553224043715847,
      "grad_norm": 1.2170422128192344,
      "learning_rate": 0.00019819165135677474,
      "loss": 14.5395,
      "step": 2032
    },
    {
      "epoch": 0.35549726775956286,
      "grad_norm": 1.1810325431879949,
      "learning_rate": 0.0001981889695118587,
      "loss": 14.7503,
      "step": 2033
    },
    {
      "epoch": 0.355672131147541,
      "grad_norm": 1.2586719946686218,
      "learning_rate": 0.00019818628569795235,
      "loss": 14.3603,
      "step": 2034
    },
    {
      "epoch": 0.3558469945355191,
      "grad_norm": 1.1786945888640479,
      "learning_rate": 0.00019818359991510947,
      "loss": 14.6476,
      "step": 2035
    },
    {
      "epoch": 0.3560218579234973,
      "grad_norm": 1.1871547556383597,
      "learning_rate": 0.00019818091216338397,
      "loss": 14.325,
      "step": 2036
    },
    {
      "epoch": 0.3561967213114754,
      "grad_norm": 1.1969020326023811,
      "learning_rate": 0.00019817822244282973,
      "loss": 14.5212,
      "step": 2037
    },
    {
      "epoch": 0.35637158469945357,
      "grad_norm": 1.2173592860939106,
      "learning_rate": 0.0001981755307535007,
      "loss": 14.4305,
      "step": 2038
    },
    {
      "epoch": 0.3565464480874317,
      "grad_norm": 1.2306703485997612,
      "learning_rate": 0.0001981728370954508,
      "loss": 14.6477,
      "step": 2039
    },
    {
      "epoch": 0.35672131147540986,
      "grad_norm": 1.1708291338521888,
      "learning_rate": 0.00019817014146873412,
      "loss": 14.5821,
      "step": 2040
    },
    {
      "epoch": 0.356896174863388,
      "grad_norm": 1.258340736728308,
      "learning_rate": 0.00019816744387340465,
      "loss": 14.4487,
      "step": 2041
    },
    {
      "epoch": 0.3570710382513661,
      "grad_norm": 1.260070010288263,
      "learning_rate": 0.00019816474430951655,
      "loss": 14.504,
      "step": 2042
    },
    {
      "epoch": 0.3572459016393443,
      "grad_norm": 1.147335697671897,
      "learning_rate": 0.0001981620427771239,
      "loss": 14.1807,
      "step": 2043
    },
    {
      "epoch": 0.3574207650273224,
      "grad_norm": 1.2198490638418382,
      "learning_rate": 0.00019815933927628092,
      "loss": 14.4614,
      "step": 2044
    },
    {
      "epoch": 0.35759562841530057,
      "grad_norm": 1.2261476714379214,
      "learning_rate": 0.00019815663380704177,
      "loss": 14.4197,
      "step": 2045
    },
    {
      "epoch": 0.3577704918032787,
      "grad_norm": 1.1305653824807724,
      "learning_rate": 0.00019815392636946073,
      "loss": 14.1985,
      "step": 2046
    },
    {
      "epoch": 0.3579453551912568,
      "grad_norm": 1.2360782919800124,
      "learning_rate": 0.00019815121696359212,
      "loss": 14.5462,
      "step": 2047
    },
    {
      "epoch": 0.358120218579235,
      "grad_norm": 1.298023433389604,
      "learning_rate": 0.00019814850558949024,
      "loss": 14.5552,
      "step": 2048
    },
    {
      "epoch": 0.3582950819672131,
      "grad_norm": 1.2737456921530048,
      "learning_rate": 0.00019814579224720945,
      "loss": 14.5088,
      "step": 2049
    },
    {
      "epoch": 0.3584699453551913,
      "grad_norm": 1.1227539251579546,
      "learning_rate": 0.0001981430769368042,
      "loss": 14.6517,
      "step": 2050
    },
    {
      "epoch": 0.3586448087431694,
      "grad_norm": 1.1759088406004907,
      "learning_rate": 0.00019814035965832889,
      "loss": 14.5846,
      "step": 2051
    },
    {
      "epoch": 0.3588196721311475,
      "grad_norm": 1.2819235318763682,
      "learning_rate": 0.00019813764041183807,
      "loss": 14.7627,
      "step": 2052
    },
    {
      "epoch": 0.3589945355191257,
      "grad_norm": 1.052466023121554,
      "learning_rate": 0.0001981349191973862,
      "loss": 14.5163,
      "step": 2053
    },
    {
      "epoch": 0.3591693989071038,
      "grad_norm": 1.1549703694391638,
      "learning_rate": 0.00019813219601502792,
      "loss": 14.5601,
      "step": 2054
    },
    {
      "epoch": 0.359344262295082,
      "grad_norm": 1.1314181949759436,
      "learning_rate": 0.0001981294708648178,
      "loss": 14.3784,
      "step": 2055
    },
    {
      "epoch": 0.3595191256830601,
      "grad_norm": 1.0919111088725644,
      "learning_rate": 0.0001981267437468105,
      "loss": 14.2381,
      "step": 2056
    },
    {
      "epoch": 0.3596939890710383,
      "grad_norm": 1.2339857174538136,
      "learning_rate": 0.00019812401466106066,
      "loss": 14.6577,
      "step": 2057
    },
    {
      "epoch": 0.3598688524590164,
      "grad_norm": 1.1933821809232639,
      "learning_rate": 0.00019812128360762309,
      "loss": 14.5059,
      "step": 2058
    },
    {
      "epoch": 0.3600437158469945,
      "grad_norm": 1.0708860843735437,
      "learning_rate": 0.00019811855058655245,
      "loss": 14.3009,
      "step": 2059
    },
    {
      "epoch": 0.3602185792349727,
      "grad_norm": 1.185627641812873,
      "learning_rate": 0.00019811581559790368,
      "loss": 14.4309,
      "step": 2060
    },
    {
      "epoch": 0.3603934426229508,
      "grad_norm": 1.191947380504678,
      "learning_rate": 0.00019811307864173153,
      "loss": 14.5829,
      "step": 2061
    },
    {
      "epoch": 0.360568306010929,
      "grad_norm": 1.260522730959767,
      "learning_rate": 0.00019811033971809088,
      "loss": 14.2674,
      "step": 2062
    },
    {
      "epoch": 0.3607431693989071,
      "grad_norm": 1.248208532925893,
      "learning_rate": 0.00019810759882703672,
      "loss": 14.6929,
      "step": 2063
    },
    {
      "epoch": 0.3609180327868852,
      "grad_norm": 1.1089568550834834,
      "learning_rate": 0.00019810485596862392,
      "loss": 14.2203,
      "step": 2064
    },
    {
      "epoch": 0.3610928961748634,
      "grad_norm": 1.246179520279392,
      "learning_rate": 0.0001981021111429076,
      "loss": 14.3669,
      "step": 2065
    },
    {
      "epoch": 0.3612677595628415,
      "grad_norm": 1.230812407635015,
      "learning_rate": 0.0001980993643499427,
      "loss": 14.5688,
      "step": 2066
    },
    {
      "epoch": 0.3614426229508197,
      "grad_norm": 1.114982335877126,
      "learning_rate": 0.00019809661558978432,
      "loss": 14.29,
      "step": 2067
    },
    {
      "epoch": 0.3616174863387978,
      "grad_norm": 1.2282211843779387,
      "learning_rate": 0.00019809386486248763,
      "loss": 14.579,
      "step": 2068
    },
    {
      "epoch": 0.36179234972677593,
      "grad_norm": 1.1543633130345645,
      "learning_rate": 0.00019809111216810777,
      "loss": 14.5766,
      "step": 2069
    },
    {
      "epoch": 0.3619672131147541,
      "grad_norm": 1.1724692370250587,
      "learning_rate": 0.0001980883575066999,
      "loss": 14.2931,
      "step": 2070
    },
    {
      "epoch": 0.3621420765027322,
      "grad_norm": 1.14890169951871,
      "learning_rate": 0.00019808560087831932,
      "loss": 14.549,
      "step": 2071
    },
    {
      "epoch": 0.3623169398907104,
      "grad_norm": 1.0997479336526743,
      "learning_rate": 0.0001980828422830213,
      "loss": 14.3714,
      "step": 2072
    },
    {
      "epoch": 0.3624918032786885,
      "grad_norm": 1.2456109635765205,
      "learning_rate": 0.0001980800817208611,
      "loss": 14.3984,
      "step": 2073
    },
    {
      "epoch": 0.3626666666666667,
      "grad_norm": 1.051893514719568,
      "learning_rate": 0.00019807731919189414,
      "loss": 14.1221,
      "step": 2074
    },
    {
      "epoch": 0.3628415300546448,
      "grad_norm": 1.1297100814392667,
      "learning_rate": 0.00019807455469617575,
      "loss": 14.1945,
      "step": 2075
    },
    {
      "epoch": 0.36301639344262293,
      "grad_norm": 1.2172743671392148,
      "learning_rate": 0.00019807178823376146,
      "loss": 14.3289,
      "step": 2076
    },
    {
      "epoch": 0.3631912568306011,
      "grad_norm": 1.1543244175406207,
      "learning_rate": 0.00019806901980470667,
      "loss": 14.2314,
      "step": 2077
    },
    {
      "epoch": 0.3633661202185792,
      "grad_norm": 1.0848057113649867,
      "learning_rate": 0.00019806624940906694,
      "loss": 14.36,
      "step": 2078
    },
    {
      "epoch": 0.3635409836065574,
      "grad_norm": 1.1601875834531086,
      "learning_rate": 0.00019806347704689778,
      "loss": 14.2966,
      "step": 2079
    },
    {
      "epoch": 0.3637158469945355,
      "grad_norm": 1.1572049816741787,
      "learning_rate": 0.00019806070271825483,
      "loss": 14.4309,
      "step": 2080
    },
    {
      "epoch": 0.36389071038251364,
      "grad_norm": 1.065319837901196,
      "learning_rate": 0.00019805792642319368,
      "loss": 14.0555,
      "step": 2081
    },
    {
      "epoch": 0.3640655737704918,
      "grad_norm": 1.2771271689864296,
      "learning_rate": 0.00019805514816177006,
      "loss": 14.2621,
      "step": 2082
    },
    {
      "epoch": 0.36424043715846993,
      "grad_norm": 1.1499755725862864,
      "learning_rate": 0.00019805236793403962,
      "loss": 14.2608,
      "step": 2083
    },
    {
      "epoch": 0.3644153005464481,
      "grad_norm": 1.1560000018677865,
      "learning_rate": 0.00019804958574005818,
      "loss": 14.3551,
      "step": 2084
    },
    {
      "epoch": 0.3645901639344262,
      "grad_norm": 1.2467451946723833,
      "learning_rate": 0.0001980468015798815,
      "loss": 14.4633,
      "step": 2085
    },
    {
      "epoch": 0.36476502732240434,
      "grad_norm": 1.2207353342029348,
      "learning_rate": 0.00019804401545356535,
      "loss": 14.0525,
      "step": 2086
    },
    {
      "epoch": 0.3649398907103825,
      "grad_norm": 1.2152290340679532,
      "learning_rate": 0.00019804122736116568,
      "loss": 14.0911,
      "step": 2087
    },
    {
      "epoch": 0.36511475409836064,
      "grad_norm": 1.1312028882383078,
      "learning_rate": 0.00019803843730273842,
      "loss": 14.3663,
      "step": 2088
    },
    {
      "epoch": 0.3652896174863388,
      "grad_norm": 1.288016708220445,
      "learning_rate": 0.0001980356452783394,
      "loss": 14.3852,
      "step": 2089
    },
    {
      "epoch": 0.36546448087431693,
      "grad_norm": 1.2138826595647072,
      "learning_rate": 0.00019803285128802472,
      "loss": 14.4097,
      "step": 2090
    },
    {
      "epoch": 0.3656393442622951,
      "grad_norm": 1.1679242924217685,
      "learning_rate": 0.00019803005533185038,
      "loss": 14.2689,
      "step": 2091
    },
    {
      "epoch": 0.3658142076502732,
      "grad_norm": 1.2946531114542794,
      "learning_rate": 0.00019802725740987242,
      "loss": 14.4184,
      "step": 2092
    },
    {
      "epoch": 0.36598907103825135,
      "grad_norm": 1.0599441120626436,
      "learning_rate": 0.000198024457522147,
      "loss": 14.2507,
      "step": 2093
    },
    {
      "epoch": 0.3661639344262295,
      "grad_norm": 1.334992456412743,
      "learning_rate": 0.00019802165566873019,
      "loss": 14.3512,
      "step": 2094
    },
    {
      "epoch": 0.36633879781420764,
      "grad_norm": 1.111275001395144,
      "learning_rate": 0.00019801885184967823,
      "loss": 14.1093,
      "step": 2095
    },
    {
      "epoch": 0.3665136612021858,
      "grad_norm": 1.2463050211023645,
      "learning_rate": 0.00019801604606504736,
      "loss": 14.4043,
      "step": 2096
    },
    {
      "epoch": 0.36668852459016393,
      "grad_norm": 1.0873792329305252,
      "learning_rate": 0.00019801323831489378,
      "loss": 14.4394,
      "step": 2097
    },
    {
      "epoch": 0.36686338797814205,
      "grad_norm": 1.2588362215866933,
      "learning_rate": 0.00019801042859927384,
      "loss": 14.4207,
      "step": 2098
    },
    {
      "epoch": 0.3670382513661202,
      "grad_norm": 1.0891599494866129,
      "learning_rate": 0.00019800761691824388,
      "loss": 14.2548,
      "step": 2099
    },
    {
      "epoch": 0.36721311475409835,
      "grad_norm": 1.2021020502521194,
      "learning_rate": 0.0001980048032718603,
      "loss": 14.2729,
      "step": 2100
    },
    {
      "epoch": 0.3673879781420765,
      "grad_norm": 1.2251304521097317,
      "learning_rate": 0.00019800198766017948,
      "loss": 14.264,
      "step": 2101
    },
    {
      "epoch": 0.36756284153005464,
      "grad_norm": 1.202472090991517,
      "learning_rate": 0.0001979991700832579,
      "loss": 14.4968,
      "step": 2102
    },
    {
      "epoch": 0.36773770491803276,
      "grad_norm": 1.1439276348972305,
      "learning_rate": 0.00019799635054115205,
      "loss": 14.344,
      "step": 2103
    },
    {
      "epoch": 0.36791256830601093,
      "grad_norm": 1.159583350204158,
      "learning_rate": 0.0001979935290339185,
      "loss": 14.1929,
      "step": 2104
    },
    {
      "epoch": 0.36808743169398905,
      "grad_norm": 1.0446799240435252,
      "learning_rate": 0.00019799070556161382,
      "loss": 14.313,
      "step": 2105
    },
    {
      "epoch": 0.36826229508196723,
      "grad_norm": 1.1694355057577992,
      "learning_rate": 0.0001979878801242946,
      "loss": 14.2656,
      "step": 2106
    },
    {
      "epoch": 0.36843715846994535,
      "grad_norm": 1.054838890194209,
      "learning_rate": 0.00019798505272201756,
      "loss": 14.0836,
      "step": 2107
    },
    {
      "epoch": 0.3686120218579235,
      "grad_norm": 1.2547096044497463,
      "learning_rate": 0.00019798222335483932,
      "loss": 14.4074,
      "step": 2108
    },
    {
      "epoch": 0.36878688524590164,
      "grad_norm": 1.2248732486548661,
      "learning_rate": 0.00019797939202281664,
      "loss": 14.1396,
      "step": 2109
    },
    {
      "epoch": 0.36896174863387976,
      "grad_norm": 1.1176335768522148,
      "learning_rate": 0.00019797655872600637,
      "loss": 14.2574,
      "step": 2110
    },
    {
      "epoch": 0.36913661202185793,
      "grad_norm": 1.0839930276490877,
      "learning_rate": 0.00019797372346446523,
      "loss": 14.1877,
      "step": 2111
    },
    {
      "epoch": 0.36931147540983605,
      "grad_norm": 1.1383481267300908,
      "learning_rate": 0.00019797088623825016,
      "loss": 14.3319,
      "step": 2112
    },
    {
      "epoch": 0.36948633879781423,
      "grad_norm": 1.1883598932623167,
      "learning_rate": 0.00019796804704741796,
      "loss": 14.3953,
      "step": 2113
    },
    {
      "epoch": 0.36966120218579235,
      "grad_norm": 1.13486260058014,
      "learning_rate": 0.00019796520589202563,
      "loss": 14.4281,
      "step": 2114
    },
    {
      "epoch": 0.36983606557377047,
      "grad_norm": 1.0475935250216375,
      "learning_rate": 0.00019796236277213014,
      "loss": 14.1379,
      "step": 2115
    },
    {
      "epoch": 0.37001092896174864,
      "grad_norm": 1.250811412143362,
      "learning_rate": 0.0001979595176877885,
      "loss": 14.4709,
      "step": 2116
    },
    {
      "epoch": 0.37018579234972676,
      "grad_norm": 1.0477773711982914,
      "learning_rate": 0.00019795667063905774,
      "loss": 14.2243,
      "step": 2117
    },
    {
      "epoch": 0.37036065573770494,
      "grad_norm": 1.1280618730809229,
      "learning_rate": 0.00019795382162599495,
      "loss": 14.4973,
      "step": 2118
    },
    {
      "epoch": 0.37053551912568305,
      "grad_norm": 1.1369948637813212,
      "learning_rate": 0.0001979509706486573,
      "loss": 14.2248,
      "step": 2119
    },
    {
      "epoch": 0.37071038251366123,
      "grad_norm": 1.0862508947066118,
      "learning_rate": 0.00019794811770710194,
      "loss": 14.3241,
      "step": 2120
    },
    {
      "epoch": 0.37088524590163935,
      "grad_norm": 1.2751997636095638,
      "learning_rate": 0.00019794526280138608,
      "loss": 14.4232,
      "step": 2121
    },
    {
      "epoch": 0.37106010928961747,
      "grad_norm": 1.0821483744777596,
      "learning_rate": 0.00019794240593156698,
      "loss": 13.9941,
      "step": 2122
    },
    {
      "epoch": 0.37123497267759564,
      "grad_norm": 1.0913845994593654,
      "learning_rate": 0.00019793954709770192,
      "loss": 14.3868,
      "step": 2123
    },
    {
      "epoch": 0.37140983606557376,
      "grad_norm": 1.2190190943590282,
      "learning_rate": 0.00019793668629984818,
      "loss": 14.3844,
      "step": 2124
    },
    {
      "epoch": 0.37158469945355194,
      "grad_norm": 1.0883902083523145,
      "learning_rate": 0.00019793382353806323,
      "loss": 14.4154,
      "step": 2125
    },
    {
      "epoch": 0.37175956284153006,
      "grad_norm": 1.2357195418730544,
      "learning_rate": 0.0001979309588124044,
      "loss": 14.3819,
      "step": 2126
    },
    {
      "epoch": 0.3719344262295082,
      "grad_norm": 1.1643575471266507,
      "learning_rate": 0.00019792809212292912,
      "loss": 14.4191,
      "step": 2127
    },
    {
      "epoch": 0.37210928961748635,
      "grad_norm": 1.13074546871809,
      "learning_rate": 0.00019792522346969498,
      "loss": 14.3541,
      "step": 2128
    },
    {
      "epoch": 0.37228415300546447,
      "grad_norm": 1.0924802028420668,
      "learning_rate": 0.0001979223528527594,
      "loss": 14.3711,
      "step": 2129
    },
    {
      "epoch": 0.37245901639344264,
      "grad_norm": 1.1296585878023655,
      "learning_rate": 0.00019791948027218,
      "loss": 14.2152,
      "step": 2130
    },
    {
      "epoch": 0.37263387978142076,
      "grad_norm": 1.1406051610879855,
      "learning_rate": 0.00019791660572801437,
      "loss": 14.2466,
      "step": 2131
    },
    {
      "epoch": 0.3728087431693989,
      "grad_norm": 1.1743261937062623,
      "learning_rate": 0.00019791372922032014,
      "loss": 14.4967,
      "step": 2132
    },
    {
      "epoch": 0.37298360655737706,
      "grad_norm": 1.1489253398033934,
      "learning_rate": 0.000197910850749155,
      "loss": 14.2821,
      "step": 2133
    },
    {
      "epoch": 0.3731584699453552,
      "grad_norm": 1.3025198557567854,
      "learning_rate": 0.0001979079703145767,
      "loss": 14.3939,
      "step": 2134
    },
    {
      "epoch": 0.37333333333333335,
      "grad_norm": 1.1128837947607193,
      "learning_rate": 0.00019790508791664294,
      "loss": 14.4696,
      "step": 2135
    },
    {
      "epoch": 0.37350819672131147,
      "grad_norm": 1.0736061784739415,
      "learning_rate": 0.0001979022035554116,
      "loss": 14.2837,
      "step": 2136
    },
    {
      "epoch": 0.37368306010928964,
      "grad_norm": 1.1475954436644555,
      "learning_rate": 0.00019789931723094046,
      "loss": 14.1548,
      "step": 2137
    },
    {
      "epoch": 0.37385792349726776,
      "grad_norm": 1.110083997468402,
      "learning_rate": 0.00019789642894328741,
      "loss": 13.9366,
      "step": 2138
    },
    {
      "epoch": 0.3740327868852459,
      "grad_norm": 1.1308039994036345,
      "learning_rate": 0.0001978935386925104,
      "loss": 14.3195,
      "step": 2139
    },
    {
      "epoch": 0.37420765027322406,
      "grad_norm": 1.149901595572649,
      "learning_rate": 0.00019789064647866736,
      "loss": 13.9427,
      "step": 2140
    },
    {
      "epoch": 0.3743825136612022,
      "grad_norm": 1.2731465281237095,
      "learning_rate": 0.0001978877523018163,
      "loss": 14.332,
      "step": 2141
    },
    {
      "epoch": 0.37455737704918035,
      "grad_norm": 1.1693322430202242,
      "learning_rate": 0.00019788485616201523,
      "loss": 14.3214,
      "step": 2142
    },
    {
      "epoch": 0.37473224043715847,
      "grad_norm": 1.0790829425554576,
      "learning_rate": 0.00019788195805932226,
      "loss": 14.159,
      "step": 2143
    },
    {
      "epoch": 0.3749071038251366,
      "grad_norm": 1.1772177496515603,
      "learning_rate": 0.0001978790579937955,
      "loss": 14.5746,
      "step": 2144
    },
    {
      "epoch": 0.37508196721311476,
      "grad_norm": 1.166431163063748,
      "learning_rate": 0.00019787615596549308,
      "loss": 14.3547,
      "step": 2145
    },
    {
      "epoch": 0.3752568306010929,
      "grad_norm": 1.1854677109964329,
      "learning_rate": 0.00019787325197447324,
      "loss": 14.4774,
      "step": 2146
    },
    {
      "epoch": 0.37543169398907106,
      "grad_norm": 1.175210602471288,
      "learning_rate": 0.00019787034602079419,
      "loss": 14.4605,
      "step": 2147
    },
    {
      "epoch": 0.3756065573770492,
      "grad_norm": 1.0631045906410121,
      "learning_rate": 0.00019786743810451415,
      "loss": 14.2081,
      "step": 2148
    },
    {
      "epoch": 0.3757814207650273,
      "grad_norm": 1.07697187354787,
      "learning_rate": 0.00019786452822569156,
      "loss": 14.6447,
      "step": 2149
    },
    {
      "epoch": 0.37595628415300547,
      "grad_norm": 1.11078579054756,
      "learning_rate": 0.00019786161638438464,
      "loss": 14.3123,
      "step": 2150
    },
    {
      "epoch": 0.3761311475409836,
      "grad_norm": 1.076062795840406,
      "learning_rate": 0.00019785870258065183,
      "loss": 14.3914,
      "step": 2151
    },
    {
      "epoch": 0.37630601092896176,
      "grad_norm": 1.035827574576478,
      "learning_rate": 0.0001978557868145516,
      "loss": 14.2974,
      "step": 2152
    },
    {
      "epoch": 0.3764808743169399,
      "grad_norm": 1.0973819110757361,
      "learning_rate": 0.00019785286908614237,
      "loss": 13.8603,
      "step": 2153
    },
    {
      "epoch": 0.37665573770491806,
      "grad_norm": 1.0863538004225481,
      "learning_rate": 0.00019784994939548266,
      "loss": 14.402,
      "step": 2154
    },
    {
      "epoch": 0.3768306010928962,
      "grad_norm": 1.0508616563425768,
      "learning_rate": 0.00019784702774263107,
      "loss": 14.0948,
      "step": 2155
    },
    {
      "epoch": 0.3770054644808743,
      "grad_norm": 1.1242930826803252,
      "learning_rate": 0.00019784410412764612,
      "loss": 14.1198,
      "step": 2156
    },
    {
      "epoch": 0.37718032786885247,
      "grad_norm": 1.1600186443228404,
      "learning_rate": 0.00019784117855058643,
      "loss": 14.218,
      "step": 2157
    },
    {
      "epoch": 0.3773551912568306,
      "grad_norm": 1.1910901734397399,
      "learning_rate": 0.00019783825101151074,
      "loss": 14.5572,
      "step": 2158
    },
    {
      "epoch": 0.37753005464480877,
      "grad_norm": 1.1876544287190731,
      "learning_rate": 0.0001978353215104777,
      "loss": 14.0718,
      "step": 2159
    },
    {
      "epoch": 0.3777049180327869,
      "grad_norm": 1.077502221916152,
      "learning_rate": 0.00019783239004754605,
      "loss": 14.3335,
      "step": 2160
    },
    {
      "epoch": 0.377879781420765,
      "grad_norm": 1.2362026822289371,
      "learning_rate": 0.00019782945662277462,
      "loss": 14.4581,
      "step": 2161
    },
    {
      "epoch": 0.3780546448087432,
      "grad_norm": 1.1789530242485782,
      "learning_rate": 0.0001978265212362222,
      "loss": 14.2793,
      "step": 2162
    },
    {
      "epoch": 0.3782295081967213,
      "grad_norm": 1.2651378118361045,
      "learning_rate": 0.00019782358388794767,
      "loss": 14.5307,
      "step": 2163
    },
    {
      "epoch": 0.3784043715846995,
      "grad_norm": 1.1282087838299615,
      "learning_rate": 0.00019782064457800992,
      "loss": 14.1895,
      "step": 2164
    },
    {
      "epoch": 0.3785792349726776,
      "grad_norm": 1.2468976785574464,
      "learning_rate": 0.0001978177033064679,
      "loss": 14.2775,
      "step": 2165
    },
    {
      "epoch": 0.3787540983606557,
      "grad_norm": 1.0211873781413552,
      "learning_rate": 0.00019781476007338058,
      "loss": 14.3623,
      "step": 2166
    },
    {
      "epoch": 0.3789289617486339,
      "grad_norm": 1.1222313638924415,
      "learning_rate": 0.00019781181487880698,
      "loss": 14.2468,
      "step": 2167
    },
    {
      "epoch": 0.379103825136612,
      "grad_norm": 1.160765353944954,
      "learning_rate": 0.00019780886772280618,
      "loss": 14.2587,
      "step": 2168
    },
    {
      "epoch": 0.3792786885245902,
      "grad_norm": 1.0933872355718937,
      "learning_rate": 0.0001978059186054373,
      "loss": 14.3566,
      "step": 2169
    },
    {
      "epoch": 0.3794535519125683,
      "grad_norm": 1.1684901753400416,
      "learning_rate": 0.0001978029675267594,
      "loss": 14.2388,
      "step": 2170
    },
    {
      "epoch": 0.3796284153005465,
      "grad_norm": 1.1103389002120239,
      "learning_rate": 0.00019780001448683175,
      "loss": 14.2398,
      "step": 2171
    },
    {
      "epoch": 0.3798032786885246,
      "grad_norm": 1.2068947493550506,
      "learning_rate": 0.00019779705948571346,
      "loss": 14.1997,
      "step": 2172
    },
    {
      "epoch": 0.3799781420765027,
      "grad_norm": 1.1346303334149639,
      "learning_rate": 0.00019779410252346392,
      "loss": 14.3575,
      "step": 2173
    },
    {
      "epoch": 0.3801530054644809,
      "grad_norm": 1.0938240434738915,
      "learning_rate": 0.00019779114360014233,
      "loss": 14.046,
      "step": 2174
    },
    {
      "epoch": 0.380327868852459,
      "grad_norm": 1.3293773636167134,
      "learning_rate": 0.00019778818271580804,
      "loss": 14.2745,
      "step": 2175
    },
    {
      "epoch": 0.3805027322404372,
      "grad_norm": 1.2241897835964295,
      "learning_rate": 0.0001977852198705205,
      "loss": 14.2889,
      "step": 2176
    },
    {
      "epoch": 0.3806775956284153,
      "grad_norm": 1.1443994198915233,
      "learning_rate": 0.00019778225506433898,
      "loss": 14.1595,
      "step": 2177
    },
    {
      "epoch": 0.3808524590163934,
      "grad_norm": 1.1621730268291803,
      "learning_rate": 0.00019777928829732306,
      "loss": 14.2915,
      "step": 2178
    },
    {
      "epoch": 0.3810273224043716,
      "grad_norm": 1.0692529277728358,
      "learning_rate": 0.00019777631956953217,
      "loss": 14.2895,
      "step": 2179
    },
    {
      "epoch": 0.3812021857923497,
      "grad_norm": 1.1225448144485226,
      "learning_rate": 0.00019777334888102587,
      "loss": 14.1477,
      "step": 2180
    },
    {
      "epoch": 0.3813770491803279,
      "grad_norm": 1.143274932964899,
      "learning_rate": 0.0001977703762318637,
      "loss": 14.1648,
      "step": 2181
    },
    {
      "epoch": 0.381551912568306,
      "grad_norm": 1.0856529381760438,
      "learning_rate": 0.0001977674016221053,
      "loss": 14.4077,
      "step": 2182
    },
    {
      "epoch": 0.3817267759562841,
      "grad_norm": 1.065537345500991,
      "learning_rate": 0.00019776442505181033,
      "loss": 14.3718,
      "step": 2183
    },
    {
      "epoch": 0.3819016393442623,
      "grad_norm": 1.2294984638632396,
      "learning_rate": 0.00019776144652103845,
      "loss": 13.9262,
      "step": 2184
    },
    {
      "epoch": 0.3820765027322404,
      "grad_norm": 1.1080464859264638,
      "learning_rate": 0.0001977584660298494,
      "loss": 14.4304,
      "step": 2185
    },
    {
      "epoch": 0.3822513661202186,
      "grad_norm": 1.189011402733918,
      "learning_rate": 0.00019775548357830295,
      "loss": 14.5697,
      "step": 2186
    },
    {
      "epoch": 0.3824262295081967,
      "grad_norm": 1.2493100847941576,
      "learning_rate": 0.0001977524991664589,
      "loss": 14.4616,
      "step": 2187
    },
    {
      "epoch": 0.3826010928961749,
      "grad_norm": 1.066144863102596,
      "learning_rate": 0.0001977495127943771,
      "loss": 14.2401,
      "step": 2188
    },
    {
      "epoch": 0.382775956284153,
      "grad_norm": 1.2100627815556853,
      "learning_rate": 0.00019774652446211745,
      "loss": 14.3338,
      "step": 2189
    },
    {
      "epoch": 0.3829508196721311,
      "grad_norm": 1.0882033678958563,
      "learning_rate": 0.0001977435341697399,
      "loss": 14.3364,
      "step": 2190
    },
    {
      "epoch": 0.3831256830601093,
      "grad_norm": 1.144071104909405,
      "learning_rate": 0.0001977405419173043,
      "loss": 14.2558,
      "step": 2191
    },
    {
      "epoch": 0.3833005464480874,
      "grad_norm": 1.111783420959502,
      "learning_rate": 0.00019773754770487076,
      "loss": 14.035,
      "step": 2192
    },
    {
      "epoch": 0.3834754098360656,
      "grad_norm": 1.105078431695514,
      "learning_rate": 0.0001977345515324993,
      "loss": 14.4215,
      "step": 2193
    },
    {
      "epoch": 0.3836502732240437,
      "grad_norm": 1.2455904678046155,
      "learning_rate": 0.00019773155340025003,
      "loss": 14.4581,
      "step": 2194
    },
    {
      "epoch": 0.38382513661202183,
      "grad_norm": 1.0572329018773414,
      "learning_rate": 0.000197728553308183,
      "loss": 14.4871,
      "step": 2195
    },
    {
      "epoch": 0.384,
      "grad_norm": 1.1568617835849482,
      "learning_rate": 0.0001977255512563584,
      "loss": 14.4069,
      "step": 2196
    },
    {
      "epoch": 0.3841748633879781,
      "grad_norm": 1.1248179922098187,
      "learning_rate": 0.00019772254724483647,
      "loss": 14.197,
      "step": 2197
    },
    {
      "epoch": 0.3843497267759563,
      "grad_norm": 1.1320328557229788,
      "learning_rate": 0.00019771954127367742,
      "loss": 14.0525,
      "step": 2198
    },
    {
      "epoch": 0.3845245901639344,
      "grad_norm": 1.1546810079681842,
      "learning_rate": 0.00019771653334294152,
      "loss": 14.0041,
      "step": 2199
    },
    {
      "epoch": 0.3846994535519126,
      "grad_norm": 1.0299667925541252,
      "learning_rate": 0.0001977135234526891,
      "loss": 14.1017,
      "step": 2200
    },
    {
      "epoch": 0.3848743169398907,
      "grad_norm": 1.13570038995109,
      "learning_rate": 0.0001977105116029805,
      "loss": 14.1217,
      "step": 2201
    },
    {
      "epoch": 0.38504918032786883,
      "grad_norm": 1.1698867713032683,
      "learning_rate": 0.00019770749779387617,
      "loss": 14.262,
      "step": 2202
    },
    {
      "epoch": 0.385224043715847,
      "grad_norm": 1.0449933649766436,
      "learning_rate": 0.00019770448202543648,
      "loss": 14.1099,
      "step": 2203
    },
    {
      "epoch": 0.38539890710382513,
      "grad_norm": 1.1562366034036586,
      "learning_rate": 0.00019770146429772193,
      "loss": 14.4549,
      "step": 2204
    },
    {
      "epoch": 0.3855737704918033,
      "grad_norm": 1.0980593489366852,
      "learning_rate": 0.00019769844461079304,
      "loss": 14.2322,
      "step": 2205
    },
    {
      "epoch": 0.3857486338797814,
      "grad_norm": 1.03844159578147,
      "learning_rate": 0.00019769542296471036,
      "loss": 14.3775,
      "step": 2206
    },
    {
      "epoch": 0.38592349726775954,
      "grad_norm": 1.0656840621428623,
      "learning_rate": 0.0001976923993595345,
      "loss": 14.1555,
      "step": 2207
    },
    {
      "epoch": 0.3860983606557377,
      "grad_norm": 1.0878982444225578,
      "learning_rate": 0.00019768937379532604,
      "loss": 14.3172,
      "step": 2208
    },
    {
      "epoch": 0.38627322404371583,
      "grad_norm": 1.0897827621251084,
      "learning_rate": 0.00019768634627214573,
      "loss": 14.3884,
      "step": 2209
    },
    {
      "epoch": 0.386448087431694,
      "grad_norm": 0.9694660277081397,
      "learning_rate": 0.00019768331679005424,
      "loss": 14.1624,
      "step": 2210
    },
    {
      "epoch": 0.38662295081967213,
      "grad_norm": 1.1282896251324115,
      "learning_rate": 0.0001976802853491123,
      "loss": 14.2045,
      "step": 2211
    },
    {
      "epoch": 0.38679781420765025,
      "grad_norm": 1.038909064872719,
      "learning_rate": 0.0001976772519493807,
      "loss": 14.2944,
      "step": 2212
    },
    {
      "epoch": 0.3869726775956284,
      "grad_norm": 1.0941734967380716,
      "learning_rate": 0.00019767421659092036,
      "loss": 14.0482,
      "step": 2213
    },
    {
      "epoch": 0.38714754098360654,
      "grad_norm": 0.972019821672408,
      "learning_rate": 0.000197671179273792,
      "loss": 14.1585,
      "step": 2214
    },
    {
      "epoch": 0.3873224043715847,
      "grad_norm": 1.1266501858949407,
      "learning_rate": 0.00019766813999805668,
      "loss": 14.0991,
      "step": 2215
    },
    {
      "epoch": 0.38749726775956284,
      "grad_norm": 1.1730302356545808,
      "learning_rate": 0.00019766509876377523,
      "loss": 14.2812,
      "step": 2216
    },
    {
      "epoch": 0.387672131147541,
      "grad_norm": 1.099905657042338,
      "learning_rate": 0.00019766205557100868,
      "loss": 14.358,
      "step": 2217
    },
    {
      "epoch": 0.38784699453551913,
      "grad_norm": 1.014990144182495,
      "learning_rate": 0.00019765901041981807,
      "loss": 14.4306,
      "step": 2218
    },
    {
      "epoch": 0.38802185792349725,
      "grad_norm": 1.0910981027772348,
      "learning_rate": 0.0001976559633102644,
      "loss": 14.135,
      "step": 2219
    },
    {
      "epoch": 0.3881967213114754,
      "grad_norm": 1.0043978272361966,
      "learning_rate": 0.0001976529142424089,
      "loss": 14.3228,
      "step": 2220
    },
    {
      "epoch": 0.38837158469945354,
      "grad_norm": 1.0398900157362536,
      "learning_rate": 0.00019764986321631258,
      "loss": 13.8111,
      "step": 2221
    },
    {
      "epoch": 0.3885464480874317,
      "grad_norm": 1.0377910494501152,
      "learning_rate": 0.0001976468102320367,
      "loss": 13.8631,
      "step": 2222
    },
    {
      "epoch": 0.38872131147540984,
      "grad_norm": 1.1455862483925159,
      "learning_rate": 0.00019764375528964244,
      "loss": 13.9735,
      "step": 2223
    },
    {
      "epoch": 0.38889617486338796,
      "grad_norm": 1.0003426345861688,
      "learning_rate": 0.0001976406983891911,
      "loss": 14.254,
      "step": 2224
    },
    {
      "epoch": 0.38907103825136613,
      "grad_norm": 1.159642918442637,
      "learning_rate": 0.00019763763953074395,
      "loss": 14.2852,
      "step": 2225
    },
    {
      "epoch": 0.38924590163934425,
      "grad_norm": 1.1058007448404696,
      "learning_rate": 0.00019763457871436235,
      "loss": 14.2746,
      "step": 2226
    },
    {
      "epoch": 0.3894207650273224,
      "grad_norm": 1.0571913820974825,
      "learning_rate": 0.00019763151594010765,
      "loss": 13.7764,
      "step": 2227
    },
    {
      "epoch": 0.38959562841530054,
      "grad_norm": 0.9953471210145025,
      "learning_rate": 0.0001976284512080413,
      "loss": 13.7983,
      "step": 2228
    },
    {
      "epoch": 0.38977049180327866,
      "grad_norm": 1.1212066390277684,
      "learning_rate": 0.00019762538451822476,
      "loss": 14.0731,
      "step": 2229
    },
    {
      "epoch": 0.38994535519125684,
      "grad_norm": 1.212062408901242,
      "learning_rate": 0.0001976223158707195,
      "loss": 14.272,
      "step": 2230
    },
    {
      "epoch": 0.39012021857923496,
      "grad_norm": 1.3144108158854881,
      "learning_rate": 0.00019761924526558706,
      "loss": 14.0842,
      "step": 2231
    },
    {
      "epoch": 0.39029508196721313,
      "grad_norm": 1.1208860742456703,
      "learning_rate": 0.00019761617270288903,
      "loss": 13.9411,
      "step": 2232
    },
    {
      "epoch": 0.39046994535519125,
      "grad_norm": 1.1819267367201047,
      "learning_rate": 0.00019761309818268702,
      "loss": 14.2086,
      "step": 2233
    },
    {
      "epoch": 0.3906448087431694,
      "grad_norm": 1.0275541054368904,
      "learning_rate": 0.00019761002170504266,
      "loss": 14.2921,
      "step": 2234
    },
    {
      "epoch": 0.39081967213114754,
      "grad_norm": 1.116834561153894,
      "learning_rate": 0.00019760694327001767,
      "loss": 14.2699,
      "step": 2235
    },
    {
      "epoch": 0.39099453551912566,
      "grad_norm": 1.1457147287441438,
      "learning_rate": 0.00019760386287767377,
      "loss": 14.327,
      "step": 2236
    },
    {
      "epoch": 0.39116939890710384,
      "grad_norm": 1.1000653191667211,
      "learning_rate": 0.00019760078052807276,
      "loss": 14.1307,
      "step": 2237
    },
    {
      "epoch": 0.39134426229508196,
      "grad_norm": 1.2083471418962057,
      "learning_rate": 0.0001975976962212764,
      "loss": 14.1853,
      "step": 2238
    },
    {
      "epoch": 0.39151912568306013,
      "grad_norm": 1.106243542453675,
      "learning_rate": 0.00019759460995734659,
      "loss": 14.2282,
      "step": 2239
    },
    {
      "epoch": 0.39169398907103825,
      "grad_norm": 1.0779310729244285,
      "learning_rate": 0.00019759152173634518,
      "loss": 13.91,
      "step": 2240
    },
    {
      "epoch": 0.39186885245901637,
      "grad_norm": 1.1195971376678555,
      "learning_rate": 0.0001975884315583341,
      "loss": 13.9777,
      "step": 2241
    },
    {
      "epoch": 0.39204371584699454,
      "grad_norm": 1.0700592944338028,
      "learning_rate": 0.00019758533942337535,
      "loss": 13.9911,
      "step": 2242
    },
    {
      "epoch": 0.39221857923497266,
      "grad_norm": 1.0816410091416362,
      "learning_rate": 0.00019758224533153094,
      "loss": 14.3154,
      "step": 2243
    },
    {
      "epoch": 0.39239344262295084,
      "grad_norm": 1.1544964402143214,
      "learning_rate": 0.00019757914928286287,
      "loss": 13.82,
      "step": 2244
    },
    {
      "epoch": 0.39256830601092896,
      "grad_norm": 1.1572488510711494,
      "learning_rate": 0.00019757605127743326,
      "loss": 14.4506,
      "step": 2245
    },
    {
      "epoch": 0.3927431693989071,
      "grad_norm": 1.1925560096227725,
      "learning_rate": 0.00019757295131530423,
      "loss": 14.0534,
      "step": 2246
    },
    {
      "epoch": 0.39291803278688525,
      "grad_norm": 1.1835019286106412,
      "learning_rate": 0.00019756984939653792,
      "loss": 14.0736,
      "step": 2247
    },
    {
      "epoch": 0.39309289617486337,
      "grad_norm": 1.0899623879961005,
      "learning_rate": 0.00019756674552119655,
      "loss": 14.1626,
      "step": 2248
    },
    {
      "epoch": 0.39326775956284155,
      "grad_norm": 1.1951981984189528,
      "learning_rate": 0.00019756363968934237,
      "loss": 13.792,
      "step": 2249
    },
    {
      "epoch": 0.39344262295081966,
      "grad_norm": 1.130822412815803,
      "learning_rate": 0.00019756053190103762,
      "loss": 13.9503,
      "step": 2250
    },
    {
      "epoch": 0.39361748633879784,
      "grad_norm": 1.1384410307614425,
      "learning_rate": 0.00019755742215634472,
      "loss": 13.8228,
      "step": 2251
    },
    {
      "epoch": 0.39379234972677596,
      "grad_norm": 1.1571331443453723,
      "learning_rate": 0.0001975543104553259,
      "loss": 14.0348,
      "step": 2252
    },
    {
      "epoch": 0.3939672131147541,
      "grad_norm": 1.0611352651742292,
      "learning_rate": 0.00019755119679804367,
      "loss": 14.1076,
      "step": 2253
    },
    {
      "epoch": 0.39414207650273225,
      "grad_norm": 1.1870477762829938,
      "learning_rate": 0.0001975480811845604,
      "loss": 14.0953,
      "step": 2254
    },
    {
      "epoch": 0.39431693989071037,
      "grad_norm": 1.2304655572706995,
      "learning_rate": 0.00019754496361493862,
      "loss": 14.295,
      "step": 2255
    },
    {
      "epoch": 0.39449180327868855,
      "grad_norm": 1.1500354504239731,
      "learning_rate": 0.0001975418440892408,
      "loss": 14.3627,
      "step": 2256
    },
    {
      "epoch": 0.39466666666666667,
      "grad_norm": 1.4182568160107452,
      "learning_rate": 0.0001975387226075295,
      "loss": 14.235,
      "step": 2257
    },
    {
      "epoch": 0.3948415300546448,
      "grad_norm": 1.3078796080453925,
      "learning_rate": 0.00019753559916986735,
      "loss": 14.1021,
      "step": 2258
    },
    {
      "epoch": 0.39501639344262296,
      "grad_norm": 1.3202170143639245,
      "learning_rate": 0.00019753247377631694,
      "loss": 14.0305,
      "step": 2259
    },
    {
      "epoch": 0.3951912568306011,
      "grad_norm": 1.5689088272419989,
      "learning_rate": 0.00019752934642694098,
      "loss": 14.0559,
      "step": 2260
    },
    {
      "epoch": 0.39536612021857925,
      "grad_norm": 1.293189955960672,
      "learning_rate": 0.00019752621712180218,
      "loss": 14.0638,
      "step": 2261
    },
    {
      "epoch": 0.3955409836065574,
      "grad_norm": 1.331804146085159,
      "learning_rate": 0.00019752308586096326,
      "loss": 14.1336,
      "step": 2262
    },
    {
      "epoch": 0.3957158469945355,
      "grad_norm": 1.2420901334135377,
      "learning_rate": 0.00019751995264448704,
      "loss": 14.1115,
      "step": 2263
    },
    {
      "epoch": 0.39589071038251367,
      "grad_norm": 1.2584949482948276,
      "learning_rate": 0.00019751681747243635,
      "loss": 14.2023,
      "step": 2264
    },
    {
      "epoch": 0.3960655737704918,
      "grad_norm": 1.177047205234822,
      "learning_rate": 0.00019751368034487404,
      "loss": 14.1132,
      "step": 2265
    },
    {
      "epoch": 0.39624043715846996,
      "grad_norm": 1.4234321033484165,
      "learning_rate": 0.00019751054126186305,
      "loss": 14.352,
      "step": 2266
    },
    {
      "epoch": 0.3964153005464481,
      "grad_norm": 1.1517069717501225,
      "learning_rate": 0.0001975074002234663,
      "loss": 14.0455,
      "step": 2267
    },
    {
      "epoch": 0.39659016393442625,
      "grad_norm": 1.2081815866301822,
      "learning_rate": 0.0001975042572297468,
      "loss": 14.2769,
      "step": 2268
    },
    {
      "epoch": 0.3967650273224044,
      "grad_norm": 1.2703704024451172,
      "learning_rate": 0.00019750111228076755,
      "loss": 14.2183,
      "step": 2269
    },
    {
      "epoch": 0.3969398907103825,
      "grad_norm": 1.0347123407852685,
      "learning_rate": 0.00019749796537659164,
      "loss": 14.1562,
      "step": 2270
    },
    {
      "epoch": 0.39711475409836067,
      "grad_norm": 1.2334577031021765,
      "learning_rate": 0.00019749481651728216,
      "loss": 14.1232,
      "step": 2271
    },
    {
      "epoch": 0.3972896174863388,
      "grad_norm": 1.080907288607022,
      "learning_rate": 0.00019749166570290228,
      "loss": 14.1847,
      "step": 2272
    },
    {
      "epoch": 0.39746448087431696,
      "grad_norm": 1.2092589731099943,
      "learning_rate": 0.00019748851293351516,
      "loss": 13.9488,
      "step": 2273
    },
    {
      "epoch": 0.3976393442622951,
      "grad_norm": 1.0861222809534683,
      "learning_rate": 0.000197485358209184,
      "loss": 13.9573,
      "step": 2274
    },
    {
      "epoch": 0.3978142076502732,
      "grad_norm": 1.2190521388587863,
      "learning_rate": 0.00019748220152997212,
      "loss": 14.129,
      "step": 2275
    },
    {
      "epoch": 0.3979890710382514,
      "grad_norm": 1.0458275471475387,
      "learning_rate": 0.00019747904289594276,
      "loss": 14.2015,
      "step": 2276
    },
    {
      "epoch": 0.3981639344262295,
      "grad_norm": 1.035612566656258,
      "learning_rate": 0.00019747588230715932,
      "loss": 14.1799,
      "step": 2277
    },
    {
      "epoch": 0.39833879781420767,
      "grad_norm": 1.04827058090976,
      "learning_rate": 0.00019747271976368512,
      "loss": 14.1093,
      "step": 2278
    },
    {
      "epoch": 0.3985136612021858,
      "grad_norm": 1.190372640491676,
      "learning_rate": 0.00019746955526558362,
      "loss": 14.305,
      "step": 2279
    },
    {
      "epoch": 0.39868852459016396,
      "grad_norm": 1.0716151375718632,
      "learning_rate": 0.00019746638881291829,
      "loss": 14.2618,
      "step": 2280
    },
    {
      "epoch": 0.3988633879781421,
      "grad_norm": 1.1529070205304546,
      "learning_rate": 0.00019746322040575256,
      "loss": 14.2033,
      "step": 2281
    },
    {
      "epoch": 0.3990382513661202,
      "grad_norm": 1.0264074865613042,
      "learning_rate": 0.00019746005004415005,
      "loss": 13.9962,
      "step": 2282
    },
    {
      "epoch": 0.3992131147540984,
      "grad_norm": 1.0361674055041792,
      "learning_rate": 0.00019745687772817428,
      "loss": 13.8923,
      "step": 2283
    },
    {
      "epoch": 0.3993879781420765,
      "grad_norm": 1.1407330557879118,
      "learning_rate": 0.00019745370345788886,
      "loss": 14.2247,
      "step": 2284
    },
    {
      "epoch": 0.39956284153005467,
      "grad_norm": 1.0776153454728417,
      "learning_rate": 0.00019745052723335745,
      "loss": 14.2613,
      "step": 2285
    },
    {
      "epoch": 0.3997377049180328,
      "grad_norm": 1.0905553463238151,
      "learning_rate": 0.00019744734905464379,
      "loss": 14.1756,
      "step": 2286
    },
    {
      "epoch": 0.3999125683060109,
      "grad_norm": 1.03949768210269,
      "learning_rate": 0.00019744416892181159,
      "loss": 14.0625,
      "step": 2287
    },
    {
      "epoch": 0.4000874316939891,
      "grad_norm": 1.022263437017207,
      "learning_rate": 0.0001974409868349246,
      "loss": 14.2236,
      "step": 2288
    },
    {
      "epoch": 0.4002622950819672,
      "grad_norm": 1.1211899845748097,
      "learning_rate": 0.0001974378027940466,
      "loss": 14.0241,
      "step": 2289
    },
    {
      "epoch": 0.4004371584699454,
      "grad_norm": 1.2010904258036825,
      "learning_rate": 0.00019743461679924153,
      "loss": 14.0007,
      "step": 2290
    },
    {
      "epoch": 0.4006120218579235,
      "grad_norm": 1.3391482306374067,
      "learning_rate": 0.00019743142885057325,
      "loss": 14.0314,
      "step": 2291
    },
    {
      "epoch": 0.4007868852459016,
      "grad_norm": 1.0659135551720744,
      "learning_rate": 0.0001974282389481056,
      "loss": 13.9507,
      "step": 2292
    },
    {
      "epoch": 0.4009617486338798,
      "grad_norm": 1.0973342167645386,
      "learning_rate": 0.00019742504709190266,
      "loss": 13.9422,
      "step": 2293
    },
    {
      "epoch": 0.4011366120218579,
      "grad_norm": 1.1230757501568347,
      "learning_rate": 0.00019742185328202837,
      "loss": 14.1791,
      "step": 2294
    },
    {
      "epoch": 0.4013114754098361,
      "grad_norm": 1.4451894631783986,
      "learning_rate": 0.0001974186575185468,
      "loss": 14.2982,
      "step": 2295
    },
    {
      "epoch": 0.4014863387978142,
      "grad_norm": 1.0299769106460146,
      "learning_rate": 0.00019741545980152205,
      "loss": 14.2457,
      "step": 2296
    },
    {
      "epoch": 0.4016612021857924,
      "grad_norm": 1.2280688440809127,
      "learning_rate": 0.0001974122601310182,
      "loss": 13.8201,
      "step": 2297
    },
    {
      "epoch": 0.4018360655737705,
      "grad_norm": 1.1024355341676286,
      "learning_rate": 0.00019740905850709948,
      "loss": 14.0468,
      "step": 2298
    },
    {
      "epoch": 0.4020109289617486,
      "grad_norm": 1.1157737465484008,
      "learning_rate": 0.00019740585492983003,
      "loss": 13.9398,
      "step": 2299
    },
    {
      "epoch": 0.4021857923497268,
      "grad_norm": 1.0398693978427282,
      "learning_rate": 0.0001974026493992741,
      "loss": 13.9991,
      "step": 2300
    },
    {
      "epoch": 0.4023606557377049,
      "grad_norm": 1.0969679953842892,
      "learning_rate": 0.000197399441915496,
      "loss": 14.1544,
      "step": 2301
    },
    {
      "epoch": 0.4025355191256831,
      "grad_norm": 1.08777445766808,
      "learning_rate": 0.00019739623247856002,
      "loss": 14.1529,
      "step": 2302
    },
    {
      "epoch": 0.4027103825136612,
      "grad_norm": 1.0952157441588175,
      "learning_rate": 0.00019739302108853055,
      "loss": 14.1226,
      "step": 2303
    },
    {
      "epoch": 0.4028852459016393,
      "grad_norm": 1.0750964904975067,
      "learning_rate": 0.00019738980774547198,
      "loss": 13.9385,
      "step": 2304
    },
    {
      "epoch": 0.4030601092896175,
      "grad_norm": 1.0695332794460886,
      "learning_rate": 0.00019738659244944874,
      "loss": 13.7887,
      "step": 2305
    },
    {
      "epoch": 0.4032349726775956,
      "grad_norm": 1.053895791934353,
      "learning_rate": 0.00019738337520052525,
      "loss": 13.7604,
      "step": 2306
    },
    {
      "epoch": 0.4034098360655738,
      "grad_norm": 1.0926558551534684,
      "learning_rate": 0.0001973801559987661,
      "loss": 14.1292,
      "step": 2307
    },
    {
      "epoch": 0.4035846994535519,
      "grad_norm": 1.0844400446170503,
      "learning_rate": 0.00019737693484423588,
      "loss": 14.1636,
      "step": 2308
    },
    {
      "epoch": 0.40375956284153003,
      "grad_norm": 1.0319441329119279,
      "learning_rate": 0.0001973737117369991,
      "loss": 13.9204,
      "step": 2309
    },
    {
      "epoch": 0.4039344262295082,
      "grad_norm": 1.182285351784395,
      "learning_rate": 0.00019737048667712039,
      "loss": 14.3212,
      "step": 2310
    },
    {
      "epoch": 0.4041092896174863,
      "grad_norm": 1.0855641080760148,
      "learning_rate": 0.0001973672596646645,
      "loss": 14.257,
      "step": 2311
    },
    {
      "epoch": 0.4042841530054645,
      "grad_norm": 1.0775217095862755,
      "learning_rate": 0.00019736403069969607,
      "loss": 14.1138,
      "step": 2312
    },
    {
      "epoch": 0.4044590163934426,
      "grad_norm": 1.1736526857197231,
      "learning_rate": 0.0001973607997822799,
      "loss": 13.8913,
      "step": 2313
    },
    {
      "epoch": 0.4046338797814208,
      "grad_norm": 1.1375864960707278,
      "learning_rate": 0.00019735756691248072,
      "loss": 13.8765,
      "step": 2314
    },
    {
      "epoch": 0.4048087431693989,
      "grad_norm": 1.1301317280646548,
      "learning_rate": 0.00019735433209036344,
      "loss": 13.8315,
      "step": 2315
    },
    {
      "epoch": 0.40498360655737703,
      "grad_norm": 1.0599951002619044,
      "learning_rate": 0.00019735109531599285,
      "loss": 13.8188,
      "step": 2316
    },
    {
      "epoch": 0.4051584699453552,
      "grad_norm": 1.1947478650136547,
      "learning_rate": 0.0001973478565894339,
      "loss": 14.1808,
      "step": 2317
    },
    {
      "epoch": 0.4053333333333333,
      "grad_norm": 1.039595699885639,
      "learning_rate": 0.00019734461591075156,
      "loss": 13.8706,
      "step": 2318
    },
    {
      "epoch": 0.4055081967213115,
      "grad_norm": 1.0594270407011068,
      "learning_rate": 0.00019734137328001076,
      "loss": 13.7896,
      "step": 2319
    },
    {
      "epoch": 0.4056830601092896,
      "grad_norm": 1.0352748789526005,
      "learning_rate": 0.00019733812869727655,
      "loss": 13.8453,
      "step": 2320
    },
    {
      "epoch": 0.40585792349726774,
      "grad_norm": 1.2442216972200733,
      "learning_rate": 0.00019733488216261397,
      "loss": 14.2564,
      "step": 2321
    },
    {
      "epoch": 0.4060327868852459,
      "grad_norm": 1.0827681705640795,
      "learning_rate": 0.0001973316336760882,
      "loss": 14.1009,
      "step": 2322
    },
    {
      "epoch": 0.40620765027322403,
      "grad_norm": 1.1644080461553554,
      "learning_rate": 0.00019732838323776428,
      "loss": 14.2862,
      "step": 2323
    },
    {
      "epoch": 0.4063825136612022,
      "grad_norm": 1.075430909409772,
      "learning_rate": 0.00019732513084770746,
      "loss": 14.2106,
      "step": 2324
    },
    {
      "epoch": 0.4065573770491803,
      "grad_norm": 1.1115481630298034,
      "learning_rate": 0.0001973218765059829,
      "loss": 14.2119,
      "step": 2325
    },
    {
      "epoch": 0.40673224043715844,
      "grad_norm": 1.0196191619832706,
      "learning_rate": 0.00019731862021265593,
      "loss": 14.0373,
      "step": 2326
    },
    {
      "epoch": 0.4069071038251366,
      "grad_norm": 1.0478156473811675,
      "learning_rate": 0.00019731536196779183,
      "loss": 14.1896,
      "step": 2327
    },
    {
      "epoch": 0.40708196721311474,
      "grad_norm": 1.2046951115720685,
      "learning_rate": 0.00019731210177145592,
      "loss": 14.0125,
      "step": 2328
    },
    {
      "epoch": 0.4072568306010929,
      "grad_norm": 1.0242085426790475,
      "learning_rate": 0.00019730883962371356,
      "loss": 14.0639,
      "step": 2329
    },
    {
      "epoch": 0.40743169398907103,
      "grad_norm": 1.0597321073795893,
      "learning_rate": 0.0001973055755246302,
      "loss": 14.0736,
      "step": 2330
    },
    {
      "epoch": 0.4076065573770492,
      "grad_norm": 1.0681283467361302,
      "learning_rate": 0.00019730230947427127,
      "loss": 14.1569,
      "step": 2331
    },
    {
      "epoch": 0.4077814207650273,
      "grad_norm": 1.0984888817080938,
      "learning_rate": 0.00019729904147270232,
      "loss": 14.2381,
      "step": 2332
    },
    {
      "epoch": 0.40795628415300544,
      "grad_norm": 1.1296139848548352,
      "learning_rate": 0.0001972957715199888,
      "loss": 14.5046,
      "step": 2333
    },
    {
      "epoch": 0.4081311475409836,
      "grad_norm": 1.0349450605917547,
      "learning_rate": 0.00019729249961619635,
      "loss": 14.0534,
      "step": 2334
    },
    {
      "epoch": 0.40830601092896174,
      "grad_norm": 1.0229796764787222,
      "learning_rate": 0.00019728922576139053,
      "loss": 13.6597,
      "step": 2335
    },
    {
      "epoch": 0.4084808743169399,
      "grad_norm": 1.1401635066144387,
      "learning_rate": 0.00019728594995563703,
      "loss": 14.0747,
      "step": 2336
    },
    {
      "epoch": 0.40865573770491803,
      "grad_norm": 1.0527474294129089,
      "learning_rate": 0.00019728267219900155,
      "loss": 14.0998,
      "step": 2337
    },
    {
      "epoch": 0.40883060109289615,
      "grad_norm": 1.1671452343478583,
      "learning_rate": 0.00019727939249154975,
      "loss": 14.084,
      "step": 2338
    },
    {
      "epoch": 0.4090054644808743,
      "grad_norm": 1.0007617316745319,
      "learning_rate": 0.00019727611083334746,
      "loss": 14.2964,
      "step": 2339
    },
    {
      "epoch": 0.40918032786885244,
      "grad_norm": 1.2640894190143566,
      "learning_rate": 0.00019727282722446047,
      "loss": 14.0529,
      "step": 2340
    },
    {
      "epoch": 0.4093551912568306,
      "grad_norm": 1.1883283485514002,
      "learning_rate": 0.00019726954166495464,
      "loss": 14.0804,
      "step": 2341
    },
    {
      "epoch": 0.40953005464480874,
      "grad_norm": 1.129324894759452,
      "learning_rate": 0.00019726625415489584,
      "loss": 14.1564,
      "step": 2342
    },
    {
      "epoch": 0.40970491803278686,
      "grad_norm": 1.1236866479324352,
      "learning_rate": 0.00019726296469435,
      "loss": 14.0002,
      "step": 2343
    },
    {
      "epoch": 0.40987978142076503,
      "grad_norm": 1.0951576652430537,
      "learning_rate": 0.0001972596732833831,
      "loss": 14.2719,
      "step": 2344
    },
    {
      "epoch": 0.41005464480874315,
      "grad_norm": 1.1699099619754123,
      "learning_rate": 0.00019725637992206108,
      "loss": 14.1338,
      "step": 2345
    },
    {
      "epoch": 0.4102295081967213,
      "grad_norm": 1.02325674117183,
      "learning_rate": 0.00019725308461045,
      "loss": 14.0158,
      "step": 2346
    },
    {
      "epoch": 0.41040437158469945,
      "grad_norm": 1.0482120372260455,
      "learning_rate": 0.00019724978734861605,
      "loss": 13.9144,
      "step": 2347
    },
    {
      "epoch": 0.4105792349726776,
      "grad_norm": 1.1642679637083184,
      "learning_rate": 0.00019724648813662524,
      "loss": 13.8814,
      "step": 2348
    },
    {
      "epoch": 0.41075409836065574,
      "grad_norm": 1.1106234927046856,
      "learning_rate": 0.00019724318697454373,
      "loss": 14.2053,
      "step": 2349
    },
    {
      "epoch": 0.41092896174863386,
      "grad_norm": 1.2270958887826482,
      "learning_rate": 0.00019723988386243776,
      "loss": 13.9688,
      "step": 2350
    },
    {
      "epoch": 0.41110382513661203,
      "grad_norm": 1.0308501110724386,
      "learning_rate": 0.00019723657880037356,
      "loss": 13.7758,
      "step": 2351
    },
    {
      "epoch": 0.41127868852459015,
      "grad_norm": 1.083363237689744,
      "learning_rate": 0.00019723327178841743,
      "loss": 14.218,
      "step": 2352
    },
    {
      "epoch": 0.4114535519125683,
      "grad_norm": 1.161198158989911,
      "learning_rate": 0.0001972299628266356,
      "loss": 14.2395,
      "step": 2353
    },
    {
      "epoch": 0.41162841530054645,
      "grad_norm": 1.1190812947414601,
      "learning_rate": 0.0001972266519150945,
      "loss": 14.0113,
      "step": 2354
    },
    {
      "epoch": 0.41180327868852457,
      "grad_norm": 1.0649054132424927,
      "learning_rate": 0.00019722333905386051,
      "loss": 14.1357,
      "step": 2355
    },
    {
      "epoch": 0.41197814207650274,
      "grad_norm": 1.222870069013402,
      "learning_rate": 0.00019722002424300006,
      "loss": 14.0869,
      "step": 2356
    },
    {
      "epoch": 0.41215300546448086,
      "grad_norm": 1.00043009774934,
      "learning_rate": 0.00019721670748257958,
      "loss": 13.9374,
      "step": 2357
    },
    {
      "epoch": 0.41232786885245903,
      "grad_norm": 1.342074158189221,
      "learning_rate": 0.00019721338877266566,
      "loss": 14.1149,
      "step": 2358
    },
    {
      "epoch": 0.41250273224043715,
      "grad_norm": 1.2195990319480199,
      "learning_rate": 0.0001972100681133248,
      "loss": 14.3075,
      "step": 2359
    },
    {
      "epoch": 0.41267759562841533,
      "grad_norm": 1.1316264651846675,
      "learning_rate": 0.00019720674550462362,
      "loss": 14.2192,
      "step": 2360
    },
    {
      "epoch": 0.41285245901639345,
      "grad_norm": 1.3330171022572432,
      "learning_rate": 0.0001972034209466287,
      "loss": 14.0214,
      "step": 2361
    },
    {
      "epoch": 0.41302732240437157,
      "grad_norm": 0.953269648406951,
      "learning_rate": 0.00019720009443940676,
      "loss": 13.8855,
      "step": 2362
    },
    {
      "epoch": 0.41320218579234974,
      "grad_norm": 1.038107428436265,
      "learning_rate": 0.00019719676598302445,
      "loss": 13.8929,
      "step": 2363
    },
    {
      "epoch": 0.41337704918032786,
      "grad_norm": 1.0871655376821996,
      "learning_rate": 0.00019719343557754858,
      "loss": 13.5883,
      "step": 2364
    },
    {
      "epoch": 0.41355191256830603,
      "grad_norm": 1.0788879503988895,
      "learning_rate": 0.0001971901032230459,
      "loss": 13.9322,
      "step": 2365
    },
    {
      "epoch": 0.41372677595628415,
      "grad_norm": 1.0960039732286446,
      "learning_rate": 0.00019718676891958322,
      "loss": 14.1106,
      "step": 2366
    },
    {
      "epoch": 0.4139016393442623,
      "grad_norm": 1.2040783171853209,
      "learning_rate": 0.00019718343266722743,
      "loss": 13.98,
      "step": 2367
    },
    {
      "epoch": 0.41407650273224045,
      "grad_norm": 1.093493487934241,
      "learning_rate": 0.00019718009446604545,
      "loss": 13.7223,
      "step": 2368
    },
    {
      "epoch": 0.41425136612021857,
      "grad_norm": 1.0319817736006907,
      "learning_rate": 0.00019717675431610415,
      "loss": 13.9823,
      "step": 2369
    },
    {
      "epoch": 0.41442622950819674,
      "grad_norm": 1.1407066492605251,
      "learning_rate": 0.00019717341221747056,
      "loss": 14.0737,
      "step": 2370
    },
    {
      "epoch": 0.41460109289617486,
      "grad_norm": 1.0409428007679413,
      "learning_rate": 0.0001971700681702117,
      "loss": 14.011,
      "step": 2371
    },
    {
      "epoch": 0.414775956284153,
      "grad_norm": 1.226509638524188,
      "learning_rate": 0.00019716672217439464,
      "loss": 13.6269,
      "step": 2372
    },
    {
      "epoch": 0.41495081967213115,
      "grad_norm": 1.082850876721327,
      "learning_rate": 0.00019716337423008646,
      "loss": 14.237,
      "step": 2373
    },
    {
      "epoch": 0.4151256830601093,
      "grad_norm": 1.064039646574152,
      "learning_rate": 0.00019716002433735428,
      "loss": 14.0208,
      "step": 2374
    },
    {
      "epoch": 0.41530054644808745,
      "grad_norm": 1.1369836632858206,
      "learning_rate": 0.00019715667249626528,
      "loss": 13.7424,
      "step": 2375
    },
    {
      "epoch": 0.41547540983606557,
      "grad_norm": 1.0699572086794698,
      "learning_rate": 0.00019715331870688672,
      "loss": 13.9829,
      "step": 2376
    },
    {
      "epoch": 0.41565027322404374,
      "grad_norm": 1.0757536338212788,
      "learning_rate": 0.00019714996296928578,
      "loss": 13.7998,
      "step": 2377
    },
    {
      "epoch": 0.41582513661202186,
      "grad_norm": 1.0336074556495467,
      "learning_rate": 0.00019714660528352983,
      "loss": 14.1894,
      "step": 2378
    },
    {
      "epoch": 0.416,
      "grad_norm": 1.1551307206010186,
      "learning_rate": 0.00019714324564968613,
      "loss": 14.1053,
      "step": 2379
    },
    {
      "epoch": 0.41617486338797816,
      "grad_norm": 1.1690113430294675,
      "learning_rate": 0.0001971398840678221,
      "loss": 14.1403,
      "step": 2380
    },
    {
      "epoch": 0.4163497267759563,
      "grad_norm": 1.3410834975871202,
      "learning_rate": 0.00019713652053800516,
      "loss": 14.0671,
      "step": 2381
    },
    {
      "epoch": 0.41652459016393445,
      "grad_norm": 1.130449589527514,
      "learning_rate": 0.0001971331550603027,
      "loss": 14.1599,
      "step": 2382
    },
    {
      "epoch": 0.41669945355191257,
      "grad_norm": 1.1445541969182857,
      "learning_rate": 0.00019712978763478228,
      "loss": 14.0559,
      "step": 2383
    },
    {
      "epoch": 0.4168743169398907,
      "grad_norm": 1.1339681806276714,
      "learning_rate": 0.00019712641826151136,
      "loss": 13.9595,
      "step": 2384
    },
    {
      "epoch": 0.41704918032786886,
      "grad_norm": 1.022141188130735,
      "learning_rate": 0.0001971230469405575,
      "loss": 13.9882,
      "step": 2385
    },
    {
      "epoch": 0.417224043715847,
      "grad_norm": 1.052822589429485,
      "learning_rate": 0.00019711967367198841,
      "loss": 13.8265,
      "step": 2386
    },
    {
      "epoch": 0.41739890710382516,
      "grad_norm": 1.0369020614719524,
      "learning_rate": 0.00019711629845587164,
      "loss": 14.0442,
      "step": 2387
    },
    {
      "epoch": 0.4175737704918033,
      "grad_norm": 1.0854054319901663,
      "learning_rate": 0.0001971129212922749,
      "loss": 14.0094,
      "step": 2388
    },
    {
      "epoch": 0.4177486338797814,
      "grad_norm": 1.1492540768174935,
      "learning_rate": 0.00019710954218126595,
      "loss": 14.0846,
      "step": 2389
    },
    {
      "epoch": 0.41792349726775957,
      "grad_norm": 1.0171052570253099,
      "learning_rate": 0.00019710616112291244,
      "loss": 14.1033,
      "step": 2390
    },
    {
      "epoch": 0.4180983606557377,
      "grad_norm": 0.9509894663059709,
      "learning_rate": 0.0001971027781172823,
      "loss": 13.8232,
      "step": 2391
    },
    {
      "epoch": 0.41827322404371586,
      "grad_norm": 0.9818472223036129,
      "learning_rate": 0.0001970993931644433,
      "loss": 13.7307,
      "step": 2392
    },
    {
      "epoch": 0.418448087431694,
      "grad_norm": 1.050066936818495,
      "learning_rate": 0.00019709600626446335,
      "loss": 13.9806,
      "step": 2393
    },
    {
      "epoch": 0.41862295081967216,
      "grad_norm": 1.0617871757911754,
      "learning_rate": 0.00019709261741741033,
      "loss": 13.9984,
      "step": 2394
    },
    {
      "epoch": 0.4187978142076503,
      "grad_norm": 0.9691087638720436,
      "learning_rate": 0.0001970892266233522,
      "loss": 13.9525,
      "step": 2395
    },
    {
      "epoch": 0.4189726775956284,
      "grad_norm": 1.0647927215804318,
      "learning_rate": 0.000197085833882357,
      "loss": 13.7747,
      "step": 2396
    },
    {
      "epoch": 0.41914754098360657,
      "grad_norm": 1.090559767730256,
      "learning_rate": 0.0001970824391944927,
      "loss": 13.7829,
      "step": 2397
    },
    {
      "epoch": 0.4193224043715847,
      "grad_norm": 1.0279991083708113,
      "learning_rate": 0.00019707904255982745,
      "loss": 13.9101,
      "step": 2398
    },
    {
      "epoch": 0.41949726775956286,
      "grad_norm": 1.0610701207211903,
      "learning_rate": 0.00019707564397842932,
      "loss": 14.0117,
      "step": 2399
    },
    {
      "epoch": 0.419672131147541,
      "grad_norm": 1.0806497756975941,
      "learning_rate": 0.00019707224345036646,
      "loss": 13.9956,
      "step": 2400
    },
    {
      "epoch": 0.4198469945355191,
      "grad_norm": 1.0533332493446343,
      "learning_rate": 0.00019706884097570707,
      "loss": 13.9697,
      "step": 2401
    },
    {
      "epoch": 0.4200218579234973,
      "grad_norm": 1.0231343642197601,
      "learning_rate": 0.00019706543655451938,
      "loss": 14.0914,
      "step": 2402
    },
    {
      "epoch": 0.4201967213114754,
      "grad_norm": 1.0609226753681282,
      "learning_rate": 0.00019706203018687166,
      "loss": 13.9504,
      "step": 2403
    },
    {
      "epoch": 0.42037158469945357,
      "grad_norm": 1.0538717297340507,
      "learning_rate": 0.0001970586218728322,
      "loss": 14.1298,
      "step": 2404
    },
    {
      "epoch": 0.4205464480874317,
      "grad_norm": 1.0335012914192225,
      "learning_rate": 0.0001970552116124694,
      "loss": 13.9746,
      "step": 2405
    },
    {
      "epoch": 0.4207213114754098,
      "grad_norm": 1.2600946677121987,
      "learning_rate": 0.00019705179940585155,
      "loss": 14.0865,
      "step": 2406
    },
    {
      "epoch": 0.420896174863388,
      "grad_norm": 1.0754158869184867,
      "learning_rate": 0.0001970483852530472,
      "loss": 14.0587,
      "step": 2407
    },
    {
      "epoch": 0.4210710382513661,
      "grad_norm": 1.1027343432309604,
      "learning_rate": 0.0001970449691541247,
      "loss": 14.0701,
      "step": 2408
    },
    {
      "epoch": 0.4212459016393443,
      "grad_norm": 1.0424982523132391,
      "learning_rate": 0.0001970415511091526,
      "loss": 14.1637,
      "step": 2409
    },
    {
      "epoch": 0.4214207650273224,
      "grad_norm": 1.1029233734491617,
      "learning_rate": 0.00019703813111819946,
      "loss": 14.0657,
      "step": 2410
    },
    {
      "epoch": 0.42159562841530057,
      "grad_norm": 1.2190838677307416,
      "learning_rate": 0.00019703470918133387,
      "loss": 13.9402,
      "step": 2411
    },
    {
      "epoch": 0.4217704918032787,
      "grad_norm": 1.1051373124876338,
      "learning_rate": 0.00019703128529862437,
      "loss": 13.7375,
      "step": 2412
    },
    {
      "epoch": 0.4219453551912568,
      "grad_norm": 1.0705264445034313,
      "learning_rate": 0.0001970278594701397,
      "loss": 14.1562,
      "step": 2413
    },
    {
      "epoch": 0.422120218579235,
      "grad_norm": 1.0686051953153888,
      "learning_rate": 0.00019702443169594856,
      "loss": 14.0556,
      "step": 2414
    },
    {
      "epoch": 0.4222950819672131,
      "grad_norm": 1.079355843077825,
      "learning_rate": 0.00019702100197611962,
      "loss": 13.9575,
      "step": 2415
    },
    {
      "epoch": 0.4224699453551913,
      "grad_norm": 0.9490705087828804,
      "learning_rate": 0.00019701757031072173,
      "loss": 13.787,
      "step": 2416
    },
    {
      "epoch": 0.4226448087431694,
      "grad_norm": 1.0427028280534336,
      "learning_rate": 0.00019701413669982366,
      "loss": 13.9581,
      "step": 2417
    },
    {
      "epoch": 0.4228196721311475,
      "grad_norm": 0.9582238260250714,
      "learning_rate": 0.00019701070114349424,
      "loss": 13.9053,
      "step": 2418
    },
    {
      "epoch": 0.4229945355191257,
      "grad_norm": 1.1411115115457853,
      "learning_rate": 0.00019700726364180247,
      "loss": 14.0918,
      "step": 2419
    },
    {
      "epoch": 0.4231693989071038,
      "grad_norm": 1.1431895938144738,
      "learning_rate": 0.00019700382419481714,
      "loss": 14.1111,
      "step": 2420
    },
    {
      "epoch": 0.423344262295082,
      "grad_norm": 1.1897787782976383,
      "learning_rate": 0.00019700038280260734,
      "loss": 13.9728,
      "step": 2421
    },
    {
      "epoch": 0.4235191256830601,
      "grad_norm": 0.9839560612685829,
      "learning_rate": 0.00019699693946524202,
      "loss": 14.0281,
      "step": 2422
    },
    {
      "epoch": 0.4236939890710382,
      "grad_norm": 1.0976036089232362,
      "learning_rate": 0.00019699349418279025,
      "loss": 13.9786,
      "step": 2423
    },
    {
      "epoch": 0.4238688524590164,
      "grad_norm": 0.9763313629641078,
      "learning_rate": 0.0001969900469553211,
      "loss": 13.9849,
      "step": 2424
    },
    {
      "epoch": 0.4240437158469945,
      "grad_norm": 1.0088439777965705,
      "learning_rate": 0.00019698659778290373,
      "loss": 13.964,
      "step": 2425
    },
    {
      "epoch": 0.4242185792349727,
      "grad_norm": 1.2034924582029214,
      "learning_rate": 0.00019698314666560725,
      "loss": 13.8691,
      "step": 2426
    },
    {
      "epoch": 0.4243934426229508,
      "grad_norm": 1.066127280044775,
      "learning_rate": 0.00019697969360350098,
      "loss": 13.9701,
      "step": 2427
    },
    {
      "epoch": 0.424568306010929,
      "grad_norm": 1.092037593332987,
      "learning_rate": 0.00019697623859665398,
      "loss": 13.951,
      "step": 2428
    },
    {
      "epoch": 0.4247431693989071,
      "grad_norm": 0.9851183019285709,
      "learning_rate": 0.00019697278164513573,
      "loss": 13.6842,
      "step": 2429
    },
    {
      "epoch": 0.4249180327868852,
      "grad_norm": 1.0200627784601217,
      "learning_rate": 0.00019696932274901544,
      "loss": 14.1306,
      "step": 2430
    },
    {
      "epoch": 0.4250928961748634,
      "grad_norm": 1.0285332191899614,
      "learning_rate": 0.00019696586190836245,
      "loss": 13.8942,
      "step": 2431
    },
    {
      "epoch": 0.4252677595628415,
      "grad_norm": 0.9937075442925144,
      "learning_rate": 0.00019696239912324624,
      "loss": 13.615,
      "step": 2432
    },
    {
      "epoch": 0.4254426229508197,
      "grad_norm": 1.0779975833833921,
      "learning_rate": 0.00019695893439373622,
      "loss": 14.0099,
      "step": 2433
    },
    {
      "epoch": 0.4256174863387978,
      "grad_norm": 1.0252877954998976,
      "learning_rate": 0.00019695546771990186,
      "loss": 14.1231,
      "step": 2434
    },
    {
      "epoch": 0.42579234972677593,
      "grad_norm": 1.0443513055769436,
      "learning_rate": 0.00019695199910181265,
      "loss": 13.8412,
      "step": 2435
    },
    {
      "epoch": 0.4259672131147541,
      "grad_norm": 0.9740194942325756,
      "learning_rate": 0.0001969485285395382,
      "loss": 13.7703,
      "step": 2436
    },
    {
      "epoch": 0.4261420765027322,
      "grad_norm": 1.0545753638530093,
      "learning_rate": 0.0001969450560331481,
      "loss": 13.5501,
      "step": 2437
    },
    {
      "epoch": 0.4263169398907104,
      "grad_norm": 1.0369265105872436,
      "learning_rate": 0.00019694158158271193,
      "loss": 13.8879,
      "step": 2438
    },
    {
      "epoch": 0.4264918032786885,
      "grad_norm": 1.0839879906026382,
      "learning_rate": 0.0001969381051882994,
      "loss": 14.1366,
      "step": 2439
    },
    {
      "epoch": 0.4266666666666667,
      "grad_norm": 1.1047983967423654,
      "learning_rate": 0.00019693462684998024,
      "loss": 14.0034,
      "step": 2440
    },
    {
      "epoch": 0.4268415300546448,
      "grad_norm": 0.9841090042087763,
      "learning_rate": 0.00019693114656782422,
      "loss": 13.9385,
      "step": 2441
    },
    {
      "epoch": 0.42701639344262293,
      "grad_norm": 1.0112873230122275,
      "learning_rate": 0.00019692766434190105,
      "loss": 14.0629,
      "step": 2442
    },
    {
      "epoch": 0.4271912568306011,
      "grad_norm": 1.0729139413048074,
      "learning_rate": 0.0001969241801722806,
      "loss": 13.7045,
      "step": 2443
    },
    {
      "epoch": 0.4273661202185792,
      "grad_norm": 1.0813474563565237,
      "learning_rate": 0.00019692069405903272,
      "loss": 14.1026,
      "step": 2444
    },
    {
      "epoch": 0.4275409836065574,
      "grad_norm": 1.0105518753767115,
      "learning_rate": 0.0001969172060022274,
      "loss": 14.2059,
      "step": 2445
    },
    {
      "epoch": 0.4277158469945355,
      "grad_norm": 0.9921472679635706,
      "learning_rate": 0.00019691371600193448,
      "loss": 13.9935,
      "step": 2446
    },
    {
      "epoch": 0.42789071038251364,
      "grad_norm": 0.9319577009707982,
      "learning_rate": 0.000196910224058224,
      "loss": 13.8727,
      "step": 2447
    },
    {
      "epoch": 0.4280655737704918,
      "grad_norm": 1.075060701778133,
      "learning_rate": 0.00019690673017116596,
      "loss": 13.7235,
      "step": 2448
    },
    {
      "epoch": 0.42824043715846993,
      "grad_norm": 1.172148982033775,
      "learning_rate": 0.00019690323434083047,
      "loss": 14.0844,
      "step": 2449
    },
    {
      "epoch": 0.4284153005464481,
      "grad_norm": 1.0432686381306813,
      "learning_rate": 0.00019689973656728758,
      "loss": 13.8767,
      "step": 2450
    },
    {
      "epoch": 0.4285901639344262,
      "grad_norm": 1.1311138436399737,
      "learning_rate": 0.00019689623685060744,
      "loss": 14.0675,
      "step": 2451
    },
    {
      "epoch": 0.42876502732240435,
      "grad_norm": 1.0400289699743712,
      "learning_rate": 0.00019689273519086026,
      "loss": 14.0049,
      "step": 2452
    },
    {
      "epoch": 0.4289398907103825,
      "grad_norm": 0.9719205497524841,
      "learning_rate": 0.00019688923158811623,
      "loss": 13.8146,
      "step": 2453
    },
    {
      "epoch": 0.42911475409836064,
      "grad_norm": 1.0557316732086777,
      "learning_rate": 0.00019688572604244562,
      "loss": 13.8075,
      "step": 2454
    },
    {
      "epoch": 0.4292896174863388,
      "grad_norm": 1.0837376756806467,
      "learning_rate": 0.00019688221855391873,
      "loss": 13.9414,
      "step": 2455
    },
    {
      "epoch": 0.42946448087431693,
      "grad_norm": 1.1357116048109472,
      "learning_rate": 0.0001968787091226059,
      "loss": 14.2013,
      "step": 2456
    },
    {
      "epoch": 0.4296393442622951,
      "grad_norm": 1.0386982859446756,
      "learning_rate": 0.00019687519774857747,
      "loss": 13.942,
      "step": 2457
    },
    {
      "epoch": 0.42981420765027323,
      "grad_norm": 1.0077505691307265,
      "learning_rate": 0.00019687168443190389,
      "loss": 13.7776,
      "step": 2458
    },
    {
      "epoch": 0.42998907103825135,
      "grad_norm": 1.0964647611674494,
      "learning_rate": 0.0001968681691726556,
      "loss": 14.1241,
      "step": 2459
    },
    {
      "epoch": 0.4301639344262295,
      "grad_norm": 0.9780683634856523,
      "learning_rate": 0.0001968646519709031,
      "loss": 13.6859,
      "step": 2460
    },
    {
      "epoch": 0.43033879781420764,
      "grad_norm": 1.1000697265738584,
      "learning_rate": 0.0001968611328267169,
      "loss": 13.9512,
      "step": 2461
    },
    {
      "epoch": 0.4305136612021858,
      "grad_norm": 1.088546489483593,
      "learning_rate": 0.00019685761174016763,
      "loss": 14.1905,
      "step": 2462
    },
    {
      "epoch": 0.43068852459016393,
      "grad_norm": 1.0095429683435346,
      "learning_rate": 0.0001968540887113258,
      "loss": 13.7311,
      "step": 2463
    },
    {
      "epoch": 0.43086338797814205,
      "grad_norm": 1.0756569516736847,
      "learning_rate": 0.00019685056374026211,
      "loss": 14.1469,
      "step": 2464
    },
    {
      "epoch": 0.43103825136612023,
      "grad_norm": 1.0393541050248545,
      "learning_rate": 0.0001968470368270473,
      "loss": 13.702,
      "step": 2465
    },
    {
      "epoch": 0.43121311475409835,
      "grad_norm": 1.0432958274384763,
      "learning_rate": 0.00019684350797175197,
      "loss": 13.9179,
      "step": 2466
    },
    {
      "epoch": 0.4313879781420765,
      "grad_norm": 0.9908511790391251,
      "learning_rate": 0.000196839977174447,
      "loss": 13.6267,
      "step": 2467
    },
    {
      "epoch": 0.43156284153005464,
      "grad_norm": 1.0106112661581963,
      "learning_rate": 0.00019683644443520314,
      "loss": 13.9424,
      "step": 2468
    },
    {
      "epoch": 0.43173770491803276,
      "grad_norm": 1.0750097000177146,
      "learning_rate": 0.00019683290975409126,
      "loss": 13.8255,
      "step": 2469
    },
    {
      "epoch": 0.43191256830601094,
      "grad_norm": 0.9986274527565063,
      "learning_rate": 0.00019682937313118218,
      "loss": 13.9988,
      "step": 2470
    },
    {
      "epoch": 0.43208743169398905,
      "grad_norm": 1.0304870877536014,
      "learning_rate": 0.00019682583456654692,
      "loss": 14.0863,
      "step": 2471
    },
    {
      "epoch": 0.43226229508196723,
      "grad_norm": 1.051790987603052,
      "learning_rate": 0.00019682229406025635,
      "loss": 14.1667,
      "step": 2472
    },
    {
      "epoch": 0.43243715846994535,
      "grad_norm": 1.023641014795343,
      "learning_rate": 0.0001968187516123815,
      "loss": 13.5536,
      "step": 2473
    },
    {
      "epoch": 0.4326120218579235,
      "grad_norm": 1.049397710642243,
      "learning_rate": 0.0001968152072229934,
      "loss": 13.9523,
      "step": 2474
    },
    {
      "epoch": 0.43278688524590164,
      "grad_norm": 1.1038408985766783,
      "learning_rate": 0.00019681166089216318,
      "loss": 13.8651,
      "step": 2475
    },
    {
      "epoch": 0.43296174863387976,
      "grad_norm": 0.9862407138407279,
      "learning_rate": 0.00019680811261996186,
      "loss": 13.9207,
      "step": 2476
    },
    {
      "epoch": 0.43313661202185794,
      "grad_norm": 1.1332319990040873,
      "learning_rate": 0.00019680456240646064,
      "loss": 13.9212,
      "step": 2477
    },
    {
      "epoch": 0.43331147540983606,
      "grad_norm": 1.049895691969903,
      "learning_rate": 0.00019680101025173073,
      "loss": 13.9949,
      "step": 2478
    },
    {
      "epoch": 0.43348633879781423,
      "grad_norm": 1.0746479094512804,
      "learning_rate": 0.00019679745615584334,
      "loss": 13.9019,
      "step": 2479
    },
    {
      "epoch": 0.43366120218579235,
      "grad_norm": 1.2329816331780448,
      "learning_rate": 0.00019679390011886975,
      "loss": 14.0636,
      "step": 2480
    },
    {
      "epoch": 0.43383606557377047,
      "grad_norm": 1.0508809532928587,
      "learning_rate": 0.0001967903421408813,
      "loss": 13.8149,
      "step": 2481
    },
    {
      "epoch": 0.43401092896174864,
      "grad_norm": 1.0400709479354242,
      "learning_rate": 0.00019678678222194925,
      "loss": 13.7735,
      "step": 2482
    },
    {
      "epoch": 0.43418579234972676,
      "grad_norm": 1.0450269884578745,
      "learning_rate": 0.0001967832203621451,
      "loss": 13.9753,
      "step": 2483
    },
    {
      "epoch": 0.43436065573770494,
      "grad_norm": 1.1094818844647496,
      "learning_rate": 0.00019677965656154017,
      "loss": 13.8895,
      "step": 2484
    },
    {
      "epoch": 0.43453551912568306,
      "grad_norm": 1.1128110853086348,
      "learning_rate": 0.00019677609082020597,
      "loss": 13.7996,
      "step": 2485
    },
    {
      "epoch": 0.4347103825136612,
      "grad_norm": 1.0266903760402293,
      "learning_rate": 0.00019677252313821404,
      "loss": 13.722,
      "step": 2486
    },
    {
      "epoch": 0.43488524590163935,
      "grad_norm": 1.075358815168796,
      "learning_rate": 0.00019676895351563588,
      "loss": 13.5894,
      "step": 2487
    },
    {
      "epoch": 0.43506010928961747,
      "grad_norm": 0.9787198445567261,
      "learning_rate": 0.00019676538195254306,
      "loss": 13.5763,
      "step": 2488
    },
    {
      "epoch": 0.43523497267759564,
      "grad_norm": 1.008961493915135,
      "learning_rate": 0.00019676180844900728,
      "loss": 13.3465,
      "step": 2489
    },
    {
      "epoch": 0.43540983606557376,
      "grad_norm": 1.1506482488492282,
      "learning_rate": 0.00019675823300510008,
      "loss": 13.977,
      "step": 2490
    },
    {
      "epoch": 0.43558469945355194,
      "grad_norm": 1.0713189088943587,
      "learning_rate": 0.00019675465562089323,
      "loss": 13.7775,
      "step": 2491
    },
    {
      "epoch": 0.43575956284153006,
      "grad_norm": 1.0316952161915538,
      "learning_rate": 0.0001967510762964585,
      "loss": 14.0078,
      "step": 2492
    },
    {
      "epoch": 0.4359344262295082,
      "grad_norm": 1.1020093700834697,
      "learning_rate": 0.0001967474950318676,
      "loss": 13.7992,
      "step": 2493
    },
    {
      "epoch": 0.43610928961748635,
      "grad_norm": 0.9808982329273453,
      "learning_rate": 0.00019674391182719237,
      "loss": 14.0797,
      "step": 2494
    },
    {
      "epoch": 0.43628415300546447,
      "grad_norm": 1.0133908182287596,
      "learning_rate": 0.00019674032668250466,
      "loss": 13.8665,
      "step": 2495
    },
    {
      "epoch": 0.43645901639344264,
      "grad_norm": 0.9874583815464799,
      "learning_rate": 0.00019673673959787639,
      "loss": 14.0477,
      "step": 2496
    },
    {
      "epoch": 0.43663387978142076,
      "grad_norm": 0.9976321065537377,
      "learning_rate": 0.00019673315057337944,
      "loss": 13.6338,
      "step": 2497
    },
    {
      "epoch": 0.4368087431693989,
      "grad_norm": 1.0722098809677607,
      "learning_rate": 0.00019672955960908583,
      "loss": 13.898,
      "step": 2498
    },
    {
      "epoch": 0.43698360655737706,
      "grad_norm": 1.0199913689000395,
      "learning_rate": 0.00019672596670506756,
      "loss": 13.8509,
      "step": 2499
    },
    {
      "epoch": 0.4371584699453552,
      "grad_norm": 1.0306035103162556,
      "learning_rate": 0.00019672237186139663,
      "loss": 14.0388,
      "step": 2500
    },
    {
      "epoch": 0.43733333333333335,
      "grad_norm": 1.0896812044096746,
      "learning_rate": 0.00019671877507814522,
      "loss": 13.8873,
      "step": 2501
    },
    {
      "epoch": 0.43750819672131147,
      "grad_norm": 1.0843351309899185,
      "learning_rate": 0.00019671517635538538,
      "loss": 13.759,
      "step": 2502
    },
    {
      "epoch": 0.4376830601092896,
      "grad_norm": 1.0105699672317197,
      "learning_rate": 0.0001967115756931893,
      "loss": 13.7712,
      "step": 2503
    },
    {
      "epoch": 0.43785792349726776,
      "grad_norm": 1.0619640240158315,
      "learning_rate": 0.00019670797309162915,
      "loss": 13.9162,
      "step": 2504
    },
    {
      "epoch": 0.4380327868852459,
      "grad_norm": 1.105192139666484,
      "learning_rate": 0.00019670436855077726,
      "loss": 13.8112,
      "step": 2505
    },
    {
      "epoch": 0.43820765027322406,
      "grad_norm": 1.00168628380512,
      "learning_rate": 0.00019670076207070584,
      "loss": 13.8477,
      "step": 2506
    },
    {
      "epoch": 0.4383825136612022,
      "grad_norm": 0.9948031160525459,
      "learning_rate": 0.00019669715365148722,
      "loss": 13.6383,
      "step": 2507
    },
    {
      "epoch": 0.43855737704918035,
      "grad_norm": 0.9710082015277027,
      "learning_rate": 0.00019669354329319378,
      "loss": 14.0052,
      "step": 2508
    },
    {
      "epoch": 0.43873224043715847,
      "grad_norm": 1.0231774715085378,
      "learning_rate": 0.00019668993099589791,
      "loss": 14.0088,
      "step": 2509
    },
    {
      "epoch": 0.4389071038251366,
      "grad_norm": 0.9717101175710734,
      "learning_rate": 0.00019668631675967205,
      "loss": 13.5105,
      "step": 2510
    },
    {
      "epoch": 0.43908196721311477,
      "grad_norm": 1.0667150990726855,
      "learning_rate": 0.00019668270058458866,
      "loss": 13.8494,
      "step": 2511
    },
    {
      "epoch": 0.4392568306010929,
      "grad_norm": 1.00635359634858,
      "learning_rate": 0.00019667908247072025,
      "loss": 13.8654,
      "step": 2512
    },
    {
      "epoch": 0.43943169398907106,
      "grad_norm": 1.0620547844030908,
      "learning_rate": 0.00019667546241813945,
      "loss": 13.7444,
      "step": 2513
    },
    {
      "epoch": 0.4396065573770492,
      "grad_norm": 0.9940856081147292,
      "learning_rate": 0.00019667184042691875,
      "loss": 13.6387,
      "step": 2514
    },
    {
      "epoch": 0.4397814207650273,
      "grad_norm": 1.660564809416758,
      "learning_rate": 0.00019666821649713085,
      "loss": 13.8755,
      "step": 2515
    },
    {
      "epoch": 0.4399562841530055,
      "grad_norm": 1.071595842519269,
      "learning_rate": 0.0001966645906288484,
      "loss": 13.8357,
      "step": 2516
    },
    {
      "epoch": 0.4401311475409836,
      "grad_norm": 1.0521886658380701,
      "learning_rate": 0.00019666096282214407,
      "loss": 13.8897,
      "step": 2517
    },
    {
      "epoch": 0.44030601092896177,
      "grad_norm": 1.018280055578924,
      "learning_rate": 0.00019665733307709069,
      "loss": 13.7245,
      "step": 2518
    },
    {
      "epoch": 0.4404808743169399,
      "grad_norm": 1.0028858613774674,
      "learning_rate": 0.000196653701393761,
      "loss": 13.9827,
      "step": 2519
    },
    {
      "epoch": 0.44065573770491806,
      "grad_norm": 0.9731481951911531,
      "learning_rate": 0.00019665006777222783,
      "loss": 13.8038,
      "step": 2520
    },
    {
      "epoch": 0.4408306010928962,
      "grad_norm": 1.1222631371219733,
      "learning_rate": 0.00019664643221256403,
      "loss": 13.8701,
      "step": 2521
    },
    {
      "epoch": 0.4410054644808743,
      "grad_norm": 1.0106495846109673,
      "learning_rate": 0.00019664279471484254,
      "loss": 13.9209,
      "step": 2522
    },
    {
      "epoch": 0.4411803278688525,
      "grad_norm": 1.1188070295194943,
      "learning_rate": 0.00019663915527913625,
      "loss": 13.909,
      "step": 2523
    },
    {
      "epoch": 0.4413551912568306,
      "grad_norm": 1.1705664114535796,
      "learning_rate": 0.00019663551390551823,
      "loss": 14.0487,
      "step": 2524
    },
    {
      "epoch": 0.44153005464480877,
      "grad_norm": 0.9984146142615324,
      "learning_rate": 0.0001966318705940614,
      "loss": 13.7995,
      "step": 2525
    },
    {
      "epoch": 0.4417049180327869,
      "grad_norm": 1.0182072081948228,
      "learning_rate": 0.00019662822534483886,
      "loss": 14.0211,
      "step": 2526
    },
    {
      "epoch": 0.441879781420765,
      "grad_norm": 1.0097591389444736,
      "learning_rate": 0.00019662457815792375,
      "loss": 13.6611,
      "step": 2527
    },
    {
      "epoch": 0.4420546448087432,
      "grad_norm": 0.9977284976863386,
      "learning_rate": 0.00019662092903338914,
      "loss": 13.9451,
      "step": 2528
    },
    {
      "epoch": 0.4422295081967213,
      "grad_norm": 0.9872038555395604,
      "learning_rate": 0.00019661727797130828,
      "loss": 13.7517,
      "step": 2529
    },
    {
      "epoch": 0.4424043715846995,
      "grad_norm": 1.0384207656160929,
      "learning_rate": 0.00019661362497175428,
      "loss": 14.0909,
      "step": 2530
    },
    {
      "epoch": 0.4425792349726776,
      "grad_norm": 1.076861253539661,
      "learning_rate": 0.00019660997003480052,
      "loss": 13.9091,
      "step": 2531
    },
    {
      "epoch": 0.4427540983606557,
      "grad_norm": 1.0455739772390444,
      "learning_rate": 0.00019660631316052021,
      "loss": 13.7675,
      "step": 2532
    },
    {
      "epoch": 0.4429289617486339,
      "grad_norm": 1.1239916557549574,
      "learning_rate": 0.00019660265434898668,
      "loss": 13.8984,
      "step": 2533
    },
    {
      "epoch": 0.443103825136612,
      "grad_norm": 0.9920522614162961,
      "learning_rate": 0.00019659899360027332,
      "loss": 13.956,
      "step": 2534
    },
    {
      "epoch": 0.4432786885245902,
      "grad_norm": 1.0668245284485485,
      "learning_rate": 0.00019659533091445356,
      "loss": 13.9992,
      "step": 2535
    },
    {
      "epoch": 0.4434535519125683,
      "grad_norm": 0.9665693782879403,
      "learning_rate": 0.00019659166629160083,
      "loss": 13.5317,
      "step": 2536
    },
    {
      "epoch": 0.4436284153005465,
      "grad_norm": 1.0648910000189888,
      "learning_rate": 0.0001965879997317886,
      "loss": 13.8641,
      "step": 2537
    },
    {
      "epoch": 0.4438032786885246,
      "grad_norm": 1.1002729128481807,
      "learning_rate": 0.00019658433123509044,
      "loss": 14.0016,
      "step": 2538
    },
    {
      "epoch": 0.4439781420765027,
      "grad_norm": 0.9654653365890074,
      "learning_rate": 0.00019658066080157984,
      "loss": 13.7259,
      "step": 2539
    },
    {
      "epoch": 0.4441530054644809,
      "grad_norm": 1.106436508439139,
      "learning_rate": 0.00019657698843133047,
      "loss": 13.8014,
      "step": 2540
    },
    {
      "epoch": 0.444327868852459,
      "grad_norm": 1.165102035745763,
      "learning_rate": 0.00019657331412441598,
      "loss": 13.615,
      "step": 2541
    },
    {
      "epoch": 0.4445027322404372,
      "grad_norm": 1.043554067715826,
      "learning_rate": 0.00019656963788091002,
      "loss": 13.9066,
      "step": 2542
    },
    {
      "epoch": 0.4446775956284153,
      "grad_norm": 1.0309453559094819,
      "learning_rate": 0.00019656595970088628,
      "loss": 13.7788,
      "step": 2543
    },
    {
      "epoch": 0.4448524590163934,
      "grad_norm": 1.0846138557378864,
      "learning_rate": 0.00019656227958441856,
      "loss": 14.1924,
      "step": 2544
    },
    {
      "epoch": 0.4450273224043716,
      "grad_norm": 1.0314521747978636,
      "learning_rate": 0.00019655859753158064,
      "loss": 13.8931,
      "step": 2545
    },
    {
      "epoch": 0.4452021857923497,
      "grad_norm": 1.1190890230996222,
      "learning_rate": 0.0001965549135424464,
      "loss": 13.8602,
      "step": 2546
    },
    {
      "epoch": 0.4453770491803279,
      "grad_norm": 1.1086222900396283,
      "learning_rate": 0.00019655122761708964,
      "loss": 13.8533,
      "step": 2547
    },
    {
      "epoch": 0.445551912568306,
      "grad_norm": 1.1501977419462563,
      "learning_rate": 0.00019654753975558434,
      "loss": 14.1384,
      "step": 2548
    },
    {
      "epoch": 0.4457267759562841,
      "grad_norm": 1.1446322188216567,
      "learning_rate": 0.00019654384995800443,
      "loss": 13.6068,
      "step": 2549
    },
    {
      "epoch": 0.4459016393442623,
      "grad_norm": 1.0482959325397132,
      "learning_rate": 0.0001965401582244239,
      "loss": 13.5587,
      "step": 2550
    },
    {
      "epoch": 0.4460765027322404,
      "grad_norm": 1.1788935553385975,
      "learning_rate": 0.00019653646455491676,
      "loss": 14.0398,
      "step": 2551
    },
    {
      "epoch": 0.4462513661202186,
      "grad_norm": 1.177807394751218,
      "learning_rate": 0.00019653276894955713,
      "loss": 13.7908,
      "step": 2552
    },
    {
      "epoch": 0.4464262295081967,
      "grad_norm": 1.0733675964632219,
      "learning_rate": 0.00019652907140841906,
      "loss": 13.8923,
      "step": 2553
    },
    {
      "epoch": 0.4466010928961749,
      "grad_norm": 1.102208611164003,
      "learning_rate": 0.00019652537193157673,
      "loss": 13.876,
      "step": 2554
    },
    {
      "epoch": 0.446775956284153,
      "grad_norm": 0.9724597420746719,
      "learning_rate": 0.00019652167051910434,
      "loss": 13.6896,
      "step": 2555
    },
    {
      "epoch": 0.4469508196721311,
      "grad_norm": 1.146106201122605,
      "learning_rate": 0.00019651796717107606,
      "loss": 13.8335,
      "step": 2556
    },
    {
      "epoch": 0.4471256830601093,
      "grad_norm": 1.1995327164986638,
      "learning_rate": 0.00019651426188756622,
      "loss": 14.1027,
      "step": 2557
    },
    {
      "epoch": 0.4473005464480874,
      "grad_norm": 1.2581794176492505,
      "learning_rate": 0.00019651055466864906,
      "loss": 13.9437,
      "step": 2558
    },
    {
      "epoch": 0.4474754098360656,
      "grad_norm": 1.3222653620211815,
      "learning_rate": 0.000196506845514399,
      "loss": 14.0054,
      "step": 2559
    },
    {
      "epoch": 0.4476502732240437,
      "grad_norm": 1.1015703848334113,
      "learning_rate": 0.00019650313442489035,
      "loss": 14.1016,
      "step": 2560
    },
    {
      "epoch": 0.44782513661202183,
      "grad_norm": 1.1650426076779747,
      "learning_rate": 0.00019649942140019755,
      "loss": 13.8276,
      "step": 2561
    },
    {
      "epoch": 0.448,
      "grad_norm": 1.0877612797416156,
      "learning_rate": 0.00019649570644039507,
      "loss": 13.9787,
      "step": 2562
    },
    {
      "epoch": 0.44817486338797813,
      "grad_norm": 1.1635613939044134,
      "learning_rate": 0.00019649198954555737,
      "loss": 13.9115,
      "step": 2563
    },
    {
      "epoch": 0.4483497267759563,
      "grad_norm": 1.2102392652415792,
      "learning_rate": 0.000196488270715759,
      "loss": 13.8699,
      "step": 2564
    },
    {
      "epoch": 0.4485245901639344,
      "grad_norm": 1.0716365046804226,
      "learning_rate": 0.0001964845499510746,
      "loss": 13.9502,
      "step": 2565
    },
    {
      "epoch": 0.44869945355191254,
      "grad_norm": 1.3898965353473622,
      "learning_rate": 0.00019648082725157867,
      "loss": 14.0084,
      "step": 2566
    },
    {
      "epoch": 0.4488743169398907,
      "grad_norm": 1.0790147176337181,
      "learning_rate": 0.00019647710261734596,
      "loss": 13.9534,
      "step": 2567
    },
    {
      "epoch": 0.44904918032786884,
      "grad_norm": 1.41308647244942,
      "learning_rate": 0.00019647337604845107,
      "loss": 13.9188,
      "step": 2568
    },
    {
      "epoch": 0.449224043715847,
      "grad_norm": 1.1219043958815977,
      "learning_rate": 0.00019646964754496885,
      "loss": 13.9426,
      "step": 2569
    },
    {
      "epoch": 0.44939890710382513,
      "grad_norm": 1.287210120509753,
      "learning_rate": 0.00019646591710697396,
      "loss": 14.0623,
      "step": 2570
    },
    {
      "epoch": 0.4495737704918033,
      "grad_norm": 1.1423667614545676,
      "learning_rate": 0.00019646218473454123,
      "loss": 13.8632,
      "step": 2571
    },
    {
      "epoch": 0.4497486338797814,
      "grad_norm": 1.234254074708637,
      "learning_rate": 0.00019645845042774553,
      "loss": 13.8379,
      "step": 2572
    },
    {
      "epoch": 0.44992349726775954,
      "grad_norm": 1.186818739983474,
      "learning_rate": 0.00019645471418666173,
      "loss": 14.0708,
      "step": 2573
    },
    {
      "epoch": 0.4500983606557377,
      "grad_norm": 1.114880804609313,
      "learning_rate": 0.00019645097601136475,
      "loss": 13.9475,
      "step": 2574
    },
    {
      "epoch": 0.45027322404371584,
      "grad_norm": 1.1834125168845164,
      "learning_rate": 0.00019644723590192954,
      "loss": 13.4621,
      "step": 2575
    },
    {
      "epoch": 0.450448087431694,
      "grad_norm": 1.0876906576982528,
      "learning_rate": 0.00019644349385843115,
      "loss": 14.0588,
      "step": 2576
    },
    {
      "epoch": 0.45062295081967213,
      "grad_norm": 1.1189816867627436,
      "learning_rate": 0.00019643974988094458,
      "loss": 13.9862,
      "step": 2577
    },
    {
      "epoch": 0.45079781420765025,
      "grad_norm": 1.0609984182895655,
      "learning_rate": 0.00019643600396954494,
      "loss": 13.7481,
      "step": 2578
    },
    {
      "epoch": 0.4509726775956284,
      "grad_norm": 1.1040933951594838,
      "learning_rate": 0.0001964322561243073,
      "loss": 14.0394,
      "step": 2579
    },
    {
      "epoch": 0.45114754098360654,
      "grad_norm": 1.0428909076699868,
      "learning_rate": 0.0001964285063453068,
      "loss": 13.9802,
      "step": 2580
    },
    {
      "epoch": 0.4513224043715847,
      "grad_norm": 1.078712334454539,
      "learning_rate": 0.00019642475463261874,
      "loss": 13.7504,
      "step": 2581
    },
    {
      "epoch": 0.45149726775956284,
      "grad_norm": 1.1719479866603142,
      "learning_rate": 0.00019642100098631824,
      "loss": 13.8418,
      "step": 2582
    },
    {
      "epoch": 0.45167213114754096,
      "grad_norm": 1.0440307772953115,
      "learning_rate": 0.00019641724540648065,
      "loss": 13.8125,
      "step": 2583
    },
    {
      "epoch": 0.45184699453551913,
      "grad_norm": 1.090984585825755,
      "learning_rate": 0.00019641348789318122,
      "loss": 13.9558,
      "step": 2584
    },
    {
      "epoch": 0.45202185792349725,
      "grad_norm": 1.0076446591941377,
      "learning_rate": 0.00019640972844649535,
      "loss": 13.6702,
      "step": 2585
    },
    {
      "epoch": 0.4521967213114754,
      "grad_norm": 1.0037070085248199,
      "learning_rate": 0.00019640596706649841,
      "loss": 13.819,
      "step": 2586
    },
    {
      "epoch": 0.45237158469945354,
      "grad_norm": 1.0176634082334775,
      "learning_rate": 0.00019640220375326584,
      "loss": 13.8489,
      "step": 2587
    },
    {
      "epoch": 0.4525464480874317,
      "grad_norm": 1.0287168523371073,
      "learning_rate": 0.00019639843850687304,
      "loss": 13.9384,
      "step": 2588
    },
    {
      "epoch": 0.45272131147540984,
      "grad_norm": 1.0629073353117022,
      "learning_rate": 0.0001963946713273956,
      "loss": 13.8922,
      "step": 2589
    },
    {
      "epoch": 0.45289617486338796,
      "grad_norm": 1.0301735973142168,
      "learning_rate": 0.00019639090221490905,
      "loss": 14.0363,
      "step": 2590
    },
    {
      "epoch": 0.45307103825136613,
      "grad_norm": 1.0209080025583852,
      "learning_rate": 0.00019638713116948891,
      "loss": 13.7473,
      "step": 2591
    },
    {
      "epoch": 0.45324590163934425,
      "grad_norm": 0.9926121553649988,
      "learning_rate": 0.00019638335819121086,
      "loss": 13.8775,
      "step": 2592
    },
    {
      "epoch": 0.4534207650273224,
      "grad_norm": 0.9788182952309901,
      "learning_rate": 0.00019637958328015055,
      "loss": 13.9436,
      "step": 2593
    },
    {
      "epoch": 0.45359562841530054,
      "grad_norm": 1.0320385978599012,
      "learning_rate": 0.00019637580643638364,
      "loss": 13.9677,
      "step": 2594
    },
    {
      "epoch": 0.45377049180327866,
      "grad_norm": 0.9888815602492301,
      "learning_rate": 0.00019637202765998592,
      "loss": 13.557,
      "step": 2595
    },
    {
      "epoch": 0.45394535519125684,
      "grad_norm": 1.0918985624970399,
      "learning_rate": 0.00019636824695103315,
      "loss": 13.4684,
      "step": 2596
    },
    {
      "epoch": 0.45412021857923496,
      "grad_norm": 1.0497562993989233,
      "learning_rate": 0.00019636446430960114,
      "loss": 13.9031,
      "step": 2597
    },
    {
      "epoch": 0.45429508196721313,
      "grad_norm": 0.9367791121907487,
      "learning_rate": 0.00019636067973576572,
      "loss": 13.8654,
      "step": 2598
    },
    {
      "epoch": 0.45446994535519125,
      "grad_norm": 1.061515828144144,
      "learning_rate": 0.00019635689322960285,
      "loss": 13.7348,
      "step": 2599
    },
    {
      "epoch": 0.4546448087431694,
      "grad_norm": 1.0782401166100366,
      "learning_rate": 0.0001963531047911884,
      "loss": 13.8129,
      "step": 2600
    },
    {
      "epoch": 0.45481967213114755,
      "grad_norm": 1.000705908514991,
      "learning_rate": 0.00019634931442059832,
      "loss": 13.8463,
      "step": 2601
    },
    {
      "epoch": 0.45499453551912566,
      "grad_norm": 1.1449287004294089,
      "learning_rate": 0.0001963455221179087,
      "loss": 13.7388,
      "step": 2602
    },
    {
      "epoch": 0.45516939890710384,
      "grad_norm": 1.152408044360486,
      "learning_rate": 0.00019634172788319553,
      "loss": 13.8015,
      "step": 2603
    },
    {
      "epoch": 0.45534426229508196,
      "grad_norm": 1.0829456006763882,
      "learning_rate": 0.00019633793171653488,
      "loss": 13.9272,
      "step": 2604
    },
    {
      "epoch": 0.45551912568306013,
      "grad_norm": 1.2231001186472517,
      "learning_rate": 0.00019633413361800291,
      "loss": 13.7362,
      "step": 2605
    },
    {
      "epoch": 0.45569398907103825,
      "grad_norm": 1.0277837805573833,
      "learning_rate": 0.00019633033358767578,
      "loss": 13.56,
      "step": 2606
    },
    {
      "epoch": 0.45586885245901637,
      "grad_norm": 1.0937384073263627,
      "learning_rate": 0.0001963265316256297,
      "loss": 13.788,
      "step": 2607
    },
    {
      "epoch": 0.45604371584699455,
      "grad_norm": 1.1790657681651193,
      "learning_rate": 0.00019632272773194092,
      "loss": 13.8793,
      "step": 2608
    },
    {
      "epoch": 0.45621857923497267,
      "grad_norm": 0.9897357741688928,
      "learning_rate": 0.00019631892190668566,
      "loss": 13.4406,
      "step": 2609
    },
    {
      "epoch": 0.45639344262295084,
      "grad_norm": 0.9690508178990758,
      "learning_rate": 0.00019631511414994028,
      "loss": 13.7134,
      "step": 2610
    },
    {
      "epoch": 0.45656830601092896,
      "grad_norm": 1.0399004827448508,
      "learning_rate": 0.00019631130446178116,
      "loss": 14.0401,
      "step": 2611
    },
    {
      "epoch": 0.4567431693989071,
      "grad_norm": 0.9838705187087681,
      "learning_rate": 0.00019630749284228467,
      "loss": 13.9022,
      "step": 2612
    },
    {
      "epoch": 0.45691803278688525,
      "grad_norm": 0.9987557181715055,
      "learning_rate": 0.00019630367929152724,
      "loss": 13.7689,
      "step": 2613
    },
    {
      "epoch": 0.4570928961748634,
      "grad_norm": 1.064566153817096,
      "learning_rate": 0.00019629986380958534,
      "loss": 13.859,
      "step": 2614
    },
    {
      "epoch": 0.45726775956284155,
      "grad_norm": 1.037739993234069,
      "learning_rate": 0.0001962960463965355,
      "loss": 13.6947,
      "step": 2615
    },
    {
      "epoch": 0.45744262295081967,
      "grad_norm": 1.1040571864776305,
      "learning_rate": 0.00019629222705245424,
      "loss": 13.7401,
      "step": 2616
    },
    {
      "epoch": 0.45761748633879784,
      "grad_norm": 1.1892653010782175,
      "learning_rate": 0.00019628840577741818,
      "loss": 13.7448,
      "step": 2617
    },
    {
      "epoch": 0.45779234972677596,
      "grad_norm": 1.032217521246257,
      "learning_rate": 0.00019628458257150397,
      "loss": 13.5975,
      "step": 2618
    },
    {
      "epoch": 0.4579672131147541,
      "grad_norm": 1.0632174652782513,
      "learning_rate": 0.00019628075743478821,
      "loss": 13.7831,
      "step": 2619
    },
    {
      "epoch": 0.45814207650273225,
      "grad_norm": 1.1552149710868276,
      "learning_rate": 0.0001962769303673477,
      "loss": 14.1046,
      "step": 2620
    },
    {
      "epoch": 0.4583169398907104,
      "grad_norm": 1.0060944525800903,
      "learning_rate": 0.00019627310136925905,
      "loss": 13.8581,
      "step": 2621
    },
    {
      "epoch": 0.45849180327868855,
      "grad_norm": 1.1043722805604594,
      "learning_rate": 0.00019626927044059914,
      "loss": 13.9629,
      "step": 2622
    },
    {
      "epoch": 0.45866666666666667,
      "grad_norm": 1.0565154862627804,
      "learning_rate": 0.0001962654375814448,
      "loss": 14.01,
      "step": 2623
    },
    {
      "epoch": 0.4588415300546448,
      "grad_norm": 1.095111250898946,
      "learning_rate": 0.00019626160279187288,
      "loss": 13.8715,
      "step": 2624
    },
    {
      "epoch": 0.45901639344262296,
      "grad_norm": 1.018507036396285,
      "learning_rate": 0.00019625776607196024,
      "loss": 13.6073,
      "step": 2625
    },
    {
      "epoch": 0.4591912568306011,
      "grad_norm": 1.04059778928093,
      "learning_rate": 0.0001962539274217838,
      "loss": 13.8844,
      "step": 2626
    },
    {
      "epoch": 0.45936612021857925,
      "grad_norm": 0.9420816328483748,
      "learning_rate": 0.00019625008684142064,
      "loss": 13.6661,
      "step": 2627
    },
    {
      "epoch": 0.4595409836065574,
      "grad_norm": 1.0326470410129935,
      "learning_rate": 0.00019624624433094766,
      "loss": 13.8977,
      "step": 2628
    },
    {
      "epoch": 0.4597158469945355,
      "grad_norm": 0.9242520982947965,
      "learning_rate": 0.000196242399890442,
      "loss": 13.4789,
      "step": 2629
    },
    {
      "epoch": 0.45989071038251367,
      "grad_norm": 1.0162403426072197,
      "learning_rate": 0.00019623855351998072,
      "loss": 13.872,
      "step": 2630
    },
    {
      "epoch": 0.4600655737704918,
      "grad_norm": 0.9509590171564628,
      "learning_rate": 0.00019623470521964092,
      "loss": 13.8442,
      "step": 2631
    },
    {
      "epoch": 0.46024043715846996,
      "grad_norm": 0.9509292692419886,
      "learning_rate": 0.00019623085498949984,
      "loss": 13.9133,
      "step": 2632
    },
    {
      "epoch": 0.4604153005464481,
      "grad_norm": 1.0270429277021516,
      "learning_rate": 0.0001962270028296346,
      "loss": 13.651,
      "step": 2633
    },
    {
      "epoch": 0.46059016393442626,
      "grad_norm": 1.0657425158191947,
      "learning_rate": 0.00019622314874012252,
      "loss": 13.5267,
      "step": 2634
    },
    {
      "epoch": 0.4607650273224044,
      "grad_norm": 0.9909042607304032,
      "learning_rate": 0.0001962192927210409,
      "loss": 13.9639,
      "step": 2635
    },
    {
      "epoch": 0.4609398907103825,
      "grad_norm": 0.9981451956903462,
      "learning_rate": 0.000196215434772467,
      "loss": 13.767,
      "step": 2636
    },
    {
      "epoch": 0.46111475409836067,
      "grad_norm": 1.0326193329014157,
      "learning_rate": 0.00019621157489447822,
      "loss": 13.8099,
      "step": 2637
    },
    {
      "epoch": 0.4612896174863388,
      "grad_norm": 0.99755228139774,
      "learning_rate": 0.00019620771308715194,
      "loss": 13.6389,
      "step": 2638
    },
    {
      "epoch": 0.46146448087431696,
      "grad_norm": 1.0618974237263046,
      "learning_rate": 0.00019620384935056562,
      "loss": 13.9231,
      "step": 2639
    },
    {
      "epoch": 0.4616393442622951,
      "grad_norm": 0.9608221548226974,
      "learning_rate": 0.00019619998368479674,
      "loss": 13.5521,
      "step": 2640
    },
    {
      "epoch": 0.4618142076502732,
      "grad_norm": 1.0087494016638074,
      "learning_rate": 0.00019619611608992282,
      "loss": 13.8345,
      "step": 2641
    },
    {
      "epoch": 0.4619890710382514,
      "grad_norm": 1.0449738251612632,
      "learning_rate": 0.0001961922465660214,
      "loss": 13.7423,
      "step": 2642
    },
    {
      "epoch": 0.4621639344262295,
      "grad_norm": 0.9664197046984325,
      "learning_rate": 0.00019618837511317007,
      "loss": 13.7193,
      "step": 2643
    },
    {
      "epoch": 0.46233879781420767,
      "grad_norm": 1.1143191228731828,
      "learning_rate": 0.00019618450173144653,
      "loss": 13.742,
      "step": 2644
    },
    {
      "epoch": 0.4625136612021858,
      "grad_norm": 1.0268000417235514,
      "learning_rate": 0.00019618062642092835,
      "loss": 13.6477,
      "step": 2645
    },
    {
      "epoch": 0.4626885245901639,
      "grad_norm": 1.0181185244797772,
      "learning_rate": 0.00019617674918169335,
      "loss": 13.9313,
      "step": 2646
    },
    {
      "epoch": 0.4628633879781421,
      "grad_norm": 1.0386985803494657,
      "learning_rate": 0.0001961728700138192,
      "loss": 13.8707,
      "step": 2647
    },
    {
      "epoch": 0.4630382513661202,
      "grad_norm": 1.0538069343293113,
      "learning_rate": 0.00019616898891738373,
      "loss": 13.7694,
      "step": 2648
    },
    {
      "epoch": 0.4632131147540984,
      "grad_norm": 1.0434518553576675,
      "learning_rate": 0.00019616510589246474,
      "loss": 13.7778,
      "step": 2649
    },
    {
      "epoch": 0.4633879781420765,
      "grad_norm": 1.1328779867905523,
      "learning_rate": 0.00019616122093914015,
      "loss": 14.0027,
      "step": 2650
    },
    {
      "epoch": 0.46356284153005467,
      "grad_norm": 1.0541918152514118,
      "learning_rate": 0.00019615733405748778,
      "loss": 13.8976,
      "step": 2651
    },
    {
      "epoch": 0.4637377049180328,
      "grad_norm": 1.0704220569763563,
      "learning_rate": 0.00019615344524758564,
      "loss": 13.6726,
      "step": 2652
    },
    {
      "epoch": 0.4639125683060109,
      "grad_norm": 0.9837505622885557,
      "learning_rate": 0.00019614955450951167,
      "loss": 13.9023,
      "step": 2653
    },
    {
      "epoch": 0.4640874316939891,
      "grad_norm": 1.0419821025577791,
      "learning_rate": 0.00019614566184334394,
      "loss": 13.6092,
      "step": 2654
    },
    {
      "epoch": 0.4642622950819672,
      "grad_norm": 1.0910760674263213,
      "learning_rate": 0.00019614176724916046,
      "loss": 14.0837,
      "step": 2655
    },
    {
      "epoch": 0.4644371584699454,
      "grad_norm": 0.9613297213453684,
      "learning_rate": 0.00019613787072703936,
      "loss": 13.6859,
      "step": 2656
    },
    {
      "epoch": 0.4646120218579235,
      "grad_norm": 1.0015459491508751,
      "learning_rate": 0.00019613397227705878,
      "loss": 13.4812,
      "step": 2657
    },
    {
      "epoch": 0.4647868852459016,
      "grad_norm": 1.04046226040371,
      "learning_rate": 0.00019613007189929688,
      "loss": 14.0154,
      "step": 2658
    },
    {
      "epoch": 0.4649617486338798,
      "grad_norm": 1.0250748025523344,
      "learning_rate": 0.0001961261695938319,
      "loss": 13.9058,
      "step": 2659
    },
    {
      "epoch": 0.4651366120218579,
      "grad_norm": 1.0336973880233982,
      "learning_rate": 0.00019612226536074205,
      "loss": 13.6752,
      "step": 2660
    },
    {
      "epoch": 0.4653114754098361,
      "grad_norm": 1.0344952891474248,
      "learning_rate": 0.00019611835920010565,
      "loss": 13.7135,
      "step": 2661
    },
    {
      "epoch": 0.4654863387978142,
      "grad_norm": 1.0352232976882978,
      "learning_rate": 0.00019611445111200105,
      "loss": 13.7505,
      "step": 2662
    },
    {
      "epoch": 0.4656612021857923,
      "grad_norm": 1.1541444164025363,
      "learning_rate": 0.00019611054109650654,
      "loss": 13.7711,
      "step": 2663
    },
    {
      "epoch": 0.4658360655737705,
      "grad_norm": 1.044593233240849,
      "learning_rate": 0.00019610662915370064,
      "loss": 13.884,
      "step": 2664
    },
    {
      "epoch": 0.4660109289617486,
      "grad_norm": 1.0575889324033507,
      "learning_rate": 0.0001961027152836617,
      "loss": 13.4015,
      "step": 2665
    },
    {
      "epoch": 0.4661857923497268,
      "grad_norm": 1.135278830867325,
      "learning_rate": 0.00019609879948646828,
      "loss": 14.0628,
      "step": 2666
    },
    {
      "epoch": 0.4663606557377049,
      "grad_norm": 0.9703963922434028,
      "learning_rate": 0.00019609488176219886,
      "loss": 13.576,
      "step": 2667
    },
    {
      "epoch": 0.4665355191256831,
      "grad_norm": 1.0325834014782114,
      "learning_rate": 0.000196090962110932,
      "loss": 13.6262,
      "step": 2668
    },
    {
      "epoch": 0.4667103825136612,
      "grad_norm": 1.0520066102915246,
      "learning_rate": 0.0001960870405327463,
      "loss": 13.7863,
      "step": 2669
    },
    {
      "epoch": 0.4668852459016393,
      "grad_norm": 1.122989086673134,
      "learning_rate": 0.0001960831170277204,
      "loss": 13.6055,
      "step": 2670
    },
    {
      "epoch": 0.4670601092896175,
      "grad_norm": 1.0030457579580943,
      "learning_rate": 0.000196079191595933,
      "loss": 13.806,
      "step": 2671
    },
    {
      "epoch": 0.4672349726775956,
      "grad_norm": 1.0159549984136218,
      "learning_rate": 0.00019607526423746287,
      "loss": 14.023,
      "step": 2672
    },
    {
      "epoch": 0.4674098360655738,
      "grad_norm": 0.9919097279684649,
      "learning_rate": 0.00019607133495238863,
      "loss": 13.7396,
      "step": 2673
    },
    {
      "epoch": 0.4675846994535519,
      "grad_norm": 0.9864738149710488,
      "learning_rate": 0.00019606740374078916,
      "loss": 14.0044,
      "step": 2674
    },
    {
      "epoch": 0.46775956284153003,
      "grad_norm": 0.9882086468448674,
      "learning_rate": 0.00019606347060274328,
      "loss": 13.6021,
      "step": 2675
    },
    {
      "epoch": 0.4679344262295082,
      "grad_norm": 1.0294247089359396,
      "learning_rate": 0.00019605953553832988,
      "loss": 13.7455,
      "step": 2676
    },
    {
      "epoch": 0.4681092896174863,
      "grad_norm": 1.0201473236869876,
      "learning_rate": 0.00019605559854762782,
      "loss": 13.8671,
      "step": 2677
    },
    {
      "epoch": 0.4682841530054645,
      "grad_norm": 1.0493122679228861,
      "learning_rate": 0.00019605165963071613,
      "loss": 13.6891,
      "step": 2678
    },
    {
      "epoch": 0.4684590163934426,
      "grad_norm": 1.035683863531544,
      "learning_rate": 0.0001960477187876737,
      "loss": 13.7566,
      "step": 2679
    },
    {
      "epoch": 0.4686338797814208,
      "grad_norm": 0.8958007054126936,
      "learning_rate": 0.00019604377601857963,
      "loss": 13.6171,
      "step": 2680
    },
    {
      "epoch": 0.4688087431693989,
      "grad_norm": 1.0473745271264012,
      "learning_rate": 0.00019603983132351294,
      "loss": 13.4582,
      "step": 2681
    },
    {
      "epoch": 0.46898360655737703,
      "grad_norm": 1.0742564353415618,
      "learning_rate": 0.00019603588470255273,
      "loss": 13.8717,
      "step": 2682
    },
    {
      "epoch": 0.4691584699453552,
      "grad_norm": 1.126778030898051,
      "learning_rate": 0.00019603193615577822,
      "loss": 13.6623,
      "step": 2683
    },
    {
      "epoch": 0.4693333333333333,
      "grad_norm": 1.237262849253663,
      "learning_rate": 0.0001960279856832685,
      "loss": 13.8871,
      "step": 2684
    },
    {
      "epoch": 0.4695081967213115,
      "grad_norm": 0.9330197722337277,
      "learning_rate": 0.0001960240332851028,
      "loss": 13.7438,
      "step": 2685
    },
    {
      "epoch": 0.4696830601092896,
      "grad_norm": 1.2836036034855758,
      "learning_rate": 0.00019602007896136044,
      "loss": 13.6775,
      "step": 2686
    },
    {
      "epoch": 0.46985792349726774,
      "grad_norm": 1.0796547882310448,
      "learning_rate": 0.00019601612271212066,
      "loss": 13.9453,
      "step": 2687
    },
    {
      "epoch": 0.4700327868852459,
      "grad_norm": 1.024670303207717,
      "learning_rate": 0.00019601216453746283,
      "loss": 13.7573,
      "step": 2688
    },
    {
      "epoch": 0.47020765027322403,
      "grad_norm": 1.1947842705125948,
      "learning_rate": 0.0001960082044374663,
      "loss": 13.6767,
      "step": 2689
    },
    {
      "epoch": 0.4703825136612022,
      "grad_norm": 0.9638463252850846,
      "learning_rate": 0.00019600424241221048,
      "loss": 13.8194,
      "step": 2690
    },
    {
      "epoch": 0.4705573770491803,
      "grad_norm": 1.2155714676672402,
      "learning_rate": 0.0001960002784617748,
      "loss": 13.8437,
      "step": 2691
    },
    {
      "epoch": 0.47073224043715844,
      "grad_norm": 1.1136924068881278,
      "learning_rate": 0.0001959963125862388,
      "loss": 13.8163,
      "step": 2692
    },
    {
      "epoch": 0.4709071038251366,
      "grad_norm": 1.0095196728123583,
      "learning_rate": 0.000195992344785682,
      "loss": 13.6908,
      "step": 2693
    },
    {
      "epoch": 0.47108196721311474,
      "grad_norm": 1.140352857286733,
      "learning_rate": 0.00019598837506018391,
      "loss": 14.0336,
      "step": 2694
    },
    {
      "epoch": 0.4712568306010929,
      "grad_norm": 0.9792269661421846,
      "learning_rate": 0.00019598440340982422,
      "loss": 13.6855,
      "step": 2695
    },
    {
      "epoch": 0.47143169398907103,
      "grad_norm": 1.241722756552639,
      "learning_rate": 0.00019598042983468253,
      "loss": 13.8106,
      "step": 2696
    },
    {
      "epoch": 0.4716065573770492,
      "grad_norm": 1.0045595046135138,
      "learning_rate": 0.0001959764543348385,
      "loss": 13.6006,
      "step": 2697
    },
    {
      "epoch": 0.4717814207650273,
      "grad_norm": 1.0751251170861742,
      "learning_rate": 0.00019597247691037188,
      "loss": 13.7772,
      "step": 2698
    },
    {
      "epoch": 0.47195628415300545,
      "grad_norm": 0.9455105399131818,
      "learning_rate": 0.00019596849756136245,
      "loss": 13.6543,
      "step": 2699
    },
    {
      "epoch": 0.4721311475409836,
      "grad_norm": 1.0606153430714838,
      "learning_rate": 0.00019596451628788993,
      "loss": 13.6388,
      "step": 2700
    },
    {
      "epoch": 0.47230601092896174,
      "grad_norm": 0.9978268275196572,
      "learning_rate": 0.00019596053309003423,
      "loss": 12.9305,
      "step": 2701
    },
    {
      "epoch": 0.4724808743169399,
      "grad_norm": 1.081200574146782,
      "learning_rate": 0.00019595654796787524,
      "loss": 13.7923,
      "step": 2702
    },
    {
      "epoch": 0.47265573770491803,
      "grad_norm": 0.9908405564309415,
      "learning_rate": 0.0001959525609214928,
      "loss": 13.4928,
      "step": 2703
    },
    {
      "epoch": 0.47283060109289615,
      "grad_norm": 1.0531099226834755,
      "learning_rate": 0.0001959485719509669,
      "loss": 13.8644,
      "step": 2704
    },
    {
      "epoch": 0.4730054644808743,
      "grad_norm": 1.0194744533109026,
      "learning_rate": 0.00019594458105637755,
      "loss": 13.4213,
      "step": 2705
    },
    {
      "epoch": 0.47318032786885245,
      "grad_norm": 1.0349234750011018,
      "learning_rate": 0.00019594058823780474,
      "loss": 13.822,
      "step": 2706
    },
    {
      "epoch": 0.4733551912568306,
      "grad_norm": 1.0793904729131365,
      "learning_rate": 0.00019593659349532858,
      "loss": 13.5784,
      "step": 2707
    },
    {
      "epoch": 0.47353005464480874,
      "grad_norm": 1.0379051618817436,
      "learning_rate": 0.00019593259682902913,
      "loss": 13.6262,
      "step": 2708
    },
    {
      "epoch": 0.47370491803278686,
      "grad_norm": 1.1235637441199646,
      "learning_rate": 0.00019592859823898659,
      "loss": 14.041,
      "step": 2709
    },
    {
      "epoch": 0.47387978142076503,
      "grad_norm": 1.0857711345989274,
      "learning_rate": 0.0001959245977252811,
      "loss": 13.7948,
      "step": 2710
    },
    {
      "epoch": 0.47405464480874315,
      "grad_norm": 0.9957185603209905,
      "learning_rate": 0.00019592059528799293,
      "loss": 13.8103,
      "step": 2711
    },
    {
      "epoch": 0.4742295081967213,
      "grad_norm": 1.0261648229854738,
      "learning_rate": 0.00019591659092720227,
      "loss": 13.7659,
      "step": 2712
    },
    {
      "epoch": 0.47440437158469945,
      "grad_norm": 0.9765459967592206,
      "learning_rate": 0.00019591258464298947,
      "loss": 13.426,
      "step": 2713
    },
    {
      "epoch": 0.4745792349726776,
      "grad_norm": 1.0529507256152548,
      "learning_rate": 0.00019590857643543488,
      "loss": 13.8186,
      "step": 2714
    },
    {
      "epoch": 0.47475409836065574,
      "grad_norm": 0.9758788864926666,
      "learning_rate": 0.0001959045663046188,
      "loss": 13.7519,
      "step": 2715
    },
    {
      "epoch": 0.47492896174863386,
      "grad_norm": 1.0858998292001636,
      "learning_rate": 0.00019590055425062176,
      "loss": 13.2631,
      "step": 2716
    },
    {
      "epoch": 0.47510382513661203,
      "grad_norm": 0.955172919022088,
      "learning_rate": 0.00019589654027352414,
      "loss": 13.832,
      "step": 2717
    },
    {
      "epoch": 0.47527868852459015,
      "grad_norm": 0.9823757914470737,
      "learning_rate": 0.00019589252437340643,
      "loss": 13.8122,
      "step": 2718
    },
    {
      "epoch": 0.47545355191256833,
      "grad_norm": 1.0864232014894253,
      "learning_rate": 0.00019588850655034918,
      "loss": 13.5325,
      "step": 2719
    },
    {
      "epoch": 0.47562841530054645,
      "grad_norm": 0.9500097323986885,
      "learning_rate": 0.00019588448680443296,
      "loss": 13.8365,
      "step": 2720
    },
    {
      "epoch": 0.47580327868852457,
      "grad_norm": 1.047611920518033,
      "learning_rate": 0.00019588046513573839,
      "loss": 13.7978,
      "step": 2721
    },
    {
      "epoch": 0.47597814207650274,
      "grad_norm": 0.9424707496443109,
      "learning_rate": 0.00019587644154434605,
      "loss": 13.8419,
      "step": 2722
    },
    {
      "epoch": 0.47615300546448086,
      "grad_norm": 1.0174943835858263,
      "learning_rate": 0.00019587241603033674,
      "loss": 13.7354,
      "step": 2723
    },
    {
      "epoch": 0.47632786885245904,
      "grad_norm": 0.9752791571887294,
      "learning_rate": 0.00019586838859379107,
      "loss": 13.6468,
      "step": 2724
    },
    {
      "epoch": 0.47650273224043715,
      "grad_norm": 1.0458457489945474,
      "learning_rate": 0.0001958643592347899,
      "loss": 13.7093,
      "step": 2725
    },
    {
      "epoch": 0.4766775956284153,
      "grad_norm": 1.0232093483527367,
      "learning_rate": 0.00019586032795341398,
      "loss": 13.7837,
      "step": 2726
    },
    {
      "epoch": 0.47685245901639345,
      "grad_norm": 0.9561222087817391,
      "learning_rate": 0.00019585629474974415,
      "loss": 13.746,
      "step": 2727
    },
    {
      "epoch": 0.47702732240437157,
      "grad_norm": 1.1148064729119609,
      "learning_rate": 0.00019585225962386128,
      "loss": 13.7702,
      "step": 2728
    },
    {
      "epoch": 0.47720218579234974,
      "grad_norm": 1.027774517325381,
      "learning_rate": 0.0001958482225758463,
      "loss": 13.7401,
      "step": 2729
    },
    {
      "epoch": 0.47737704918032786,
      "grad_norm": 1.056377907155402,
      "learning_rate": 0.00019584418360578016,
      "loss": 13.6516,
      "step": 2730
    },
    {
      "epoch": 0.47755191256830604,
      "grad_norm": 0.9540716560689773,
      "learning_rate": 0.00019584014271374387,
      "loss": 13.6757,
      "step": 2731
    },
    {
      "epoch": 0.47772677595628416,
      "grad_norm": 1.0424093704970994,
      "learning_rate": 0.00019583609989981847,
      "loss": 13.8274,
      "step": 2732
    },
    {
      "epoch": 0.4779016393442623,
      "grad_norm": 0.9776794866596291,
      "learning_rate": 0.00019583205516408495,
      "loss": 13.8345,
      "step": 2733
    },
    {
      "epoch": 0.47807650273224045,
      "grad_norm": 1.007365315085525,
      "learning_rate": 0.00019582800850662455,
      "loss": 13.7747,
      "step": 2734
    },
    {
      "epoch": 0.47825136612021857,
      "grad_norm": 1.0749625631824524,
      "learning_rate": 0.00019582395992751833,
      "loss": 13.9528,
      "step": 2735
    },
    {
      "epoch": 0.47842622950819674,
      "grad_norm": 0.9435655577147557,
      "learning_rate": 0.00019581990942684746,
      "loss": 13.9215,
      "step": 2736
    },
    {
      "epoch": 0.47860109289617486,
      "grad_norm": 0.9763515189293828,
      "learning_rate": 0.00019581585700469324,
      "loss": 13.3989,
      "step": 2737
    },
    {
      "epoch": 0.478775956284153,
      "grad_norm": 0.9517505251518643,
      "learning_rate": 0.00019581180266113686,
      "loss": 13.7244,
      "step": 2738
    },
    {
      "epoch": 0.47895081967213116,
      "grad_norm": 1.1287528323345009,
      "learning_rate": 0.00019580774639625968,
      "loss": 13.7205,
      "step": 2739
    },
    {
      "epoch": 0.4791256830601093,
      "grad_norm": 0.9960056506241635,
      "learning_rate": 0.00019580368821014302,
      "loss": 13.548,
      "step": 2740
    },
    {
      "epoch": 0.47930054644808745,
      "grad_norm": 1.0454397447807775,
      "learning_rate": 0.00019579962810286822,
      "loss": 13.7561,
      "step": 2741
    },
    {
      "epoch": 0.47947540983606557,
      "grad_norm": 0.994305604816776,
      "learning_rate": 0.00019579556607451676,
      "loss": 13.7588,
      "step": 2742
    },
    {
      "epoch": 0.4796502732240437,
      "grad_norm": 0.9469005433506799,
      "learning_rate": 0.00019579150212517007,
      "loss": 13.4303,
      "step": 2743
    },
    {
      "epoch": 0.47982513661202186,
      "grad_norm": 0.8975159016080743,
      "learning_rate": 0.00019578743625490964,
      "loss": 13.6108,
      "step": 2744
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0661463575141834,
      "learning_rate": 0.000195783368463817,
      "loss": 13.6488,
      "step": 2745
    },
    {
      "epoch": 0.48017486338797816,
      "grad_norm": 1.0438482122527193,
      "learning_rate": 0.00019577929875197377,
      "loss": 13.6085,
      "step": 2746
    },
    {
      "epoch": 0.4803497267759563,
      "grad_norm": 0.9397794524266546,
      "learning_rate": 0.00019577522711946148,
      "loss": 13.7544,
      "step": 2747
    },
    {
      "epoch": 0.48052459016393445,
      "grad_norm": 1.0271607894747528,
      "learning_rate": 0.00019577115356636182,
      "loss": 13.6221,
      "step": 2748
    },
    {
      "epoch": 0.48069945355191257,
      "grad_norm": 0.9839405058624341,
      "learning_rate": 0.00019576707809275648,
      "loss": 13.6797,
      "step": 2749
    },
    {
      "epoch": 0.4808743169398907,
      "grad_norm": 1.0648424469037987,
      "learning_rate": 0.00019576300069872718,
      "loss": 13.7768,
      "step": 2750
    },
    {
      "epoch": 0.48104918032786886,
      "grad_norm": 0.9834882785975742,
      "learning_rate": 0.00019575892138435567,
      "loss": 13.7727,
      "step": 2751
    },
    {
      "epoch": 0.481224043715847,
      "grad_norm": 0.9829775846669789,
      "learning_rate": 0.0001957548401497238,
      "loss": 13.8122,
      "step": 2752
    },
    {
      "epoch": 0.48139890710382516,
      "grad_norm": 0.9599453605715416,
      "learning_rate": 0.00019575075699491336,
      "loss": 13.8521,
      "step": 2753
    },
    {
      "epoch": 0.4815737704918033,
      "grad_norm": 0.9730805461610138,
      "learning_rate": 0.00019574667192000624,
      "loss": 13.6121,
      "step": 2754
    },
    {
      "epoch": 0.4817486338797814,
      "grad_norm": 1.074730705258843,
      "learning_rate": 0.00019574258492508439,
      "loss": 13.4574,
      "step": 2755
    },
    {
      "epoch": 0.48192349726775957,
      "grad_norm": 1.0211068958650966,
      "learning_rate": 0.00019573849601022973,
      "loss": 13.87,
      "step": 2756
    },
    {
      "epoch": 0.4820983606557377,
      "grad_norm": 0.9528415545134309,
      "learning_rate": 0.00019573440517552427,
      "loss": 13.7676,
      "step": 2757
    },
    {
      "epoch": 0.48227322404371586,
      "grad_norm": 1.0335351343845902,
      "learning_rate": 0.00019573031242105005,
      "loss": 13.6737,
      "step": 2758
    },
    {
      "epoch": 0.482448087431694,
      "grad_norm": 0.9426218061868703,
      "learning_rate": 0.00019572621774688914,
      "loss": 13.4633,
      "step": 2759
    },
    {
      "epoch": 0.48262295081967216,
      "grad_norm": 1.001992012728544,
      "learning_rate": 0.00019572212115312362,
      "loss": 13.8014,
      "step": 2760
    },
    {
      "epoch": 0.4827978142076503,
      "grad_norm": 1.0210008326077695,
      "learning_rate": 0.00019571802263983567,
      "loss": 13.8089,
      "step": 2761
    },
    {
      "epoch": 0.4829726775956284,
      "grad_norm": 0.9720096049379108,
      "learning_rate": 0.0001957139222071075,
      "loss": 13.7349,
      "step": 2762
    },
    {
      "epoch": 0.48314754098360657,
      "grad_norm": 0.9950131541914806,
      "learning_rate": 0.0001957098198550213,
      "loss": 13.5074,
      "step": 2763
    },
    {
      "epoch": 0.4833224043715847,
      "grad_norm": 1.1566475583966214,
      "learning_rate": 0.00019570571558365934,
      "loss": 13.7652,
      "step": 2764
    },
    {
      "epoch": 0.48349726775956287,
      "grad_norm": 0.9331654167356419,
      "learning_rate": 0.0001957016093931039,
      "loss": 13.5994,
      "step": 2765
    },
    {
      "epoch": 0.483672131147541,
      "grad_norm": 1.1300859696671042,
      "learning_rate": 0.0001956975012834374,
      "loss": 13.5213,
      "step": 2766
    },
    {
      "epoch": 0.4838469945355191,
      "grad_norm": 0.962948325379912,
      "learning_rate": 0.0001956933912547421,
      "loss": 13.534,
      "step": 2767
    },
    {
      "epoch": 0.4840218579234973,
      "grad_norm": 1.0647262299764435,
      "learning_rate": 0.00019568927930710056,
      "loss": 13.7166,
      "step": 2768
    },
    {
      "epoch": 0.4841967213114754,
      "grad_norm": 1.2857003158033866,
      "learning_rate": 0.0001956851654405951,
      "loss": 13.8061,
      "step": 2769
    },
    {
      "epoch": 0.4843715846994536,
      "grad_norm": 1.0594586671780533,
      "learning_rate": 0.00019568104965530836,
      "loss": 13.4612,
      "step": 2770
    },
    {
      "epoch": 0.4845464480874317,
      "grad_norm": 0.9841042970284212,
      "learning_rate": 0.00019567693195132274,
      "loss": 13.6326,
      "step": 2771
    },
    {
      "epoch": 0.4847213114754098,
      "grad_norm": 1.0939924846608313,
      "learning_rate": 0.00019567281232872087,
      "loss": 13.5481,
      "step": 2772
    },
    {
      "epoch": 0.484896174863388,
      "grad_norm": 0.9686585997782109,
      "learning_rate": 0.0001956686907875854,
      "loss": 13.6561,
      "step": 2773
    },
    {
      "epoch": 0.4850710382513661,
      "grad_norm": 1.0542003849606387,
      "learning_rate": 0.00019566456732799886,
      "loss": 13.7142,
      "step": 2774
    },
    {
      "epoch": 0.4852459016393443,
      "grad_norm": 1.0354477260867583,
      "learning_rate": 0.0001956604419500441,
      "loss": 13.638,
      "step": 2775
    },
    {
      "epoch": 0.4854207650273224,
      "grad_norm": 0.9519187320533862,
      "learning_rate": 0.00019565631465380372,
      "loss": 13.6558,
      "step": 2776
    },
    {
      "epoch": 0.4855956284153006,
      "grad_norm": 1.0484271758713486,
      "learning_rate": 0.00019565218543936053,
      "loss": 13.9707,
      "step": 2777
    },
    {
      "epoch": 0.4857704918032787,
      "grad_norm": 1.0048633031198386,
      "learning_rate": 0.00019564805430679735,
      "loss": 13.6412,
      "step": 2778
    },
    {
      "epoch": 0.4859453551912568,
      "grad_norm": 0.9965820033738458,
      "learning_rate": 0.00019564392125619696,
      "loss": 13.5269,
      "step": 2779
    },
    {
      "epoch": 0.486120218579235,
      "grad_norm": 1.0563116916871251,
      "learning_rate": 0.0001956397862876423,
      "loss": 13.7482,
      "step": 2780
    },
    {
      "epoch": 0.4862950819672131,
      "grad_norm": 0.9945480140901124,
      "learning_rate": 0.0001956356494012163,
      "loss": 13.6303,
      "step": 2781
    },
    {
      "epoch": 0.4864699453551913,
      "grad_norm": 0.9988036090079917,
      "learning_rate": 0.00019563151059700187,
      "loss": 13.3717,
      "step": 2782
    },
    {
      "epoch": 0.4866448087431694,
      "grad_norm": 1.043257890418062,
      "learning_rate": 0.00019562736987508204,
      "loss": 13.7611,
      "step": 2783
    },
    {
      "epoch": 0.4868196721311475,
      "grad_norm": 1.0052223390323887,
      "learning_rate": 0.00019562322723553984,
      "loss": 13.7159,
      "step": 2784
    },
    {
      "epoch": 0.4869945355191257,
      "grad_norm": 1.1207412010734632,
      "learning_rate": 0.0001956190826784583,
      "loss": 13.7937,
      "step": 2785
    },
    {
      "epoch": 0.4871693989071038,
      "grad_norm": 0.9749479136415329,
      "learning_rate": 0.00019561493620392055,
      "loss": 13.6345,
      "step": 2786
    },
    {
      "epoch": 0.487344262295082,
      "grad_norm": 0.9653063902967477,
      "learning_rate": 0.00019561078781200978,
      "loss": 13.7805,
      "step": 2787
    },
    {
      "epoch": 0.4875191256830601,
      "grad_norm": 0.9362274668365378,
      "learning_rate": 0.00019560663750280914,
      "loss": 13.9104,
      "step": 2788
    },
    {
      "epoch": 0.4876939890710382,
      "grad_norm": 0.8927644542433627,
      "learning_rate": 0.00019560248527640187,
      "loss": 13.6845,
      "step": 2789
    },
    {
      "epoch": 0.4878688524590164,
      "grad_norm": 1.1927882453693432,
      "learning_rate": 0.00019559833113287124,
      "loss": 14.0009,
      "step": 2790
    },
    {
      "epoch": 0.4880437158469945,
      "grad_norm": 0.9868049888516308,
      "learning_rate": 0.00019559417507230054,
      "loss": 13.6026,
      "step": 2791
    },
    {
      "epoch": 0.4882185792349727,
      "grad_norm": 1.0291713521357282,
      "learning_rate": 0.00019559001709477315,
      "loss": 13.9716,
      "step": 2792
    },
    {
      "epoch": 0.4883934426229508,
      "grad_norm": 0.9426065066612458,
      "learning_rate": 0.00019558585720037236,
      "loss": 13.8122,
      "step": 2793
    },
    {
      "epoch": 0.488568306010929,
      "grad_norm": 1.0039042378618366,
      "learning_rate": 0.00019558169538918169,
      "loss": 13.8102,
      "step": 2794
    },
    {
      "epoch": 0.4887431693989071,
      "grad_norm": 0.9653705088755274,
      "learning_rate": 0.00019557753166128452,
      "loss": 13.5247,
      "step": 2795
    },
    {
      "epoch": 0.4889180327868852,
      "grad_norm": 1.0031795935314276,
      "learning_rate": 0.0001955733660167644,
      "loss": 13.8929,
      "step": 2796
    },
    {
      "epoch": 0.4890928961748634,
      "grad_norm": 1.1284742944247717,
      "learning_rate": 0.00019556919845570482,
      "loss": 13.8154,
      "step": 2797
    },
    {
      "epoch": 0.4892677595628415,
      "grad_norm": 1.0220232145940706,
      "learning_rate": 0.0001955650289781894,
      "loss": 13.5459,
      "step": 2798
    },
    {
      "epoch": 0.4894426229508197,
      "grad_norm": 0.8870023207483291,
      "learning_rate": 0.00019556085758430173,
      "loss": 13.3406,
      "step": 2799
    },
    {
      "epoch": 0.4896174863387978,
      "grad_norm": 1.0059281901237613,
      "learning_rate": 0.00019555668427412542,
      "loss": 13.8822,
      "step": 2800
    },
    {
      "epoch": 0.48979234972677593,
      "grad_norm": 0.9634024407759194,
      "learning_rate": 0.00019555250904774424,
      "loss": 13.9125,
      "step": 2801
    },
    {
      "epoch": 0.4899672131147541,
      "grad_norm": 0.9610296942498886,
      "learning_rate": 0.00019554833190524182,
      "loss": 13.7967,
      "step": 2802
    },
    {
      "epoch": 0.4901420765027322,
      "grad_norm": 0.9994702483191884,
      "learning_rate": 0.000195544152846702,
      "loss": 13.5713,
      "step": 2803
    },
    {
      "epoch": 0.4903169398907104,
      "grad_norm": 1.0683235691962534,
      "learning_rate": 0.00019553997187220855,
      "loss": 13.6572,
      "step": 2804
    },
    {
      "epoch": 0.4904918032786885,
      "grad_norm": 0.9999520473616972,
      "learning_rate": 0.00019553578898184531,
      "loss": 13.6035,
      "step": 2805
    },
    {
      "epoch": 0.49066666666666664,
      "grad_norm": 0.9340242716479895,
      "learning_rate": 0.0001955316041756962,
      "loss": 13.3731,
      "step": 2806
    },
    {
      "epoch": 0.4908415300546448,
      "grad_norm": 0.9926432179063194,
      "learning_rate": 0.00019552741745384507,
      "loss": 13.5444,
      "step": 2807
    },
    {
      "epoch": 0.49101639344262293,
      "grad_norm": 0.921856385306385,
      "learning_rate": 0.0001955232288163759,
      "loss": 13.9049,
      "step": 2808
    },
    {
      "epoch": 0.4911912568306011,
      "grad_norm": 0.8962939060191968,
      "learning_rate": 0.00019551903826337273,
      "loss": 13.611,
      "step": 2809
    },
    {
      "epoch": 0.4913661202185792,
      "grad_norm": 1.0431760734532087,
      "learning_rate": 0.0001955148457949196,
      "loss": 13.7954,
      "step": 2810
    },
    {
      "epoch": 0.4915409836065574,
      "grad_norm": 0.9623330677913781,
      "learning_rate": 0.00019551065141110047,
      "loss": 13.673,
      "step": 2811
    },
    {
      "epoch": 0.4917158469945355,
      "grad_norm": 0.9769885578634874,
      "learning_rate": 0.00019550645511199957,
      "loss": 13.6283,
      "step": 2812
    },
    {
      "epoch": 0.49189071038251364,
      "grad_norm": 1.120475308175384,
      "learning_rate": 0.00019550225689770096,
      "loss": 13.5755,
      "step": 2813
    },
    {
      "epoch": 0.4920655737704918,
      "grad_norm": 0.9141136879836506,
      "learning_rate": 0.00019549805676828889,
      "loss": 13.5157,
      "step": 2814
    },
    {
      "epoch": 0.49224043715846993,
      "grad_norm": 1.0063853303427304,
      "learning_rate": 0.0001954938547238476,
      "loss": 13.7154,
      "step": 2815
    },
    {
      "epoch": 0.4924153005464481,
      "grad_norm": 1.0998297145680918,
      "learning_rate": 0.00019548965076446125,
      "loss": 13.7466,
      "step": 2816
    },
    {
      "epoch": 0.49259016393442623,
      "grad_norm": 0.9673848302559848,
      "learning_rate": 0.00019548544489021423,
      "loss": 13.6858,
      "step": 2817
    },
    {
      "epoch": 0.49276502732240435,
      "grad_norm": 1.0184800940281475,
      "learning_rate": 0.00019548123710119087,
      "loss": 13.676,
      "step": 2818
    },
    {
      "epoch": 0.4929398907103825,
      "grad_norm": 1.2814819088004707,
      "learning_rate": 0.00019547702739747556,
      "loss": 13.4413,
      "step": 2819
    },
    {
      "epoch": 0.49311475409836064,
      "grad_norm": 1.0472206894751455,
      "learning_rate": 0.00019547281577915267,
      "loss": 13.7678,
      "step": 2820
    },
    {
      "epoch": 0.4932896174863388,
      "grad_norm": 1.0618062697002077,
      "learning_rate": 0.00019546860224630672,
      "loss": 13.3155,
      "step": 2821
    },
    {
      "epoch": 0.49346448087431694,
      "grad_norm": 1.101667493533273,
      "learning_rate": 0.00019546438679902216,
      "loss": 13.7958,
      "step": 2822
    },
    {
      "epoch": 0.49363934426229505,
      "grad_norm": 0.982803677887882,
      "learning_rate": 0.00019546016943738348,
      "loss": 13.7166,
      "step": 2823
    },
    {
      "epoch": 0.49381420765027323,
      "grad_norm": 1.0907539931736854,
      "learning_rate": 0.00019545595016147536,
      "loss": 13.8263,
      "step": 2824
    },
    {
      "epoch": 0.49398907103825135,
      "grad_norm": 1.1474937101082838,
      "learning_rate": 0.0001954517289713823,
      "loss": 13.3652,
      "step": 2825
    },
    {
      "epoch": 0.4941639344262295,
      "grad_norm": 0.9959224811855497,
      "learning_rate": 0.00019544750586718903,
      "loss": 13.6761,
      "step": 2826
    },
    {
      "epoch": 0.49433879781420764,
      "grad_norm": 1.093119679315707,
      "learning_rate": 0.0001954432808489802,
      "loss": 13.5311,
      "step": 2827
    },
    {
      "epoch": 0.4945136612021858,
      "grad_norm": 1.094100776339299,
      "learning_rate": 0.00019543905391684055,
      "loss": 13.5932,
      "step": 2828
    },
    {
      "epoch": 0.49468852459016394,
      "grad_norm": 1.0201265027274289,
      "learning_rate": 0.00019543482507085482,
      "loss": 13.6799,
      "step": 2829
    },
    {
      "epoch": 0.49486338797814206,
      "grad_norm": 0.9598137921003934,
      "learning_rate": 0.0001954305943111078,
      "loss": 13.5645,
      "step": 2830
    },
    {
      "epoch": 0.49503825136612023,
      "grad_norm": 0.9202046847053671,
      "learning_rate": 0.0001954263616376844,
      "loss": 13.5328,
      "step": 2831
    },
    {
      "epoch": 0.49521311475409835,
      "grad_norm": 0.9268589251023858,
      "learning_rate": 0.00019542212705066944,
      "loss": 13.8486,
      "step": 2832
    },
    {
      "epoch": 0.4953879781420765,
      "grad_norm": 1.0332631762617273,
      "learning_rate": 0.00019541789055014784,
      "loss": 13.4394,
      "step": 2833
    },
    {
      "epoch": 0.49556284153005464,
      "grad_norm": 0.9733566132209243,
      "learning_rate": 0.00019541365213620457,
      "loss": 13.7761,
      "step": 2834
    },
    {
      "epoch": 0.49573770491803276,
      "grad_norm": 1.1341250013740818,
      "learning_rate": 0.0001954094118089246,
      "loss": 13.3585,
      "step": 2835
    },
    {
      "epoch": 0.49591256830601094,
      "grad_norm": 1.2793917971576667,
      "learning_rate": 0.000195405169568393,
      "loss": 13.4722,
      "step": 2836
    },
    {
      "epoch": 0.49608743169398906,
      "grad_norm": 1.0560238732130471,
      "learning_rate": 0.00019540092541469478,
      "loss": 13.8426,
      "step": 2837
    },
    {
      "epoch": 0.49626229508196723,
      "grad_norm": 1.099947401268941,
      "learning_rate": 0.00019539667934791513,
      "loss": 13.6166,
      "step": 2838
    },
    {
      "epoch": 0.49643715846994535,
      "grad_norm": 1.119995556283767,
      "learning_rate": 0.00019539243136813914,
      "loss": 13.6012,
      "step": 2839
    },
    {
      "epoch": 0.4966120218579235,
      "grad_norm": 0.9883028412682773,
      "learning_rate": 0.000195388181475452,
      "loss": 13.4669,
      "step": 2840
    },
    {
      "epoch": 0.49678688524590164,
      "grad_norm": 1.0832615967686265,
      "learning_rate": 0.00019538392966993894,
      "loss": 13.8768,
      "step": 2841
    },
    {
      "epoch": 0.49696174863387976,
      "grad_norm": 1.0866854388936558,
      "learning_rate": 0.00019537967595168522,
      "loss": 13.77,
      "step": 2842
    },
    {
      "epoch": 0.49713661202185794,
      "grad_norm": 1.127739432616613,
      "learning_rate": 0.00019537542032077614,
      "loss": 13.8826,
      "step": 2843
    },
    {
      "epoch": 0.49731147540983606,
      "grad_norm": 0.9651344544750868,
      "learning_rate": 0.00019537116277729705,
      "loss": 13.4843,
      "step": 2844
    },
    {
      "epoch": 0.49748633879781423,
      "grad_norm": 1.0375087194547985,
      "learning_rate": 0.00019536690332133333,
      "loss": 13.6002,
      "step": 2845
    },
    {
      "epoch": 0.49766120218579235,
      "grad_norm": 0.8927417101597307,
      "learning_rate": 0.0001953626419529704,
      "loss": 13.6036,
      "step": 2846
    },
    {
      "epoch": 0.49783606557377047,
      "grad_norm": 0.9395658176084891,
      "learning_rate": 0.00019535837867229363,
      "loss": 13.4893,
      "step": 2847
    },
    {
      "epoch": 0.49801092896174864,
      "grad_norm": 1.0115623501541902,
      "learning_rate": 0.00019535411347938863,
      "loss": 13.7436,
      "step": 2848
    },
    {
      "epoch": 0.49818579234972676,
      "grad_norm": 1.0262545030715977,
      "learning_rate": 0.00019534984637434088,
      "loss": 13.9304,
      "step": 2849
    },
    {
      "epoch": 0.49836065573770494,
      "grad_norm": 1.006193217305135,
      "learning_rate": 0.00019534557735723592,
      "loss": 13.8449,
      "step": 2850
    },
    {
      "epoch": 0.49853551912568306,
      "grad_norm": 0.9776017719998007,
      "learning_rate": 0.00019534130642815943,
      "loss": 13.7921,
      "step": 2851
    },
    {
      "epoch": 0.4987103825136612,
      "grad_norm": 0.9793931538612354,
      "learning_rate": 0.00019533703358719695,
      "loss": 13.7439,
      "step": 2852
    },
    {
      "epoch": 0.49888524590163935,
      "grad_norm": 0.9853461294739757,
      "learning_rate": 0.00019533275883443425,
      "loss": 13.7132,
      "step": 2853
    },
    {
      "epoch": 0.49906010928961747,
      "grad_norm": 0.9314983253229577,
      "learning_rate": 0.00019532848216995702,
      "loss": 13.6646,
      "step": 2854
    },
    {
      "epoch": 0.49923497267759565,
      "grad_norm": 0.9305022271973571,
      "learning_rate": 0.00019532420359385104,
      "loss": 13.481,
      "step": 2855
    },
    {
      "epoch": 0.49940983606557376,
      "grad_norm": 0.9617230399465803,
      "learning_rate": 0.0001953199231062021,
      "loss": 13.4687,
      "step": 2856
    },
    {
      "epoch": 0.49958469945355194,
      "grad_norm": 1.0237625710469123,
      "learning_rate": 0.00019531564070709602,
      "loss": 13.5051,
      "step": 2857
    },
    {
      "epoch": 0.49975956284153006,
      "grad_norm": 0.9788802041478847,
      "learning_rate": 0.0001953113563966187,
      "loss": 13.6848,
      "step": 2858
    },
    {
      "epoch": 0.4999344262295082,
      "grad_norm": 1.0745865107073258,
      "learning_rate": 0.000195307070174856,
      "loss": 13.7783,
      "step": 2859
    },
    {
      "epoch": 0.5001092896174864,
      "grad_norm": 1.0923567826812028,
      "learning_rate": 0.00019530278204189395,
      "loss": 13.58,
      "step": 2860
    },
    {
      "epoch": 0.5002841530054645,
      "grad_norm": 0.9982116568641932,
      "learning_rate": 0.00019529849199781852,
      "loss": 13.6141,
      "step": 2861
    },
    {
      "epoch": 0.5004590163934426,
      "grad_norm": 1.0366307899345448,
      "learning_rate": 0.00019529420004271567,
      "loss": 13.9208,
      "step": 2862
    },
    {
      "epoch": 0.5006338797814207,
      "grad_norm": 0.9684380031035101,
      "learning_rate": 0.00019528990617667154,
      "loss": 13.6801,
      "step": 2863
    },
    {
      "epoch": 0.5008087431693989,
      "grad_norm": 1.108341014065766,
      "learning_rate": 0.0001952856103997722,
      "loss": 13.7209,
      "step": 2864
    },
    {
      "epoch": 0.5009836065573771,
      "grad_norm": 1.0187490558694987,
      "learning_rate": 0.00019528131271210383,
      "loss": 13.7078,
      "step": 2865
    },
    {
      "epoch": 0.5011584699453552,
      "grad_norm": 1.0294794875249458,
      "learning_rate": 0.00019527701311375256,
      "loss": 13.8591,
      "step": 2866
    },
    {
      "epoch": 0.5013333333333333,
      "grad_norm": 0.9892042602942029,
      "learning_rate": 0.00019527271160480467,
      "loss": 13.6454,
      "step": 2867
    },
    {
      "epoch": 0.5015081967213115,
      "grad_norm": 1.073628491576119,
      "learning_rate": 0.00019526840818534638,
      "loss": 13.8168,
      "step": 2868
    },
    {
      "epoch": 0.5016830601092896,
      "grad_norm": 1.0004883794693855,
      "learning_rate": 0.00019526410285546398,
      "loss": 13.7328,
      "step": 2869
    },
    {
      "epoch": 0.5018579234972678,
      "grad_norm": 0.9989889722054103,
      "learning_rate": 0.0001952597956152438,
      "loss": 13.7332,
      "step": 2870
    },
    {
      "epoch": 0.5020327868852459,
      "grad_norm": 1.0272336551097891,
      "learning_rate": 0.00019525548646477225,
      "loss": 13.6431,
      "step": 2871
    },
    {
      "epoch": 0.502207650273224,
      "grad_norm": 0.959121532509824,
      "learning_rate": 0.00019525117540413573,
      "loss": 13.6464,
      "step": 2872
    },
    {
      "epoch": 0.5023825136612022,
      "grad_norm": 1.1281966180707945,
      "learning_rate": 0.0001952468624334207,
      "loss": 13.667,
      "step": 2873
    },
    {
      "epoch": 0.5025573770491804,
      "grad_norm": 0.9751407882734925,
      "learning_rate": 0.0001952425475527136,
      "loss": 13.3744,
      "step": 2874
    },
    {
      "epoch": 0.5027322404371585,
      "grad_norm": 1.0658413953449255,
      "learning_rate": 0.00019523823076210103,
      "loss": 13.5401,
      "step": 2875
    },
    {
      "epoch": 0.5029071038251366,
      "grad_norm": 1.1073748897550597,
      "learning_rate": 0.00019523391206166947,
      "loss": 13.6665,
      "step": 2876
    },
    {
      "epoch": 0.5030819672131147,
      "grad_norm": 1.004134188037413,
      "learning_rate": 0.00019522959145150558,
      "loss": 13.7312,
      "step": 2877
    },
    {
      "epoch": 0.5032568306010929,
      "grad_norm": 1.0872704966195879,
      "learning_rate": 0.00019522526893169597,
      "loss": 13.76,
      "step": 2878
    },
    {
      "epoch": 0.5034316939890711,
      "grad_norm": 0.9720876842872359,
      "learning_rate": 0.0001952209445023274,
      "loss": 13.7877,
      "step": 2879
    },
    {
      "epoch": 0.5036065573770492,
      "grad_norm": 0.932436911110571,
      "learning_rate": 0.00019521661816348647,
      "loss": 13.3555,
      "step": 2880
    },
    {
      "epoch": 0.5037814207650273,
      "grad_norm": 1.098915436043757,
      "learning_rate": 0.00019521228991525999,
      "loss": 13.4408,
      "step": 2881
    },
    {
      "epoch": 0.5039562841530054,
      "grad_norm": 1.0814917312132295,
      "learning_rate": 0.00019520795975773477,
      "loss": 13.7672,
      "step": 2882
    },
    {
      "epoch": 0.5041311475409836,
      "grad_norm": 0.9985216510935128,
      "learning_rate": 0.00019520362769099764,
      "loss": 13.6643,
      "step": 2883
    },
    {
      "epoch": 0.5043060109289618,
      "grad_norm": 0.9922081481094325,
      "learning_rate": 0.00019519929371513545,
      "loss": 13.6191,
      "step": 2884
    },
    {
      "epoch": 0.5044808743169399,
      "grad_norm": 1.0898686518112355,
      "learning_rate": 0.00019519495783023515,
      "loss": 13.7407,
      "step": 2885
    },
    {
      "epoch": 0.504655737704918,
      "grad_norm": 0.9461689312142367,
      "learning_rate": 0.00019519062003638363,
      "loss": 13.6112,
      "step": 2886
    },
    {
      "epoch": 0.5048306010928961,
      "grad_norm": 0.9908690873825207,
      "learning_rate": 0.00019518628033366787,
      "loss": 13.5717,
      "step": 2887
    },
    {
      "epoch": 0.5050054644808744,
      "grad_norm": 0.9988626556012546,
      "learning_rate": 0.00019518193872217497,
      "loss": 13.5023,
      "step": 2888
    },
    {
      "epoch": 0.5051803278688525,
      "grad_norm": 1.0771993602646042,
      "learning_rate": 0.00019517759520199196,
      "loss": 13.3956,
      "step": 2889
    },
    {
      "epoch": 0.5053551912568306,
      "grad_norm": 1.013175230315544,
      "learning_rate": 0.00019517324977320595,
      "loss": 13.8256,
      "step": 2890
    },
    {
      "epoch": 0.5055300546448087,
      "grad_norm": 1.053640888395228,
      "learning_rate": 0.000195168902435904,
      "loss": 13.4845,
      "step": 2891
    },
    {
      "epoch": 0.5057049180327868,
      "grad_norm": 1.0625608913208857,
      "learning_rate": 0.0001951645531901734,
      "loss": 13.7666,
      "step": 2892
    },
    {
      "epoch": 0.5058797814207651,
      "grad_norm": 1.0243636924871085,
      "learning_rate": 0.0001951602020361013,
      "loss": 13.644,
      "step": 2893
    },
    {
      "epoch": 0.5060546448087432,
      "grad_norm": 1.1262788850242975,
      "learning_rate": 0.00019515584897377496,
      "loss": 13.5686,
      "step": 2894
    },
    {
      "epoch": 0.5062295081967213,
      "grad_norm": 0.9878203497566111,
      "learning_rate": 0.0001951514940032817,
      "loss": 13.6912,
      "step": 2895
    },
    {
      "epoch": 0.5064043715846994,
      "grad_norm": 1.0593087720950045,
      "learning_rate": 0.0001951471371247088,
      "loss": 13.4638,
      "step": 2896
    },
    {
      "epoch": 0.5065792349726777,
      "grad_norm": 1.015843034801462,
      "learning_rate": 0.0001951427783381437,
      "loss": 13.7507,
      "step": 2897
    },
    {
      "epoch": 0.5067540983606558,
      "grad_norm": 1.1078023240303687,
      "learning_rate": 0.00019513841764367374,
      "loss": 13.5352,
      "step": 2898
    },
    {
      "epoch": 0.5069289617486339,
      "grad_norm": 1.0636839185317222,
      "learning_rate": 0.0001951340550413864,
      "loss": 13.507,
      "step": 2899
    },
    {
      "epoch": 0.507103825136612,
      "grad_norm": 1.0277026599540113,
      "learning_rate": 0.00019512969053136917,
      "loss": 13.2742,
      "step": 2900
    },
    {
      "epoch": 0.5072786885245901,
      "grad_norm": 1.029213581422598,
      "learning_rate": 0.00019512532411370954,
      "loss": 13.3933,
      "step": 2901
    },
    {
      "epoch": 0.5074535519125684,
      "grad_norm": 1.119401350051224,
      "learning_rate": 0.0001951209557884951,
      "loss": 13.629,
      "step": 2902
    },
    {
      "epoch": 0.5076284153005465,
      "grad_norm": 1.1349685719958207,
      "learning_rate": 0.00019511658555581343,
      "loss": 13.6605,
      "step": 2903
    },
    {
      "epoch": 0.5078032786885246,
      "grad_norm": 1.1381058343030108,
      "learning_rate": 0.0001951122134157522,
      "loss": 13.6082,
      "step": 2904
    },
    {
      "epoch": 0.5079781420765027,
      "grad_norm": 1.1440555607705603,
      "learning_rate": 0.000195107839368399,
      "loss": 13.8162,
      "step": 2905
    },
    {
      "epoch": 0.5081530054644808,
      "grad_norm": 1.2158775842908993,
      "learning_rate": 0.00019510346341384163,
      "loss": 13.699,
      "step": 2906
    },
    {
      "epoch": 0.5083278688524591,
      "grad_norm": 0.9764371518616559,
      "learning_rate": 0.00019509908555216782,
      "loss": 13.229,
      "step": 2907
    },
    {
      "epoch": 0.5085027322404372,
      "grad_norm": 0.903032231693118,
      "learning_rate": 0.00019509470578346535,
      "loss": 13.5763,
      "step": 2908
    },
    {
      "epoch": 0.5086775956284153,
      "grad_norm": 1.070024632071431,
      "learning_rate": 0.00019509032410782205,
      "loss": 13.4245,
      "step": 2909
    },
    {
      "epoch": 0.5088524590163934,
      "grad_norm": 1.0214148245443746,
      "learning_rate": 0.0001950859405253258,
      "loss": 13.7427,
      "step": 2910
    },
    {
      "epoch": 0.5090273224043715,
      "grad_norm": 0.9526222296765148,
      "learning_rate": 0.00019508155503606447,
      "loss": 13.6937,
      "step": 2911
    },
    {
      "epoch": 0.5092021857923498,
      "grad_norm": 0.9740636205895397,
      "learning_rate": 0.00019507716764012603,
      "loss": 13.6836,
      "step": 2912
    },
    {
      "epoch": 0.5093770491803279,
      "grad_norm": 0.9639461915170574,
      "learning_rate": 0.00019507277833759844,
      "loss": 13.5913,
      "step": 2913
    },
    {
      "epoch": 0.509551912568306,
      "grad_norm": 1.0714543800613354,
      "learning_rate": 0.00019506838712856977,
      "loss": 13.4495,
      "step": 2914
    },
    {
      "epoch": 0.5097267759562841,
      "grad_norm": 1.002681906087875,
      "learning_rate": 0.00019506399401312798,
      "loss": 13.6132,
      "step": 2915
    },
    {
      "epoch": 0.5099016393442622,
      "grad_norm": 1.1053915063014559,
      "learning_rate": 0.0001950595989913613,
      "loss": 13.678,
      "step": 2916
    },
    {
      "epoch": 0.5100765027322405,
      "grad_norm": 0.9523854537017795,
      "learning_rate": 0.0001950552020633577,
      "loss": 13.677,
      "step": 2917
    },
    {
      "epoch": 0.5102513661202186,
      "grad_norm": 0.9293060819419329,
      "learning_rate": 0.0001950508032292055,
      "loss": 13.6551,
      "step": 2918
    },
    {
      "epoch": 0.5104262295081967,
      "grad_norm": 0.9514721096651271,
      "learning_rate": 0.00019504640248899286,
      "loss": 13.5476,
      "step": 2919
    },
    {
      "epoch": 0.5106010928961748,
      "grad_norm": 0.9669424942421951,
      "learning_rate": 0.00019504199984280799,
      "loss": 13.8268,
      "step": 2920
    },
    {
      "epoch": 0.510775956284153,
      "grad_norm": 0.9486681397417882,
      "learning_rate": 0.0001950375952907392,
      "loss": 13.5247,
      "step": 2921
    },
    {
      "epoch": 0.5109508196721312,
      "grad_norm": 1.0235409983522883,
      "learning_rate": 0.00019503318883287485,
      "loss": 13.8011,
      "step": 2922
    },
    {
      "epoch": 0.5111256830601093,
      "grad_norm": 0.967362403159219,
      "learning_rate": 0.00019502878046930326,
      "loss": 13.3595,
      "step": 2923
    },
    {
      "epoch": 0.5113005464480874,
      "grad_norm": 0.9988876768349352,
      "learning_rate": 0.00019502437020011283,
      "loss": 13.5574,
      "step": 2924
    },
    {
      "epoch": 0.5114754098360655,
      "grad_norm": 1.0301892796410608,
      "learning_rate": 0.00019501995802539206,
      "loss": 13.7533,
      "step": 2925
    },
    {
      "epoch": 0.5116502732240437,
      "grad_norm": 1.2650998906613986,
      "learning_rate": 0.00019501554394522934,
      "loss": 13.4181,
      "step": 2926
    },
    {
      "epoch": 0.5118251366120219,
      "grad_norm": 1.0490176918237737,
      "learning_rate": 0.00019501112795971323,
      "loss": 13.5352,
      "step": 2927
    },
    {
      "epoch": 0.512,
      "grad_norm": 1.064748030413938,
      "learning_rate": 0.0001950067100689323,
      "loss": 13.7271,
      "step": 2928
    },
    {
      "epoch": 0.5121748633879781,
      "grad_norm": 1.2667886547258584,
      "learning_rate": 0.00019500229027297512,
      "loss": 13.6401,
      "step": 2929
    },
    {
      "epoch": 0.5123497267759562,
      "grad_norm": 1.0609693979257764,
      "learning_rate": 0.00019499786857193028,
      "loss": 13.6713,
      "step": 2930
    },
    {
      "epoch": 0.5125245901639345,
      "grad_norm": 1.1260994861704958,
      "learning_rate": 0.00019499344496588656,
      "loss": 13.5591,
      "step": 2931
    },
    {
      "epoch": 0.5126994535519126,
      "grad_norm": 1.0703698471586334,
      "learning_rate": 0.00019498901945493258,
      "loss": 13.7113,
      "step": 2932
    },
    {
      "epoch": 0.5128743169398907,
      "grad_norm": 1.020696814017387,
      "learning_rate": 0.00019498459203915708,
      "loss": 13.5268,
      "step": 2933
    },
    {
      "epoch": 0.5130491803278688,
      "grad_norm": 0.9617058300165185,
      "learning_rate": 0.0001949801627186489,
      "loss": 13.6009,
      "step": 2934
    },
    {
      "epoch": 0.513224043715847,
      "grad_norm": 0.9684345977366567,
      "learning_rate": 0.0001949757314934968,
      "loss": 13.6445,
      "step": 2935
    },
    {
      "epoch": 0.5133989071038252,
      "grad_norm": 0.8861052611041361,
      "learning_rate": 0.0001949712983637897,
      "loss": 13.5627,
      "step": 2936
    },
    {
      "epoch": 0.5135737704918033,
      "grad_norm": 1.0535758067486167,
      "learning_rate": 0.00019496686332961646,
      "loss": 13.723,
      "step": 2937
    },
    {
      "epoch": 0.5137486338797814,
      "grad_norm": 1.1014643132373467,
      "learning_rate": 0.000194962426391066,
      "loss": 13.3278,
      "step": 2938
    },
    {
      "epoch": 0.5139234972677595,
      "grad_norm": 1.0546911502655996,
      "learning_rate": 0.00019495798754822733,
      "loss": 13.5538,
      "step": 2939
    },
    {
      "epoch": 0.5140983606557377,
      "grad_norm": 1.1502820969959655,
      "learning_rate": 0.00019495354680118947,
      "loss": 13.6343,
      "step": 2940
    },
    {
      "epoch": 0.5142732240437159,
      "grad_norm": 0.9382422321050949,
      "learning_rate": 0.00019494910415004144,
      "loss": 13.7127,
      "step": 2941
    },
    {
      "epoch": 0.514448087431694,
      "grad_norm": 1.1504890291035341,
      "learning_rate": 0.00019494465959487233,
      "loss": 13.6645,
      "step": 2942
    },
    {
      "epoch": 0.5146229508196721,
      "grad_norm": 1.1896685818754817,
      "learning_rate": 0.0001949402131357713,
      "loss": 13.8492,
      "step": 2943
    },
    {
      "epoch": 0.5147978142076503,
      "grad_norm": 1.024264169590649,
      "learning_rate": 0.00019493576477282748,
      "loss": 13.7007,
      "step": 2944
    },
    {
      "epoch": 0.5149726775956284,
      "grad_norm": 1.2418414833386526,
      "learning_rate": 0.00019493131450613006,
      "loss": 13.4311,
      "step": 2945
    },
    {
      "epoch": 0.5151475409836066,
      "grad_norm": 0.9229285168676195,
      "learning_rate": 0.00019492686233576833,
      "loss": 13.6094,
      "step": 2946
    },
    {
      "epoch": 0.5153224043715847,
      "grad_norm": 1.2574877896233718,
      "learning_rate": 0.00019492240826183156,
      "loss": 13.7844,
      "step": 2947
    },
    {
      "epoch": 0.5154972677595628,
      "grad_norm": 1.0255145197577828,
      "learning_rate": 0.00019491795228440903,
      "loss": 13.6173,
      "step": 2948
    },
    {
      "epoch": 0.515672131147541,
      "grad_norm": 1.1180770029955462,
      "learning_rate": 0.00019491349440359015,
      "loss": 13.7828,
      "step": 2949
    },
    {
      "epoch": 0.5158469945355191,
      "grad_norm": 1.1350022606958559,
      "learning_rate": 0.00019490903461946422,
      "loss": 13.6086,
      "step": 2950
    },
    {
      "epoch": 0.5160218579234973,
      "grad_norm": 0.9239148035151327,
      "learning_rate": 0.0001949045729321208,
      "loss": 13.6084,
      "step": 2951
    },
    {
      "epoch": 0.5161967213114754,
      "grad_norm": 1.0843533941839316,
      "learning_rate": 0.00019490010934164928,
      "loss": 13.752,
      "step": 2952
    },
    {
      "epoch": 0.5163715846994535,
      "grad_norm": 0.9779323543021564,
      "learning_rate": 0.0001948956438481392,
      "loss": 13.5174,
      "step": 2953
    },
    {
      "epoch": 0.5165464480874317,
      "grad_norm": 1.1087536148080934,
      "learning_rate": 0.00019489117645168005,
      "loss": 13.5284,
      "step": 2954
    },
    {
      "epoch": 0.5167213114754098,
      "grad_norm": 0.9340028526230504,
      "learning_rate": 0.0001948867071523615,
      "loss": 13.5341,
      "step": 2955
    },
    {
      "epoch": 0.516896174863388,
      "grad_norm": 0.9775130727191992,
      "learning_rate": 0.0001948822359502731,
      "loss": 13.6239,
      "step": 2956
    },
    {
      "epoch": 0.5170710382513661,
      "grad_norm": 0.9875639326992199,
      "learning_rate": 0.00019487776284550458,
      "loss": 13.5972,
      "step": 2957
    },
    {
      "epoch": 0.5172459016393443,
      "grad_norm": 0.9473030724693283,
      "learning_rate": 0.00019487328783814557,
      "loss": 13.5709,
      "step": 2958
    },
    {
      "epoch": 0.5174207650273224,
      "grad_norm": 1.0042030281215866,
      "learning_rate": 0.00019486881092828587,
      "loss": 13.5047,
      "step": 2959
    },
    {
      "epoch": 0.5175956284153006,
      "grad_norm": 0.944020784056162,
      "learning_rate": 0.00019486433211601521,
      "loss": 13.6246,
      "step": 2960
    },
    {
      "epoch": 0.5177704918032787,
      "grad_norm": 0.8635211354006453,
      "learning_rate": 0.0001948598514014234,
      "loss": 13.5895,
      "step": 2961
    },
    {
      "epoch": 0.5179453551912568,
      "grad_norm": 0.9722453899419032,
      "learning_rate": 0.00019485536878460031,
      "loss": 13.6597,
      "step": 2962
    },
    {
      "epoch": 0.518120218579235,
      "grad_norm": 0.9458947801195775,
      "learning_rate": 0.00019485088426563586,
      "loss": 13.5878,
      "step": 2963
    },
    {
      "epoch": 0.5182950819672131,
      "grad_norm": 1.114349005435432,
      "learning_rate": 0.00019484639784461994,
      "loss": 13.4376,
      "step": 2964
    },
    {
      "epoch": 0.5184699453551913,
      "grad_norm": 1.143165948856259,
      "learning_rate": 0.00019484190952164253,
      "loss": 13.6856,
      "step": 2965
    },
    {
      "epoch": 0.5186448087431694,
      "grad_norm": 0.9745350884612844,
      "learning_rate": 0.00019483741929679363,
      "loss": 13.8349,
      "step": 2966
    },
    {
      "epoch": 0.5188196721311475,
      "grad_norm": 0.9679264689788613,
      "learning_rate": 0.00019483292717016327,
      "loss": 13.6412,
      "step": 2967
    },
    {
      "epoch": 0.5189945355191257,
      "grad_norm": 1.0267374165799446,
      "learning_rate": 0.00019482843314184154,
      "loss": 13.818,
      "step": 2968
    },
    {
      "epoch": 0.5191693989071038,
      "grad_norm": 0.9964767831721149,
      "learning_rate": 0.0001948239372119186,
      "loss": 13.4969,
      "step": 2969
    },
    {
      "epoch": 0.519344262295082,
      "grad_norm": 0.9748847227032275,
      "learning_rate": 0.00019481943938048452,
      "loss": 13.4632,
      "step": 2970
    },
    {
      "epoch": 0.5195191256830601,
      "grad_norm": 0.9259240952586736,
      "learning_rate": 0.00019481493964762958,
      "loss": 13.5738,
      "step": 2971
    },
    {
      "epoch": 0.5196939890710383,
      "grad_norm": 0.9589700534158534,
      "learning_rate": 0.000194810438013444,
      "loss": 13.299,
      "step": 2972
    },
    {
      "epoch": 0.5198688524590164,
      "grad_norm": 1.010628243704928,
      "learning_rate": 0.00019480593447801799,
      "loss": 13.4315,
      "step": 2973
    },
    {
      "epoch": 0.5200437158469945,
      "grad_norm": 0.9439327652767526,
      "learning_rate": 0.00019480142904144192,
      "loss": 13.7155,
      "step": 2974
    },
    {
      "epoch": 0.5202185792349727,
      "grad_norm": 0.926665627655319,
      "learning_rate": 0.0001947969217038061,
      "loss": 13.6833,
      "step": 2975
    },
    {
      "epoch": 0.5203934426229508,
      "grad_norm": 0.9262975824884631,
      "learning_rate": 0.00019479241246520097,
      "loss": 13.5552,
      "step": 2976
    },
    {
      "epoch": 0.520568306010929,
      "grad_norm": 1.021567428733198,
      "learning_rate": 0.0001947879013257169,
      "loss": 13.6483,
      "step": 2977
    },
    {
      "epoch": 0.5207431693989071,
      "grad_norm": 0.9190931488410761,
      "learning_rate": 0.00019478338828544435,
      "loss": 13.3555,
      "step": 2978
    },
    {
      "epoch": 0.5209180327868852,
      "grad_norm": 0.9237742084403218,
      "learning_rate": 0.00019477887334447389,
      "loss": 13.4273,
      "step": 2979
    },
    {
      "epoch": 0.5210928961748634,
      "grad_norm": 0.9469167056156241,
      "learning_rate": 0.00019477435650289596,
      "loss": 13.623,
      "step": 2980
    },
    {
      "epoch": 0.5212677595628415,
      "grad_norm": 0.9945250267421993,
      "learning_rate": 0.00019476983776080123,
      "loss": 13.7546,
      "step": 2981
    },
    {
      "epoch": 0.5214426229508197,
      "grad_norm": 0.9427124062036015,
      "learning_rate": 0.00019476531711828027,
      "loss": 13.21,
      "step": 2982
    },
    {
      "epoch": 0.5216174863387978,
      "grad_norm": 1.0460146701820578,
      "learning_rate": 0.0001947607945754237,
      "loss": 13.4254,
      "step": 2983
    },
    {
      "epoch": 0.5217923497267759,
      "grad_norm": 0.901699494054996,
      "learning_rate": 0.00019475627013232228,
      "loss": 13.0612,
      "step": 2984
    },
    {
      "epoch": 0.5219672131147541,
      "grad_norm": 0.9318123198961787,
      "learning_rate": 0.00019475174378906667,
      "loss": 13.6192,
      "step": 2985
    },
    {
      "epoch": 0.5221420765027323,
      "grad_norm": 1.0104847740899432,
      "learning_rate": 0.0001947472155457477,
      "loss": 13.4805,
      "step": 2986
    },
    {
      "epoch": 0.5223169398907104,
      "grad_norm": 0.9536549445467554,
      "learning_rate": 0.0001947426854024561,
      "loss": 13.4305,
      "step": 2987
    },
    {
      "epoch": 0.5224918032786885,
      "grad_norm": 0.969089552748059,
      "learning_rate": 0.0001947381533592828,
      "loss": 13.5487,
      "step": 2988
    },
    {
      "epoch": 0.5226666666666666,
      "grad_norm": 0.940598402104475,
      "learning_rate": 0.00019473361941631866,
      "loss": 13.4179,
      "step": 2989
    },
    {
      "epoch": 0.5228415300546448,
      "grad_norm": 0.9770685112677214,
      "learning_rate": 0.00019472908357365455,
      "loss": 13.6319,
      "step": 2990
    },
    {
      "epoch": 0.523016393442623,
      "grad_norm": 1.1997404517094918,
      "learning_rate": 0.00019472454583138144,
      "loss": 13.7838,
      "step": 2991
    },
    {
      "epoch": 0.5231912568306011,
      "grad_norm": 1.0369980348104295,
      "learning_rate": 0.00019472000618959038,
      "loss": 13.7013,
      "step": 2992
    },
    {
      "epoch": 0.5233661202185792,
      "grad_norm": 1.0837916680469388,
      "learning_rate": 0.00019471546464837233,
      "loss": 13.6473,
      "step": 2993
    },
    {
      "epoch": 0.5235409836065574,
      "grad_norm": 1.0569737534581791,
      "learning_rate": 0.00019471092120781843,
      "loss": 13.6226,
      "step": 2994
    },
    {
      "epoch": 0.5237158469945355,
      "grad_norm": 1.0875872152095936,
      "learning_rate": 0.00019470637586801973,
      "loss": 13.2983,
      "step": 2995
    },
    {
      "epoch": 0.5238907103825137,
      "grad_norm": 1.0708994985522358,
      "learning_rate": 0.00019470182862906744,
      "loss": 13.4304,
      "step": 2996
    },
    {
      "epoch": 0.5240655737704918,
      "grad_norm": 1.0457127757546056,
      "learning_rate": 0.0001946972794910527,
      "loss": 13.7335,
      "step": 2997
    },
    {
      "epoch": 0.5242404371584699,
      "grad_norm": 1.0353473296401687,
      "learning_rate": 0.00019469272845406676,
      "loss": 13.6345,
      "step": 2998
    },
    {
      "epoch": 0.5244153005464481,
      "grad_norm": 0.9058416411460175,
      "learning_rate": 0.00019468817551820085,
      "loss": 13.6347,
      "step": 2999
    },
    {
      "epoch": 0.5245901639344263,
      "grad_norm": 1.0217767581537038,
      "learning_rate": 0.0001946836206835463,
      "loss": 13.5373,
      "step": 3000
    },
    {
      "epoch": 0.5247650273224044,
      "grad_norm": 0.9581124048119303,
      "learning_rate": 0.00019467906395019443,
      "loss": 13.7891,
      "step": 3001
    },
    {
      "epoch": 0.5249398907103825,
      "grad_norm": 0.923511161484937,
      "learning_rate": 0.00019467450531823665,
      "loss": 13.5341,
      "step": 3002
    },
    {
      "epoch": 0.5251147540983606,
      "grad_norm": 0.8731742762576573,
      "learning_rate": 0.00019466994478776432,
      "loss": 13.4648,
      "step": 3003
    },
    {
      "epoch": 0.5252896174863388,
      "grad_norm": 0.9961283646335458,
      "learning_rate": 0.00019466538235886894,
      "loss": 13.6097,
      "step": 3004
    },
    {
      "epoch": 0.525464480874317,
      "grad_norm": 0.8882361127242414,
      "learning_rate": 0.00019466081803164195,
      "loss": 13.4844,
      "step": 3005
    },
    {
      "epoch": 0.5256393442622951,
      "grad_norm": 0.9317236310298542,
      "learning_rate": 0.00019465625180617495,
      "loss": 13.4303,
      "step": 3006
    },
    {
      "epoch": 0.5258142076502732,
      "grad_norm": 0.9961694117292905,
      "learning_rate": 0.00019465168368255946,
      "loss": 13.6866,
      "step": 3007
    },
    {
      "epoch": 0.5259890710382513,
      "grad_norm": 0.9462189294988509,
      "learning_rate": 0.0001946471136608871,
      "loss": 13.4539,
      "step": 3008
    },
    {
      "epoch": 0.5261639344262296,
      "grad_norm": 0.9629506041068492,
      "learning_rate": 0.0001946425417412495,
      "loss": 13.5455,
      "step": 3009
    },
    {
      "epoch": 0.5263387978142077,
      "grad_norm": 1.0135845082133033,
      "learning_rate": 0.0001946379679237383,
      "loss": 13.6125,
      "step": 3010
    },
    {
      "epoch": 0.5265136612021858,
      "grad_norm": 1.0095947055370746,
      "learning_rate": 0.00019463339220844535,
      "loss": 13.3648,
      "step": 3011
    },
    {
      "epoch": 0.5266885245901639,
      "grad_norm": 1.0444123582848337,
      "learning_rate": 0.00019462881459546229,
      "loss": 13.6098,
      "step": 3012
    },
    {
      "epoch": 0.526863387978142,
      "grad_norm": 1.0891146653353996,
      "learning_rate": 0.0001946242350848809,
      "loss": 13.6263,
      "step": 3013
    },
    {
      "epoch": 0.5270382513661203,
      "grad_norm": 1.006791767866901,
      "learning_rate": 0.0001946196536767931,
      "loss": 13.7341,
      "step": 3014
    },
    {
      "epoch": 0.5272131147540984,
      "grad_norm": 0.9480970810172408,
      "learning_rate": 0.00019461507037129075,
      "loss": 13.5631,
      "step": 3015
    },
    {
      "epoch": 0.5273879781420765,
      "grad_norm": 0.9822458583786237,
      "learning_rate": 0.0001946104851684657,
      "loss": 13.5903,
      "step": 3016
    },
    {
      "epoch": 0.5275628415300546,
      "grad_norm": 0.9090489096520588,
      "learning_rate": 0.00019460589806840993,
      "loss": 13.6441,
      "step": 3017
    },
    {
      "epoch": 0.5277377049180327,
      "grad_norm": 0.9186438954024314,
      "learning_rate": 0.00019460130907121545,
      "loss": 13.5995,
      "step": 3018
    },
    {
      "epoch": 0.527912568306011,
      "grad_norm": 0.8903996887584451,
      "learning_rate": 0.00019459671817697425,
      "loss": 13.2291,
      "step": 3019
    },
    {
      "epoch": 0.5280874316939891,
      "grad_norm": 0.9528447010247791,
      "learning_rate": 0.0001945921253857784,
      "loss": 13.4604,
      "step": 3020
    },
    {
      "epoch": 0.5282622950819672,
      "grad_norm": 0.8742148136260849,
      "learning_rate": 0.00019458753069771998,
      "loss": 13.4817,
      "step": 3021
    },
    {
      "epoch": 0.5284371584699453,
      "grad_norm": 0.9554395146730532,
      "learning_rate": 0.00019458293411289118,
      "loss": 13.3677,
      "step": 3022
    },
    {
      "epoch": 0.5286120218579234,
      "grad_norm": 1.1552268208638172,
      "learning_rate": 0.00019457833563138411,
      "loss": 13.42,
      "step": 3023
    },
    {
      "epoch": 0.5287868852459017,
      "grad_norm": 0.9000148489957122,
      "learning_rate": 0.00019457373525329103,
      "loss": 13.225,
      "step": 3024
    },
    {
      "epoch": 0.5289617486338798,
      "grad_norm": 0.9973048530607069,
      "learning_rate": 0.00019456913297870418,
      "loss": 13.3248,
      "step": 3025
    },
    {
      "epoch": 0.5291366120218579,
      "grad_norm": 1.0102229604719282,
      "learning_rate": 0.00019456452880771584,
      "loss": 13.3687,
      "step": 3026
    },
    {
      "epoch": 0.529311475409836,
      "grad_norm": 0.9765458537889247,
      "learning_rate": 0.00019455992274041835,
      "loss": 13.6661,
      "step": 3027
    },
    {
      "epoch": 0.5294863387978143,
      "grad_norm": 0.9490828767111555,
      "learning_rate": 0.00019455531477690407,
      "loss": 13.6995,
      "step": 3028
    },
    {
      "epoch": 0.5296612021857924,
      "grad_norm": 0.8706449491915638,
      "learning_rate": 0.00019455070491726541,
      "loss": 13.4381,
      "step": 3029
    },
    {
      "epoch": 0.5298360655737705,
      "grad_norm": 1.0346562685239498,
      "learning_rate": 0.00019454609316159482,
      "loss": 13.0425,
      "step": 3030
    },
    {
      "epoch": 0.5300109289617486,
      "grad_norm": 0.9716619278996901,
      "learning_rate": 0.00019454147950998476,
      "loss": 13.4915,
      "step": 3031
    },
    {
      "epoch": 0.5301857923497267,
      "grad_norm": 1.0422489517940812,
      "learning_rate": 0.00019453686396252775,
      "loss": 13.6012,
      "step": 3032
    },
    {
      "epoch": 0.530360655737705,
      "grad_norm": 0.9141210825143061,
      "learning_rate": 0.00019453224651931635,
      "loss": 13.5127,
      "step": 3033
    },
    {
      "epoch": 0.5305355191256831,
      "grad_norm": 1.0182945253346312,
      "learning_rate": 0.00019452762718044316,
      "loss": 13.5446,
      "step": 3034
    },
    {
      "epoch": 0.5307103825136612,
      "grad_norm": 1.0678078654487866,
      "learning_rate": 0.00019452300594600082,
      "loss": 13.3491,
      "step": 3035
    },
    {
      "epoch": 0.5308852459016393,
      "grad_norm": 0.9666731236085663,
      "learning_rate": 0.00019451838281608197,
      "loss": 13.597,
      "step": 3036
    },
    {
      "epoch": 0.5310601092896174,
      "grad_norm": 0.9493092428358406,
      "learning_rate": 0.00019451375779077932,
      "loss": 13.4366,
      "step": 3037
    },
    {
      "epoch": 0.5312349726775957,
      "grad_norm": 0.9282835578570336,
      "learning_rate": 0.00019450913087018566,
      "loss": 13.6027,
      "step": 3038
    },
    {
      "epoch": 0.5314098360655738,
      "grad_norm": 0.9889830678390822,
      "learning_rate": 0.0001945045020543937,
      "loss": 13.3566,
      "step": 3039
    },
    {
      "epoch": 0.5315846994535519,
      "grad_norm": 1.06829996328561,
      "learning_rate": 0.00019449987134349637,
      "loss": 13.4416,
      "step": 3040
    },
    {
      "epoch": 0.53175956284153,
      "grad_norm": 0.9870929704085504,
      "learning_rate": 0.00019449523873758646,
      "loss": 13.705,
      "step": 3041
    },
    {
      "epoch": 0.5319344262295082,
      "grad_norm": 1.204447587763946,
      "learning_rate": 0.0001944906042367568,
      "loss": 13.6622,
      "step": 3042
    },
    {
      "epoch": 0.5321092896174864,
      "grad_norm": 1.0789850699348593,
      "learning_rate": 0.0001944859678411005,
      "loss": 13.6206,
      "step": 3043
    },
    {
      "epoch": 0.5322841530054645,
      "grad_norm": 1.0889022102673833,
      "learning_rate": 0.00019448132955071038,
      "loss": 13.4631,
      "step": 3044
    },
    {
      "epoch": 0.5324590163934426,
      "grad_norm": 1.2336553897803582,
      "learning_rate": 0.00019447668936567952,
      "loss": 13.4297,
      "step": 3045
    },
    {
      "epoch": 0.5326338797814207,
      "grad_norm": 0.8696917142956884,
      "learning_rate": 0.00019447204728610093,
      "loss": 13.202,
      "step": 3046
    },
    {
      "epoch": 0.5328087431693989,
      "grad_norm": 1.2286856183300683,
      "learning_rate": 0.00019446740331206778,
      "loss": 13.4247,
      "step": 3047
    },
    {
      "epoch": 0.5329836065573771,
      "grad_norm": 1.0087663471921937,
      "learning_rate": 0.00019446275744367309,
      "loss": 13.2396,
      "step": 3048
    },
    {
      "epoch": 0.5331584699453552,
      "grad_norm": 0.9270972249593211,
      "learning_rate": 0.00019445810968101007,
      "loss": 13.4697,
      "step": 3049
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 1.0499256533493722,
      "learning_rate": 0.00019445346002417194,
      "loss": 13.5297,
      "step": 3050
    },
    {
      "epoch": 0.5335081967213114,
      "grad_norm": 0.968952519462038,
      "learning_rate": 0.00019444880847325195,
      "loss": 13.5379,
      "step": 3051
    },
    {
      "epoch": 0.5336830601092896,
      "grad_norm": 1.337126236968979,
      "learning_rate": 0.00019444415502834335,
      "loss": 13.5782,
      "step": 3052
    },
    {
      "epoch": 0.5338579234972678,
      "grad_norm": 1.0488325652550752,
      "learning_rate": 0.00019443949968953945,
      "loss": 13.5253,
      "step": 3053
    },
    {
      "epoch": 0.5340327868852459,
      "grad_norm": 1.063502137754246,
      "learning_rate": 0.0001944348424569336,
      "loss": 13.6559,
      "step": 3054
    },
    {
      "epoch": 0.534207650273224,
      "grad_norm": 0.9169689622468017,
      "learning_rate": 0.00019443018333061923,
      "loss": 13.0279,
      "step": 3055
    },
    {
      "epoch": 0.5343825136612022,
      "grad_norm": 1.2291141482171017,
      "learning_rate": 0.00019442552231068973,
      "loss": 13.7557,
      "step": 3056
    },
    {
      "epoch": 0.5345573770491804,
      "grad_norm": 1.008088447195288,
      "learning_rate": 0.00019442085939723857,
      "loss": 13.7031,
      "step": 3057
    },
    {
      "epoch": 0.5347322404371585,
      "grad_norm": 1.1202727508154962,
      "learning_rate": 0.0001944161945903593,
      "loss": 13.4083,
      "step": 3058
    },
    {
      "epoch": 0.5349071038251366,
      "grad_norm": 1.00005360956544,
      "learning_rate": 0.0001944115278901454,
      "loss": 13.6728,
      "step": 3059
    },
    {
      "epoch": 0.5350819672131147,
      "grad_norm": 1.0596457826789192,
      "learning_rate": 0.00019440685929669052,
      "loss": 13.1665,
      "step": 3060
    },
    {
      "epoch": 0.5352568306010929,
      "grad_norm": 1.0047261411246602,
      "learning_rate": 0.00019440218881008824,
      "loss": 13.5445,
      "step": 3061
    },
    {
      "epoch": 0.5354316939890711,
      "grad_norm": 1.0558936361926792,
      "learning_rate": 0.0001943975164304322,
      "loss": 13.6962,
      "step": 3062
    },
    {
      "epoch": 0.5356065573770492,
      "grad_norm": 1.1224163564739325,
      "learning_rate": 0.00019439284215781613,
      "loss": 13.6719,
      "step": 3063
    },
    {
      "epoch": 0.5357814207650273,
      "grad_norm": 1.0142584209332712,
      "learning_rate": 0.00019438816599233374,
      "loss": 13.6892,
      "step": 3064
    },
    {
      "epoch": 0.5359562841530054,
      "grad_norm": 1.0983161069080956,
      "learning_rate": 0.00019438348793407881,
      "loss": 13.5979,
      "step": 3065
    },
    {
      "epoch": 0.5361311475409836,
      "grad_norm": 0.891230804429564,
      "learning_rate": 0.0001943788079831452,
      "loss": 13.3116,
      "step": 3066
    },
    {
      "epoch": 0.5363060109289618,
      "grad_norm": 1.0674464621024404,
      "learning_rate": 0.0001943741261396266,
      "loss": 13.5436,
      "step": 3067
    },
    {
      "epoch": 0.5364808743169399,
      "grad_norm": 1.0049142041815677,
      "learning_rate": 0.00019436944240361709,
      "loss": 13.3692,
      "step": 3068
    },
    {
      "epoch": 0.536655737704918,
      "grad_norm": 1.007505229733278,
      "learning_rate": 0.00019436475677521052,
      "loss": 13.442,
      "step": 3069
    },
    {
      "epoch": 0.5368306010928962,
      "grad_norm": 0.9655057000874749,
      "learning_rate": 0.00019436006925450082,
      "loss": 13.2892,
      "step": 3070
    },
    {
      "epoch": 0.5370054644808743,
      "grad_norm": 1.2940240031220205,
      "learning_rate": 0.000194355379841582,
      "loss": 13.3976,
      "step": 3071
    },
    {
      "epoch": 0.5371803278688525,
      "grad_norm": 0.930304580123482,
      "learning_rate": 0.00019435068853654807,
      "loss": 13.7016,
      "step": 3072
    },
    {
      "epoch": 0.5373551912568306,
      "grad_norm": 1.073267529386429,
      "learning_rate": 0.00019434599533949318,
      "loss": 13.6618,
      "step": 3073
    },
    {
      "epoch": 0.5375300546448087,
      "grad_norm": 0.9198712683790711,
      "learning_rate": 0.0001943413002505114,
      "loss": 13.4367,
      "step": 3074
    },
    {
      "epoch": 0.5377049180327869,
      "grad_norm": 0.9855546039886469,
      "learning_rate": 0.00019433660326969687,
      "loss": 13.4353,
      "step": 3075
    },
    {
      "epoch": 0.537879781420765,
      "grad_norm": 0.9245621694475789,
      "learning_rate": 0.0001943319043971438,
      "loss": 13.6168,
      "step": 3076
    },
    {
      "epoch": 0.5380546448087432,
      "grad_norm": 0.8929498646382743,
      "learning_rate": 0.00019432720363294644,
      "loss": 13.6907,
      "step": 3077
    },
    {
      "epoch": 0.5382295081967213,
      "grad_norm": 0.8934499178956162,
      "learning_rate": 0.00019432250097719896,
      "loss": 13.2924,
      "step": 3078
    },
    {
      "epoch": 0.5384043715846994,
      "grad_norm": 1.033823542019808,
      "learning_rate": 0.00019431779642999577,
      "loss": 13.3227,
      "step": 3079
    },
    {
      "epoch": 0.5385792349726776,
      "grad_norm": 1.3368439576784334,
      "learning_rate": 0.00019431308999143116,
      "loss": 13.6571,
      "step": 3080
    },
    {
      "epoch": 0.5387540983606557,
      "grad_norm": 0.9239256222345374,
      "learning_rate": 0.00019430838166159954,
      "loss": 13.5977,
      "step": 3081
    },
    {
      "epoch": 0.5389289617486339,
      "grad_norm": 1.0865055239070485,
      "learning_rate": 0.00019430367144059526,
      "loss": 13.4001,
      "step": 3082
    },
    {
      "epoch": 0.539103825136612,
      "grad_norm": 1.0162470383270605,
      "learning_rate": 0.00019429895932851284,
      "loss": 13.5145,
      "step": 3083
    },
    {
      "epoch": 0.5392786885245902,
      "grad_norm": 1.029016338775725,
      "learning_rate": 0.00019429424532544677,
      "loss": 13.5367,
      "step": 3084
    },
    {
      "epoch": 0.5394535519125683,
      "grad_norm": 1.0371987706339918,
      "learning_rate": 0.00019428952943149154,
      "loss": 13.6819,
      "step": 3085
    },
    {
      "epoch": 0.5396284153005464,
      "grad_norm": 1.043865443570372,
      "learning_rate": 0.00019428481164674173,
      "loss": 13.7894,
      "step": 3086
    },
    {
      "epoch": 0.5398032786885246,
      "grad_norm": 1.0772753456362114,
      "learning_rate": 0.00019428009197129195,
      "loss": 13.5344,
      "step": 3087
    },
    {
      "epoch": 0.5399781420765027,
      "grad_norm": 1.019547000216977,
      "learning_rate": 0.00019427537040523686,
      "loss": 13.6489,
      "step": 3088
    },
    {
      "epoch": 0.5401530054644809,
      "grad_norm": 1.0607338597402074,
      "learning_rate": 0.00019427064694867112,
      "loss": 13.5197,
      "step": 3089
    },
    {
      "epoch": 0.540327868852459,
      "grad_norm": 0.9341003798393159,
      "learning_rate": 0.0001942659216016895,
      "loss": 13.3812,
      "step": 3090
    },
    {
      "epoch": 0.5405027322404372,
      "grad_norm": 1.1291889428498938,
      "learning_rate": 0.00019426119436438666,
      "loss": 13.5789,
      "step": 3091
    },
    {
      "epoch": 0.5406775956284153,
      "grad_norm": 0.915776986329911,
      "learning_rate": 0.00019425646523685747,
      "loss": 13.1553,
      "step": 3092
    },
    {
      "epoch": 0.5408524590163934,
      "grad_norm": 1.0487801284265326,
      "learning_rate": 0.00019425173421919677,
      "loss": 13.2843,
      "step": 3093
    },
    {
      "epoch": 0.5410273224043716,
      "grad_norm": 0.9774866630777646,
      "learning_rate": 0.0001942470013114994,
      "loss": 13.319,
      "step": 3094
    },
    {
      "epoch": 0.5412021857923497,
      "grad_norm": 1.005080252698169,
      "learning_rate": 0.00019424226651386026,
      "loss": 13.3959,
      "step": 3095
    },
    {
      "epoch": 0.5413770491803279,
      "grad_norm": 1.11698905399647,
      "learning_rate": 0.00019423752982637433,
      "loss": 13.5562,
      "step": 3096
    },
    {
      "epoch": 0.541551912568306,
      "grad_norm": 1.0065847959178953,
      "learning_rate": 0.00019423279124913657,
      "loss": 13.6967,
      "step": 3097
    },
    {
      "epoch": 0.5417267759562842,
      "grad_norm": 0.9734238457686308,
      "learning_rate": 0.00019422805078224206,
      "loss": 13.2404,
      "step": 3098
    },
    {
      "epoch": 0.5419016393442623,
      "grad_norm": 1.0229539948552857,
      "learning_rate": 0.00019422330842578577,
      "loss": 13.6248,
      "step": 3099
    },
    {
      "epoch": 0.5420765027322404,
      "grad_norm": 0.9309659568578983,
      "learning_rate": 0.0001942185641798628,
      "loss": 13.7443,
      "step": 3100
    },
    {
      "epoch": 0.5422513661202186,
      "grad_norm": 0.9323369544495618,
      "learning_rate": 0.00019421381804456838,
      "loss": 13.5926,
      "step": 3101
    },
    {
      "epoch": 0.5424262295081967,
      "grad_norm": 0.9845017393731502,
      "learning_rate": 0.00019420907001999765,
      "loss": 13.6641,
      "step": 3102
    },
    {
      "epoch": 0.5426010928961749,
      "grad_norm": 0.9292865807098959,
      "learning_rate": 0.00019420432010624578,
      "loss": 13.5403,
      "step": 3103
    },
    {
      "epoch": 0.542775956284153,
      "grad_norm": 0.969776699204987,
      "learning_rate": 0.00019419956830340804,
      "loss": 13.3791,
      "step": 3104
    },
    {
      "epoch": 0.5429508196721311,
      "grad_norm": 1.011707958038552,
      "learning_rate": 0.00019419481461157973,
      "loss": 13.5372,
      "step": 3105
    },
    {
      "epoch": 0.5431256830601093,
      "grad_norm": 0.8972131672805524,
      "learning_rate": 0.0001941900590308562,
      "loss": 13.5077,
      "step": 3106
    },
    {
      "epoch": 0.5433005464480875,
      "grad_norm": 1.0332903151207435,
      "learning_rate": 0.00019418530156133275,
      "loss": 13.4966,
      "step": 3107
    },
    {
      "epoch": 0.5434754098360656,
      "grad_norm": 1.063643326455163,
      "learning_rate": 0.00019418054220310483,
      "loss": 13.4903,
      "step": 3108
    },
    {
      "epoch": 0.5436502732240437,
      "grad_norm": 1.0888998617492767,
      "learning_rate": 0.00019417578095626788,
      "loss": 13.786,
      "step": 3109
    },
    {
      "epoch": 0.5438251366120218,
      "grad_norm": 1.0240359931480465,
      "learning_rate": 0.00019417101782091733,
      "loss": 13.3309,
      "step": 3110
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.9974610763029776,
      "learning_rate": 0.00019416625279714876,
      "loss": 13.5433,
      "step": 3111
    },
    {
      "epoch": 0.5441748633879782,
      "grad_norm": 1.0091639887003714,
      "learning_rate": 0.00019416148588505768,
      "loss": 13.5757,
      "step": 3112
    },
    {
      "epoch": 0.5443497267759563,
      "grad_norm": 1.1341500990569164,
      "learning_rate": 0.0001941567170847397,
      "loss": 13.8217,
      "step": 3113
    },
    {
      "epoch": 0.5445245901639344,
      "grad_norm": 0.9571517756928334,
      "learning_rate": 0.00019415194639629043,
      "loss": 13.4467,
      "step": 3114
    },
    {
      "epoch": 0.5446994535519125,
      "grad_norm": 1.058431283061758,
      "learning_rate": 0.00019414717381980556,
      "loss": 13.5581,
      "step": 3115
    },
    {
      "epoch": 0.5448743169398907,
      "grad_norm": 1.0120397863501078,
      "learning_rate": 0.00019414239935538075,
      "loss": 13.5753,
      "step": 3116
    },
    {
      "epoch": 0.5450491803278689,
      "grad_norm": 1.0147600364429754,
      "learning_rate": 0.00019413762300311182,
      "loss": 13.583,
      "step": 3117
    },
    {
      "epoch": 0.545224043715847,
      "grad_norm": 0.955041994541866,
      "learning_rate": 0.00019413284476309454,
      "loss": 13.5887,
      "step": 3118
    },
    {
      "epoch": 0.5453989071038251,
      "grad_norm": 0.944098036121129,
      "learning_rate": 0.00019412806463542463,
      "loss": 13.5028,
      "step": 3119
    },
    {
      "epoch": 0.5455737704918033,
      "grad_norm": 1.010158647955324,
      "learning_rate": 0.00019412328262019806,
      "loss": 13.4089,
      "step": 3120
    },
    {
      "epoch": 0.5457486338797815,
      "grad_norm": 1.003446143125721,
      "learning_rate": 0.00019411849871751063,
      "loss": 13.4881,
      "step": 3121
    },
    {
      "epoch": 0.5459234972677596,
      "grad_norm": 1.0969957555382233,
      "learning_rate": 0.00019411371292745838,
      "loss": 13.4019,
      "step": 3122
    },
    {
      "epoch": 0.5460983606557377,
      "grad_norm": 0.9757338969954107,
      "learning_rate": 0.0001941089252501372,
      "loss": 13.5132,
      "step": 3123
    },
    {
      "epoch": 0.5462732240437158,
      "grad_norm": 1.0384075107315236,
      "learning_rate": 0.00019410413568564307,
      "loss": 13.3358,
      "step": 3124
    },
    {
      "epoch": 0.546448087431694,
      "grad_norm": 0.9416875897323799,
      "learning_rate": 0.00019409934423407211,
      "loss": 13.5192,
      "step": 3125
    },
    {
      "epoch": 0.5466229508196722,
      "grad_norm": 1.089736765045309,
      "learning_rate": 0.00019409455089552038,
      "loss": 13.4615,
      "step": 3126
    },
    {
      "epoch": 0.5467978142076503,
      "grad_norm": 0.9644280363886458,
      "learning_rate": 0.00019408975567008398,
      "loss": 13.4862,
      "step": 3127
    },
    {
      "epoch": 0.5469726775956284,
      "grad_norm": 0.9358474730033601,
      "learning_rate": 0.0001940849585578591,
      "loss": 13.4062,
      "step": 3128
    },
    {
      "epoch": 0.5471475409836065,
      "grad_norm": 0.9367621092726899,
      "learning_rate": 0.00019408015955894194,
      "loss": 13.5818,
      "step": 3129
    },
    {
      "epoch": 0.5473224043715847,
      "grad_norm": 1.0149592736860327,
      "learning_rate": 0.0001940753586734287,
      "loss": 13.2992,
      "step": 3130
    },
    {
      "epoch": 0.5474972677595629,
      "grad_norm": 0.9601025360695273,
      "learning_rate": 0.00019407055590141568,
      "loss": 13.305,
      "step": 3131
    },
    {
      "epoch": 0.547672131147541,
      "grad_norm": 0.9966311085809375,
      "learning_rate": 0.00019406575124299917,
      "loss": 13.741,
      "step": 3132
    },
    {
      "epoch": 0.5478469945355191,
      "grad_norm": 1.0744534225296158,
      "learning_rate": 0.00019406094469827552,
      "loss": 13.6433,
      "step": 3133
    },
    {
      "epoch": 0.5480218579234972,
      "grad_norm": 0.961266397466507,
      "learning_rate": 0.00019405613626734112,
      "loss": 13.5507,
      "step": 3134
    },
    {
      "epoch": 0.5481967213114755,
      "grad_norm": 0.9585978287877849,
      "learning_rate": 0.0001940513259502924,
      "loss": 13.6421,
      "step": 3135
    },
    {
      "epoch": 0.5483715846994536,
      "grad_norm": 0.8748092724299013,
      "learning_rate": 0.00019404651374722582,
      "loss": 13.4823,
      "step": 3136
    },
    {
      "epoch": 0.5485464480874317,
      "grad_norm": 1.020439076919462,
      "learning_rate": 0.0001940416996582379,
      "loss": 13.2651,
      "step": 3137
    },
    {
      "epoch": 0.5487213114754098,
      "grad_norm": 1.0938769994519904,
      "learning_rate": 0.00019403688368342515,
      "loss": 13.3936,
      "step": 3138
    },
    {
      "epoch": 0.5488961748633879,
      "grad_norm": 0.9153364406950043,
      "learning_rate": 0.00019403206582288413,
      "loss": 13.6958,
      "step": 3139
    },
    {
      "epoch": 0.5490710382513662,
      "grad_norm": 0.9154172879093757,
      "learning_rate": 0.0001940272460767115,
      "loss": 13.5812,
      "step": 3140
    },
    {
      "epoch": 0.5492459016393443,
      "grad_norm": 1.0346244860484373,
      "learning_rate": 0.00019402242444500388,
      "loss": 13.464,
      "step": 3141
    },
    {
      "epoch": 0.5494207650273224,
      "grad_norm": 0.9619386901395786,
      "learning_rate": 0.00019401760092785795,
      "loss": 13.4541,
      "step": 3142
    },
    {
      "epoch": 0.5495956284153005,
      "grad_norm": 0.9905634707038713,
      "learning_rate": 0.00019401277552537046,
      "loss": 13.4612,
      "step": 3143
    },
    {
      "epoch": 0.5497704918032786,
      "grad_norm": 1.2295274632162179,
      "learning_rate": 0.00019400794823763815,
      "loss": 13.6981,
      "step": 3144
    },
    {
      "epoch": 0.5499453551912569,
      "grad_norm": 0.9462202989150925,
      "learning_rate": 0.0001940031190647578,
      "loss": 13.535,
      "step": 3145
    },
    {
      "epoch": 0.550120218579235,
      "grad_norm": 1.0467750467988362,
      "learning_rate": 0.00019399828800682636,
      "loss": 13.4118,
      "step": 3146
    },
    {
      "epoch": 0.5502950819672131,
      "grad_norm": 1.0100060393371029,
      "learning_rate": 0.0001939934550639406,
      "loss": 13.7087,
      "step": 3147
    },
    {
      "epoch": 0.5504699453551912,
      "grad_norm": 0.869755730799023,
      "learning_rate": 0.00019398862023619747,
      "loss": 13.6287,
      "step": 3148
    },
    {
      "epoch": 0.5506448087431693,
      "grad_norm": 0.9877756091167166,
      "learning_rate": 0.0001939837835236939,
      "loss": 13.4266,
      "step": 3149
    },
    {
      "epoch": 0.5508196721311476,
      "grad_norm": 0.9229823961279358,
      "learning_rate": 0.00019397894492652692,
      "loss": 13.5654,
      "step": 3150
    },
    {
      "epoch": 0.5509945355191257,
      "grad_norm": 0.9384627523665443,
      "learning_rate": 0.00019397410444479355,
      "loss": 13.2579,
      "step": 3151
    },
    {
      "epoch": 0.5511693989071038,
      "grad_norm": 0.9067649998174367,
      "learning_rate": 0.00019396926207859084,
      "loss": 13.4037,
      "step": 3152
    },
    {
      "epoch": 0.5513442622950819,
      "grad_norm": 0.9622598867424002,
      "learning_rate": 0.00019396441782801592,
      "loss": 13.3102,
      "step": 3153
    },
    {
      "epoch": 0.5515191256830602,
      "grad_norm": 0.9823163290271485,
      "learning_rate": 0.00019395957169316587,
      "loss": 13.2173,
      "step": 3154
    },
    {
      "epoch": 0.5516939890710383,
      "grad_norm": 0.9800510111930507,
      "learning_rate": 0.00019395472367413796,
      "loss": 13.5196,
      "step": 3155
    },
    {
      "epoch": 0.5518688524590164,
      "grad_norm": 0.8874562136286474,
      "learning_rate": 0.00019394987377102934,
      "loss": 13.5386,
      "step": 3156
    },
    {
      "epoch": 0.5520437158469945,
      "grad_norm": 1.089369471497548,
      "learning_rate": 0.0001939450219839373,
      "loss": 13.3465,
      "step": 3157
    },
    {
      "epoch": 0.5522185792349726,
      "grad_norm": 0.9289650471966843,
      "learning_rate": 0.0001939401683129591,
      "loss": 13.7452,
      "step": 3158
    },
    {
      "epoch": 0.5523934426229509,
      "grad_norm": 1.0209102292114272,
      "learning_rate": 0.0001939353127581921,
      "loss": 13.5961,
      "step": 3159
    },
    {
      "epoch": 0.552568306010929,
      "grad_norm": 0.8930177396743438,
      "learning_rate": 0.00019393045531973364,
      "loss": 13.501,
      "step": 3160
    },
    {
      "epoch": 0.5527431693989071,
      "grad_norm": 1.0565266194213612,
      "learning_rate": 0.00019392559599768118,
      "loss": 13.1795,
      "step": 3161
    },
    {
      "epoch": 0.5529180327868852,
      "grad_norm": 1.15714046776709,
      "learning_rate": 0.00019392073479213213,
      "loss": 13.4475,
      "step": 3162
    },
    {
      "epoch": 0.5530928961748633,
      "grad_norm": 1.0026275295742564,
      "learning_rate": 0.00019391587170318394,
      "loss": 13.5515,
      "step": 3163
    },
    {
      "epoch": 0.5532677595628416,
      "grad_norm": 0.9864628391131909,
      "learning_rate": 0.00019391100673093418,
      "loss": 13.5333,
      "step": 3164
    },
    {
      "epoch": 0.5534426229508197,
      "grad_norm": 1.017082926915241,
      "learning_rate": 0.0001939061398754804,
      "loss": 13.6163,
      "step": 3165
    },
    {
      "epoch": 0.5536174863387978,
      "grad_norm": 1.0749854194959676,
      "learning_rate": 0.0001939012711369202,
      "loss": 13.4086,
      "step": 3166
    },
    {
      "epoch": 0.5537923497267759,
      "grad_norm": 0.8662236180117302,
      "learning_rate": 0.00019389640051535115,
      "loss": 13.6291,
      "step": 3167
    },
    {
      "epoch": 0.553967213114754,
      "grad_norm": 1.075667211685345,
      "learning_rate": 0.000193891528010871,
      "loss": 13.4025,
      "step": 3168
    },
    {
      "epoch": 0.5541420765027323,
      "grad_norm": 0.9530911031262244,
      "learning_rate": 0.00019388665362357743,
      "loss": 13.6351,
      "step": 3169
    },
    {
      "epoch": 0.5543169398907104,
      "grad_norm": 0.9027765238563402,
      "learning_rate": 0.00019388177735356818,
      "loss": 13.202,
      "step": 3170
    },
    {
      "epoch": 0.5544918032786885,
      "grad_norm": 1.052055100846841,
      "learning_rate": 0.00019387689920094103,
      "loss": 13.5336,
      "step": 3171
    },
    {
      "epoch": 0.5546666666666666,
      "grad_norm": 0.9478985323806716,
      "learning_rate": 0.00019387201916579383,
      "loss": 13.3687,
      "step": 3172
    },
    {
      "epoch": 0.5548415300546448,
      "grad_norm": 0.993630494668098,
      "learning_rate": 0.00019386713724822442,
      "loss": 13.358,
      "step": 3173
    },
    {
      "epoch": 0.555016393442623,
      "grad_norm": 0.9588585577774756,
      "learning_rate": 0.00019386225344833068,
      "loss": 13.626,
      "step": 3174
    },
    {
      "epoch": 0.5551912568306011,
      "grad_norm": 1.1662835661675548,
      "learning_rate": 0.00019385736776621058,
      "loss": 13.5107,
      "step": 3175
    },
    {
      "epoch": 0.5553661202185792,
      "grad_norm": 0.9010129689350684,
      "learning_rate": 0.00019385248020196207,
      "loss": 13.2962,
      "step": 3176
    },
    {
      "epoch": 0.5555409836065573,
      "grad_norm": 1.097181501409815,
      "learning_rate": 0.00019384759075568314,
      "loss": 13.1945,
      "step": 3177
    },
    {
      "epoch": 0.5557158469945355,
      "grad_norm": 0.962526804159563,
      "learning_rate": 0.00019384269942747192,
      "loss": 13.5611,
      "step": 3178
    },
    {
      "epoch": 0.5558907103825137,
      "grad_norm": 1.1921537531619415,
      "learning_rate": 0.00019383780621742643,
      "loss": 13.6145,
      "step": 3179
    },
    {
      "epoch": 0.5560655737704918,
      "grad_norm": 1.063122266071612,
      "learning_rate": 0.00019383291112564478,
      "loss": 13.3959,
      "step": 3180
    },
    {
      "epoch": 0.5562404371584699,
      "grad_norm": 1.042595177661906,
      "learning_rate": 0.00019382801415222516,
      "loss": 13.3084,
      "step": 3181
    },
    {
      "epoch": 0.556415300546448,
      "grad_norm": 1.0499098494473624,
      "learning_rate": 0.00019382311529726577,
      "loss": 13.488,
      "step": 3182
    },
    {
      "epoch": 0.5565901639344262,
      "grad_norm": 0.9407627368390374,
      "learning_rate": 0.00019381821456086485,
      "loss": 13.6532,
      "step": 3183
    },
    {
      "epoch": 0.5567650273224044,
      "grad_norm": 1.1799405056384458,
      "learning_rate": 0.00019381331194312065,
      "loss": 13.5539,
      "step": 3184
    },
    {
      "epoch": 0.5569398907103825,
      "grad_norm": 0.9540794875092369,
      "learning_rate": 0.00019380840744413153,
      "loss": 13.6319,
      "step": 3185
    },
    {
      "epoch": 0.5571147540983606,
      "grad_norm": 1.1133954275345972,
      "learning_rate": 0.00019380350106399578,
      "loss": 13.5181,
      "step": 3186
    },
    {
      "epoch": 0.5572896174863388,
      "grad_norm": 0.8934395865122524,
      "learning_rate": 0.00019379859280281182,
      "loss": 13.4579,
      "step": 3187
    },
    {
      "epoch": 0.557464480874317,
      "grad_norm": 1.0976473693480804,
      "learning_rate": 0.0001937936826606781,
      "loss": 13.3263,
      "step": 3188
    },
    {
      "epoch": 0.5576393442622951,
      "grad_norm": 0.9263543188500467,
      "learning_rate": 0.00019378877063769309,
      "loss": 13.3024,
      "step": 3189
    },
    {
      "epoch": 0.5578142076502732,
      "grad_norm": 1.0754446641923832,
      "learning_rate": 0.0001937838567339552,
      "loss": 13.4283,
      "step": 3190
    },
    {
      "epoch": 0.5579890710382513,
      "grad_norm": 1.0280366884493797,
      "learning_rate": 0.00019377894094956303,
      "loss": 13.4116,
      "step": 3191
    },
    {
      "epoch": 0.5581639344262295,
      "grad_norm": 0.9607883714322091,
      "learning_rate": 0.00019377402328461517,
      "loss": 13.5555,
      "step": 3192
    },
    {
      "epoch": 0.5583387978142077,
      "grad_norm": 1.0520557904243926,
      "learning_rate": 0.00019376910373921022,
      "loss": 13.6659,
      "step": 3193
    },
    {
      "epoch": 0.5585136612021858,
      "grad_norm": 0.9107020694382304,
      "learning_rate": 0.00019376418231344683,
      "loss": 13.3072,
      "step": 3194
    },
    {
      "epoch": 0.5586885245901639,
      "grad_norm": 0.9252654660877263,
      "learning_rate": 0.00019375925900742372,
      "loss": 13.5406,
      "step": 3195
    },
    {
      "epoch": 0.5588633879781421,
      "grad_norm": 0.9677383154535906,
      "learning_rate": 0.00019375433382123953,
      "loss": 13.4359,
      "step": 3196
    },
    {
      "epoch": 0.5590382513661202,
      "grad_norm": 1.2207912192298587,
      "learning_rate": 0.0001937494067549931,
      "loss": 13.3398,
      "step": 3197
    },
    {
      "epoch": 0.5592131147540984,
      "grad_norm": 1.0327299925515636,
      "learning_rate": 0.00019374447780878327,
      "loss": 13.1123,
      "step": 3198
    },
    {
      "epoch": 0.5593879781420765,
      "grad_norm": 1.0299285444789947,
      "learning_rate": 0.00019373954698270876,
      "loss": 13.539,
      "step": 3199
    },
    {
      "epoch": 0.5595628415300546,
      "grad_norm": 1.0069742877495904,
      "learning_rate": 0.00019373461427686853,
      "loss": 13.4754,
      "step": 3200
    },
    {
      "epoch": 0.5597377049180328,
      "grad_norm": 1.0366353194064744,
      "learning_rate": 0.0001937296796913615,
      "loss": 13.1238,
      "step": 3201
    },
    {
      "epoch": 0.5599125683060109,
      "grad_norm": 1.0279063487001696,
      "learning_rate": 0.00019372474322628663,
      "loss": 13.448,
      "step": 3202
    },
    {
      "epoch": 0.5600874316939891,
      "grad_norm": 0.9130322574050452,
      "learning_rate": 0.00019371980488174285,
      "loss": 13.231,
      "step": 3203
    },
    {
      "epoch": 0.5602622950819672,
      "grad_norm": 1.175841242129236,
      "learning_rate": 0.0001937148646578292,
      "loss": 13.3844,
      "step": 3204
    },
    {
      "epoch": 0.5604371584699454,
      "grad_norm": 0.9909934521895752,
      "learning_rate": 0.0001937099225546448,
      "loss": 13.54,
      "step": 3205
    },
    {
      "epoch": 0.5606120218579235,
      "grad_norm": 1.2410723775216788,
      "learning_rate": 0.00019370497857228874,
      "loss": 13.6784,
      "step": 3206
    },
    {
      "epoch": 0.5607868852459016,
      "grad_norm": 0.9966496398420365,
      "learning_rate": 0.0001937000327108601,
      "loss": 13.4523,
      "step": 3207
    },
    {
      "epoch": 0.5609617486338798,
      "grad_norm": 1.1916029012484317,
      "learning_rate": 0.00019369508497045813,
      "loss": 13.5509,
      "step": 3208
    },
    {
      "epoch": 0.5611366120218579,
      "grad_norm": 0.956760013605472,
      "learning_rate": 0.00019369013535118206,
      "loss": 13.4159,
      "step": 3209
    },
    {
      "epoch": 0.5613114754098361,
      "grad_norm": 1.0316996957035023,
      "learning_rate": 0.00019368518385313107,
      "loss": 13.4914,
      "step": 3210
    },
    {
      "epoch": 0.5614863387978142,
      "grad_norm": 1.007299079495291,
      "learning_rate": 0.00019368023047640448,
      "loss": 13.538,
      "step": 3211
    },
    {
      "epoch": 0.5616612021857923,
      "grad_norm": 1.0312459087053443,
      "learning_rate": 0.0001936752752211017,
      "loss": 13.3806,
      "step": 3212
    },
    {
      "epoch": 0.5618360655737705,
      "grad_norm": 0.9444176716782282,
      "learning_rate": 0.00019367031808732198,
      "loss": 13.3661,
      "step": 3213
    },
    {
      "epoch": 0.5620109289617486,
      "grad_norm": 0.9308913066831762,
      "learning_rate": 0.0001936653590751648,
      "loss": 13.2405,
      "step": 3214
    },
    {
      "epoch": 0.5621857923497268,
      "grad_norm": 1.1261625514414786,
      "learning_rate": 0.00019366039818472953,
      "loss": 13.3495,
      "step": 3215
    },
    {
      "epoch": 0.5623606557377049,
      "grad_norm": 0.9306977431819692,
      "learning_rate": 0.00019365543541611575,
      "loss": 13.427,
      "step": 3216
    },
    {
      "epoch": 0.5625355191256831,
      "grad_norm": 1.0067297918792997,
      "learning_rate": 0.00019365047076942292,
      "loss": 13.5,
      "step": 3217
    },
    {
      "epoch": 0.5627103825136612,
      "grad_norm": 0.9206215397995078,
      "learning_rate": 0.0001936455042447506,
      "loss": 13.4318,
      "step": 3218
    },
    {
      "epoch": 0.5628852459016394,
      "grad_norm": 1.0255113742380504,
      "learning_rate": 0.00019364053584219838,
      "loss": 13.4486,
      "step": 3219
    },
    {
      "epoch": 0.5630601092896175,
      "grad_norm": 0.9694791560500665,
      "learning_rate": 0.00019363556556186592,
      "loss": 13.4753,
      "step": 3220
    },
    {
      "epoch": 0.5632349726775956,
      "grad_norm": 1.073545964814841,
      "learning_rate": 0.00019363059340385288,
      "loss": 13.4499,
      "step": 3221
    },
    {
      "epoch": 0.5634098360655738,
      "grad_norm": 0.9389985326410469,
      "learning_rate": 0.00019362561936825894,
      "loss": 13.3389,
      "step": 3222
    },
    {
      "epoch": 0.5635846994535519,
      "grad_norm": 0.9879605834653101,
      "learning_rate": 0.00019362064345518384,
      "loss": 13.5552,
      "step": 3223
    },
    {
      "epoch": 0.5637595628415301,
      "grad_norm": 1.0342302163165231,
      "learning_rate": 0.00019361566566472745,
      "loss": 13.5067,
      "step": 3224
    },
    {
      "epoch": 0.5639344262295082,
      "grad_norm": 0.9022785140705755,
      "learning_rate": 0.00019361068599698945,
      "loss": 13.6647,
      "step": 3225
    },
    {
      "epoch": 0.5641092896174863,
      "grad_norm": 0.9245216806715479,
      "learning_rate": 0.00019360570445206983,
      "loss": 13.0612,
      "step": 3226
    },
    {
      "epoch": 0.5642841530054645,
      "grad_norm": 0.9980914940785754,
      "learning_rate": 0.0001936007210300684,
      "loss": 13.58,
      "step": 3227
    },
    {
      "epoch": 0.5644590163934426,
      "grad_norm": 0.9113504818739426,
      "learning_rate": 0.00019359573573108512,
      "loss": 13.273,
      "step": 3228
    },
    {
      "epoch": 0.5646338797814208,
      "grad_norm": 0.8747376685266343,
      "learning_rate": 0.00019359074855522,
      "loss": 13.5063,
      "step": 3229
    },
    {
      "epoch": 0.5648087431693989,
      "grad_norm": 0.8989504207340704,
      "learning_rate": 0.00019358575950257297,
      "loss": 13.435,
      "step": 3230
    },
    {
      "epoch": 0.564983606557377,
      "grad_norm": 0.8889398282184461,
      "learning_rate": 0.00019358076857324414,
      "loss": 13.1211,
      "step": 3231
    },
    {
      "epoch": 0.5651584699453552,
      "grad_norm": 0.9722765484398351,
      "learning_rate": 0.00019357577576733357,
      "loss": 13.4077,
      "step": 3232
    },
    {
      "epoch": 0.5653333333333334,
      "grad_norm": 1.0883617908441237,
      "learning_rate": 0.0001935707810849414,
      "loss": 13.249,
      "step": 3233
    },
    {
      "epoch": 0.5655081967213115,
      "grad_norm": 0.9903644717797279,
      "learning_rate": 0.00019356578452616772,
      "loss": 13.3011,
      "step": 3234
    },
    {
      "epoch": 0.5656830601092896,
      "grad_norm": 1.2360605152290385,
      "learning_rate": 0.0001935607860911128,
      "loss": 13.669,
      "step": 3235
    },
    {
      "epoch": 0.5658579234972677,
      "grad_norm": 0.927346696218434,
      "learning_rate": 0.00019355578577987683,
      "loss": 13.2054,
      "step": 3236
    },
    {
      "epoch": 0.5660327868852459,
      "grad_norm": 0.9854599694672229,
      "learning_rate": 0.0001935507835925601,
      "loss": 13.5794,
      "step": 3237
    },
    {
      "epoch": 0.5662076502732241,
      "grad_norm": 0.8734517010676143,
      "learning_rate": 0.00019354577952926296,
      "loss": 13.4042,
      "step": 3238
    },
    {
      "epoch": 0.5663825136612022,
      "grad_norm": 1.0425364605069634,
      "learning_rate": 0.0001935407735900857,
      "loss": 13.273,
      "step": 3239
    },
    {
      "epoch": 0.5665573770491803,
      "grad_norm": 0.8738494801148361,
      "learning_rate": 0.00019353576577512872,
      "loss": 13.4373,
      "step": 3240
    },
    {
      "epoch": 0.5667322404371584,
      "grad_norm": 1.0902073821857352,
      "learning_rate": 0.00019353075608449243,
      "loss": 13.601,
      "step": 3241
    },
    {
      "epoch": 0.5669071038251366,
      "grad_norm": 1.0391079331253814,
      "learning_rate": 0.0001935257445182773,
      "loss": 13.5388,
      "step": 3242
    },
    {
      "epoch": 0.5670819672131148,
      "grad_norm": 0.9484119938529344,
      "learning_rate": 0.00019352073107658385,
      "loss": 13.3998,
      "step": 3243
    },
    {
      "epoch": 0.5672568306010929,
      "grad_norm": 1.0264191082765515,
      "learning_rate": 0.00019351571575951257,
      "loss": 13.3257,
      "step": 3244
    },
    {
      "epoch": 0.567431693989071,
      "grad_norm": 0.856859129875892,
      "learning_rate": 0.00019351069856716408,
      "loss": 13.1742,
      "step": 3245
    },
    {
      "epoch": 0.5676065573770491,
      "grad_norm": 1.1309626971876432,
      "learning_rate": 0.00019350567949963897,
      "loss": 13.6781,
      "step": 3246
    },
    {
      "epoch": 0.5677814207650274,
      "grad_norm": 0.9303448848909287,
      "learning_rate": 0.00019350065855703788,
      "loss": 13.4151,
      "step": 3247
    },
    {
      "epoch": 0.5679562841530055,
      "grad_norm": 0.9981037452680137,
      "learning_rate": 0.0001934956357394615,
      "loss": 13.3058,
      "step": 3248
    },
    {
      "epoch": 0.5681311475409836,
      "grad_norm": 0.9815161416106643,
      "learning_rate": 0.00019349061104701054,
      "loss": 13.323,
      "step": 3249
    },
    {
      "epoch": 0.5683060109289617,
      "grad_norm": 1.0599222737782164,
      "learning_rate": 0.0001934855844797858,
      "loss": 13.5334,
      "step": 3250
    },
    {
      "epoch": 0.5684808743169399,
      "grad_norm": 1.061318326301577,
      "learning_rate": 0.00019348055603788807,
      "loss": 13.5758,
      "step": 3251
    },
    {
      "epoch": 0.5686557377049181,
      "grad_norm": 0.981717280202072,
      "learning_rate": 0.0001934755257214181,
      "loss": 13.3057,
      "step": 3252
    },
    {
      "epoch": 0.5688306010928962,
      "grad_norm": 0.9964172624201081,
      "learning_rate": 0.00019347049353047688,
      "loss": 13.3553,
      "step": 3253
    },
    {
      "epoch": 0.5690054644808743,
      "grad_norm": 1.066078143032975,
      "learning_rate": 0.00019346545946516527,
      "loss": 13.3787,
      "step": 3254
    },
    {
      "epoch": 0.5691803278688524,
      "grad_norm": 1.1091588264371244,
      "learning_rate": 0.0001934604235255842,
      "loss": 13.5333,
      "step": 3255
    },
    {
      "epoch": 0.5693551912568307,
      "grad_norm": 1.0048089886273717,
      "learning_rate": 0.00019345538571183472,
      "loss": 13.6604,
      "step": 3256
    },
    {
      "epoch": 0.5695300546448088,
      "grad_norm": 1.2307898567953082,
      "learning_rate": 0.0001934503460240178,
      "loss": 13.6019,
      "step": 3257
    },
    {
      "epoch": 0.5697049180327869,
      "grad_norm": 0.8989594230269732,
      "learning_rate": 0.00019344530446223447,
      "loss": 13.544,
      "step": 3258
    },
    {
      "epoch": 0.569879781420765,
      "grad_norm": 1.0349793600807542,
      "learning_rate": 0.0001934402610265859,
      "loss": 13.5164,
      "step": 3259
    },
    {
      "epoch": 0.5700546448087431,
      "grad_norm": 0.9148634997524413,
      "learning_rate": 0.0001934352157171732,
      "loss": 13.1655,
      "step": 3260
    },
    {
      "epoch": 0.5702295081967214,
      "grad_norm": 0.967682206424733,
      "learning_rate": 0.00019343016853409754,
      "loss": 13.4619,
      "step": 3261
    },
    {
      "epoch": 0.5704043715846995,
      "grad_norm": 0.877242631235064,
      "learning_rate": 0.00019342511947746012,
      "loss": 13.4967,
      "step": 3262
    },
    {
      "epoch": 0.5705792349726776,
      "grad_norm": 1.0275728597954477,
      "learning_rate": 0.0001934200685473622,
      "loss": 13.3892,
      "step": 3263
    },
    {
      "epoch": 0.5707540983606557,
      "grad_norm": 1.0051527557056334,
      "learning_rate": 0.0001934150157439051,
      "loss": 13.3554,
      "step": 3264
    },
    {
      "epoch": 0.5709289617486338,
      "grad_norm": 0.994502933678613,
      "learning_rate": 0.0001934099610671901,
      "loss": 13.4409,
      "step": 3265
    },
    {
      "epoch": 0.5711038251366121,
      "grad_norm": 0.9116037929441663,
      "learning_rate": 0.00019340490451731854,
      "loss": 13.4789,
      "step": 3266
    },
    {
      "epoch": 0.5712786885245902,
      "grad_norm": 1.1202827050349773,
      "learning_rate": 0.0001933998460943919,
      "loss": 13.6842,
      "step": 3267
    },
    {
      "epoch": 0.5714535519125683,
      "grad_norm": 0.9711686275674912,
      "learning_rate": 0.00019339478579851155,
      "loss": 13.2265,
      "step": 3268
    },
    {
      "epoch": 0.5716284153005464,
      "grad_norm": 1.085473591969275,
      "learning_rate": 0.00019338972362977896,
      "loss": 13.4123,
      "step": 3269
    },
    {
      "epoch": 0.5718032786885245,
      "grad_norm": 0.9132320301273744,
      "learning_rate": 0.00019338465958829572,
      "loss": 13.4216,
      "step": 3270
    },
    {
      "epoch": 0.5719781420765028,
      "grad_norm": 0.9003539940804539,
      "learning_rate": 0.00019337959367416328,
      "loss": 13.3591,
      "step": 3271
    },
    {
      "epoch": 0.5721530054644809,
      "grad_norm": 0.9696090938267932,
      "learning_rate": 0.00019337452588748328,
      "loss": 13.4341,
      "step": 3272
    },
    {
      "epoch": 0.572327868852459,
      "grad_norm": 0.9186107581344224,
      "learning_rate": 0.00019336945622835735,
      "loss": 13.1742,
      "step": 3273
    },
    {
      "epoch": 0.5725027322404371,
      "grad_norm": 0.9522711336648456,
      "learning_rate": 0.00019336438469688715,
      "loss": 13.3175,
      "step": 3274
    },
    {
      "epoch": 0.5726775956284152,
      "grad_norm": 0.8700037044132861,
      "learning_rate": 0.00019335931129317436,
      "loss": 13.3864,
      "step": 3275
    },
    {
      "epoch": 0.5728524590163935,
      "grad_norm": 1.0070818252850076,
      "learning_rate": 0.00019335423601732075,
      "loss": 13.2604,
      "step": 3276
    },
    {
      "epoch": 0.5730273224043716,
      "grad_norm": 0.823206349942672,
      "learning_rate": 0.00019334915886942802,
      "loss": 13.1639,
      "step": 3277
    },
    {
      "epoch": 0.5732021857923497,
      "grad_norm": 1.0265853193071681,
      "learning_rate": 0.00019334407984959807,
      "loss": 13.5172,
      "step": 3278
    },
    {
      "epoch": 0.5733770491803278,
      "grad_norm": 0.9137115176213453,
      "learning_rate": 0.00019333899895793272,
      "loss": 13.3803,
      "step": 3279
    },
    {
      "epoch": 0.5735519125683061,
      "grad_norm": 0.9463356787986195,
      "learning_rate": 0.00019333391619453386,
      "loss": 13.1994,
      "step": 3280
    },
    {
      "epoch": 0.5737267759562842,
      "grad_norm": 0.9133960476996321,
      "learning_rate": 0.00019332883155950337,
      "loss": 13.487,
      "step": 3281
    },
    {
      "epoch": 0.5739016393442623,
      "grad_norm": 0.9518679738626422,
      "learning_rate": 0.0001933237450529433,
      "loss": 13.7381,
      "step": 3282
    },
    {
      "epoch": 0.5740765027322404,
      "grad_norm": 0.9172585369296977,
      "learning_rate": 0.00019331865667495554,
      "loss": 13.3929,
      "step": 3283
    },
    {
      "epoch": 0.5742513661202185,
      "grad_norm": 0.9341853137008599,
      "learning_rate": 0.00019331356642564225,
      "loss": 13.5307,
      "step": 3284
    },
    {
      "epoch": 0.5744262295081968,
      "grad_norm": 0.8482952407736153,
      "learning_rate": 0.00019330847430510542,
      "loss": 13.0497,
      "step": 3285
    },
    {
      "epoch": 0.5746010928961749,
      "grad_norm": 1.0343994227708289,
      "learning_rate": 0.00019330338031344716,
      "loss": 13.3406,
      "step": 3286
    },
    {
      "epoch": 0.574775956284153,
      "grad_norm": 1.1561334827588468,
      "learning_rate": 0.00019329828445076967,
      "loss": 13.2344,
      "step": 3287
    },
    {
      "epoch": 0.5749508196721311,
      "grad_norm": 0.9682162374518497,
      "learning_rate": 0.0001932931867171751,
      "loss": 13.2637,
      "step": 3288
    },
    {
      "epoch": 0.5751256830601092,
      "grad_norm": 1.0067432646357226,
      "learning_rate": 0.00019328808711276571,
      "loss": 13.6177,
      "step": 3289
    },
    {
      "epoch": 0.5753005464480875,
      "grad_norm": 0.8849862207665187,
      "learning_rate": 0.00019328298563764374,
      "loss": 13.4049,
      "step": 3290
    },
    {
      "epoch": 0.5754754098360656,
      "grad_norm": 0.9424102137247068,
      "learning_rate": 0.00019327788229191146,
      "loss": 13.4196,
      "step": 3291
    },
    {
      "epoch": 0.5756502732240437,
      "grad_norm": 1.059172720581931,
      "learning_rate": 0.00019327277707567125,
      "loss": 13.2921,
      "step": 3292
    },
    {
      "epoch": 0.5758251366120218,
      "grad_norm": 0.9947135656698878,
      "learning_rate": 0.0001932676699890255,
      "loss": 13.2545,
      "step": 3293
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.9455182664415199,
      "learning_rate": 0.00019326256103207657,
      "loss": 13.5204,
      "step": 3294
    },
    {
      "epoch": 0.5761748633879782,
      "grad_norm": 1.001382207710697,
      "learning_rate": 0.00019325745020492695,
      "loss": 13.3778,
      "step": 3295
    },
    {
      "epoch": 0.5763497267759563,
      "grad_norm": 0.9190880586277861,
      "learning_rate": 0.00019325233750767913,
      "loss": 13.2799,
      "step": 3296
    },
    {
      "epoch": 0.5765245901639344,
      "grad_norm": 1.0810654765443324,
      "learning_rate": 0.00019324722294043558,
      "loss": 13.6168,
      "step": 3297
    },
    {
      "epoch": 0.5766994535519125,
      "grad_norm": 0.963827076550258,
      "learning_rate": 0.00019324210650329895,
      "loss": 13.3694,
      "step": 3298
    },
    {
      "epoch": 0.5768743169398907,
      "grad_norm": 0.8635986449628229,
      "learning_rate": 0.00019323698819637177,
      "loss": 13.3692,
      "step": 3299
    },
    {
      "epoch": 0.5770491803278689,
      "grad_norm": 0.9311307421265559,
      "learning_rate": 0.00019323186801975667,
      "loss": 13.358,
      "step": 3300
    },
    {
      "epoch": 0.577224043715847,
      "grad_norm": 0.9202514854803981,
      "learning_rate": 0.0001932267459735564,
      "loss": 13.7414,
      "step": 3301
    },
    {
      "epoch": 0.5773989071038251,
      "grad_norm": 0.9258692890884112,
      "learning_rate": 0.00019322162205787362,
      "loss": 13.3164,
      "step": 3302
    },
    {
      "epoch": 0.5775737704918033,
      "grad_norm": 0.8502120076403898,
      "learning_rate": 0.00019321649627281108,
      "loss": 13.4569,
      "step": 3303
    },
    {
      "epoch": 0.5777486338797814,
      "grad_norm": 0.9178606596948925,
      "learning_rate": 0.0001932113686184716,
      "loss": 13.4939,
      "step": 3304
    },
    {
      "epoch": 0.5779234972677596,
      "grad_norm": 1.0424671050386514,
      "learning_rate": 0.00019320623909495796,
      "loss": 13.5939,
      "step": 3305
    },
    {
      "epoch": 0.5780983606557377,
      "grad_norm": 0.8530567617001807,
      "learning_rate": 0.00019320110770237308,
      "loss": 13.4677,
      "step": 3306
    },
    {
      "epoch": 0.5782732240437158,
      "grad_norm": 0.9064571790189699,
      "learning_rate": 0.0001931959744408198,
      "loss": 13.5938,
      "step": 3307
    },
    {
      "epoch": 0.578448087431694,
      "grad_norm": 0.9082194512137348,
      "learning_rate": 0.00019319083931040109,
      "loss": 13.446,
      "step": 3308
    },
    {
      "epoch": 0.5786229508196721,
      "grad_norm": 0.8839202785248176,
      "learning_rate": 0.0001931857023112199,
      "loss": 13.3962,
      "step": 3309
    },
    {
      "epoch": 0.5787978142076503,
      "grad_norm": 0.9004465160752998,
      "learning_rate": 0.00019318056344337926,
      "loss": 13.6385,
      "step": 3310
    },
    {
      "epoch": 0.5789726775956284,
      "grad_norm": 0.9440743312051407,
      "learning_rate": 0.00019317542270698226,
      "loss": 13.2625,
      "step": 3311
    },
    {
      "epoch": 0.5791475409836065,
      "grad_norm": 0.9355186013978259,
      "learning_rate": 0.00019317028010213191,
      "loss": 13.2929,
      "step": 3312
    },
    {
      "epoch": 0.5793224043715847,
      "grad_norm": 1.1036547717115974,
      "learning_rate": 0.0001931651356289314,
      "loss": 13.3917,
      "step": 3313
    },
    {
      "epoch": 0.5794972677595629,
      "grad_norm": 0.8926009422714287,
      "learning_rate": 0.00019315998928748384,
      "loss": 13.5381,
      "step": 3314
    },
    {
      "epoch": 0.579672131147541,
      "grad_norm": 0.9005678284774663,
      "learning_rate": 0.00019315484107789246,
      "loss": 13.5071,
      "step": 3315
    },
    {
      "epoch": 0.5798469945355191,
      "grad_norm": 0.9204434536033187,
      "learning_rate": 0.0001931496910002605,
      "loss": 13.5469,
      "step": 3316
    },
    {
      "epoch": 0.5800218579234973,
      "grad_norm": 0.9197244233583779,
      "learning_rate": 0.00019314453905469123,
      "loss": 13.4277,
      "step": 3317
    },
    {
      "epoch": 0.5801967213114754,
      "grad_norm": 1.1123254302541048,
      "learning_rate": 0.00019313938524128798,
      "loss": 13.5912,
      "step": 3318
    },
    {
      "epoch": 0.5803715846994536,
      "grad_norm": 0.8828074761822713,
      "learning_rate": 0.00019313422956015406,
      "loss": 13.3397,
      "step": 3319
    },
    {
      "epoch": 0.5805464480874317,
      "grad_norm": 0.9478227231907408,
      "learning_rate": 0.0001931290720113929,
      "loss": 13.5577,
      "step": 3320
    },
    {
      "epoch": 0.5807213114754098,
      "grad_norm": 0.9312531525924556,
      "learning_rate": 0.00019312391259510787,
      "loss": 13.5223,
      "step": 3321
    },
    {
      "epoch": 0.580896174863388,
      "grad_norm": 1.0104164847833523,
      "learning_rate": 0.00019311875131140246,
      "loss": 13.3836,
      "step": 3322
    },
    {
      "epoch": 0.5810710382513661,
      "grad_norm": 0.9491683806189339,
      "learning_rate": 0.0001931135881603802,
      "loss": 13.4312,
      "step": 3323
    },
    {
      "epoch": 0.5812459016393443,
      "grad_norm": 0.9607502287575771,
      "learning_rate": 0.00019310842314214458,
      "loss": 13.2946,
      "step": 3324
    },
    {
      "epoch": 0.5814207650273224,
      "grad_norm": 0.9384496962046968,
      "learning_rate": 0.00019310325625679923,
      "loss": 13.6419,
      "step": 3325
    },
    {
      "epoch": 0.5815956284153005,
      "grad_norm": 1.0608032854952938,
      "learning_rate": 0.0001930980875044477,
      "loss": 13.2808,
      "step": 3326
    },
    {
      "epoch": 0.5817704918032787,
      "grad_norm": 0.9512220044007649,
      "learning_rate": 0.00019309291688519365,
      "loss": 13.1205,
      "step": 3327
    },
    {
      "epoch": 0.5819453551912568,
      "grad_norm": 0.9905990347938309,
      "learning_rate": 0.00019308774439914082,
      "loss": 13.3282,
      "step": 3328
    },
    {
      "epoch": 0.582120218579235,
      "grad_norm": 0.9371748758831044,
      "learning_rate": 0.00019308257004639287,
      "loss": 13.5489,
      "step": 3329
    },
    {
      "epoch": 0.5822950819672131,
      "grad_norm": 1.08968278658363,
      "learning_rate": 0.00019307739382705362,
      "loss": 13.4473,
      "step": 3330
    },
    {
      "epoch": 0.5824699453551913,
      "grad_norm": 0.8852055126825602,
      "learning_rate": 0.0001930722157412268,
      "loss": 13.4642,
      "step": 3331
    },
    {
      "epoch": 0.5826448087431694,
      "grad_norm": 0.9656688939139875,
      "learning_rate": 0.0001930670357890163,
      "loss": 13.5768,
      "step": 3332
    },
    {
      "epoch": 0.5828196721311475,
      "grad_norm": 0.939240830927579,
      "learning_rate": 0.000193061853970526,
      "loss": 13.1663,
      "step": 3333
    },
    {
      "epoch": 0.5829945355191257,
      "grad_norm": 0.9582475085389617,
      "learning_rate": 0.0001930566702858598,
      "loss": 13.3404,
      "step": 3334
    },
    {
      "epoch": 0.5831693989071038,
      "grad_norm": 0.9212746529921539,
      "learning_rate": 0.0001930514847351216,
      "loss": 13.5068,
      "step": 3335
    },
    {
      "epoch": 0.583344262295082,
      "grad_norm": 1.0271650083397654,
      "learning_rate": 0.00019304629731841543,
      "loss": 13.2219,
      "step": 3336
    },
    {
      "epoch": 0.5835191256830601,
      "grad_norm": 1.1279985096818776,
      "learning_rate": 0.00019304110803584532,
      "loss": 13.5296,
      "step": 3337
    },
    {
      "epoch": 0.5836939890710382,
      "grad_norm": 0.943547133036531,
      "learning_rate": 0.00019303591688751533,
      "loss": 13.2475,
      "step": 3338
    },
    {
      "epoch": 0.5838688524590164,
      "grad_norm": 1.0938514217521507,
      "learning_rate": 0.00019303072387352952,
      "loss": 13.2444,
      "step": 3339
    },
    {
      "epoch": 0.5840437158469945,
      "grad_norm": 1.1072380719254407,
      "learning_rate": 0.0001930255289939921,
      "loss": 13.4724,
      "step": 3340
    },
    {
      "epoch": 0.5842185792349727,
      "grad_norm": 0.9167303771087932,
      "learning_rate": 0.00019302033224900716,
      "loss": 13.3804,
      "step": 3341
    },
    {
      "epoch": 0.5843934426229508,
      "grad_norm": 0.9654702766154888,
      "learning_rate": 0.00019301513363867895,
      "loss": 13.3353,
      "step": 3342
    },
    {
      "epoch": 0.5845683060109289,
      "grad_norm": 0.9386471324224445,
      "learning_rate": 0.0001930099331631117,
      "loss": 12.9869,
      "step": 3343
    },
    {
      "epoch": 0.5847431693989071,
      "grad_norm": 1.1044929975948932,
      "learning_rate": 0.00019300473082240976,
      "loss": 13.4681,
      "step": 3344
    },
    {
      "epoch": 0.5849180327868853,
      "grad_norm": 0.9006043794506492,
      "learning_rate": 0.0001929995266166774,
      "loss": 13.5291,
      "step": 3345
    },
    {
      "epoch": 0.5850928961748634,
      "grad_norm": 0.9921124439252578,
      "learning_rate": 0.00019299432054601897,
      "loss": 13.1867,
      "step": 3346
    },
    {
      "epoch": 0.5852677595628415,
      "grad_norm": 0.9350438137257049,
      "learning_rate": 0.00019298911261053886,
      "loss": 13.3479,
      "step": 3347
    },
    {
      "epoch": 0.5854426229508197,
      "grad_norm": 0.9150306386797483,
      "learning_rate": 0.00019298390281034155,
      "loss": 13.4642,
      "step": 3348
    },
    {
      "epoch": 0.5856174863387978,
      "grad_norm": 0.8752116725969363,
      "learning_rate": 0.0001929786911455315,
      "loss": 13.3695,
      "step": 3349
    },
    {
      "epoch": 0.585792349726776,
      "grad_norm": 0.8098269397887538,
      "learning_rate": 0.0001929734776162132,
      "loss": 13.2713,
      "step": 3350
    },
    {
      "epoch": 0.5859672131147541,
      "grad_norm": 0.9953183954110749,
      "learning_rate": 0.0001929682622224912,
      "loss": 13.5907,
      "step": 3351
    },
    {
      "epoch": 0.5861420765027322,
      "grad_norm": 0.9522677275832104,
      "learning_rate": 0.0001929630449644701,
      "loss": 13.4462,
      "step": 3352
    },
    {
      "epoch": 0.5863169398907104,
      "grad_norm": 0.8566368684903238,
      "learning_rate": 0.0001929578258422545,
      "loss": 13.3603,
      "step": 3353
    },
    {
      "epoch": 0.5864918032786886,
      "grad_norm": 1.000012066690896,
      "learning_rate": 0.0001929526048559491,
      "loss": 13.4442,
      "step": 3354
    },
    {
      "epoch": 0.5866666666666667,
      "grad_norm": 0.9869798760999616,
      "learning_rate": 0.00019294738200565856,
      "loss": 13.4125,
      "step": 3355
    },
    {
      "epoch": 0.5868415300546448,
      "grad_norm": 0.9595813969880091,
      "learning_rate": 0.00019294215729148763,
      "loss": 13.2643,
      "step": 3356
    },
    {
      "epoch": 0.5870163934426229,
      "grad_norm": 0.9547351976634597,
      "learning_rate": 0.0001929369307135411,
      "loss": 13.317,
      "step": 3357
    },
    {
      "epoch": 0.5871912568306011,
      "grad_norm": 1.0191030577465534,
      "learning_rate": 0.0001929317022719237,
      "loss": 13.4643,
      "step": 3358
    },
    {
      "epoch": 0.5873661202185793,
      "grad_norm": 0.9138671918739881,
      "learning_rate": 0.00019292647196674035,
      "loss": 13.5353,
      "step": 3359
    },
    {
      "epoch": 0.5875409836065574,
      "grad_norm": 1.0270735107018072,
      "learning_rate": 0.00019292123979809594,
      "loss": 13.4005,
      "step": 3360
    },
    {
      "epoch": 0.5877158469945355,
      "grad_norm": 1.1252192604882332,
      "learning_rate": 0.00019291600576609535,
      "loss": 13.01,
      "step": 3361
    },
    {
      "epoch": 0.5878907103825136,
      "grad_norm": 0.9830116213320326,
      "learning_rate": 0.00019291076987084353,
      "loss": 13.5608,
      "step": 3362
    },
    {
      "epoch": 0.5880655737704918,
      "grad_norm": 0.9260209884019034,
      "learning_rate": 0.00019290553211244552,
      "loss": 13.3384,
      "step": 3363
    },
    {
      "epoch": 0.58824043715847,
      "grad_norm": 1.0095382692416206,
      "learning_rate": 0.00019290029249100635,
      "loss": 13.2408,
      "step": 3364
    },
    {
      "epoch": 0.5884153005464481,
      "grad_norm": 0.9846171478642992,
      "learning_rate": 0.00019289505100663105,
      "loss": 13.5822,
      "step": 3365
    },
    {
      "epoch": 0.5885901639344262,
      "grad_norm": 0.9958568393725268,
      "learning_rate": 0.00019288980765942476,
      "loss": 13.1981,
      "step": 3366
    },
    {
      "epoch": 0.5887650273224043,
      "grad_norm": 0.9300870624877433,
      "learning_rate": 0.00019288456244949265,
      "loss": 13.3417,
      "step": 3367
    },
    {
      "epoch": 0.5889398907103826,
      "grad_norm": 0.9166805468123227,
      "learning_rate": 0.00019287931537693984,
      "loss": 13.3202,
      "step": 3368
    },
    {
      "epoch": 0.5891147540983607,
      "grad_norm": 1.0503950167571021,
      "learning_rate": 0.00019287406644187156,
      "loss": 13.4779,
      "step": 3369
    },
    {
      "epoch": 0.5892896174863388,
      "grad_norm": 1.0618166432513556,
      "learning_rate": 0.00019286881564439313,
      "loss": 13.3783,
      "step": 3370
    },
    {
      "epoch": 0.5894644808743169,
      "grad_norm": 0.9532876021348261,
      "learning_rate": 0.00019286356298460976,
      "loss": 13.2864,
      "step": 3371
    },
    {
      "epoch": 0.589639344262295,
      "grad_norm": 0.8877156146710182,
      "learning_rate": 0.00019285830846262685,
      "loss": 13.5235,
      "step": 3372
    },
    {
      "epoch": 0.5898142076502733,
      "grad_norm": 1.027481254929611,
      "learning_rate": 0.00019285305207854974,
      "loss": 13.2716,
      "step": 3373
    },
    {
      "epoch": 0.5899890710382514,
      "grad_norm": 0.9883259476006967,
      "learning_rate": 0.00019284779383248383,
      "loss": 13.467,
      "step": 3374
    },
    {
      "epoch": 0.5901639344262295,
      "grad_norm": 0.916244939796447,
      "learning_rate": 0.00019284253372453457,
      "loss": 13.2383,
      "step": 3375
    },
    {
      "epoch": 0.5903387978142076,
      "grad_norm": 0.9234199867271728,
      "learning_rate": 0.00019283727175480747,
      "loss": 13.3677,
      "step": 3376
    },
    {
      "epoch": 0.5905136612021858,
      "grad_norm": 1.0393653767078301,
      "learning_rate": 0.000192832007923408,
      "loss": 13.2273,
      "step": 3377
    },
    {
      "epoch": 0.590688524590164,
      "grad_norm": 0.9043054048553698,
      "learning_rate": 0.00019282674223044177,
      "loss": 13.3336,
      "step": 3378
    },
    {
      "epoch": 0.5908633879781421,
      "grad_norm": 1.051663722904142,
      "learning_rate": 0.0001928214746760143,
      "loss": 13.347,
      "step": 3379
    },
    {
      "epoch": 0.5910382513661202,
      "grad_norm": 0.9215639938286387,
      "learning_rate": 0.00019281620526023126,
      "loss": 13.4092,
      "step": 3380
    },
    {
      "epoch": 0.5912131147540983,
      "grad_norm": 0.9669140531907878,
      "learning_rate": 0.00019281093398319836,
      "loss": 13.4232,
      "step": 3381
    },
    {
      "epoch": 0.5913879781420766,
      "grad_norm": 0.9592383815952578,
      "learning_rate": 0.00019280566084502128,
      "loss": 13.408,
      "step": 3382
    },
    {
      "epoch": 0.5915628415300547,
      "grad_norm": 0.9219781434584031,
      "learning_rate": 0.00019280038584580572,
      "loss": 13.4984,
      "step": 3383
    },
    {
      "epoch": 0.5917377049180328,
      "grad_norm": 0.8784377856574199,
      "learning_rate": 0.0001927951089856575,
      "loss": 13.3218,
      "step": 3384
    },
    {
      "epoch": 0.5919125683060109,
      "grad_norm": 0.934985967055692,
      "learning_rate": 0.0001927898302646824,
      "loss": 13.2228,
      "step": 3385
    },
    {
      "epoch": 0.592087431693989,
      "grad_norm": 0.9487593017754544,
      "learning_rate": 0.00019278454968298632,
      "loss": 13.3431,
      "step": 3386
    },
    {
      "epoch": 0.5922622950819673,
      "grad_norm": 0.9346319921873653,
      "learning_rate": 0.0001927792672406751,
      "loss": 13.3292,
      "step": 3387
    },
    {
      "epoch": 0.5924371584699454,
      "grad_norm": 0.9735607112474648,
      "learning_rate": 0.00019277398293785476,
      "loss": 13.3425,
      "step": 3388
    },
    {
      "epoch": 0.5926120218579235,
      "grad_norm": 0.8290359432494369,
      "learning_rate": 0.00019276869677463116,
      "loss": 13.3444,
      "step": 3389
    },
    {
      "epoch": 0.5927868852459016,
      "grad_norm": 0.9095604663146332,
      "learning_rate": 0.0001927634087511104,
      "loss": 13.2358,
      "step": 3390
    },
    {
      "epoch": 0.5929617486338797,
      "grad_norm": 0.8778952770593683,
      "learning_rate": 0.00019275811886739844,
      "loss": 13.1715,
      "step": 3391
    },
    {
      "epoch": 0.593136612021858,
      "grad_norm": 0.909764765261126,
      "learning_rate": 0.00019275282712360138,
      "loss": 12.9322,
      "step": 3392
    },
    {
      "epoch": 0.5933114754098361,
      "grad_norm": 1.0575249662975337,
      "learning_rate": 0.00019274753351982538,
      "loss": 13.3634,
      "step": 3393
    },
    {
      "epoch": 0.5934863387978142,
      "grad_norm": 1.009265088795415,
      "learning_rate": 0.00019274223805617653,
      "loss": 13.6936,
      "step": 3394
    },
    {
      "epoch": 0.5936612021857923,
      "grad_norm": 0.9023552633296991,
      "learning_rate": 0.00019273694073276108,
      "loss": 13.6111,
      "step": 3395
    },
    {
      "epoch": 0.5938360655737704,
      "grad_norm": 0.9569359700687948,
      "learning_rate": 0.00019273164154968522,
      "loss": 13.4658,
      "step": 3396
    },
    {
      "epoch": 0.5940109289617487,
      "grad_norm": 0.9254511208319233,
      "learning_rate": 0.00019272634050705522,
      "loss": 13.5408,
      "step": 3397
    },
    {
      "epoch": 0.5941857923497268,
      "grad_norm": 0.8722589077081844,
      "learning_rate": 0.00019272103760497737,
      "loss": 13.2474,
      "step": 3398
    },
    {
      "epoch": 0.5943606557377049,
      "grad_norm": 0.850503812483353,
      "learning_rate": 0.00019271573284355804,
      "loss": 13.3851,
      "step": 3399
    },
    {
      "epoch": 0.594535519125683,
      "grad_norm": 0.9856901631195033,
      "learning_rate": 0.00019271042622290359,
      "loss": 13.5997,
      "step": 3400
    },
    {
      "epoch": 0.5947103825136612,
      "grad_norm": 0.8870817965946608,
      "learning_rate": 0.00019270511774312045,
      "loss": 12.9278,
      "step": 3401
    },
    {
      "epoch": 0.5948852459016394,
      "grad_norm": 0.9069728329231082,
      "learning_rate": 0.00019269980740431504,
      "loss": 13.3202,
      "step": 3402
    },
    {
      "epoch": 0.5950601092896175,
      "grad_norm": 0.9484421900659736,
      "learning_rate": 0.0001926944952065939,
      "loss": 13.0972,
      "step": 3403
    },
    {
      "epoch": 0.5952349726775956,
      "grad_norm": 0.9226792431225694,
      "learning_rate": 0.00019268918115006347,
      "loss": 13.2712,
      "step": 3404
    },
    {
      "epoch": 0.5954098360655737,
      "grad_norm": 0.9657906901908947,
      "learning_rate": 0.00019268386523483037,
      "loss": 13.3501,
      "step": 3405
    },
    {
      "epoch": 0.5955846994535519,
      "grad_norm": 0.9902416485983488,
      "learning_rate": 0.00019267854746100123,
      "loss": 13.3454,
      "step": 3406
    },
    {
      "epoch": 0.5957595628415301,
      "grad_norm": 0.8257605517353969,
      "learning_rate": 0.00019267322782868263,
      "loss": 13.2582,
      "step": 3407
    },
    {
      "epoch": 0.5959344262295082,
      "grad_norm": 0.9865666913384424,
      "learning_rate": 0.00019266790633798127,
      "loss": 13.3983,
      "step": 3408
    },
    {
      "epoch": 0.5961092896174863,
      "grad_norm": 0.8376730656683831,
      "learning_rate": 0.00019266258298900385,
      "loss": 13.5645,
      "step": 3409
    },
    {
      "epoch": 0.5962841530054644,
      "grad_norm": 0.9470514646972285,
      "learning_rate": 0.00019265725778185713,
      "loss": 12.966,
      "step": 3410
    },
    {
      "epoch": 0.5964590163934427,
      "grad_norm": 0.9628811263555448,
      "learning_rate": 0.00019265193071664794,
      "loss": 13.3598,
      "step": 3411
    },
    {
      "epoch": 0.5966338797814208,
      "grad_norm": 1.0258111608770022,
      "learning_rate": 0.000192646601793483,
      "loss": 13.1761,
      "step": 3412
    },
    {
      "epoch": 0.5968087431693989,
      "grad_norm": 1.1052198912251197,
      "learning_rate": 0.0001926412710124693,
      "loss": 13.6213,
      "step": 3413
    },
    {
      "epoch": 0.596983606557377,
      "grad_norm": 0.8778759347679054,
      "learning_rate": 0.0001926359383737136,
      "loss": 13.2268,
      "step": 3414
    },
    {
      "epoch": 0.5971584699453552,
      "grad_norm": 0.9738985435359917,
      "learning_rate": 0.00019263060387732293,
      "loss": 12.965,
      "step": 3415
    },
    {
      "epoch": 0.5973333333333334,
      "grad_norm": 0.9300848331172848,
      "learning_rate": 0.00019262526752340426,
      "loss": 13.4193,
      "step": 3416
    },
    {
      "epoch": 0.5975081967213115,
      "grad_norm": 0.9000104182152654,
      "learning_rate": 0.00019261992931206455,
      "loss": 13.4647,
      "step": 3417
    },
    {
      "epoch": 0.5976830601092896,
      "grad_norm": 0.8973350618325717,
      "learning_rate": 0.00019261458924341092,
      "loss": 13.1739,
      "step": 3418
    },
    {
      "epoch": 0.5978579234972677,
      "grad_norm": 0.879967705012116,
      "learning_rate": 0.0001926092473175504,
      "loss": 13.1172,
      "step": 3419
    },
    {
      "epoch": 0.5980327868852459,
      "grad_norm": 1.151769935887163,
      "learning_rate": 0.00019260390353459007,
      "loss": 13.545,
      "step": 3420
    },
    {
      "epoch": 0.5982076502732241,
      "grad_norm": 1.2822535732940632,
      "learning_rate": 0.0001925985578946372,
      "loss": 13.4208,
      "step": 3421
    },
    {
      "epoch": 0.5983825136612022,
      "grad_norm": 0.8980922335069762,
      "learning_rate": 0.0001925932103977989,
      "loss": 13.19,
      "step": 3422
    },
    {
      "epoch": 0.5985573770491803,
      "grad_norm": 0.9429679480300031,
      "learning_rate": 0.00019258786104418244,
      "loss": 13.2368,
      "step": 3423
    },
    {
      "epoch": 0.5987322404371584,
      "grad_norm": 0.8567121939015205,
      "learning_rate": 0.0001925825098338951,
      "loss": 13.3721,
      "step": 3424
    },
    {
      "epoch": 0.5989071038251366,
      "grad_norm": 0.9490376707147286,
      "learning_rate": 0.00019257715676704415,
      "loss": 13.4963,
      "step": 3425
    },
    {
      "epoch": 0.5990819672131148,
      "grad_norm": 0.9212088107510737,
      "learning_rate": 0.00019257180184373697,
      "loss": 13.446,
      "step": 3426
    },
    {
      "epoch": 0.5992568306010929,
      "grad_norm": 1.0019547774566084,
      "learning_rate": 0.0001925664450640809,
      "loss": 13.4103,
      "step": 3427
    },
    {
      "epoch": 0.599431693989071,
      "grad_norm": 1.0358862224650416,
      "learning_rate": 0.00019256108642818346,
      "loss": 13.4385,
      "step": 3428
    },
    {
      "epoch": 0.5996065573770492,
      "grad_norm": 0.9831693552064481,
      "learning_rate": 0.00019255572593615197,
      "loss": 13.5003,
      "step": 3429
    },
    {
      "epoch": 0.5997814207650273,
      "grad_norm": 0.9274091758485385,
      "learning_rate": 0.000192550363588094,
      "loss": 13.3043,
      "step": 3430
    },
    {
      "epoch": 0.5999562841530055,
      "grad_norm": 0.9414743140964301,
      "learning_rate": 0.0001925449993841171,
      "loss": 13.5803,
      "step": 3431
    },
    {
      "epoch": 0.6001311475409836,
      "grad_norm": 0.9270627598938721,
      "learning_rate": 0.00019253963332432878,
      "loss": 13.5469,
      "step": 3432
    },
    {
      "epoch": 0.6003060109289617,
      "grad_norm": 1.1502414909769576,
      "learning_rate": 0.0001925342654088367,
      "loss": 13.3307,
      "step": 3433
    },
    {
      "epoch": 0.6004808743169399,
      "grad_norm": 0.8153127181293015,
      "learning_rate": 0.0001925288956377485,
      "loss": 13.4171,
      "step": 3434
    },
    {
      "epoch": 0.600655737704918,
      "grad_norm": 0.9356811445879311,
      "learning_rate": 0.00019252352401117183,
      "loss": 13.144,
      "step": 3435
    },
    {
      "epoch": 0.6008306010928962,
      "grad_norm": 0.9670001553915507,
      "learning_rate": 0.0001925181505292144,
      "loss": 12.8645,
      "step": 3436
    },
    {
      "epoch": 0.6010054644808743,
      "grad_norm": 0.9429467216498449,
      "learning_rate": 0.000192512775191984,
      "loss": 13.4298,
      "step": 3437
    },
    {
      "epoch": 0.6011803278688524,
      "grad_norm": 0.9957876178385731,
      "learning_rate": 0.00019250739799958838,
      "loss": 13.2842,
      "step": 3438
    },
    {
      "epoch": 0.6013551912568306,
      "grad_norm": 0.9683644930761984,
      "learning_rate": 0.00019250201895213543,
      "loss": 13.4397,
      "step": 3439
    },
    {
      "epoch": 0.6015300546448088,
      "grad_norm": 0.9632185265608649,
      "learning_rate": 0.00019249663804973296,
      "loss": 13.456,
      "step": 3440
    },
    {
      "epoch": 0.6017049180327869,
      "grad_norm": 0.954271246926664,
      "learning_rate": 0.0001924912552924889,
      "loss": 13.5078,
      "step": 3441
    },
    {
      "epoch": 0.601879781420765,
      "grad_norm": 0.9639639972291945,
      "learning_rate": 0.0001924858706805112,
      "loss": 13.4988,
      "step": 3442
    },
    {
      "epoch": 0.6020546448087432,
      "grad_norm": 0.8870730722727116,
      "learning_rate": 0.00019248048421390782,
      "loss": 13.4816,
      "step": 3443
    },
    {
      "epoch": 0.6022295081967213,
      "grad_norm": 1.0312161191402973,
      "learning_rate": 0.00019247509589278674,
      "loss": 13.5833,
      "step": 3444
    },
    {
      "epoch": 0.6024043715846995,
      "grad_norm": 0.965209868771187,
      "learning_rate": 0.00019246970571725608,
      "loss": 13.3722,
      "step": 3445
    },
    {
      "epoch": 0.6025792349726776,
      "grad_norm": 0.9852361976508546,
      "learning_rate": 0.0001924643136874239,
      "loss": 13.4742,
      "step": 3446
    },
    {
      "epoch": 0.6027540983606557,
      "grad_norm": 1.0231048566194505,
      "learning_rate": 0.00019245891980339833,
      "loss": 13.2411,
      "step": 3447
    },
    {
      "epoch": 0.6029289617486339,
      "grad_norm": 0.9458199790406236,
      "learning_rate": 0.00019245352406528754,
      "loss": 13.4552,
      "step": 3448
    },
    {
      "epoch": 0.603103825136612,
      "grad_norm": 1.1784658647552189,
      "learning_rate": 0.0001924481264731997,
      "loss": 13.5461,
      "step": 3449
    },
    {
      "epoch": 0.6032786885245902,
      "grad_norm": 0.9937829655374179,
      "learning_rate": 0.0001924427270272431,
      "loss": 13.3228,
      "step": 3450
    },
    {
      "epoch": 0.6034535519125683,
      "grad_norm": 1.0363841566542669,
      "learning_rate": 0.00019243732572752598,
      "loss": 13.3047,
      "step": 3451
    },
    {
      "epoch": 0.6036284153005465,
      "grad_norm": 0.949973030547785,
      "learning_rate": 0.00019243192257415665,
      "loss": 13.3562,
      "step": 3452
    },
    {
      "epoch": 0.6038032786885246,
      "grad_norm": 1.0143830799437255,
      "learning_rate": 0.00019242651756724345,
      "loss": 13.2478,
      "step": 3453
    },
    {
      "epoch": 0.6039781420765027,
      "grad_norm": 0.9624575524582181,
      "learning_rate": 0.00019242111070689482,
      "loss": 13.4015,
      "step": 3454
    },
    {
      "epoch": 0.6041530054644809,
      "grad_norm": 1.0131738643504842,
      "learning_rate": 0.00019241570199321915,
      "loss": 13.4909,
      "step": 3455
    },
    {
      "epoch": 0.604327868852459,
      "grad_norm": 0.8722018329400817,
      "learning_rate": 0.00019241029142632487,
      "loss": 12.8638,
      "step": 3456
    },
    {
      "epoch": 0.6045027322404372,
      "grad_norm": 0.8587920699735142,
      "learning_rate": 0.00019240487900632053,
      "loss": 13.2534,
      "step": 3457
    },
    {
      "epoch": 0.6046775956284153,
      "grad_norm": 0.9264490001616246,
      "learning_rate": 0.00019239946473331466,
      "loss": 13.4692,
      "step": 3458
    },
    {
      "epoch": 0.6048524590163934,
      "grad_norm": 0.9171091516013077,
      "learning_rate": 0.00019239404860741578,
      "loss": 13.5191,
      "step": 3459
    },
    {
      "epoch": 0.6050273224043716,
      "grad_norm": 0.9721579468707222,
      "learning_rate": 0.00019238863062873257,
      "loss": 13.6025,
      "step": 3460
    },
    {
      "epoch": 0.6052021857923497,
      "grad_norm": 0.9904162521932679,
      "learning_rate": 0.00019238321079737366,
      "loss": 13.4262,
      "step": 3461
    },
    {
      "epoch": 0.6053770491803279,
      "grad_norm": 0.9114773016197839,
      "learning_rate": 0.00019237778911344769,
      "loss": 13.44,
      "step": 3462
    },
    {
      "epoch": 0.605551912568306,
      "grad_norm": 0.9341922624874072,
      "learning_rate": 0.00019237236557706342,
      "loss": 13.3191,
      "step": 3463
    },
    {
      "epoch": 0.6057267759562841,
      "grad_norm": 0.9554637274122989,
      "learning_rate": 0.00019236694018832958,
      "loss": 13.1431,
      "step": 3464
    },
    {
      "epoch": 0.6059016393442623,
      "grad_norm": 0.982435887701405,
      "learning_rate": 0.000192361512947355,
      "loss": 13.4178,
      "step": 3465
    },
    {
      "epoch": 0.6060765027322405,
      "grad_norm": 0.9425301589342484,
      "learning_rate": 0.0001923560838542485,
      "loss": 13.4305,
      "step": 3466
    },
    {
      "epoch": 0.6062513661202186,
      "grad_norm": 1.1682570942336061,
      "learning_rate": 0.00019235065290911898,
      "loss": 13.2422,
      "step": 3467
    },
    {
      "epoch": 0.6064262295081967,
      "grad_norm": 1.1411194542337153,
      "learning_rate": 0.00019234522011207528,
      "loss": 13.4044,
      "step": 3468
    },
    {
      "epoch": 0.6066010928961748,
      "grad_norm": 1.0122258978176049,
      "learning_rate": 0.00019233978546322636,
      "loss": 13.5125,
      "step": 3469
    },
    {
      "epoch": 0.606775956284153,
      "grad_norm": 0.9468718456169798,
      "learning_rate": 0.00019233434896268127,
      "loss": 13.3893,
      "step": 3470
    },
    {
      "epoch": 0.6069508196721312,
      "grad_norm": 1.0160910086558146,
      "learning_rate": 0.00019232891061054895,
      "loss": 13.2328,
      "step": 3471
    },
    {
      "epoch": 0.6071256830601093,
      "grad_norm": 1.0262612719315338,
      "learning_rate": 0.0001923234704069385,
      "loss": 13.5596,
      "step": 3472
    },
    {
      "epoch": 0.6073005464480874,
      "grad_norm": 0.8900241163775249,
      "learning_rate": 0.000192318028351959,
      "loss": 13.0845,
      "step": 3473
    },
    {
      "epoch": 0.6074754098360656,
      "grad_norm": 0.9234238890825686,
      "learning_rate": 0.00019231258444571953,
      "loss": 12.9408,
      "step": 3474
    },
    {
      "epoch": 0.6076502732240437,
      "grad_norm": 0.8686872437342148,
      "learning_rate": 0.00019230713868832936,
      "loss": 13.3333,
      "step": 3475
    },
    {
      "epoch": 0.6078251366120219,
      "grad_norm": 1.0535777200074716,
      "learning_rate": 0.00019230169107989763,
      "loss": 13.2924,
      "step": 3476
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.9696768707595104,
      "learning_rate": 0.00019229624162053356,
      "loss": 13.5117,
      "step": 3477
    },
    {
      "epoch": 0.6081748633879781,
      "grad_norm": 0.9292961203166727,
      "learning_rate": 0.00019229079031034648,
      "loss": 13.3014,
      "step": 3478
    },
    {
      "epoch": 0.6083497267759563,
      "grad_norm": 1.0669291774966039,
      "learning_rate": 0.0001922853371494457,
      "loss": 13.4367,
      "step": 3479
    },
    {
      "epoch": 0.6085245901639345,
      "grad_norm": 0.9101984312036492,
      "learning_rate": 0.00019227988213794053,
      "loss": 13.2735,
      "step": 3480
    },
    {
      "epoch": 0.6086994535519126,
      "grad_norm": 0.9799456475569815,
      "learning_rate": 0.0001922744252759404,
      "loss": 13.1263,
      "step": 3481
    },
    {
      "epoch": 0.6088743169398907,
      "grad_norm": 1.0391230328884489,
      "learning_rate": 0.00019226896656355472,
      "loss": 13.2455,
      "step": 3482
    },
    {
      "epoch": 0.6090491803278688,
      "grad_norm": 0.9133871646044726,
      "learning_rate": 0.00019226350600089296,
      "loss": 13.1869,
      "step": 3483
    },
    {
      "epoch": 0.609224043715847,
      "grad_norm": 1.0441781640973407,
      "learning_rate": 0.00019225804358806464,
      "loss": 13.4717,
      "step": 3484
    },
    {
      "epoch": 0.6093989071038252,
      "grad_norm": 1.018411612260401,
      "learning_rate": 0.00019225257932517923,
      "loss": 13.4828,
      "step": 3485
    },
    {
      "epoch": 0.6095737704918033,
      "grad_norm": 1.0349349652326212,
      "learning_rate": 0.0001922471132123464,
      "loss": 13.5391,
      "step": 3486
    },
    {
      "epoch": 0.6097486338797814,
      "grad_norm": 1.6352751075101728,
      "learning_rate": 0.00019224164524967566,
      "loss": 13.4316,
      "step": 3487
    },
    {
      "epoch": 0.6099234972677595,
      "grad_norm": 0.8934608367929646,
      "learning_rate": 0.00019223617543727675,
      "loss": 13.1976,
      "step": 3488
    },
    {
      "epoch": 0.6100983606557377,
      "grad_norm": 1.144187242199568,
      "learning_rate": 0.0001922307037752593,
      "loss": 13.5986,
      "step": 3489
    },
    {
      "epoch": 0.6102732240437159,
      "grad_norm": 1.0089801473860975,
      "learning_rate": 0.0001922252302637331,
      "loss": 13.2874,
      "step": 3490
    },
    {
      "epoch": 0.610448087431694,
      "grad_norm": 1.0258869328996223,
      "learning_rate": 0.00019221975490280784,
      "loss": 13.4225,
      "step": 3491
    },
    {
      "epoch": 0.6106229508196721,
      "grad_norm": 1.0269534160367035,
      "learning_rate": 0.00019221427769259333,
      "loss": 13.1161,
      "step": 3492
    },
    {
      "epoch": 0.6107978142076502,
      "grad_norm": 0.9830630819799887,
      "learning_rate": 0.00019220879863319942,
      "loss": 13.2553,
      "step": 3493
    },
    {
      "epoch": 0.6109726775956285,
      "grad_norm": 1.1268731475530123,
      "learning_rate": 0.00019220331772473598,
      "loss": 13.3678,
      "step": 3494
    },
    {
      "epoch": 0.6111475409836066,
      "grad_norm": 0.8731995904573239,
      "learning_rate": 0.00019219783496731292,
      "loss": 12.9479,
      "step": 3495
    },
    {
      "epoch": 0.6113224043715847,
      "grad_norm": 1.1144284639267614,
      "learning_rate": 0.0001921923503610402,
      "loss": 13.203,
      "step": 3496
    },
    {
      "epoch": 0.6114972677595628,
      "grad_norm": 0.8742218203277479,
      "learning_rate": 0.00019218686390602775,
      "loss": 13.0538,
      "step": 3497
    },
    {
      "epoch": 0.6116721311475409,
      "grad_norm": 1.1854248802886953,
      "learning_rate": 0.00019218137560238566,
      "loss": 13.5565,
      "step": 3498
    },
    {
      "epoch": 0.6118469945355192,
      "grad_norm": 0.9190514481579467,
      "learning_rate": 0.00019217588545022393,
      "loss": 13.2503,
      "step": 3499
    },
    {
      "epoch": 0.6120218579234973,
      "grad_norm": 1.0194125041127875,
      "learning_rate": 0.0001921703934496527,
      "loss": 13.505,
      "step": 3500
    },
    {
      "epoch": 0.6121967213114754,
      "grad_norm": 0.9557807412580962,
      "learning_rate": 0.0001921648996007821,
      "loss": 13.6055,
      "step": 3501
    },
    {
      "epoch": 0.6123715846994535,
      "grad_norm": 0.9281307603164918,
      "learning_rate": 0.00019215940390372222,
      "loss": 13.1982,
      "step": 3502
    },
    {
      "epoch": 0.6125464480874316,
      "grad_norm": 0.9207248035589718,
      "learning_rate": 0.00019215390635858337,
      "loss": 13.4119,
      "step": 3503
    },
    {
      "epoch": 0.6127213114754099,
      "grad_norm": 1.0515455288358435,
      "learning_rate": 0.00019214840696547575,
      "loss": 13.1625,
      "step": 3504
    },
    {
      "epoch": 0.612896174863388,
      "grad_norm": 1.0666082078969317,
      "learning_rate": 0.00019214290572450962,
      "loss": 13.5784,
      "step": 3505
    },
    {
      "epoch": 0.6130710382513661,
      "grad_norm": 0.9327090848011365,
      "learning_rate": 0.0001921374026357953,
      "loss": 13.3554,
      "step": 3506
    },
    {
      "epoch": 0.6132459016393442,
      "grad_norm": 1.0776258306136326,
      "learning_rate": 0.0001921318976994432,
      "loss": 13.479,
      "step": 3507
    },
    {
      "epoch": 0.6134207650273225,
      "grad_norm": 1.0756726533846817,
      "learning_rate": 0.00019212639091556363,
      "loss": 13.397,
      "step": 3508
    },
    {
      "epoch": 0.6135956284153006,
      "grad_norm": 1.0744970053250242,
      "learning_rate": 0.0001921208822842671,
      "loss": 13.1072,
      "step": 3509
    },
    {
      "epoch": 0.6137704918032787,
      "grad_norm": 0.8855009324895006,
      "learning_rate": 0.000192115371805664,
      "loss": 13.632,
      "step": 3510
    },
    {
      "epoch": 0.6139453551912568,
      "grad_norm": 0.9378640975253942,
      "learning_rate": 0.0001921098594798649,
      "loss": 13.2757,
      "step": 3511
    },
    {
      "epoch": 0.6141202185792349,
      "grad_norm": 0.9855267907480656,
      "learning_rate": 0.00019210434530698024,
      "loss": 13.134,
      "step": 3512
    },
    {
      "epoch": 0.6142950819672132,
      "grad_norm": 0.9196458360354567,
      "learning_rate": 0.0001920988292871207,
      "loss": 13.3083,
      "step": 3513
    },
    {
      "epoch": 0.6144699453551913,
      "grad_norm": 0.9054541308859502,
      "learning_rate": 0.00019209331142039683,
      "loss": 13.3021,
      "step": 3514
    },
    {
      "epoch": 0.6146448087431694,
      "grad_norm": 0.9240235551764099,
      "learning_rate": 0.00019208779170691934,
      "loss": 13.3518,
      "step": 3515
    },
    {
      "epoch": 0.6148196721311475,
      "grad_norm": 0.9510682003630361,
      "learning_rate": 0.00019208227014679886,
      "loss": 13.1179,
      "step": 3516
    },
    {
      "epoch": 0.6149945355191256,
      "grad_norm": 0.9000843995935793,
      "learning_rate": 0.00019207674674014613,
      "loss": 13.2041,
      "step": 3517
    },
    {
      "epoch": 0.6151693989071039,
      "grad_norm": 0.877769745296891,
      "learning_rate": 0.0001920712214870719,
      "loss": 13.2522,
      "step": 3518
    },
    {
      "epoch": 0.615344262295082,
      "grad_norm": 0.9454304840206179,
      "learning_rate": 0.000192065694387687,
      "loss": 13.4857,
      "step": 3519
    },
    {
      "epoch": 0.6155191256830601,
      "grad_norm": 0.8895897076972947,
      "learning_rate": 0.00019206016544210224,
      "loss": 13.1096,
      "step": 3520
    },
    {
      "epoch": 0.6156939890710382,
      "grad_norm": 0.9472496383773201,
      "learning_rate": 0.00019205463465042853,
      "loss": 13.4485,
      "step": 3521
    },
    {
      "epoch": 0.6158688524590163,
      "grad_norm": 0.9161868709037898,
      "learning_rate": 0.00019204910201277672,
      "loss": 13.3858,
      "step": 3522
    },
    {
      "epoch": 0.6160437158469946,
      "grad_norm": 0.9002651376648938,
      "learning_rate": 0.00019204356752925778,
      "loss": 13.1942,
      "step": 3523
    },
    {
      "epoch": 0.6162185792349727,
      "grad_norm": 0.991553643375817,
      "learning_rate": 0.0001920380311999827,
      "loss": 13.5542,
      "step": 3524
    },
    {
      "epoch": 0.6163934426229508,
      "grad_norm": 1.0278446023344916,
      "learning_rate": 0.00019203249302506256,
      "loss": 13.1726,
      "step": 3525
    },
    {
      "epoch": 0.6165683060109289,
      "grad_norm": 0.9124537262352423,
      "learning_rate": 0.0001920269530046083,
      "loss": 13.2042,
      "step": 3526
    },
    {
      "epoch": 0.616743169398907,
      "grad_norm": 1.057490157562251,
      "learning_rate": 0.0001920214111387311,
      "loss": 13.0208,
      "step": 3527
    },
    {
      "epoch": 0.6169180327868853,
      "grad_norm": 1.0066702655422388,
      "learning_rate": 0.00019201586742754204,
      "loss": 13.1113,
      "step": 3528
    },
    {
      "epoch": 0.6170928961748634,
      "grad_norm": 0.8523094919705652,
      "learning_rate": 0.00019201032187115234,
      "loss": 13.381,
      "step": 3529
    },
    {
      "epoch": 0.6172677595628415,
      "grad_norm": 1.11990560382511,
      "learning_rate": 0.00019200477446967314,
      "loss": 13.4887,
      "step": 3530
    },
    {
      "epoch": 0.6174426229508196,
      "grad_norm": 1.0700455740414807,
      "learning_rate": 0.00019199922522321574,
      "loss": 13.3411,
      "step": 3531
    },
    {
      "epoch": 0.6176174863387978,
      "grad_norm": 0.9032482693023566,
      "learning_rate": 0.00019199367413189143,
      "loss": 13.6213,
      "step": 3532
    },
    {
      "epoch": 0.617792349726776,
      "grad_norm": 1.012294508316553,
      "learning_rate": 0.00019198812119581145,
      "loss": 13.4383,
      "step": 3533
    },
    {
      "epoch": 0.6179672131147541,
      "grad_norm": 1.0223395800696216,
      "learning_rate": 0.00019198256641508722,
      "loss": 13.3845,
      "step": 3534
    },
    {
      "epoch": 0.6181420765027322,
      "grad_norm": 0.9216105781909367,
      "learning_rate": 0.00019197700978983007,
      "loss": 13.3376,
      "step": 3535
    },
    {
      "epoch": 0.6183169398907103,
      "grad_norm": 0.9369490633041329,
      "learning_rate": 0.00019197145132015153,
      "loss": 13.1073,
      "step": 3536
    },
    {
      "epoch": 0.6184918032786886,
      "grad_norm": 1.003904261201751,
      "learning_rate": 0.00019196589100616297,
      "loss": 13.2864,
      "step": 3537
    },
    {
      "epoch": 0.6186666666666667,
      "grad_norm": 0.8598018137300193,
      "learning_rate": 0.00019196032884797592,
      "loss": 13.384,
      "step": 3538
    },
    {
      "epoch": 0.6188415300546448,
      "grad_norm": 0.9039323045891883,
      "learning_rate": 0.00019195476484570195,
      "loss": 13.3696,
      "step": 3539
    },
    {
      "epoch": 0.6190163934426229,
      "grad_norm": 1.163033856058523,
      "learning_rate": 0.0001919491989994526,
      "loss": 13.5459,
      "step": 3540
    },
    {
      "epoch": 0.6191912568306011,
      "grad_norm": 1.0162840937268323,
      "learning_rate": 0.00019194363130933944,
      "loss": 13.0905,
      "step": 3541
    },
    {
      "epoch": 0.6193661202185793,
      "grad_norm": 1.0421710354062281,
      "learning_rate": 0.0001919380617754742,
      "loss": 13.2384,
      "step": 3542
    },
    {
      "epoch": 0.6195409836065574,
      "grad_norm": 1.2965004632364772,
      "learning_rate": 0.00019193249039796854,
      "loss": 13.6779,
      "step": 3543
    },
    {
      "epoch": 0.6197158469945355,
      "grad_norm": 0.8886013469185051,
      "learning_rate": 0.00019192691717693418,
      "loss": 13.406,
      "step": 3544
    },
    {
      "epoch": 0.6198907103825136,
      "grad_norm": 1.0781533032062622,
      "learning_rate": 0.00019192134211248288,
      "loss": 13.3894,
      "step": 3545
    },
    {
      "epoch": 0.6200655737704918,
      "grad_norm": 1.0442637285833054,
      "learning_rate": 0.00019191576520472643,
      "loss": 13.3198,
      "step": 3546
    },
    {
      "epoch": 0.62024043715847,
      "grad_norm": 0.9342182694141637,
      "learning_rate": 0.00019191018645377668,
      "loss": 13.394,
      "step": 3547
    },
    {
      "epoch": 0.6204153005464481,
      "grad_norm": 1.0415705048483057,
      "learning_rate": 0.00019190460585974547,
      "loss": 13.4045,
      "step": 3548
    },
    {
      "epoch": 0.6205901639344262,
      "grad_norm": 1.0337049609167699,
      "learning_rate": 0.00019189902342274471,
      "loss": 13.2152,
      "step": 3549
    },
    {
      "epoch": 0.6207650273224044,
      "grad_norm": 0.9903588240701532,
      "learning_rate": 0.0001918934391428864,
      "loss": 13.1816,
      "step": 3550
    },
    {
      "epoch": 0.6209398907103825,
      "grad_norm": 0.9914862275632412,
      "learning_rate": 0.00019188785302028246,
      "loss": 13.3929,
      "step": 3551
    },
    {
      "epoch": 0.6211147540983607,
      "grad_norm": 0.9546325139967877,
      "learning_rate": 0.00019188226505504496,
      "loss": 13.4179,
      "step": 3552
    },
    {
      "epoch": 0.6212896174863388,
      "grad_norm": 0.9538564516354288,
      "learning_rate": 0.00019187667524728593,
      "loss": 13.0439,
      "step": 3553
    },
    {
      "epoch": 0.6214644808743169,
      "grad_norm": 0.985284839735479,
      "learning_rate": 0.00019187108359711746,
      "loss": 13.1881,
      "step": 3554
    },
    {
      "epoch": 0.6216393442622951,
      "grad_norm": 0.885691235787334,
      "learning_rate": 0.00019186549010465168,
      "loss": 13.4157,
      "step": 3555
    },
    {
      "epoch": 0.6218142076502732,
      "grad_norm": 0.92214016429088,
      "learning_rate": 0.00019185989477000072,
      "loss": 13.3028,
      "step": 3556
    },
    {
      "epoch": 0.6219890710382514,
      "grad_norm": 1.0054528127774334,
      "learning_rate": 0.00019185429759327685,
      "loss": 13.4359,
      "step": 3557
    },
    {
      "epoch": 0.6221639344262295,
      "grad_norm": 0.9023359822038655,
      "learning_rate": 0.00019184869857459232,
      "loss": 13.4096,
      "step": 3558
    },
    {
      "epoch": 0.6223387978142076,
      "grad_norm": 0.9200076019901049,
      "learning_rate": 0.00019184309771405934,
      "loss": 13.2701,
      "step": 3559
    },
    {
      "epoch": 0.6225136612021858,
      "grad_norm": 0.989131573034619,
      "learning_rate": 0.00019183749501179025,
      "loss": 13.4283,
      "step": 3560
    },
    {
      "epoch": 0.6226885245901639,
      "grad_norm": 0.9301163757789965,
      "learning_rate": 0.00019183189046789742,
      "loss": 13.0254,
      "step": 3561
    },
    {
      "epoch": 0.6228633879781421,
      "grad_norm": 1.010359166296698,
      "learning_rate": 0.00019182628408249318,
      "loss": 13.2111,
      "step": 3562
    },
    {
      "epoch": 0.6230382513661202,
      "grad_norm": 1.0563020603319102,
      "learning_rate": 0.00019182067585569006,
      "loss": 13.3451,
      "step": 3563
    },
    {
      "epoch": 0.6232131147540984,
      "grad_norm": 0.9270242837443403,
      "learning_rate": 0.00019181506578760046,
      "loss": 13.3658,
      "step": 3564
    },
    {
      "epoch": 0.6233879781420765,
      "grad_norm": 1.0338329459122109,
      "learning_rate": 0.00019180945387833686,
      "loss": 13.4594,
      "step": 3565
    },
    {
      "epoch": 0.6235628415300546,
      "grad_norm": 1.0397602159565351,
      "learning_rate": 0.00019180384012801183,
      "loss": 13.3571,
      "step": 3566
    },
    {
      "epoch": 0.6237377049180328,
      "grad_norm": 0.9557166644712396,
      "learning_rate": 0.0001917982245367379,
      "loss": 13.2797,
      "step": 3567
    },
    {
      "epoch": 0.6239125683060109,
      "grad_norm": 1.0334796940421536,
      "learning_rate": 0.00019179260710462773,
      "loss": 13.254,
      "step": 3568
    },
    {
      "epoch": 0.6240874316939891,
      "grad_norm": 0.8725282792660929,
      "learning_rate": 0.00019178698783179395,
      "loss": 13.0327,
      "step": 3569
    },
    {
      "epoch": 0.6242622950819672,
      "grad_norm": 1.1353044303174384,
      "learning_rate": 0.00019178136671834924,
      "loss": 13.2635,
      "step": 3570
    },
    {
      "epoch": 0.6244371584699454,
      "grad_norm": 0.9496006550974703,
      "learning_rate": 0.00019177574376440633,
      "loss": 13.6032,
      "step": 3571
    },
    {
      "epoch": 0.6246120218579235,
      "grad_norm": 0.9856778042363141,
      "learning_rate": 0.00019177011897007797,
      "loss": 13.2434,
      "step": 3572
    },
    {
      "epoch": 0.6247868852459016,
      "grad_norm": 1.0308616433811466,
      "learning_rate": 0.00019176449233547694,
      "loss": 13.3986,
      "step": 3573
    },
    {
      "epoch": 0.6249617486338798,
      "grad_norm": 1.0658299739907293,
      "learning_rate": 0.00019175886386071608,
      "loss": 13.3455,
      "step": 3574
    },
    {
      "epoch": 0.6251366120218579,
      "grad_norm": 1.0074751488837057,
      "learning_rate": 0.00019175323354590827,
      "loss": 13.4953,
      "step": 3575
    },
    {
      "epoch": 0.6253114754098361,
      "grad_norm": 1.0755281909648557,
      "learning_rate": 0.00019174760139116642,
      "loss": 13.126,
      "step": 3576
    },
    {
      "epoch": 0.6254863387978142,
      "grad_norm": 0.8705066267273387,
      "learning_rate": 0.00019174196739660344,
      "loss": 13.4533,
      "step": 3577
    },
    {
      "epoch": 0.6256612021857924,
      "grad_norm": 0.9664935515640899,
      "learning_rate": 0.00019173633156233235,
      "loss": 13.388,
      "step": 3578
    },
    {
      "epoch": 0.6258360655737705,
      "grad_norm": 0.9602855745092426,
      "learning_rate": 0.00019173069388846612,
      "loss": 13.2515,
      "step": 3579
    },
    {
      "epoch": 0.6260109289617486,
      "grad_norm": 0.9309228626099669,
      "learning_rate": 0.00019172505437511786,
      "loss": 13.2337,
      "step": 3580
    },
    {
      "epoch": 0.6261857923497268,
      "grad_norm": 0.9198313033933357,
      "learning_rate": 0.00019171941302240057,
      "loss": 13.371,
      "step": 3581
    },
    {
      "epoch": 0.6263606557377049,
      "grad_norm": 0.9192050550554396,
      "learning_rate": 0.00019171376983042747,
      "loss": 13.2747,
      "step": 3582
    },
    {
      "epoch": 0.6265355191256831,
      "grad_norm": 0.9065961316842472,
      "learning_rate": 0.00019170812479931171,
      "loss": 13.3128,
      "step": 3583
    },
    {
      "epoch": 0.6267103825136612,
      "grad_norm": 0.9219654243500286,
      "learning_rate": 0.00019170247792916643,
      "loss": 13.1804,
      "step": 3584
    },
    {
      "epoch": 0.6268852459016393,
      "grad_norm": 0.865152438422764,
      "learning_rate": 0.00019169682922010492,
      "loss": 13.2837,
      "step": 3585
    },
    {
      "epoch": 0.6270601092896175,
      "grad_norm": 0.9281839398289898,
      "learning_rate": 0.00019169117867224043,
      "loss": 13.2468,
      "step": 3586
    },
    {
      "epoch": 0.6272349726775956,
      "grad_norm": 0.932409093444275,
      "learning_rate": 0.00019168552628568631,
      "loss": 13.3939,
      "step": 3587
    },
    {
      "epoch": 0.6274098360655738,
      "grad_norm": 0.9210429717218429,
      "learning_rate": 0.00019167987206055584,
      "loss": 12.9839,
      "step": 3588
    },
    {
      "epoch": 0.6275846994535519,
      "grad_norm": 0.8754675736993764,
      "learning_rate": 0.00019167421599696244,
      "loss": 13.069,
      "step": 3589
    },
    {
      "epoch": 0.62775956284153,
      "grad_norm": 0.9400603310327735,
      "learning_rate": 0.00019166855809501953,
      "loss": 13.3209,
      "step": 3590
    },
    {
      "epoch": 0.6279344262295082,
      "grad_norm": 0.8421239963459289,
      "learning_rate": 0.00019166289835484063,
      "loss": 13.159,
      "step": 3591
    },
    {
      "epoch": 0.6281092896174864,
      "grad_norm": 0.8098267928196508,
      "learning_rate": 0.0001916572367765391,
      "loss": 13.2892,
      "step": 3592
    },
    {
      "epoch": 0.6282841530054645,
      "grad_norm": 1.0735942763273507,
      "learning_rate": 0.00019165157336022858,
      "loss": 13.2467,
      "step": 3593
    },
    {
      "epoch": 0.6284590163934426,
      "grad_norm": 1.1022960163040687,
      "learning_rate": 0.00019164590810602262,
      "loss": 13.4036,
      "step": 3594
    },
    {
      "epoch": 0.6286338797814207,
      "grad_norm": 0.8355706145378303,
      "learning_rate": 0.0001916402410140348,
      "loss": 13.0969,
      "step": 3595
    },
    {
      "epoch": 0.6288087431693989,
      "grad_norm": 0.9431702250396629,
      "learning_rate": 0.00019163457208437877,
      "loss": 12.8832,
      "step": 3596
    },
    {
      "epoch": 0.6289836065573771,
      "grad_norm": 0.9302053393402668,
      "learning_rate": 0.0001916289013171682,
      "loss": 13.4554,
      "step": 3597
    },
    {
      "epoch": 0.6291584699453552,
      "grad_norm": 0.9297181869156238,
      "learning_rate": 0.00019162322871251687,
      "loss": 13.2422,
      "step": 3598
    },
    {
      "epoch": 0.6293333333333333,
      "grad_norm": 1.0651723421847739,
      "learning_rate": 0.00019161755427053842,
      "loss": 13.4621,
      "step": 3599
    },
    {
      "epoch": 0.6295081967213115,
      "grad_norm": 1.0274381091049043,
      "learning_rate": 0.00019161187799134673,
      "loss": 13.2751,
      "step": 3600
    },
    {
      "epoch": 0.6296830601092896,
      "grad_norm": 1.1232082225184332,
      "learning_rate": 0.00019160619987505562,
      "loss": 12.9489,
      "step": 3601
    },
    {
      "epoch": 0.6298579234972678,
      "grad_norm": 1.0530385566345342,
      "learning_rate": 0.0001916005199217789,
      "loss": 13.3156,
      "step": 3602
    },
    {
      "epoch": 0.6300327868852459,
      "grad_norm": 1.1663477562683207,
      "learning_rate": 0.00019159483813163054,
      "loss": 13.3832,
      "step": 3603
    },
    {
      "epoch": 0.630207650273224,
      "grad_norm": 0.8924431634745812,
      "learning_rate": 0.00019158915450472446,
      "loss": 13.189,
      "step": 3604
    },
    {
      "epoch": 0.6303825136612022,
      "grad_norm": 1.0291626851782465,
      "learning_rate": 0.00019158346904117458,
      "loss": 13.2755,
      "step": 3605
    },
    {
      "epoch": 0.6305573770491804,
      "grad_norm": 0.9480809163577634,
      "learning_rate": 0.00019157778174109497,
      "loss": 13.0095,
      "step": 3606
    },
    {
      "epoch": 0.6307322404371585,
      "grad_norm": 1.0423609672898957,
      "learning_rate": 0.00019157209260459962,
      "loss": 13.495,
      "step": 3607
    },
    {
      "epoch": 0.6309071038251366,
      "grad_norm": 0.9005209670694083,
      "learning_rate": 0.00019156640163180266,
      "loss": 13.1909,
      "step": 3608
    },
    {
      "epoch": 0.6310819672131147,
      "grad_norm": 0.8959594123740139,
      "learning_rate": 0.00019156070882281822,
      "loss": 13.3662,
      "step": 3609
    },
    {
      "epoch": 0.6312568306010929,
      "grad_norm": 0.9257389882476014,
      "learning_rate": 0.00019155501417776044,
      "loss": 13.2912,
      "step": 3610
    },
    {
      "epoch": 0.6314316939890711,
      "grad_norm": 0.9270736512631559,
      "learning_rate": 0.00019154931769674348,
      "loss": 13.4091,
      "step": 3611
    },
    {
      "epoch": 0.6316065573770492,
      "grad_norm": 0.9590132842486703,
      "learning_rate": 0.00019154361937988163,
      "loss": 13.3203,
      "step": 3612
    },
    {
      "epoch": 0.6317814207650273,
      "grad_norm": 0.8370905814624119,
      "learning_rate": 0.00019153791922728916,
      "loss": 13.3336,
      "step": 3613
    },
    {
      "epoch": 0.6319562841530054,
      "grad_norm": 0.9999241541815758,
      "learning_rate": 0.00019153221723908033,
      "loss": 13.3939,
      "step": 3614
    },
    {
      "epoch": 0.6321311475409837,
      "grad_norm": 0.8747258545771104,
      "learning_rate": 0.00019152651341536949,
      "loss": 13.0209,
      "step": 3615
    },
    {
      "epoch": 0.6323060109289618,
      "grad_norm": 0.9077763184171183,
      "learning_rate": 0.00019152080775627103,
      "loss": 13.4672,
      "step": 3616
    },
    {
      "epoch": 0.6324808743169399,
      "grad_norm": 0.9804461590401496,
      "learning_rate": 0.0001915151002618994,
      "loss": 13.4046,
      "step": 3617
    },
    {
      "epoch": 0.632655737704918,
      "grad_norm": 0.9009034916950297,
      "learning_rate": 0.00019150939093236902,
      "loss": 12.8938,
      "step": 3618
    },
    {
      "epoch": 0.6328306010928961,
      "grad_norm": 0.8703846389030221,
      "learning_rate": 0.00019150367976779434,
      "loss": 13.4381,
      "step": 3619
    },
    {
      "epoch": 0.6330054644808744,
      "grad_norm": 0.9377977352949444,
      "learning_rate": 0.00019149796676828996,
      "loss": 13.385,
      "step": 3620
    },
    {
      "epoch": 0.6331803278688525,
      "grad_norm": 0.8627436180159695,
      "learning_rate": 0.00019149225193397043,
      "loss": 13.2947,
      "step": 3621
    },
    {
      "epoch": 0.6333551912568306,
      "grad_norm": 0.8730607399557845,
      "learning_rate": 0.00019148653526495027,
      "loss": 13.545,
      "step": 3622
    },
    {
      "epoch": 0.6335300546448087,
      "grad_norm": 0.9791802662655331,
      "learning_rate": 0.00019148081676134422,
      "loss": 13.2716,
      "step": 3623
    },
    {
      "epoch": 0.6337049180327868,
      "grad_norm": 0.8655358803220099,
      "learning_rate": 0.00019147509642326692,
      "loss": 13.1813,
      "step": 3624
    },
    {
      "epoch": 0.6338797814207651,
      "grad_norm": 1.0119123933058383,
      "learning_rate": 0.00019146937425083306,
      "loss": 13.34,
      "step": 3625
    },
    {
      "epoch": 0.6340546448087432,
      "grad_norm": 1.0780541510643264,
      "learning_rate": 0.0001914636502441574,
      "loss": 13.1624,
      "step": 3626
    },
    {
      "epoch": 0.6342295081967213,
      "grad_norm": 0.9097794139641081,
      "learning_rate": 0.0001914579244033547,
      "loss": 13.1116,
      "step": 3627
    },
    {
      "epoch": 0.6344043715846994,
      "grad_norm": 1.093291522783342,
      "learning_rate": 0.0001914521967285398,
      "loss": 13.2847,
      "step": 3628
    },
    {
      "epoch": 0.6345792349726775,
      "grad_norm": 0.9879732736308767,
      "learning_rate": 0.0001914464672198276,
      "loss": 13.3857,
      "step": 3629
    },
    {
      "epoch": 0.6347540983606558,
      "grad_norm": 0.951014546481635,
      "learning_rate": 0.00019144073587733294,
      "loss": 12.9713,
      "step": 3630
    },
    {
      "epoch": 0.6349289617486339,
      "grad_norm": 1.1091285381574687,
      "learning_rate": 0.00019143500270117075,
      "loss": 13.298,
      "step": 3631
    },
    {
      "epoch": 0.635103825136612,
      "grad_norm": 0.866378342383825,
      "learning_rate": 0.000191429267691456,
      "loss": 13.2777,
      "step": 3632
    },
    {
      "epoch": 0.6352786885245901,
      "grad_norm": 0.9130000433824786,
      "learning_rate": 0.00019142353084830376,
      "loss": 13.2265,
      "step": 3633
    },
    {
      "epoch": 0.6354535519125684,
      "grad_norm": 0.9206724661957488,
      "learning_rate": 0.000191417792171829,
      "loss": 13.074,
      "step": 3634
    },
    {
      "epoch": 0.6356284153005465,
      "grad_norm": 0.910500155660248,
      "learning_rate": 0.0001914120516621468,
      "loss": 13.4429,
      "step": 3635
    },
    {
      "epoch": 0.6358032786885246,
      "grad_norm": 0.9002509339114957,
      "learning_rate": 0.0001914063093193723,
      "loss": 13.1505,
      "step": 3636
    },
    {
      "epoch": 0.6359781420765027,
      "grad_norm": 0.944940738406728,
      "learning_rate": 0.00019140056514362064,
      "loss": 12.8412,
      "step": 3637
    },
    {
      "epoch": 0.6361530054644808,
      "grad_norm": 0.8763298081077708,
      "learning_rate": 0.000191394819135007,
      "loss": 13.4958,
      "step": 3638
    },
    {
      "epoch": 0.6363278688524591,
      "grad_norm": 1.058629936352624,
      "learning_rate": 0.00019138907129364664,
      "loss": 13.1726,
      "step": 3639
    },
    {
      "epoch": 0.6365027322404372,
      "grad_norm": 0.9191471385810188,
      "learning_rate": 0.0001913833216196548,
      "loss": 13.426,
      "step": 3640
    },
    {
      "epoch": 0.6366775956284153,
      "grad_norm": 0.898962330118014,
      "learning_rate": 0.00019137757011314678,
      "loss": 13.1796,
      "step": 3641
    },
    {
      "epoch": 0.6368524590163934,
      "grad_norm": 0.8867251177401742,
      "learning_rate": 0.0001913718167742379,
      "loss": 13.2285,
      "step": 3642
    },
    {
      "epoch": 0.6370273224043715,
      "grad_norm": 0.866178381982188,
      "learning_rate": 0.00019136606160304355,
      "loss": 13.1791,
      "step": 3643
    },
    {
      "epoch": 0.6372021857923498,
      "grad_norm": 0.9391399989685768,
      "learning_rate": 0.00019136030459967915,
      "loss": 13.3437,
      "step": 3644
    },
    {
      "epoch": 0.6373770491803279,
      "grad_norm": 0.8953975031700151,
      "learning_rate": 0.0001913545457642601,
      "loss": 12.9604,
      "step": 3645
    },
    {
      "epoch": 0.637551912568306,
      "grad_norm": 0.8941018746125172,
      "learning_rate": 0.00019134878509690192,
      "loss": 13.1142,
      "step": 3646
    },
    {
      "epoch": 0.6377267759562841,
      "grad_norm": 1.0045107747919606,
      "learning_rate": 0.00019134302259772015,
      "loss": 13.413,
      "step": 3647
    },
    {
      "epoch": 0.6379016393442623,
      "grad_norm": 0.9880185213674761,
      "learning_rate": 0.0001913372582668303,
      "loss": 13.2314,
      "step": 3648
    },
    {
      "epoch": 0.6380765027322405,
      "grad_norm": 0.9595577592740299,
      "learning_rate": 0.00019133149210434795,
      "loss": 13.1539,
      "step": 3649
    },
    {
      "epoch": 0.6382513661202186,
      "grad_norm": 0.9494497568252476,
      "learning_rate": 0.00019132572411038879,
      "loss": 13.565,
      "step": 3650
    },
    {
      "epoch": 0.6384262295081967,
      "grad_norm": 1.0764137024190794,
      "learning_rate": 0.00019131995428506844,
      "loss": 13.3229,
      "step": 3651
    },
    {
      "epoch": 0.6386010928961748,
      "grad_norm": 0.8879058089467222,
      "learning_rate": 0.00019131418262850262,
      "loss": 12.8705,
      "step": 3652
    },
    {
      "epoch": 0.638775956284153,
      "grad_norm": 0.9931204456298843,
      "learning_rate": 0.00019130840914080707,
      "loss": 13.1383,
      "step": 3653
    },
    {
      "epoch": 0.6389508196721312,
      "grad_norm": 0.9483184438606509,
      "learning_rate": 0.00019130263382209755,
      "loss": 13.1862,
      "step": 3654
    },
    {
      "epoch": 0.6391256830601093,
      "grad_norm": 0.8733437266460906,
      "learning_rate": 0.00019129685667248992,
      "loss": 13.1989,
      "step": 3655
    },
    {
      "epoch": 0.6393005464480874,
      "grad_norm": 1.04817877531781,
      "learning_rate": 0.00019129107769209997,
      "loss": 13.3282,
      "step": 3656
    },
    {
      "epoch": 0.6394754098360655,
      "grad_norm": 0.8237516704913128,
      "learning_rate": 0.00019128529688104364,
      "loss": 13.0719,
      "step": 3657
    },
    {
      "epoch": 0.6396502732240437,
      "grad_norm": 0.9551899842864967,
      "learning_rate": 0.0001912795142394368,
      "loss": 13.2396,
      "step": 3658
    },
    {
      "epoch": 0.6398251366120219,
      "grad_norm": 1.0045527236105494,
      "learning_rate": 0.00019127372976739544,
      "loss": 13.3404,
      "step": 3659
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9302124397773424,
      "learning_rate": 0.00019126794346503554,
      "loss": 13.0671,
      "step": 3660
    },
    {
      "epoch": 0.6401748633879781,
      "grad_norm": 0.9240026034987119,
      "learning_rate": 0.00019126215533247316,
      "loss": 13.2276,
      "step": 3661
    },
    {
      "epoch": 0.6403497267759563,
      "grad_norm": 0.8979022231326702,
      "learning_rate": 0.00019125636536982432,
      "loss": 13.1558,
      "step": 3662
    },
    {
      "epoch": 0.6405245901639344,
      "grad_norm": 1.0134393105934298,
      "learning_rate": 0.00019125057357720522,
      "loss": 13.2842,
      "step": 3663
    },
    {
      "epoch": 0.6406994535519126,
      "grad_norm": 0.8490780041917523,
      "learning_rate": 0.00019124477995473188,
      "loss": 13.1423,
      "step": 3664
    },
    {
      "epoch": 0.6408743169398907,
      "grad_norm": 0.9824851199422934,
      "learning_rate": 0.0001912389845025206,
      "loss": 13.3018,
      "step": 3665
    },
    {
      "epoch": 0.6410491803278688,
      "grad_norm": 0.9171219095703207,
      "learning_rate": 0.0001912331872206875,
      "loss": 13.3208,
      "step": 3666
    },
    {
      "epoch": 0.641224043715847,
      "grad_norm": 0.9206172689962532,
      "learning_rate": 0.0001912273881093489,
      "loss": 12.9128,
      "step": 3667
    },
    {
      "epoch": 0.6413989071038252,
      "grad_norm": 0.9572049268212505,
      "learning_rate": 0.00019122158716862103,
      "loss": 13.186,
      "step": 3668
    },
    {
      "epoch": 0.6415737704918033,
      "grad_norm": 0.842095213282674,
      "learning_rate": 0.00019121578439862028,
      "loss": 13.2961,
      "step": 3669
    },
    {
      "epoch": 0.6417486338797814,
      "grad_norm": 0.9309465420383427,
      "learning_rate": 0.000191209979799463,
      "loss": 13.4475,
      "step": 3670
    },
    {
      "epoch": 0.6419234972677595,
      "grad_norm": 0.8696699504107078,
      "learning_rate": 0.00019120417337126553,
      "loss": 13.1149,
      "step": 3671
    },
    {
      "epoch": 0.6420983606557377,
      "grad_norm": 0.8925022339770885,
      "learning_rate": 0.00019119836511414438,
      "loss": 13.2311,
      "step": 3672
    },
    {
      "epoch": 0.6422732240437159,
      "grad_norm": 0.9249150482394956,
      "learning_rate": 0.00019119255502821595,
      "loss": 13.4231,
      "step": 3673
    },
    {
      "epoch": 0.642448087431694,
      "grad_norm": 0.9693512670203125,
      "learning_rate": 0.00019118674311359684,
      "loss": 13.4426,
      "step": 3674
    },
    {
      "epoch": 0.6426229508196721,
      "grad_norm": 0.8223076212575996,
      "learning_rate": 0.00019118092937040352,
      "loss": 13.1828,
      "step": 3675
    },
    {
      "epoch": 0.6427978142076503,
      "grad_norm": 0.926165460095612,
      "learning_rate": 0.00019117511379875265,
      "loss": 13.3546,
      "step": 3676
    },
    {
      "epoch": 0.6429726775956284,
      "grad_norm": 0.8150856417939993,
      "learning_rate": 0.00019116929639876074,
      "loss": 13.2132,
      "step": 3677
    },
    {
      "epoch": 0.6431475409836066,
      "grad_norm": 0.9502474963183173,
      "learning_rate": 0.00019116347717054456,
      "loss": 13.3331,
      "step": 3678
    },
    {
      "epoch": 0.6433224043715847,
      "grad_norm": 0.8555584350322237,
      "learning_rate": 0.00019115765611422077,
      "loss": 13.4085,
      "step": 3679
    },
    {
      "epoch": 0.6434972677595628,
      "grad_norm": 0.9494553786853492,
      "learning_rate": 0.00019115183322990605,
      "loss": 13.2282,
      "step": 3680
    },
    {
      "epoch": 0.643672131147541,
      "grad_norm": 0.9154991131469874,
      "learning_rate": 0.00019114600851771721,
      "loss": 13.2943,
      "step": 3681
    },
    {
      "epoch": 0.6438469945355191,
      "grad_norm": 0.8787384219258332,
      "learning_rate": 0.00019114018197777103,
      "loss": 13.4581,
      "step": 3682
    },
    {
      "epoch": 0.6440218579234973,
      "grad_norm": 0.9741454647967305,
      "learning_rate": 0.00019113435361018438,
      "loss": 12.9971,
      "step": 3683
    },
    {
      "epoch": 0.6441967213114754,
      "grad_norm": 0.9869890461933061,
      "learning_rate": 0.0001911285234150741,
      "loss": 13.3245,
      "step": 3684
    },
    {
      "epoch": 0.6443715846994535,
      "grad_norm": 1.0208758883930569,
      "learning_rate": 0.00019112269139255714,
      "loss": 13.196,
      "step": 3685
    },
    {
      "epoch": 0.6445464480874317,
      "grad_norm": 1.0141752090399263,
      "learning_rate": 0.00019111685754275045,
      "loss": 13.2066,
      "step": 3686
    },
    {
      "epoch": 0.6447213114754098,
      "grad_norm": 0.8144784889086392,
      "learning_rate": 0.00019111102186577096,
      "loss": 13.2571,
      "step": 3687
    },
    {
      "epoch": 0.644896174863388,
      "grad_norm": 0.8573800899017006,
      "learning_rate": 0.00019110518436173576,
      "loss": 13.1642,
      "step": 3688
    },
    {
      "epoch": 0.6450710382513661,
      "grad_norm": 0.9276581025170995,
      "learning_rate": 0.00019109934503076188,
      "loss": 13.2888,
      "step": 3689
    },
    {
      "epoch": 0.6452459016393443,
      "grad_norm": 0.8762179826764428,
      "learning_rate": 0.00019109350387296642,
      "loss": 13.2527,
      "step": 3690
    },
    {
      "epoch": 0.6454207650273224,
      "grad_norm": 0.9154087842127469,
      "learning_rate": 0.00019108766088846653,
      "loss": 13.1564,
      "step": 3691
    },
    {
      "epoch": 0.6455956284153005,
      "grad_norm": 0.9165173117302744,
      "learning_rate": 0.00019108181607737935,
      "loss": 13.328,
      "step": 3692
    },
    {
      "epoch": 0.6457704918032787,
      "grad_norm": 0.9041407602374264,
      "learning_rate": 0.0001910759694398221,
      "loss": 13.2224,
      "step": 3693
    },
    {
      "epoch": 0.6459453551912568,
      "grad_norm": 0.8827964462455048,
      "learning_rate": 0.000191070120975912,
      "loss": 13.0714,
      "step": 3694
    },
    {
      "epoch": 0.646120218579235,
      "grad_norm": 0.9175686441609919,
      "learning_rate": 0.00019106427068576638,
      "loss": 13.2076,
      "step": 3695
    },
    {
      "epoch": 0.6462950819672131,
      "grad_norm": 0.8869580607774395,
      "learning_rate": 0.00019105841856950252,
      "loss": 13.3406,
      "step": 3696
    },
    {
      "epoch": 0.6464699453551913,
      "grad_norm": 0.8395454211372709,
      "learning_rate": 0.0001910525646272378,
      "loss": 13.2326,
      "step": 3697
    },
    {
      "epoch": 0.6466448087431694,
      "grad_norm": 0.9741827179569801,
      "learning_rate": 0.00019104670885908956,
      "loss": 13.4062,
      "step": 3698
    },
    {
      "epoch": 0.6468196721311475,
      "grad_norm": 0.9674573523570276,
      "learning_rate": 0.00019104085126517528,
      "loss": 13.2005,
      "step": 3699
    },
    {
      "epoch": 0.6469945355191257,
      "grad_norm": 0.8406459062973882,
      "learning_rate": 0.0001910349918456124,
      "loss": 13.3452,
      "step": 3700
    },
    {
      "epoch": 0.6471693989071038,
      "grad_norm": 1.0391355672362939,
      "learning_rate": 0.00019102913060051839,
      "loss": 13.2323,
      "step": 3701
    },
    {
      "epoch": 0.647344262295082,
      "grad_norm": 0.8812423981785957,
      "learning_rate": 0.00019102326753001086,
      "loss": 12.9094,
      "step": 3702
    },
    {
      "epoch": 0.6475191256830601,
      "grad_norm": 1.2050457200734093,
      "learning_rate": 0.0001910174026342073,
      "loss": 13.1482,
      "step": 3703
    },
    {
      "epoch": 0.6476939890710383,
      "grad_norm": 0.9489129863374999,
      "learning_rate": 0.00019101153591322538,
      "loss": 13.1289,
      "step": 3704
    },
    {
      "epoch": 0.6478688524590164,
      "grad_norm": 0.9676197432589859,
      "learning_rate": 0.00019100566736718273,
      "loss": 13.2689,
      "step": 3705
    },
    {
      "epoch": 0.6480437158469945,
      "grad_norm": 0.965903784279026,
      "learning_rate": 0.000190999796996197,
      "loss": 13.4403,
      "step": 3706
    },
    {
      "epoch": 0.6482185792349727,
      "grad_norm": 0.9621152502737645,
      "learning_rate": 0.00019099392480038593,
      "loss": 13.4353,
      "step": 3707
    },
    {
      "epoch": 0.6483934426229508,
      "grad_norm": 0.9963768841582776,
      "learning_rate": 0.0001909880507798673,
      "loss": 13.3305,
      "step": 3708
    },
    {
      "epoch": 0.648568306010929,
      "grad_norm": 0.9582300759501763,
      "learning_rate": 0.0001909821749347589,
      "loss": 13.2957,
      "step": 3709
    },
    {
      "epoch": 0.6487431693989071,
      "grad_norm": 0.9068298838993132,
      "learning_rate": 0.0001909762972651785,
      "loss": 13.1868,
      "step": 3710
    },
    {
      "epoch": 0.6489180327868852,
      "grad_norm": 0.9021493312731304,
      "learning_rate": 0.000190970417771244,
      "loss": 13.3626,
      "step": 3711
    },
    {
      "epoch": 0.6490928961748634,
      "grad_norm": 0.8362667824941309,
      "learning_rate": 0.00019096453645307335,
      "loss": 13.1561,
      "step": 3712
    },
    {
      "epoch": 0.6492677595628416,
      "grad_norm": 0.8742195290587053,
      "learning_rate": 0.00019095865331078443,
      "loss": 13.4305,
      "step": 3713
    },
    {
      "epoch": 0.6494426229508197,
      "grad_norm": 0.9164974026848297,
      "learning_rate": 0.00019095276834449522,
      "loss": 13.1997,
      "step": 3714
    },
    {
      "epoch": 0.6496174863387978,
      "grad_norm": 0.9439019925201335,
      "learning_rate": 0.00019094688155432373,
      "loss": 13.3026,
      "step": 3715
    },
    {
      "epoch": 0.6497923497267759,
      "grad_norm": 0.9212808237867899,
      "learning_rate": 0.00019094099294038807,
      "loss": 13.3976,
      "step": 3716
    },
    {
      "epoch": 0.6499672131147541,
      "grad_norm": 1.0325094065944418,
      "learning_rate": 0.00019093510250280625,
      "loss": 13.2774,
      "step": 3717
    },
    {
      "epoch": 0.6501420765027323,
      "grad_norm": 0.9791368630791887,
      "learning_rate": 0.00019092921024169642,
      "loss": 13.3729,
      "step": 3718
    },
    {
      "epoch": 0.6503169398907104,
      "grad_norm": 0.8987417681890425,
      "learning_rate": 0.0001909233161571767,
      "loss": 13.224,
      "step": 3719
    },
    {
      "epoch": 0.6504918032786885,
      "grad_norm": 0.9252235483285964,
      "learning_rate": 0.00019091742024936537,
      "loss": 13.5249,
      "step": 3720
    },
    {
      "epoch": 0.6506666666666666,
      "grad_norm": 1.018078389975224,
      "learning_rate": 0.00019091152251838058,
      "loss": 13.2233,
      "step": 3721
    },
    {
      "epoch": 0.6508415300546448,
      "grad_norm": 0.9749884888051431,
      "learning_rate": 0.00019090562296434068,
      "loss": 13.37,
      "step": 3722
    },
    {
      "epoch": 0.651016393442623,
      "grad_norm": 0.9387921366337414,
      "learning_rate": 0.00019089972158736385,
      "loss": 13.3645,
      "step": 3723
    },
    {
      "epoch": 0.6511912568306011,
      "grad_norm": 1.0186917537811289,
      "learning_rate": 0.00019089381838756855,
      "loss": 13.4194,
      "step": 3724
    },
    {
      "epoch": 0.6513661202185792,
      "grad_norm": 0.8174718733658719,
      "learning_rate": 0.00019088791336507307,
      "loss": 13.5225,
      "step": 3725
    },
    {
      "epoch": 0.6515409836065573,
      "grad_norm": 1.0243304080142186,
      "learning_rate": 0.0001908820065199959,
      "loss": 13.1595,
      "step": 3726
    },
    {
      "epoch": 0.6517158469945356,
      "grad_norm": 0.988402367043013,
      "learning_rate": 0.00019087609785245545,
      "loss": 13.4816,
      "step": 3727
    },
    {
      "epoch": 0.6518907103825137,
      "grad_norm": 0.9189083124486576,
      "learning_rate": 0.0001908701873625702,
      "loss": 13.1129,
      "step": 3728
    },
    {
      "epoch": 0.6520655737704918,
      "grad_norm": 1.0735845172222926,
      "learning_rate": 0.0001908642750504587,
      "loss": 13.0777,
      "step": 3729
    },
    {
      "epoch": 0.6522404371584699,
      "grad_norm": 0.963548807167117,
      "learning_rate": 0.0001908583609162395,
      "loss": 13.2599,
      "step": 3730
    },
    {
      "epoch": 0.6524153005464481,
      "grad_norm": 0.9346205487687754,
      "learning_rate": 0.00019085244496003115,
      "loss": 13.1415,
      "step": 3731
    },
    {
      "epoch": 0.6525901639344263,
      "grad_norm": 0.9626582647331113,
      "learning_rate": 0.00019084652718195238,
      "loss": 13.1177,
      "step": 3732
    },
    {
      "epoch": 0.6527650273224044,
      "grad_norm": 0.8857523188857402,
      "learning_rate": 0.00019084060758212178,
      "loss": 13.1133,
      "step": 3733
    },
    {
      "epoch": 0.6529398907103825,
      "grad_norm": 0.8703419497558966,
      "learning_rate": 0.00019083468616065807,
      "loss": 13.3667,
      "step": 3734
    },
    {
      "epoch": 0.6531147540983606,
      "grad_norm": 0.9743479759967195,
      "learning_rate": 0.00019082876291768002,
      "loss": 13.3948,
      "step": 3735
    },
    {
      "epoch": 0.6532896174863388,
      "grad_norm": 0.9899359066262141,
      "learning_rate": 0.0001908228378533064,
      "loss": 13.3893,
      "step": 3736
    },
    {
      "epoch": 0.653464480874317,
      "grad_norm": 0.8827158928538401,
      "learning_rate": 0.000190816910967656,
      "loss": 13.2441,
      "step": 3737
    },
    {
      "epoch": 0.6536393442622951,
      "grad_norm": 0.9031151383344904,
      "learning_rate": 0.0001908109822608477,
      "loss": 13.0026,
      "step": 3738
    },
    {
      "epoch": 0.6538142076502732,
      "grad_norm": 0.918172207665738,
      "learning_rate": 0.00019080505173300037,
      "loss": 13.2889,
      "step": 3739
    },
    {
      "epoch": 0.6539890710382513,
      "grad_norm": 0.9221818746630599,
      "learning_rate": 0.00019079911938423293,
      "loss": 13.3355,
      "step": 3740
    },
    {
      "epoch": 0.6541639344262296,
      "grad_norm": 0.8904224814100627,
      "learning_rate": 0.00019079318521466438,
      "loss": 13.1325,
      "step": 3741
    },
    {
      "epoch": 0.6543387978142077,
      "grad_norm": 0.8566654838168806,
      "learning_rate": 0.0001907872492244137,
      "loss": 13.2374,
      "step": 3742
    },
    {
      "epoch": 0.6545136612021858,
      "grad_norm": 0.8619036647192887,
      "learning_rate": 0.0001907813114135999,
      "loss": 13.1512,
      "step": 3743
    },
    {
      "epoch": 0.6546885245901639,
      "grad_norm": 0.8777849530530881,
      "learning_rate": 0.00019077537178234207,
      "loss": 13.1046,
      "step": 3744
    },
    {
      "epoch": 0.654863387978142,
      "grad_norm": 0.8328302634225089,
      "learning_rate": 0.00019076943033075936,
      "loss": 12.9123,
      "step": 3745
    },
    {
      "epoch": 0.6550382513661203,
      "grad_norm": 0.9085253212334607,
      "learning_rate": 0.00019076348705897084,
      "loss": 13.145,
      "step": 3746
    },
    {
      "epoch": 0.6552131147540984,
      "grad_norm": 0.9824327516727756,
      "learning_rate": 0.00019075754196709572,
      "loss": 13.0599,
      "step": 3747
    },
    {
      "epoch": 0.6553879781420765,
      "grad_norm": 0.9255134978359781,
      "learning_rate": 0.00019075159505525322,
      "loss": 13.3883,
      "step": 3748
    },
    {
      "epoch": 0.6555628415300546,
      "grad_norm": 1.0486250150893386,
      "learning_rate": 0.00019074564632356263,
      "loss": 13.3422,
      "step": 3749
    },
    {
      "epoch": 0.6557377049180327,
      "grad_norm": 1.1044740106529924,
      "learning_rate": 0.00019073969577214315,
      "loss": 13.2356,
      "step": 3750
    },
    {
      "epoch": 0.655912568306011,
      "grad_norm": 0.8942559229439718,
      "learning_rate": 0.0001907337434011142,
      "loss": 13.2539,
      "step": 3751
    },
    {
      "epoch": 0.6560874316939891,
      "grad_norm": 1.1116944727126477,
      "learning_rate": 0.0001907277892105951,
      "loss": 13.1507,
      "step": 3752
    },
    {
      "epoch": 0.6562622950819672,
      "grad_norm": 0.8523415118706545,
      "learning_rate": 0.00019072183320070524,
      "loss": 13.1734,
      "step": 3753
    },
    {
      "epoch": 0.6564371584699453,
      "grad_norm": 0.9612966134014285,
      "learning_rate": 0.0001907158753715641,
      "loss": 13.1621,
      "step": 3754
    },
    {
      "epoch": 0.6566120218579234,
      "grad_norm": 0.9903410782499905,
      "learning_rate": 0.0001907099157232911,
      "loss": 13.2434,
      "step": 3755
    },
    {
      "epoch": 0.6567868852459017,
      "grad_norm": 1.0149433913342172,
      "learning_rate": 0.00019070395425600578,
      "loss": 13.2122,
      "step": 3756
    },
    {
      "epoch": 0.6569617486338798,
      "grad_norm": 1.1374454980573439,
      "learning_rate": 0.0001906979909698277,
      "loss": 13.3717,
      "step": 3757
    },
    {
      "epoch": 0.6571366120218579,
      "grad_norm": 0.9318849960462098,
      "learning_rate": 0.00019069202586487639,
      "loss": 13.133,
      "step": 3758
    },
    {
      "epoch": 0.657311475409836,
      "grad_norm": 0.9578719188416831,
      "learning_rate": 0.0001906860589412715,
      "loss": 13.1811,
      "step": 3759
    },
    {
      "epoch": 0.6574863387978143,
      "grad_norm": 0.9320334044978401,
      "learning_rate": 0.0001906800901991327,
      "loss": 13.3332,
      "step": 3760
    },
    {
      "epoch": 0.6576612021857924,
      "grad_norm": 0.8889603682096104,
      "learning_rate": 0.00019067411963857967,
      "loss": 13.273,
      "step": 3761
    },
    {
      "epoch": 0.6578360655737705,
      "grad_norm": 0.9561427949750504,
      "learning_rate": 0.00019066814725973212,
      "loss": 13.2589,
      "step": 3762
    },
    {
      "epoch": 0.6580109289617486,
      "grad_norm": 0.9464423900255711,
      "learning_rate": 0.00019066217306270983,
      "loss": 13.2602,
      "step": 3763
    },
    {
      "epoch": 0.6581857923497267,
      "grad_norm": 0.8594958835757792,
      "learning_rate": 0.00019065619704763264,
      "loss": 13.217,
      "step": 3764
    },
    {
      "epoch": 0.658360655737705,
      "grad_norm": 0.9047839496438367,
      "learning_rate": 0.0001906502192146203,
      "loss": 13.3183,
      "step": 3765
    },
    {
      "epoch": 0.6585355191256831,
      "grad_norm": 0.8369119119717521,
      "learning_rate": 0.00019064423956379275,
      "loss": 13.3843,
      "step": 3766
    },
    {
      "epoch": 0.6587103825136612,
      "grad_norm": 0.8613075409949721,
      "learning_rate": 0.0001906382580952699,
      "loss": 13.1111,
      "step": 3767
    },
    {
      "epoch": 0.6588852459016393,
      "grad_norm": 0.9359391968985578,
      "learning_rate": 0.00019063227480917166,
      "loss": 13.2563,
      "step": 3768
    },
    {
      "epoch": 0.6590601092896174,
      "grad_norm": 0.8520697100309609,
      "learning_rate": 0.00019062628970561806,
      "loss": 13.2144,
      "step": 3769
    },
    {
      "epoch": 0.6592349726775957,
      "grad_norm": 0.9384486827280636,
      "learning_rate": 0.00019062030278472905,
      "loss": 13.351,
      "step": 3770
    },
    {
      "epoch": 0.6594098360655738,
      "grad_norm": 0.9056066080560504,
      "learning_rate": 0.00019061431404662473,
      "loss": 13.4818,
      "step": 3771
    },
    {
      "epoch": 0.6595846994535519,
      "grad_norm": 0.8225782081389472,
      "learning_rate": 0.0001906083234914252,
      "loss": 12.9771,
      "step": 3772
    },
    {
      "epoch": 0.65975956284153,
      "grad_norm": 0.9077006360083703,
      "learning_rate": 0.0001906023311192506,
      "loss": 13.2287,
      "step": 3773
    },
    {
      "epoch": 0.6599344262295082,
      "grad_norm": 0.9179400227399187,
      "learning_rate": 0.00019059633693022104,
      "loss": 13.104,
      "step": 3774
    },
    {
      "epoch": 0.6601092896174864,
      "grad_norm": 0.927907532529622,
      "learning_rate": 0.0001905903409244568,
      "loss": 13.3971,
      "step": 3775
    },
    {
      "epoch": 0.6602841530054645,
      "grad_norm": 0.9433750539057824,
      "learning_rate": 0.00019058434310207804,
      "loss": 13.1626,
      "step": 3776
    },
    {
      "epoch": 0.6604590163934426,
      "grad_norm": 1.0080622487408168,
      "learning_rate": 0.0001905783434632051,
      "loss": 13.4346,
      "step": 3777
    },
    {
      "epoch": 0.6606338797814207,
      "grad_norm": 0.9799401258282238,
      "learning_rate": 0.00019057234200795824,
      "loss": 13.2575,
      "step": 3778
    },
    {
      "epoch": 0.6608087431693989,
      "grad_norm": 0.926728844513868,
      "learning_rate": 0.00019056633873645783,
      "loss": 13.0903,
      "step": 3779
    },
    {
      "epoch": 0.6609836065573771,
      "grad_norm": 0.8566426608544483,
      "learning_rate": 0.00019056033364882427,
      "loss": 13.2038,
      "step": 3780
    },
    {
      "epoch": 0.6611584699453552,
      "grad_norm": 0.8709469661700339,
      "learning_rate": 0.00019055432674517796,
      "loss": 12.8201,
      "step": 3781
    },
    {
      "epoch": 0.6613333333333333,
      "grad_norm": 1.0084516592927784,
      "learning_rate": 0.00019054831802563934,
      "loss": 13.1927,
      "step": 3782
    },
    {
      "epoch": 0.6615081967213114,
      "grad_norm": 0.8637058999866941,
      "learning_rate": 0.00019054230749032894,
      "loss": 12.8926,
      "step": 3783
    },
    {
      "epoch": 0.6616830601092896,
      "grad_norm": 0.9355323753541519,
      "learning_rate": 0.00019053629513936726,
      "loss": 13.1757,
      "step": 3784
    },
    {
      "epoch": 0.6618579234972678,
      "grad_norm": 0.8034455650223516,
      "learning_rate": 0.00019053028097287488,
      "loss": 13.4036,
      "step": 3785
    },
    {
      "epoch": 0.6620327868852459,
      "grad_norm": 0.8982172471769391,
      "learning_rate": 0.0001905242649909724,
      "loss": 13.237,
      "step": 3786
    },
    {
      "epoch": 0.662207650273224,
      "grad_norm": 0.9493317665883115,
      "learning_rate": 0.0001905182471937805,
      "loss": 13.3452,
      "step": 3787
    },
    {
      "epoch": 0.6623825136612022,
      "grad_norm": 1.0359032573737754,
      "learning_rate": 0.0001905122275814198,
      "loss": 13.1341,
      "step": 3788
    },
    {
      "epoch": 0.6625573770491803,
      "grad_norm": 0.8789782438755326,
      "learning_rate": 0.000190506206154011,
      "loss": 13.1583,
      "step": 3789
    },
    {
      "epoch": 0.6627322404371585,
      "grad_norm": 0.9288521636016424,
      "learning_rate": 0.0001905001829116749,
      "loss": 13.0375,
      "step": 3790
    },
    {
      "epoch": 0.6629071038251366,
      "grad_norm": 0.9715120753853351,
      "learning_rate": 0.00019049415785453227,
      "loss": 13.3195,
      "step": 3791
    },
    {
      "epoch": 0.6630819672131147,
      "grad_norm": 0.9837214708836891,
      "learning_rate": 0.00019048813098270387,
      "loss": 13.243,
      "step": 3792
    },
    {
      "epoch": 0.6632568306010929,
      "grad_norm": 1.262804171953384,
      "learning_rate": 0.00019048210229631067,
      "loss": 13.1975,
      "step": 3793
    },
    {
      "epoch": 0.6634316939890711,
      "grad_norm": 0.9223225322758913,
      "learning_rate": 0.00019047607179547348,
      "loss": 13.167,
      "step": 3794
    },
    {
      "epoch": 0.6636065573770492,
      "grad_norm": 1.0249940393272785,
      "learning_rate": 0.00019047003948031326,
      "loss": 13.4128,
      "step": 3795
    },
    {
      "epoch": 0.6637814207650273,
      "grad_norm": 0.9429006938053692,
      "learning_rate": 0.00019046400535095094,
      "loss": 13.2079,
      "step": 3796
    },
    {
      "epoch": 0.6639562841530054,
      "grad_norm": 0.9490683442499493,
      "learning_rate": 0.00019045796940750758,
      "loss": 13.3659,
      "step": 3797
    },
    {
      "epoch": 0.6641311475409836,
      "grad_norm": 1.0278233571557678,
      "learning_rate": 0.00019045193165010416,
      "loss": 13.136,
      "step": 3798
    },
    {
      "epoch": 0.6643060109289618,
      "grad_norm": 0.9209551365661126,
      "learning_rate": 0.00019044589207886181,
      "loss": 13.1848,
      "step": 3799
    },
    {
      "epoch": 0.6644808743169399,
      "grad_norm": 1.1991638372622166,
      "learning_rate": 0.00019043985069390162,
      "loss": 13.0036,
      "step": 3800
    },
    {
      "epoch": 0.664655737704918,
      "grad_norm": 0.9380135918193236,
      "learning_rate": 0.00019043380749534473,
      "loss": 13.0412,
      "step": 3801
    },
    {
      "epoch": 0.6648306010928962,
      "grad_norm": 1.1310965193667681,
      "learning_rate": 0.00019042776248331234,
      "loss": 13.1269,
      "step": 3802
    },
    {
      "epoch": 0.6650054644808743,
      "grad_norm": 1.0246698032672834,
      "learning_rate": 0.00019042171565792563,
      "loss": 13.344,
      "step": 3803
    },
    {
      "epoch": 0.6651803278688525,
      "grad_norm": 1.0492193963307572,
      "learning_rate": 0.00019041566701930593,
      "loss": 13.3231,
      "step": 3804
    },
    {
      "epoch": 0.6653551912568306,
      "grad_norm": 1.1524099772758918,
      "learning_rate": 0.00019040961656757447,
      "loss": 13.3112,
      "step": 3805
    },
    {
      "epoch": 0.6655300546448087,
      "grad_norm": 1.2980949654377723,
      "learning_rate": 0.00019040356430285264,
      "loss": 13.1498,
      "step": 3806
    },
    {
      "epoch": 0.6657049180327869,
      "grad_norm": 1.115784506821576,
      "learning_rate": 0.00019039751022526173,
      "loss": 13.1465,
      "step": 3807
    },
    {
      "epoch": 0.665879781420765,
      "grad_norm": 0.9109594744198574,
      "learning_rate": 0.00019039145433492319,
      "loss": 13.36,
      "step": 3808
    },
    {
      "epoch": 0.6660546448087432,
      "grad_norm": 1.0597761801212222,
      "learning_rate": 0.00019038539663195848,
      "loss": 13.1366,
      "step": 3809
    },
    {
      "epoch": 0.6662295081967213,
      "grad_norm": 1.1675637248525992,
      "learning_rate": 0.000190379337116489,
      "loss": 13.5947,
      "step": 3810
    },
    {
      "epoch": 0.6664043715846995,
      "grad_norm": 1.1132807402527392,
      "learning_rate": 0.00019037327578863634,
      "loss": 13.3815,
      "step": 3811
    },
    {
      "epoch": 0.6665792349726776,
      "grad_norm": 0.9320284771964691,
      "learning_rate": 0.00019036721264852204,
      "loss": 13.0685,
      "step": 3812
    },
    {
      "epoch": 0.6667540983606557,
      "grad_norm": 1.0606838613145786,
      "learning_rate": 0.00019036114769626763,
      "loss": 13.2149,
      "step": 3813
    },
    {
      "epoch": 0.6669289617486339,
      "grad_norm": 1.0690981549239469,
      "learning_rate": 0.00019035508093199476,
      "loss": 13.3101,
      "step": 3814
    },
    {
      "epoch": 0.667103825136612,
      "grad_norm": 0.9718544141179539,
      "learning_rate": 0.0001903490123558251,
      "loss": 13.4535,
      "step": 3815
    },
    {
      "epoch": 0.6672786885245902,
      "grad_norm": 1.0882427507700394,
      "learning_rate": 0.00019034294196788033,
      "loss": 12.9673,
      "step": 3816
    },
    {
      "epoch": 0.6674535519125683,
      "grad_norm": 0.8274670324542005,
      "learning_rate": 0.0001903368697682822,
      "loss": 13.2508,
      "step": 3817
    },
    {
      "epoch": 0.6676284153005464,
      "grad_norm": 1.0583313935914347,
      "learning_rate": 0.00019033079575715248,
      "loss": 12.9978,
      "step": 3818
    },
    {
      "epoch": 0.6678032786885246,
      "grad_norm": 0.9329130139527168,
      "learning_rate": 0.0001903247199346129,
      "loss": 13.2239,
      "step": 3819
    },
    {
      "epoch": 0.6679781420765027,
      "grad_norm": 0.9540031243921263,
      "learning_rate": 0.00019031864230078537,
      "loss": 12.9689,
      "step": 3820
    },
    {
      "epoch": 0.6681530054644809,
      "grad_norm": 0.9829660530443889,
      "learning_rate": 0.00019031256285579175,
      "loss": 13.0909,
      "step": 3821
    },
    {
      "epoch": 0.668327868852459,
      "grad_norm": 0.9270922980121188,
      "learning_rate": 0.00019030648159975395,
      "loss": 12.8659,
      "step": 3822
    },
    {
      "epoch": 0.6685027322404371,
      "grad_norm": 1.0594179393578183,
      "learning_rate": 0.00019030039853279394,
      "loss": 12.9186,
      "step": 3823
    },
    {
      "epoch": 0.6686775956284153,
      "grad_norm": 0.898915369817315,
      "learning_rate": 0.00019029431365503365,
      "loss": 13.3367,
      "step": 3824
    },
    {
      "epoch": 0.6688524590163935,
      "grad_norm": 0.8908084065946564,
      "learning_rate": 0.00019028822696659514,
      "loss": 12.9679,
      "step": 3825
    },
    {
      "epoch": 0.6690273224043716,
      "grad_norm": 0.8887185087888042,
      "learning_rate": 0.00019028213846760046,
      "loss": 13.2542,
      "step": 3826
    },
    {
      "epoch": 0.6692021857923497,
      "grad_norm": 0.9155471965168476,
      "learning_rate": 0.0001902760481581717,
      "loss": 13.2454,
      "step": 3827
    },
    {
      "epoch": 0.6693770491803279,
      "grad_norm": 0.8521890692776897,
      "learning_rate": 0.00019026995603843097,
      "loss": 13.169,
      "step": 3828
    },
    {
      "epoch": 0.669551912568306,
      "grad_norm": 0.8829273492627311,
      "learning_rate": 0.00019026386210850048,
      "loss": 13.4434,
      "step": 3829
    },
    {
      "epoch": 0.6697267759562842,
      "grad_norm": 0.8961161205058256,
      "learning_rate": 0.00019025776636850242,
      "loss": 13.1195,
      "step": 3830
    },
    {
      "epoch": 0.6699016393442623,
      "grad_norm": 0.810909112225887,
      "learning_rate": 0.000190251668818559,
      "loss": 13.108,
      "step": 3831
    },
    {
      "epoch": 0.6700765027322404,
      "grad_norm": 1.4104501155095097,
      "learning_rate": 0.00019024556945879253,
      "loss": 13.2233,
      "step": 3832
    },
    {
      "epoch": 0.6702513661202186,
      "grad_norm": 1.0329470904128553,
      "learning_rate": 0.00019023946828932526,
      "loss": 13.2905,
      "step": 3833
    },
    {
      "epoch": 0.6704262295081967,
      "grad_norm": 0.9183222405698005,
      "learning_rate": 0.0001902333653102796,
      "loss": 13.305,
      "step": 3834
    },
    {
      "epoch": 0.6706010928961749,
      "grad_norm": 0.8891450743234129,
      "learning_rate": 0.00019022726052177793,
      "loss": 13.1795,
      "step": 3835
    },
    {
      "epoch": 0.670775956284153,
      "grad_norm": 0.8344285561032143,
      "learning_rate": 0.00019022115392394265,
      "loss": 13.1514,
      "step": 3836
    },
    {
      "epoch": 0.6709508196721311,
      "grad_norm": 0.8693473179838513,
      "learning_rate": 0.0001902150455168962,
      "loss": 13.2117,
      "step": 3837
    },
    {
      "epoch": 0.6711256830601093,
      "grad_norm": 0.8983589880693434,
      "learning_rate": 0.00019020893530076108,
      "loss": 13.0662,
      "step": 3838
    },
    {
      "epoch": 0.6713005464480875,
      "grad_norm": 0.9034173017888193,
      "learning_rate": 0.00019020282327565985,
      "loss": 13.3325,
      "step": 3839
    },
    {
      "epoch": 0.6714754098360656,
      "grad_norm": 0.899237397864575,
      "learning_rate": 0.00019019670944171507,
      "loss": 13.1287,
      "step": 3840
    },
    {
      "epoch": 0.6716502732240437,
      "grad_norm": 0.9950992685337566,
      "learning_rate": 0.0001901905937990493,
      "loss": 13.2222,
      "step": 3841
    },
    {
      "epoch": 0.6718251366120218,
      "grad_norm": 0.8623372608051825,
      "learning_rate": 0.00019018447634778523,
      "loss": 13.2246,
      "step": 3842
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.9150714776141629,
      "learning_rate": 0.00019017835708804546,
      "loss": 13.2657,
      "step": 3843
    },
    {
      "epoch": 0.6721748633879782,
      "grad_norm": 0.8286176004166821,
      "learning_rate": 0.00019017223601995278,
      "loss": 13.0373,
      "step": 3844
    },
    {
      "epoch": 0.6723497267759563,
      "grad_norm": 0.8538653297110903,
      "learning_rate": 0.0001901661131436299,
      "loss": 13.4349,
      "step": 3845
    },
    {
      "epoch": 0.6725245901639344,
      "grad_norm": 0.9132824844804699,
      "learning_rate": 0.0001901599884591996,
      "loss": 13.3693,
      "step": 3846
    },
    {
      "epoch": 0.6726994535519125,
      "grad_norm": 0.9478318354254138,
      "learning_rate": 0.0001901538619667847,
      "loss": 13.2908,
      "step": 3847
    },
    {
      "epoch": 0.6728743169398907,
      "grad_norm": 0.9067148900259258,
      "learning_rate": 0.00019014773366650807,
      "loss": 12.8005,
      "step": 3848
    },
    {
      "epoch": 0.6730491803278689,
      "grad_norm": 0.9449763765065027,
      "learning_rate": 0.0001901416035584926,
      "loss": 12.8296,
      "step": 3849
    },
    {
      "epoch": 0.673224043715847,
      "grad_norm": 0.9569426554486518,
      "learning_rate": 0.00019013547164286115,
      "loss": 13.2013,
      "step": 3850
    },
    {
      "epoch": 0.6733989071038251,
      "grad_norm": 0.876682386954186,
      "learning_rate": 0.00019012933791973682,
      "loss": 13.2164,
      "step": 3851
    },
    {
      "epoch": 0.6735737704918032,
      "grad_norm": 0.9065566005143275,
      "learning_rate": 0.00019012320238924249,
      "loss": 13.044,
      "step": 3852
    },
    {
      "epoch": 0.6737486338797815,
      "grad_norm": 0.9219008665115129,
      "learning_rate": 0.00019011706505150122,
      "loss": 13.2779,
      "step": 3853
    },
    {
      "epoch": 0.6739234972677596,
      "grad_norm": 0.9133428217917966,
      "learning_rate": 0.00019011092590663612,
      "loss": 13.3005,
      "step": 3854
    },
    {
      "epoch": 0.6740983606557377,
      "grad_norm": 0.9121757579943661,
      "learning_rate": 0.0001901047849547703,
      "loss": 13.4291,
      "step": 3855
    },
    {
      "epoch": 0.6742732240437158,
      "grad_norm": 0.9157828131455258,
      "learning_rate": 0.00019009864219602686,
      "loss": 13.1047,
      "step": 3856
    },
    {
      "epoch": 0.674448087431694,
      "grad_norm": 0.8517151043085508,
      "learning_rate": 0.000190092497630529,
      "loss": 13.259,
      "step": 3857
    },
    {
      "epoch": 0.6746229508196722,
      "grad_norm": 0.9465743312748149,
      "learning_rate": 0.00019008635125839995,
      "loss": 13.0634,
      "step": 3858
    },
    {
      "epoch": 0.6747978142076503,
      "grad_norm": 1.1095826156883228,
      "learning_rate": 0.00019008020307976296,
      "loss": 13.3283,
      "step": 3859
    },
    {
      "epoch": 0.6749726775956284,
      "grad_norm": 1.0270643000664843,
      "learning_rate": 0.00019007405309474131,
      "loss": 13.2017,
      "step": 3860
    },
    {
      "epoch": 0.6751475409836065,
      "grad_norm": 0.8422933076543295,
      "learning_rate": 0.00019006790130345836,
      "loss": 12.9259,
      "step": 3861
    },
    {
      "epoch": 0.6753224043715848,
      "grad_norm": 0.9986659949793735,
      "learning_rate": 0.0001900617477060374,
      "loss": 13.4152,
      "step": 3862
    },
    {
      "epoch": 0.6754972677595629,
      "grad_norm": 0.8919301006344047,
      "learning_rate": 0.0001900555923026019,
      "loss": 12.9948,
      "step": 3863
    },
    {
      "epoch": 0.675672131147541,
      "grad_norm": 0.9459311783551202,
      "learning_rate": 0.00019004943509327523,
      "loss": 13.0182,
      "step": 3864
    },
    {
      "epoch": 0.6758469945355191,
      "grad_norm": 0.9798662381529445,
      "learning_rate": 0.00019004327607818094,
      "loss": 13.1577,
      "step": 3865
    },
    {
      "epoch": 0.6760218579234972,
      "grad_norm": 0.8495069431187524,
      "learning_rate": 0.0001900371152574425,
      "loss": 13.0607,
      "step": 3866
    },
    {
      "epoch": 0.6761967213114755,
      "grad_norm": 0.9987346241477776,
      "learning_rate": 0.00019003095263118343,
      "loss": 13.3972,
      "step": 3867
    },
    {
      "epoch": 0.6763715846994536,
      "grad_norm": 0.9273499361643016,
      "learning_rate": 0.0001900247881995273,
      "loss": 13.4439,
      "step": 3868
    },
    {
      "epoch": 0.6765464480874317,
      "grad_norm": 0.8842747174579997,
      "learning_rate": 0.00019001862196259782,
      "loss": 13.1604,
      "step": 3869
    },
    {
      "epoch": 0.6767213114754098,
      "grad_norm": 0.7990430794814416,
      "learning_rate": 0.0001900124539205185,
      "loss": 13.0644,
      "step": 3870
    },
    {
      "epoch": 0.6768961748633879,
      "grad_norm": 0.927712595481299,
      "learning_rate": 0.00019000628407341317,
      "loss": 13.4393,
      "step": 3871
    },
    {
      "epoch": 0.6770710382513662,
      "grad_norm": 0.8770597059638237,
      "learning_rate": 0.00019000011242140547,
      "loss": 13.2764,
      "step": 3872
    },
    {
      "epoch": 0.6772459016393443,
      "grad_norm": 0.8729135233058697,
      "learning_rate": 0.00018999393896461917,
      "loss": 13.3464,
      "step": 3873
    },
    {
      "epoch": 0.6774207650273224,
      "grad_norm": 0.9428759358776241,
      "learning_rate": 0.00018998776370317809,
      "loss": 13.0423,
      "step": 3874
    },
    {
      "epoch": 0.6775956284153005,
      "grad_norm": 0.8976286098149034,
      "learning_rate": 0.00018998158663720603,
      "loss": 13.0973,
      "step": 3875
    },
    {
      "epoch": 0.6777704918032786,
      "grad_norm": 1.013797740234126,
      "learning_rate": 0.00018997540776682687,
      "loss": 12.8732,
      "step": 3876
    },
    {
      "epoch": 0.6779453551912569,
      "grad_norm": 0.9623190370925615,
      "learning_rate": 0.00018996922709216455,
      "loss": 13.3211,
      "step": 3877
    },
    {
      "epoch": 0.678120218579235,
      "grad_norm": 1.0412555004055748,
      "learning_rate": 0.00018996304461334296,
      "loss": 13.3113,
      "step": 3878
    },
    {
      "epoch": 0.6782950819672131,
      "grad_norm": 0.9287411973133,
      "learning_rate": 0.00018995686033048613,
      "loss": 13.1755,
      "step": 3879
    },
    {
      "epoch": 0.6784699453551912,
      "grad_norm": 0.9259392458508525,
      "learning_rate": 0.00018995067424371805,
      "loss": 13.2525,
      "step": 3880
    },
    {
      "epoch": 0.6786448087431693,
      "grad_norm": 0.879579477982813,
      "learning_rate": 0.0001899444863531627,
      "loss": 13.0884,
      "step": 3881
    },
    {
      "epoch": 0.6788196721311476,
      "grad_norm": 0.9314616600973008,
      "learning_rate": 0.0001899382966589443,
      "loss": 13.406,
      "step": 3882
    },
    {
      "epoch": 0.6789945355191257,
      "grad_norm": 0.835151766229153,
      "learning_rate": 0.0001899321051611869,
      "loss": 13.0294,
      "step": 3883
    },
    {
      "epoch": 0.6791693989071038,
      "grad_norm": 0.8545562438043741,
      "learning_rate": 0.00018992591186001466,
      "loss": 13.2073,
      "step": 3884
    },
    {
      "epoch": 0.6793442622950819,
      "grad_norm": 0.8898073129236367,
      "learning_rate": 0.00018991971675555176,
      "loss": 13.2407,
      "step": 3885
    },
    {
      "epoch": 0.67951912568306,
      "grad_norm": 0.9572910807104029,
      "learning_rate": 0.00018991351984792246,
      "loss": 13.4074,
      "step": 3886
    },
    {
      "epoch": 0.6796939890710383,
      "grad_norm": 0.9426011754285617,
      "learning_rate": 0.00018990732113725102,
      "loss": 13.246,
      "step": 3887
    },
    {
      "epoch": 0.6798688524590164,
      "grad_norm": 0.8677181529968921,
      "learning_rate": 0.00018990112062366172,
      "loss": 13.3298,
      "step": 3888
    },
    {
      "epoch": 0.6800437158469945,
      "grad_norm": 0.9286501645951222,
      "learning_rate": 0.00018989491830727895,
      "loss": 13.2114,
      "step": 3889
    },
    {
      "epoch": 0.6802185792349726,
      "grad_norm": 0.911881918475213,
      "learning_rate": 0.000189888714188227,
      "loss": 13.1666,
      "step": 3890
    },
    {
      "epoch": 0.6803934426229509,
      "grad_norm": 1.1728404861272572,
      "learning_rate": 0.0001898825082666304,
      "loss": 13.3196,
      "step": 3891
    },
    {
      "epoch": 0.680568306010929,
      "grad_norm": 0.8745405970010403,
      "learning_rate": 0.00018987630054261348,
      "loss": 13.1101,
      "step": 3892
    },
    {
      "epoch": 0.6807431693989071,
      "grad_norm": 0.9414913214386322,
      "learning_rate": 0.00018987009101630086,
      "loss": 13.3122,
      "step": 3893
    },
    {
      "epoch": 0.6809180327868852,
      "grad_norm": 0.8754463767956993,
      "learning_rate": 0.0001898638796878169,
      "loss": 13.1798,
      "step": 3894
    },
    {
      "epoch": 0.6810928961748633,
      "grad_norm": 0.9135529427627417,
      "learning_rate": 0.00018985766655728626,
      "loss": 13.0455,
      "step": 3895
    },
    {
      "epoch": 0.6812677595628416,
      "grad_norm": 1.0205910989032465,
      "learning_rate": 0.0001898514516248335,
      "loss": 13.3099,
      "step": 3896
    },
    {
      "epoch": 0.6814426229508197,
      "grad_norm": 0.9307345589372218,
      "learning_rate": 0.00018984523489058326,
      "loss": 13.3168,
      "step": 3897
    },
    {
      "epoch": 0.6816174863387978,
      "grad_norm": 0.9365286025973448,
      "learning_rate": 0.0001898390163546602,
      "loss": 13.3318,
      "step": 3898
    },
    {
      "epoch": 0.6817923497267759,
      "grad_norm": 1.0155847074113802,
      "learning_rate": 0.000189832796017189,
      "loss": 13.0565,
      "step": 3899
    },
    {
      "epoch": 0.6819672131147541,
      "grad_norm": 0.9085302423299785,
      "learning_rate": 0.00018982657387829445,
      "loss": 12.9928,
      "step": 3900
    },
    {
      "epoch": 0.6821420765027323,
      "grad_norm": 0.8565228887709407,
      "learning_rate": 0.00018982034993810128,
      "loss": 13.1897,
      "step": 3901
    },
    {
      "epoch": 0.6823169398907104,
      "grad_norm": 0.9964518495240798,
      "learning_rate": 0.00018981412419673429,
      "loss": 13.385,
      "step": 3902
    },
    {
      "epoch": 0.6824918032786885,
      "grad_norm": 0.9901033928126028,
      "learning_rate": 0.00018980789665431837,
      "loss": 12.9138,
      "step": 3903
    },
    {
      "epoch": 0.6826666666666666,
      "grad_norm": 0.9360576258347943,
      "learning_rate": 0.00018980166731097836,
      "loss": 13.3299,
      "step": 3904
    },
    {
      "epoch": 0.6828415300546448,
      "grad_norm": 1.190814570087446,
      "learning_rate": 0.00018979543616683917,
      "loss": 13.3427,
      "step": 3905
    },
    {
      "epoch": 0.683016393442623,
      "grad_norm": 0.9054406124240282,
      "learning_rate": 0.00018978920322202582,
      "loss": 13.2018,
      "step": 3906
    },
    {
      "epoch": 0.6831912568306011,
      "grad_norm": 1.2305451482802534,
      "learning_rate": 0.00018978296847666323,
      "loss": 13.2037,
      "step": 3907
    },
    {
      "epoch": 0.6833661202185792,
      "grad_norm": 0.9145461420443479,
      "learning_rate": 0.00018977673193087648,
      "loss": 13.1944,
      "step": 3908
    },
    {
      "epoch": 0.6835409836065574,
      "grad_norm": 1.2880475400193996,
      "learning_rate": 0.00018977049358479057,
      "loss": 13.2563,
      "step": 3909
    },
    {
      "epoch": 0.6837158469945355,
      "grad_norm": 1.0415174848782656,
      "learning_rate": 0.00018976425343853067,
      "loss": 13.1661,
      "step": 3910
    },
    {
      "epoch": 0.6838907103825137,
      "grad_norm": 1.1699382069610498,
      "learning_rate": 0.00018975801149222183,
      "loss": 13.0733,
      "step": 3911
    },
    {
      "epoch": 0.6840655737704918,
      "grad_norm": 1.0058159266524684,
      "learning_rate": 0.00018975176774598928,
      "loss": 13.1661,
      "step": 3912
    },
    {
      "epoch": 0.6842404371584699,
      "grad_norm": 1.0757535850558009,
      "learning_rate": 0.00018974552219995822,
      "loss": 13.3079,
      "step": 3913
    },
    {
      "epoch": 0.6844153005464481,
      "grad_norm": 1.1226195756754953,
      "learning_rate": 0.0001897392748542539,
      "loss": 13.3139,
      "step": 3914
    },
    {
      "epoch": 0.6845901639344262,
      "grad_norm": 0.8941298856305411,
      "learning_rate": 0.00018973302570900159,
      "loss": 13.3346,
      "step": 3915
    },
    {
      "epoch": 0.6847650273224044,
      "grad_norm": 1.0082165172721618,
      "learning_rate": 0.00018972677476432657,
      "loss": 13.2636,
      "step": 3916
    },
    {
      "epoch": 0.6849398907103825,
      "grad_norm": 0.917127339035043,
      "learning_rate": 0.0001897205220203542,
      "loss": 13.0833,
      "step": 3917
    },
    {
      "epoch": 0.6851147540983606,
      "grad_norm": 0.911557354208069,
      "learning_rate": 0.00018971426747720993,
      "loss": 13.0766,
      "step": 3918
    },
    {
      "epoch": 0.6852896174863388,
      "grad_norm": 0.8093817807760927,
      "learning_rate": 0.00018970801113501912,
      "loss": 13.0944,
      "step": 3919
    },
    {
      "epoch": 0.685464480874317,
      "grad_norm": 0.9117897741639117,
      "learning_rate": 0.00018970175299390726,
      "loss": 13.0101,
      "step": 3920
    },
    {
      "epoch": 0.6856393442622951,
      "grad_norm": 0.8191793148151615,
      "learning_rate": 0.00018969549305399982,
      "loss": 13.0983,
      "step": 3921
    },
    {
      "epoch": 0.6858142076502732,
      "grad_norm": 0.8393453224352623,
      "learning_rate": 0.00018968923131542232,
      "loss": 13.0552,
      "step": 3922
    },
    {
      "epoch": 0.6859890710382514,
      "grad_norm": 0.8877275493913934,
      "learning_rate": 0.0001896829677783004,
      "loss": 13.1425,
      "step": 3923
    },
    {
      "epoch": 0.6861639344262295,
      "grad_norm": 0.8599397512432422,
      "learning_rate": 0.00018967670244275956,
      "loss": 12.8947,
      "step": 3924
    },
    {
      "epoch": 0.6863387978142077,
      "grad_norm": 0.8815049497827895,
      "learning_rate": 0.00018967043530892553,
      "loss": 12.8705,
      "step": 3925
    },
    {
      "epoch": 0.6865136612021858,
      "grad_norm": 1.0819446271355462,
      "learning_rate": 0.00018966416637692394,
      "loss": 13.2542,
      "step": 3926
    },
    {
      "epoch": 0.6866885245901639,
      "grad_norm": 0.8935824622278327,
      "learning_rate": 0.0001896578956468805,
      "loss": 13.1023,
      "step": 3927
    },
    {
      "epoch": 0.6868633879781421,
      "grad_norm": 0.8775904765632491,
      "learning_rate": 0.00018965162311892098,
      "loss": 12.937,
      "step": 3928
    },
    {
      "epoch": 0.6870382513661202,
      "grad_norm": 0.9930756449129193,
      "learning_rate": 0.00018964534879317113,
      "loss": 13.258,
      "step": 3929
    },
    {
      "epoch": 0.6872131147540984,
      "grad_norm": 0.9650820479271923,
      "learning_rate": 0.00018963907266975681,
      "loss": 13.2495,
      "step": 3930
    },
    {
      "epoch": 0.6873879781420765,
      "grad_norm": 0.9467397081005186,
      "learning_rate": 0.0001896327947488038,
      "loss": 13.2775,
      "step": 3931
    },
    {
      "epoch": 0.6875628415300546,
      "grad_norm": 0.9496770209407756,
      "learning_rate": 0.00018962651503043811,
      "loss": 13.2185,
      "step": 3932
    },
    {
      "epoch": 0.6877377049180328,
      "grad_norm": 1.0018176671060464,
      "learning_rate": 0.0001896202335147856,
      "loss": 13.3068,
      "step": 3933
    },
    {
      "epoch": 0.6879125683060109,
      "grad_norm": 0.870096057172097,
      "learning_rate": 0.00018961395020197222,
      "loss": 13.1696,
      "step": 3934
    },
    {
      "epoch": 0.6880874316939891,
      "grad_norm": 0.9783246297700356,
      "learning_rate": 0.000189607665092124,
      "loss": 13.393,
      "step": 3935
    },
    {
      "epoch": 0.6882622950819672,
      "grad_norm": 0.875809159438067,
      "learning_rate": 0.00018960137818536694,
      "loss": 13.3084,
      "step": 3936
    },
    {
      "epoch": 0.6884371584699454,
      "grad_norm": 0.9829505493404562,
      "learning_rate": 0.00018959508948182717,
      "loss": 13.3453,
      "step": 3937
    },
    {
      "epoch": 0.6886120218579235,
      "grad_norm": 0.915583734992481,
      "learning_rate": 0.00018958879898163072,
      "loss": 13.2931,
      "step": 3938
    },
    {
      "epoch": 0.6887868852459016,
      "grad_norm": 0.8183021004209811,
      "learning_rate": 0.0001895825066849038,
      "loss": 13.2555,
      "step": 3939
    },
    {
      "epoch": 0.6889617486338798,
      "grad_norm": 0.970104308877271,
      "learning_rate": 0.00018957621259177256,
      "loss": 13.115,
      "step": 3940
    },
    {
      "epoch": 0.6891366120218579,
      "grad_norm": 0.8439408173909542,
      "learning_rate": 0.00018956991670236324,
      "loss": 13.1687,
      "step": 3941
    },
    {
      "epoch": 0.6893114754098361,
      "grad_norm": 1.1105793817221734,
      "learning_rate": 0.00018956361901680205,
      "loss": 12.9891,
      "step": 3942
    },
    {
      "epoch": 0.6894863387978142,
      "grad_norm": 0.9062209087086757,
      "learning_rate": 0.00018955731953521535,
      "loss": 13.1739,
      "step": 3943
    },
    {
      "epoch": 0.6896612021857923,
      "grad_norm": 1.13778931372349,
      "learning_rate": 0.00018955101825772936,
      "loss": 12.969,
      "step": 3944
    },
    {
      "epoch": 0.6898360655737705,
      "grad_norm": 1.106409295573368,
      "learning_rate": 0.00018954471518447052,
      "loss": 13.2466,
      "step": 3945
    },
    {
      "epoch": 0.6900109289617486,
      "grad_norm": 0.9620000529372114,
      "learning_rate": 0.00018953841031556522,
      "loss": 13.4088,
      "step": 3946
    },
    {
      "epoch": 0.6901857923497268,
      "grad_norm": 1.077612838517263,
      "learning_rate": 0.00018953210365113982,
      "loss": 13.1839,
      "step": 3947
    },
    {
      "epoch": 0.6903606557377049,
      "grad_norm": 0.9129201175644241,
      "learning_rate": 0.00018952579519132087,
      "loss": 13.2681,
      "step": 3948
    },
    {
      "epoch": 0.690535519125683,
      "grad_norm": 1.1177601088100135,
      "learning_rate": 0.00018951948493623486,
      "loss": 13.053,
      "step": 3949
    },
    {
      "epoch": 0.6907103825136612,
      "grad_norm": 0.9695708780588636,
      "learning_rate": 0.0001895131728860083,
      "loss": 13.1136,
      "step": 3950
    },
    {
      "epoch": 0.6908852459016394,
      "grad_norm": 1.0578601673160357,
      "learning_rate": 0.0001895068590407678,
      "loss": 13.1383,
      "step": 3951
    },
    {
      "epoch": 0.6910601092896175,
      "grad_norm": 0.8786078738289076,
      "learning_rate": 0.00018950054340063994,
      "loss": 12.9155,
      "step": 3952
    },
    {
      "epoch": 0.6912349726775956,
      "grad_norm": 0.9966757243283413,
      "learning_rate": 0.00018949422596575136,
      "loss": 13.2099,
      "step": 3953
    },
    {
      "epoch": 0.6914098360655738,
      "grad_norm": 0.9480607186298594,
      "learning_rate": 0.00018948790673622884,
      "loss": 13.2601,
      "step": 3954
    },
    {
      "epoch": 0.6915846994535519,
      "grad_norm": 0.897422817102646,
      "learning_rate": 0.00018948158571219895,
      "loss": 13.2185,
      "step": 3955
    },
    {
      "epoch": 0.6917595628415301,
      "grad_norm": 0.9651898556310812,
      "learning_rate": 0.00018947526289378854,
      "loss": 13.2431,
      "step": 3956
    },
    {
      "epoch": 0.6919344262295082,
      "grad_norm": 0.856959890772078,
      "learning_rate": 0.0001894689382811244,
      "loss": 13.1046,
      "step": 3957
    },
    {
      "epoch": 0.6921092896174863,
      "grad_norm": 0.9567559443158246,
      "learning_rate": 0.00018946261187433334,
      "loss": 13.1491,
      "step": 3958
    },
    {
      "epoch": 0.6922841530054645,
      "grad_norm": 0.9614009674709161,
      "learning_rate": 0.0001894562836735422,
      "loss": 13.2031,
      "step": 3959
    },
    {
      "epoch": 0.6924590163934427,
      "grad_norm": 0.9420613808209674,
      "learning_rate": 0.00018944995367887796,
      "loss": 13.0671,
      "step": 3960
    },
    {
      "epoch": 0.6926338797814208,
      "grad_norm": 0.8635820927724773,
      "learning_rate": 0.00018944362189046748,
      "loss": 13.4257,
      "step": 3961
    },
    {
      "epoch": 0.6928087431693989,
      "grad_norm": 1.0163659524706246,
      "learning_rate": 0.00018943728830843773,
      "loss": 13.1551,
      "step": 3962
    },
    {
      "epoch": 0.692983606557377,
      "grad_norm": 1.0508332066383586,
      "learning_rate": 0.00018943095293291572,
      "loss": 13.4963,
      "step": 3963
    },
    {
      "epoch": 0.6931584699453552,
      "grad_norm": 0.8900217963114947,
      "learning_rate": 0.00018942461576402857,
      "loss": 13.2957,
      "step": 3964
    },
    {
      "epoch": 0.6933333333333334,
      "grad_norm": 0.9895074218458361,
      "learning_rate": 0.00018941827680190327,
      "loss": 13.3642,
      "step": 3965
    },
    {
      "epoch": 0.6935081967213115,
      "grad_norm": 0.8273009006153099,
      "learning_rate": 0.000189411936046667,
      "loss": 12.6991,
      "step": 3966
    },
    {
      "epoch": 0.6936830601092896,
      "grad_norm": 0.970673793542197,
      "learning_rate": 0.00018940559349844683,
      "loss": 13.4448,
      "step": 3967
    },
    {
      "epoch": 0.6938579234972677,
      "grad_norm": 1.0068815981372445,
      "learning_rate": 0.00018939924915737004,
      "loss": 13.0224,
      "step": 3968
    },
    {
      "epoch": 0.694032786885246,
      "grad_norm": 0.999667011155052,
      "learning_rate": 0.0001893929030235638,
      "loss": 13.2961,
      "step": 3969
    },
    {
      "epoch": 0.6942076502732241,
      "grad_norm": 0.9361224016558369,
      "learning_rate": 0.00018938655509715538,
      "loss": 13.0183,
      "step": 3970
    },
    {
      "epoch": 0.6943825136612022,
      "grad_norm": 1.0185623855350945,
      "learning_rate": 0.0001893802053782721,
      "loss": 13.3526,
      "step": 3971
    },
    {
      "epoch": 0.6945573770491803,
      "grad_norm": 1.0816104095755426,
      "learning_rate": 0.00018937385386704126,
      "loss": 13.3947,
      "step": 3972
    },
    {
      "epoch": 0.6947322404371584,
      "grad_norm": 0.8093093079908391,
      "learning_rate": 0.00018936750056359022,
      "loss": 13.0857,
      "step": 3973
    },
    {
      "epoch": 0.6949071038251367,
      "grad_norm": 1.0068716560988822,
      "learning_rate": 0.0001893611454680464,
      "loss": 13.3249,
      "step": 3974
    },
    {
      "epoch": 0.6950819672131148,
      "grad_norm": 0.8735420817869544,
      "learning_rate": 0.00018935478858053724,
      "loss": 13.0879,
      "step": 3975
    },
    {
      "epoch": 0.6952568306010929,
      "grad_norm": 1.0219548926638777,
      "learning_rate": 0.00018934842990119024,
      "loss": 13.2087,
      "step": 3976
    },
    {
      "epoch": 0.695431693989071,
      "grad_norm": 0.9155017645958079,
      "learning_rate": 0.00018934206943013287,
      "loss": 13.2045,
      "step": 3977
    },
    {
      "epoch": 0.6956065573770491,
      "grad_norm": 1.0227131852535034,
      "learning_rate": 0.00018933570716749267,
      "loss": 13.2492,
      "step": 3978
    },
    {
      "epoch": 0.6957814207650274,
      "grad_norm": 0.9766174744893799,
      "learning_rate": 0.00018932934311339728,
      "loss": 13.1903,
      "step": 3979
    },
    {
      "epoch": 0.6959562841530055,
      "grad_norm": 1.041558764963351,
      "learning_rate": 0.00018932297726797427,
      "loss": 13.2309,
      "step": 3980
    },
    {
      "epoch": 0.6961311475409836,
      "grad_norm": 0.9314819375675513,
      "learning_rate": 0.00018931660963135126,
      "loss": 13.1978,
      "step": 3981
    },
    {
      "epoch": 0.6963060109289617,
      "grad_norm": 0.9533653944871688,
      "learning_rate": 0.00018931024020365605,
      "loss": 13.3095,
      "step": 3982
    },
    {
      "epoch": 0.6964808743169398,
      "grad_norm": 1.0477932886957277,
      "learning_rate": 0.0001893038689850163,
      "loss": 13.4099,
      "step": 3983
    },
    {
      "epoch": 0.6966557377049181,
      "grad_norm": 0.9010804770165132,
      "learning_rate": 0.00018929749597555975,
      "loss": 13.198,
      "step": 3984
    },
    {
      "epoch": 0.6968306010928962,
      "grad_norm": 0.953594722808253,
      "learning_rate": 0.00018929112117541425,
      "loss": 13.3483,
      "step": 3985
    },
    {
      "epoch": 0.6970054644808743,
      "grad_norm": 1.014991442504839,
      "learning_rate": 0.0001892847445847076,
      "loss": 13.148,
      "step": 3986
    },
    {
      "epoch": 0.6971803278688524,
      "grad_norm": 0.8655711630557998,
      "learning_rate": 0.00018927836620356769,
      "loss": 13.4185,
      "step": 3987
    },
    {
      "epoch": 0.6973551912568307,
      "grad_norm": 0.8881349224921089,
      "learning_rate": 0.0001892719860321224,
      "loss": 13.2154,
      "step": 3988
    },
    {
      "epoch": 0.6975300546448088,
      "grad_norm": 0.866997521040014,
      "learning_rate": 0.0001892656040704997,
      "loss": 12.8726,
      "step": 3989
    },
    {
      "epoch": 0.6977049180327869,
      "grad_norm": 0.879693235880319,
      "learning_rate": 0.00018925922031882758,
      "loss": 13.2264,
      "step": 3990
    },
    {
      "epoch": 0.697879781420765,
      "grad_norm": 0.8562601903972453,
      "learning_rate": 0.000189252834777234,
      "loss": 12.9287,
      "step": 3991
    },
    {
      "epoch": 0.6980546448087431,
      "grad_norm": 0.9973908500539523,
      "learning_rate": 0.00018924644744584704,
      "loss": 13.3088,
      "step": 3992
    },
    {
      "epoch": 0.6982295081967214,
      "grad_norm": 0.8374978571495812,
      "learning_rate": 0.00018924005832479478,
      "loss": 13.1453,
      "step": 3993
    },
    {
      "epoch": 0.6984043715846995,
      "grad_norm": 0.9851301813266657,
      "learning_rate": 0.00018923366741420536,
      "loss": 13.2577,
      "step": 3994
    },
    {
      "epoch": 0.6985792349726776,
      "grad_norm": 0.9384391670908399,
      "learning_rate": 0.00018922727471420693,
      "loss": 13.0886,
      "step": 3995
    },
    {
      "epoch": 0.6987540983606557,
      "grad_norm": 0.9610741822359429,
      "learning_rate": 0.0001892208802249277,
      "loss": 13.1012,
      "step": 3996
    },
    {
      "epoch": 0.6989289617486338,
      "grad_norm": 0.8757061051824352,
      "learning_rate": 0.00018921448394649583,
      "loss": 12.8123,
      "step": 3997
    },
    {
      "epoch": 0.6991038251366121,
      "grad_norm": 0.8790150583084548,
      "learning_rate": 0.00018920808587903967,
      "loss": 13.1436,
      "step": 3998
    },
    {
      "epoch": 0.6992786885245902,
      "grad_norm": 0.9073401426794416,
      "learning_rate": 0.00018920168602268748,
      "loss": 13.1148,
      "step": 3999
    },
    {
      "epoch": 0.6994535519125683,
      "grad_norm": 0.8473716640516493,
      "learning_rate": 0.00018919528437756758,
      "loss": 13.3071,
      "step": 4000
    },
    {
      "epoch": 0.6996284153005464,
      "grad_norm": 0.996158404366075,
      "learning_rate": 0.00018918888094380834,
      "loss": 13.1442,
      "step": 4001
    },
    {
      "epoch": 0.6998032786885245,
      "grad_norm": 0.8924682916984016,
      "learning_rate": 0.00018918247572153823,
      "loss": 13.3437,
      "step": 4002
    },
    {
      "epoch": 0.6999781420765028,
      "grad_norm": 0.9458856019897139,
      "learning_rate": 0.00018917606871088566,
      "loss": 13.4136,
      "step": 4003
    },
    {
      "epoch": 0.7001530054644809,
      "grad_norm": 0.874984482871073,
      "learning_rate": 0.00018916965991197908,
      "loss": 13.2938,
      "step": 4004
    },
    {
      "epoch": 0.700327868852459,
      "grad_norm": 0.8734905939020229,
      "learning_rate": 0.00018916324932494706,
      "loss": 13.3416,
      "step": 4005
    },
    {
      "epoch": 0.7005027322404371,
      "grad_norm": 0.869506209813023,
      "learning_rate": 0.0001891568369499181,
      "loss": 13.4232,
      "step": 4006
    },
    {
      "epoch": 0.7006775956284153,
      "grad_norm": 0.7683156709798674,
      "learning_rate": 0.00018915042278702082,
      "loss": 12.9946,
      "step": 4007
    },
    {
      "epoch": 0.7008524590163935,
      "grad_norm": 0.8921585049567621,
      "learning_rate": 0.00018914400683638384,
      "loss": 13.3188,
      "step": 4008
    },
    {
      "epoch": 0.7010273224043716,
      "grad_norm": 0.9481315804644961,
      "learning_rate": 0.0001891375890981358,
      "loss": 13.199,
      "step": 4009
    },
    {
      "epoch": 0.7012021857923497,
      "grad_norm": 0.8325364237719337,
      "learning_rate": 0.00018913116957240545,
      "loss": 13.2107,
      "step": 4010
    },
    {
      "epoch": 0.7013770491803278,
      "grad_norm": 0.8897250504687936,
      "learning_rate": 0.00018912474825932143,
      "loss": 13.1046,
      "step": 4011
    },
    {
      "epoch": 0.701551912568306,
      "grad_norm": 0.9551277969925197,
      "learning_rate": 0.0001891183251590126,
      "loss": 13.2618,
      "step": 4012
    },
    {
      "epoch": 0.7017267759562842,
      "grad_norm": 0.8790357565227483,
      "learning_rate": 0.0001891119002716077,
      "loss": 13.3309,
      "step": 4013
    },
    {
      "epoch": 0.7019016393442623,
      "grad_norm": 0.8570332614265423,
      "learning_rate": 0.00018910547359723558,
      "loss": 13.2854,
      "step": 4014
    },
    {
      "epoch": 0.7020765027322404,
      "grad_norm": 0.8401263302883903,
      "learning_rate": 0.00018909904513602515,
      "loss": 13.1674,
      "step": 4015
    },
    {
      "epoch": 0.7022513661202185,
      "grad_norm": 0.8446458001252787,
      "learning_rate": 0.00018909261488810527,
      "loss": 13.1404,
      "step": 4016
    },
    {
      "epoch": 0.7024262295081968,
      "grad_norm": 0.8624525210156683,
      "learning_rate": 0.00018908618285360494,
      "loss": 12.8692,
      "step": 4017
    },
    {
      "epoch": 0.7026010928961749,
      "grad_norm": 0.8978781194579646,
      "learning_rate": 0.00018907974903265308,
      "loss": 13.106,
      "step": 4018
    },
    {
      "epoch": 0.702775956284153,
      "grad_norm": 0.8821049019113392,
      "learning_rate": 0.00018907331342537877,
      "loss": 13.0518,
      "step": 4019
    },
    {
      "epoch": 0.7029508196721311,
      "grad_norm": 0.9557221574262257,
      "learning_rate": 0.00018906687603191102,
      "loss": 13.248,
      "step": 4020
    },
    {
      "epoch": 0.7031256830601093,
      "grad_norm": 0.802194871874074,
      "learning_rate": 0.00018906043685237893,
      "loss": 13.0382,
      "step": 4021
    },
    {
      "epoch": 0.7033005464480875,
      "grad_norm": 0.8177081637993799,
      "learning_rate": 0.00018905399588691163,
      "loss": 12.9694,
      "step": 4022
    },
    {
      "epoch": 0.7034754098360656,
      "grad_norm": 0.9322103055436632,
      "learning_rate": 0.00018904755313563828,
      "loss": 13.1526,
      "step": 4023
    },
    {
      "epoch": 0.7036502732240437,
      "grad_norm": 0.943883375348273,
      "learning_rate": 0.0001890411085986881,
      "loss": 13.1809,
      "step": 4024
    },
    {
      "epoch": 0.7038251366120218,
      "grad_norm": 0.9004384327991489,
      "learning_rate": 0.00018903466227619028,
      "loss": 13.3539,
      "step": 4025
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.8893079895598054,
      "learning_rate": 0.00018902821416827412,
      "loss": 13.2314,
      "step": 4026
    },
    {
      "epoch": 0.7041748633879782,
      "grad_norm": 0.8196470073434647,
      "learning_rate": 0.0001890217642750689,
      "loss": 12.9315,
      "step": 4027
    },
    {
      "epoch": 0.7043497267759563,
      "grad_norm": 0.9654631605724908,
      "learning_rate": 0.000189015312596704,
      "loss": 13.3118,
      "step": 4028
    },
    {
      "epoch": 0.7045245901639344,
      "grad_norm": 0.8804396447838686,
      "learning_rate": 0.00018900885913330874,
      "loss": 13.2641,
      "step": 4029
    },
    {
      "epoch": 0.7046994535519125,
      "grad_norm": 0.9253062621528432,
      "learning_rate": 0.00018900240388501258,
      "loss": 13.2775,
      "step": 4030
    },
    {
      "epoch": 0.7048743169398907,
      "grad_norm": 0.9769569157893598,
      "learning_rate": 0.00018899594685194493,
      "loss": 13.082,
      "step": 4031
    },
    {
      "epoch": 0.7050491803278689,
      "grad_norm": 0.8791806984475703,
      "learning_rate": 0.00018898948803423531,
      "loss": 12.9697,
      "step": 4032
    },
    {
      "epoch": 0.705224043715847,
      "grad_norm": 1.0834198517413893,
      "learning_rate": 0.00018898302743201322,
      "loss": 13.0974,
      "step": 4033
    },
    {
      "epoch": 0.7053989071038251,
      "grad_norm": 0.914323046656777,
      "learning_rate": 0.00018897656504540818,
      "loss": 13.1156,
      "step": 4034
    },
    {
      "epoch": 0.7055737704918033,
      "grad_norm": 0.839743319352127,
      "learning_rate": 0.00018897010087454987,
      "loss": 13.232,
      "step": 4035
    },
    {
      "epoch": 0.7057486338797814,
      "grad_norm": 0.8773923732421085,
      "learning_rate": 0.00018896363491956785,
      "loss": 12.7421,
      "step": 4036
    },
    {
      "epoch": 0.7059234972677596,
      "grad_norm": 0.9747824251475596,
      "learning_rate": 0.00018895716718059179,
      "loss": 12.8512,
      "step": 4037
    },
    {
      "epoch": 0.7060983606557377,
      "grad_norm": 0.925341658543888,
      "learning_rate": 0.0001889506976577514,
      "loss": 13.2352,
      "step": 4038
    },
    {
      "epoch": 0.7062732240437158,
      "grad_norm": 0.9252328034699986,
      "learning_rate": 0.00018894422635117638,
      "loss": 13.2631,
      "step": 4039
    },
    {
      "epoch": 0.706448087431694,
      "grad_norm": 0.9483755744918804,
      "learning_rate": 0.00018893775326099657,
      "loss": 13.2459,
      "step": 4040
    },
    {
      "epoch": 0.7066229508196721,
      "grad_norm": 0.9327835357682973,
      "learning_rate": 0.00018893127838734167,
      "loss": 12.9235,
      "step": 4041
    },
    {
      "epoch": 0.7067978142076503,
      "grad_norm": 1.1177137586384875,
      "learning_rate": 0.00018892480173034163,
      "loss": 13.2217,
      "step": 4042
    },
    {
      "epoch": 0.7069726775956284,
      "grad_norm": 0.8523522238558175,
      "learning_rate": 0.00018891832329012627,
      "loss": 12.979,
      "step": 4043
    },
    {
      "epoch": 0.7071475409836065,
      "grad_norm": 0.8727799507349638,
      "learning_rate": 0.00018891184306682552,
      "loss": 13.1845,
      "step": 4044
    },
    {
      "epoch": 0.7073224043715847,
      "grad_norm": 0.9028837546513504,
      "learning_rate": 0.00018890536106056934,
      "loss": 13.1894,
      "step": 4045
    },
    {
      "epoch": 0.7074972677595628,
      "grad_norm": 0.8901745718352507,
      "learning_rate": 0.00018889887727148764,
      "loss": 13.205,
      "step": 4046
    },
    {
      "epoch": 0.707672131147541,
      "grad_norm": 0.8648231778523332,
      "learning_rate": 0.00018889239169971053,
      "loss": 13.058,
      "step": 4047
    },
    {
      "epoch": 0.7078469945355191,
      "grad_norm": 0.9224083681267404,
      "learning_rate": 0.00018888590434536803,
      "loss": 13.231,
      "step": 4048
    },
    {
      "epoch": 0.7080218579234973,
      "grad_norm": 0.822145934989213,
      "learning_rate": 0.00018887941520859023,
      "loss": 13.1446,
      "step": 4049
    },
    {
      "epoch": 0.7081967213114754,
      "grad_norm": 0.982632552619366,
      "learning_rate": 0.00018887292428950727,
      "loss": 12.9542,
      "step": 4050
    },
    {
      "epoch": 0.7083715846994536,
      "grad_norm": 0.9302986006933008,
      "learning_rate": 0.0001888664315882493,
      "loss": 13.136,
      "step": 4051
    },
    {
      "epoch": 0.7085464480874317,
      "grad_norm": 0.9606468433772013,
      "learning_rate": 0.0001888599371049465,
      "loss": 13.2028,
      "step": 4052
    },
    {
      "epoch": 0.7087213114754098,
      "grad_norm": 1.047266333483547,
      "learning_rate": 0.00018885344083972914,
      "loss": 13.3011,
      "step": 4053
    },
    {
      "epoch": 0.708896174863388,
      "grad_norm": 0.8914787094924358,
      "learning_rate": 0.0001888469427927275,
      "loss": 12.9293,
      "step": 4054
    },
    {
      "epoch": 0.7090710382513661,
      "grad_norm": 0.8115193349798684,
      "learning_rate": 0.0001888404429640718,
      "loss": 13.2744,
      "step": 4055
    },
    {
      "epoch": 0.7092459016393443,
      "grad_norm": 0.8580435165493602,
      "learning_rate": 0.00018883394135389247,
      "loss": 12.9949,
      "step": 4056
    },
    {
      "epoch": 0.7094207650273224,
      "grad_norm": 0.9403575651764684,
      "learning_rate": 0.00018882743796231987,
      "loss": 12.7854,
      "step": 4057
    },
    {
      "epoch": 0.7095956284153005,
      "grad_norm": 0.8405701962342029,
      "learning_rate": 0.0001888209327894844,
      "loss": 12.9883,
      "step": 4058
    },
    {
      "epoch": 0.7097704918032787,
      "grad_norm": 0.8823909070173216,
      "learning_rate": 0.00018881442583551652,
      "loss": 12.8116,
      "step": 4059
    },
    {
      "epoch": 0.7099453551912568,
      "grad_norm": 0.836229694813897,
      "learning_rate": 0.0001888079171005467,
      "loss": 13.0348,
      "step": 4060
    },
    {
      "epoch": 0.710120218579235,
      "grad_norm": 0.8503097800952055,
      "learning_rate": 0.00018880140658470546,
      "loss": 12.9581,
      "step": 4061
    },
    {
      "epoch": 0.7102950819672131,
      "grad_norm": 0.8799587825656612,
      "learning_rate": 0.00018879489428812334,
      "loss": 13.2174,
      "step": 4062
    },
    {
      "epoch": 0.7104699453551913,
      "grad_norm": 0.8482848395611124,
      "learning_rate": 0.000188788380210931,
      "loss": 13.1607,
      "step": 4063
    },
    {
      "epoch": 0.7106448087431694,
      "grad_norm": 0.8086602077870227,
      "learning_rate": 0.000188781864353259,
      "loss": 13.0804,
      "step": 4064
    },
    {
      "epoch": 0.7108196721311475,
      "grad_norm": 0.7855661642282339,
      "learning_rate": 0.000188775346715238,
      "loss": 13.1755,
      "step": 4065
    },
    {
      "epoch": 0.7109945355191257,
      "grad_norm": 0.8989511846942084,
      "learning_rate": 0.00018876882729699875,
      "loss": 13.0107,
      "step": 4066
    },
    {
      "epoch": 0.7111693989071038,
      "grad_norm": 0.8711306682451498,
      "learning_rate": 0.00018876230609867197,
      "loss": 13.1977,
      "step": 4067
    },
    {
      "epoch": 0.711344262295082,
      "grad_norm": 0.8113596024273556,
      "learning_rate": 0.00018875578312038837,
      "loss": 13.0582,
      "step": 4068
    },
    {
      "epoch": 0.7115191256830601,
      "grad_norm": 0.8949630394288852,
      "learning_rate": 0.00018874925836227883,
      "loss": 13.0772,
      "step": 4069
    },
    {
      "epoch": 0.7116939890710382,
      "grad_norm": 0.9303259556649619,
      "learning_rate": 0.00018874273182447417,
      "loss": 13.224,
      "step": 4070
    },
    {
      "epoch": 0.7118688524590164,
      "grad_norm": 0.9462174574997743,
      "learning_rate": 0.00018873620350710527,
      "loss": 13.3634,
      "step": 4071
    },
    {
      "epoch": 0.7120437158469946,
      "grad_norm": 0.840039556061599,
      "learning_rate": 0.000188729673410303,
      "loss": 13.3452,
      "step": 4072
    },
    {
      "epoch": 0.7122185792349727,
      "grad_norm": 0.8547601520504886,
      "learning_rate": 0.00018872314153419836,
      "loss": 13.2487,
      "step": 4073
    },
    {
      "epoch": 0.7123934426229508,
      "grad_norm": 0.9362443967122622,
      "learning_rate": 0.00018871660787892234,
      "loss": 13.1427,
      "step": 4074
    },
    {
      "epoch": 0.7125683060109289,
      "grad_norm": 0.8575726378358111,
      "learning_rate": 0.0001887100724446059,
      "loss": 13.0088,
      "step": 4075
    },
    {
      "epoch": 0.7127431693989071,
      "grad_norm": 0.8417431880070025,
      "learning_rate": 0.00018870353523138013,
      "loss": 13.3757,
      "step": 4076
    },
    {
      "epoch": 0.7129180327868853,
      "grad_norm": 0.8676329582202514,
      "learning_rate": 0.00018869699623937615,
      "loss": 13.2029,
      "step": 4077
    },
    {
      "epoch": 0.7130928961748634,
      "grad_norm": 0.8028039110496785,
      "learning_rate": 0.00018869045546872503,
      "loss": 13.2027,
      "step": 4078
    },
    {
      "epoch": 0.7132677595628415,
      "grad_norm": 0.9791503011646082,
      "learning_rate": 0.000188683912919558,
      "loss": 13.3061,
      "step": 4079
    },
    {
      "epoch": 0.7134426229508197,
      "grad_norm": 0.9407625672174644,
      "learning_rate": 0.0001886773685920062,
      "loss": 13.2957,
      "step": 4080
    },
    {
      "epoch": 0.7136174863387978,
      "grad_norm": 0.9608079317758326,
      "learning_rate": 0.00018867082248620086,
      "loss": 13.2443,
      "step": 4081
    },
    {
      "epoch": 0.713792349726776,
      "grad_norm": 0.9854457143470297,
      "learning_rate": 0.00018866427460227333,
      "loss": 13.2684,
      "step": 4082
    },
    {
      "epoch": 0.7139672131147541,
      "grad_norm": 0.9988116488775629,
      "learning_rate": 0.0001886577249403548,
      "loss": 13.0486,
      "step": 4083
    },
    {
      "epoch": 0.7141420765027322,
      "grad_norm": 0.8806571933210828,
      "learning_rate": 0.00018865117350057672,
      "loss": 13.3031,
      "step": 4084
    },
    {
      "epoch": 0.7143169398907104,
      "grad_norm": 0.996303360792958,
      "learning_rate": 0.0001886446202830704,
      "loss": 13.3851,
      "step": 4085
    },
    {
      "epoch": 0.7144918032786886,
      "grad_norm": 0.8947900455154818,
      "learning_rate": 0.00018863806528796726,
      "loss": 13.128,
      "step": 4086
    },
    {
      "epoch": 0.7146666666666667,
      "grad_norm": 0.9230731256989211,
      "learning_rate": 0.00018863150851539877,
      "loss": 13.058,
      "step": 4087
    },
    {
      "epoch": 0.7148415300546448,
      "grad_norm": 0.9912041660922319,
      "learning_rate": 0.00018862494996549638,
      "loss": 13.1745,
      "step": 4088
    },
    {
      "epoch": 0.7150163934426229,
      "grad_norm": 0.9393766467087403,
      "learning_rate": 0.00018861838963839164,
      "loss": 13.4445,
      "step": 4089
    },
    {
      "epoch": 0.7151912568306011,
      "grad_norm": 0.8886029156113783,
      "learning_rate": 0.00018861182753421608,
      "loss": 13.2328,
      "step": 4090
    },
    {
      "epoch": 0.7153661202185793,
      "grad_norm": 0.9446451428101212,
      "learning_rate": 0.0001886052636531013,
      "loss": 13.2427,
      "step": 4091
    },
    {
      "epoch": 0.7155409836065574,
      "grad_norm": 0.8961810678939895,
      "learning_rate": 0.00018859869799517896,
      "loss": 13.1897,
      "step": 4092
    },
    {
      "epoch": 0.7157158469945355,
      "grad_norm": 0.8702249073729101,
      "learning_rate": 0.0001885921305605807,
      "loss": 13.2502,
      "step": 4093
    },
    {
      "epoch": 0.7158907103825136,
      "grad_norm": 0.927553323592842,
      "learning_rate": 0.0001885855613494382,
      "loss": 13.1401,
      "step": 4094
    },
    {
      "epoch": 0.7160655737704918,
      "grad_norm": 1.0144600402834267,
      "learning_rate": 0.00018857899036188317,
      "loss": 13.0012,
      "step": 4095
    },
    {
      "epoch": 0.71624043715847,
      "grad_norm": 0.9071186653354006,
      "learning_rate": 0.0001885724175980474,
      "loss": 13.1873,
      "step": 4096
    },
    {
      "epoch": 0.7164153005464481,
      "grad_norm": 0.8923956567917745,
      "learning_rate": 0.00018856584305806275,
      "loss": 13.2419,
      "step": 4097
    },
    {
      "epoch": 0.7165901639344262,
      "grad_norm": 0.9119793426631168,
      "learning_rate": 0.00018855926674206098,
      "loss": 13.0153,
      "step": 4098
    },
    {
      "epoch": 0.7167650273224043,
      "grad_norm": 0.8411532622571355,
      "learning_rate": 0.00018855268865017403,
      "loss": 12.7455,
      "step": 4099
    },
    {
      "epoch": 0.7169398907103826,
      "grad_norm": 0.8511190398245823,
      "learning_rate": 0.00018854610878253376,
      "loss": 13.2411,
      "step": 4100
    },
    {
      "epoch": 0.7171147540983607,
      "grad_norm": 0.9930259386004845,
      "learning_rate": 0.00018853952713927214,
      "loss": 13.3391,
      "step": 4101
    },
    {
      "epoch": 0.7172896174863388,
      "grad_norm": 0.8350595933689491,
      "learning_rate": 0.00018853294372052115,
      "loss": 13.3614,
      "step": 4102
    },
    {
      "epoch": 0.7174644808743169,
      "grad_norm": 0.8359939445682155,
      "learning_rate": 0.0001885263585264128,
      "loss": 13.1074,
      "step": 4103
    },
    {
      "epoch": 0.717639344262295,
      "grad_norm": 0.7831555619148708,
      "learning_rate": 0.00018851977155707916,
      "loss": 12.9711,
      "step": 4104
    },
    {
      "epoch": 0.7178142076502733,
      "grad_norm": 0.8759415309731127,
      "learning_rate": 0.00018851318281265232,
      "loss": 13.158,
      "step": 4105
    },
    {
      "epoch": 0.7179890710382514,
      "grad_norm": 0.8983033370816691,
      "learning_rate": 0.00018850659229326436,
      "loss": 13.3474,
      "step": 4106
    },
    {
      "epoch": 0.7181639344262295,
      "grad_norm": 0.9155935988239491,
      "learning_rate": 0.0001884999999990475,
      "loss": 13.0786,
      "step": 4107
    },
    {
      "epoch": 0.7183387978142076,
      "grad_norm": 0.7613722541002295,
      "learning_rate": 0.00018849340593013388,
      "loss": 13.1822,
      "step": 4108
    },
    {
      "epoch": 0.7185136612021857,
      "grad_norm": 0.8601729657383342,
      "learning_rate": 0.00018848681008665582,
      "loss": 13.202,
      "step": 4109
    },
    {
      "epoch": 0.718688524590164,
      "grad_norm": 0.9584281411978518,
      "learning_rate": 0.0001884802124687455,
      "loss": 13.2381,
      "step": 4110
    },
    {
      "epoch": 0.7188633879781421,
      "grad_norm": 0.8195284053060873,
      "learning_rate": 0.0001884736130765352,
      "loss": 13.0821,
      "step": 4111
    },
    {
      "epoch": 0.7190382513661202,
      "grad_norm": 0.9870414646984293,
      "learning_rate": 0.00018846701191015737,
      "loss": 13.3594,
      "step": 4112
    },
    {
      "epoch": 0.7192131147540983,
      "grad_norm": 0.8666094886863844,
      "learning_rate": 0.0001884604089697443,
      "loss": 13.1504,
      "step": 4113
    },
    {
      "epoch": 0.7193879781420766,
      "grad_norm": 0.9299889607175971,
      "learning_rate": 0.00018845380425542842,
      "loss": 13.239,
      "step": 4114
    },
    {
      "epoch": 0.7195628415300547,
      "grad_norm": 0.8884827328719773,
      "learning_rate": 0.0001884471977673422,
      "loss": 13.1886,
      "step": 4115
    },
    {
      "epoch": 0.7197377049180328,
      "grad_norm": 0.9913914769528728,
      "learning_rate": 0.00018844058950561805,
      "loss": 13.2157,
      "step": 4116
    },
    {
      "epoch": 0.7199125683060109,
      "grad_norm": 0.9145928929817774,
      "learning_rate": 0.00018843397947038856,
      "loss": 13.1536,
      "step": 4117
    },
    {
      "epoch": 0.720087431693989,
      "grad_norm": 0.9857106189278774,
      "learning_rate": 0.00018842736766178626,
      "loss": 13.0499,
      "step": 4118
    },
    {
      "epoch": 0.7202622950819673,
      "grad_norm": 1.0727982878973163,
      "learning_rate": 0.0001884207540799437,
      "loss": 13.0112,
      "step": 4119
    },
    {
      "epoch": 0.7204371584699454,
      "grad_norm": 0.8737766712127921,
      "learning_rate": 0.00018841413872499354,
      "loss": 13.1692,
      "step": 4120
    },
    {
      "epoch": 0.7206120218579235,
      "grad_norm": 0.9079614816963495,
      "learning_rate": 0.00018840752159706846,
      "loss": 13.058,
      "step": 4121
    },
    {
      "epoch": 0.7207868852459016,
      "grad_norm": 1.028898587215858,
      "learning_rate": 0.00018840090269630112,
      "loss": 12.8295,
      "step": 4122
    },
    {
      "epoch": 0.7209617486338797,
      "grad_norm": 0.9072894022361203,
      "learning_rate": 0.00018839428202282424,
      "loss": 13.2487,
      "step": 4123
    },
    {
      "epoch": 0.721136612021858,
      "grad_norm": 0.9231376140449741,
      "learning_rate": 0.0001883876595767706,
      "loss": 13.2135,
      "step": 4124
    },
    {
      "epoch": 0.7213114754098361,
      "grad_norm": 0.9626194119168926,
      "learning_rate": 0.00018838103535827297,
      "loss": 13.1232,
      "step": 4125
    },
    {
      "epoch": 0.7214863387978142,
      "grad_norm": 0.9264748841466092,
      "learning_rate": 0.0001883744093674643,
      "loss": 13.1319,
      "step": 4126
    },
    {
      "epoch": 0.7216612021857923,
      "grad_norm": 0.9218579433083265,
      "learning_rate": 0.0001883677816044773,
      "loss": 13.0367,
      "step": 4127
    },
    {
      "epoch": 0.7218360655737704,
      "grad_norm": 0.8123352244245288,
      "learning_rate": 0.000188361152069445,
      "loss": 13.3305,
      "step": 4128
    },
    {
      "epoch": 0.7220109289617487,
      "grad_norm": 1.1865481083745673,
      "learning_rate": 0.00018835452076250027,
      "loss": 13.037,
      "step": 4129
    },
    {
      "epoch": 0.7221857923497268,
      "grad_norm": 1.0513277448222367,
      "learning_rate": 0.00018834788768377613,
      "loss": 12.8513,
      "step": 4130
    },
    {
      "epoch": 0.7223606557377049,
      "grad_norm": 0.8690949881605594,
      "learning_rate": 0.00018834125283340557,
      "loss": 13.2385,
      "step": 4131
    },
    {
      "epoch": 0.722535519125683,
      "grad_norm": 0.8814208525066318,
      "learning_rate": 0.00018833461621152163,
      "loss": 13.1235,
      "step": 4132
    },
    {
      "epoch": 0.7227103825136612,
      "grad_norm": 0.8690416375645997,
      "learning_rate": 0.00018832797781825745,
      "loss": 12.9195,
      "step": 4133
    },
    {
      "epoch": 0.7228852459016394,
      "grad_norm": 1.0002640323745942,
      "learning_rate": 0.00018832133765374606,
      "loss": 13.1912,
      "step": 4134
    },
    {
      "epoch": 0.7230601092896175,
      "grad_norm": 0.8779230991148442,
      "learning_rate": 0.00018831469571812073,
      "loss": 13.2114,
      "step": 4135
    },
    {
      "epoch": 0.7232349726775956,
      "grad_norm": 0.8228912702216339,
      "learning_rate": 0.00018830805201151455,
      "loss": 13.0434,
      "step": 4136
    },
    {
      "epoch": 0.7234098360655737,
      "grad_norm": 0.8991032198428323,
      "learning_rate": 0.0001883014065340608,
      "loss": 13.2105,
      "step": 4137
    },
    {
      "epoch": 0.7235846994535519,
      "grad_norm": 0.858837100749513,
      "learning_rate": 0.00018829475928589271,
      "loss": 13.2342,
      "step": 4138
    },
    {
      "epoch": 0.7237595628415301,
      "grad_norm": 0.8793974836223738,
      "learning_rate": 0.0001882881102671436,
      "loss": 13.3582,
      "step": 4139
    },
    {
      "epoch": 0.7239344262295082,
      "grad_norm": 0.9192874088637081,
      "learning_rate": 0.00018828145947794682,
      "loss": 13.1924,
      "step": 4140
    },
    {
      "epoch": 0.7241092896174863,
      "grad_norm": 0.7946399383484449,
      "learning_rate": 0.00018827480691843568,
      "loss": 12.9691,
      "step": 4141
    },
    {
      "epoch": 0.7242841530054644,
      "grad_norm": 0.8716230496849169,
      "learning_rate": 0.00018826815258874365,
      "loss": 13.3383,
      "step": 4142
    },
    {
      "epoch": 0.7244590163934426,
      "grad_norm": 0.8168608846828802,
      "learning_rate": 0.00018826149648900416,
      "loss": 13.0569,
      "step": 4143
    },
    {
      "epoch": 0.7246338797814208,
      "grad_norm": 0.9215069702666782,
      "learning_rate": 0.00018825483861935064,
      "loss": 13.1387,
      "step": 4144
    },
    {
      "epoch": 0.7248087431693989,
      "grad_norm": 0.9451651445548762,
      "learning_rate": 0.00018824817897991665,
      "loss": 13.2194,
      "step": 4145
    },
    {
      "epoch": 0.724983606557377,
      "grad_norm": 0.8667337297761897,
      "learning_rate": 0.00018824151757083568,
      "loss": 13.071,
      "step": 4146
    },
    {
      "epoch": 0.7251584699453552,
      "grad_norm": 0.9716498458434871,
      "learning_rate": 0.00018823485439224138,
      "loss": 13.2071,
      "step": 4147
    },
    {
      "epoch": 0.7253333333333334,
      "grad_norm": 0.8403778361619276,
      "learning_rate": 0.0001882281894442673,
      "loss": 13.3522,
      "step": 4148
    },
    {
      "epoch": 0.7255081967213115,
      "grad_norm": 0.9016954307281794,
      "learning_rate": 0.00018822152272704714,
      "loss": 12.9031,
      "step": 4149
    },
    {
      "epoch": 0.7256830601092896,
      "grad_norm": 0.8896632290160993,
      "learning_rate": 0.00018821485424071457,
      "loss": 13.3801,
      "step": 4150
    },
    {
      "epoch": 0.7258579234972677,
      "grad_norm": 0.9120534781350197,
      "learning_rate": 0.0001882081839854033,
      "loss": 12.9905,
      "step": 4151
    },
    {
      "epoch": 0.7260327868852459,
      "grad_norm": 1.040183096068941,
      "learning_rate": 0.00018820151196124717,
      "loss": 13.0249,
      "step": 4152
    },
    {
      "epoch": 0.7262076502732241,
      "grad_norm": 0.9008465651760534,
      "learning_rate": 0.00018819483816837984,
      "loss": 12.9735,
      "step": 4153
    },
    {
      "epoch": 0.7263825136612022,
      "grad_norm": 1.0860726576123643,
      "learning_rate": 0.00018818816260693526,
      "loss": 13.1465,
      "step": 4154
    },
    {
      "epoch": 0.7265573770491803,
      "grad_norm": 0.8331734989032478,
      "learning_rate": 0.00018818148527704722,
      "loss": 13.1322,
      "step": 4155
    },
    {
      "epoch": 0.7267322404371584,
      "grad_norm": 0.9202345909279748,
      "learning_rate": 0.00018817480617884966,
      "loss": 13.0463,
      "step": 4156
    },
    {
      "epoch": 0.7269071038251366,
      "grad_norm": 0.9759716066026897,
      "learning_rate": 0.0001881681253124765,
      "loss": 13.3701,
      "step": 4157
    },
    {
      "epoch": 0.7270819672131148,
      "grad_norm": 0.8644055750773849,
      "learning_rate": 0.00018816144267806172,
      "loss": 13.2095,
      "step": 4158
    },
    {
      "epoch": 0.7272568306010929,
      "grad_norm": 0.7955432881733248,
      "learning_rate": 0.0001881547582757393,
      "loss": 12.8967,
      "step": 4159
    },
    {
      "epoch": 0.727431693989071,
      "grad_norm": 0.9625354920478602,
      "learning_rate": 0.00018814807210564333,
      "loss": 13.0421,
      "step": 4160
    },
    {
      "epoch": 0.7276065573770492,
      "grad_norm": 1.0648396195119272,
      "learning_rate": 0.00018814138416790787,
      "loss": 13.2235,
      "step": 4161
    },
    {
      "epoch": 0.7277814207650273,
      "grad_norm": 0.8451609355679205,
      "learning_rate": 0.00018813469446266702,
      "loss": 13.0518,
      "step": 4162
    },
    {
      "epoch": 0.7279562841530055,
      "grad_norm": 1.0255086043697292,
      "learning_rate": 0.0001881280029900549,
      "loss": 13.1074,
      "step": 4163
    },
    {
      "epoch": 0.7281311475409836,
      "grad_norm": 0.9393616610949129,
      "learning_rate": 0.0001881213097502058,
      "loss": 13.3434,
      "step": 4164
    },
    {
      "epoch": 0.7283060109289617,
      "grad_norm": 1.0071862377928318,
      "learning_rate": 0.0001881146147432538,
      "loss": 13.1569,
      "step": 4165
    },
    {
      "epoch": 0.7284808743169399,
      "grad_norm": 0.9403715601426388,
      "learning_rate": 0.0001881079179693333,
      "loss": 13.3171,
      "step": 4166
    },
    {
      "epoch": 0.728655737704918,
      "grad_norm": 1.0515697821045469,
      "learning_rate": 0.00018810121942857845,
      "loss": 13.064,
      "step": 4167
    },
    {
      "epoch": 0.7288306010928962,
      "grad_norm": 1.0465164432694016,
      "learning_rate": 0.00018809451912112367,
      "loss": 13.1813,
      "step": 4168
    },
    {
      "epoch": 0.7290054644808743,
      "grad_norm": 0.9723477334943973,
      "learning_rate": 0.00018808781704710332,
      "loss": 13.2822,
      "step": 4169
    },
    {
      "epoch": 0.7291803278688525,
      "grad_norm": 1.1694332334305277,
      "learning_rate": 0.00018808111320665173,
      "loss": 12.9013,
      "step": 4170
    },
    {
      "epoch": 0.7293551912568306,
      "grad_norm": 0.9155811441531884,
      "learning_rate": 0.00018807440759990342,
      "loss": 13.1518,
      "step": 4171
    },
    {
      "epoch": 0.7295300546448087,
      "grad_norm": 0.913282197768497,
      "learning_rate": 0.00018806770022699278,
      "loss": 13.1985,
      "step": 4172
    },
    {
      "epoch": 0.7297049180327869,
      "grad_norm": 0.8888048670202967,
      "learning_rate": 0.00018806099108805438,
      "loss": 13.1056,
      "step": 4173
    },
    {
      "epoch": 0.729879781420765,
      "grad_norm": 1.071349854039852,
      "learning_rate": 0.0001880542801832227,
      "loss": 13.3768,
      "step": 4174
    },
    {
      "epoch": 0.7300546448087432,
      "grad_norm": 0.9486334314945027,
      "learning_rate": 0.00018804756751263234,
      "loss": 12.9683,
      "step": 4175
    },
    {
      "epoch": 0.7302295081967213,
      "grad_norm": 1.0915730825226013,
      "learning_rate": 0.00018804085307641793,
      "loss": 13.0725,
      "step": 4176
    },
    {
      "epoch": 0.7304043715846995,
      "grad_norm": 0.9607005050004059,
      "learning_rate": 0.00018803413687471407,
      "loss": 13.1583,
      "step": 4177
    },
    {
      "epoch": 0.7305792349726776,
      "grad_norm": 1.057742082080554,
      "learning_rate": 0.00018802741890765548,
      "loss": 13.1585,
      "step": 4178
    },
    {
      "epoch": 0.7307540983606557,
      "grad_norm": 1.1098636040138838,
      "learning_rate": 0.00018802069917537686,
      "loss": 13.1816,
      "step": 4179
    },
    {
      "epoch": 0.7309289617486339,
      "grad_norm": 0.9513710435606801,
      "learning_rate": 0.00018801397767801293,
      "loss": 13.0174,
      "step": 4180
    },
    {
      "epoch": 0.731103825136612,
      "grad_norm": 1.068338531351298,
      "learning_rate": 0.00018800725441569854,
      "loss": 13.3381,
      "step": 4181
    },
    {
      "epoch": 0.7312786885245902,
      "grad_norm": 0.9490751121274508,
      "learning_rate": 0.00018800052938856848,
      "loss": 13.0429,
      "step": 4182
    },
    {
      "epoch": 0.7314535519125683,
      "grad_norm": 1.1073178299466282,
      "learning_rate": 0.00018799380259675762,
      "loss": 13.1225,
      "step": 4183
    },
    {
      "epoch": 0.7316284153005465,
      "grad_norm": 0.8708380057056248,
      "learning_rate": 0.00018798707404040085,
      "loss": 13.1583,
      "step": 4184
    },
    {
      "epoch": 0.7318032786885246,
      "grad_norm": 1.2813734714229172,
      "learning_rate": 0.00018798034371963308,
      "loss": 12.9818,
      "step": 4185
    },
    {
      "epoch": 0.7319781420765027,
      "grad_norm": 0.8962543273360708,
      "learning_rate": 0.00018797361163458925,
      "loss": 13.0905,
      "step": 4186
    },
    {
      "epoch": 0.7321530054644809,
      "grad_norm": 1.027364580060604,
      "learning_rate": 0.00018796687778540442,
      "loss": 13.249,
      "step": 4187
    },
    {
      "epoch": 0.732327868852459,
      "grad_norm": 0.9637749169018156,
      "learning_rate": 0.0001879601421722136,
      "loss": 13.1466,
      "step": 4188
    },
    {
      "epoch": 0.7325027322404372,
      "grad_norm": 0.9472337140389634,
      "learning_rate": 0.00018795340479515183,
      "loss": 12.9988,
      "step": 4189
    },
    {
      "epoch": 0.7326775956284153,
      "grad_norm": 0.8668623803146764,
      "learning_rate": 0.00018794666565435426,
      "loss": 13.1826,
      "step": 4190
    },
    {
      "epoch": 0.7328524590163934,
      "grad_norm": 0.9707227855078788,
      "learning_rate": 0.00018793992474995603,
      "loss": 13.1045,
      "step": 4191
    },
    {
      "epoch": 0.7330273224043716,
      "grad_norm": 0.926451504105174,
      "learning_rate": 0.00018793318208209226,
      "loss": 13.3966,
      "step": 4192
    },
    {
      "epoch": 0.7332021857923497,
      "grad_norm": 0.8686678679342976,
      "learning_rate": 0.00018792643765089823,
      "loss": 13.0134,
      "step": 4193
    },
    {
      "epoch": 0.7333770491803279,
      "grad_norm": 0.9071025203261227,
      "learning_rate": 0.00018791969145650917,
      "loss": 12.9044,
      "step": 4194
    },
    {
      "epoch": 0.733551912568306,
      "grad_norm": 0.9403530409855069,
      "learning_rate": 0.0001879129434990603,
      "loss": 13.1938,
      "step": 4195
    },
    {
      "epoch": 0.7337267759562841,
      "grad_norm": 0.982398202742574,
      "learning_rate": 0.00018790619377868703,
      "loss": 13.0783,
      "step": 4196
    },
    {
      "epoch": 0.7339016393442623,
      "grad_norm": 0.8560179498660967,
      "learning_rate": 0.0001878994422955246,
      "loss": 12.9888,
      "step": 4197
    },
    {
      "epoch": 0.7340765027322405,
      "grad_norm": 1.0673321651634673,
      "learning_rate": 0.00018789268904970854,
      "loss": 13.0828,
      "step": 4198
    },
    {
      "epoch": 0.7342513661202186,
      "grad_norm": 0.9503378356924301,
      "learning_rate": 0.00018788593404137416,
      "loss": 13.1468,
      "step": 4199
    },
    {
      "epoch": 0.7344262295081967,
      "grad_norm": 1.0655686210151933,
      "learning_rate": 0.00018787917727065696,
      "loss": 13.1375,
      "step": 4200
    },
    {
      "epoch": 0.7346010928961748,
      "grad_norm": 0.9189788562457275,
      "learning_rate": 0.00018787241873769242,
      "loss": 13.2716,
      "step": 4201
    },
    {
      "epoch": 0.734775956284153,
      "grad_norm": 0.9772604471768427,
      "learning_rate": 0.0001878656584426161,
      "loss": 13.2066,
      "step": 4202
    },
    {
      "epoch": 0.7349508196721312,
      "grad_norm": 0.9731621882342197,
      "learning_rate": 0.00018785889638556354,
      "loss": 13.2962,
      "step": 4203
    },
    {
      "epoch": 0.7351256830601093,
      "grad_norm": 0.9673516345980977,
      "learning_rate": 0.00018785213256667036,
      "loss": 13.1488,
      "step": 4204
    },
    {
      "epoch": 0.7353005464480874,
      "grad_norm": 0.9131277401631698,
      "learning_rate": 0.00018784536698607216,
      "loss": 13.2238,
      "step": 4205
    },
    {
      "epoch": 0.7354754098360655,
      "grad_norm": 0.8300784596340204,
      "learning_rate": 0.00018783859964390464,
      "loss": 13.086,
      "step": 4206
    },
    {
      "epoch": 0.7356502732240437,
      "grad_norm": 0.8881139036209553,
      "learning_rate": 0.0001878318305403035,
      "loss": 13.2795,
      "step": 4207
    },
    {
      "epoch": 0.7358251366120219,
      "grad_norm": 0.9220868576862524,
      "learning_rate": 0.00018782505967540448,
      "loss": 13.3582,
      "step": 4208
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.8514948746702287,
      "learning_rate": 0.00018781828704934336,
      "loss": 13.1132,
      "step": 4209
    },
    {
      "epoch": 0.7361748633879781,
      "grad_norm": 1.0243779953090797,
      "learning_rate": 0.00018781151266225593,
      "loss": 13.3603,
      "step": 4210
    },
    {
      "epoch": 0.7363497267759563,
      "grad_norm": 0.8651795484584337,
      "learning_rate": 0.00018780473651427805,
      "loss": 13.1938,
      "step": 4211
    },
    {
      "epoch": 0.7365245901639345,
      "grad_norm": 0.8821535243822647,
      "learning_rate": 0.0001877979586055456,
      "loss": 13.025,
      "step": 4212
    },
    {
      "epoch": 0.7366994535519126,
      "grad_norm": 0.9538333644534484,
      "learning_rate": 0.00018779117893619453,
      "loss": 13.2321,
      "step": 4213
    },
    {
      "epoch": 0.7368743169398907,
      "grad_norm": 0.8317512400234746,
      "learning_rate": 0.00018778439750636076,
      "loss": 13.1514,
      "step": 4214
    },
    {
      "epoch": 0.7370491803278688,
      "grad_norm": 0.9180733140783203,
      "learning_rate": 0.0001877776143161803,
      "loss": 13.3312,
      "step": 4215
    },
    {
      "epoch": 0.737224043715847,
      "grad_norm": 0.9564182107178075,
      "learning_rate": 0.00018777082936578915,
      "loss": 13.2338,
      "step": 4216
    },
    {
      "epoch": 0.7373989071038252,
      "grad_norm": 1.0232371311805675,
      "learning_rate": 0.00018776404265532338,
      "loss": 13.268,
      "step": 4217
    },
    {
      "epoch": 0.7375737704918033,
      "grad_norm": 0.8776796086748511,
      "learning_rate": 0.00018775725418491908,
      "loss": 13.104,
      "step": 4218
    },
    {
      "epoch": 0.7377486338797814,
      "grad_norm": 0.8478624166348283,
      "learning_rate": 0.00018775046395471239,
      "loss": 13.0758,
      "step": 4219
    },
    {
      "epoch": 0.7379234972677595,
      "grad_norm": 0.8948195345618628,
      "learning_rate": 0.00018774367196483947,
      "loss": 13.2798,
      "step": 4220
    },
    {
      "epoch": 0.7380983606557378,
      "grad_norm": 0.8390927062547622,
      "learning_rate": 0.0001877368782154365,
      "loss": 13.0128,
      "step": 4221
    },
    {
      "epoch": 0.7382732240437159,
      "grad_norm": 0.8591100629207741,
      "learning_rate": 0.00018773008270663973,
      "loss": 13.0443,
      "step": 4222
    },
    {
      "epoch": 0.738448087431694,
      "grad_norm": 0.8821405675123234,
      "learning_rate": 0.00018772328543858547,
      "loss": 13.0411,
      "step": 4223
    },
    {
      "epoch": 0.7386229508196721,
      "grad_norm": 0.863333272710549,
      "learning_rate": 0.00018771648641140995,
      "loss": 12.943,
      "step": 4224
    },
    {
      "epoch": 0.7387978142076502,
      "grad_norm": 1.0368840336817875,
      "learning_rate": 0.0001877096856252496,
      "loss": 13.2329,
      "step": 4225
    },
    {
      "epoch": 0.7389726775956285,
      "grad_norm": 0.9391374652426079,
      "learning_rate": 0.0001877028830802407,
      "loss": 13.1824,
      "step": 4226
    },
    {
      "epoch": 0.7391475409836066,
      "grad_norm": 0.9266466629215173,
      "learning_rate": 0.00018769607877651976,
      "loss": 13.1698,
      "step": 4227
    },
    {
      "epoch": 0.7393224043715847,
      "grad_norm": 0.8382140098110991,
      "learning_rate": 0.0001876892727142231,
      "loss": 13.0374,
      "step": 4228
    },
    {
      "epoch": 0.7394972677595628,
      "grad_norm": 0.8524043201752244,
      "learning_rate": 0.00018768246489348736,
      "loss": 12.9978,
      "step": 4229
    },
    {
      "epoch": 0.7396721311475409,
      "grad_norm": 0.9409742502717178,
      "learning_rate": 0.00018767565531444897,
      "loss": 13.3517,
      "step": 4230
    },
    {
      "epoch": 0.7398469945355192,
      "grad_norm": 0.8508114492304483,
      "learning_rate": 0.00018766884397724445,
      "loss": 13.1587,
      "step": 4231
    },
    {
      "epoch": 0.7400218579234973,
      "grad_norm": 0.8541432963344809,
      "learning_rate": 0.00018766203088201047,
      "loss": 13.2473,
      "step": 4232
    },
    {
      "epoch": 0.7401967213114754,
      "grad_norm": 0.8289483030735648,
      "learning_rate": 0.0001876552160288836,
      "loss": 13.0297,
      "step": 4233
    },
    {
      "epoch": 0.7403715846994535,
      "grad_norm": 0.8515648205790185,
      "learning_rate": 0.0001876483994180005,
      "loss": 13.2113,
      "step": 4234
    },
    {
      "epoch": 0.7405464480874316,
      "grad_norm": 0.7512023815432122,
      "learning_rate": 0.00018764158104949788,
      "loss": 13.1295,
      "step": 4235
    },
    {
      "epoch": 0.7407213114754099,
      "grad_norm": 0.8930573241216395,
      "learning_rate": 0.00018763476092351246,
      "loss": 13.0178,
      "step": 4236
    },
    {
      "epoch": 0.740896174863388,
      "grad_norm": 0.8118864575249483,
      "learning_rate": 0.00018762793904018103,
      "loss": 12.9444,
      "step": 4237
    },
    {
      "epoch": 0.7410710382513661,
      "grad_norm": 0.9112127844974662,
      "learning_rate": 0.00018762111539964032,
      "loss": 13.1221,
      "step": 4238
    },
    {
      "epoch": 0.7412459016393442,
      "grad_norm": 0.9276958252005181,
      "learning_rate": 0.00018761429000202728,
      "loss": 13.1241,
      "step": 4239
    },
    {
      "epoch": 0.7414207650273225,
      "grad_norm": 0.7682524870445818,
      "learning_rate": 0.00018760746284747867,
      "loss": 13.0492,
      "step": 4240
    },
    {
      "epoch": 0.7415956284153006,
      "grad_norm": 0.8594035738832976,
      "learning_rate": 0.00018760063393613146,
      "loss": 13.2161,
      "step": 4241
    },
    {
      "epoch": 0.7417704918032787,
      "grad_norm": 0.8158107288660685,
      "learning_rate": 0.00018759380326812257,
      "loss": 12.8208,
      "step": 4242
    },
    {
      "epoch": 0.7419453551912568,
      "grad_norm": 0.9027628659442992,
      "learning_rate": 0.00018758697084358898,
      "loss": 12.866,
      "step": 4243
    },
    {
      "epoch": 0.7421202185792349,
      "grad_norm": 0.8222137505126516,
      "learning_rate": 0.00018758013666266766,
      "loss": 13.0836,
      "step": 4244
    },
    {
      "epoch": 0.7422950819672132,
      "grad_norm": 0.8786999879822057,
      "learning_rate": 0.00018757330072549574,
      "loss": 13.0885,
      "step": 4245
    },
    {
      "epoch": 0.7424699453551913,
      "grad_norm": 0.831209854917171,
      "learning_rate": 0.00018756646303221024,
      "loss": 12.9152,
      "step": 4246
    },
    {
      "epoch": 0.7426448087431694,
      "grad_norm": 0.895296242026207,
      "learning_rate": 0.00018755962358294828,
      "loss": 13.2401,
      "step": 4247
    },
    {
      "epoch": 0.7428196721311475,
      "grad_norm": 0.9346159878318642,
      "learning_rate": 0.00018755278237784703,
      "loss": 13.2101,
      "step": 4248
    },
    {
      "epoch": 0.7429945355191256,
      "grad_norm": 0.7888289328979712,
      "learning_rate": 0.00018754593941704367,
      "loss": 13.2142,
      "step": 4249
    },
    {
      "epoch": 0.7431693989071039,
      "grad_norm": 1.0563861611524767,
      "learning_rate": 0.00018753909470067544,
      "loss": 13.256,
      "step": 4250
    },
    {
      "epoch": 0.743344262295082,
      "grad_norm": 1.2064614357913752,
      "learning_rate": 0.00018753224822887958,
      "loss": 13.2986,
      "step": 4251
    },
    {
      "epoch": 0.7435191256830601,
      "grad_norm": 0.8184881285042064,
      "learning_rate": 0.00018752540000179337,
      "loss": 12.9527,
      "step": 4252
    },
    {
      "epoch": 0.7436939890710382,
      "grad_norm": 1.106081375728628,
      "learning_rate": 0.00018751855001955414,
      "loss": 12.7378,
      "step": 4253
    },
    {
      "epoch": 0.7438688524590163,
      "grad_norm": 0.8837202879334238,
      "learning_rate": 0.00018751169828229927,
      "loss": 12.9661,
      "step": 4254
    },
    {
      "epoch": 0.7440437158469946,
      "grad_norm": 1.0893144952047846,
      "learning_rate": 0.00018750484479016617,
      "loss": 13.2611,
      "step": 4255
    },
    {
      "epoch": 0.7442185792349727,
      "grad_norm": 1.1271632749985894,
      "learning_rate": 0.00018749798954329223,
      "loss": 13.1652,
      "step": 4256
    },
    {
      "epoch": 0.7443934426229508,
      "grad_norm": 0.8037060593537427,
      "learning_rate": 0.00018749113254181498,
      "loss": 13.2002,
      "step": 4257
    },
    {
      "epoch": 0.7445683060109289,
      "grad_norm": 1.0399745906777838,
      "learning_rate": 0.00018748427378587187,
      "loss": 13.1257,
      "step": 4258
    },
    {
      "epoch": 0.7447431693989071,
      "grad_norm": 0.8278923212993362,
      "learning_rate": 0.00018747741327560045,
      "loss": 13.0032,
      "step": 4259
    },
    {
      "epoch": 0.7449180327868853,
      "grad_norm": 1.121943727563691,
      "learning_rate": 0.00018747055101113832,
      "loss": 13.0259,
      "step": 4260
    },
    {
      "epoch": 0.7450928961748634,
      "grad_norm": 0.9122300649088257,
      "learning_rate": 0.0001874636869926231,
      "loss": 13.0531,
      "step": 4261
    },
    {
      "epoch": 0.7452677595628415,
      "grad_norm": 0.8808590881912942,
      "learning_rate": 0.00018745682122019234,
      "loss": 13.0514,
      "step": 4262
    },
    {
      "epoch": 0.7454426229508196,
      "grad_norm": 0.7969170667976675,
      "learning_rate": 0.00018744995369398383,
      "loss": 12.9744,
      "step": 4263
    },
    {
      "epoch": 0.7456174863387978,
      "grad_norm": 0.925649557864892,
      "learning_rate": 0.00018744308441413522,
      "loss": 12.9232,
      "step": 4264
    },
    {
      "epoch": 0.745792349726776,
      "grad_norm": 0.800238033178706,
      "learning_rate": 0.0001874362133807843,
      "loss": 13.1557,
      "step": 4265
    },
    {
      "epoch": 0.7459672131147541,
      "grad_norm": 0.8713950818199037,
      "learning_rate": 0.00018742934059406884,
      "loss": 12.8537,
      "step": 4266
    },
    {
      "epoch": 0.7461420765027322,
      "grad_norm": 0.8954283709555866,
      "learning_rate": 0.00018742246605412666,
      "loss": 12.6702,
      "step": 4267
    },
    {
      "epoch": 0.7463169398907104,
      "grad_norm": 1.0375667393705137,
      "learning_rate": 0.00018741558976109558,
      "loss": 12.9582,
      "step": 4268
    },
    {
      "epoch": 0.7464918032786885,
      "grad_norm": 0.8714523790052439,
      "learning_rate": 0.00018740871171511357,
      "loss": 12.8486,
      "step": 4269
    },
    {
      "epoch": 0.7466666666666667,
      "grad_norm": 0.9251387315574461,
      "learning_rate": 0.00018740183191631848,
      "loss": 13.0485,
      "step": 4270
    },
    {
      "epoch": 0.7468415300546448,
      "grad_norm": 0.7570216177454931,
      "learning_rate": 0.0001873949503648483,
      "loss": 13.1031,
      "step": 4271
    },
    {
      "epoch": 0.7470163934426229,
      "grad_norm": 1.0510600534827912,
      "learning_rate": 0.00018738806706084104,
      "loss": 13.1105,
      "step": 4272
    },
    {
      "epoch": 0.7471912568306011,
      "grad_norm": 1.031854526286104,
      "learning_rate": 0.00018738118200443472,
      "loss": 13.2429,
      "step": 4273
    },
    {
      "epoch": 0.7473661202185793,
      "grad_norm": 0.893172672230583,
      "learning_rate": 0.00018737429519576738,
      "loss": 13.1407,
      "step": 4274
    },
    {
      "epoch": 0.7475409836065574,
      "grad_norm": 0.9706096581957765,
      "learning_rate": 0.00018736740663497718,
      "loss": 13.1348,
      "step": 4275
    },
    {
      "epoch": 0.7477158469945355,
      "grad_norm": 0.8803893117709871,
      "learning_rate": 0.00018736051632220222,
      "loss": 13.0972,
      "step": 4276
    },
    {
      "epoch": 0.7478907103825136,
      "grad_norm": 1.0762593374412928,
      "learning_rate": 0.00018735362425758065,
      "loss": 12.8972,
      "step": 4277
    },
    {
      "epoch": 0.7480655737704918,
      "grad_norm": 0.8992438519272067,
      "learning_rate": 0.00018734673044125072,
      "loss": 13.307,
      "step": 4278
    },
    {
      "epoch": 0.74824043715847,
      "grad_norm": 1.0010142431563656,
      "learning_rate": 0.00018733983487335065,
      "loss": 12.8625,
      "step": 4279
    },
    {
      "epoch": 0.7484153005464481,
      "grad_norm": 0.787730156196862,
      "learning_rate": 0.00018733293755401873,
      "loss": 12.8798,
      "step": 4280
    },
    {
      "epoch": 0.7485901639344262,
      "grad_norm": 1.0063154569712252,
      "learning_rate": 0.00018732603848339326,
      "loss": 12.88,
      "step": 4281
    },
    {
      "epoch": 0.7487650273224044,
      "grad_norm": 0.7805840327892076,
      "learning_rate": 0.0001873191376616126,
      "loss": 13.1913,
      "step": 4282
    },
    {
      "epoch": 0.7489398907103825,
      "grad_norm": 0.9472841377714645,
      "learning_rate": 0.0001873122350888151,
      "loss": 12.9665,
      "step": 4283
    },
    {
      "epoch": 0.7491147540983607,
      "grad_norm": 0.8171430679172685,
      "learning_rate": 0.00018730533076513923,
      "loss": 13.0142,
      "step": 4284
    },
    {
      "epoch": 0.7492896174863388,
      "grad_norm": 0.9034867131237674,
      "learning_rate": 0.00018729842469072342,
      "loss": 13.2323,
      "step": 4285
    },
    {
      "epoch": 0.7494644808743169,
      "grad_norm": 0.8250696942712231,
      "learning_rate": 0.00018729151686570613,
      "loss": 12.9135,
      "step": 4286
    },
    {
      "epoch": 0.7496393442622951,
      "grad_norm": 0.8718234243261027,
      "learning_rate": 0.0001872846072902259,
      "loss": 13.1528,
      "step": 4287
    },
    {
      "epoch": 0.7498142076502732,
      "grad_norm": 0.8896969932043961,
      "learning_rate": 0.00018727769596442134,
      "loss": 13.2151,
      "step": 4288
    },
    {
      "epoch": 0.7499890710382514,
      "grad_norm": 0.8856813141678269,
      "learning_rate": 0.00018727078288843094,
      "loss": 13.0168,
      "step": 4289
    },
    {
      "epoch": 0.7501639344262295,
      "grad_norm": 0.8158681148704361,
      "learning_rate": 0.00018726386806239344,
      "loss": 13.0119,
      "step": 4290
    },
    {
      "epoch": 0.7503387978142076,
      "grad_norm": 0.8117932484812898,
      "learning_rate": 0.00018725695148644742,
      "loss": 12.7037,
      "step": 4291
    },
    {
      "epoch": 0.7505136612021858,
      "grad_norm": 0.7671385826930722,
      "learning_rate": 0.00018725003316073162,
      "loss": 13.016,
      "step": 4292
    },
    {
      "epoch": 0.7506885245901639,
      "grad_norm": 0.9621444475464843,
      "learning_rate": 0.00018724311308538475,
      "loss": 13.0569,
      "step": 4293
    },
    {
      "epoch": 0.7508633879781421,
      "grad_norm": 0.9940343112834904,
      "learning_rate": 0.0001872361912605456,
      "loss": 13.0566,
      "step": 4294
    },
    {
      "epoch": 0.7510382513661202,
      "grad_norm": 0.8478068556389106,
      "learning_rate": 0.00018722926768635295,
      "loss": 13.0329,
      "step": 4295
    },
    {
      "epoch": 0.7512131147540984,
      "grad_norm": 1.0701208975180572,
      "learning_rate": 0.00018722234236294568,
      "loss": 13.0326,
      "step": 4296
    },
    {
      "epoch": 0.7513879781420765,
      "grad_norm": 0.8943930023075227,
      "learning_rate": 0.00018721541529046262,
      "loss": 13.2262,
      "step": 4297
    },
    {
      "epoch": 0.7515628415300546,
      "grad_norm": 0.8789407665512289,
      "learning_rate": 0.0001872084864690427,
      "loss": 13.0309,
      "step": 4298
    },
    {
      "epoch": 0.7517377049180328,
      "grad_norm": 0.9081629700970265,
      "learning_rate": 0.0001872015558988249,
      "loss": 13.2455,
      "step": 4299
    },
    {
      "epoch": 0.7519125683060109,
      "grad_norm": 0.8532658686642215,
      "learning_rate": 0.0001871946235799481,
      "loss": 13.072,
      "step": 4300
    },
    {
      "epoch": 0.7520874316939891,
      "grad_norm": 0.9414852306147136,
      "learning_rate": 0.00018718768951255142,
      "loss": 13.227,
      "step": 4301
    },
    {
      "epoch": 0.7522622950819672,
      "grad_norm": 0.8324310167130027,
      "learning_rate": 0.0001871807536967738,
      "loss": 13.0198,
      "step": 4302
    },
    {
      "epoch": 0.7524371584699453,
      "grad_norm": 0.8824130248994585,
      "learning_rate": 0.00018717381613275445,
      "loss": 13.1017,
      "step": 4303
    },
    {
      "epoch": 0.7526120218579235,
      "grad_norm": 0.8543331089072418,
      "learning_rate": 0.0001871668768206324,
      "loss": 13.1152,
      "step": 4304
    },
    {
      "epoch": 0.7527868852459016,
      "grad_norm": 0.9323443616077941,
      "learning_rate": 0.00018715993576054685,
      "loss": 13.139,
      "step": 4305
    },
    {
      "epoch": 0.7529617486338798,
      "grad_norm": 0.8682603124197843,
      "learning_rate": 0.00018715299295263694,
      "loss": 13.1443,
      "step": 4306
    },
    {
      "epoch": 0.7531366120218579,
      "grad_norm": 0.7895475852048973,
      "learning_rate": 0.00018714604839704198,
      "loss": 13.0035,
      "step": 4307
    },
    {
      "epoch": 0.7533114754098361,
      "grad_norm": 0.9209956644746143,
      "learning_rate": 0.00018713910209390113,
      "loss": 12.8021,
      "step": 4308
    },
    {
      "epoch": 0.7534863387978142,
      "grad_norm": 1.0006601054670343,
      "learning_rate": 0.00018713215404335377,
      "loss": 13.0422,
      "step": 4309
    },
    {
      "epoch": 0.7536612021857924,
      "grad_norm": 0.8815482152581616,
      "learning_rate": 0.00018712520424553914,
      "loss": 13.056,
      "step": 4310
    },
    {
      "epoch": 0.7538360655737705,
      "grad_norm": 0.9778374843153445,
      "learning_rate": 0.00018711825270059667,
      "loss": 13.2053,
      "step": 4311
    },
    {
      "epoch": 0.7540109289617486,
      "grad_norm": 0.931309710388436,
      "learning_rate": 0.00018711129940866575,
      "loss": 12.8016,
      "step": 4312
    },
    {
      "epoch": 0.7541857923497268,
      "grad_norm": 0.8292360191090439,
      "learning_rate": 0.00018710434436988585,
      "loss": 12.9425,
      "step": 4313
    },
    {
      "epoch": 0.7543606557377049,
      "grad_norm": 0.9223508859751586,
      "learning_rate": 0.00018709738758439635,
      "loss": 13.0561,
      "step": 4314
    },
    {
      "epoch": 0.7545355191256831,
      "grad_norm": 0.9367298695112038,
      "learning_rate": 0.00018709042905233684,
      "loss": 13.1913,
      "step": 4315
    },
    {
      "epoch": 0.7547103825136612,
      "grad_norm": 0.9703742416942936,
      "learning_rate": 0.0001870834687738468,
      "loss": 13.0796,
      "step": 4316
    },
    {
      "epoch": 0.7548852459016393,
      "grad_norm": 0.9005367103352067,
      "learning_rate": 0.00018707650674906583,
      "loss": 12.7756,
      "step": 4317
    },
    {
      "epoch": 0.7550601092896175,
      "grad_norm": 1.0460317939091313,
      "learning_rate": 0.00018706954297813356,
      "loss": 13.171,
      "step": 4318
    },
    {
      "epoch": 0.7552349726775957,
      "grad_norm": 0.9962064037743162,
      "learning_rate": 0.00018706257746118962,
      "loss": 13.1901,
      "step": 4319
    },
    {
      "epoch": 0.7554098360655738,
      "grad_norm": 0.8344597275412756,
      "learning_rate": 0.00018705561019837364,
      "loss": 12.9679,
      "step": 4320
    },
    {
      "epoch": 0.7555846994535519,
      "grad_norm": 0.9486356526519284,
      "learning_rate": 0.0001870486411898254,
      "loss": 13.3124,
      "step": 4321
    },
    {
      "epoch": 0.75575956284153,
      "grad_norm": 0.9065670048544139,
      "learning_rate": 0.00018704167043568463,
      "loss": 13.2862,
      "step": 4322
    },
    {
      "epoch": 0.7559344262295082,
      "grad_norm": 0.8497338765666598,
      "learning_rate": 0.00018703469793609112,
      "loss": 13.03,
      "step": 4323
    },
    {
      "epoch": 0.7561092896174864,
      "grad_norm": 0.8424277689048006,
      "learning_rate": 0.00018702772369118468,
      "loss": 13.0383,
      "step": 4324
    },
    {
      "epoch": 0.7562841530054645,
      "grad_norm": 1.0112430493653484,
      "learning_rate": 0.0001870207477011052,
      "loss": 13.1452,
      "step": 4325
    },
    {
      "epoch": 0.7564590163934426,
      "grad_norm": 0.8816804187509703,
      "learning_rate": 0.0001870137699659925,
      "loss": 13.0032,
      "step": 4326
    },
    {
      "epoch": 0.7566338797814207,
      "grad_norm": 0.8935351297542474,
      "learning_rate": 0.00018700679048598654,
      "loss": 13.0461,
      "step": 4327
    },
    {
      "epoch": 0.756808743169399,
      "grad_norm": 0.9602427916589747,
      "learning_rate": 0.00018699980926122732,
      "loss": 12.8569,
      "step": 4328
    },
    {
      "epoch": 0.7569836065573771,
      "grad_norm": 0.9447867943517606,
      "learning_rate": 0.00018699282629185476,
      "loss": 13.1749,
      "step": 4329
    },
    {
      "epoch": 0.7571584699453552,
      "grad_norm": 0.8808757332715813,
      "learning_rate": 0.00018698584157800897,
      "loss": 12.9118,
      "step": 4330
    },
    {
      "epoch": 0.7573333333333333,
      "grad_norm": 0.9196525035404506,
      "learning_rate": 0.00018697885511982995,
      "loss": 12.9862,
      "step": 4331
    },
    {
      "epoch": 0.7575081967213114,
      "grad_norm": 0.8622177128875647,
      "learning_rate": 0.00018697186691745782,
      "loss": 13.0392,
      "step": 4332
    },
    {
      "epoch": 0.7576830601092897,
      "grad_norm": 0.8450204005129126,
      "learning_rate": 0.00018696487697103276,
      "loss": 12.6551,
      "step": 4333
    },
    {
      "epoch": 0.7578579234972678,
      "grad_norm": 1.0700086347705036,
      "learning_rate": 0.00018695788528069485,
      "loss": 12.851,
      "step": 4334
    },
    {
      "epoch": 0.7580327868852459,
      "grad_norm": 1.132715668016166,
      "learning_rate": 0.0001869508918465844,
      "loss": 13.1076,
      "step": 4335
    },
    {
      "epoch": 0.758207650273224,
      "grad_norm": 0.9131723654414043,
      "learning_rate": 0.00018694389666884154,
      "loss": 12.8533,
      "step": 4336
    },
    {
      "epoch": 0.7583825136612022,
      "grad_norm": 0.9673777023697148,
      "learning_rate": 0.0001869368997476066,
      "loss": 13.0261,
      "step": 4337
    },
    {
      "epoch": 0.7585573770491804,
      "grad_norm": 0.9301763925614568,
      "learning_rate": 0.00018692990108301993,
      "loss": 13.0063,
      "step": 4338
    },
    {
      "epoch": 0.7587322404371585,
      "grad_norm": 0.9848098854676577,
      "learning_rate": 0.00018692290067522183,
      "loss": 12.9874,
      "step": 4339
    },
    {
      "epoch": 0.7589071038251366,
      "grad_norm": 1.0003235419802794,
      "learning_rate": 0.00018691589852435264,
      "loss": 13.1981,
      "step": 4340
    },
    {
      "epoch": 0.7590819672131147,
      "grad_norm": 0.8335065682681365,
      "learning_rate": 0.00018690889463055283,
      "loss": 12.9742,
      "step": 4341
    },
    {
      "epoch": 0.759256830601093,
      "grad_norm": 1.0707007402505337,
      "learning_rate": 0.00018690188899396287,
      "loss": 13.1297,
      "step": 4342
    },
    {
      "epoch": 0.7594316939890711,
      "grad_norm": 0.8446289643224234,
      "learning_rate": 0.00018689488161472318,
      "loss": 12.9898,
      "step": 4343
    },
    {
      "epoch": 0.7596065573770492,
      "grad_norm": 0.9524588555409903,
      "learning_rate": 0.00018688787249297433,
      "loss": 12.8498,
      "step": 4344
    },
    {
      "epoch": 0.7597814207650273,
      "grad_norm": 0.8877164019059847,
      "learning_rate": 0.00018688086162885683,
      "loss": 13.1995,
      "step": 4345
    },
    {
      "epoch": 0.7599562841530054,
      "grad_norm": 0.8776299557674725,
      "learning_rate": 0.00018687384902251134,
      "loss": 12.8962,
      "step": 4346
    },
    {
      "epoch": 0.7601311475409837,
      "grad_norm": 1.0127913150937442,
      "learning_rate": 0.00018686683467407839,
      "loss": 13.0244,
      "step": 4347
    },
    {
      "epoch": 0.7603060109289618,
      "grad_norm": 0.817955543959612,
      "learning_rate": 0.00018685981858369873,
      "loss": 12.9503,
      "step": 4348
    },
    {
      "epoch": 0.7604808743169399,
      "grad_norm": 0.9465299407146798,
      "learning_rate": 0.000186852800751513,
      "loss": 12.9854,
      "step": 4349
    },
    {
      "epoch": 0.760655737704918,
      "grad_norm": 0.8598343632117665,
      "learning_rate": 0.0001868457811776619,
      "loss": 13.0726,
      "step": 4350
    },
    {
      "epoch": 0.7608306010928961,
      "grad_norm": 0.9468196856557861,
      "learning_rate": 0.00018683875986228628,
      "loss": 13.1427,
      "step": 4351
    },
    {
      "epoch": 0.7610054644808744,
      "grad_norm": 0.9193679116454953,
      "learning_rate": 0.00018683173680552687,
      "loss": 13.0273,
      "step": 4352
    },
    {
      "epoch": 0.7611803278688525,
      "grad_norm": 0.9478607501780423,
      "learning_rate": 0.00018682471200752457,
      "loss": 13.0453,
      "step": 4353
    },
    {
      "epoch": 0.7613551912568306,
      "grad_norm": 1.0818839816060444,
      "learning_rate": 0.00018681768546842018,
      "loss": 12.8517,
      "step": 4354
    },
    {
      "epoch": 0.7615300546448087,
      "grad_norm": 0.8592152053202299,
      "learning_rate": 0.00018681065718835463,
      "loss": 12.9667,
      "step": 4355
    },
    {
      "epoch": 0.7617049180327868,
      "grad_norm": 0.8852117589107479,
      "learning_rate": 0.00018680362716746887,
      "loss": 12.8507,
      "step": 4356
    },
    {
      "epoch": 0.7618797814207651,
      "grad_norm": 0.7969345636506437,
      "learning_rate": 0.00018679659540590384,
      "loss": 13.0829,
      "step": 4357
    },
    {
      "epoch": 0.7620546448087432,
      "grad_norm": 0.9656744235693251,
      "learning_rate": 0.00018678956190380062,
      "loss": 12.7498,
      "step": 4358
    },
    {
      "epoch": 0.7622295081967213,
      "grad_norm": 0.7787979269870801,
      "learning_rate": 0.00018678252666130013,
      "loss": 12.9483,
      "step": 4359
    },
    {
      "epoch": 0.7624043715846994,
      "grad_norm": 0.9140920204637635,
      "learning_rate": 0.0001867754896785436,
      "loss": 13.1412,
      "step": 4360
    },
    {
      "epoch": 0.7625792349726775,
      "grad_norm": 0.9350995796649575,
      "learning_rate": 0.00018676845095567203,
      "loss": 12.9894,
      "step": 4361
    },
    {
      "epoch": 0.7627540983606558,
      "grad_norm": 0.8391375625614473,
      "learning_rate": 0.00018676141049282658,
      "loss": 13.0512,
      "step": 4362
    },
    {
      "epoch": 0.7629289617486339,
      "grad_norm": 0.925939773544372,
      "learning_rate": 0.0001867543682901485,
      "loss": 13.0546,
      "step": 4363
    },
    {
      "epoch": 0.763103825136612,
      "grad_norm": 0.8798297719442454,
      "learning_rate": 0.00018674732434777896,
      "loss": 13.1094,
      "step": 4364
    },
    {
      "epoch": 0.7632786885245901,
      "grad_norm": 0.9082255158441378,
      "learning_rate": 0.0001867402786658592,
      "loss": 13.05,
      "step": 4365
    },
    {
      "epoch": 0.7634535519125683,
      "grad_norm": 0.8911586119441169,
      "learning_rate": 0.00018673323124453054,
      "loss": 13.0429,
      "step": 4366
    },
    {
      "epoch": 0.7636284153005465,
      "grad_norm": 0.9036791118805022,
      "learning_rate": 0.00018672618208393426,
      "loss": 12.9824,
      "step": 4367
    },
    {
      "epoch": 0.7638032786885246,
      "grad_norm": 0.9140459811941436,
      "learning_rate": 0.00018671913118421175,
      "loss": 13.0069,
      "step": 4368
    },
    {
      "epoch": 0.7639781420765027,
      "grad_norm": 0.8329072728049557,
      "learning_rate": 0.00018671207854550442,
      "loss": 13.0725,
      "step": 4369
    },
    {
      "epoch": 0.7641530054644808,
      "grad_norm": 0.9199604335790511,
      "learning_rate": 0.00018670502416795367,
      "loss": 13.0628,
      "step": 4370
    },
    {
      "epoch": 0.7643278688524591,
      "grad_norm": 0.9231558811885493,
      "learning_rate": 0.00018669796805170097,
      "loss": 12.9974,
      "step": 4371
    },
    {
      "epoch": 0.7645027322404372,
      "grad_norm": 0.9043546434221917,
      "learning_rate": 0.0001866909101968878,
      "loss": 13.1005,
      "step": 4372
    },
    {
      "epoch": 0.7646775956284153,
      "grad_norm": 1.0054975863368918,
      "learning_rate": 0.0001866838506036557,
      "loss": 13.0524,
      "step": 4373
    },
    {
      "epoch": 0.7648524590163934,
      "grad_norm": 0.7895992656586681,
      "learning_rate": 0.00018667678927214624,
      "loss": 13.0944,
      "step": 4374
    },
    {
      "epoch": 0.7650273224043715,
      "grad_norm": 0.9260986020509465,
      "learning_rate": 0.00018666972620250106,
      "loss": 12.7429,
      "step": 4375
    },
    {
      "epoch": 0.7652021857923498,
      "grad_norm": 0.8818938543182351,
      "learning_rate": 0.0001866626613948617,
      "loss": 13.1163,
      "step": 4376
    },
    {
      "epoch": 0.7653770491803279,
      "grad_norm": 0.9221638273458037,
      "learning_rate": 0.0001866555948493699,
      "loss": 13.3336,
      "step": 4377
    },
    {
      "epoch": 0.765551912568306,
      "grad_norm": 0.9559679300784573,
      "learning_rate": 0.0001866485265661674,
      "loss": 13.1723,
      "step": 4378
    },
    {
      "epoch": 0.7657267759562841,
      "grad_norm": 0.826377501368777,
      "learning_rate": 0.00018664145654539587,
      "loss": 12.9619,
      "step": 4379
    },
    {
      "epoch": 0.7659016393442623,
      "grad_norm": 0.7890432208274545,
      "learning_rate": 0.00018663438478719712,
      "loss": 13.0621,
      "step": 4380
    },
    {
      "epoch": 0.7660765027322405,
      "grad_norm": 0.8690580892354087,
      "learning_rate": 0.00018662731129171295,
      "loss": 13.0377,
      "step": 4381
    },
    {
      "epoch": 0.7662513661202186,
      "grad_norm": 0.7870564641197332,
      "learning_rate": 0.00018662023605908517,
      "loss": 13.0702,
      "step": 4382
    },
    {
      "epoch": 0.7664262295081967,
      "grad_norm": 0.8255199472624445,
      "learning_rate": 0.00018661315908945574,
      "loss": 12.9828,
      "step": 4383
    },
    {
      "epoch": 0.7666010928961748,
      "grad_norm": 0.8550729063458986,
      "learning_rate": 0.00018660608038296655,
      "loss": 13.104,
      "step": 4384
    },
    {
      "epoch": 0.766775956284153,
      "grad_norm": 0.8763544546691113,
      "learning_rate": 0.0001865989999397595,
      "loss": 13.2744,
      "step": 4385
    },
    {
      "epoch": 0.7669508196721312,
      "grad_norm": 0.8094346604178572,
      "learning_rate": 0.0001865919177599766,
      "loss": 12.7831,
      "step": 4386
    },
    {
      "epoch": 0.7671256830601093,
      "grad_norm": 0.9330666272939641,
      "learning_rate": 0.0001865848338437599,
      "loss": 13.203,
      "step": 4387
    },
    {
      "epoch": 0.7673005464480874,
      "grad_norm": 0.7842696138501852,
      "learning_rate": 0.0001865777481912514,
      "loss": 12.7938,
      "step": 4388
    },
    {
      "epoch": 0.7674754098360655,
      "grad_norm": 0.8908420867568214,
      "learning_rate": 0.00018657066080259328,
      "loss": 12.9442,
      "step": 4389
    },
    {
      "epoch": 0.7676502732240437,
      "grad_norm": 0.9707485729868298,
      "learning_rate": 0.00018656357167792757,
      "loss": 13.1021,
      "step": 4390
    },
    {
      "epoch": 0.7678251366120219,
      "grad_norm": 0.9125462863092552,
      "learning_rate": 0.00018655648081739644,
      "loss": 12.9769,
      "step": 4391
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.9430689895929144,
      "learning_rate": 0.00018654938822114213,
      "loss": 13.3873,
      "step": 4392
    },
    {
      "epoch": 0.7681748633879781,
      "grad_norm": 0.9790969035443324,
      "learning_rate": 0.00018654229388930684,
      "loss": 13.2544,
      "step": 4393
    },
    {
      "epoch": 0.7683497267759563,
      "grad_norm": 0.844468659059475,
      "learning_rate": 0.00018653519782203283,
      "loss": 13.1388,
      "step": 4394
    },
    {
      "epoch": 0.7685245901639344,
      "grad_norm": 0.9370073318361424,
      "learning_rate": 0.00018652810001946243,
      "loss": 13.2511,
      "step": 4395
    },
    {
      "epoch": 0.7686994535519126,
      "grad_norm": 0.875428076269333,
      "learning_rate": 0.00018652100048173793,
      "loss": 13.0935,
      "step": 4396
    },
    {
      "epoch": 0.7688743169398907,
      "grad_norm": 0.9503613669362156,
      "learning_rate": 0.00018651389920900174,
      "loss": 13.1817,
      "step": 4397
    },
    {
      "epoch": 0.7690491803278688,
      "grad_norm": 0.8053354808400365,
      "learning_rate": 0.0001865067962013962,
      "loss": 12.8944,
      "step": 4398
    },
    {
      "epoch": 0.769224043715847,
      "grad_norm": 0.9265140311403497,
      "learning_rate": 0.0001864996914590638,
      "loss": 13.0784,
      "step": 4399
    },
    {
      "epoch": 0.7693989071038252,
      "grad_norm": 0.8779161458578703,
      "learning_rate": 0.00018649258498214702,
      "loss": 13.1524,
      "step": 4400
    },
    {
      "epoch": 0.7695737704918033,
      "grad_norm": 0.8377496682808456,
      "learning_rate": 0.00018648547677078828,
      "loss": 13.0105,
      "step": 4401
    },
    {
      "epoch": 0.7697486338797814,
      "grad_norm": 0.9446549674361421,
      "learning_rate": 0.00018647836682513023,
      "loss": 13.1316,
      "step": 4402
    },
    {
      "epoch": 0.7699234972677595,
      "grad_norm": 0.8435874243431253,
      "learning_rate": 0.00018647125514531542,
      "loss": 12.6877,
      "step": 4403
    },
    {
      "epoch": 0.7700983606557377,
      "grad_norm": 0.9398482443375666,
      "learning_rate": 0.00018646414173148642,
      "loss": 13.1567,
      "step": 4404
    },
    {
      "epoch": 0.7702732240437159,
      "grad_norm": 0.8093129495770292,
      "learning_rate": 0.0001864570265837859,
      "loss": 13.0073,
      "step": 4405
    },
    {
      "epoch": 0.770448087431694,
      "grad_norm": 0.8517277079406366,
      "learning_rate": 0.00018644990970235654,
      "loss": 12.8487,
      "step": 4406
    },
    {
      "epoch": 0.7706229508196721,
      "grad_norm": 0.8461157450127769,
      "learning_rate": 0.00018644279108734105,
      "loss": 13.0576,
      "step": 4407
    },
    {
      "epoch": 0.7707978142076503,
      "grad_norm": 0.7913475895981326,
      "learning_rate": 0.0001864356707388822,
      "loss": 13.0668,
      "step": 4408
    },
    {
      "epoch": 0.7709726775956284,
      "grad_norm": 0.8623645161089742,
      "learning_rate": 0.00018642854865712276,
      "loss": 13.0548,
      "step": 4409
    },
    {
      "epoch": 0.7711475409836066,
      "grad_norm": 0.8120244844361075,
      "learning_rate": 0.0001864214248422055,
      "loss": 13.3037,
      "step": 4410
    },
    {
      "epoch": 0.7713224043715847,
      "grad_norm": 0.8837711102003103,
      "learning_rate": 0.00018641429929427337,
      "loss": 12.9898,
      "step": 4411
    },
    {
      "epoch": 0.7714972677595628,
      "grad_norm": 0.7709875950148212,
      "learning_rate": 0.00018640717201346923,
      "loss": 12.7658,
      "step": 4412
    },
    {
      "epoch": 0.771672131147541,
      "grad_norm": 0.9321658987650633,
      "learning_rate": 0.00018640004299993597,
      "loss": 13.1537,
      "step": 4413
    },
    {
      "epoch": 0.7718469945355191,
      "grad_norm": 0.8447641223397993,
      "learning_rate": 0.00018639291225381653,
      "loss": 12.9184,
      "step": 4414
    },
    {
      "epoch": 0.7720218579234973,
      "grad_norm": 0.95635206630464,
      "learning_rate": 0.00018638577977525396,
      "loss": 13.0416,
      "step": 4415
    },
    {
      "epoch": 0.7721967213114754,
      "grad_norm": 1.0147719380261262,
      "learning_rate": 0.00018637864556439126,
      "loss": 13.271,
      "step": 4416
    },
    {
      "epoch": 0.7723715846994536,
      "grad_norm": 0.8787798330939349,
      "learning_rate": 0.00018637150962137153,
      "loss": 13.1445,
      "step": 4417
    },
    {
      "epoch": 0.7725464480874317,
      "grad_norm": 1.0045468126459989,
      "learning_rate": 0.0001863643719463378,
      "loss": 12.9399,
      "step": 4418
    },
    {
      "epoch": 0.7727213114754098,
      "grad_norm": 0.8987714219510784,
      "learning_rate": 0.00018635723253943327,
      "loss": 13.128,
      "step": 4419
    },
    {
      "epoch": 0.772896174863388,
      "grad_norm": 1.0814780351106015,
      "learning_rate": 0.00018635009140080105,
      "loss": 13.2391,
      "step": 4420
    },
    {
      "epoch": 0.7730710382513661,
      "grad_norm": 0.9713733486381786,
      "learning_rate": 0.00018634294853058438,
      "loss": 13.0091,
      "step": 4421
    },
    {
      "epoch": 0.7732459016393443,
      "grad_norm": 1.0106263316086526,
      "learning_rate": 0.00018633580392892648,
      "loss": 12.9333,
      "step": 4422
    },
    {
      "epoch": 0.7734207650273224,
      "grad_norm": 0.8675133191265849,
      "learning_rate": 0.00018632865759597063,
      "loss": 13.281,
      "step": 4423
    },
    {
      "epoch": 0.7735956284153005,
      "grad_norm": 0.9017916259419191,
      "learning_rate": 0.00018632150953186015,
      "loss": 12.8783,
      "step": 4424
    },
    {
      "epoch": 0.7737704918032787,
      "grad_norm": 0.9359968804131416,
      "learning_rate": 0.00018631435973673831,
      "loss": 13.0807,
      "step": 4425
    },
    {
      "epoch": 0.7739453551912568,
      "grad_norm": 0.8031451446756595,
      "learning_rate": 0.00018630720821074859,
      "loss": 12.9893,
      "step": 4426
    },
    {
      "epoch": 0.774120218579235,
      "grad_norm": 0.9020819640413408,
      "learning_rate": 0.0001863000549540343,
      "loss": 13.1495,
      "step": 4427
    },
    {
      "epoch": 0.7742950819672131,
      "grad_norm": 0.9887920036085621,
      "learning_rate": 0.00018629289996673897,
      "loss": 12.9923,
      "step": 4428
    },
    {
      "epoch": 0.7744699453551912,
      "grad_norm": 0.9438531855056883,
      "learning_rate": 0.00018628574324900602,
      "loss": 13.2343,
      "step": 4429
    },
    {
      "epoch": 0.7746448087431694,
      "grad_norm": 0.8801229265728491,
      "learning_rate": 0.00018627858480097903,
      "loss": 13.2064,
      "step": 4430
    },
    {
      "epoch": 0.7748196721311476,
      "grad_norm": 0.8339487233428352,
      "learning_rate": 0.00018627142462280144,
      "loss": 13.0236,
      "step": 4431
    },
    {
      "epoch": 0.7749945355191257,
      "grad_norm": 1.026801839524429,
      "learning_rate": 0.00018626426271461694,
      "loss": 13.0787,
      "step": 4432
    },
    {
      "epoch": 0.7751693989071038,
      "grad_norm": 0.8308703327732179,
      "learning_rate": 0.0001862570990765691,
      "loss": 12.9823,
      "step": 4433
    },
    {
      "epoch": 0.775344262295082,
      "grad_norm": 1.1291676395201535,
      "learning_rate": 0.00018624993370880156,
      "loss": 12.7143,
      "step": 4434
    },
    {
      "epoch": 0.7755191256830601,
      "grad_norm": 0.9317449774988998,
      "learning_rate": 0.00018624276661145805,
      "loss": 13.1186,
      "step": 4435
    },
    {
      "epoch": 0.7756939890710383,
      "grad_norm": 1.1050700396517201,
      "learning_rate": 0.00018623559778468228,
      "loss": 13.3621,
      "step": 4436
    },
    {
      "epoch": 0.7758688524590164,
      "grad_norm": 1.0338638467364942,
      "learning_rate": 0.00018622842722861797,
      "loss": 13.2733,
      "step": 4437
    },
    {
      "epoch": 0.7760437158469945,
      "grad_norm": 0.9235751607773303,
      "learning_rate": 0.00018622125494340895,
      "loss": 13.0162,
      "step": 4438
    },
    {
      "epoch": 0.7762185792349727,
      "grad_norm": 0.8979792999478459,
      "learning_rate": 0.000186214080929199,
      "loss": 13.1202,
      "step": 4439
    },
    {
      "epoch": 0.7763934426229508,
      "grad_norm": 0.9339387035152704,
      "learning_rate": 0.00018620690518613206,
      "loss": 13.0588,
      "step": 4440
    },
    {
      "epoch": 0.776568306010929,
      "grad_norm": 0.92246401408509,
      "learning_rate": 0.00018619972771435198,
      "loss": 13.2516,
      "step": 4441
    },
    {
      "epoch": 0.7767431693989071,
      "grad_norm": 0.8192845829491209,
      "learning_rate": 0.00018619254851400267,
      "loss": 12.8299,
      "step": 4442
    },
    {
      "epoch": 0.7769180327868852,
      "grad_norm": 0.8672081091889084,
      "learning_rate": 0.00018618536758522813,
      "loss": 13.1966,
      "step": 4443
    },
    {
      "epoch": 0.7770928961748634,
      "grad_norm": 0.8811985973893071,
      "learning_rate": 0.00018617818492817234,
      "loss": 13.0123,
      "step": 4444
    },
    {
      "epoch": 0.7772677595628416,
      "grad_norm": 0.7823534102184468,
      "learning_rate": 0.00018617100054297935,
      "loss": 13.0754,
      "step": 4445
    },
    {
      "epoch": 0.7774426229508197,
      "grad_norm": 0.9576432316239759,
      "learning_rate": 0.0001861638144297932,
      "loss": 12.7808,
      "step": 4446
    },
    {
      "epoch": 0.7776174863387978,
      "grad_norm": 1.0451547430541233,
      "learning_rate": 0.00018615662658875802,
      "loss": 13.1243,
      "step": 4447
    },
    {
      "epoch": 0.7777923497267759,
      "grad_norm": 0.9619411013698926,
      "learning_rate": 0.00018614943702001795,
      "loss": 12.967,
      "step": 4448
    },
    {
      "epoch": 0.7779672131147541,
      "grad_norm": 0.9467447917272875,
      "learning_rate": 0.00018614224572371715,
      "loss": 13.2076,
      "step": 4449
    },
    {
      "epoch": 0.7781420765027323,
      "grad_norm": 0.9884118269031313,
      "learning_rate": 0.00018613505269999983,
      "loss": 12.813,
      "step": 4450
    },
    {
      "epoch": 0.7783169398907104,
      "grad_norm": 0.8813886022379118,
      "learning_rate": 0.00018612785794901022,
      "loss": 12.7372,
      "step": 4451
    },
    {
      "epoch": 0.7784918032786885,
      "grad_norm": 1.0166533138339886,
      "learning_rate": 0.00018612066147089261,
      "loss": 12.8908,
      "step": 4452
    },
    {
      "epoch": 0.7786666666666666,
      "grad_norm": 0.9405695793632635,
      "learning_rate": 0.00018611346326579135,
      "loss": 13.1925,
      "step": 4453
    },
    {
      "epoch": 0.7788415300546448,
      "grad_norm": 0.8711353092656021,
      "learning_rate": 0.00018610626333385072,
      "loss": 13.0308,
      "step": 4454
    },
    {
      "epoch": 0.779016393442623,
      "grad_norm": 0.8923074526991694,
      "learning_rate": 0.00018609906167521513,
      "loss": 13.0795,
      "step": 4455
    },
    {
      "epoch": 0.7791912568306011,
      "grad_norm": 0.8994521028023668,
      "learning_rate": 0.00018609185829002902,
      "loss": 12.9942,
      "step": 4456
    },
    {
      "epoch": 0.7793661202185792,
      "grad_norm": 1.0924807496478097,
      "learning_rate": 0.00018608465317843678,
      "loss": 13.1161,
      "step": 4457
    },
    {
      "epoch": 0.7795409836065573,
      "grad_norm": 0.9261036834243465,
      "learning_rate": 0.00018607744634058294,
      "loss": 12.8664,
      "step": 4458
    },
    {
      "epoch": 0.7797158469945356,
      "grad_norm": 1.0416814025239485,
      "learning_rate": 0.000186070237776612,
      "loss": 13.1768,
      "step": 4459
    },
    {
      "epoch": 0.7798907103825137,
      "grad_norm": 0.9349190713299667,
      "learning_rate": 0.00018606302748666855,
      "loss": 13.0048,
      "step": 4460
    },
    {
      "epoch": 0.7800655737704918,
      "grad_norm": 0.9732593493172847,
      "learning_rate": 0.00018605581547089715,
      "loss": 13.1308,
      "step": 4461
    },
    {
      "epoch": 0.7802404371584699,
      "grad_norm": 0.9982062834826657,
      "learning_rate": 0.00018604860172944241,
      "loss": 12.8828,
      "step": 4462
    },
    {
      "epoch": 0.780415300546448,
      "grad_norm": 0.7914209197593218,
      "learning_rate": 0.000186041386262449,
      "loss": 12.9217,
      "step": 4463
    },
    {
      "epoch": 0.7805901639344263,
      "grad_norm": 1.1676268415411881,
      "learning_rate": 0.0001860341690700616,
      "loss": 13.0929,
      "step": 4464
    },
    {
      "epoch": 0.7807650273224044,
      "grad_norm": 0.9337911879159089,
      "learning_rate": 0.00018602695015242494,
      "loss": 12.9751,
      "step": 4465
    },
    {
      "epoch": 0.7809398907103825,
      "grad_norm": 1.0199084134881147,
      "learning_rate": 0.00018601972950968386,
      "loss": 13.2687,
      "step": 4466
    },
    {
      "epoch": 0.7811147540983606,
      "grad_norm": 0.9730634150356064,
      "learning_rate": 0.00018601250714198302,
      "loss": 13.2574,
      "step": 4467
    },
    {
      "epoch": 0.7812896174863388,
      "grad_norm": 0.9085221932025609,
      "learning_rate": 0.00018600528304946735,
      "loss": 13.0157,
      "step": 4468
    },
    {
      "epoch": 0.781464480874317,
      "grad_norm": 1.0808445101138004,
      "learning_rate": 0.00018599805723228168,
      "loss": 13.028,
      "step": 4469
    },
    {
      "epoch": 0.7816393442622951,
      "grad_norm": 0.898961166068904,
      "learning_rate": 0.00018599082969057089,
      "loss": 13.1039,
      "step": 4470
    },
    {
      "epoch": 0.7818142076502732,
      "grad_norm": 1.0829117002887876,
      "learning_rate": 0.00018598360042447996,
      "loss": 13.2155,
      "step": 4471
    },
    {
      "epoch": 0.7819890710382513,
      "grad_norm": 0.9437305111392001,
      "learning_rate": 0.00018597636943415386,
      "loss": 13.1906,
      "step": 4472
    },
    {
      "epoch": 0.7821639344262296,
      "grad_norm": 0.9754020925891284,
      "learning_rate": 0.00018596913671973755,
      "loss": 13.0654,
      "step": 4473
    },
    {
      "epoch": 0.7823387978142077,
      "grad_norm": 1.073022555899563,
      "learning_rate": 0.00018596190228137606,
      "loss": 12.8433,
      "step": 4474
    },
    {
      "epoch": 0.7825136612021858,
      "grad_norm": 0.8406700173663305,
      "learning_rate": 0.00018595466611921453,
      "loss": 12.7002,
      "step": 4475
    },
    {
      "epoch": 0.7826885245901639,
      "grad_norm": 1.0828910461361216,
      "learning_rate": 0.00018594742823339802,
      "loss": 12.7866,
      "step": 4476
    },
    {
      "epoch": 0.782863387978142,
      "grad_norm": 0.8493971831121068,
      "learning_rate": 0.00018594018862407168,
      "loss": 13.0051,
      "step": 4477
    },
    {
      "epoch": 0.7830382513661203,
      "grad_norm": 1.0792534076350742,
      "learning_rate": 0.00018593294729138065,
      "loss": 13.179,
      "step": 4478
    },
    {
      "epoch": 0.7832131147540984,
      "grad_norm": 0.7951725070846749,
      "learning_rate": 0.00018592570423547024,
      "loss": 12.9065,
      "step": 4479
    },
    {
      "epoch": 0.7833879781420765,
      "grad_norm": 0.9475204546734106,
      "learning_rate": 0.0001859184594564856,
      "loss": 12.9673,
      "step": 4480
    },
    {
      "epoch": 0.7835628415300546,
      "grad_norm": 0.857620227330129,
      "learning_rate": 0.00018591121295457206,
      "loss": 13.1505,
      "step": 4481
    },
    {
      "epoch": 0.7837377049180327,
      "grad_norm": 0.8244441646701097,
      "learning_rate": 0.0001859039647298749,
      "loss": 12.7192,
      "step": 4482
    },
    {
      "epoch": 0.783912568306011,
      "grad_norm": 0.8954672204319973,
      "learning_rate": 0.00018589671478253948,
      "loss": 12.9662,
      "step": 4483
    },
    {
      "epoch": 0.7840874316939891,
      "grad_norm": 0.9224393723527201,
      "learning_rate": 0.00018588946311271122,
      "loss": 13.2826,
      "step": 4484
    },
    {
      "epoch": 0.7842622950819672,
      "grad_norm": 0.8561226674110402,
      "learning_rate": 0.0001858822097205355,
      "loss": 13.1224,
      "step": 4485
    },
    {
      "epoch": 0.7844371584699453,
      "grad_norm": 0.9183917924584435,
      "learning_rate": 0.00018587495460615778,
      "loss": 13.0661,
      "step": 4486
    },
    {
      "epoch": 0.7846120218579234,
      "grad_norm": 0.8407926577888704,
      "learning_rate": 0.00018586769776972353,
      "loss": 12.8353,
      "step": 4487
    },
    {
      "epoch": 0.7847868852459017,
      "grad_norm": 0.8934479935682621,
      "learning_rate": 0.00018586043921137833,
      "loss": 13.2808,
      "step": 4488
    },
    {
      "epoch": 0.7849617486338798,
      "grad_norm": 0.8863207894028607,
      "learning_rate": 0.00018585317893126768,
      "loss": 13.1058,
      "step": 4489
    },
    {
      "epoch": 0.7851366120218579,
      "grad_norm": 0.8824234917939632,
      "learning_rate": 0.00018584591692953717,
      "loss": 13.1817,
      "step": 4490
    },
    {
      "epoch": 0.785311475409836,
      "grad_norm": 0.8800015894808786,
      "learning_rate": 0.00018583865320633244,
      "loss": 13.1636,
      "step": 4491
    },
    {
      "epoch": 0.7854863387978142,
      "grad_norm": 0.8692634195930591,
      "learning_rate": 0.0001858313877617992,
      "loss": 13.1947,
      "step": 4492
    },
    {
      "epoch": 0.7856612021857924,
      "grad_norm": 0.8863049108948712,
      "learning_rate": 0.00018582412059608308,
      "loss": 13.0993,
      "step": 4493
    },
    {
      "epoch": 0.7858360655737705,
      "grad_norm": 0.7709519927790456,
      "learning_rate": 0.0001858168517093298,
      "loss": 13.193,
      "step": 4494
    },
    {
      "epoch": 0.7860109289617486,
      "grad_norm": 0.8239077826367494,
      "learning_rate": 0.00018580958110168516,
      "loss": 12.981,
      "step": 4495
    },
    {
      "epoch": 0.7861857923497267,
      "grad_norm": 0.878054289223864,
      "learning_rate": 0.00018580230877329496,
      "loss": 13.2415,
      "step": 4496
    },
    {
      "epoch": 0.786360655737705,
      "grad_norm": 0.8820037208383252,
      "learning_rate": 0.00018579503472430502,
      "loss": 12.8437,
      "step": 4497
    },
    {
      "epoch": 0.7865355191256831,
      "grad_norm": 0.8832738293319355,
      "learning_rate": 0.0001857877589548612,
      "loss": 13.1788,
      "step": 4498
    },
    {
      "epoch": 0.7867103825136612,
      "grad_norm": 0.7929622980387997,
      "learning_rate": 0.00018578048146510943,
      "loss": 12.9553,
      "step": 4499
    },
    {
      "epoch": 0.7868852459016393,
      "grad_norm": 1.0142042796389028,
      "learning_rate": 0.00018577320225519558,
      "loss": 13.0902,
      "step": 4500
    },
    {
      "epoch": 0.7870601092896174,
      "grad_norm": 0.8050186829224729,
      "learning_rate": 0.00018576592132526573,
      "loss": 13.2038,
      "step": 4501
    },
    {
      "epoch": 0.7872349726775957,
      "grad_norm": 0.8508946097741067,
      "learning_rate": 0.0001857586386754658,
      "loss": 12.8953,
      "step": 4502
    },
    {
      "epoch": 0.7874098360655738,
      "grad_norm": 0.9031014708512654,
      "learning_rate": 0.00018575135430594185,
      "loss": 12.9363,
      "step": 4503
    },
    {
      "epoch": 0.7875846994535519,
      "grad_norm": 0.817631730260539,
      "learning_rate": 0.00018574406821683992,
      "loss": 13.1268,
      "step": 4504
    },
    {
      "epoch": 0.78775956284153,
      "grad_norm": 0.8685941650310538,
      "learning_rate": 0.0001857367804083062,
      "loss": 12.9879,
      "step": 4505
    },
    {
      "epoch": 0.7879344262295082,
      "grad_norm": 0.8141761576142599,
      "learning_rate": 0.00018572949088048674,
      "loss": 12.9542,
      "step": 4506
    },
    {
      "epoch": 0.7881092896174864,
      "grad_norm": 0.9087923096579753,
      "learning_rate": 0.0001857221996335278,
      "loss": 13.038,
      "step": 4507
    },
    {
      "epoch": 0.7882841530054645,
      "grad_norm": 0.7905904871915521,
      "learning_rate": 0.00018571490666757554,
      "loss": 12.8526,
      "step": 4508
    },
    {
      "epoch": 0.7884590163934426,
      "grad_norm": 0.943409537501243,
      "learning_rate": 0.00018570761198277624,
      "loss": 13.1595,
      "step": 4509
    },
    {
      "epoch": 0.7886338797814207,
      "grad_norm": 0.8444987862588287,
      "learning_rate": 0.00018570031557927616,
      "loss": 13.0313,
      "step": 4510
    },
    {
      "epoch": 0.7888087431693989,
      "grad_norm": 0.8631495147942966,
      "learning_rate": 0.0001856930174572216,
      "loss": 13.125,
      "step": 4511
    },
    {
      "epoch": 0.7889836065573771,
      "grad_norm": 0.8911350447092343,
      "learning_rate": 0.00018568571761675893,
      "loss": 13.2804,
      "step": 4512
    },
    {
      "epoch": 0.7891584699453552,
      "grad_norm": 0.7877776842193764,
      "learning_rate": 0.00018567841605803453,
      "loss": 12.9183,
      "step": 4513
    },
    {
      "epoch": 0.7893333333333333,
      "grad_norm": 0.8631747465922006,
      "learning_rate": 0.00018567111278119486,
      "loss": 13.0268,
      "step": 4514
    },
    {
      "epoch": 0.7895081967213115,
      "grad_norm": 0.8398678605361003,
      "learning_rate": 0.00018566380778638628,
      "loss": 13.0821,
      "step": 4515
    },
    {
      "epoch": 0.7896830601092896,
      "grad_norm": 0.8567746631648218,
      "learning_rate": 0.00018565650107375535,
      "loss": 13.2616,
      "step": 4516
    },
    {
      "epoch": 0.7898579234972678,
      "grad_norm": 0.850415228806156,
      "learning_rate": 0.00018564919264344858,
      "loss": 13.0799,
      "step": 4517
    },
    {
      "epoch": 0.7900327868852459,
      "grad_norm": 0.8301524308826939,
      "learning_rate": 0.00018564188249561254,
      "loss": 13.0704,
      "step": 4518
    },
    {
      "epoch": 0.790207650273224,
      "grad_norm": 0.865587035735593,
      "learning_rate": 0.00018563457063039377,
      "loss": 13.288,
      "step": 4519
    },
    {
      "epoch": 0.7903825136612022,
      "grad_norm": 0.7679496418661235,
      "learning_rate": 0.00018562725704793893,
      "loss": 12.9053,
      "step": 4520
    },
    {
      "epoch": 0.7905573770491803,
      "grad_norm": 0.8584023592386525,
      "learning_rate": 0.00018561994174839467,
      "loss": 12.9991,
      "step": 4521
    },
    {
      "epoch": 0.7907322404371585,
      "grad_norm": 0.8589539118722412,
      "learning_rate": 0.0001856126247319077,
      "loss": 12.8562,
      "step": 4522
    },
    {
      "epoch": 0.7909071038251366,
      "grad_norm": 0.9381999521306359,
      "learning_rate": 0.00018560530599862475,
      "loss": 13.225,
      "step": 4523
    },
    {
      "epoch": 0.7910819672131147,
      "grad_norm": 0.8271143522237913,
      "learning_rate": 0.00018559798554869256,
      "loss": 13.1358,
      "step": 4524
    },
    {
      "epoch": 0.7912568306010929,
      "grad_norm": 0.9178125315846564,
      "learning_rate": 0.00018559066338225794,
      "loss": 13.2608,
      "step": 4525
    },
    {
      "epoch": 0.791431693989071,
      "grad_norm": 0.8501948606470516,
      "learning_rate": 0.00018558333949946772,
      "loss": 12.9221,
      "step": 4526
    },
    {
      "epoch": 0.7916065573770492,
      "grad_norm": 0.9061523920627327,
      "learning_rate": 0.0001855760139004688,
      "loss": 12.8884,
      "step": 4527
    },
    {
      "epoch": 0.7917814207650273,
      "grad_norm": 0.8255532563237684,
      "learning_rate": 0.000185568686585408,
      "loss": 12.8543,
      "step": 4528
    },
    {
      "epoch": 0.7919562841530055,
      "grad_norm": 0.9625675837235699,
      "learning_rate": 0.00018556135755443235,
      "loss": 12.714,
      "step": 4529
    },
    {
      "epoch": 0.7921311475409836,
      "grad_norm": 0.8887414289972978,
      "learning_rate": 0.0001855540268076887,
      "loss": 12.7974,
      "step": 4530
    },
    {
      "epoch": 0.7923060109289618,
      "grad_norm": 0.8798107786180811,
      "learning_rate": 0.00018554669434532419,
      "loss": 13.0756,
      "step": 4531
    },
    {
      "epoch": 0.7924808743169399,
      "grad_norm": 0.8402196892432146,
      "learning_rate": 0.0001855393601674858,
      "loss": 12.877,
      "step": 4532
    },
    {
      "epoch": 0.792655737704918,
      "grad_norm": 0.8530387002728137,
      "learning_rate": 0.00018553202427432053,
      "loss": 12.9703,
      "step": 4533
    },
    {
      "epoch": 0.7928306010928962,
      "grad_norm": 0.8295906245215301,
      "learning_rate": 0.00018552468666597563,
      "loss": 13.0103,
      "step": 4534
    },
    {
      "epoch": 0.7930054644808743,
      "grad_norm": 0.810546720931482,
      "learning_rate": 0.00018551734734259813,
      "loss": 13.2196,
      "step": 4535
    },
    {
      "epoch": 0.7931803278688525,
      "grad_norm": 0.7857825692195052,
      "learning_rate": 0.00018551000630433526,
      "loss": 12.659,
      "step": 4536
    },
    {
      "epoch": 0.7933551912568306,
      "grad_norm": 0.8752042030148781,
      "learning_rate": 0.0001855026635513342,
      "loss": 13.0624,
      "step": 4537
    },
    {
      "epoch": 0.7935300546448087,
      "grad_norm": 0.8604319394152696,
      "learning_rate": 0.00018549531908374222,
      "loss": 12.7734,
      "step": 4538
    },
    {
      "epoch": 0.7937049180327869,
      "grad_norm": 0.8743206287187385,
      "learning_rate": 0.0001854879729017066,
      "loss": 13.1933,
      "step": 4539
    },
    {
      "epoch": 0.793879781420765,
      "grad_norm": 0.8241028100322526,
      "learning_rate": 0.00018548062500537462,
      "loss": 13.175,
      "step": 4540
    },
    {
      "epoch": 0.7940546448087432,
      "grad_norm": 1.0316971573353881,
      "learning_rate": 0.00018547327539489362,
      "loss": 13.1378,
      "step": 4541
    },
    {
      "epoch": 0.7942295081967213,
      "grad_norm": 0.8230640855051178,
      "learning_rate": 0.00018546592407041107,
      "loss": 12.8974,
      "step": 4542
    },
    {
      "epoch": 0.7944043715846995,
      "grad_norm": 0.8788165817635661,
      "learning_rate": 0.0001854585710320743,
      "loss": 13.0261,
      "step": 4543
    },
    {
      "epoch": 0.7945792349726776,
      "grad_norm": 0.8617627586186789,
      "learning_rate": 0.00018545121628003077,
      "loss": 13.172,
      "step": 4544
    },
    {
      "epoch": 0.7947540983606557,
      "grad_norm": 0.7361012360157816,
      "learning_rate": 0.000185443859814428,
      "loss": 12.8602,
      "step": 4545
    },
    {
      "epoch": 0.7949289617486339,
      "grad_norm": 0.9494379099937822,
      "learning_rate": 0.0001854365016354135,
      "loss": 12.9179,
      "step": 4546
    },
    {
      "epoch": 0.795103825136612,
      "grad_norm": 1.0188039370806357,
      "learning_rate": 0.0001854291417431348,
      "loss": 13.1792,
      "step": 4547
    },
    {
      "epoch": 0.7952786885245902,
      "grad_norm": 0.8889263593276471,
      "learning_rate": 0.00018542178013773955,
      "loss": 12.9081,
      "step": 4548
    },
    {
      "epoch": 0.7954535519125683,
      "grad_norm": 0.9702253972329549,
      "learning_rate": 0.00018541441681937528,
      "loss": 13.1055,
      "step": 4549
    },
    {
      "epoch": 0.7956284153005464,
      "grad_norm": 0.8322049269636218,
      "learning_rate": 0.0001854070517881897,
      "loss": 12.7542,
      "step": 4550
    },
    {
      "epoch": 0.7958032786885246,
      "grad_norm": 0.9015722446686116,
      "learning_rate": 0.00018539968504433052,
      "loss": 13.038,
      "step": 4551
    },
    {
      "epoch": 0.7959781420765027,
      "grad_norm": 1.1904169139755019,
      "learning_rate": 0.00018539231658794544,
      "loss": 13.1027,
      "step": 4552
    },
    {
      "epoch": 0.7961530054644809,
      "grad_norm": 0.9004368109032831,
      "learning_rate": 0.00018538494641918223,
      "loss": 12.8618,
      "step": 4553
    },
    {
      "epoch": 0.796327868852459,
      "grad_norm": 0.9312521147839151,
      "learning_rate": 0.00018537757453818867,
      "loss": 13.1052,
      "step": 4554
    },
    {
      "epoch": 0.7965027322404371,
      "grad_norm": 0.8447893071166619,
      "learning_rate": 0.0001853702009451126,
      "loss": 12.8365,
      "step": 4555
    },
    {
      "epoch": 0.7966775956284153,
      "grad_norm": 0.9510830646293631,
      "learning_rate": 0.00018536282564010186,
      "loss": 12.7039,
      "step": 4556
    },
    {
      "epoch": 0.7968524590163935,
      "grad_norm": 0.992348561564195,
      "learning_rate": 0.00018535544862330436,
      "loss": 13.1589,
      "step": 4557
    },
    {
      "epoch": 0.7970273224043716,
      "grad_norm": 0.9234728511782819,
      "learning_rate": 0.00018534806989486805,
      "loss": 13.0386,
      "step": 4558
    },
    {
      "epoch": 0.7972021857923497,
      "grad_norm": 0.9018357233625138,
      "learning_rate": 0.00018534068945494093,
      "loss": 12.9352,
      "step": 4559
    },
    {
      "epoch": 0.7973770491803279,
      "grad_norm": 0.8939564238971047,
      "learning_rate": 0.0001853333073036709,
      "loss": 13.0569,
      "step": 4560
    },
    {
      "epoch": 0.797551912568306,
      "grad_norm": 1.0481369532109495,
      "learning_rate": 0.0001853259234412061,
      "loss": 12.822,
      "step": 4561
    },
    {
      "epoch": 0.7977267759562842,
      "grad_norm": 0.8335881544556353,
      "learning_rate": 0.00018531853786769452,
      "loss": 13.1117,
      "step": 4562
    },
    {
      "epoch": 0.7979016393442623,
      "grad_norm": 0.8457701736072499,
      "learning_rate": 0.0001853111505832843,
      "loss": 12.9326,
      "step": 4563
    },
    {
      "epoch": 0.7980765027322404,
      "grad_norm": 0.893971794484999,
      "learning_rate": 0.00018530376158812354,
      "loss": 12.8719,
      "step": 4564
    },
    {
      "epoch": 0.7982513661202186,
      "grad_norm": 0.8049947791324711,
      "learning_rate": 0.00018529637088236046,
      "loss": 12.8391,
      "step": 4565
    },
    {
      "epoch": 0.7984262295081967,
      "grad_norm": 0.852927859954717,
      "learning_rate": 0.0001852889784661433,
      "loss": 13.0905,
      "step": 4566
    },
    {
      "epoch": 0.7986010928961749,
      "grad_norm": 1.0344343162317478,
      "learning_rate": 0.0001852815843396202,
      "loss": 13.1768,
      "step": 4567
    },
    {
      "epoch": 0.798775956284153,
      "grad_norm": 0.8486513752983521,
      "learning_rate": 0.0001852741885029395,
      "loss": 12.9934,
      "step": 4568
    },
    {
      "epoch": 0.7989508196721311,
      "grad_norm": 0.9020697271113723,
      "learning_rate": 0.0001852667909562495,
      "loss": 13.107,
      "step": 4569
    },
    {
      "epoch": 0.7991256830601093,
      "grad_norm": 0.899800090867052,
      "learning_rate": 0.0001852593916996985,
      "loss": 12.7169,
      "step": 4570
    },
    {
      "epoch": 0.7993005464480875,
      "grad_norm": 1.090662137615553,
      "learning_rate": 0.00018525199073343495,
      "loss": 12.6929,
      "step": 4571
    },
    {
      "epoch": 0.7994754098360656,
      "grad_norm": 0.8311429487239118,
      "learning_rate": 0.00018524458805760724,
      "loss": 13.1171,
      "step": 4572
    },
    {
      "epoch": 0.7996502732240437,
      "grad_norm": 1.0389407282025747,
      "learning_rate": 0.0001852371836723638,
      "loss": 12.9383,
      "step": 4573
    },
    {
      "epoch": 0.7998251366120218,
      "grad_norm": 0.8371955693232839,
      "learning_rate": 0.0001852297775778531,
      "loss": 12.8037,
      "step": 4574
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1809106905511833,
      "learning_rate": 0.00018522236977422363,
      "loss": 13.1294,
      "step": 4575
    },
    {
      "epoch": 0.8001748633879782,
      "grad_norm": 1.0177375210578161,
      "learning_rate": 0.00018521496026162405,
      "loss": 12.9194,
      "step": 4576
    },
    {
      "epoch": 0.8003497267759563,
      "grad_norm": 0.9206251286917895,
      "learning_rate": 0.00018520754904020285,
      "loss": 12.6047,
      "step": 4577
    },
    {
      "epoch": 0.8005245901639344,
      "grad_norm": 1.003193548021234,
      "learning_rate": 0.00018520013611010865,
      "loss": 12.8728,
      "step": 4578
    },
    {
      "epoch": 0.8006994535519125,
      "grad_norm": 0.8651790232988724,
      "learning_rate": 0.00018519272147149015,
      "loss": 13.2526,
      "step": 4579
    },
    {
      "epoch": 0.8008743169398908,
      "grad_norm": 0.9295573217973061,
      "learning_rate": 0.00018518530512449596,
      "loss": 13.0035,
      "step": 4580
    },
    {
      "epoch": 0.8010491803278689,
      "grad_norm": 0.8347135489997757,
      "learning_rate": 0.0001851778870692749,
      "loss": 13.1392,
      "step": 4581
    },
    {
      "epoch": 0.801224043715847,
      "grad_norm": 0.8119634154624836,
      "learning_rate": 0.00018517046730597564,
      "loss": 13.1206,
      "step": 4582
    },
    {
      "epoch": 0.8013989071038251,
      "grad_norm": 0.8836754265230475,
      "learning_rate": 0.000185163045834747,
      "loss": 12.835,
      "step": 4583
    },
    {
      "epoch": 0.8015737704918032,
      "grad_norm": 0.9152291120313703,
      "learning_rate": 0.00018515562265573784,
      "loss": 13.2332,
      "step": 4584
    },
    {
      "epoch": 0.8017486338797815,
      "grad_norm": 0.9020996490050398,
      "learning_rate": 0.00018514819776909697,
      "loss": 12.8572,
      "step": 4585
    },
    {
      "epoch": 0.8019234972677596,
      "grad_norm": 1.0131677553408203,
      "learning_rate": 0.0001851407711749733,
      "loss": 13.2309,
      "step": 4586
    },
    {
      "epoch": 0.8020983606557377,
      "grad_norm": 0.856859449296939,
      "learning_rate": 0.00018513334287351577,
      "loss": 13.0086,
      "step": 4587
    },
    {
      "epoch": 0.8022732240437158,
      "grad_norm": 0.839797320366295,
      "learning_rate": 0.00018512591286487328,
      "loss": 13.2346,
      "step": 4588
    },
    {
      "epoch": 0.8024480874316939,
      "grad_norm": 0.8888173635136442,
      "learning_rate": 0.00018511848114919488,
      "loss": 13.0289,
      "step": 4589
    },
    {
      "epoch": 0.8026229508196722,
      "grad_norm": 0.8685189702711975,
      "learning_rate": 0.00018511104772662957,
      "loss": 12.9,
      "step": 4590
    },
    {
      "epoch": 0.8027978142076503,
      "grad_norm": 0.7771634092091698,
      "learning_rate": 0.00018510361259732646,
      "loss": 13.0146,
      "step": 4591
    },
    {
      "epoch": 0.8029726775956284,
      "grad_norm": 0.7916476602938558,
      "learning_rate": 0.00018509617576143459,
      "loss": 13.0432,
      "step": 4592
    },
    {
      "epoch": 0.8031475409836065,
      "grad_norm": 0.8570658258364061,
      "learning_rate": 0.00018508873721910315,
      "loss": 13.1474,
      "step": 4593
    },
    {
      "epoch": 0.8033224043715848,
      "grad_norm": 0.9344456473559416,
      "learning_rate": 0.00018508129697048125,
      "loss": 13.1372,
      "step": 4594
    },
    {
      "epoch": 0.8034972677595629,
      "grad_norm": 0.8657572722189788,
      "learning_rate": 0.0001850738550157181,
      "loss": 12.9274,
      "step": 4595
    },
    {
      "epoch": 0.803672131147541,
      "grad_norm": 0.8414838766819653,
      "learning_rate": 0.00018506641135496293,
      "loss": 13.1758,
      "step": 4596
    },
    {
      "epoch": 0.8038469945355191,
      "grad_norm": 0.8599761071331414,
      "learning_rate": 0.00018505896598836508,
      "loss": 13.092,
      "step": 4597
    },
    {
      "epoch": 0.8040218579234972,
      "grad_norm": 0.8260597863265878,
      "learning_rate": 0.00018505151891607375,
      "loss": 12.999,
      "step": 4598
    },
    {
      "epoch": 0.8041967213114755,
      "grad_norm": 0.8625659527880096,
      "learning_rate": 0.00018504407013823836,
      "loss": 13.2037,
      "step": 4599
    },
    {
      "epoch": 0.8043715846994536,
      "grad_norm": 0.7605771754603672,
      "learning_rate": 0.00018503661965500824,
      "loss": 12.911,
      "step": 4600
    },
    {
      "epoch": 0.8045464480874317,
      "grad_norm": 0.7991403110172289,
      "learning_rate": 0.00018502916746653277,
      "loss": 13.0451,
      "step": 4601
    },
    {
      "epoch": 0.8047213114754098,
      "grad_norm": 0.8763505659363816,
      "learning_rate": 0.00018502171357296144,
      "loss": 12.9723,
      "step": 4602
    },
    {
      "epoch": 0.8048961748633879,
      "grad_norm": 0.7991587332122074,
      "learning_rate": 0.0001850142579744437,
      "loss": 12.9098,
      "step": 4603
    },
    {
      "epoch": 0.8050710382513662,
      "grad_norm": 0.8368967828035558,
      "learning_rate": 0.00018500680067112904,
      "loss": 13.1143,
      "step": 4604
    },
    {
      "epoch": 0.8052459016393443,
      "grad_norm": 0.9113869551839487,
      "learning_rate": 0.000184999341663167,
      "loss": 13.0943,
      "step": 4605
    },
    {
      "epoch": 0.8054207650273224,
      "grad_norm": 0.8121564947693238,
      "learning_rate": 0.00018499188095070723,
      "loss": 12.6515,
      "step": 4606
    },
    {
      "epoch": 0.8055956284153005,
      "grad_norm": 0.8513555527279024,
      "learning_rate": 0.00018498441853389929,
      "loss": 13.2451,
      "step": 4607
    },
    {
      "epoch": 0.8057704918032786,
      "grad_norm": 0.8017439039614723,
      "learning_rate": 0.0001849769544128928,
      "loss": 12.4588,
      "step": 4608
    },
    {
      "epoch": 0.8059453551912569,
      "grad_norm": 0.8071737706500862,
      "learning_rate": 0.00018496948858783748,
      "loss": 13.0098,
      "step": 4609
    },
    {
      "epoch": 0.806120218579235,
      "grad_norm": 0.8632066546087966,
      "learning_rate": 0.00018496202105888299,
      "loss": 13.0115,
      "step": 4610
    },
    {
      "epoch": 0.8062950819672131,
      "grad_norm": 0.7928865257664099,
      "learning_rate": 0.00018495455182617913,
      "loss": 13.0753,
      "step": 4611
    },
    {
      "epoch": 0.8064699453551912,
      "grad_norm": 0.8645734754023564,
      "learning_rate": 0.00018494708088987564,
      "loss": 12.8196,
      "step": 4612
    },
    {
      "epoch": 0.8066448087431694,
      "grad_norm": 0.8188947770107377,
      "learning_rate": 0.00018493960825012237,
      "loss": 12.9153,
      "step": 4613
    },
    {
      "epoch": 0.8068196721311476,
      "grad_norm": 0.7950520789737016,
      "learning_rate": 0.00018493213390706917,
      "loss": 12.9291,
      "step": 4614
    },
    {
      "epoch": 0.8069945355191257,
      "grad_norm": 0.7975193228539132,
      "learning_rate": 0.0001849246578608659,
      "loss": 12.9266,
      "step": 4615
    },
    {
      "epoch": 0.8071693989071038,
      "grad_norm": 0.7325791222704094,
      "learning_rate": 0.00018491718011166247,
      "loss": 13.2002,
      "step": 4616
    },
    {
      "epoch": 0.8073442622950819,
      "grad_norm": 0.8224135597213633,
      "learning_rate": 0.0001849097006596089,
      "loss": 12.9137,
      "step": 4617
    },
    {
      "epoch": 0.8075191256830601,
      "grad_norm": 0.8234239849598005,
      "learning_rate": 0.00018490221950485507,
      "loss": 13.0461,
      "step": 4618
    },
    {
      "epoch": 0.8076939890710383,
      "grad_norm": 0.910154271072494,
      "learning_rate": 0.00018489473664755107,
      "loss": 12.9863,
      "step": 4619
    },
    {
      "epoch": 0.8078688524590164,
      "grad_norm": 0.8515818782561526,
      "learning_rate": 0.00018488725208784694,
      "loss": 12.7818,
      "step": 4620
    },
    {
      "epoch": 0.8080437158469945,
      "grad_norm": 0.8248649216617209,
      "learning_rate": 0.00018487976582589276,
      "loss": 13.2755,
      "step": 4621
    },
    {
      "epoch": 0.8082185792349726,
      "grad_norm": 0.8628103344437962,
      "learning_rate": 0.0001848722778618387,
      "loss": 13.0698,
      "step": 4622
    },
    {
      "epoch": 0.8083934426229509,
      "grad_norm": 0.9700094248035281,
      "learning_rate": 0.00018486478819583485,
      "loss": 12.9714,
      "step": 4623
    },
    {
      "epoch": 0.808568306010929,
      "grad_norm": 0.860849877705533,
      "learning_rate": 0.00018485729682803142,
      "loss": 13.0757,
      "step": 4624
    },
    {
      "epoch": 0.8087431693989071,
      "grad_norm": 1.0737214310149972,
      "learning_rate": 0.00018484980375857863,
      "loss": 13.0273,
      "step": 4625
    },
    {
      "epoch": 0.8089180327868852,
      "grad_norm": 0.8436889857147509,
      "learning_rate": 0.00018484230898762677,
      "loss": 13.0311,
      "step": 4626
    },
    {
      "epoch": 0.8090928961748634,
      "grad_norm": 1.1099011424406107,
      "learning_rate": 0.00018483481251532614,
      "loss": 13.0162,
      "step": 4627
    },
    {
      "epoch": 0.8092677595628416,
      "grad_norm": 0.8646050864707974,
      "learning_rate": 0.00018482731434182702,
      "loss": 13.0992,
      "step": 4628
    },
    {
      "epoch": 0.8094426229508197,
      "grad_norm": 1.0625926924219917,
      "learning_rate": 0.00018481981446727977,
      "loss": 12.8322,
      "step": 4629
    },
    {
      "epoch": 0.8096174863387978,
      "grad_norm": 0.9701287447698783,
      "learning_rate": 0.00018481231289183483,
      "loss": 12.9182,
      "step": 4630
    },
    {
      "epoch": 0.8097923497267759,
      "grad_norm": 0.9338441661987319,
      "learning_rate": 0.0001848048096156426,
      "loss": 12.998,
      "step": 4631
    },
    {
      "epoch": 0.8099672131147541,
      "grad_norm": 0.954839547612619,
      "learning_rate": 0.00018479730463885356,
      "loss": 12.9976,
      "step": 4632
    },
    {
      "epoch": 0.8101420765027323,
      "grad_norm": 0.8003751326326799,
      "learning_rate": 0.0001847897979616182,
      "loss": 12.8299,
      "step": 4633
    },
    {
      "epoch": 0.8103169398907104,
      "grad_norm": 0.8969273681766741,
      "learning_rate": 0.00018478228958408705,
      "loss": 12.8881,
      "step": 4634
    },
    {
      "epoch": 0.8104918032786885,
      "grad_norm": 0.8542773993093283,
      "learning_rate": 0.0001847747795064107,
      "loss": 13.0936,
      "step": 4635
    },
    {
      "epoch": 0.8106666666666666,
      "grad_norm": 1.019000731135211,
      "learning_rate": 0.0001847672677287397,
      "loss": 12.846,
      "step": 4636
    },
    {
      "epoch": 0.8108415300546448,
      "grad_norm": 0.8814918147535944,
      "learning_rate": 0.0001847597542512247,
      "loss": 13.0154,
      "step": 4637
    },
    {
      "epoch": 0.811016393442623,
      "grad_norm": 1.099822873439941,
      "learning_rate": 0.00018475223907401638,
      "loss": 12.8196,
      "step": 4638
    },
    {
      "epoch": 0.8111912568306011,
      "grad_norm": 0.8504309266977255,
      "learning_rate": 0.00018474472219726548,
      "loss": 12.9972,
      "step": 4639
    },
    {
      "epoch": 0.8113661202185792,
      "grad_norm": 0.8736424115243158,
      "learning_rate": 0.0001847372036211227,
      "loss": 12.9135,
      "step": 4640
    },
    {
      "epoch": 0.8115409836065574,
      "grad_norm": 0.8558184730154671,
      "learning_rate": 0.00018472968334573879,
      "loss": 13.1164,
      "step": 4641
    },
    {
      "epoch": 0.8117158469945355,
      "grad_norm": 0.8574538658410223,
      "learning_rate": 0.0001847221613712646,
      "loss": 13.1458,
      "step": 4642
    },
    {
      "epoch": 0.8118907103825137,
      "grad_norm": 0.7957097334811848,
      "learning_rate": 0.0001847146376978509,
      "loss": 12.8645,
      "step": 4643
    },
    {
      "epoch": 0.8120655737704918,
      "grad_norm": 0.830316048202096,
      "learning_rate": 0.0001847071123256486,
      "loss": 12.9738,
      "step": 4644
    },
    {
      "epoch": 0.8122404371584699,
      "grad_norm": 0.8633154753046743,
      "learning_rate": 0.00018469958525480865,
      "loss": 13.3002,
      "step": 4645
    },
    {
      "epoch": 0.8124153005464481,
      "grad_norm": 0.9550587471622713,
      "learning_rate": 0.00018469205648548195,
      "loss": 13.14,
      "step": 4646
    },
    {
      "epoch": 0.8125901639344262,
      "grad_norm": 0.8723566913180508,
      "learning_rate": 0.00018468452601781948,
      "loss": 13.0462,
      "step": 4647
    },
    {
      "epoch": 0.8127650273224044,
      "grad_norm": 0.8309357109810089,
      "learning_rate": 0.00018467699385197226,
      "loss": 13.1014,
      "step": 4648
    },
    {
      "epoch": 0.8129398907103825,
      "grad_norm": 0.8172392352712932,
      "learning_rate": 0.00018466945998809128,
      "loss": 13.0169,
      "step": 4649
    },
    {
      "epoch": 0.8131147540983606,
      "grad_norm": 0.9407393073292563,
      "learning_rate": 0.0001846619244263277,
      "loss": 13.0395,
      "step": 4650
    },
    {
      "epoch": 0.8132896174863388,
      "grad_norm": 0.7978911492173271,
      "learning_rate": 0.00018465438716683255,
      "loss": 12.9114,
      "step": 4651
    },
    {
      "epoch": 0.8134644808743169,
      "grad_norm": 1.0822247535179406,
      "learning_rate": 0.00018464684820975702,
      "loss": 13.0972,
      "step": 4652
    },
    {
      "epoch": 0.8136393442622951,
      "grad_norm": 0.9625678971124508,
      "learning_rate": 0.0001846393075552523,
      "loss": 12.8389,
      "step": 4653
    },
    {
      "epoch": 0.8138142076502732,
      "grad_norm": 0.9336360409059964,
      "learning_rate": 0.00018463176520346959,
      "loss": 12.8501,
      "step": 4654
    },
    {
      "epoch": 0.8139890710382514,
      "grad_norm": 0.8898552614691401,
      "learning_rate": 0.00018462422115456012,
      "loss": 12.9252,
      "step": 4655
    },
    {
      "epoch": 0.8141639344262295,
      "grad_norm": 0.8142804977514362,
      "learning_rate": 0.0001846166754086752,
      "loss": 12.9374,
      "step": 4656
    },
    {
      "epoch": 0.8143387978142077,
      "grad_norm": 0.8325495437275643,
      "learning_rate": 0.00018460912796596613,
      "loss": 13.1154,
      "step": 4657
    },
    {
      "epoch": 0.8145136612021858,
      "grad_norm": 0.8933245346073163,
      "learning_rate": 0.00018460157882658424,
      "loss": 13.0689,
      "step": 4658
    },
    {
      "epoch": 0.8146885245901639,
      "grad_norm": 0.7799884863427453,
      "learning_rate": 0.00018459402799068096,
      "loss": 13.0962,
      "step": 4659
    },
    {
      "epoch": 0.8148633879781421,
      "grad_norm": 0.7990310228921478,
      "learning_rate": 0.00018458647545840763,
      "loss": 13.078,
      "step": 4660
    },
    {
      "epoch": 0.8150382513661202,
      "grad_norm": 0.8719917015591419,
      "learning_rate": 0.0001845789212299158,
      "loss": 13.0546,
      "step": 4661
    },
    {
      "epoch": 0.8152131147540984,
      "grad_norm": 0.7327878708738997,
      "learning_rate": 0.0001845713653053569,
      "loss": 12.7949,
      "step": 4662
    },
    {
      "epoch": 0.8153879781420765,
      "grad_norm": 0.8767497819771385,
      "learning_rate": 0.00018456380768488246,
      "loss": 13.0052,
      "step": 4663
    },
    {
      "epoch": 0.8155628415300546,
      "grad_norm": 0.7654931148094029,
      "learning_rate": 0.00018455624836864404,
      "loss": 13.053,
      "step": 4664
    },
    {
      "epoch": 0.8157377049180328,
      "grad_norm": 0.9462928152894021,
      "learning_rate": 0.0001845486873567932,
      "loss": 12.9258,
      "step": 4665
    },
    {
      "epoch": 0.8159125683060109,
      "grad_norm": 0.8675095268874271,
      "learning_rate": 0.00018454112464948158,
      "loss": 13.0349,
      "step": 4666
    },
    {
      "epoch": 0.8160874316939891,
      "grad_norm": 0.7952774051725775,
      "learning_rate": 0.00018453356024686083,
      "loss": 13.0005,
      "step": 4667
    },
    {
      "epoch": 0.8162622950819672,
      "grad_norm": 0.8980015393754894,
      "learning_rate": 0.00018452599414908264,
      "loss": 13.032,
      "step": 4668
    },
    {
      "epoch": 0.8164371584699454,
      "grad_norm": 0.8497872349984429,
      "learning_rate": 0.00018451842635629876,
      "loss": 13.1867,
      "step": 4669
    },
    {
      "epoch": 0.8166120218579235,
      "grad_norm": 0.9594131747852597,
      "learning_rate": 0.0001845108568686609,
      "loss": 12.9402,
      "step": 4670
    },
    {
      "epoch": 0.8167868852459016,
      "grad_norm": 0.8973102585094729,
      "learning_rate": 0.0001845032856863209,
      "loss": 12.6487,
      "step": 4671
    },
    {
      "epoch": 0.8169617486338798,
      "grad_norm": 0.9703195494592313,
      "learning_rate": 0.00018449571280943058,
      "loss": 13.0534,
      "step": 4672
    },
    {
      "epoch": 0.8171366120218579,
      "grad_norm": 1.0734284519533412,
      "learning_rate": 0.00018448813823814179,
      "loss": 13.1215,
      "step": 4673
    },
    {
      "epoch": 0.8173114754098361,
      "grad_norm": 0.8708149105422482,
      "learning_rate": 0.00018448056197260636,
      "loss": 13.1286,
      "step": 4674
    },
    {
      "epoch": 0.8174863387978142,
      "grad_norm": 1.047331593210076,
      "learning_rate": 0.0001844729840129763,
      "loss": 13.1811,
      "step": 4675
    },
    {
      "epoch": 0.8176612021857923,
      "grad_norm": 0.8500650302729282,
      "learning_rate": 0.00018446540435940355,
      "loss": 13.2871,
      "step": 4676
    },
    {
      "epoch": 0.8178360655737705,
      "grad_norm": 1.060977956453804,
      "learning_rate": 0.0001844578230120401,
      "loss": 13.005,
      "step": 4677
    },
    {
      "epoch": 0.8180109289617487,
      "grad_norm": 0.8962207935941724,
      "learning_rate": 0.00018445023997103795,
      "loss": 12.832,
      "step": 4678
    },
    {
      "epoch": 0.8181857923497268,
      "grad_norm": 1.0170865401319906,
      "learning_rate": 0.00018444265523654923,
      "loss": 13.0589,
      "step": 4679
    },
    {
      "epoch": 0.8183606557377049,
      "grad_norm": 0.8713133463642145,
      "learning_rate": 0.00018443506880872596,
      "loss": 13.0438,
      "step": 4680
    },
    {
      "epoch": 0.818535519125683,
      "grad_norm": 1.0171920758877464,
      "learning_rate": 0.00018442748068772034,
      "loss": 13.2741,
      "step": 4681
    },
    {
      "epoch": 0.8187103825136612,
      "grad_norm": 0.9469702332530525,
      "learning_rate": 0.00018441989087368449,
      "loss": 12.8193,
      "step": 4682
    },
    {
      "epoch": 0.8188852459016394,
      "grad_norm": 0.9382612948666541,
      "learning_rate": 0.00018441229936677064,
      "loss": 12.8546,
      "step": 4683
    },
    {
      "epoch": 0.8190601092896175,
      "grad_norm": 0.9265520135196568,
      "learning_rate": 0.000184404706167131,
      "loss": 13.1361,
      "step": 4684
    },
    {
      "epoch": 0.8192349726775956,
      "grad_norm": 0.8629022403872516,
      "learning_rate": 0.00018439711127491787,
      "loss": 13.0176,
      "step": 4685
    },
    {
      "epoch": 0.8194098360655737,
      "grad_norm": 0.8814724350493484,
      "learning_rate": 0.00018438951469028353,
      "loss": 12.8593,
      "step": 4686
    },
    {
      "epoch": 0.819584699453552,
      "grad_norm": 0.9856465922889744,
      "learning_rate": 0.00018438191641338025,
      "loss": 13.1332,
      "step": 4687
    },
    {
      "epoch": 0.8197595628415301,
      "grad_norm": 1.0981385787991786,
      "learning_rate": 0.0001843743164443605,
      "loss": 13.1074,
      "step": 4688
    },
    {
      "epoch": 0.8199344262295082,
      "grad_norm": 0.8591257323568943,
      "learning_rate": 0.00018436671478337666,
      "loss": 13.1575,
      "step": 4689
    },
    {
      "epoch": 0.8201092896174863,
      "grad_norm": 1.0861624659197602,
      "learning_rate": 0.0001843591114305811,
      "loss": 13.3129,
      "step": 4690
    },
    {
      "epoch": 0.8202841530054645,
      "grad_norm": 0.8344587468113175,
      "learning_rate": 0.00018435150638612636,
      "loss": 12.9777,
      "step": 4691
    },
    {
      "epoch": 0.8204590163934427,
      "grad_norm": 0.9391461862025224,
      "learning_rate": 0.00018434389965016495,
      "loss": 13.1046,
      "step": 4692
    },
    {
      "epoch": 0.8206338797814208,
      "grad_norm": 0.8357614674397769,
      "learning_rate": 0.00018433629122284936,
      "loss": 13.198,
      "step": 4693
    },
    {
      "epoch": 0.8208087431693989,
      "grad_norm": 0.8834284007939651,
      "learning_rate": 0.00018432868110433218,
      "loss": 13.2428,
      "step": 4694
    },
    {
      "epoch": 0.820983606557377,
      "grad_norm": 0.8967409732727508,
      "learning_rate": 0.00018432106929476604,
      "loss": 12.9931,
      "step": 4695
    },
    {
      "epoch": 0.8211584699453552,
      "grad_norm": 0.9048250657970538,
      "learning_rate": 0.00018431345579430355,
      "loss": 13.0184,
      "step": 4696
    },
    {
      "epoch": 0.8213333333333334,
      "grad_norm": 0.8153215925749631,
      "learning_rate": 0.00018430584060309736,
      "loss": 13.0191,
      "step": 4697
    },
    {
      "epoch": 0.8215081967213115,
      "grad_norm": 1.0489082953724354,
      "learning_rate": 0.00018429822372130025,
      "loss": 12.8501,
      "step": 4698
    },
    {
      "epoch": 0.8216830601092896,
      "grad_norm": 0.7731841172150439,
      "learning_rate": 0.00018429060514906493,
      "loss": 12.9221,
      "step": 4699
    },
    {
      "epoch": 0.8218579234972677,
      "grad_norm": 0.942767747615448,
      "learning_rate": 0.00018428298488654417,
      "loss": 13.0679,
      "step": 4700
    },
    {
      "epoch": 0.822032786885246,
      "grad_norm": 0.8548556166043145,
      "learning_rate": 0.00018427536293389075,
      "loss": 13.0926,
      "step": 4701
    },
    {
      "epoch": 0.8222076502732241,
      "grad_norm": 0.9675790868858596,
      "learning_rate": 0.00018426773929125756,
      "loss": 12.9638,
      "step": 4702
    },
    {
      "epoch": 0.8223825136612022,
      "grad_norm": 0.9007177901384611,
      "learning_rate": 0.00018426011395879748,
      "loss": 13.0801,
      "step": 4703
    },
    {
      "epoch": 0.8225573770491803,
      "grad_norm": 0.8614291157335459,
      "learning_rate": 0.0001842524869366634,
      "loss": 12.6457,
      "step": 4704
    },
    {
      "epoch": 0.8227322404371584,
      "grad_norm": 0.9712340390386881,
      "learning_rate": 0.00018424485822500825,
      "loss": 12.9308,
      "step": 4705
    },
    {
      "epoch": 0.8229071038251367,
      "grad_norm": 0.8351079648638174,
      "learning_rate": 0.00018423722782398504,
      "loss": 13.1235,
      "step": 4706
    },
    {
      "epoch": 0.8230819672131148,
      "grad_norm": 0.8839933769722246,
      "learning_rate": 0.00018422959573374676,
      "loss": 12.9063,
      "step": 4707
    },
    {
      "epoch": 0.8232568306010929,
      "grad_norm": 0.8206244095008738,
      "learning_rate": 0.00018422196195444645,
      "loss": 12.978,
      "step": 4708
    },
    {
      "epoch": 0.823431693989071,
      "grad_norm": 0.868997154247671,
      "learning_rate": 0.00018421432648623722,
      "loss": 13.0588,
      "step": 4709
    },
    {
      "epoch": 0.8236065573770491,
      "grad_norm": 0.8168521172781122,
      "learning_rate": 0.0001842066893292722,
      "loss": 12.9074,
      "step": 4710
    },
    {
      "epoch": 0.8237814207650274,
      "grad_norm": 0.9936914094041828,
      "learning_rate": 0.0001841990504837045,
      "loss": 12.6517,
      "step": 4711
    },
    {
      "epoch": 0.8239562841530055,
      "grad_norm": 0.7509011823529878,
      "learning_rate": 0.0001841914099496873,
      "loss": 13.0102,
      "step": 4712
    },
    {
      "epoch": 0.8241311475409836,
      "grad_norm": 0.9276433527138535,
      "learning_rate": 0.00018418376772737382,
      "loss": 12.9887,
      "step": 4713
    },
    {
      "epoch": 0.8243060109289617,
      "grad_norm": 0.8542027767315943,
      "learning_rate": 0.00018417612381691734,
      "loss": 13.102,
      "step": 4714
    },
    {
      "epoch": 0.8244808743169398,
      "grad_norm": 1.1503073863333875,
      "learning_rate": 0.0001841684782184711,
      "loss": 12.834,
      "step": 4715
    },
    {
      "epoch": 0.8246557377049181,
      "grad_norm": 0.9512335465284796,
      "learning_rate": 0.00018416083093218845,
      "loss": 12.7959,
      "step": 4716
    },
    {
      "epoch": 0.8248306010928962,
      "grad_norm": 1.0379810023537626,
      "learning_rate": 0.00018415318195822274,
      "loss": 12.9645,
      "step": 4717
    },
    {
      "epoch": 0.8250054644808743,
      "grad_norm": 0.984987657310595,
      "learning_rate": 0.00018414553129672732,
      "loss": 12.9172,
      "step": 4718
    },
    {
      "epoch": 0.8251803278688524,
      "grad_norm": 0.9522347254450738,
      "learning_rate": 0.0001841378789478557,
      "loss": 13.1378,
      "step": 4719
    },
    {
      "epoch": 0.8253551912568307,
      "grad_norm": 1.0323828843843097,
      "learning_rate": 0.0001841302249117612,
      "loss": 12.7429,
      "step": 4720
    },
    {
      "epoch": 0.8255300546448088,
      "grad_norm": 0.9569921027529233,
      "learning_rate": 0.00018412256918859743,
      "loss": 13.3789,
      "step": 4721
    },
    {
      "epoch": 0.8257049180327869,
      "grad_norm": 1.0170510739694423,
      "learning_rate": 0.00018411491177851782,
      "loss": 12.9617,
      "step": 4722
    },
    {
      "epoch": 0.825879781420765,
      "grad_norm": 1.0383149423008096,
      "learning_rate": 0.00018410725268167596,
      "loss": 13.1791,
      "step": 4723
    },
    {
      "epoch": 0.8260546448087431,
      "grad_norm": 1.0149761831882793,
      "learning_rate": 0.0001840995918982255,
      "loss": 12.9174,
      "step": 4724
    },
    {
      "epoch": 0.8262295081967214,
      "grad_norm": 1.2173977112141663,
      "learning_rate": 0.00018409192942831995,
      "loss": 13.0177,
      "step": 4725
    },
    {
      "epoch": 0.8264043715846995,
      "grad_norm": 1.0641942143797776,
      "learning_rate": 0.00018408426527211303,
      "loss": 12.8838,
      "step": 4726
    },
    {
      "epoch": 0.8265792349726776,
      "grad_norm": 1.1043111700359811,
      "learning_rate": 0.0001840765994297584,
      "loss": 13.101,
      "step": 4727
    },
    {
      "epoch": 0.8267540983606557,
      "grad_norm": 0.9061742003754679,
      "learning_rate": 0.0001840689319014098,
      "loss": 12.9303,
      "step": 4728
    },
    {
      "epoch": 0.8269289617486338,
      "grad_norm": 1.1016316306590672,
      "learning_rate": 0.00018406126268722103,
      "loss": 13.0813,
      "step": 4729
    },
    {
      "epoch": 0.8271038251366121,
      "grad_norm": 0.9078113495716112,
      "learning_rate": 0.00018405359178734582,
      "loss": 13.0653,
      "step": 4730
    },
    {
      "epoch": 0.8272786885245902,
      "grad_norm": 1.0681208917592613,
      "learning_rate": 0.000184045919201938,
      "loss": 12.8081,
      "step": 4731
    },
    {
      "epoch": 0.8274535519125683,
      "grad_norm": 0.8329227186445749,
      "learning_rate": 0.00018403824493115148,
      "loss": 13.0415,
      "step": 4732
    },
    {
      "epoch": 0.8276284153005464,
      "grad_norm": 0.9546541812708762,
      "learning_rate": 0.0001840305689751401,
      "loss": 12.8082,
      "step": 4733
    },
    {
      "epoch": 0.8278032786885245,
      "grad_norm": 0.7946493503677374,
      "learning_rate": 0.0001840228913340578,
      "loss": 12.891,
      "step": 4734
    },
    {
      "epoch": 0.8279781420765028,
      "grad_norm": 1.1553425844589473,
      "learning_rate": 0.00018401521200805856,
      "loss": 13.2459,
      "step": 4735
    },
    {
      "epoch": 0.8281530054644809,
      "grad_norm": 0.7999747031648378,
      "learning_rate": 0.00018400753099729635,
      "loss": 12.9817,
      "step": 4736
    },
    {
      "epoch": 0.828327868852459,
      "grad_norm": 0.8787765871513966,
      "learning_rate": 0.00018399984830192522,
      "loss": 13.0758,
      "step": 4737
    },
    {
      "epoch": 0.8285027322404371,
      "grad_norm": 0.8252056071620607,
      "learning_rate": 0.00018399216392209916,
      "loss": 13.1344,
      "step": 4738
    },
    {
      "epoch": 0.8286775956284153,
      "grad_norm": 0.8395908904666436,
      "learning_rate": 0.00018398447785797237,
      "loss": 12.8688,
      "step": 4739
    },
    {
      "epoch": 0.8288524590163935,
      "grad_norm": 0.7403614284518193,
      "learning_rate": 0.00018397679010969891,
      "loss": 12.9762,
      "step": 4740
    },
    {
      "epoch": 0.8290273224043716,
      "grad_norm": 0.8854177330516009,
      "learning_rate": 0.00018396910067743302,
      "loss": 12.7295,
      "step": 4741
    },
    {
      "epoch": 0.8292021857923497,
      "grad_norm": 0.8170403016131584,
      "learning_rate": 0.00018396140956132882,
      "loss": 13.0783,
      "step": 4742
    },
    {
      "epoch": 0.8293770491803278,
      "grad_norm": 0.8611920103812186,
      "learning_rate": 0.00018395371676154055,
      "loss": 12.654,
      "step": 4743
    },
    {
      "epoch": 0.829551912568306,
      "grad_norm": 0.7711589452772212,
      "learning_rate": 0.00018394602227822247,
      "loss": 12.9113,
      "step": 4744
    },
    {
      "epoch": 0.8297267759562842,
      "grad_norm": 0.8479989334937529,
      "learning_rate": 0.00018393832611152892,
      "loss": 13.0081,
      "step": 4745
    },
    {
      "epoch": 0.8299016393442623,
      "grad_norm": 0.8642139699112351,
      "learning_rate": 0.00018393062826161418,
      "loss": 12.8725,
      "step": 4746
    },
    {
      "epoch": 0.8300765027322404,
      "grad_norm": 0.8952636453607528,
      "learning_rate": 0.00018392292872863267,
      "loss": 12.8429,
      "step": 4747
    },
    {
      "epoch": 0.8302513661202185,
      "grad_norm": 0.833359301049981,
      "learning_rate": 0.00018391522751273875,
      "loss": 12.9462,
      "step": 4748
    },
    {
      "epoch": 0.8304262295081967,
      "grad_norm": 0.8514574209731881,
      "learning_rate": 0.0001839075246140869,
      "loss": 12.7646,
      "step": 4749
    },
    {
      "epoch": 0.8306010928961749,
      "grad_norm": 0.8656480524787546,
      "learning_rate": 0.0001838998200328315,
      "loss": 13.0325,
      "step": 4750
    },
    {
      "epoch": 0.830775956284153,
      "grad_norm": 0.8584676065392577,
      "learning_rate": 0.00018389211376912713,
      "loss": 13.0113,
      "step": 4751
    },
    {
      "epoch": 0.8309508196721311,
      "grad_norm": 0.8702749564513673,
      "learning_rate": 0.00018388440582312827,
      "loss": 12.826,
      "step": 4752
    },
    {
      "epoch": 0.8311256830601093,
      "grad_norm": 0.9068539230325576,
      "learning_rate": 0.00018387669619498954,
      "loss": 12.9226,
      "step": 4753
    },
    {
      "epoch": 0.8313005464480875,
      "grad_norm": 0.8617637283769218,
      "learning_rate": 0.0001838689848848655,
      "loss": 13.0484,
      "step": 4754
    },
    {
      "epoch": 0.8314754098360656,
      "grad_norm": 0.9412424042707643,
      "learning_rate": 0.00018386127189291084,
      "loss": 13.2215,
      "step": 4755
    },
    {
      "epoch": 0.8316502732240437,
      "grad_norm": 0.7981543377385956,
      "learning_rate": 0.00018385355721928012,
      "loss": 13.0317,
      "step": 4756
    },
    {
      "epoch": 0.8318251366120218,
      "grad_norm": 0.8716834961252706,
      "learning_rate": 0.00018384584086412816,
      "loss": 13.1479,
      "step": 4757
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.861363474910785,
      "learning_rate": 0.00018383812282760963,
      "loss": 12.9839,
      "step": 4758
    },
    {
      "epoch": 0.8321748633879782,
      "grad_norm": 0.8141479342063884,
      "learning_rate": 0.00018383040310987936,
      "loss": 12.7204,
      "step": 4759
    },
    {
      "epoch": 0.8323497267759563,
      "grad_norm": 0.8463845990564114,
      "learning_rate": 0.00018382268171109208,
      "loss": 13.005,
      "step": 4760
    },
    {
      "epoch": 0.8325245901639344,
      "grad_norm": 0.8028673729403353,
      "learning_rate": 0.00018381495863140266,
      "loss": 12.9257,
      "step": 4761
    },
    {
      "epoch": 0.8326994535519125,
      "grad_norm": 0.8383669241874673,
      "learning_rate": 0.000183807233870966,
      "loss": 13.0594,
      "step": 4762
    },
    {
      "epoch": 0.8328743169398907,
      "grad_norm": 0.9319503121470188,
      "learning_rate": 0.00018379950742993697,
      "loss": 13.093,
      "step": 4763
    },
    {
      "epoch": 0.8330491803278689,
      "grad_norm": 0.7461312018461344,
      "learning_rate": 0.0001837917793084705,
      "loss": 13.0886,
      "step": 4764
    },
    {
      "epoch": 0.833224043715847,
      "grad_norm": 0.8843801589493352,
      "learning_rate": 0.0001837840495067216,
      "loss": 13.1715,
      "step": 4765
    },
    {
      "epoch": 0.8333989071038251,
      "grad_norm": 0.8516641459201589,
      "learning_rate": 0.00018377631802484525,
      "loss": 13.0492,
      "step": 4766
    },
    {
      "epoch": 0.8335737704918033,
      "grad_norm": 0.9120588942076658,
      "learning_rate": 0.00018376858486299647,
      "loss": 13.0602,
      "step": 4767
    },
    {
      "epoch": 0.8337486338797814,
      "grad_norm": 0.8957299124157733,
      "learning_rate": 0.0001837608500213304,
      "loss": 13.1127,
      "step": 4768
    },
    {
      "epoch": 0.8339234972677596,
      "grad_norm": 0.853255877739365,
      "learning_rate": 0.00018375311350000208,
      "loss": 12.9668,
      "step": 4769
    },
    {
      "epoch": 0.8340983606557377,
      "grad_norm": 0.8962671375624662,
      "learning_rate": 0.00018374537529916669,
      "loss": 12.8454,
      "step": 4770
    },
    {
      "epoch": 0.8342732240437158,
      "grad_norm": 0.8499956027107035,
      "learning_rate": 0.0001837376354189794,
      "loss": 12.8947,
      "step": 4771
    },
    {
      "epoch": 0.834448087431694,
      "grad_norm": 0.7701437070623244,
      "learning_rate": 0.0001837298938595954,
      "loss": 12.9321,
      "step": 4772
    },
    {
      "epoch": 0.8346229508196721,
      "grad_norm": 0.8320802118491071,
      "learning_rate": 0.00018372215062116998,
      "loss": 13.0445,
      "step": 4773
    },
    {
      "epoch": 0.8347978142076503,
      "grad_norm": 0.8720721779239207,
      "learning_rate": 0.0001837144057038583,
      "loss": 12.9493,
      "step": 4774
    },
    {
      "epoch": 0.8349726775956284,
      "grad_norm": 0.7982984745089403,
      "learning_rate": 0.0001837066591078158,
      "loss": 12.9636,
      "step": 4775
    },
    {
      "epoch": 0.8351475409836066,
      "grad_norm": 0.7704758458289632,
      "learning_rate": 0.00018369891083319778,
      "loss": 13.0088,
      "step": 4776
    },
    {
      "epoch": 0.8353224043715847,
      "grad_norm": 0.735493375722011,
      "learning_rate": 0.0001836911608801596,
      "loss": 13.0379,
      "step": 4777
    },
    {
      "epoch": 0.8354972677595628,
      "grad_norm": 0.8580437097230658,
      "learning_rate": 0.0001836834092488567,
      "loss": 12.729,
      "step": 4778
    },
    {
      "epoch": 0.835672131147541,
      "grad_norm": 0.7656875403563825,
      "learning_rate": 0.00018367565593944448,
      "loss": 13.0417,
      "step": 4779
    },
    {
      "epoch": 0.8358469945355191,
      "grad_norm": 0.857782652044013,
      "learning_rate": 0.00018366790095207842,
      "loss": 12.7337,
      "step": 4780
    },
    {
      "epoch": 0.8360218579234973,
      "grad_norm": 0.8012027360179552,
      "learning_rate": 0.00018366014428691408,
      "loss": 12.938,
      "step": 4781
    },
    {
      "epoch": 0.8361967213114754,
      "grad_norm": 0.8983852257858803,
      "learning_rate": 0.00018365238594410696,
      "loss": 13.0385,
      "step": 4782
    },
    {
      "epoch": 0.8363715846994536,
      "grad_norm": 0.7877051574805813,
      "learning_rate": 0.00018364462592381265,
      "loss": 12.7643,
      "step": 4783
    },
    {
      "epoch": 0.8365464480874317,
      "grad_norm": 0.9410865446424388,
      "learning_rate": 0.0001836368642261868,
      "loss": 12.98,
      "step": 4784
    },
    {
      "epoch": 0.8367213114754098,
      "grad_norm": 0.9225015793498734,
      "learning_rate": 0.000183629100851385,
      "loss": 12.8144,
      "step": 4785
    },
    {
      "epoch": 0.836896174863388,
      "grad_norm": 0.8598389626983255,
      "learning_rate": 0.00018362133579956296,
      "loss": 12.9185,
      "step": 4786
    },
    {
      "epoch": 0.8370710382513661,
      "grad_norm": 0.8216435074591385,
      "learning_rate": 0.00018361356907087636,
      "loss": 12.9529,
      "step": 4787
    },
    {
      "epoch": 0.8372459016393443,
      "grad_norm": 0.8432991387438193,
      "learning_rate": 0.000183605800665481,
      "loss": 12.976,
      "step": 4788
    },
    {
      "epoch": 0.8374207650273224,
      "grad_norm": 0.8058108501477795,
      "learning_rate": 0.00018359803058353265,
      "loss": 13.0225,
      "step": 4789
    },
    {
      "epoch": 0.8375956284153006,
      "grad_norm": 0.8816441434686637,
      "learning_rate": 0.00018359025882518708,
      "loss": 12.8666,
      "step": 4790
    },
    {
      "epoch": 0.8377704918032787,
      "grad_norm": 0.9264116691551338,
      "learning_rate": 0.00018358248539060017,
      "loss": 12.9454,
      "step": 4791
    },
    {
      "epoch": 0.8379453551912568,
      "grad_norm": 0.9689294445478641,
      "learning_rate": 0.0001835747102799278,
      "loss": 12.8916,
      "step": 4792
    },
    {
      "epoch": 0.838120218579235,
      "grad_norm": 0.8444845209799907,
      "learning_rate": 0.00018356693349332588,
      "loss": 12.9814,
      "step": 4793
    },
    {
      "epoch": 0.8382950819672131,
      "grad_norm": 0.8617446200275046,
      "learning_rate": 0.00018355915503095034,
      "loss": 12.9825,
      "step": 4794
    },
    {
      "epoch": 0.8384699453551913,
      "grad_norm": 0.8520642981744976,
      "learning_rate": 0.0001835513748929572,
      "loss": 13.0143,
      "step": 4795
    },
    {
      "epoch": 0.8386448087431694,
      "grad_norm": 0.868376450787687,
      "learning_rate": 0.00018354359307950246,
      "loss": 13.0448,
      "step": 4796
    },
    {
      "epoch": 0.8388196721311475,
      "grad_norm": 1.0017813710205303,
      "learning_rate": 0.00018353580959074215,
      "loss": 12.9768,
      "step": 4797
    },
    {
      "epoch": 0.8389945355191257,
      "grad_norm": 0.804216761642095,
      "learning_rate": 0.00018352802442683238,
      "loss": 13.0401,
      "step": 4798
    },
    {
      "epoch": 0.8391693989071038,
      "grad_norm": 0.8706689294028404,
      "learning_rate": 0.00018352023758792926,
      "loss": 12.792,
      "step": 4799
    },
    {
      "epoch": 0.839344262295082,
      "grad_norm": 0.846058985106636,
      "learning_rate": 0.00018351244907418893,
      "loss": 12.8331,
      "step": 4800
    },
    {
      "epoch": 0.8395191256830601,
      "grad_norm": 0.976106856782025,
      "learning_rate": 0.00018350465888576758,
      "loss": 12.8628,
      "step": 4801
    },
    {
      "epoch": 0.8396939890710382,
      "grad_norm": 0.9802180714686045,
      "learning_rate": 0.00018349686702282144,
      "loss": 13.0154,
      "step": 4802
    },
    {
      "epoch": 0.8398688524590164,
      "grad_norm": 0.850494875514107,
      "learning_rate": 0.00018348907348550672,
      "loss": 12.8864,
      "step": 4803
    },
    {
      "epoch": 0.8400437158469946,
      "grad_norm": 0.9976217982893906,
      "learning_rate": 0.00018348127827397975,
      "loss": 12.937,
      "step": 4804
    },
    {
      "epoch": 0.8402185792349727,
      "grad_norm": 0.8063741500918847,
      "learning_rate": 0.00018347348138839683,
      "loss": 12.9879,
      "step": 4805
    },
    {
      "epoch": 0.8403934426229508,
      "grad_norm": 1.0102983409879565,
      "learning_rate": 0.0001834656828289143,
      "loss": 13.1963,
      "step": 4806
    },
    {
      "epoch": 0.8405683060109289,
      "grad_norm": 0.9464455540598642,
      "learning_rate": 0.00018345788259568858,
      "loss": 13.1939,
      "step": 4807
    },
    {
      "epoch": 0.8407431693989071,
      "grad_norm": 1.0402120905974326,
      "learning_rate": 0.00018345008068887603,
      "loss": 12.9643,
      "step": 4808
    },
    {
      "epoch": 0.8409180327868853,
      "grad_norm": 0.9189292912356276,
      "learning_rate": 0.00018344227710863316,
      "loss": 13.0621,
      "step": 4809
    },
    {
      "epoch": 0.8410928961748634,
      "grad_norm": 0.9219126699553718,
      "learning_rate": 0.00018343447185511643,
      "loss": 13.0567,
      "step": 4810
    },
    {
      "epoch": 0.8412677595628415,
      "grad_norm": 0.8711585040769686,
      "learning_rate": 0.00018342666492848235,
      "loss": 12.8561,
      "step": 4811
    },
    {
      "epoch": 0.8414426229508196,
      "grad_norm": 0.8504399387693526,
      "learning_rate": 0.0001834188563288875,
      "loss": 13.0462,
      "step": 4812
    },
    {
      "epoch": 0.8416174863387978,
      "grad_norm": 1.046039751530231,
      "learning_rate": 0.00018341104605648845,
      "loss": 12.9464,
      "step": 4813
    },
    {
      "epoch": 0.841792349726776,
      "grad_norm": 0.8934603808360626,
      "learning_rate": 0.00018340323411144177,
      "loss": 12.9613,
      "step": 4814
    },
    {
      "epoch": 0.8419672131147541,
      "grad_norm": 1.1975982680673949,
      "learning_rate": 0.00018339542049390422,
      "loss": 13.0419,
      "step": 4815
    },
    {
      "epoch": 0.8421420765027322,
      "grad_norm": 0.8785926849826969,
      "learning_rate": 0.00018338760520403242,
      "loss": 13.127,
      "step": 4816
    },
    {
      "epoch": 0.8423169398907104,
      "grad_norm": 0.9737025424201794,
      "learning_rate": 0.0001833797882419831,
      "loss": 13.0714,
      "step": 4817
    },
    {
      "epoch": 0.8424918032786886,
      "grad_norm": 0.9324299774283168,
      "learning_rate": 0.00018337196960791302,
      "loss": 13.1787,
      "step": 4818
    },
    {
      "epoch": 0.8426666666666667,
      "grad_norm": 0.9385945427317481,
      "learning_rate": 0.00018336414930197897,
      "loss": 13.0241,
      "step": 4819
    },
    {
      "epoch": 0.8428415300546448,
      "grad_norm": 1.0520305543150057,
      "learning_rate": 0.00018335632732433777,
      "loss": 13.1256,
      "step": 4820
    },
    {
      "epoch": 0.8430163934426229,
      "grad_norm": 0.7944853813279499,
      "learning_rate": 0.00018334850367514625,
      "loss": 12.9788,
      "step": 4821
    },
    {
      "epoch": 0.8431912568306011,
      "grad_norm": 1.0099482920372906,
      "learning_rate": 0.00018334067835456131,
      "loss": 13.0512,
      "step": 4822
    },
    {
      "epoch": 0.8433661202185793,
      "grad_norm": 0.7516820292719344,
      "learning_rate": 0.0001833328513627399,
      "loss": 13.0308,
      "step": 4823
    },
    {
      "epoch": 0.8435409836065574,
      "grad_norm": 0.9524087379044863,
      "learning_rate": 0.00018332502269983897,
      "loss": 12.9229,
      "step": 4824
    },
    {
      "epoch": 0.8437158469945355,
      "grad_norm": 0.896721757584273,
      "learning_rate": 0.00018331719236601546,
      "loss": 12.8645,
      "step": 4825
    },
    {
      "epoch": 0.8438907103825136,
      "grad_norm": 0.8854147229357686,
      "learning_rate": 0.00018330936036142643,
      "loss": 13.0308,
      "step": 4826
    },
    {
      "epoch": 0.8440655737704919,
      "grad_norm": 0.787584547736169,
      "learning_rate": 0.00018330152668622892,
      "loss": 13.1331,
      "step": 4827
    },
    {
      "epoch": 0.84424043715847,
      "grad_norm": 0.9957413841031462,
      "learning_rate": 0.00018329369134058006,
      "loss": 12.9786,
      "step": 4828
    },
    {
      "epoch": 0.8444153005464481,
      "grad_norm": 0.8685789109058935,
      "learning_rate": 0.00018328585432463693,
      "loss": 12.9918,
      "step": 4829
    },
    {
      "epoch": 0.8445901639344262,
      "grad_norm": 0.7562077523586201,
      "learning_rate": 0.0001832780156385567,
      "loss": 12.8358,
      "step": 4830
    },
    {
      "epoch": 0.8447650273224043,
      "grad_norm": 0.8732578560733691,
      "learning_rate": 0.00018327017528249658,
      "loss": 12.9836,
      "step": 4831
    },
    {
      "epoch": 0.8449398907103826,
      "grad_norm": 0.8327824369773398,
      "learning_rate": 0.00018326233325661377,
      "loss": 12.4674,
      "step": 4832
    },
    {
      "epoch": 0.8451147540983607,
      "grad_norm": 0.8000194043748627,
      "learning_rate": 0.0001832544895610655,
      "loss": 13.1207,
      "step": 4833
    },
    {
      "epoch": 0.8452896174863388,
      "grad_norm": 0.7853806704100009,
      "learning_rate": 0.0001832466441960091,
      "loss": 12.9494,
      "step": 4834
    },
    {
      "epoch": 0.8454644808743169,
      "grad_norm": 0.8756708660513212,
      "learning_rate": 0.0001832387971616019,
      "loss": 13.0954,
      "step": 4835
    },
    {
      "epoch": 0.845639344262295,
      "grad_norm": 0.9045869577866741,
      "learning_rate": 0.00018323094845800123,
      "loss": 13.0261,
      "step": 4836
    },
    {
      "epoch": 0.8458142076502733,
      "grad_norm": 0.887502311742074,
      "learning_rate": 0.00018322309808536447,
      "loss": 12.8956,
      "step": 4837
    },
    {
      "epoch": 0.8459890710382514,
      "grad_norm": 0.8512436470823048,
      "learning_rate": 0.00018321524604384909,
      "loss": 12.9514,
      "step": 4838
    },
    {
      "epoch": 0.8461639344262295,
      "grad_norm": 0.8562006065778318,
      "learning_rate": 0.00018320739233361252,
      "loss": 13.13,
      "step": 4839
    },
    {
      "epoch": 0.8463387978142076,
      "grad_norm": 1.0427994476333722,
      "learning_rate": 0.00018319953695481226,
      "loss": 12.8454,
      "step": 4840
    },
    {
      "epoch": 0.8465136612021857,
      "grad_norm": 0.8764054344001772,
      "learning_rate": 0.0001831916799076058,
      "loss": 12.9187,
      "step": 4841
    },
    {
      "epoch": 0.846688524590164,
      "grad_norm": 0.8439323972627489,
      "learning_rate": 0.00018318382119215075,
      "loss": 12.8488,
      "step": 4842
    },
    {
      "epoch": 0.8468633879781421,
      "grad_norm": 0.8414664618705763,
      "learning_rate": 0.00018317596080860463,
      "loss": 12.7854,
      "step": 4843
    },
    {
      "epoch": 0.8470382513661202,
      "grad_norm": 1.031803427113204,
      "learning_rate": 0.00018316809875712516,
      "loss": 12.9489,
      "step": 4844
    },
    {
      "epoch": 0.8472131147540983,
      "grad_norm": 0.8403133842226045,
      "learning_rate": 0.00018316023503786997,
      "loss": 13.0723,
      "step": 4845
    },
    {
      "epoch": 0.8473879781420764,
      "grad_norm": 0.8337187595471343,
      "learning_rate": 0.0001831523696509967,
      "loss": 12.9148,
      "step": 4846
    },
    {
      "epoch": 0.8475628415300547,
      "grad_norm": 0.8839787090473783,
      "learning_rate": 0.00018314450259666312,
      "loss": 13.1021,
      "step": 4847
    },
    {
      "epoch": 0.8477377049180328,
      "grad_norm": 0.8926687629599822,
      "learning_rate": 0.00018313663387502696,
      "loss": 12.9795,
      "step": 4848
    },
    {
      "epoch": 0.8479125683060109,
      "grad_norm": 0.9883994706652799,
      "learning_rate": 0.00018312876348624603,
      "loss": 13.1887,
      "step": 4849
    },
    {
      "epoch": 0.848087431693989,
      "grad_norm": 0.8501210580081711,
      "learning_rate": 0.00018312089143047815,
      "loss": 13.1247,
      "step": 4850
    },
    {
      "epoch": 0.8482622950819673,
      "grad_norm": 0.9632353811022404,
      "learning_rate": 0.00018311301770788122,
      "loss": 13.019,
      "step": 4851
    },
    {
      "epoch": 0.8484371584699454,
      "grad_norm": 1.0253057646276116,
      "learning_rate": 0.00018310514231861303,
      "loss": 12.9738,
      "step": 4852
    },
    {
      "epoch": 0.8486120218579235,
      "grad_norm": 0.8490433436388339,
      "learning_rate": 0.0001830972652628316,
      "loss": 12.7115,
      "step": 4853
    },
    {
      "epoch": 0.8487868852459016,
      "grad_norm": 0.9941421230839238,
      "learning_rate": 0.00018308938654069487,
      "loss": 13.1582,
      "step": 4854
    },
    {
      "epoch": 0.8489617486338797,
      "grad_norm": 0.9244574668705767,
      "learning_rate": 0.00018308150615236083,
      "loss": 13.0257,
      "step": 4855
    },
    {
      "epoch": 0.849136612021858,
      "grad_norm": 1.1062681548472089,
      "learning_rate": 0.00018307362409798747,
      "loss": 13.1892,
      "step": 4856
    },
    {
      "epoch": 0.8493114754098361,
      "grad_norm": 0.9459903264110432,
      "learning_rate": 0.0001830657403777329,
      "loss": 13.0402,
      "step": 4857
    },
    {
      "epoch": 0.8494863387978142,
      "grad_norm": 0.9490637753216535,
      "learning_rate": 0.0001830578549917552,
      "loss": 12.9019,
      "step": 4858
    },
    {
      "epoch": 0.8496612021857923,
      "grad_norm": 0.9283826961302879,
      "learning_rate": 0.00018304996794021245,
      "loss": 12.9872,
      "step": 4859
    },
    {
      "epoch": 0.8498360655737704,
      "grad_norm": 1.063298618717862,
      "learning_rate": 0.00018304207922326286,
      "loss": 12.9246,
      "step": 4860
    },
    {
      "epoch": 0.8500109289617487,
      "grad_norm": 0.9300506012441135,
      "learning_rate": 0.00018303418884106464,
      "loss": 12.9272,
      "step": 4861
    },
    {
      "epoch": 0.8501857923497268,
      "grad_norm": 1.0527541523102435,
      "learning_rate": 0.00018302629679377592,
      "loss": 13.0543,
      "step": 4862
    },
    {
      "epoch": 0.8503606557377049,
      "grad_norm": 0.9730116393302632,
      "learning_rate": 0.00018301840308155507,
      "loss": 13.0297,
      "step": 4863
    },
    {
      "epoch": 0.850535519125683,
      "grad_norm": 0.8657613663001166,
      "learning_rate": 0.00018301050770456033,
      "loss": 13.0555,
      "step": 4864
    },
    {
      "epoch": 0.8507103825136612,
      "grad_norm": 0.8499703210280837,
      "learning_rate": 0.00018300261066295004,
      "loss": 12.8945,
      "step": 4865
    },
    {
      "epoch": 0.8508852459016394,
      "grad_norm": 0.8136383262597989,
      "learning_rate": 0.00018299471195688254,
      "loss": 12.9127,
      "step": 4866
    },
    {
      "epoch": 0.8510601092896175,
      "grad_norm": 0.8588322520251251,
      "learning_rate": 0.00018298681158651628,
      "loss": 12.8236,
      "step": 4867
    },
    {
      "epoch": 0.8512349726775956,
      "grad_norm": 0.8369422716820792,
      "learning_rate": 0.0001829789095520096,
      "loss": 13.1462,
      "step": 4868
    },
    {
      "epoch": 0.8514098360655737,
      "grad_norm": 0.9007593576448905,
      "learning_rate": 0.00018297100585352102,
      "loss": 13.0488,
      "step": 4869
    },
    {
      "epoch": 0.8515846994535519,
      "grad_norm": 0.8522981982885824,
      "learning_rate": 0.00018296310049120899,
      "loss": 12.7565,
      "step": 4870
    },
    {
      "epoch": 0.8517595628415301,
      "grad_norm": 0.8477754778338491,
      "learning_rate": 0.00018295519346523211,
      "loss": 13.1451,
      "step": 4871
    },
    {
      "epoch": 0.8519344262295082,
      "grad_norm": 0.7653802657128832,
      "learning_rate": 0.00018294728477574886,
      "loss": 13.0329,
      "step": 4872
    },
    {
      "epoch": 0.8521092896174863,
      "grad_norm": 0.9815102034815141,
      "learning_rate": 0.0001829393744229179,
      "loss": 13.0733,
      "step": 4873
    },
    {
      "epoch": 0.8522841530054645,
      "grad_norm": 0.8078331957594964,
      "learning_rate": 0.0001829314624068978,
      "loss": 12.9362,
      "step": 4874
    },
    {
      "epoch": 0.8524590163934426,
      "grad_norm": 0.855811071411528,
      "learning_rate": 0.00018292354872784727,
      "loss": 13.0504,
      "step": 4875
    },
    {
      "epoch": 0.8526338797814208,
      "grad_norm": 0.8184787437900317,
      "learning_rate": 0.00018291563338592495,
      "loss": 12.9252,
      "step": 4876
    },
    {
      "epoch": 0.8528087431693989,
      "grad_norm": 0.8121646314191275,
      "learning_rate": 0.0001829077163812896,
      "loss": 13.2327,
      "step": 4877
    },
    {
      "epoch": 0.852983606557377,
      "grad_norm": 0.8637765780157028,
      "learning_rate": 0.0001828997977141,
      "loss": 12.8038,
      "step": 4878
    },
    {
      "epoch": 0.8531584699453552,
      "grad_norm": 0.9423964965317431,
      "learning_rate": 0.00018289187738451492,
      "loss": 13.0624,
      "step": 4879
    },
    {
      "epoch": 0.8533333333333334,
      "grad_norm": 0.8276502234812027,
      "learning_rate": 0.00018288395539269318,
      "loss": 13.056,
      "step": 4880
    },
    {
      "epoch": 0.8535081967213115,
      "grad_norm": 1.0179655179130893,
      "learning_rate": 0.00018287603173879364,
      "loss": 13.0163,
      "step": 4881
    },
    {
      "epoch": 0.8536830601092896,
      "grad_norm": 0.8070562866193829,
      "learning_rate": 0.0001828681064229752,
      "loss": 12.8428,
      "step": 4882
    },
    {
      "epoch": 0.8538579234972677,
      "grad_norm": 0.9332799489250764,
      "learning_rate": 0.0001828601794453968,
      "loss": 13.0836,
      "step": 4883
    },
    {
      "epoch": 0.8540327868852459,
      "grad_norm": 0.7756845137631445,
      "learning_rate": 0.00018285225080621738,
      "loss": 12.9291,
      "step": 4884
    },
    {
      "epoch": 0.8542076502732241,
      "grad_norm": 0.9923463074466138,
      "learning_rate": 0.00018284432050559595,
      "loss": 13.1196,
      "step": 4885
    },
    {
      "epoch": 0.8543825136612022,
      "grad_norm": 0.7907623601973034,
      "learning_rate": 0.0001828363885436915,
      "loss": 12.9675,
      "step": 4886
    },
    {
      "epoch": 0.8545573770491803,
      "grad_norm": 0.9611218083171433,
      "learning_rate": 0.00018282845492066313,
      "loss": 13.1054,
      "step": 4887
    },
    {
      "epoch": 0.8547322404371585,
      "grad_norm": 0.8261911163150485,
      "learning_rate": 0.00018282051963666994,
      "loss": 13.0679,
      "step": 4888
    },
    {
      "epoch": 0.8549071038251366,
      "grad_norm": 0.8560580908141333,
      "learning_rate": 0.00018281258269187102,
      "loss": 12.7187,
      "step": 4889
    },
    {
      "epoch": 0.8550819672131148,
      "grad_norm": 0.8526227058264787,
      "learning_rate": 0.00018280464408642556,
      "loss": 12.9235,
      "step": 4890
    },
    {
      "epoch": 0.8552568306010929,
      "grad_norm": 0.8159860900672241,
      "learning_rate": 0.00018279670382049273,
      "loss": 12.9535,
      "step": 4891
    },
    {
      "epoch": 0.855431693989071,
      "grad_norm": 0.7975231607385823,
      "learning_rate": 0.00018278876189423179,
      "loss": 12.9563,
      "step": 4892
    },
    {
      "epoch": 0.8556065573770492,
      "grad_norm": 0.787797947368635,
      "learning_rate": 0.00018278081830780195,
      "loss": 12.9988,
      "step": 4893
    },
    {
      "epoch": 0.8557814207650273,
      "grad_norm": 0.8923700516594165,
      "learning_rate": 0.00018277287306136256,
      "loss": 13.2136,
      "step": 4894
    },
    {
      "epoch": 0.8559562841530055,
      "grad_norm": 0.8758111560728867,
      "learning_rate": 0.00018276492615507294,
      "loss": 12.9616,
      "step": 4895
    },
    {
      "epoch": 0.8561311475409836,
      "grad_norm": 0.8536063044399287,
      "learning_rate": 0.00018275697758909237,
      "loss": 13.0594,
      "step": 4896
    },
    {
      "epoch": 0.8563060109289617,
      "grad_norm": 0.8855376039218679,
      "learning_rate": 0.00018274902736358035,
      "loss": 12.8765,
      "step": 4897
    },
    {
      "epoch": 0.8564808743169399,
      "grad_norm": 0.7954852575547106,
      "learning_rate": 0.00018274107547869623,
      "loss": 12.8,
      "step": 4898
    },
    {
      "epoch": 0.856655737704918,
      "grad_norm": 0.8610215290269846,
      "learning_rate": 0.00018273312193459952,
      "loss": 13.2085,
      "step": 4899
    },
    {
      "epoch": 0.8568306010928962,
      "grad_norm": 0.8856776722877262,
      "learning_rate": 0.00018272516673144965,
      "loss": 12.772,
      "step": 4900
    },
    {
      "epoch": 0.8570054644808743,
      "grad_norm": 0.782643763852774,
      "learning_rate": 0.00018271720986940624,
      "loss": 13.022,
      "step": 4901
    },
    {
      "epoch": 0.8571803278688525,
      "grad_norm": 0.9031467574149458,
      "learning_rate": 0.00018270925134862876,
      "loss": 12.7603,
      "step": 4902
    },
    {
      "epoch": 0.8573551912568306,
      "grad_norm": 0.8666830255666823,
      "learning_rate": 0.00018270129116927685,
      "loss": 12.6505,
      "step": 4903
    },
    {
      "epoch": 0.8575300546448087,
      "grad_norm": 0.8687041453668108,
      "learning_rate": 0.00018269332933151014,
      "loss": 12.9948,
      "step": 4904
    },
    {
      "epoch": 0.8577049180327869,
      "grad_norm": 0.8272715331438835,
      "learning_rate": 0.00018268536583548826,
      "loss": 13.0569,
      "step": 4905
    },
    {
      "epoch": 0.857879781420765,
      "grad_norm": 0.8783713358847662,
      "learning_rate": 0.0001826774006813709,
      "loss": 13.0594,
      "step": 4906
    },
    {
      "epoch": 0.8580546448087432,
      "grad_norm": 0.7950529835379627,
      "learning_rate": 0.00018266943386931783,
      "loss": 13.1841,
      "step": 4907
    },
    {
      "epoch": 0.8582295081967213,
      "grad_norm": 0.9854934358705887,
      "learning_rate": 0.00018266146539948878,
      "loss": 12.9938,
      "step": 4908
    },
    {
      "epoch": 0.8584043715846994,
      "grad_norm": 0.950926986198173,
      "learning_rate": 0.00018265349527204355,
      "loss": 12.7197,
      "step": 4909
    },
    {
      "epoch": 0.8585792349726776,
      "grad_norm": 0.97771062256112,
      "learning_rate": 0.00018264552348714195,
      "loss": 12.9993,
      "step": 4910
    },
    {
      "epoch": 0.8587540983606557,
      "grad_norm": 0.9196531802386714,
      "learning_rate": 0.00018263755004494384,
      "loss": 12.9087,
      "step": 4911
    },
    {
      "epoch": 0.8589289617486339,
      "grad_norm": 0.8293343493733731,
      "learning_rate": 0.00018262957494560915,
      "loss": 12.9815,
      "step": 4912
    },
    {
      "epoch": 0.859103825136612,
      "grad_norm": 0.9001203063718254,
      "learning_rate": 0.00018262159818929774,
      "loss": 13.1025,
      "step": 4913
    },
    {
      "epoch": 0.8592786885245902,
      "grad_norm": 0.7720329506376667,
      "learning_rate": 0.00018261361977616963,
      "loss": 12.7255,
      "step": 4914
    },
    {
      "epoch": 0.8594535519125683,
      "grad_norm": 0.8783199428948903,
      "learning_rate": 0.00018260563970638478,
      "loss": 12.7111,
      "step": 4915
    },
    {
      "epoch": 0.8596284153005465,
      "grad_norm": 0.8325891942065007,
      "learning_rate": 0.0001825976579801032,
      "loss": 12.8026,
      "step": 4916
    },
    {
      "epoch": 0.8598032786885246,
      "grad_norm": 0.8639301237854797,
      "learning_rate": 0.000182589674597485,
      "loss": 12.9182,
      "step": 4917
    },
    {
      "epoch": 0.8599781420765027,
      "grad_norm": 0.8064528859064158,
      "learning_rate": 0.00018258168955869022,
      "loss": 12.652,
      "step": 4918
    },
    {
      "epoch": 0.8601530054644809,
      "grad_norm": 0.8018820948074616,
      "learning_rate": 0.00018257370286387903,
      "loss": 12.9276,
      "step": 4919
    },
    {
      "epoch": 0.860327868852459,
      "grad_norm": 0.8799585873364891,
      "learning_rate": 0.00018256571451321155,
      "loss": 12.8317,
      "step": 4920
    },
    {
      "epoch": 0.8605027322404372,
      "grad_norm": 0.8676157034184109,
      "learning_rate": 0.00018255772450684798,
      "loss": 12.756,
      "step": 4921
    },
    {
      "epoch": 0.8606775956284153,
      "grad_norm": 0.8269993944256732,
      "learning_rate": 0.00018254973284494855,
      "loss": 12.8312,
      "step": 4922
    },
    {
      "epoch": 0.8608524590163934,
      "grad_norm": 0.841954034151068,
      "learning_rate": 0.00018254173952767354,
      "loss": 13.0368,
      "step": 4923
    },
    {
      "epoch": 0.8610273224043716,
      "grad_norm": 0.8008502996127622,
      "learning_rate": 0.00018253374455518318,
      "loss": 12.8753,
      "step": 4924
    },
    {
      "epoch": 0.8612021857923498,
      "grad_norm": 0.7861718768539425,
      "learning_rate": 0.00018252574792763782,
      "loss": 12.7404,
      "step": 4925
    },
    {
      "epoch": 0.8613770491803279,
      "grad_norm": 0.9037093393778752,
      "learning_rate": 0.00018251774964519785,
      "loss": 12.8717,
      "step": 4926
    },
    {
      "epoch": 0.861551912568306,
      "grad_norm": 0.7983007856161819,
      "learning_rate": 0.00018250974970802366,
      "loss": 12.9795,
      "step": 4927
    },
    {
      "epoch": 0.8617267759562841,
      "grad_norm": 0.8566710801720497,
      "learning_rate": 0.00018250174811627561,
      "loss": 12.8813,
      "step": 4928
    },
    {
      "epoch": 0.8619016393442623,
      "grad_norm": 0.7948612187267203,
      "learning_rate": 0.00018249374487011423,
      "loss": 12.8718,
      "step": 4929
    },
    {
      "epoch": 0.8620765027322405,
      "grad_norm": 0.880069224818431,
      "learning_rate": 0.00018248573996969993,
      "loss": 12.8997,
      "step": 4930
    },
    {
      "epoch": 0.8622513661202186,
      "grad_norm": 0.8261622402961216,
      "learning_rate": 0.00018247773341519333,
      "loss": 12.7229,
      "step": 4931
    },
    {
      "epoch": 0.8624262295081967,
      "grad_norm": 0.9338861339236916,
      "learning_rate": 0.00018246972520675488,
      "loss": 13.0799,
      "step": 4932
    },
    {
      "epoch": 0.8626010928961748,
      "grad_norm": 0.8155159061323668,
      "learning_rate": 0.0001824617153445453,
      "loss": 12.9211,
      "step": 4933
    },
    {
      "epoch": 0.862775956284153,
      "grad_norm": 0.7522035790556908,
      "learning_rate": 0.00018245370382872507,
      "loss": 12.9086,
      "step": 4934
    },
    {
      "epoch": 0.8629508196721312,
      "grad_norm": 0.8869624600974336,
      "learning_rate": 0.00018244569065945494,
      "loss": 12.8467,
      "step": 4935
    },
    {
      "epoch": 0.8631256830601093,
      "grad_norm": 0.878336392014373,
      "learning_rate": 0.00018243767583689556,
      "loss": 13.0147,
      "step": 4936
    },
    {
      "epoch": 0.8633005464480874,
      "grad_norm": 0.9277332183339475,
      "learning_rate": 0.00018242965936120768,
      "loss": 13.1505,
      "step": 4937
    },
    {
      "epoch": 0.8634754098360655,
      "grad_norm": 0.8508122908850747,
      "learning_rate": 0.00018242164123255202,
      "loss": 12.8016,
      "step": 4938
    },
    {
      "epoch": 0.8636502732240438,
      "grad_norm": 0.9872288745524671,
      "learning_rate": 0.00018241362145108938,
      "loss": 12.9107,
      "step": 4939
    },
    {
      "epoch": 0.8638251366120219,
      "grad_norm": 0.9569312464116585,
      "learning_rate": 0.0001824056000169806,
      "loss": 13.0658,
      "step": 4940
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.7707733618421386,
      "learning_rate": 0.00018239757693038652,
      "loss": 13.0954,
      "step": 4941
    },
    {
      "epoch": 0.8641748633879781,
      "grad_norm": 0.9595794063454106,
      "learning_rate": 0.00018238955219146804,
      "loss": 12.9384,
      "step": 4942
    },
    {
      "epoch": 0.8643497267759563,
      "grad_norm": 0.8915682020859146,
      "learning_rate": 0.00018238152580038605,
      "loss": 12.9446,
      "step": 4943
    },
    {
      "epoch": 0.8645245901639345,
      "grad_norm": 0.8353903465916706,
      "learning_rate": 0.00018237349775730152,
      "loss": 12.9557,
      "step": 4944
    },
    {
      "epoch": 0.8646994535519126,
      "grad_norm": 0.8359820757085688,
      "learning_rate": 0.00018236546806237546,
      "loss": 12.7232,
      "step": 4945
    },
    {
      "epoch": 0.8648743169398907,
      "grad_norm": 0.8354937592715883,
      "learning_rate": 0.00018235743671576887,
      "loss": 13.0434,
      "step": 4946
    },
    {
      "epoch": 0.8650491803278688,
      "grad_norm": 0.7604594287689994,
      "learning_rate": 0.00018234940371764278,
      "loss": 12.8319,
      "step": 4947
    },
    {
      "epoch": 0.865224043715847,
      "grad_norm": 0.8667774398205197,
      "learning_rate": 0.0001823413690681583,
      "loss": 12.9892,
      "step": 4948
    },
    {
      "epoch": 0.8653989071038252,
      "grad_norm": 0.9483470458235085,
      "learning_rate": 0.0001823333327674766,
      "loss": 13.0555,
      "step": 4949
    },
    {
      "epoch": 0.8655737704918033,
      "grad_norm": 0.8941624633797852,
      "learning_rate": 0.00018232529481575872,
      "loss": 12.9468,
      "step": 4950
    },
    {
      "epoch": 0.8657486338797814,
      "grad_norm": 0.9569014659266046,
      "learning_rate": 0.00018231725521316596,
      "loss": 12.9623,
      "step": 4951
    },
    {
      "epoch": 0.8659234972677595,
      "grad_norm": 0.8103517341633838,
      "learning_rate": 0.00018230921395985945,
      "loss": 12.7674,
      "step": 4952
    },
    {
      "epoch": 0.8660983606557378,
      "grad_norm": 0.778616907755513,
      "learning_rate": 0.00018230117105600047,
      "loss": 12.4862,
      "step": 4953
    },
    {
      "epoch": 0.8662732240437159,
      "grad_norm": 0.8271702489585694,
      "learning_rate": 0.00018229312650175036,
      "loss": 12.9936,
      "step": 4954
    },
    {
      "epoch": 0.866448087431694,
      "grad_norm": 0.8696740562900135,
      "learning_rate": 0.00018228508029727035,
      "loss": 12.9813,
      "step": 4955
    },
    {
      "epoch": 0.8666229508196721,
      "grad_norm": 0.874354552316342,
      "learning_rate": 0.00018227703244272182,
      "loss": 12.856,
      "step": 4956
    },
    {
      "epoch": 0.8667978142076502,
      "grad_norm": 0.7871347877625051,
      "learning_rate": 0.0001822689829382662,
      "loss": 12.9272,
      "step": 4957
    },
    {
      "epoch": 0.8669726775956285,
      "grad_norm": 0.9563718529410224,
      "learning_rate": 0.00018226093178406484,
      "loss": 13.0955,
      "step": 4958
    },
    {
      "epoch": 0.8671475409836066,
      "grad_norm": 0.7526368909994993,
      "learning_rate": 0.00018225287898027926,
      "loss": 12.8968,
      "step": 4959
    },
    {
      "epoch": 0.8673224043715847,
      "grad_norm": 0.8669492936441474,
      "learning_rate": 0.00018224482452707089,
      "loss": 13.0758,
      "step": 4960
    },
    {
      "epoch": 0.8674972677595628,
      "grad_norm": 0.7640029407601656,
      "learning_rate": 0.00018223676842460127,
      "loss": 12.8696,
      "step": 4961
    },
    {
      "epoch": 0.8676721311475409,
      "grad_norm": 0.9719486605966562,
      "learning_rate": 0.00018222871067303192,
      "loss": 13.0323,
      "step": 4962
    },
    {
      "epoch": 0.8678469945355192,
      "grad_norm": 0.9267034503422648,
      "learning_rate": 0.00018222065127252446,
      "loss": 12.963,
      "step": 4963
    },
    {
      "epoch": 0.8680218579234973,
      "grad_norm": 0.8441225924162612,
      "learning_rate": 0.00018221259022324052,
      "loss": 12.9948,
      "step": 4964
    },
    {
      "epoch": 0.8681967213114754,
      "grad_norm": 1.0669925037132282,
      "learning_rate": 0.00018220452752534168,
      "loss": 12.8218,
      "step": 4965
    },
    {
      "epoch": 0.8683715846994535,
      "grad_norm": 0.8573899701399403,
      "learning_rate": 0.00018219646317898966,
      "loss": 12.9167,
      "step": 4966
    },
    {
      "epoch": 0.8685464480874316,
      "grad_norm": 1.1606845994377728,
      "learning_rate": 0.0001821883971843462,
      "loss": 13.0006,
      "step": 4967
    },
    {
      "epoch": 0.8687213114754099,
      "grad_norm": 0.9810286433453311,
      "learning_rate": 0.00018218032954157303,
      "loss": 12.7856,
      "step": 4968
    },
    {
      "epoch": 0.868896174863388,
      "grad_norm": 0.9401181085057273,
      "learning_rate": 0.00018217226025083193,
      "loss": 12.7428,
      "step": 4969
    },
    {
      "epoch": 0.8690710382513661,
      "grad_norm": 0.857207520248114,
      "learning_rate": 0.00018216418931228467,
      "loss": 12.8589,
      "step": 4970
    },
    {
      "epoch": 0.8692459016393442,
      "grad_norm": 0.8600274663401442,
      "learning_rate": 0.00018215611672609317,
      "loss": 12.9694,
      "step": 4971
    },
    {
      "epoch": 0.8694207650273224,
      "grad_norm": 0.8524719273127117,
      "learning_rate": 0.00018214804249241926,
      "loss": 12.9426,
      "step": 4972
    },
    {
      "epoch": 0.8695956284153006,
      "grad_norm": 0.8182720107804757,
      "learning_rate": 0.00018213996661142488,
      "loss": 12.734,
      "step": 4973
    },
    {
      "epoch": 0.8697704918032787,
      "grad_norm": 0.8981503915567619,
      "learning_rate": 0.00018213188908327194,
      "loss": 13.0135,
      "step": 4974
    },
    {
      "epoch": 0.8699453551912568,
      "grad_norm": 0.8841399436864892,
      "learning_rate": 0.00018212380990812247,
      "loss": 12.9887,
      "step": 4975
    },
    {
      "epoch": 0.8701202185792349,
      "grad_norm": 0.7769964476322018,
      "learning_rate": 0.00018211572908613847,
      "loss": 12.6587,
      "step": 4976
    },
    {
      "epoch": 0.8702950819672132,
      "grad_norm": 0.9250827117117301,
      "learning_rate": 0.00018210764661748198,
      "loss": 12.8857,
      "step": 4977
    },
    {
      "epoch": 0.8704699453551913,
      "grad_norm": 0.8830600475647242,
      "learning_rate": 0.00018209956250231506,
      "loss": 13.0222,
      "step": 4978
    },
    {
      "epoch": 0.8706448087431694,
      "grad_norm": 0.9183282777655944,
      "learning_rate": 0.00018209147674079983,
      "loss": 13.1805,
      "step": 4979
    },
    {
      "epoch": 0.8708196721311475,
      "grad_norm": 0.8109670387054206,
      "learning_rate": 0.00018208338933309843,
      "loss": 12.9667,
      "step": 4980
    },
    {
      "epoch": 0.8709945355191256,
      "grad_norm": 0.8782144016027559,
      "learning_rate": 0.00018207530027937303,
      "loss": 13.153,
      "step": 4981
    },
    {
      "epoch": 0.8711693989071039,
      "grad_norm": 0.814197738414931,
      "learning_rate": 0.00018206720957978588,
      "loss": 12.85,
      "step": 4982
    },
    {
      "epoch": 0.871344262295082,
      "grad_norm": 0.8003727253401668,
      "learning_rate": 0.0001820591172344992,
      "loss": 12.9224,
      "step": 4983
    },
    {
      "epoch": 0.8715191256830601,
      "grad_norm": 0.7377638327347836,
      "learning_rate": 0.00018205102324367524,
      "loss": 12.7895,
      "step": 4984
    },
    {
      "epoch": 0.8716939890710382,
      "grad_norm": 0.806576183845769,
      "learning_rate": 0.00018204292760747633,
      "loss": 13.3108,
      "step": 4985
    },
    {
      "epoch": 0.8718688524590164,
      "grad_norm": 0.7953648673445098,
      "learning_rate": 0.00018203483032606488,
      "loss": 12.7826,
      "step": 4986
    },
    {
      "epoch": 0.8720437158469946,
      "grad_norm": 0.8527741927021802,
      "learning_rate": 0.00018202673139960316,
      "loss": 13.2312,
      "step": 4987
    },
    {
      "epoch": 0.8722185792349727,
      "grad_norm": 0.807886341459609,
      "learning_rate": 0.0001820186308282536,
      "loss": 12.8414,
      "step": 4988
    },
    {
      "epoch": 0.8723934426229508,
      "grad_norm": 0.8177182697785293,
      "learning_rate": 0.0001820105286121787,
      "loss": 12.9049,
      "step": 4989
    },
    {
      "epoch": 0.8725683060109289,
      "grad_norm": 0.771866150447619,
      "learning_rate": 0.0001820024247515409,
      "loss": 12.9637,
      "step": 4990
    },
    {
      "epoch": 0.8727431693989071,
      "grad_norm": 0.9144372316788378,
      "learning_rate": 0.00018199431924650265,
      "loss": 12.9444,
      "step": 4991
    },
    {
      "epoch": 0.8729180327868853,
      "grad_norm": 0.9340302624701333,
      "learning_rate": 0.00018198621209722658,
      "loss": 12.876,
      "step": 4992
    },
    {
      "epoch": 0.8730928961748634,
      "grad_norm": 0.9211265594412189,
      "learning_rate": 0.0001819781033038752,
      "loss": 12.7659,
      "step": 4993
    },
    {
      "epoch": 0.8732677595628415,
      "grad_norm": 1.0049528132066106,
      "learning_rate": 0.00018196999286661117,
      "loss": 13.0151,
      "step": 4994
    },
    {
      "epoch": 0.8734426229508196,
      "grad_norm": 0.8883093218208795,
      "learning_rate": 0.00018196188078559711,
      "loss": 13.1908,
      "step": 4995
    },
    {
      "epoch": 0.8736174863387978,
      "grad_norm": 1.0033489081225417,
      "learning_rate": 0.00018195376706099566,
      "loss": 12.9372,
      "step": 4996
    },
    {
      "epoch": 0.873792349726776,
      "grad_norm": 0.811260965143563,
      "learning_rate": 0.0001819456516929696,
      "loss": 13.0532,
      "step": 4997
    },
    {
      "epoch": 0.8739672131147541,
      "grad_norm": 0.9858669259757534,
      "learning_rate": 0.00018193753468168154,
      "loss": 12.8307,
      "step": 4998
    },
    {
      "epoch": 0.8741420765027322,
      "grad_norm": 0.9709574074745613,
      "learning_rate": 0.00018192941602729437,
      "loss": 12.9116,
      "step": 4999
    },
    {
      "epoch": 0.8743169398907104,
      "grad_norm": 0.9244387671536272,
      "learning_rate": 0.00018192129572997085,
      "loss": 12.8268,
      "step": 5000
    },
    {
      "epoch": 0.8744918032786885,
      "grad_norm": 0.9036445127441105,
      "learning_rate": 0.0001819131737898738,
      "loss": 12.9172,
      "step": 5001
    },
    {
      "epoch": 0.8746666666666667,
      "grad_norm": 0.8662892642170015,
      "learning_rate": 0.00018190505020716614,
      "loss": 12.9175,
      "step": 5002
    },
    {
      "epoch": 0.8748415300546448,
      "grad_norm": 0.9379137991603183,
      "learning_rate": 0.00018189692498201074,
      "loss": 12.9685,
      "step": 5003
    },
    {
      "epoch": 0.8750163934426229,
      "grad_norm": 0.7873525243424977,
      "learning_rate": 0.00018188879811457053,
      "loss": 12.905,
      "step": 5004
    },
    {
      "epoch": 0.8751912568306011,
      "grad_norm": 0.983062721024225,
      "learning_rate": 0.00018188066960500847,
      "loss": 12.9156,
      "step": 5005
    },
    {
      "epoch": 0.8753661202185792,
      "grad_norm": 0.7415261438661287,
      "learning_rate": 0.0001818725394534876,
      "loss": 12.8023,
      "step": 5006
    },
    {
      "epoch": 0.8755409836065574,
      "grad_norm": 1.0367050072947914,
      "learning_rate": 0.0001818644076601709,
      "loss": 12.8978,
      "step": 5007
    },
    {
      "epoch": 0.8757158469945355,
      "grad_norm": 0.8341134612119188,
      "learning_rate": 0.00018185627422522148,
      "loss": 12.9004,
      "step": 5008
    },
    {
      "epoch": 0.8758907103825136,
      "grad_norm": 0.860485962091035,
      "learning_rate": 0.00018184813914880245,
      "loss": 12.9613,
      "step": 5009
    },
    {
      "epoch": 0.8760655737704918,
      "grad_norm": 0.8499460583006558,
      "learning_rate": 0.00018184000243107694,
      "loss": 12.8047,
      "step": 5010
    },
    {
      "epoch": 0.87624043715847,
      "grad_norm": 0.8260903434241186,
      "learning_rate": 0.00018183186407220807,
      "loss": 12.83,
      "step": 5011
    },
    {
      "epoch": 0.8764153005464481,
      "grad_norm": 0.9512242788853243,
      "learning_rate": 0.00018182372407235907,
      "loss": 12.9908,
      "step": 5012
    },
    {
      "epoch": 0.8765901639344262,
      "grad_norm": 0.8200018153446678,
      "learning_rate": 0.00018181558243169317,
      "loss": 12.9577,
      "step": 5013
    },
    {
      "epoch": 0.8767650273224044,
      "grad_norm": 0.8882318759900254,
      "learning_rate": 0.00018180743915037365,
      "loss": 12.8347,
      "step": 5014
    },
    {
      "epoch": 0.8769398907103825,
      "grad_norm": 0.8203396928561446,
      "learning_rate": 0.00018179929422856378,
      "loss": 13.0119,
      "step": 5015
    },
    {
      "epoch": 0.8771147540983607,
      "grad_norm": 0.8372176510255713,
      "learning_rate": 0.0001817911476664269,
      "loss": 12.9167,
      "step": 5016
    },
    {
      "epoch": 0.8772896174863388,
      "grad_norm": 0.8215182343236468,
      "learning_rate": 0.00018178299946412639,
      "loss": 12.9436,
      "step": 5017
    },
    {
      "epoch": 0.8774644808743169,
      "grad_norm": 0.8258456180875542,
      "learning_rate": 0.00018177484962182564,
      "loss": 12.8947,
      "step": 5018
    },
    {
      "epoch": 0.8776393442622951,
      "grad_norm": 0.9275980606029592,
      "learning_rate": 0.00018176669813968802,
      "loss": 12.895,
      "step": 5019
    },
    {
      "epoch": 0.8778142076502732,
      "grad_norm": 0.7920004408265243,
      "learning_rate": 0.0001817585450178771,
      "loss": 13.0314,
      "step": 5020
    },
    {
      "epoch": 0.8779890710382514,
      "grad_norm": 0.8755041729493717,
      "learning_rate": 0.00018175039025655632,
      "loss": 13.0279,
      "step": 5021
    },
    {
      "epoch": 0.8781639344262295,
      "grad_norm": 0.8614165071158185,
      "learning_rate": 0.00018174223385588917,
      "loss": 13.0226,
      "step": 5022
    },
    {
      "epoch": 0.8783387978142077,
      "grad_norm": 0.8203539532875869,
      "learning_rate": 0.00018173407581603928,
      "loss": 13.0617,
      "step": 5023
    },
    {
      "epoch": 0.8785136612021858,
      "grad_norm": 0.7608834993828517,
      "learning_rate": 0.00018172591613717023,
      "loss": 12.6974,
      "step": 5024
    },
    {
      "epoch": 0.8786885245901639,
      "grad_norm": 0.8859495890693035,
      "learning_rate": 0.00018171775481944563,
      "loss": 12.973,
      "step": 5025
    },
    {
      "epoch": 0.8788633879781421,
      "grad_norm": 0.8323672940315546,
      "learning_rate": 0.0001817095918630291,
      "loss": 12.897,
      "step": 5026
    },
    {
      "epoch": 0.8790382513661202,
      "grad_norm": 0.9133201163335395,
      "learning_rate": 0.00018170142726808437,
      "loss": 12.8761,
      "step": 5027
    },
    {
      "epoch": 0.8792131147540984,
      "grad_norm": 0.8624497976873102,
      "learning_rate": 0.0001816932610347752,
      "loss": 12.7963,
      "step": 5028
    },
    {
      "epoch": 0.8793879781420765,
      "grad_norm": 0.8641775326010481,
      "learning_rate": 0.0001816850931632653,
      "loss": 12.8019,
      "step": 5029
    },
    {
      "epoch": 0.8795628415300546,
      "grad_norm": 0.858138072377368,
      "learning_rate": 0.00018167692365371848,
      "loss": 13.1872,
      "step": 5030
    },
    {
      "epoch": 0.8797377049180328,
      "grad_norm": 0.8149450604058743,
      "learning_rate": 0.00018166875250629855,
      "loss": 13.0099,
      "step": 5031
    },
    {
      "epoch": 0.879912568306011,
      "grad_norm": 0.8335531023309888,
      "learning_rate": 0.0001816605797211694,
      "loss": 12.6851,
      "step": 5032
    },
    {
      "epoch": 0.8800874316939891,
      "grad_norm": 0.8910066228768578,
      "learning_rate": 0.00018165240529849488,
      "loss": 12.8911,
      "step": 5033
    },
    {
      "epoch": 0.8802622950819672,
      "grad_norm": 0.8644255580274802,
      "learning_rate": 0.00018164422923843893,
      "loss": 13.0717,
      "step": 5034
    },
    {
      "epoch": 0.8804371584699453,
      "grad_norm": 0.8410589925554137,
      "learning_rate": 0.00018163605154116548,
      "loss": 12.943,
      "step": 5035
    },
    {
      "epoch": 0.8806120218579235,
      "grad_norm": 0.8363465171352145,
      "learning_rate": 0.00018162787220683856,
      "loss": 12.7086,
      "step": 5036
    },
    {
      "epoch": 0.8807868852459017,
      "grad_norm": 0.8529661673760514,
      "learning_rate": 0.0001816196912356222,
      "loss": 12.9573,
      "step": 5037
    },
    {
      "epoch": 0.8809617486338798,
      "grad_norm": 0.8034317072790451,
      "learning_rate": 0.00018161150862768037,
      "loss": 13.045,
      "step": 5038
    },
    {
      "epoch": 0.8811366120218579,
      "grad_norm": 0.7491101133399273,
      "learning_rate": 0.00018160332438317724,
      "loss": 12.7182,
      "step": 5039
    },
    {
      "epoch": 0.8813114754098361,
      "grad_norm": 0.7591700998897302,
      "learning_rate": 0.0001815951385022769,
      "loss": 12.8977,
      "step": 5040
    },
    {
      "epoch": 0.8814863387978142,
      "grad_norm": 0.7550239940287613,
      "learning_rate": 0.00018158695098514352,
      "loss": 12.8195,
      "step": 5041
    },
    {
      "epoch": 0.8816612021857924,
      "grad_norm": 0.7697621264312651,
      "learning_rate": 0.00018157876183194126,
      "loss": 12.9813,
      "step": 5042
    },
    {
      "epoch": 0.8818360655737705,
      "grad_norm": 0.8284044045076739,
      "learning_rate": 0.00018157057104283431,
      "loss": 12.9083,
      "step": 5043
    },
    {
      "epoch": 0.8820109289617486,
      "grad_norm": 0.7852090593642284,
      "learning_rate": 0.000181562378617987,
      "loss": 12.76,
      "step": 5044
    },
    {
      "epoch": 0.8821857923497268,
      "grad_norm": 0.7633480281053234,
      "learning_rate": 0.00018155418455756354,
      "loss": 12.8617,
      "step": 5045
    },
    {
      "epoch": 0.882360655737705,
      "grad_norm": 0.9003930989743453,
      "learning_rate": 0.00018154598886172828,
      "loss": 12.9303,
      "step": 5046
    },
    {
      "epoch": 0.8825355191256831,
      "grad_norm": 0.89626535855259,
      "learning_rate": 0.00018153779153064559,
      "loss": 12.771,
      "step": 5047
    },
    {
      "epoch": 0.8827103825136612,
      "grad_norm": 0.8042042713710291,
      "learning_rate": 0.00018152959256447981,
      "loss": 12.9947,
      "step": 5048
    },
    {
      "epoch": 0.8828852459016393,
      "grad_norm": 0.9881733275745935,
      "learning_rate": 0.00018152139196339536,
      "loss": 12.6016,
      "step": 5049
    },
    {
      "epoch": 0.8830601092896175,
      "grad_norm": 0.8452500943205921,
      "learning_rate": 0.0001815131897275567,
      "loss": 13.0487,
      "step": 5050
    },
    {
      "epoch": 0.8832349726775957,
      "grad_norm": 0.9025021614734147,
      "learning_rate": 0.00018150498585712833,
      "loss": 13.0897,
      "step": 5051
    },
    {
      "epoch": 0.8834098360655738,
      "grad_norm": 0.8980639793683033,
      "learning_rate": 0.00018149678035227473,
      "loss": 13.1978,
      "step": 5052
    },
    {
      "epoch": 0.8835846994535519,
      "grad_norm": 0.8223185372186664,
      "learning_rate": 0.00018148857321316044,
      "loss": 12.8315,
      "step": 5053
    },
    {
      "epoch": 0.88375956284153,
      "grad_norm": 1.051713967857811,
      "learning_rate": 0.00018148036443995006,
      "loss": 13.0308,
      "step": 5054
    },
    {
      "epoch": 0.8839344262295082,
      "grad_norm": 0.7998081825353253,
      "learning_rate": 0.0001814721540328082,
      "loss": 12.8295,
      "step": 5055
    },
    {
      "epoch": 0.8841092896174864,
      "grad_norm": 1.0181655636886169,
      "learning_rate": 0.00018146394199189948,
      "loss": 13.2191,
      "step": 5056
    },
    {
      "epoch": 0.8842841530054645,
      "grad_norm": 0.925915309591187,
      "learning_rate": 0.00018145572831738865,
      "loss": 12.9316,
      "step": 5057
    },
    {
      "epoch": 0.8844590163934426,
      "grad_norm": 0.8832635832576193,
      "learning_rate": 0.00018144751300944032,
      "loss": 12.8776,
      "step": 5058
    },
    {
      "epoch": 0.8846338797814207,
      "grad_norm": 1.0687229420487068,
      "learning_rate": 0.00018143929606821927,
      "loss": 13.0781,
      "step": 5059
    },
    {
      "epoch": 0.884808743169399,
      "grad_norm": 0.8444604093410542,
      "learning_rate": 0.0001814310774938903,
      "loss": 12.6955,
      "step": 5060
    },
    {
      "epoch": 0.8849836065573771,
      "grad_norm": 1.0774113139251527,
      "learning_rate": 0.0001814228572866182,
      "loss": 12.7255,
      "step": 5061
    },
    {
      "epoch": 0.8851584699453552,
      "grad_norm": 0.9005416566249096,
      "learning_rate": 0.00018141463544656784,
      "loss": 13.1436,
      "step": 5062
    },
    {
      "epoch": 0.8853333333333333,
      "grad_norm": 0.9302885426289832,
      "learning_rate": 0.00018140641197390402,
      "loss": 13.0543,
      "step": 5063
    },
    {
      "epoch": 0.8855081967213114,
      "grad_norm": 0.9520799679033946,
      "learning_rate": 0.00018139818686879172,
      "loss": 13.0195,
      "step": 5064
    },
    {
      "epoch": 0.8856830601092897,
      "grad_norm": 0.8299301163695484,
      "learning_rate": 0.0001813899601313958,
      "loss": 13.0156,
      "step": 5065
    },
    {
      "epoch": 0.8858579234972678,
      "grad_norm": 0.8934245172773586,
      "learning_rate": 0.00018138173176188133,
      "loss": 12.9738,
      "step": 5066
    },
    {
      "epoch": 0.8860327868852459,
      "grad_norm": 0.7607375166837687,
      "learning_rate": 0.00018137350176041322,
      "loss": 12.9623,
      "step": 5067
    },
    {
      "epoch": 0.886207650273224,
      "grad_norm": 1.1596950267717132,
      "learning_rate": 0.0001813652701271566,
      "loss": 12.9612,
      "step": 5068
    },
    {
      "epoch": 0.8863825136612021,
      "grad_norm": 0.8649663784160138,
      "learning_rate": 0.0001813570368622765,
      "loss": 12.8491,
      "step": 5069
    },
    {
      "epoch": 0.8865573770491804,
      "grad_norm": 0.8304703809992873,
      "learning_rate": 0.00018134880196593795,
      "loss": 13.0199,
      "step": 5070
    },
    {
      "epoch": 0.8867322404371585,
      "grad_norm": 0.8428782170516026,
      "learning_rate": 0.00018134056543830614,
      "loss": 12.7093,
      "step": 5071
    },
    {
      "epoch": 0.8869071038251366,
      "grad_norm": 0.9116216475213433,
      "learning_rate": 0.0001813323272795463,
      "loss": 13.0403,
      "step": 5072
    },
    {
      "epoch": 0.8870819672131147,
      "grad_norm": 0.9263770357477399,
      "learning_rate": 0.00018132408748982356,
      "loss": 13.1311,
      "step": 5073
    },
    {
      "epoch": 0.887256830601093,
      "grad_norm": 0.942586066926971,
      "learning_rate": 0.00018131584606930317,
      "loss": 12.6654,
      "step": 5074
    },
    {
      "epoch": 0.8874316939890711,
      "grad_norm": 0.8514058087107222,
      "learning_rate": 0.00018130760301815038,
      "loss": 12.524,
      "step": 5075
    },
    {
      "epoch": 0.8876065573770492,
      "grad_norm": 1.0912454857931853,
      "learning_rate": 0.0001812993583365305,
      "loss": 12.982,
      "step": 5076
    },
    {
      "epoch": 0.8877814207650273,
      "grad_norm": 0.8954183172516741,
      "learning_rate": 0.00018129111202460886,
      "loss": 13.0291,
      "step": 5077
    },
    {
      "epoch": 0.8879562841530054,
      "grad_norm": 1.0014073803761385,
      "learning_rate": 0.00018128286408255085,
      "loss": 13.0266,
      "step": 5078
    },
    {
      "epoch": 0.8881311475409837,
      "grad_norm": 0.8554614208646585,
      "learning_rate": 0.00018127461451052183,
      "loss": 12.8669,
      "step": 5079
    },
    {
      "epoch": 0.8883060109289618,
      "grad_norm": 0.9959543815171495,
      "learning_rate": 0.00018126636330868725,
      "loss": 12.9522,
      "step": 5080
    },
    {
      "epoch": 0.8884808743169399,
      "grad_norm": 0.8418167454117547,
      "learning_rate": 0.00018125811047721256,
      "loss": 12.9248,
      "step": 5081
    },
    {
      "epoch": 0.888655737704918,
      "grad_norm": 0.9948769892243389,
      "learning_rate": 0.00018124985601626324,
      "loss": 12.9383,
      "step": 5082
    },
    {
      "epoch": 0.8888306010928961,
      "grad_norm": 0.9938678133539685,
      "learning_rate": 0.00018124159992600488,
      "loss": 13.0127,
      "step": 5083
    },
    {
      "epoch": 0.8890054644808744,
      "grad_norm": 0.980132306049049,
      "learning_rate": 0.00018123334220660297,
      "loss": 12.7409,
      "step": 5084
    },
    {
      "epoch": 0.8891803278688525,
      "grad_norm": 1.0383227209660626,
      "learning_rate": 0.00018122508285822313,
      "loss": 13.0835,
      "step": 5085
    },
    {
      "epoch": 0.8893551912568306,
      "grad_norm": 1.0120364236400465,
      "learning_rate": 0.000181216821881031,
      "loss": 13.102,
      "step": 5086
    },
    {
      "epoch": 0.8895300546448087,
      "grad_norm": 1.1146537656887965,
      "learning_rate": 0.00018120855927519222,
      "loss": 12.9398,
      "step": 5087
    },
    {
      "epoch": 0.8897049180327868,
      "grad_norm": 0.9356264673072099,
      "learning_rate": 0.00018120029504087247,
      "loss": 12.8644,
      "step": 5088
    },
    {
      "epoch": 0.8898797814207651,
      "grad_norm": 1.0848949317384937,
      "learning_rate": 0.00018119202917823749,
      "loss": 13.0906,
      "step": 5089
    },
    {
      "epoch": 0.8900546448087432,
      "grad_norm": 1.016251396288286,
      "learning_rate": 0.00018118376168745304,
      "loss": 13.1503,
      "step": 5090
    },
    {
      "epoch": 0.8902295081967213,
      "grad_norm": 1.0471597992763684,
      "learning_rate": 0.0001811754925686849,
      "loss": 12.9826,
      "step": 5091
    },
    {
      "epoch": 0.8904043715846994,
      "grad_norm": 1.0749885341416052,
      "learning_rate": 0.00018116722182209887,
      "loss": 13.0301,
      "step": 5092
    },
    {
      "epoch": 0.8905792349726775,
      "grad_norm": 1.0049037766800872,
      "learning_rate": 0.00018115894944786086,
      "loss": 12.9898,
      "step": 5093
    },
    {
      "epoch": 0.8907540983606558,
      "grad_norm": 1.0343870001532744,
      "learning_rate": 0.0001811506754461367,
      "loss": 13.0413,
      "step": 5094
    },
    {
      "epoch": 0.8909289617486339,
      "grad_norm": 1.0117935874224555,
      "learning_rate": 0.00018114239981709232,
      "loss": 13.0212,
      "step": 5095
    },
    {
      "epoch": 0.891103825136612,
      "grad_norm": 0.9806314710675558,
      "learning_rate": 0.00018113412256089373,
      "loss": 12.9622,
      "step": 5096
    },
    {
      "epoch": 0.8912786885245901,
      "grad_norm": 1.0127157625273302,
      "learning_rate": 0.00018112584367770685,
      "loss": 13.0091,
      "step": 5097
    },
    {
      "epoch": 0.8914535519125683,
      "grad_norm": 1.032846083056268,
      "learning_rate": 0.0001811175631676977,
      "loss": 12.9172,
      "step": 5098
    },
    {
      "epoch": 0.8916284153005465,
      "grad_norm": 1.1202799812272217,
      "learning_rate": 0.00018110928103103234,
      "loss": 13.1086,
      "step": 5099
    },
    {
      "epoch": 0.8918032786885246,
      "grad_norm": 0.8840038372579545,
      "learning_rate": 0.00018110099726787685,
      "loss": 12.9451,
      "step": 5100
    },
    {
      "epoch": 0.8919781420765027,
      "grad_norm": 0.8392516082475039,
      "learning_rate": 0.00018109271187839737,
      "loss": 13.0691,
      "step": 5101
    },
    {
      "epoch": 0.8921530054644808,
      "grad_norm": 0.9229198130673434,
      "learning_rate": 0.00018108442486276005,
      "loss": 12.7973,
      "step": 5102
    },
    {
      "epoch": 0.8923278688524591,
      "grad_norm": 0.9222450563823058,
      "learning_rate": 0.000181076136221131,
      "loss": 12.8781,
      "step": 5103
    },
    {
      "epoch": 0.8925027322404372,
      "grad_norm": 0.9174184800739467,
      "learning_rate": 0.00018106784595367653,
      "loss": 12.9729,
      "step": 5104
    },
    {
      "epoch": 0.8926775956284153,
      "grad_norm": 0.807092914821775,
      "learning_rate": 0.00018105955406056283,
      "loss": 12.8352,
      "step": 5105
    },
    {
      "epoch": 0.8928524590163934,
      "grad_norm": 0.844500715650439,
      "learning_rate": 0.00018105126054195617,
      "loss": 12.7797,
      "step": 5106
    },
    {
      "epoch": 0.8930273224043715,
      "grad_norm": 0.8629316342353693,
      "learning_rate": 0.0001810429653980229,
      "loss": 13.0047,
      "step": 5107
    },
    {
      "epoch": 0.8932021857923498,
      "grad_norm": 0.9379629527577505,
      "learning_rate": 0.0001810346686289293,
      "loss": 12.8758,
      "step": 5108
    },
    {
      "epoch": 0.8933770491803279,
      "grad_norm": 0.8958677336967958,
      "learning_rate": 0.00018102637023484182,
      "loss": 12.7112,
      "step": 5109
    },
    {
      "epoch": 0.893551912568306,
      "grad_norm": 0.905195548054946,
      "learning_rate": 0.00018101807021592682,
      "loss": 12.882,
      "step": 5110
    },
    {
      "epoch": 0.8937267759562841,
      "grad_norm": 0.9335134111880777,
      "learning_rate": 0.00018100976857235074,
      "loss": 12.8816,
      "step": 5111
    },
    {
      "epoch": 0.8939016393442623,
      "grad_norm": 0.8706113201428148,
      "learning_rate": 0.0001810014653042801,
      "loss": 12.8862,
      "step": 5112
    },
    {
      "epoch": 0.8940765027322405,
      "grad_norm": 0.9607726174282045,
      "learning_rate": 0.00018099316041188135,
      "loss": 13.1507,
      "step": 5113
    },
    {
      "epoch": 0.8942513661202186,
      "grad_norm": 0.7767957982287768,
      "learning_rate": 0.00018098485389532107,
      "loss": 12.9735,
      "step": 5114
    },
    {
      "epoch": 0.8944262295081967,
      "grad_norm": 0.8876851221397062,
      "learning_rate": 0.00018097654575476577,
      "loss": 12.9024,
      "step": 5115
    },
    {
      "epoch": 0.8946010928961748,
      "grad_norm": 0.9006343087446246,
      "learning_rate": 0.00018096823599038215,
      "loss": 13.2545,
      "step": 5116
    },
    {
      "epoch": 0.894775956284153,
      "grad_norm": 0.8650218870909009,
      "learning_rate": 0.00018095992460233674,
      "loss": 13.032,
      "step": 5117
    },
    {
      "epoch": 0.8949508196721312,
      "grad_norm": 0.8268367627913896,
      "learning_rate": 0.0001809516115907963,
      "loss": 13.0366,
      "step": 5118
    },
    {
      "epoch": 0.8951256830601093,
      "grad_norm": 0.8415495259574407,
      "learning_rate": 0.0001809432969559275,
      "loss": 12.7229,
      "step": 5119
    },
    {
      "epoch": 0.8953005464480874,
      "grad_norm": 0.758302651176997,
      "learning_rate": 0.00018093498069789705,
      "loss": 12.7485,
      "step": 5120
    },
    {
      "epoch": 0.8954754098360656,
      "grad_norm": 0.850002869499218,
      "learning_rate": 0.00018092666281687174,
      "loss": 13.0399,
      "step": 5121
    },
    {
      "epoch": 0.8956502732240437,
      "grad_norm": 0.8491881569616166,
      "learning_rate": 0.00018091834331301836,
      "loss": 13.1546,
      "step": 5122
    },
    {
      "epoch": 0.8958251366120219,
      "grad_norm": 0.8220415686985929,
      "learning_rate": 0.00018091002218650375,
      "loss": 12.9068,
      "step": 5123
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.8321460393479059,
      "learning_rate": 0.00018090169943749476,
      "loss": 12.8271,
      "step": 5124
    },
    {
      "epoch": 0.8961748633879781,
      "grad_norm": 0.8319327429567185,
      "learning_rate": 0.00018089337506615828,
      "loss": 12.7245,
      "step": 5125
    },
    {
      "epoch": 0.8963497267759563,
      "grad_norm": 0.8502166055849788,
      "learning_rate": 0.00018088504907266129,
      "loss": 13.1669,
      "step": 5126
    },
    {
      "epoch": 0.8965245901639344,
      "grad_norm": 0.9478416776715762,
      "learning_rate": 0.0001808767214571707,
      "loss": 12.9557,
      "step": 5127
    },
    {
      "epoch": 0.8966994535519126,
      "grad_norm": 0.8521104005398017,
      "learning_rate": 0.0001808683922198535,
      "loss": 12.8805,
      "step": 5128
    },
    {
      "epoch": 0.8968743169398907,
      "grad_norm": 0.9306150759175077,
      "learning_rate": 0.00018086006136087677,
      "loss": 13.0113,
      "step": 5129
    },
    {
      "epoch": 0.8970491803278688,
      "grad_norm": 0.7481623987048152,
      "learning_rate": 0.00018085172888040752,
      "loss": 12.913,
      "step": 5130
    },
    {
      "epoch": 0.897224043715847,
      "grad_norm": 1.171470520568206,
      "learning_rate": 0.00018084339477861288,
      "loss": 13.0303,
      "step": 5131
    },
    {
      "epoch": 0.8973989071038251,
      "grad_norm": 1.0878588229358424,
      "learning_rate": 0.00018083505905565992,
      "loss": 12.6991,
      "step": 5132
    },
    {
      "epoch": 0.8975737704918033,
      "grad_norm": 0.9273951403479361,
      "learning_rate": 0.00018082672171171584,
      "loss": 12.9352,
      "step": 5133
    },
    {
      "epoch": 0.8977486338797814,
      "grad_norm": 1.2002020984003445,
      "learning_rate": 0.00018081838274694784,
      "loss": 12.9798,
      "step": 5134
    },
    {
      "epoch": 0.8979234972677596,
      "grad_norm": 0.927197043274542,
      "learning_rate": 0.00018081004216152308,
      "loss": 13.031,
      "step": 5135
    },
    {
      "epoch": 0.8980983606557377,
      "grad_norm": 1.053080354928748,
      "learning_rate": 0.0001808016999556089,
      "loss": 13.1826,
      "step": 5136
    },
    {
      "epoch": 0.8982732240437159,
      "grad_norm": 0.7337111130350256,
      "learning_rate": 0.00018079335612937252,
      "loss": 12.8214,
      "step": 5137
    },
    {
      "epoch": 0.898448087431694,
      "grad_norm": 0.937220439660989,
      "learning_rate": 0.00018078501068298129,
      "loss": 12.9531,
      "step": 5138
    },
    {
      "epoch": 0.8986229508196721,
      "grad_norm": 0.8052773093047391,
      "learning_rate": 0.00018077666361660254,
      "loss": 12.9345,
      "step": 5139
    },
    {
      "epoch": 0.8987978142076503,
      "grad_norm": 1.0600957662132253,
      "learning_rate": 0.00018076831493040368,
      "loss": 12.8224,
      "step": 5140
    },
    {
      "epoch": 0.8989726775956284,
      "grad_norm": 0.9591087852955881,
      "learning_rate": 0.00018075996462455212,
      "loss": 13.0325,
      "step": 5141
    },
    {
      "epoch": 0.8991475409836066,
      "grad_norm": 0.9179524589377536,
      "learning_rate": 0.00018075161269921527,
      "loss": 13.0842,
      "step": 5142
    },
    {
      "epoch": 0.8993224043715847,
      "grad_norm": 0.9441291311893211,
      "learning_rate": 0.00018074325915456069,
      "loss": 12.837,
      "step": 5143
    },
    {
      "epoch": 0.8994972677595628,
      "grad_norm": 0.9635860568885037,
      "learning_rate": 0.0001807349039907558,
      "loss": 13.0021,
      "step": 5144
    },
    {
      "epoch": 0.899672131147541,
      "grad_norm": 0.7947993659312179,
      "learning_rate": 0.00018072654720796823,
      "loss": 12.8793,
      "step": 5145
    },
    {
      "epoch": 0.8998469945355191,
      "grad_norm": 0.8745793930713331,
      "learning_rate": 0.00018071818880636552,
      "loss": 13.0529,
      "step": 5146
    },
    {
      "epoch": 0.9000218579234973,
      "grad_norm": 0.8319213280682765,
      "learning_rate": 0.00018070982878611532,
      "loss": 12.881,
      "step": 5147
    },
    {
      "epoch": 0.9001967213114754,
      "grad_norm": 0.936610028834533,
      "learning_rate": 0.0001807014671473852,
      "loss": 12.7527,
      "step": 5148
    },
    {
      "epoch": 0.9003715846994536,
      "grad_norm": 0.8628409783392306,
      "learning_rate": 0.00018069310389034286,
      "loss": 12.9266,
      "step": 5149
    },
    {
      "epoch": 0.9005464480874317,
      "grad_norm": 0.8360238540785179,
      "learning_rate": 0.00018068473901515608,
      "loss": 12.9557,
      "step": 5150
    },
    {
      "epoch": 0.9007213114754098,
      "grad_norm": 0.787913627091436,
      "learning_rate": 0.0001806763725219925,
      "loss": 12.8312,
      "step": 5151
    },
    {
      "epoch": 0.900896174863388,
      "grad_norm": 0.9038167197720968,
      "learning_rate": 0.00018066800441101996,
      "loss": 12.8377,
      "step": 5152
    },
    {
      "epoch": 0.9010710382513661,
      "grad_norm": 0.7963810081881981,
      "learning_rate": 0.00018065963468240625,
      "loss": 13.0151,
      "step": 5153
    },
    {
      "epoch": 0.9012459016393443,
      "grad_norm": 0.9803648059041727,
      "learning_rate": 0.0001806512633363192,
      "loss": 12.9679,
      "step": 5154
    },
    {
      "epoch": 0.9014207650273224,
      "grad_norm": 0.7922144303867487,
      "learning_rate": 0.0001806428903729267,
      "loss": 12.866,
      "step": 5155
    },
    {
      "epoch": 0.9015956284153005,
      "grad_norm": 1.0182811077001679,
      "learning_rate": 0.00018063451579239666,
      "loss": 12.8906,
      "step": 5156
    },
    {
      "epoch": 0.9017704918032787,
      "grad_norm": 0.88537152610552,
      "learning_rate": 0.00018062613959489695,
      "loss": 12.7927,
      "step": 5157
    },
    {
      "epoch": 0.9019453551912568,
      "grad_norm": 0.9087126834822449,
      "learning_rate": 0.00018061776178059563,
      "loss": 12.9486,
      "step": 5158
    },
    {
      "epoch": 0.902120218579235,
      "grad_norm": 0.8888808115535052,
      "learning_rate": 0.00018060938234966062,
      "loss": 13.0714,
      "step": 5159
    },
    {
      "epoch": 0.9022950819672131,
      "grad_norm": 0.844769450029912,
      "learning_rate": 0.00018060100130226002,
      "loss": 12.8143,
      "step": 5160
    },
    {
      "epoch": 0.9024699453551912,
      "grad_norm": 0.7485638089152369,
      "learning_rate": 0.00018059261863856185,
      "loss": 12.7945,
      "step": 5161
    },
    {
      "epoch": 0.9026448087431694,
      "grad_norm": 0.8086703123286283,
      "learning_rate": 0.00018058423435873423,
      "loss": 12.8486,
      "step": 5162
    },
    {
      "epoch": 0.9028196721311476,
      "grad_norm": 0.7646372329485213,
      "learning_rate": 0.0001805758484629453,
      "loss": 12.9792,
      "step": 5163
    },
    {
      "epoch": 0.9029945355191257,
      "grad_norm": 0.859366979471135,
      "learning_rate": 0.0001805674609513632,
      "loss": 12.8985,
      "step": 5164
    },
    {
      "epoch": 0.9031693989071038,
      "grad_norm": 0.9321231194819121,
      "learning_rate": 0.0001805590718241561,
      "loss": 12.8481,
      "step": 5165
    },
    {
      "epoch": 0.9033442622950819,
      "grad_norm": 0.8769244037157088,
      "learning_rate": 0.00018055068108149228,
      "loss": 12.5546,
      "step": 5166
    },
    {
      "epoch": 0.9035191256830601,
      "grad_norm": 0.8171167807483429,
      "learning_rate": 0.00018054228872354,
      "loss": 12.8791,
      "step": 5167
    },
    {
      "epoch": 0.9036939890710383,
      "grad_norm": 0.8769196055254687,
      "learning_rate": 0.00018053389475046754,
      "loss": 12.8551,
      "step": 5168
    },
    {
      "epoch": 0.9038688524590164,
      "grad_norm": 0.931891881392433,
      "learning_rate": 0.0001805254991624432,
      "loss": 12.7935,
      "step": 5169
    },
    {
      "epoch": 0.9040437158469945,
      "grad_norm": 0.8339717783288031,
      "learning_rate": 0.00018051710195963536,
      "loss": 12.9066,
      "step": 5170
    },
    {
      "epoch": 0.9042185792349727,
      "grad_norm": 0.9132245042576608,
      "learning_rate": 0.0001805087031422124,
      "loss": 13.0582,
      "step": 5171
    },
    {
      "epoch": 0.9043934426229508,
      "grad_norm": 0.8518193863158954,
      "learning_rate": 0.00018050030271034274,
      "loss": 12.8618,
      "step": 5172
    },
    {
      "epoch": 0.904568306010929,
      "grad_norm": 0.9004654938246212,
      "learning_rate": 0.00018049190066419485,
      "loss": 13.0011,
      "step": 5173
    },
    {
      "epoch": 0.9047431693989071,
      "grad_norm": 0.8289247850657498,
      "learning_rate": 0.00018048349700393722,
      "loss": 12.6,
      "step": 5174
    },
    {
      "epoch": 0.9049180327868852,
      "grad_norm": 1.0003389919811743,
      "learning_rate": 0.00018047509172973834,
      "loss": 12.9065,
      "step": 5175
    },
    {
      "epoch": 0.9050928961748634,
      "grad_norm": 0.7594824624568506,
      "learning_rate": 0.0001804666848417668,
      "loss": 13.0918,
      "step": 5176
    },
    {
      "epoch": 0.9052677595628416,
      "grad_norm": 0.9322068744344133,
      "learning_rate": 0.00018045827634019114,
      "loss": 12.9499,
      "step": 5177
    },
    {
      "epoch": 0.9054426229508197,
      "grad_norm": 0.8540910680156777,
      "learning_rate": 0.00018044986622518002,
      "loss": 12.8138,
      "step": 5178
    },
    {
      "epoch": 0.9056174863387978,
      "grad_norm": 0.899697970760239,
      "learning_rate": 0.00018044145449690207,
      "loss": 12.7953,
      "step": 5179
    },
    {
      "epoch": 0.9057923497267759,
      "grad_norm": 0.8067176116058846,
      "learning_rate": 0.00018043304115552598,
      "loss": 12.9218,
      "step": 5180
    },
    {
      "epoch": 0.9059672131147541,
      "grad_norm": 0.915527793348976,
      "learning_rate": 0.0001804246262012204,
      "loss": 12.9285,
      "step": 5181
    },
    {
      "epoch": 0.9061420765027323,
      "grad_norm": 0.9480081854450587,
      "learning_rate": 0.00018041620963415417,
      "loss": 13.1672,
      "step": 5182
    },
    {
      "epoch": 0.9063169398907104,
      "grad_norm": 0.9644356888236707,
      "learning_rate": 0.000180407791454496,
      "loss": 12.705,
      "step": 5183
    },
    {
      "epoch": 0.9064918032786885,
      "grad_norm": 0.8664018636381196,
      "learning_rate": 0.00018039937166241477,
      "loss": 12.9907,
      "step": 5184
    },
    {
      "epoch": 0.9066666666666666,
      "grad_norm": 0.8480444869956631,
      "learning_rate": 0.00018039095025807925,
      "loss": 12.9564,
      "step": 5185
    },
    {
      "epoch": 0.9068415300546449,
      "grad_norm": 0.8499451791204102,
      "learning_rate": 0.00018038252724165838,
      "loss": 12.9312,
      "step": 5186
    },
    {
      "epoch": 0.907016393442623,
      "grad_norm": 0.9626956189645027,
      "learning_rate": 0.000180374102613321,
      "loss": 12.9508,
      "step": 5187
    },
    {
      "epoch": 0.9071912568306011,
      "grad_norm": 0.8533075716541787,
      "learning_rate": 0.00018036567637323608,
      "loss": 12.9035,
      "step": 5188
    },
    {
      "epoch": 0.9073661202185792,
      "grad_norm": 0.9492972814794391,
      "learning_rate": 0.0001803572485215726,
      "loss": 12.8627,
      "step": 5189
    },
    {
      "epoch": 0.9075409836065573,
      "grad_norm": 0.8359028576971924,
      "learning_rate": 0.00018034881905849956,
      "loss": 12.7082,
      "step": 5190
    },
    {
      "epoch": 0.9077158469945356,
      "grad_norm": 0.9273886025104433,
      "learning_rate": 0.00018034038798418598,
      "loss": 12.5898,
      "step": 5191
    },
    {
      "epoch": 0.9078907103825137,
      "grad_norm": 0.7994799599536361,
      "learning_rate": 0.000180331955298801,
      "loss": 13.0705,
      "step": 5192
    },
    {
      "epoch": 0.9080655737704918,
      "grad_norm": 0.8712725971198009,
      "learning_rate": 0.0001803235210025136,
      "loss": 12.9636,
      "step": 5193
    },
    {
      "epoch": 0.9082404371584699,
      "grad_norm": 0.7752935241184385,
      "learning_rate": 0.00018031508509549297,
      "loss": 12.9727,
      "step": 5194
    },
    {
      "epoch": 0.908415300546448,
      "grad_norm": 0.8829222162142736,
      "learning_rate": 0.00018030664757790832,
      "loss": 13.105,
      "step": 5195
    },
    {
      "epoch": 0.9085901639344263,
      "grad_norm": 0.7859721666924514,
      "learning_rate": 0.00018029820844992883,
      "loss": 12.9823,
      "step": 5196
    },
    {
      "epoch": 0.9087650273224044,
      "grad_norm": 0.863562628357023,
      "learning_rate": 0.00018028976771172368,
      "loss": 12.716,
      "step": 5197
    },
    {
      "epoch": 0.9089398907103825,
      "grad_norm": 0.9612205599005401,
      "learning_rate": 0.00018028132536346216,
      "loss": 12.8908,
      "step": 5198
    },
    {
      "epoch": 0.9091147540983606,
      "grad_norm": 0.8284554445183626,
      "learning_rate": 0.00018027288140531357,
      "loss": 13.1731,
      "step": 5199
    },
    {
      "epoch": 0.9092896174863389,
      "grad_norm": 0.906937819577629,
      "learning_rate": 0.0001802644358374473,
      "loss": 12.8422,
      "step": 5200
    },
    {
      "epoch": 0.909464480874317,
      "grad_norm": 0.8871982612384192,
      "learning_rate": 0.00018025598866003257,
      "loss": 12.8395,
      "step": 5201
    },
    {
      "epoch": 0.9096393442622951,
      "grad_norm": 0.8234830858232606,
      "learning_rate": 0.00018024753987323888,
      "loss": 12.9718,
      "step": 5202
    },
    {
      "epoch": 0.9098142076502732,
      "grad_norm": 0.7935516611941797,
      "learning_rate": 0.00018023908947723563,
      "loss": 12.9371,
      "step": 5203
    },
    {
      "epoch": 0.9099890710382513,
      "grad_norm": 0.7857027900259291,
      "learning_rate": 0.00018023063747219228,
      "loss": 13.0832,
      "step": 5204
    },
    {
      "epoch": 0.9101639344262296,
      "grad_norm": 0.84860718168229,
      "learning_rate": 0.00018022218385827828,
      "loss": 12.7997,
      "step": 5205
    },
    {
      "epoch": 0.9103387978142077,
      "grad_norm": 0.745245360591727,
      "learning_rate": 0.00018021372863566315,
      "loss": 12.9509,
      "step": 5206
    },
    {
      "epoch": 0.9105136612021858,
      "grad_norm": 0.8649539365633137,
      "learning_rate": 0.00018020527180451652,
      "loss": 12.8899,
      "step": 5207
    },
    {
      "epoch": 0.9106885245901639,
      "grad_norm": 0.8250776483873616,
      "learning_rate": 0.00018019681336500792,
      "loss": 13.1031,
      "step": 5208
    },
    {
      "epoch": 0.910863387978142,
      "grad_norm": 0.7724978984929214,
      "learning_rate": 0.000180188353317307,
      "loss": 12.9686,
      "step": 5209
    },
    {
      "epoch": 0.9110382513661203,
      "grad_norm": 0.7633657587579257,
      "learning_rate": 0.00018017989166158334,
      "loss": 12.8712,
      "step": 5210
    },
    {
      "epoch": 0.9112131147540984,
      "grad_norm": 0.7993847776640524,
      "learning_rate": 0.00018017142839800668,
      "loss": 12.7224,
      "step": 5211
    },
    {
      "epoch": 0.9113879781420765,
      "grad_norm": 0.8378054795553618,
      "learning_rate": 0.00018016296352674674,
      "loss": 12.8908,
      "step": 5212
    },
    {
      "epoch": 0.9115628415300546,
      "grad_norm": 0.8030062274027296,
      "learning_rate": 0.00018015449704797323,
      "loss": 12.8376,
      "step": 5213
    },
    {
      "epoch": 0.9117377049180327,
      "grad_norm": 0.7633811493308538,
      "learning_rate": 0.00018014602896185595,
      "loss": 12.9562,
      "step": 5214
    },
    {
      "epoch": 0.911912568306011,
      "grad_norm": 0.7947928804035849,
      "learning_rate": 0.0001801375592685647,
      "loss": 12.875,
      "step": 5215
    },
    {
      "epoch": 0.9120874316939891,
      "grad_norm": 0.8447161497822839,
      "learning_rate": 0.00018012908796826937,
      "loss": 12.911,
      "step": 5216
    },
    {
      "epoch": 0.9122622950819672,
      "grad_norm": 0.7703539422429438,
      "learning_rate": 0.00018012061506113977,
      "loss": 12.7574,
      "step": 5217
    },
    {
      "epoch": 0.9124371584699453,
      "grad_norm": 0.8859603108306675,
      "learning_rate": 0.00018011214054734583,
      "loss": 12.8593,
      "step": 5218
    },
    {
      "epoch": 0.9126120218579234,
      "grad_norm": 0.8628055909888982,
      "learning_rate": 0.00018010366442705751,
      "loss": 13.0996,
      "step": 5219
    },
    {
      "epoch": 0.9127868852459017,
      "grad_norm": 0.8241218669636062,
      "learning_rate": 0.00018009518670044475,
      "loss": 12.7437,
      "step": 5220
    },
    {
      "epoch": 0.9129617486338798,
      "grad_norm": 0.8509153591703911,
      "learning_rate": 0.0001800867073676776,
      "loss": 12.86,
      "step": 5221
    },
    {
      "epoch": 0.9131366120218579,
      "grad_norm": 0.944221945311615,
      "learning_rate": 0.00018007822642892606,
      "loss": 12.8462,
      "step": 5222
    },
    {
      "epoch": 0.913311475409836,
      "grad_norm": 0.8149290022699504,
      "learning_rate": 0.0001800697438843602,
      "loss": 12.9516,
      "step": 5223
    },
    {
      "epoch": 0.9134863387978142,
      "grad_norm": 0.9076346752209998,
      "learning_rate": 0.00018006125973415012,
      "loss": 12.7079,
      "step": 5224
    },
    {
      "epoch": 0.9136612021857924,
      "grad_norm": 0.9152665585612225,
      "learning_rate": 0.00018005277397846597,
      "loss": 12.8462,
      "step": 5225
    },
    {
      "epoch": 0.9138360655737705,
      "grad_norm": 0.7797902338741713,
      "learning_rate": 0.0001800442866174779,
      "loss": 12.9317,
      "step": 5226
    },
    {
      "epoch": 0.9140109289617486,
      "grad_norm": 0.8598294206413257,
      "learning_rate": 0.00018003579765135612,
      "loss": 12.919,
      "step": 5227
    },
    {
      "epoch": 0.9141857923497267,
      "grad_norm": 0.9369745669625763,
      "learning_rate": 0.00018002730708027086,
      "loss": 12.7851,
      "step": 5228
    },
    {
      "epoch": 0.9143606557377049,
      "grad_norm": 0.8889618686787129,
      "learning_rate": 0.0001800188149043924,
      "loss": 12.8022,
      "step": 5229
    },
    {
      "epoch": 0.9145355191256831,
      "grad_norm": 0.8589080962094103,
      "learning_rate": 0.00018001032112389097,
      "loss": 12.8997,
      "step": 5230
    },
    {
      "epoch": 0.9147103825136612,
      "grad_norm": 0.897873996656303,
      "learning_rate": 0.00018000182573893697,
      "loss": 13.0285,
      "step": 5231
    },
    {
      "epoch": 0.9148852459016393,
      "grad_norm": 0.8105431725926908,
      "learning_rate": 0.0001799933287497007,
      "loss": 12.9463,
      "step": 5232
    },
    {
      "epoch": 0.9150601092896175,
      "grad_norm": 0.8811164416370576,
      "learning_rate": 0.00017998483015635257,
      "loss": 12.9523,
      "step": 5233
    },
    {
      "epoch": 0.9152349726775957,
      "grad_norm": 0.9063399968677726,
      "learning_rate": 0.00017997632995906304,
      "loss": 12.7809,
      "step": 5234
    },
    {
      "epoch": 0.9154098360655738,
      "grad_norm": 0.8472252037565964,
      "learning_rate": 0.00017996782815800252,
      "loss": 12.7435,
      "step": 5235
    },
    {
      "epoch": 0.9155846994535519,
      "grad_norm": 0.7737193945032769,
      "learning_rate": 0.0001799593247533415,
      "loss": 12.9146,
      "step": 5236
    },
    {
      "epoch": 0.91575956284153,
      "grad_norm": 0.8541174759302976,
      "learning_rate": 0.00017995081974525053,
      "loss": 13.0135,
      "step": 5237
    },
    {
      "epoch": 0.9159344262295082,
      "grad_norm": 0.8252194182784662,
      "learning_rate": 0.00017994231313390015,
      "loss": 12.9059,
      "step": 5238
    },
    {
      "epoch": 0.9161092896174864,
      "grad_norm": 0.8069246249833315,
      "learning_rate": 0.0001799338049194609,
      "loss": 13.09,
      "step": 5239
    },
    {
      "epoch": 0.9162841530054645,
      "grad_norm": 0.7920817124821872,
      "learning_rate": 0.00017992529510210348,
      "loss": 12.8105,
      "step": 5240
    },
    {
      "epoch": 0.9164590163934426,
      "grad_norm": 0.8743514545099742,
      "learning_rate": 0.00017991678368199846,
      "loss": 12.9041,
      "step": 5241
    },
    {
      "epoch": 0.9166338797814207,
      "grad_norm": 0.9413026585042086,
      "learning_rate": 0.00017990827065931657,
      "loss": 12.8482,
      "step": 5242
    },
    {
      "epoch": 0.9168087431693989,
      "grad_norm": 0.8891112891700894,
      "learning_rate": 0.00017989975603422846,
      "loss": 12.9789,
      "step": 5243
    },
    {
      "epoch": 0.9169836065573771,
      "grad_norm": 0.7786301700344975,
      "learning_rate": 0.00017989123980690494,
      "loss": 12.8,
      "step": 5244
    },
    {
      "epoch": 0.9171584699453552,
      "grad_norm": 0.775753928348788,
      "learning_rate": 0.00017988272197751678,
      "loss": 12.9226,
      "step": 5245
    },
    {
      "epoch": 0.9173333333333333,
      "grad_norm": 0.8327333086222415,
      "learning_rate": 0.00017987420254623478,
      "loss": 12.7074,
      "step": 5246
    },
    {
      "epoch": 0.9175081967213115,
      "grad_norm": 0.8668269572511896,
      "learning_rate": 0.00017986568151322975,
      "loss": 12.6428,
      "step": 5247
    },
    {
      "epoch": 0.9176830601092896,
      "grad_norm": 0.8037530848961026,
      "learning_rate": 0.00017985715887867257,
      "loss": 12.9264,
      "step": 5248
    },
    {
      "epoch": 0.9178579234972678,
      "grad_norm": 0.9786905491686075,
      "learning_rate": 0.0001798486346427342,
      "loss": 12.7392,
      "step": 5249
    },
    {
      "epoch": 0.9180327868852459,
      "grad_norm": 0.7662531810510088,
      "learning_rate": 0.00017984010880558554,
      "loss": 12.9165,
      "step": 5250
    },
    {
      "epoch": 0.918207650273224,
      "grad_norm": 0.9358317394213745,
      "learning_rate": 0.0001798315813673975,
      "loss": 12.8395,
      "step": 5251
    },
    {
      "epoch": 0.9183825136612022,
      "grad_norm": 0.8871578852124747,
      "learning_rate": 0.0001798230523283412,
      "loss": 12.855,
      "step": 5252
    },
    {
      "epoch": 0.9185573770491803,
      "grad_norm": 0.7625973527369725,
      "learning_rate": 0.0001798145216885876,
      "loss": 12.6649,
      "step": 5253
    },
    {
      "epoch": 0.9187322404371585,
      "grad_norm": 0.845841117085825,
      "learning_rate": 0.00017980598944830774,
      "loss": 12.8502,
      "step": 5254
    },
    {
      "epoch": 0.9189071038251366,
      "grad_norm": 0.7928132343377575,
      "learning_rate": 0.00017979745560767278,
      "loss": 12.8442,
      "step": 5255
    },
    {
      "epoch": 0.9190819672131147,
      "grad_norm": 0.8439048014267267,
      "learning_rate": 0.00017978892016685383,
      "loss": 12.8624,
      "step": 5256
    },
    {
      "epoch": 0.9192568306010929,
      "grad_norm": 0.792344677604097,
      "learning_rate": 0.00017978038312602203,
      "loss": 12.779,
      "step": 5257
    },
    {
      "epoch": 0.919431693989071,
      "grad_norm": 0.8307661984380567,
      "learning_rate": 0.00017977184448534863,
      "loss": 12.6894,
      "step": 5258
    },
    {
      "epoch": 0.9196065573770492,
      "grad_norm": 0.8000412104469738,
      "learning_rate": 0.00017976330424500478,
      "loss": 12.9773,
      "step": 5259
    },
    {
      "epoch": 0.9197814207650273,
      "grad_norm": 0.9048271608783677,
      "learning_rate": 0.00017975476240516178,
      "loss": 12.9264,
      "step": 5260
    },
    {
      "epoch": 0.9199562841530055,
      "grad_norm": 0.8299968458312347,
      "learning_rate": 0.0001797462189659909,
      "loss": 12.4066,
      "step": 5261
    },
    {
      "epoch": 0.9201311475409836,
      "grad_norm": 0.7842901803259982,
      "learning_rate": 0.0001797376739276635,
      "loss": 12.9224,
      "step": 5262
    },
    {
      "epoch": 0.9203060109289618,
      "grad_norm": 0.9577479238047104,
      "learning_rate": 0.00017972912729035092,
      "loss": 12.97,
      "step": 5263
    },
    {
      "epoch": 0.9204808743169399,
      "grad_norm": 0.8453622150375667,
      "learning_rate": 0.0001797205790542245,
      "loss": 12.8801,
      "step": 5264
    },
    {
      "epoch": 0.920655737704918,
      "grad_norm": 0.8031631752311005,
      "learning_rate": 0.00017971202921945572,
      "loss": 12.9139,
      "step": 5265
    },
    {
      "epoch": 0.9208306010928962,
      "grad_norm": 0.9072311320745692,
      "learning_rate": 0.00017970347778621602,
      "loss": 13.0342,
      "step": 5266
    },
    {
      "epoch": 0.9210054644808743,
      "grad_norm": 0.8172287108038649,
      "learning_rate": 0.00017969492475467685,
      "loss": 12.8333,
      "step": 5267
    },
    {
      "epoch": 0.9211803278688525,
      "grad_norm": 0.8814260972426263,
      "learning_rate": 0.00017968637012500972,
      "loss": 12.6263,
      "step": 5268
    },
    {
      "epoch": 0.9213551912568306,
      "grad_norm": 0.7823023677537639,
      "learning_rate": 0.00017967781389738625,
      "loss": 13.1438,
      "step": 5269
    },
    {
      "epoch": 0.9215300546448087,
      "grad_norm": 0.8868776051715113,
      "learning_rate": 0.00017966925607197794,
      "loss": 12.99,
      "step": 5270
    },
    {
      "epoch": 0.9217049180327869,
      "grad_norm": 0.8243857451168536,
      "learning_rate": 0.00017966069664895643,
      "loss": 12.6974,
      "step": 5271
    },
    {
      "epoch": 0.921879781420765,
      "grad_norm": 0.7456992913466955,
      "learning_rate": 0.00017965213562849333,
      "loss": 12.8054,
      "step": 5272
    },
    {
      "epoch": 0.9220546448087432,
      "grad_norm": 0.8172707163931762,
      "learning_rate": 0.0001796435730107604,
      "loss": 12.6755,
      "step": 5273
    },
    {
      "epoch": 0.9222295081967213,
      "grad_norm": 0.7971375842199492,
      "learning_rate": 0.00017963500879592927,
      "loss": 12.9704,
      "step": 5274
    },
    {
      "epoch": 0.9224043715846995,
      "grad_norm": 0.7860074933555617,
      "learning_rate": 0.00017962644298417173,
      "loss": 12.7676,
      "step": 5275
    },
    {
      "epoch": 0.9225792349726776,
      "grad_norm": 0.8061664189753899,
      "learning_rate": 0.0001796178755756595,
      "loss": 12.9827,
      "step": 5276
    },
    {
      "epoch": 0.9227540983606557,
      "grad_norm": 0.8877232384967559,
      "learning_rate": 0.00017960930657056438,
      "loss": 12.9049,
      "step": 5277
    },
    {
      "epoch": 0.9229289617486339,
      "grad_norm": 0.8751454263251308,
      "learning_rate": 0.00017960073596905827,
      "loss": 12.6528,
      "step": 5278
    },
    {
      "epoch": 0.923103825136612,
      "grad_norm": 0.7836167960533988,
      "learning_rate": 0.00017959216377131298,
      "loss": 12.8981,
      "step": 5279
    },
    {
      "epoch": 0.9232786885245902,
      "grad_norm": 0.9972676646131452,
      "learning_rate": 0.00017958358997750043,
      "loss": 12.991,
      "step": 5280
    },
    {
      "epoch": 0.9234535519125683,
      "grad_norm": 1.0844504205744099,
      "learning_rate": 0.00017957501458779252,
      "loss": 12.951,
      "step": 5281
    },
    {
      "epoch": 0.9236284153005464,
      "grad_norm": 0.7973692302548707,
      "learning_rate": 0.00017956643760236128,
      "loss": 12.9454,
      "step": 5282
    },
    {
      "epoch": 0.9238032786885246,
      "grad_norm": 0.9440887974491746,
      "learning_rate": 0.00017955785902137864,
      "loss": 12.9259,
      "step": 5283
    },
    {
      "epoch": 0.9239781420765028,
      "grad_norm": 0.796790342215155,
      "learning_rate": 0.0001795492788450167,
      "loss": 12.9603,
      "step": 5284
    },
    {
      "epoch": 0.9241530054644809,
      "grad_norm": 0.9282750163529793,
      "learning_rate": 0.00017954069707344742,
      "loss": 12.8859,
      "step": 5285
    },
    {
      "epoch": 0.924327868852459,
      "grad_norm": 0.8052099051908356,
      "learning_rate": 0.00017953211370684295,
      "loss": 12.9499,
      "step": 5286
    },
    {
      "epoch": 0.9245027322404371,
      "grad_norm": 0.831512778044425,
      "learning_rate": 0.0001795235287453754,
      "loss": 12.7342,
      "step": 5287
    },
    {
      "epoch": 0.9246775956284153,
      "grad_norm": 0.8381537544811299,
      "learning_rate": 0.00017951494218921695,
      "loss": 12.8616,
      "step": 5288
    },
    {
      "epoch": 0.9248524590163935,
      "grad_norm": 0.7963841767009509,
      "learning_rate": 0.00017950635403853974,
      "loss": 12.6006,
      "step": 5289
    },
    {
      "epoch": 0.9250273224043716,
      "grad_norm": 0.8335533233248866,
      "learning_rate": 0.00017949776429351604,
      "loss": 12.908,
      "step": 5290
    },
    {
      "epoch": 0.9252021857923497,
      "grad_norm": 0.8790926543145458,
      "learning_rate": 0.00017948917295431807,
      "loss": 12.9064,
      "step": 5291
    },
    {
      "epoch": 0.9253770491803278,
      "grad_norm": 1.038690096894774,
      "learning_rate": 0.0001794805800211181,
      "loss": 12.6047,
      "step": 5292
    },
    {
      "epoch": 0.925551912568306,
      "grad_norm": 0.8870955640038997,
      "learning_rate": 0.0001794719854940885,
      "loss": 12.7138,
      "step": 5293
    },
    {
      "epoch": 0.9257267759562842,
      "grad_norm": 0.833612820106668,
      "learning_rate": 0.00017946338937340154,
      "loss": 12.6701,
      "step": 5294
    },
    {
      "epoch": 0.9259016393442623,
      "grad_norm": 0.9196200115361064,
      "learning_rate": 0.00017945479165922966,
      "loss": 12.8416,
      "step": 5295
    },
    {
      "epoch": 0.9260765027322404,
      "grad_norm": 0.818633724701512,
      "learning_rate": 0.00017944619235174525,
      "loss": 13.0231,
      "step": 5296
    },
    {
      "epoch": 0.9262513661202186,
      "grad_norm": 0.9164808318987887,
      "learning_rate": 0.00017943759145112073,
      "loss": 13.1203,
      "step": 5297
    },
    {
      "epoch": 0.9264262295081968,
      "grad_norm": 0.8166295285845646,
      "learning_rate": 0.0001794289889575286,
      "loss": 12.8402,
      "step": 5298
    },
    {
      "epoch": 0.9266010928961749,
      "grad_norm": 1.0340239367864843,
      "learning_rate": 0.0001794203848711414,
      "loss": 12.9705,
      "step": 5299
    },
    {
      "epoch": 0.926775956284153,
      "grad_norm": 0.8376680902914289,
      "learning_rate": 0.00017941177919213161,
      "loss": 12.8761,
      "step": 5300
    },
    {
      "epoch": 0.9269508196721311,
      "grad_norm": 1.061228842436927,
      "learning_rate": 0.0001794031719206718,
      "loss": 12.778,
      "step": 5301
    },
    {
      "epoch": 0.9271256830601093,
      "grad_norm": 0.7626299211151462,
      "learning_rate": 0.0001793945630569346,
      "loss": 12.781,
      "step": 5302
    },
    {
      "epoch": 0.9273005464480875,
      "grad_norm": 0.9867771212989489,
      "learning_rate": 0.00017938595260109266,
      "loss": 12.7196,
      "step": 5303
    },
    {
      "epoch": 0.9274754098360656,
      "grad_norm": 0.9622849675142041,
      "learning_rate": 0.0001793773405533186,
      "loss": 12.9105,
      "step": 5304
    },
    {
      "epoch": 0.9276502732240437,
      "grad_norm": 1.1721390075355103,
      "learning_rate": 0.00017936872691378517,
      "loss": 13.1532,
      "step": 5305
    },
    {
      "epoch": 0.9278251366120218,
      "grad_norm": 0.9049937893184722,
      "learning_rate": 0.00017936011168266503,
      "loss": 12.8414,
      "step": 5306
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.9468259820406735,
      "learning_rate": 0.00017935149486013102,
      "loss": 12.9603,
      "step": 5307
    },
    {
      "epoch": 0.9281748633879782,
      "grad_norm": 0.8272152650837928,
      "learning_rate": 0.00017934287644635587,
      "loss": 12.9254,
      "step": 5308
    },
    {
      "epoch": 0.9283497267759563,
      "grad_norm": 0.8605223352299329,
      "learning_rate": 0.00017933425644151246,
      "loss": 12.7841,
      "step": 5309
    },
    {
      "epoch": 0.9285245901639344,
      "grad_norm": 0.9322390373665379,
      "learning_rate": 0.00017932563484577362,
      "loss": 12.6906,
      "step": 5310
    },
    {
      "epoch": 0.9286994535519125,
      "grad_norm": 0.7968605315440959,
      "learning_rate": 0.0001793170116593122,
      "loss": 12.8275,
      "step": 5311
    },
    {
      "epoch": 0.9288743169398908,
      "grad_norm": 1.0149430034713371,
      "learning_rate": 0.00017930838688230122,
      "loss": 12.7894,
      "step": 5312
    },
    {
      "epoch": 0.9290491803278689,
      "grad_norm": 0.78202546406527,
      "learning_rate": 0.0001792997605149135,
      "loss": 12.7639,
      "step": 5313
    },
    {
      "epoch": 0.929224043715847,
      "grad_norm": 0.838869476707418,
      "learning_rate": 0.00017929113255732215,
      "loss": 12.7711,
      "step": 5314
    },
    {
      "epoch": 0.9293989071038251,
      "grad_norm": 0.8680598153486293,
      "learning_rate": 0.00017928250300970012,
      "loss": 12.7896,
      "step": 5315
    },
    {
      "epoch": 0.9295737704918032,
      "grad_norm": 0.8678947403044841,
      "learning_rate": 0.0001792738718722205,
      "loss": 12.9497,
      "step": 5316
    },
    {
      "epoch": 0.9297486338797815,
      "grad_norm": 0.823359173176945,
      "learning_rate": 0.00017926523914505634,
      "loss": 13.0474,
      "step": 5317
    },
    {
      "epoch": 0.9299234972677596,
      "grad_norm": 0.8702005759884737,
      "learning_rate": 0.00017925660482838074,
      "loss": 12.6481,
      "step": 5318
    },
    {
      "epoch": 0.9300983606557377,
      "grad_norm": 0.7634075485794224,
      "learning_rate": 0.00017924796892236687,
      "loss": 12.8759,
      "step": 5319
    },
    {
      "epoch": 0.9302732240437158,
      "grad_norm": 0.7998131562331222,
      "learning_rate": 0.0001792393314271879,
      "loss": 12.6808,
      "step": 5320
    },
    {
      "epoch": 0.9304480874316939,
      "grad_norm": 0.754033392377483,
      "learning_rate": 0.00017923069234301704,
      "loss": 12.9606,
      "step": 5321
    },
    {
      "epoch": 0.9306229508196722,
      "grad_norm": 1.0145939421317192,
      "learning_rate": 0.00017922205167002754,
      "loss": 12.9352,
      "step": 5322
    },
    {
      "epoch": 0.9307978142076503,
      "grad_norm": 0.8072242334637242,
      "learning_rate": 0.00017921340940839264,
      "loss": 12.7699,
      "step": 5323
    },
    {
      "epoch": 0.9309726775956284,
      "grad_norm": 0.9384567841557956,
      "learning_rate": 0.00017920476555828566,
      "loss": 13.2329,
      "step": 5324
    },
    {
      "epoch": 0.9311475409836065,
      "grad_norm": 0.8417799590165822,
      "learning_rate": 0.00017919612011987995,
      "loss": 12.7547,
      "step": 5325
    },
    {
      "epoch": 0.9313224043715846,
      "grad_norm": 0.9721432426040845,
      "learning_rate": 0.0001791874730933489,
      "loss": 12.9954,
      "step": 5326
    },
    {
      "epoch": 0.9314972677595629,
      "grad_norm": 1.1195096230702908,
      "learning_rate": 0.00017917882447886582,
      "loss": 12.9624,
      "step": 5327
    },
    {
      "epoch": 0.931672131147541,
      "grad_norm": 0.7707182935480317,
      "learning_rate": 0.00017917017427660422,
      "loss": 12.8986,
      "step": 5328
    },
    {
      "epoch": 0.9318469945355191,
      "grad_norm": 1.0027259456739783,
      "learning_rate": 0.00017916152248673758,
      "loss": 12.9188,
      "step": 5329
    },
    {
      "epoch": 0.9320218579234972,
      "grad_norm": 0.7731959482679363,
      "learning_rate": 0.0001791528691094393,
      "loss": 12.8525,
      "step": 5330
    },
    {
      "epoch": 0.9321967213114755,
      "grad_norm": 0.8271685580949573,
      "learning_rate": 0.00017914421414488298,
      "loss": 12.9845,
      "step": 5331
    },
    {
      "epoch": 0.9323715846994536,
      "grad_norm": 0.7564160660317267,
      "learning_rate": 0.00017913555759324217,
      "loss": 12.9114,
      "step": 5332
    },
    {
      "epoch": 0.9325464480874317,
      "grad_norm": 0.8547720410282978,
      "learning_rate": 0.00017912689945469046,
      "loss": 12.8568,
      "step": 5333
    },
    {
      "epoch": 0.9327213114754098,
      "grad_norm": 0.8255407833331474,
      "learning_rate": 0.00017911823972940143,
      "loss": 12.7656,
      "step": 5334
    },
    {
      "epoch": 0.9328961748633879,
      "grad_norm": 0.9510432415590794,
      "learning_rate": 0.0001791095784175488,
      "loss": 12.8751,
      "step": 5335
    },
    {
      "epoch": 0.9330710382513662,
      "grad_norm": 0.8024031042356249,
      "learning_rate": 0.00017910091551930624,
      "loss": 12.9899,
      "step": 5336
    },
    {
      "epoch": 0.9332459016393443,
      "grad_norm": 0.8578331165832629,
      "learning_rate": 0.0001790922510348474,
      "loss": 12.7019,
      "step": 5337
    },
    {
      "epoch": 0.9334207650273224,
      "grad_norm": 0.8783990748726863,
      "learning_rate": 0.0001790835849643461,
      "loss": 12.962,
      "step": 5338
    },
    {
      "epoch": 0.9335956284153005,
      "grad_norm": 0.9089442004476244,
      "learning_rate": 0.00017907491730797612,
      "loss": 12.9344,
      "step": 5339
    },
    {
      "epoch": 0.9337704918032786,
      "grad_norm": 0.87198514979435,
      "learning_rate": 0.00017906624806591126,
      "loss": 12.8942,
      "step": 5340
    },
    {
      "epoch": 0.9339453551912569,
      "grad_norm": 0.8094408767033184,
      "learning_rate": 0.00017905757723832535,
      "loss": 12.5777,
      "step": 5341
    },
    {
      "epoch": 0.934120218579235,
      "grad_norm": 0.8756458740070404,
      "learning_rate": 0.00017904890482539228,
      "loss": 12.8936,
      "step": 5342
    },
    {
      "epoch": 0.9342950819672131,
      "grad_norm": 0.96616161389653,
      "learning_rate": 0.00017904023082728596,
      "loss": 12.7846,
      "step": 5343
    },
    {
      "epoch": 0.9344699453551912,
      "grad_norm": 1.100572202219528,
      "learning_rate": 0.00017903155524418032,
      "loss": 12.8467,
      "step": 5344
    },
    {
      "epoch": 0.9346448087431694,
      "grad_norm": 0.8576172218494414,
      "learning_rate": 0.00017902287807624933,
      "loss": 12.7493,
      "step": 5345
    },
    {
      "epoch": 0.9348196721311476,
      "grad_norm": 1.1730539681618415,
      "learning_rate": 0.00017901419932366702,
      "loss": 12.7442,
      "step": 5346
    },
    {
      "epoch": 0.9349945355191257,
      "grad_norm": 0.7775351991416516,
      "learning_rate": 0.0001790055189866074,
      "loss": 12.6593,
      "step": 5347
    },
    {
      "epoch": 0.9351693989071038,
      "grad_norm": 1.1624806554941205,
      "learning_rate": 0.00017899683706524454,
      "loss": 12.9199,
      "step": 5348
    },
    {
      "epoch": 0.9353442622950819,
      "grad_norm": 0.8331420495178339,
      "learning_rate": 0.00017898815355975255,
      "loss": 13.0538,
      "step": 5349
    },
    {
      "epoch": 0.9355191256830601,
      "grad_norm": 1.136848917267625,
      "learning_rate": 0.00017897946847030558,
      "loss": 12.9809,
      "step": 5350
    },
    {
      "epoch": 0.9356939890710383,
      "grad_norm": 0.8342441477105814,
      "learning_rate": 0.00017897078179707773,
      "loss": 12.8493,
      "step": 5351
    },
    {
      "epoch": 0.9358688524590164,
      "grad_norm": 1.1074411003040632,
      "learning_rate": 0.00017896209354024322,
      "loss": 13.0018,
      "step": 5352
    },
    {
      "epoch": 0.9360437158469945,
      "grad_norm": 0.9934321628912054,
      "learning_rate": 0.00017895340369997634,
      "loss": 12.9668,
      "step": 5353
    },
    {
      "epoch": 0.9362185792349726,
      "grad_norm": 0.993810040367422,
      "learning_rate": 0.00017894471227645128,
      "loss": 12.8522,
      "step": 5354
    },
    {
      "epoch": 0.9363934426229508,
      "grad_norm": 0.9892622423101287,
      "learning_rate": 0.00017893601926984235,
      "loss": 13.0705,
      "step": 5355
    },
    {
      "epoch": 0.936568306010929,
      "grad_norm": 0.8850160628806815,
      "learning_rate": 0.00017892732468032386,
      "loss": 13.0075,
      "step": 5356
    },
    {
      "epoch": 0.9367431693989071,
      "grad_norm": 1.0155768828441731,
      "learning_rate": 0.00017891862850807016,
      "loss": 12.848,
      "step": 5357
    },
    {
      "epoch": 0.9369180327868852,
      "grad_norm": 0.7333414068858152,
      "learning_rate": 0.00017890993075325565,
      "loss": 12.7163,
      "step": 5358
    },
    {
      "epoch": 0.9370928961748634,
      "grad_norm": 1.0430056232070373,
      "learning_rate": 0.00017890123141605475,
      "loss": 12.9844,
      "step": 5359
    },
    {
      "epoch": 0.9372677595628416,
      "grad_norm": 0.7608921444720358,
      "learning_rate": 0.0001788925304966419,
      "loss": 12.4609,
      "step": 5360
    },
    {
      "epoch": 0.9374426229508197,
      "grad_norm": 0.9303717119918913,
      "learning_rate": 0.00017888382799519158,
      "loss": 12.8013,
      "step": 5361
    },
    {
      "epoch": 0.9376174863387978,
      "grad_norm": 0.9059755568909829,
      "learning_rate": 0.00017887512391187831,
      "loss": 12.8192,
      "step": 5362
    },
    {
      "epoch": 0.9377923497267759,
      "grad_norm": 0.8447866026142776,
      "learning_rate": 0.00017886641824687663,
      "loss": 12.8974,
      "step": 5363
    },
    {
      "epoch": 0.9379672131147541,
      "grad_norm": 0.8245659669445853,
      "learning_rate": 0.00017885771100036108,
      "loss": 12.6863,
      "step": 5364
    },
    {
      "epoch": 0.9381420765027323,
      "grad_norm": 0.9993431835625843,
      "learning_rate": 0.00017884900217250636,
      "loss": 12.7233,
      "step": 5365
    },
    {
      "epoch": 0.9383169398907104,
      "grad_norm": 0.7671403867430514,
      "learning_rate": 0.000178840291763487,
      "loss": 12.9775,
      "step": 5366
    },
    {
      "epoch": 0.9384918032786885,
      "grad_norm": 0.8549577163934534,
      "learning_rate": 0.00017883157977347774,
      "loss": 12.7841,
      "step": 5367
    },
    {
      "epoch": 0.9386666666666666,
      "grad_norm": 0.8746622801721646,
      "learning_rate": 0.00017882286620265325,
      "loss": 12.6889,
      "step": 5368
    },
    {
      "epoch": 0.9388415300546448,
      "grad_norm": 0.9390101199088637,
      "learning_rate": 0.00017881415105118825,
      "loss": 12.8934,
      "step": 5369
    },
    {
      "epoch": 0.939016393442623,
      "grad_norm": 0.7834494595482842,
      "learning_rate": 0.00017880543431925757,
      "loss": 12.6165,
      "step": 5370
    },
    {
      "epoch": 0.9391912568306011,
      "grad_norm": 0.9946540238608499,
      "learning_rate": 0.00017879671600703597,
      "loss": 12.58,
      "step": 5371
    },
    {
      "epoch": 0.9393661202185792,
      "grad_norm": 0.833283595908824,
      "learning_rate": 0.00017878799611469826,
      "loss": 12.8172,
      "step": 5372
    },
    {
      "epoch": 0.9395409836065574,
      "grad_norm": 1.0398870220557568,
      "learning_rate": 0.00017877927464241932,
      "loss": 13.0177,
      "step": 5373
    },
    {
      "epoch": 0.9397158469945355,
      "grad_norm": 0.7764465883836256,
      "learning_rate": 0.00017877055159037401,
      "loss": 12.5897,
      "step": 5374
    },
    {
      "epoch": 0.9398907103825137,
      "grad_norm": 1.118723752335751,
      "learning_rate": 0.00017876182695873732,
      "loss": 12.8197,
      "step": 5375
    },
    {
      "epoch": 0.9400655737704918,
      "grad_norm": 0.8437315027513863,
      "learning_rate": 0.00017875310074768418,
      "loss": 13.1136,
      "step": 5376
    },
    {
      "epoch": 0.9402404371584699,
      "grad_norm": 1.0547429817473561,
      "learning_rate": 0.00017874437295738952,
      "loss": 12.8364,
      "step": 5377
    },
    {
      "epoch": 0.9404153005464481,
      "grad_norm": 0.7495267141644609,
      "learning_rate": 0.00017873564358802843,
      "loss": 12.8491,
      "step": 5378
    },
    {
      "epoch": 0.9405901639344262,
      "grad_norm": 1.2059256868213484,
      "learning_rate": 0.00017872691263977596,
      "loss": 12.8546,
      "step": 5379
    },
    {
      "epoch": 0.9407650273224044,
      "grad_norm": 0.8456109976421528,
      "learning_rate": 0.00017871818011280715,
      "loss": 12.9278,
      "step": 5380
    },
    {
      "epoch": 0.9409398907103825,
      "grad_norm": 0.9892056729941389,
      "learning_rate": 0.0001787094460072971,
      "loss": 12.8702,
      "step": 5381
    },
    {
      "epoch": 0.9411147540983607,
      "grad_norm": 0.8018438522792647,
      "learning_rate": 0.00017870071032342102,
      "loss": 12.7688,
      "step": 5382
    },
    {
      "epoch": 0.9412896174863388,
      "grad_norm": 0.9036943265047128,
      "learning_rate": 0.00017869197306135407,
      "loss": 12.8113,
      "step": 5383
    },
    {
      "epoch": 0.9414644808743169,
      "grad_norm": 0.7817343134016483,
      "learning_rate": 0.00017868323422127143,
      "loss": 12.9547,
      "step": 5384
    },
    {
      "epoch": 0.9416393442622951,
      "grad_norm": 0.9475717170959164,
      "learning_rate": 0.00017867449380334834,
      "loss": 12.864,
      "step": 5385
    },
    {
      "epoch": 0.9418142076502732,
      "grad_norm": 0.805136882666296,
      "learning_rate": 0.0001786657518077601,
      "loss": 13.0261,
      "step": 5386
    },
    {
      "epoch": 0.9419890710382514,
      "grad_norm": 1.0448126742250032,
      "learning_rate": 0.000178657008234682,
      "loss": 12.9159,
      "step": 5387
    },
    {
      "epoch": 0.9421639344262295,
      "grad_norm": 0.7766790235055709,
      "learning_rate": 0.00017864826308428935,
      "loss": 12.7719,
      "step": 5388
    },
    {
      "epoch": 0.9423387978142076,
      "grad_norm": 1.0867580235563405,
      "learning_rate": 0.00017863951635675759,
      "loss": 12.9099,
      "step": 5389
    },
    {
      "epoch": 0.9425136612021858,
      "grad_norm": 0.8690593097731596,
      "learning_rate": 0.00017863076805226205,
      "loss": 12.9619,
      "step": 5390
    },
    {
      "epoch": 0.942688524590164,
      "grad_norm": 0.9197575717069003,
      "learning_rate": 0.00017862201817097818,
      "loss": 12.602,
      "step": 5391
    },
    {
      "epoch": 0.9428633879781421,
      "grad_norm": 0.8720555086315062,
      "learning_rate": 0.00017861326671308144,
      "loss": 12.6225,
      "step": 5392
    },
    {
      "epoch": 0.9430382513661202,
      "grad_norm": 0.8853181195179024,
      "learning_rate": 0.0001786045136787473,
      "loss": 12.7871,
      "step": 5393
    },
    {
      "epoch": 0.9432131147540984,
      "grad_norm": 0.8125408370787591,
      "learning_rate": 0.00017859575906815139,
      "loss": 12.9594,
      "step": 5394
    },
    {
      "epoch": 0.9433879781420765,
      "grad_norm": 0.8759999219827205,
      "learning_rate": 0.00017858700288146911,
      "loss": 13.046,
      "step": 5395
    },
    {
      "epoch": 0.9435628415300547,
      "grad_norm": 0.7140072778852576,
      "learning_rate": 0.00017857824511887616,
      "loss": 12.8055,
      "step": 5396
    },
    {
      "epoch": 0.9437377049180328,
      "grad_norm": 0.8879260593963652,
      "learning_rate": 0.00017856948578054814,
      "loss": 12.9955,
      "step": 5397
    },
    {
      "epoch": 0.9439125683060109,
      "grad_norm": 0.8302006823090727,
      "learning_rate": 0.0001785607248666607,
      "loss": 12.7038,
      "step": 5398
    },
    {
      "epoch": 0.9440874316939891,
      "grad_norm": 0.8188175156893397,
      "learning_rate": 0.0001785519623773895,
      "loss": 13.1159,
      "step": 5399
    },
    {
      "epoch": 0.9442622950819672,
      "grad_norm": 0.8887024210200167,
      "learning_rate": 0.00017854319831291026,
      "loss": 13.0512,
      "step": 5400
    },
    {
      "epoch": 0.9444371584699454,
      "grad_norm": 0.8216720504292548,
      "learning_rate": 0.00017853443267339873,
      "loss": 12.8194,
      "step": 5401
    },
    {
      "epoch": 0.9446120218579235,
      "grad_norm": 0.8038161586889213,
      "learning_rate": 0.00017852566545903072,
      "loss": 13.0207,
      "step": 5402
    },
    {
      "epoch": 0.9447868852459016,
      "grad_norm": 0.7798404862898488,
      "learning_rate": 0.00017851689666998198,
      "loss": 12.7921,
      "step": 5403
    },
    {
      "epoch": 0.9449617486338798,
      "grad_norm": 0.86956061372186,
      "learning_rate": 0.00017850812630642842,
      "loss": 12.7037,
      "step": 5404
    },
    {
      "epoch": 0.945136612021858,
      "grad_norm": 0.8581232840403991,
      "learning_rate": 0.00017849935436854583,
      "loss": 13.0678,
      "step": 5405
    },
    {
      "epoch": 0.9453114754098361,
      "grad_norm": 0.8550262874824378,
      "learning_rate": 0.00017849058085651017,
      "loss": 13.0092,
      "step": 5406
    },
    {
      "epoch": 0.9454863387978142,
      "grad_norm": 0.8414675384870756,
      "learning_rate": 0.0001784818057704974,
      "loss": 12.7813,
      "step": 5407
    },
    {
      "epoch": 0.9456612021857923,
      "grad_norm": 0.8519111975945498,
      "learning_rate": 0.0001784730291106834,
      "loss": 12.9357,
      "step": 5408
    },
    {
      "epoch": 0.9458360655737705,
      "grad_norm": 1.0311369453174426,
      "learning_rate": 0.00017846425087724428,
      "loss": 13.1722,
      "step": 5409
    },
    {
      "epoch": 0.9460109289617487,
      "grad_norm": 0.8317790237526603,
      "learning_rate": 0.00017845547107035598,
      "loss": 12.8444,
      "step": 5410
    },
    {
      "epoch": 0.9461857923497268,
      "grad_norm": 1.112747049432959,
      "learning_rate": 0.00017844668969019462,
      "loss": 12.7901,
      "step": 5411
    },
    {
      "epoch": 0.9463606557377049,
      "grad_norm": 0.8039547645525987,
      "learning_rate": 0.00017843790673693625,
      "loss": 12.9965,
      "step": 5412
    },
    {
      "epoch": 0.946535519125683,
      "grad_norm": 1.2886631858473099,
      "learning_rate": 0.00017842912221075703,
      "loss": 12.5608,
      "step": 5413
    },
    {
      "epoch": 0.9467103825136612,
      "grad_norm": 0.7984195940926664,
      "learning_rate": 0.00017842033611183307,
      "loss": 12.92,
      "step": 5414
    },
    {
      "epoch": 0.9468852459016394,
      "grad_norm": 1.2985214352939904,
      "learning_rate": 0.0001784115484403406,
      "loss": 13.0954,
      "step": 5415
    },
    {
      "epoch": 0.9470601092896175,
      "grad_norm": 0.8506933339339493,
      "learning_rate": 0.00017840275919645584,
      "loss": 12.9693,
      "step": 5416
    },
    {
      "epoch": 0.9472349726775956,
      "grad_norm": 1.0682003181695598,
      "learning_rate": 0.00017839396838035501,
      "loss": 12.8815,
      "step": 5417
    },
    {
      "epoch": 0.9474098360655737,
      "grad_norm": 0.8119320507692593,
      "learning_rate": 0.00017838517599221443,
      "loss": 12.7675,
      "step": 5418
    },
    {
      "epoch": 0.947584699453552,
      "grad_norm": 0.9528157504217195,
      "learning_rate": 0.00017837638203221042,
      "loss": 13.0674,
      "step": 5419
    },
    {
      "epoch": 0.9477595628415301,
      "grad_norm": 0.7807799278039523,
      "learning_rate": 0.00017836758650051926,
      "loss": 13.0901,
      "step": 5420
    },
    {
      "epoch": 0.9479344262295082,
      "grad_norm": 0.8982737058424797,
      "learning_rate": 0.0001783587893973174,
      "loss": 12.415,
      "step": 5421
    },
    {
      "epoch": 0.9481092896174863,
      "grad_norm": 0.8256130334617987,
      "learning_rate": 0.0001783499907227812,
      "loss": 13.0622,
      "step": 5422
    },
    {
      "epoch": 0.9482841530054645,
      "grad_norm": 1.0414030982123474,
      "learning_rate": 0.00017834119047708714,
      "loss": 12.9207,
      "step": 5423
    },
    {
      "epoch": 0.9484590163934427,
      "grad_norm": 0.8279748486883098,
      "learning_rate": 0.00017833238866041162,
      "loss": 12.7967,
      "step": 5424
    },
    {
      "epoch": 0.9486338797814208,
      "grad_norm": 1.018218014017526,
      "learning_rate": 0.00017832358527293125,
      "loss": 12.7349,
      "step": 5425
    },
    {
      "epoch": 0.9488087431693989,
      "grad_norm": 0.8974354904031446,
      "learning_rate": 0.0001783147803148225,
      "loss": 12.9848,
      "step": 5426
    },
    {
      "epoch": 0.948983606557377,
      "grad_norm": 0.879386576372689,
      "learning_rate": 0.00017830597378626193,
      "loss": 12.6491,
      "step": 5427
    },
    {
      "epoch": 0.9491584699453552,
      "grad_norm": 0.8874206772620661,
      "learning_rate": 0.00017829716568742619,
      "loss": 12.6155,
      "step": 5428
    },
    {
      "epoch": 0.9493333333333334,
      "grad_norm": 0.866396581870154,
      "learning_rate": 0.00017828835601849187,
      "loss": 12.8611,
      "step": 5429
    },
    {
      "epoch": 0.9495081967213115,
      "grad_norm": 0.8818682366316475,
      "learning_rate": 0.00017827954477963557,
      "loss": 12.7864,
      "step": 5430
    },
    {
      "epoch": 0.9496830601092896,
      "grad_norm": 0.9776919445677904,
      "learning_rate": 0.00017827073197103412,
      "loss": 13.014,
      "step": 5431
    },
    {
      "epoch": 0.9498579234972677,
      "grad_norm": 0.9148388033314732,
      "learning_rate": 0.00017826191759286414,
      "loss": 13.0236,
      "step": 5432
    },
    {
      "epoch": 0.950032786885246,
      "grad_norm": 0.7870887894139538,
      "learning_rate": 0.0001782531016453024,
      "loss": 13.0392,
      "step": 5433
    },
    {
      "epoch": 0.9502076502732241,
      "grad_norm": 0.7791742474865999,
      "learning_rate": 0.00017824428412852573,
      "loss": 12.6212,
      "step": 5434
    },
    {
      "epoch": 0.9503825136612022,
      "grad_norm": 0.8612949799326965,
      "learning_rate": 0.00017823546504271094,
      "loss": 12.7254,
      "step": 5435
    },
    {
      "epoch": 0.9505573770491803,
      "grad_norm": 0.9726288393246836,
      "learning_rate": 0.0001782266443880348,
      "loss": 12.8651,
      "step": 5436
    },
    {
      "epoch": 0.9507322404371584,
      "grad_norm": 0.7359337576556473,
      "learning_rate": 0.00017821782216467433,
      "loss": 12.6529,
      "step": 5437
    },
    {
      "epoch": 0.9509071038251367,
      "grad_norm": 0.9097167759951317,
      "learning_rate": 0.00017820899837280631,
      "loss": 12.593,
      "step": 5438
    },
    {
      "epoch": 0.9510819672131148,
      "grad_norm": 0.7025017224795906,
      "learning_rate": 0.00017820017301260776,
      "loss": 12.8483,
      "step": 5439
    },
    {
      "epoch": 0.9512568306010929,
      "grad_norm": 0.7848557950416679,
      "learning_rate": 0.00017819134608425562,
      "loss": 12.8284,
      "step": 5440
    },
    {
      "epoch": 0.951431693989071,
      "grad_norm": 0.8330367404178189,
      "learning_rate": 0.00017818251758792693,
      "loss": 12.8488,
      "step": 5441
    },
    {
      "epoch": 0.9516065573770491,
      "grad_norm": 0.869666957843523,
      "learning_rate": 0.0001781736875237987,
      "loss": 12.8698,
      "step": 5442
    },
    {
      "epoch": 0.9517814207650274,
      "grad_norm": 0.7984653484059224,
      "learning_rate": 0.00017816485589204801,
      "loss": 13.1134,
      "step": 5443
    },
    {
      "epoch": 0.9519562841530055,
      "grad_norm": 0.7744104887116301,
      "learning_rate": 0.00017815602269285197,
      "loss": 12.7062,
      "step": 5444
    },
    {
      "epoch": 0.9521311475409836,
      "grad_norm": 0.7524020557613991,
      "learning_rate": 0.0001781471879263877,
      "loss": 12.7507,
      "step": 5445
    },
    {
      "epoch": 0.9523060109289617,
      "grad_norm": 0.8708706952085737,
      "learning_rate": 0.00017813835159283237,
      "loss": 12.6828,
      "step": 5446
    },
    {
      "epoch": 0.9524808743169398,
      "grad_norm": 0.7883244168813448,
      "learning_rate": 0.00017812951369236316,
      "loss": 12.73,
      "step": 5447
    },
    {
      "epoch": 0.9526557377049181,
      "grad_norm": 0.7831505360069915,
      "learning_rate": 0.00017812067422515732,
      "loss": 12.9777,
      "step": 5448
    },
    {
      "epoch": 0.9528306010928962,
      "grad_norm": 0.8035612110986171,
      "learning_rate": 0.0001781118331913921,
      "loss": 12.71,
      "step": 5449
    },
    {
      "epoch": 0.9530054644808743,
      "grad_norm": 0.8592077860204013,
      "learning_rate": 0.0001781029905912448,
      "loss": 12.8327,
      "step": 5450
    },
    {
      "epoch": 0.9531803278688524,
      "grad_norm": 0.8373262215274324,
      "learning_rate": 0.00017809414642489272,
      "loss": 13.0498,
      "step": 5451
    },
    {
      "epoch": 0.9533551912568305,
      "grad_norm": 0.7905624749959614,
      "learning_rate": 0.00017808530069251323,
      "loss": 12.8118,
      "step": 5452
    },
    {
      "epoch": 0.9535300546448088,
      "grad_norm": 0.8034238405230001,
      "learning_rate": 0.00017807645339428373,
      "loss": 12.856,
      "step": 5453
    },
    {
      "epoch": 0.9537049180327869,
      "grad_norm": 0.8574493733742414,
      "learning_rate": 0.00017806760453038155,
      "loss": 13.0481,
      "step": 5454
    },
    {
      "epoch": 0.953879781420765,
      "grad_norm": 0.8092574771724781,
      "learning_rate": 0.00017805875410098425,
      "loss": 13.0142,
      "step": 5455
    },
    {
      "epoch": 0.9540546448087431,
      "grad_norm": 0.7102967392310275,
      "learning_rate": 0.00017804990210626924,
      "loss": 12.6516,
      "step": 5456
    },
    {
      "epoch": 0.9542295081967214,
      "grad_norm": 0.8365088683292677,
      "learning_rate": 0.00017804104854641408,
      "loss": 12.8648,
      "step": 5457
    },
    {
      "epoch": 0.9544043715846995,
      "grad_norm": 0.8300477209482454,
      "learning_rate": 0.00017803219342159624,
      "loss": 12.5116,
      "step": 5458
    },
    {
      "epoch": 0.9545792349726776,
      "grad_norm": 0.7922753297276717,
      "learning_rate": 0.00017802333673199334,
      "loss": 12.8582,
      "step": 5459
    },
    {
      "epoch": 0.9547540983606557,
      "grad_norm": 0.8240020757766849,
      "learning_rate": 0.000178014478477783,
      "loss": 12.7625,
      "step": 5460
    },
    {
      "epoch": 0.9549289617486338,
      "grad_norm": 0.7504665636206106,
      "learning_rate": 0.0001780056186591428,
      "loss": 12.913,
      "step": 5461
    },
    {
      "epoch": 0.9551038251366121,
      "grad_norm": 0.9648717750403807,
      "learning_rate": 0.00017799675727625048,
      "loss": 12.91,
      "step": 5462
    },
    {
      "epoch": 0.9552786885245902,
      "grad_norm": 0.8849704985384119,
      "learning_rate": 0.00017798789432928365,
      "loss": 12.4047,
      "step": 5463
    },
    {
      "epoch": 0.9554535519125683,
      "grad_norm": 0.834716319456095,
      "learning_rate": 0.00017797902981842007,
      "loss": 12.7178,
      "step": 5464
    },
    {
      "epoch": 0.9556284153005464,
      "grad_norm": 0.8251470518990999,
      "learning_rate": 0.00017797016374383756,
      "loss": 12.9482,
      "step": 5465
    },
    {
      "epoch": 0.9558032786885245,
      "grad_norm": 0.7738967177583456,
      "learning_rate": 0.00017796129610571384,
      "loss": 12.9742,
      "step": 5466
    },
    {
      "epoch": 0.9559781420765028,
      "grad_norm": 0.8223836240676923,
      "learning_rate": 0.00017795242690422678,
      "loss": 13.0386,
      "step": 5467
    },
    {
      "epoch": 0.9561530054644809,
      "grad_norm": 0.788361218398444,
      "learning_rate": 0.0001779435561395542,
      "loss": 12.9436,
      "step": 5468
    },
    {
      "epoch": 0.956327868852459,
      "grad_norm": 0.8286541997879033,
      "learning_rate": 0.00017793468381187402,
      "loss": 12.4155,
      "step": 5469
    },
    {
      "epoch": 0.9565027322404371,
      "grad_norm": 0.7948543861543097,
      "learning_rate": 0.0001779258099213641,
      "loss": 12.9842,
      "step": 5470
    },
    {
      "epoch": 0.9566775956284153,
      "grad_norm": 0.8551448862906744,
      "learning_rate": 0.00017791693446820247,
      "loss": 12.8361,
      "step": 5471
    },
    {
      "epoch": 0.9568524590163935,
      "grad_norm": 0.9221416137471259,
      "learning_rate": 0.00017790805745256704,
      "loss": 12.8691,
      "step": 5472
    },
    {
      "epoch": 0.9570273224043716,
      "grad_norm": 0.7862875462955403,
      "learning_rate": 0.0001778991788746359,
      "loss": 12.8275,
      "step": 5473
    },
    {
      "epoch": 0.9572021857923497,
      "grad_norm": 0.8903944283540698,
      "learning_rate": 0.00017789029873458697,
      "loss": 12.7762,
      "step": 5474
    },
    {
      "epoch": 0.9573770491803278,
      "grad_norm": 0.826733197487361,
      "learning_rate": 0.0001778814170325984,
      "loss": 12.8315,
      "step": 5475
    },
    {
      "epoch": 0.957551912568306,
      "grad_norm": 0.91859666218542,
      "learning_rate": 0.00017787253376884836,
      "loss": 12.6788,
      "step": 5476
    },
    {
      "epoch": 0.9577267759562842,
      "grad_norm": 0.7498170540992268,
      "learning_rate": 0.00017786364894351487,
      "loss": 12.6438,
      "step": 5477
    },
    {
      "epoch": 0.9579016393442623,
      "grad_norm": 0.9210794224000834,
      "learning_rate": 0.00017785476255677616,
      "loss": 12.9763,
      "step": 5478
    },
    {
      "epoch": 0.9580765027322404,
      "grad_norm": 0.8838211071638569,
      "learning_rate": 0.00017784587460881039,
      "loss": 12.808,
      "step": 5479
    },
    {
      "epoch": 0.9582513661202186,
      "grad_norm": 0.8344435737681407,
      "learning_rate": 0.00017783698509979585,
      "loss": 12.9406,
      "step": 5480
    },
    {
      "epoch": 0.9584262295081967,
      "grad_norm": 0.7680952397130875,
      "learning_rate": 0.00017782809402991075,
      "loss": 12.9543,
      "step": 5481
    },
    {
      "epoch": 0.9586010928961749,
      "grad_norm": 0.7848042002911588,
      "learning_rate": 0.0001778192013993334,
      "loss": 12.6635,
      "step": 5482
    },
    {
      "epoch": 0.958775956284153,
      "grad_norm": 0.7401545572559095,
      "learning_rate": 0.00017781030720824211,
      "loss": 12.9615,
      "step": 5483
    },
    {
      "epoch": 0.9589508196721311,
      "grad_norm": 0.7838840463320057,
      "learning_rate": 0.0001778014114568153,
      "loss": 12.6585,
      "step": 5484
    },
    {
      "epoch": 0.9591256830601093,
      "grad_norm": 0.7555411363388339,
      "learning_rate": 0.0001777925141452313,
      "loss": 12.4839,
      "step": 5485
    },
    {
      "epoch": 0.9593005464480874,
      "grad_norm": 0.840246458925768,
      "learning_rate": 0.00017778361527366848,
      "loss": 12.7669,
      "step": 5486
    },
    {
      "epoch": 0.9594754098360656,
      "grad_norm": 0.8252108969484632,
      "learning_rate": 0.00017777471484230537,
      "loss": 12.6635,
      "step": 5487
    },
    {
      "epoch": 0.9596502732240437,
      "grad_norm": 0.8108580162198427,
      "learning_rate": 0.00017776581285132044,
      "loss": 12.6754,
      "step": 5488
    },
    {
      "epoch": 0.9598251366120218,
      "grad_norm": 0.778248785074034,
      "learning_rate": 0.0001777569093008922,
      "loss": 12.946,
      "step": 5489
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9054758961735601,
      "learning_rate": 0.00017774800419119918,
      "loss": 12.8914,
      "step": 5490
    },
    {
      "epoch": 0.9601748633879782,
      "grad_norm": 0.7926995222330835,
      "learning_rate": 0.00017773909752241996,
      "loss": 12.9094,
      "step": 5491
    },
    {
      "epoch": 0.9603497267759563,
      "grad_norm": 0.9897518836482692,
      "learning_rate": 0.00017773018929473312,
      "loss": 13.0708,
      "step": 5492
    },
    {
      "epoch": 0.9605245901639344,
      "grad_norm": 0.8583794170209462,
      "learning_rate": 0.00017772127950831733,
      "loss": 12.7619,
      "step": 5493
    },
    {
      "epoch": 0.9606994535519126,
      "grad_norm": 0.8660798186480134,
      "learning_rate": 0.00017771236816335127,
      "loss": 12.6166,
      "step": 5494
    },
    {
      "epoch": 0.9608743169398907,
      "grad_norm": 0.8684628439806696,
      "learning_rate": 0.00017770345526001358,
      "loss": 12.9289,
      "step": 5495
    },
    {
      "epoch": 0.9610491803278689,
      "grad_norm": 0.8948110460816365,
      "learning_rate": 0.00017769454079848309,
      "loss": 12.8939,
      "step": 5496
    },
    {
      "epoch": 0.961224043715847,
      "grad_norm": 1.0032777688965908,
      "learning_rate": 0.00017768562477893845,
      "loss": 12.8721,
      "step": 5497
    },
    {
      "epoch": 0.9613989071038251,
      "grad_norm": 0.9450772159177091,
      "learning_rate": 0.00017767670720155853,
      "loss": 12.9724,
      "step": 5498
    },
    {
      "epoch": 0.9615737704918033,
      "grad_norm": 0.969052722190214,
      "learning_rate": 0.00017766778806652215,
      "loss": 12.9416,
      "step": 5499
    },
    {
      "epoch": 0.9617486338797814,
      "grad_norm": 0.777083678525566,
      "learning_rate": 0.00017765886737400812,
      "loss": 12.8022,
      "step": 5500
    },
    {
      "epoch": 0.9619234972677596,
      "grad_norm": 0.9418471321843668,
      "learning_rate": 0.00017764994512419534,
      "loss": 12.9821,
      "step": 5501
    },
    {
      "epoch": 0.9620983606557377,
      "grad_norm": 0.8698468854712209,
      "learning_rate": 0.0001776410213172628,
      "loss": 12.9218,
      "step": 5502
    },
    {
      "epoch": 0.9622732240437158,
      "grad_norm": 0.8660418675722242,
      "learning_rate": 0.00017763209595338935,
      "loss": 12.8174,
      "step": 5503
    },
    {
      "epoch": 0.962448087431694,
      "grad_norm": 0.7575755740729384,
      "learning_rate": 0.000177623169032754,
      "loss": 12.7709,
      "step": 5504
    },
    {
      "epoch": 0.9626229508196721,
      "grad_norm": 0.9676366671799735,
      "learning_rate": 0.00017761424055553584,
      "loss": 12.8469,
      "step": 5505
    },
    {
      "epoch": 0.9627978142076503,
      "grad_norm": 0.7635973065073489,
      "learning_rate": 0.0001776053105219138,
      "loss": 12.7345,
      "step": 5506
    },
    {
      "epoch": 0.9629726775956284,
      "grad_norm": 0.8296886467019895,
      "learning_rate": 0.000177596378932067,
      "loss": 12.9865,
      "step": 5507
    },
    {
      "epoch": 0.9631475409836066,
      "grad_norm": 0.7971576046961911,
      "learning_rate": 0.0001775874457861746,
      "loss": 13.17,
      "step": 5508
    },
    {
      "epoch": 0.9633224043715847,
      "grad_norm": 0.9320554555864495,
      "learning_rate": 0.00017757851108441563,
      "loss": 13.0311,
      "step": 5509
    },
    {
      "epoch": 0.9634972677595628,
      "grad_norm": 0.8110747550234892,
      "learning_rate": 0.00017756957482696934,
      "loss": 12.8062,
      "step": 5510
    },
    {
      "epoch": 0.963672131147541,
      "grad_norm": 0.8789485995877468,
      "learning_rate": 0.00017756063701401492,
      "loss": 13.0167,
      "step": 5511
    },
    {
      "epoch": 0.9638469945355191,
      "grad_norm": 0.8024484983866682,
      "learning_rate": 0.00017755169764573157,
      "loss": 12.6728,
      "step": 5512
    },
    {
      "epoch": 0.9640218579234973,
      "grad_norm": 0.8250826968737487,
      "learning_rate": 0.00017754275672229855,
      "loss": 12.8067,
      "step": 5513
    },
    {
      "epoch": 0.9641967213114754,
      "grad_norm": 0.8903753721956893,
      "learning_rate": 0.0001775338142438952,
      "loss": 12.7589,
      "step": 5514
    },
    {
      "epoch": 0.9643715846994535,
      "grad_norm": 0.9204912862467727,
      "learning_rate": 0.00017752487021070078,
      "loss": 12.6273,
      "step": 5515
    },
    {
      "epoch": 0.9645464480874317,
      "grad_norm": 0.9127513699106486,
      "learning_rate": 0.0001775159246228947,
      "loss": 13.0893,
      "step": 5516
    },
    {
      "epoch": 0.9647213114754098,
      "grad_norm": 0.8149257494443887,
      "learning_rate": 0.00017750697748065633,
      "loss": 12.7542,
      "step": 5517
    },
    {
      "epoch": 0.964896174863388,
      "grad_norm": 0.9155624113728377,
      "learning_rate": 0.00017749802878416507,
      "loss": 12.9439,
      "step": 5518
    },
    {
      "epoch": 0.9650710382513661,
      "grad_norm": 0.8437695605478126,
      "learning_rate": 0.0001774890785336004,
      "loss": 12.3981,
      "step": 5519
    },
    {
      "epoch": 0.9652459016393443,
      "grad_norm": 0.9596010539493668,
      "learning_rate": 0.00017748012672914176,
      "loss": 12.8049,
      "step": 5520
    },
    {
      "epoch": 0.9654207650273224,
      "grad_norm": 0.7624958196934972,
      "learning_rate": 0.0001774711733709687,
      "loss": 13.0296,
      "step": 5521
    },
    {
      "epoch": 0.9655956284153006,
      "grad_norm": 0.8717062584868994,
      "learning_rate": 0.00017746221845926078,
      "loss": 12.8067,
      "step": 5522
    },
    {
      "epoch": 0.9657704918032787,
      "grad_norm": 0.9282089996578966,
      "learning_rate": 0.0001774532619941975,
      "loss": 12.9294,
      "step": 5523
    },
    {
      "epoch": 0.9659453551912568,
      "grad_norm": 0.821044086014568,
      "learning_rate": 0.0001774443039759585,
      "loss": 12.894,
      "step": 5524
    },
    {
      "epoch": 0.966120218579235,
      "grad_norm": 0.8053932616872711,
      "learning_rate": 0.00017743534440472342,
      "loss": 12.7734,
      "step": 5525
    },
    {
      "epoch": 0.9662950819672131,
      "grad_norm": 0.8737618024859286,
      "learning_rate": 0.00017742638328067194,
      "loss": 13.1037,
      "step": 5526
    },
    {
      "epoch": 0.9664699453551913,
      "grad_norm": 0.7363039439577072,
      "learning_rate": 0.00017741742060398374,
      "loss": 12.6715,
      "step": 5527
    },
    {
      "epoch": 0.9666448087431694,
      "grad_norm": 0.8388227673719586,
      "learning_rate": 0.00017740845637483855,
      "loss": 12.6929,
      "step": 5528
    },
    {
      "epoch": 0.9668196721311475,
      "grad_norm": 0.8220001159000943,
      "learning_rate": 0.00017739949059341617,
      "loss": 12.8511,
      "step": 5529
    },
    {
      "epoch": 0.9669945355191257,
      "grad_norm": 0.7890542478551507,
      "learning_rate": 0.0001773905232598963,
      "loss": 12.895,
      "step": 5530
    },
    {
      "epoch": 0.9671693989071038,
      "grad_norm": 0.756032701387264,
      "learning_rate": 0.00017738155437445886,
      "loss": 12.8813,
      "step": 5531
    },
    {
      "epoch": 0.967344262295082,
      "grad_norm": 1.0571439551978072,
      "learning_rate": 0.00017737258393728364,
      "loss": 12.9788,
      "step": 5532
    },
    {
      "epoch": 0.9675191256830601,
      "grad_norm": 0.7961714947539552,
      "learning_rate": 0.00017736361194855055,
      "loss": 12.8841,
      "step": 5533
    },
    {
      "epoch": 0.9676939890710382,
      "grad_norm": 0.8153489071381118,
      "learning_rate": 0.0001773546384084395,
      "loss": 12.9025,
      "step": 5534
    },
    {
      "epoch": 0.9678688524590164,
      "grad_norm": 0.7808184867324267,
      "learning_rate": 0.00017734566331713047,
      "loss": 12.6828,
      "step": 5535
    },
    {
      "epoch": 0.9680437158469946,
      "grad_norm": 0.8821569575958307,
      "learning_rate": 0.00017733668667480336,
      "loss": 12.9887,
      "step": 5536
    },
    {
      "epoch": 0.9682185792349727,
      "grad_norm": 0.8799237278959051,
      "learning_rate": 0.00017732770848163826,
      "loss": 12.8393,
      "step": 5537
    },
    {
      "epoch": 0.9683934426229508,
      "grad_norm": 0.7945505286904734,
      "learning_rate": 0.00017731872873781517,
      "loss": 12.7814,
      "step": 5538
    },
    {
      "epoch": 0.9685683060109289,
      "grad_norm": 0.9112214875783055,
      "learning_rate": 0.00017730974744351418,
      "loss": 12.8639,
      "step": 5539
    },
    {
      "epoch": 0.9687431693989071,
      "grad_norm": 0.8094518533339831,
      "learning_rate": 0.00017730076459891535,
      "loss": 12.8646,
      "step": 5540
    },
    {
      "epoch": 0.9689180327868853,
      "grad_norm": 0.8611465412841025,
      "learning_rate": 0.00017729178020419888,
      "loss": 12.8467,
      "step": 5541
    },
    {
      "epoch": 0.9690928961748634,
      "grad_norm": 0.7939349227003548,
      "learning_rate": 0.0001772827942595449,
      "loss": 12.8031,
      "step": 5542
    },
    {
      "epoch": 0.9692677595628415,
      "grad_norm": 0.7915963415888485,
      "learning_rate": 0.00017727380676513358,
      "loss": 12.9625,
      "step": 5543
    },
    {
      "epoch": 0.9694426229508196,
      "grad_norm": 0.8308173009550835,
      "learning_rate": 0.00017726481772114517,
      "loss": 12.8315,
      "step": 5544
    },
    {
      "epoch": 0.9696174863387979,
      "grad_norm": 0.8137665435673519,
      "learning_rate": 0.00017725582712775997,
      "loss": 12.678,
      "step": 5545
    },
    {
      "epoch": 0.969792349726776,
      "grad_norm": 0.938475870246557,
      "learning_rate": 0.00017724683498515818,
      "loss": 12.5702,
      "step": 5546
    },
    {
      "epoch": 0.9699672131147541,
      "grad_norm": 0.8246184620436696,
      "learning_rate": 0.00017723784129352018,
      "loss": 12.6981,
      "step": 5547
    },
    {
      "epoch": 0.9701420765027322,
      "grad_norm": 0.7922633602661693,
      "learning_rate": 0.00017722884605302633,
      "loss": 12.9579,
      "step": 5548
    },
    {
      "epoch": 0.9703169398907103,
      "grad_norm": 0.8333915625253274,
      "learning_rate": 0.000177219849263857,
      "loss": 12.9174,
      "step": 5549
    },
    {
      "epoch": 0.9704918032786886,
      "grad_norm": 0.8861229270650908,
      "learning_rate": 0.0001772108509261926,
      "loss": 12.966,
      "step": 5550
    },
    {
      "epoch": 0.9706666666666667,
      "grad_norm": 0.8196686145606655,
      "learning_rate": 0.00017720185104021355,
      "loss": 13.1373,
      "step": 5551
    },
    {
      "epoch": 0.9708415300546448,
      "grad_norm": 0.9137618857245425,
      "learning_rate": 0.00017719284960610035,
      "loss": 12.8919,
      "step": 5552
    },
    {
      "epoch": 0.9710163934426229,
      "grad_norm": 0.9224109613593424,
      "learning_rate": 0.0001771838466240335,
      "loss": 12.9771,
      "step": 5553
    },
    {
      "epoch": 0.9711912568306011,
      "grad_norm": 0.7773137699233633,
      "learning_rate": 0.00017717484209419352,
      "loss": 12.7699,
      "step": 5554
    },
    {
      "epoch": 0.9713661202185793,
      "grad_norm": 0.8365673542753388,
      "learning_rate": 0.00017716583601676104,
      "loss": 12.7205,
      "step": 5555
    },
    {
      "epoch": 0.9715409836065574,
      "grad_norm": 0.7581643456488552,
      "learning_rate": 0.0001771568283919166,
      "loss": 12.9417,
      "step": 5556
    },
    {
      "epoch": 0.9717158469945355,
      "grad_norm": 0.8543858482079308,
      "learning_rate": 0.00017714781921984084,
      "loss": 12.6908,
      "step": 5557
    },
    {
      "epoch": 0.9718907103825136,
      "grad_norm": 0.7798737505534057,
      "learning_rate": 0.00017713880850071445,
      "loss": 12.9734,
      "step": 5558
    },
    {
      "epoch": 0.9720655737704919,
      "grad_norm": 0.9058693988140651,
      "learning_rate": 0.00017712979623471807,
      "loss": 13.2163,
      "step": 5559
    },
    {
      "epoch": 0.97224043715847,
      "grad_norm": 1.0527685772056692,
      "learning_rate": 0.00017712078242203248,
      "loss": 12.8653,
      "step": 5560
    },
    {
      "epoch": 0.9724153005464481,
      "grad_norm": 0.9668111618668123,
      "learning_rate": 0.00017711176706283838,
      "loss": 12.9895,
      "step": 5561
    },
    {
      "epoch": 0.9725901639344262,
      "grad_norm": 0.888882857214729,
      "learning_rate": 0.00017710275015731664,
      "loss": 12.4963,
      "step": 5562
    },
    {
      "epoch": 0.9727650273224043,
      "grad_norm": 0.8462637177694591,
      "learning_rate": 0.00017709373170564797,
      "loss": 13.0782,
      "step": 5563
    },
    {
      "epoch": 0.9729398907103826,
      "grad_norm": 0.9686035306379255,
      "learning_rate": 0.0001770847117080133,
      "loss": 12.8497,
      "step": 5564
    },
    {
      "epoch": 0.9731147540983607,
      "grad_norm": 0.8482490553192247,
      "learning_rate": 0.00017707569016459348,
      "loss": 12.924,
      "step": 5565
    },
    {
      "epoch": 0.9732896174863388,
      "grad_norm": 0.8555356910123989,
      "learning_rate": 0.0001770666670755694,
      "loss": 12.9514,
      "step": 5566
    },
    {
      "epoch": 0.9734644808743169,
      "grad_norm": 0.8643375318551747,
      "learning_rate": 0.00017705764244112205,
      "loss": 12.6601,
      "step": 5567
    },
    {
      "epoch": 0.973639344262295,
      "grad_norm": 0.9025027068418806,
      "learning_rate": 0.00017704861626143234,
      "loss": 12.8565,
      "step": 5568
    },
    {
      "epoch": 0.9738142076502733,
      "grad_norm": 0.7542409239117833,
      "learning_rate": 0.00017703958853668131,
      "loss": 12.8227,
      "step": 5569
    },
    {
      "epoch": 0.9739890710382514,
      "grad_norm": 0.7747496392428026,
      "learning_rate": 0.00017703055926705002,
      "loss": 12.7853,
      "step": 5570
    },
    {
      "epoch": 0.9741639344262295,
      "grad_norm": 0.7785606544208911,
      "learning_rate": 0.00017702152845271946,
      "loss": 12.6818,
      "step": 5571
    },
    {
      "epoch": 0.9743387978142076,
      "grad_norm": 0.8124472509458652,
      "learning_rate": 0.00017701249609387078,
      "loss": 12.8883,
      "step": 5572
    },
    {
      "epoch": 0.9745136612021857,
      "grad_norm": 0.7791650531758858,
      "learning_rate": 0.0001770034621906851,
      "loss": 12.5511,
      "step": 5573
    },
    {
      "epoch": 0.974688524590164,
      "grad_norm": 0.8314904254324628,
      "learning_rate": 0.00017699442674334358,
      "loss": 13.0245,
      "step": 5574
    },
    {
      "epoch": 0.9748633879781421,
      "grad_norm": 0.7255698140596163,
      "learning_rate": 0.00017698538975202736,
      "loss": 12.6662,
      "step": 5575
    },
    {
      "epoch": 0.9750382513661202,
      "grad_norm": 0.8316743861806473,
      "learning_rate": 0.00017697635121691774,
      "loss": 12.9771,
      "step": 5576
    },
    {
      "epoch": 0.9752131147540983,
      "grad_norm": 0.8413303916769861,
      "learning_rate": 0.00017696731113819592,
      "loss": 12.6727,
      "step": 5577
    },
    {
      "epoch": 0.9753879781420765,
      "grad_norm": 0.8230697336387662,
      "learning_rate": 0.00017695826951604317,
      "loss": 12.9164,
      "step": 5578
    },
    {
      "epoch": 0.9755628415300547,
      "grad_norm": 0.8723209205881405,
      "learning_rate": 0.00017694922635064084,
      "loss": 13.1363,
      "step": 5579
    },
    {
      "epoch": 0.9757377049180328,
      "grad_norm": 0.8448270111610715,
      "learning_rate": 0.00017694018164217027,
      "loss": 12.6181,
      "step": 5580
    },
    {
      "epoch": 0.9759125683060109,
      "grad_norm": 0.77063410562581,
      "learning_rate": 0.00017693113539081284,
      "loss": 12.8406,
      "step": 5581
    },
    {
      "epoch": 0.976087431693989,
      "grad_norm": 0.7862408452660603,
      "learning_rate": 0.00017692208759674988,
      "loss": 12.741,
      "step": 5582
    },
    {
      "epoch": 0.9762622950819673,
      "grad_norm": 0.7447185199122145,
      "learning_rate": 0.0001769130382601629,
      "loss": 12.9129,
      "step": 5583
    },
    {
      "epoch": 0.9764371584699454,
      "grad_norm": 0.8372058949477078,
      "learning_rate": 0.00017690398738123334,
      "loss": 12.9285,
      "step": 5584
    },
    {
      "epoch": 0.9766120218579235,
      "grad_norm": 0.7290707156751103,
      "learning_rate": 0.00017689493496014273,
      "loss": 12.73,
      "step": 5585
    },
    {
      "epoch": 0.9767868852459016,
      "grad_norm": 0.8077018651295063,
      "learning_rate": 0.00017688588099707257,
      "loss": 13.0393,
      "step": 5586
    },
    {
      "epoch": 0.9769617486338797,
      "grad_norm": 0.7729196012635821,
      "learning_rate": 0.00017687682549220442,
      "loss": 12.8012,
      "step": 5587
    },
    {
      "epoch": 0.977136612021858,
      "grad_norm": 0.8843960742688168,
      "learning_rate": 0.00017686776844571988,
      "loss": 13.0286,
      "step": 5588
    },
    {
      "epoch": 0.9773114754098361,
      "grad_norm": 0.745558612099418,
      "learning_rate": 0.00017685870985780056,
      "loss": 12.6877,
      "step": 5589
    },
    {
      "epoch": 0.9774863387978142,
      "grad_norm": 0.9924152401792767,
      "learning_rate": 0.00017684964972862813,
      "loss": 12.854,
      "step": 5590
    },
    {
      "epoch": 0.9776612021857923,
      "grad_norm": 0.8481925934207298,
      "learning_rate": 0.00017684058805838426,
      "loss": 12.916,
      "step": 5591
    },
    {
      "epoch": 0.9778360655737705,
      "grad_norm": 0.8466060321336636,
      "learning_rate": 0.00017683152484725066,
      "loss": 12.6729,
      "step": 5592
    },
    {
      "epoch": 0.9780109289617487,
      "grad_norm": 0.9389240974189996,
      "learning_rate": 0.0001768224600954091,
      "loss": 12.659,
      "step": 5593
    },
    {
      "epoch": 0.9781857923497268,
      "grad_norm": 0.8718701959827347,
      "learning_rate": 0.0001768133938030413,
      "loss": 12.8107,
      "step": 5594
    },
    {
      "epoch": 0.9783606557377049,
      "grad_norm": 0.8174460537852266,
      "learning_rate": 0.00017680432597032918,
      "loss": 12.8389,
      "step": 5595
    },
    {
      "epoch": 0.978535519125683,
      "grad_norm": 0.7758837178810004,
      "learning_rate": 0.00017679525659745443,
      "loss": 12.7055,
      "step": 5596
    },
    {
      "epoch": 0.9787103825136612,
      "grad_norm": 0.7685683151817299,
      "learning_rate": 0.00017678618568459903,
      "loss": 12.9699,
      "step": 5597
    },
    {
      "epoch": 0.9788852459016394,
      "grad_norm": 0.8855357615514103,
      "learning_rate": 0.00017677711323194485,
      "loss": 12.7654,
      "step": 5598
    },
    {
      "epoch": 0.9790601092896175,
      "grad_norm": 0.9929147161666734,
      "learning_rate": 0.00017676803923967377,
      "loss": 12.5836,
      "step": 5599
    },
    {
      "epoch": 0.9792349726775956,
      "grad_norm": 0.8309982367793829,
      "learning_rate": 0.00017675896370796784,
      "loss": 12.9254,
      "step": 5600
    },
    {
      "epoch": 0.9794098360655737,
      "grad_norm": 0.9882890959528033,
      "learning_rate": 0.00017674988663700898,
      "loss": 12.9354,
      "step": 5601
    },
    {
      "epoch": 0.9795846994535519,
      "grad_norm": 0.8211838147084414,
      "learning_rate": 0.00017674080802697925,
      "loss": 13.104,
      "step": 5602
    },
    {
      "epoch": 0.9797595628415301,
      "grad_norm": 0.9150145643565433,
      "learning_rate": 0.00017673172787806068,
      "loss": 12.9374,
      "step": 5603
    },
    {
      "epoch": 0.9799344262295082,
      "grad_norm": 0.7267629682470045,
      "learning_rate": 0.0001767226461904354,
      "loss": 12.9512,
      "step": 5604
    },
    {
      "epoch": 0.9801092896174863,
      "grad_norm": 0.9070208271266976,
      "learning_rate": 0.00017671356296428549,
      "loss": 12.7537,
      "step": 5605
    },
    {
      "epoch": 0.9802841530054645,
      "grad_norm": 0.7540425084828905,
      "learning_rate": 0.00017670447819979307,
      "loss": 12.5464,
      "step": 5606
    },
    {
      "epoch": 0.9804590163934426,
      "grad_norm": 0.9405298383223933,
      "learning_rate": 0.00017669539189714038,
      "loss": 12.863,
      "step": 5607
    },
    {
      "epoch": 0.9806338797814208,
      "grad_norm": 0.9039431850568402,
      "learning_rate": 0.00017668630405650955,
      "loss": 12.8896,
      "step": 5608
    },
    {
      "epoch": 0.9808087431693989,
      "grad_norm": 0.8377596688887782,
      "learning_rate": 0.0001766772146780829,
      "loss": 12.8657,
      "step": 5609
    },
    {
      "epoch": 0.980983606557377,
      "grad_norm": 0.9393256869054044,
      "learning_rate": 0.00017666812376204266,
      "loss": 12.7998,
      "step": 5610
    },
    {
      "epoch": 0.9811584699453552,
      "grad_norm": 0.7902374362894282,
      "learning_rate": 0.00017665903130857115,
      "loss": 12.8989,
      "step": 5611
    },
    {
      "epoch": 0.9813333333333333,
      "grad_norm": 0.856407440686011,
      "learning_rate": 0.0001766499373178507,
      "loss": 12.8343,
      "step": 5612
    },
    {
      "epoch": 0.9815081967213115,
      "grad_norm": 0.8007149843337797,
      "learning_rate": 0.00017664084179006362,
      "loss": 12.8817,
      "step": 5613
    },
    {
      "epoch": 0.9816830601092896,
      "grad_norm": 0.8470389055250079,
      "learning_rate": 0.00017663174472539237,
      "loss": 12.9001,
      "step": 5614
    },
    {
      "epoch": 0.9818579234972677,
      "grad_norm": 0.8744584932117897,
      "learning_rate": 0.00017662264612401934,
      "loss": 12.8497,
      "step": 5615
    },
    {
      "epoch": 0.9820327868852459,
      "grad_norm": 0.7309406618058045,
      "learning_rate": 0.000176613545986127,
      "loss": 12.7628,
      "step": 5616
    },
    {
      "epoch": 0.9822076502732241,
      "grad_norm": 0.8769633214585669,
      "learning_rate": 0.0001766044443118978,
      "loss": 12.8363,
      "step": 5617
    },
    {
      "epoch": 0.9823825136612022,
      "grad_norm": 0.8672204453941926,
      "learning_rate": 0.00017659534110151432,
      "loss": 12.8854,
      "step": 5618
    },
    {
      "epoch": 0.9825573770491803,
      "grad_norm": 0.7652646664229532,
      "learning_rate": 0.00017658623635515906,
      "loss": 12.9649,
      "step": 5619
    },
    {
      "epoch": 0.9827322404371585,
      "grad_norm": 0.9105462707836653,
      "learning_rate": 0.00017657713007301462,
      "loss": 12.9293,
      "step": 5620
    },
    {
      "epoch": 0.9829071038251366,
      "grad_norm": 0.7802406482865478,
      "learning_rate": 0.00017656802225526357,
      "loss": 13.0266,
      "step": 5621
    },
    {
      "epoch": 0.9830819672131148,
      "grad_norm": 0.9781277082918342,
      "learning_rate": 0.00017655891290208859,
      "loss": 12.8551,
      "step": 5622
    },
    {
      "epoch": 0.9832568306010929,
      "grad_norm": 0.7882723235082223,
      "learning_rate": 0.00017654980201367235,
      "loss": 12.8467,
      "step": 5623
    },
    {
      "epoch": 0.983431693989071,
      "grad_norm": 0.8772167557730505,
      "learning_rate": 0.00017654068959019754,
      "loss": 12.8575,
      "step": 5624
    },
    {
      "epoch": 0.9836065573770492,
      "grad_norm": 0.8304170858071337,
      "learning_rate": 0.0001765315756318469,
      "loss": 12.7323,
      "step": 5625
    },
    {
      "epoch": 0.9837814207650273,
      "grad_norm": 0.8174831169401123,
      "learning_rate": 0.00017652246013880317,
      "loss": 12.6948,
      "step": 5626
    },
    {
      "epoch": 0.9839562841530055,
      "grad_norm": 0.8040934279134748,
      "learning_rate": 0.00017651334311124915,
      "loss": 12.8451,
      "step": 5627
    },
    {
      "epoch": 0.9841311475409836,
      "grad_norm": 0.8330454059423907,
      "learning_rate": 0.00017650422454936772,
      "loss": 12.9513,
      "step": 5628
    },
    {
      "epoch": 0.9843060109289617,
      "grad_norm": 0.828321865763306,
      "learning_rate": 0.00017649510445334166,
      "loss": 12.8108,
      "step": 5629
    },
    {
      "epoch": 0.9844808743169399,
      "grad_norm": 0.7835969743141918,
      "learning_rate": 0.00017648598282335386,
      "loss": 12.9234,
      "step": 5630
    },
    {
      "epoch": 0.984655737704918,
      "grad_norm": 0.9618746038728024,
      "learning_rate": 0.00017647685965958729,
      "loss": 13.0048,
      "step": 5631
    },
    {
      "epoch": 0.9848306010928962,
      "grad_norm": 0.8051460903561881,
      "learning_rate": 0.00017646773496222487,
      "loss": 12.9163,
      "step": 5632
    },
    {
      "epoch": 0.9850054644808743,
      "grad_norm": 0.8502133501301286,
      "learning_rate": 0.00017645860873144958,
      "loss": 12.9848,
      "step": 5633
    },
    {
      "epoch": 0.9851803278688525,
      "grad_norm": 0.8098443583468704,
      "learning_rate": 0.0001764494809674444,
      "loss": 12.7674,
      "step": 5634
    },
    {
      "epoch": 0.9853551912568306,
      "grad_norm": 0.8231037425666412,
      "learning_rate": 0.00017644035167039244,
      "loss": 12.7917,
      "step": 5635
    },
    {
      "epoch": 0.9855300546448087,
      "grad_norm": 0.8071083193453644,
      "learning_rate": 0.0001764312208404767,
      "loss": 12.869,
      "step": 5636
    },
    {
      "epoch": 0.9857049180327869,
      "grad_norm": 0.8700232740222695,
      "learning_rate": 0.00017642208847788032,
      "loss": 12.5632,
      "step": 5637
    },
    {
      "epoch": 0.985879781420765,
      "grad_norm": 0.7445772741034912,
      "learning_rate": 0.00017641295458278638,
      "loss": 12.5077,
      "step": 5638
    },
    {
      "epoch": 0.9860546448087432,
      "grad_norm": 0.8053746317955044,
      "learning_rate": 0.0001764038191553781,
      "loss": 12.6533,
      "step": 5639
    },
    {
      "epoch": 0.9862295081967213,
      "grad_norm": 0.8229226639292568,
      "learning_rate": 0.00017639468219583868,
      "loss": 12.6693,
      "step": 5640
    },
    {
      "epoch": 0.9864043715846994,
      "grad_norm": 0.7478430909887022,
      "learning_rate": 0.0001763855437043513,
      "loss": 12.8906,
      "step": 5641
    },
    {
      "epoch": 0.9865792349726776,
      "grad_norm": 0.8121549354078954,
      "learning_rate": 0.00017637640368109924,
      "loss": 12.9464,
      "step": 5642
    },
    {
      "epoch": 0.9867540983606558,
      "grad_norm": 0.8350768284173523,
      "learning_rate": 0.00017636726212626576,
      "loss": 12.8235,
      "step": 5643
    },
    {
      "epoch": 0.9869289617486339,
      "grad_norm": 0.8440494127948424,
      "learning_rate": 0.00017635811904003423,
      "loss": 12.8052,
      "step": 5644
    },
    {
      "epoch": 0.987103825136612,
      "grad_norm": 0.8397251786058159,
      "learning_rate": 0.00017634897442258795,
      "loss": 13.051,
      "step": 5645
    },
    {
      "epoch": 0.9872786885245901,
      "grad_norm": 0.790291693735974,
      "learning_rate": 0.00017633982827411032,
      "loss": 13.0225,
      "step": 5646
    },
    {
      "epoch": 0.9874535519125683,
      "grad_norm": 0.7795837515785095,
      "learning_rate": 0.00017633068059478473,
      "loss": 12.9444,
      "step": 5647
    },
    {
      "epoch": 0.9876284153005465,
      "grad_norm": 0.8430478523923287,
      "learning_rate": 0.00017632153138479463,
      "loss": 12.9335,
      "step": 5648
    },
    {
      "epoch": 0.9878032786885246,
      "grad_norm": 0.854577501229973,
      "learning_rate": 0.00017631238064432346,
      "loss": 12.5294,
      "step": 5649
    },
    {
      "epoch": 0.9879781420765027,
      "grad_norm": 0.823396265392788,
      "learning_rate": 0.00017630322837355476,
      "loss": 13.027,
      "step": 5650
    },
    {
      "epoch": 0.9881530054644809,
      "grad_norm": 0.7590974470664756,
      "learning_rate": 0.00017629407457267208,
      "loss": 12.7697,
      "step": 5651
    },
    {
      "epoch": 0.988327868852459,
      "grad_norm": 0.7787068337728749,
      "learning_rate": 0.00017628491924185892,
      "loss": 12.856,
      "step": 5652
    },
    {
      "epoch": 0.9885027322404372,
      "grad_norm": 0.8459364886572085,
      "learning_rate": 0.00017627576238129892,
      "loss": 12.9255,
      "step": 5653
    },
    {
      "epoch": 0.9886775956284153,
      "grad_norm": 0.7923768017517762,
      "learning_rate": 0.0001762666039911757,
      "loss": 12.9376,
      "step": 5654
    },
    {
      "epoch": 0.9888524590163934,
      "grad_norm": 0.7703745942289201,
      "learning_rate": 0.00017625744407167288,
      "loss": 12.9167,
      "step": 5655
    },
    {
      "epoch": 0.9890273224043716,
      "grad_norm": 0.711318139269569,
      "learning_rate": 0.00017624828262297415,
      "loss": 12.6926,
      "step": 5656
    },
    {
      "epoch": 0.9892021857923498,
      "grad_norm": 0.8034207632456646,
      "learning_rate": 0.00017623911964526325,
      "loss": 13.0194,
      "step": 5657
    },
    {
      "epoch": 0.9893770491803279,
      "grad_norm": 0.8772916986710858,
      "learning_rate": 0.00017622995513872394,
      "loss": 12.6844,
      "step": 5658
    },
    {
      "epoch": 0.989551912568306,
      "grad_norm": 0.7629145245903011,
      "learning_rate": 0.00017622078910353996,
      "loss": 12.7356,
      "step": 5659
    },
    {
      "epoch": 0.9897267759562841,
      "grad_norm": 0.7914731836184504,
      "learning_rate": 0.0001762116215398951,
      "loss": 12.6503,
      "step": 5660
    },
    {
      "epoch": 0.9899016393442623,
      "grad_norm": 0.7113474482378267,
      "learning_rate": 0.0001762024524479733,
      "loss": 12.8616,
      "step": 5661
    },
    {
      "epoch": 0.9900765027322405,
      "grad_norm": 0.8036896623176655,
      "learning_rate": 0.0001761932818279583,
      "loss": 12.8397,
      "step": 5662
    },
    {
      "epoch": 0.9902513661202186,
      "grad_norm": 0.7623807277249437,
      "learning_rate": 0.00017618410968003406,
      "loss": 12.6464,
      "step": 5663
    },
    {
      "epoch": 0.9904262295081967,
      "grad_norm": 0.868262183954439,
      "learning_rate": 0.0001761749360043845,
      "loss": 12.9581,
      "step": 5664
    },
    {
      "epoch": 0.9906010928961748,
      "grad_norm": 0.8580192969396879,
      "learning_rate": 0.0001761657608011936,
      "loss": 12.6733,
      "step": 5665
    },
    {
      "epoch": 0.990775956284153,
      "grad_norm": 0.7930404749103727,
      "learning_rate": 0.00017615658407064535,
      "loss": 13.112,
      "step": 5666
    },
    {
      "epoch": 0.9909508196721312,
      "grad_norm": 0.7533607615655907,
      "learning_rate": 0.00017614740581292372,
      "loss": 12.7322,
      "step": 5667
    },
    {
      "epoch": 0.9911256830601093,
      "grad_norm": 0.765434168066667,
      "learning_rate": 0.00017613822602821283,
      "loss": 12.6914,
      "step": 5668
    },
    {
      "epoch": 0.9913005464480874,
      "grad_norm": 0.7809274230442166,
      "learning_rate": 0.00017612904471669673,
      "loss": 12.9553,
      "step": 5669
    },
    {
      "epoch": 0.9914754098360655,
      "grad_norm": 0.7954349055671707,
      "learning_rate": 0.00017611986187855954,
      "loss": 12.8523,
      "step": 5670
    },
    {
      "epoch": 0.9916502732240438,
      "grad_norm": 0.7667871002148201,
      "learning_rate": 0.0001761106775139854,
      "loss": 12.7246,
      "step": 5671
    },
    {
      "epoch": 0.9918251366120219,
      "grad_norm": 0.791186483069234,
      "learning_rate": 0.00017610149162315846,
      "loss": 12.8398,
      "step": 5672
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.772053293824559,
      "learning_rate": 0.00017609230420626297,
      "loss": 12.6843,
      "step": 5673
    },
    {
      "epoch": 0.9921748633879781,
      "grad_norm": 0.8786285309349217,
      "learning_rate": 0.00017608311526348314,
      "loss": 12.6885,
      "step": 5674
    },
    {
      "epoch": 0.9923497267759562,
      "grad_norm": 0.8692430573636772,
      "learning_rate": 0.00017607392479500325,
      "loss": 12.756,
      "step": 5675
    },
    {
      "epoch": 0.9925245901639345,
      "grad_norm": 0.7780227990451996,
      "learning_rate": 0.00017606473280100756,
      "loss": 12.8472,
      "step": 5676
    },
    {
      "epoch": 0.9926994535519126,
      "grad_norm": 0.8593971779520992,
      "learning_rate": 0.0001760555392816805,
      "loss": 12.8643,
      "step": 5677
    },
    {
      "epoch": 0.9928743169398907,
      "grad_norm": 0.8407203037283457,
      "learning_rate": 0.0001760463442372063,
      "loss": 12.8205,
      "step": 5678
    },
    {
      "epoch": 0.9930491803278688,
      "grad_norm": 0.765644760034387,
      "learning_rate": 0.00017603714766776943,
      "loss": 12.8149,
      "step": 5679
    },
    {
      "epoch": 0.993224043715847,
      "grad_norm": 0.8405443027150902,
      "learning_rate": 0.00017602794957355427,
      "loss": 12.8261,
      "step": 5680
    },
    {
      "epoch": 0.9933989071038252,
      "grad_norm": 0.8529612295207806,
      "learning_rate": 0.00017601874995474526,
      "loss": 12.7087,
      "step": 5681
    },
    {
      "epoch": 0.9935737704918033,
      "grad_norm": 0.9582425923277105,
      "learning_rate": 0.00017600954881152693,
      "loss": 12.7122,
      "step": 5682
    },
    {
      "epoch": 0.9937486338797814,
      "grad_norm": 0.824301369251614,
      "learning_rate": 0.0001760003461440838,
      "loss": 12.7651,
      "step": 5683
    },
    {
      "epoch": 0.9939234972677595,
      "grad_norm": 0.930399872485115,
      "learning_rate": 0.00017599114195260033,
      "loss": 13.0221,
      "step": 5684
    },
    {
      "epoch": 0.9940983606557378,
      "grad_norm": 1.0201931029677764,
      "learning_rate": 0.00017598193623726116,
      "loss": 12.6545,
      "step": 5685
    },
    {
      "epoch": 0.9942732240437159,
      "grad_norm": 0.8077297407767694,
      "learning_rate": 0.00017597272899825086,
      "loss": 12.5154,
      "step": 5686
    },
    {
      "epoch": 0.994448087431694,
      "grad_norm": 0.9713969026438488,
      "learning_rate": 0.0001759635202357541,
      "loss": 13.0276,
      "step": 5687
    },
    {
      "epoch": 0.9946229508196721,
      "grad_norm": 0.8096624032534688,
      "learning_rate": 0.0001759543099499555,
      "loss": 12.9839,
      "step": 5688
    },
    {
      "epoch": 0.9947978142076502,
      "grad_norm": 0.9679577461394525,
      "learning_rate": 0.00017594509814103975,
      "loss": 12.9683,
      "step": 5689
    },
    {
      "epoch": 0.9949726775956285,
      "grad_norm": 0.8487484816561336,
      "learning_rate": 0.00017593588480919166,
      "loss": 12.2874,
      "step": 5690
    },
    {
      "epoch": 0.9951475409836066,
      "grad_norm": 0.8068569156907133,
      "learning_rate": 0.0001759266699545959,
      "loss": 12.807,
      "step": 5691
    },
    {
      "epoch": 0.9953224043715847,
      "grad_norm": 0.7774238508602642,
      "learning_rate": 0.00017591745357743726,
      "loss": 12.9238,
      "step": 5692
    },
    {
      "epoch": 0.9954972677595628,
      "grad_norm": 0.8358514841164671,
      "learning_rate": 0.00017590823567790063,
      "loss": 12.731,
      "step": 5693
    },
    {
      "epoch": 0.9956721311475409,
      "grad_norm": 0.7448975773458913,
      "learning_rate": 0.00017589901625617076,
      "loss": 12.6672,
      "step": 5694
    },
    {
      "epoch": 0.9958469945355192,
      "grad_norm": 0.8145539936151599,
      "learning_rate": 0.0001758897953124326,
      "loss": 12.4793,
      "step": 5695
    },
    {
      "epoch": 0.9960218579234973,
      "grad_norm": 0.8197832141794944,
      "learning_rate": 0.00017588057284687102,
      "loss": 12.8258,
      "step": 5696
    },
    {
      "epoch": 0.9961967213114754,
      "grad_norm": 0.8248058734889726,
      "learning_rate": 0.00017587134885967097,
      "loss": 12.8241,
      "step": 5697
    },
    {
      "epoch": 0.9963715846994535,
      "grad_norm": 0.794888105708432,
      "learning_rate": 0.0001758621233510174,
      "loss": 12.9108,
      "step": 5698
    },
    {
      "epoch": 0.9965464480874316,
      "grad_norm": 0.7448998664380841,
      "learning_rate": 0.00017585289632109536,
      "loss": 12.7217,
      "step": 5699
    },
    {
      "epoch": 0.9967213114754099,
      "grad_norm": 0.8988406444510526,
      "learning_rate": 0.00017584366777008984,
      "loss": 12.889,
      "step": 5700
    },
    {
      "epoch": 0.996896174863388,
      "grad_norm": 0.7984053937143537,
      "learning_rate": 0.00017583443769818592,
      "loss": 12.8331,
      "step": 5701
    },
    {
      "epoch": 0.9970710382513661,
      "grad_norm": 0.8874474277339607,
      "learning_rate": 0.00017582520610556865,
      "loss": 12.6977,
      "step": 5702
    },
    {
      "epoch": 0.9972459016393442,
      "grad_norm": 0.7445044733553431,
      "learning_rate": 0.00017581597299242324,
      "loss": 12.8139,
      "step": 5703
    },
    {
      "epoch": 0.9974207650273224,
      "grad_norm": 0.8479721007680259,
      "learning_rate": 0.00017580673835893473,
      "loss": 12.7304,
      "step": 5704
    },
    {
      "epoch": 0.9975956284153006,
      "grad_norm": 0.7941689571697209,
      "learning_rate": 0.00017579750220528837,
      "loss": 12.788,
      "step": 5705
    },
    {
      "epoch": 0.9977704918032787,
      "grad_norm": 0.8000568603633336,
      "learning_rate": 0.0001757882645316694,
      "loss": 12.8497,
      "step": 5706
    },
    {
      "epoch": 0.9979453551912568,
      "grad_norm": 0.845580582194168,
      "learning_rate": 0.000175779025338263,
      "loss": 12.8389,
      "step": 5707
    },
    {
      "epoch": 0.9981202185792349,
      "grad_norm": 0.7897300483971529,
      "learning_rate": 0.00017576978462525443,
      "loss": 12.8213,
      "step": 5708
    },
    {
      "epoch": 0.9982950819672131,
      "grad_norm": 0.7243332861590495,
      "learning_rate": 0.0001757605423928291,
      "loss": 12.5071,
      "step": 5709
    },
    {
      "epoch": 0.9984699453551913,
      "grad_norm": 0.8334000285868253,
      "learning_rate": 0.00017575129864117224,
      "loss": 12.9657,
      "step": 5710
    },
    {
      "epoch": 0.9986448087431694,
      "grad_norm": 0.8214379042591812,
      "learning_rate": 0.00017574205337046928,
      "loss": 12.6491,
      "step": 5711
    },
    {
      "epoch": 0.9988196721311475,
      "grad_norm": 0.8299055886825063,
      "learning_rate": 0.00017573280658090556,
      "loss": 12.8479,
      "step": 5712
    },
    {
      "epoch": 0.9989945355191256,
      "grad_norm": 0.8776920162667662,
      "learning_rate": 0.00017572355827266656,
      "loss": 12.5871,
      "step": 5713
    },
    {
      "epoch": 0.9991693989071039,
      "grad_norm": 0.8596616100132742,
      "learning_rate": 0.0001757143084459377,
      "loss": 12.7958,
      "step": 5714
    },
    {
      "epoch": 0.999344262295082,
      "grad_norm": 1.0359945579411598,
      "learning_rate": 0.0001757050571009045,
      "loss": 12.9942,
      "step": 5715
    },
    {
      "epoch": 0.9995191256830601,
      "grad_norm": 0.8636543731097387,
      "learning_rate": 0.00017569580423775245,
      "loss": 12.8126,
      "step": 5716
    },
    {
      "epoch": 0.9996939890710382,
      "grad_norm": 0.8760481846879824,
      "learning_rate": 0.0001756865498566671,
      "loss": 12.7696,
      "step": 5717
    },
    {
      "epoch": 0.9998688524590164,
      "grad_norm": 0.8003840631518347,
      "learning_rate": 0.00017567729395783405,
      "loss": 12.7241,
      "step": 5718
    },
    {
      "epoch": 1.0000437158469946,
      "grad_norm": 0.9150489894239181,
      "learning_rate": 0.0001756680365414389,
      "loss": 12.6622,
      "step": 5719
    },
    {
      "epoch": 1.0002185792349727,
      "grad_norm": 0.8519469026014624,
      "learning_rate": 0.00017565877760766728,
      "loss": 12.4006,
      "step": 5720
    },
    {
      "epoch": 1.0003934426229508,
      "grad_norm": 0.8204190031608104,
      "learning_rate": 0.00017564951715670488,
      "loss": 12.65,
      "step": 5721
    },
    {
      "epoch": 1.000568306010929,
      "grad_norm": 0.8166555381788109,
      "learning_rate": 0.00017564025518873738,
      "loss": 12.7251,
      "step": 5722
    },
    {
      "epoch": 1.000743169398907,
      "grad_norm": 0.8306573543330523,
      "learning_rate": 0.00017563099170395048,
      "loss": 12.6174,
      "step": 5723
    },
    {
      "epoch": 1.0009180327868852,
      "grad_norm": 0.8530517904787471,
      "learning_rate": 0.00017562172670253,
      "loss": 12.6902,
      "step": 5724
    },
    {
      "epoch": 1.0010928961748633,
      "grad_norm": 0.8956253228567301,
      "learning_rate": 0.00017561246018466174,
      "loss": 12.6224,
      "step": 5725
    },
    {
      "epoch": 1.0012677595628414,
      "grad_norm": 0.815136649212103,
      "learning_rate": 0.00017560319215053147,
      "loss": 12.4986,
      "step": 5726
    },
    {
      "epoch": 1.0014426229508198,
      "grad_norm": 0.9416486589545071,
      "learning_rate": 0.00017559392260032506,
      "loss": 12.6203,
      "step": 5727
    },
    {
      "epoch": 1.0016174863387979,
      "grad_norm": 0.742694053009519,
      "learning_rate": 0.0001755846515342284,
      "loss": 12.8132,
      "step": 5728
    },
    {
      "epoch": 1.001792349726776,
      "grad_norm": 0.8313907737641714,
      "learning_rate": 0.0001755753789524274,
      "loss": 12.6501,
      "step": 5729
    },
    {
      "epoch": 1.0019672131147541,
      "grad_norm": 0.8025801114198043,
      "learning_rate": 0.00017556610485510801,
      "loss": 12.6032,
      "step": 5730
    },
    {
      "epoch": 1.0021420765027322,
      "grad_norm": 0.7829142444713206,
      "learning_rate": 0.00017555682924245622,
      "loss": 12.8201,
      "step": 5731
    },
    {
      "epoch": 1.0023169398907104,
      "grad_norm": 0.7909635444909767,
      "learning_rate": 0.000175547552114658,
      "loss": 12.7359,
      "step": 5732
    },
    {
      "epoch": 1.0024918032786885,
      "grad_norm": 0.7650778453157628,
      "learning_rate": 0.00017553827347189938,
      "loss": 12.4513,
      "step": 5733
    },
    {
      "epoch": 1.0026666666666666,
      "grad_norm": 0.7779884165357858,
      "learning_rate": 0.00017552899331436646,
      "loss": 12.4947,
      "step": 5734
    },
    {
      "epoch": 1.0028415300546447,
      "grad_norm": 0.8763166886743834,
      "learning_rate": 0.00017551971164224533,
      "loss": 12.4019,
      "step": 5735
    },
    {
      "epoch": 1.003016393442623,
      "grad_norm": 0.9306022647323137,
      "learning_rate": 0.00017551042845572208,
      "loss": 12.6722,
      "step": 5736
    },
    {
      "epoch": 1.0031912568306012,
      "grad_norm": 0.7132093153876273,
      "learning_rate": 0.0001755011437549829,
      "loss": 12.5796,
      "step": 5737
    },
    {
      "epoch": 1.0033661202185793,
      "grad_norm": 0.8874506589633955,
      "learning_rate": 0.00017549185754021397,
      "loss": 12.4501,
      "step": 5738
    },
    {
      "epoch": 1.0035409836065574,
      "grad_norm": 0.7122287292953569,
      "learning_rate": 0.00017548256981160152,
      "loss": 12.7592,
      "step": 5739
    },
    {
      "epoch": 1.0037158469945355,
      "grad_norm": 0.9078350783340938,
      "learning_rate": 0.00017547328056933176,
      "loss": 12.72,
      "step": 5740
    },
    {
      "epoch": 1.0038907103825137,
      "grad_norm": 0.8006668526748517,
      "learning_rate": 0.00017546398981359103,
      "loss": 12.5844,
      "step": 5741
    },
    {
      "epoch": 1.0040655737704918,
      "grad_norm": 0.8960018897618263,
      "learning_rate": 0.00017545469754456553,
      "loss": 12.4919,
      "step": 5742
    },
    {
      "epoch": 1.00424043715847,
      "grad_norm": 0.8279700179766089,
      "learning_rate": 0.00017544540376244173,
      "loss": 12.4952,
      "step": 5743
    },
    {
      "epoch": 1.004415300546448,
      "grad_norm": 0.7782836519832557,
      "learning_rate": 0.00017543610846740593,
      "loss": 12.5629,
      "step": 5744
    },
    {
      "epoch": 1.0045901639344261,
      "grad_norm": 0.9595696931309365,
      "learning_rate": 0.0001754268116596445,
      "loss": 12.5455,
      "step": 5745
    },
    {
      "epoch": 1.0047650273224045,
      "grad_norm": 0.739575104208006,
      "learning_rate": 0.00017541751333934396,
      "loss": 12.4343,
      "step": 5746
    },
    {
      "epoch": 1.0049398907103826,
      "grad_norm": 0.9240931642142548,
      "learning_rate": 0.00017540821350669065,
      "loss": 12.6544,
      "step": 5747
    },
    {
      "epoch": 1.0051147540983607,
      "grad_norm": 0.8612497980761474,
      "learning_rate": 0.00017539891216187119,
      "loss": 12.5671,
      "step": 5748
    },
    {
      "epoch": 1.0052896174863388,
      "grad_norm": 0.7649998881078872,
      "learning_rate": 0.000175389609305072,
      "loss": 12.5722,
      "step": 5749
    },
    {
      "epoch": 1.005464480874317,
      "grad_norm": 0.8529825245384849,
      "learning_rate": 0.00017538030493647968,
      "loss": 12.5736,
      "step": 5750
    },
    {
      "epoch": 1.005639344262295,
      "grad_norm": 0.8657801594945541,
      "learning_rate": 0.0001753709990562808,
      "loss": 12.717,
      "step": 5751
    },
    {
      "epoch": 1.0058142076502732,
      "grad_norm": 0.7669430650712411,
      "learning_rate": 0.00017536169166466195,
      "loss": 12.6724,
      "step": 5752
    },
    {
      "epoch": 1.0059890710382513,
      "grad_norm": 0.7687278606649389,
      "learning_rate": 0.00017535238276180978,
      "loss": 12.5231,
      "step": 5753
    },
    {
      "epoch": 1.0061639344262294,
      "grad_norm": 0.8680228461136215,
      "learning_rate": 0.00017534307234791098,
      "loss": 12.639,
      "step": 5754
    },
    {
      "epoch": 1.0063387978142075,
      "grad_norm": 0.9050188798223338,
      "learning_rate": 0.00017533376042315227,
      "loss": 12.498,
      "step": 5755
    },
    {
      "epoch": 1.0065136612021859,
      "grad_norm": 0.7781041277951375,
      "learning_rate": 0.00017532444698772036,
      "loss": 12.646,
      "step": 5756
    },
    {
      "epoch": 1.006688524590164,
      "grad_norm": 0.7514892619773957,
      "learning_rate": 0.00017531513204180202,
      "loss": 12.4461,
      "step": 5757
    },
    {
      "epoch": 1.0068633879781421,
      "grad_norm": 0.8012153428572004,
      "learning_rate": 0.00017530581558558402,
      "loss": 12.6926,
      "step": 5758
    },
    {
      "epoch": 1.0070382513661202,
      "grad_norm": 0.7575970062834946,
      "learning_rate": 0.00017529649761925317,
      "loss": 12.4709,
      "step": 5759
    },
    {
      "epoch": 1.0072131147540984,
      "grad_norm": 0.872691979519393,
      "learning_rate": 0.00017528717814299638,
      "loss": 12.6052,
      "step": 5760
    },
    {
      "epoch": 1.0073879781420765,
      "grad_norm": 0.7581084576624101,
      "learning_rate": 0.00017527785715700048,
      "loss": 12.5477,
      "step": 5761
    },
    {
      "epoch": 1.0075628415300546,
      "grad_norm": 0.8502675284271759,
      "learning_rate": 0.00017526853466145244,
      "loss": 12.5756,
      "step": 5762
    },
    {
      "epoch": 1.0077377049180327,
      "grad_norm": 0.8810639617958641,
      "learning_rate": 0.00017525921065653918,
      "loss": 12.6725,
      "step": 5763
    },
    {
      "epoch": 1.0079125683060108,
      "grad_norm": 0.8754278007996512,
      "learning_rate": 0.00017524988514244763,
      "loss": 12.741,
      "step": 5764
    },
    {
      "epoch": 1.0080874316939892,
      "grad_norm": 0.7683052674437576,
      "learning_rate": 0.0001752405581193649,
      "loss": 12.7249,
      "step": 5765
    },
    {
      "epoch": 1.0082622950819673,
      "grad_norm": 1.0596921398239059,
      "learning_rate": 0.00017523122958747794,
      "loss": 12.7571,
      "step": 5766
    },
    {
      "epoch": 1.0084371584699454,
      "grad_norm": 0.852578006515661,
      "learning_rate": 0.0001752218995469738,
      "loss": 12.4808,
      "step": 5767
    },
    {
      "epoch": 1.0086120218579235,
      "grad_norm": 0.9405791412083487,
      "learning_rate": 0.00017521256799803964,
      "loss": 12.6017,
      "step": 5768
    },
    {
      "epoch": 1.0087868852459017,
      "grad_norm": 0.9976530476777512,
      "learning_rate": 0.00017520323494086255,
      "loss": 12.6726,
      "step": 5769
    },
    {
      "epoch": 1.0089617486338798,
      "grad_norm": 0.7633417916697496,
      "learning_rate": 0.00017519390037562969,
      "loss": 12.7163,
      "step": 5770
    },
    {
      "epoch": 1.009136612021858,
      "grad_norm": 0.8567014236466253,
      "learning_rate": 0.00017518456430252828,
      "loss": 12.7269,
      "step": 5771
    },
    {
      "epoch": 1.009311475409836,
      "grad_norm": 0.7426648062625595,
      "learning_rate": 0.00017517522672174548,
      "loss": 12.8351,
      "step": 5772
    },
    {
      "epoch": 1.0094863387978141,
      "grad_norm": 0.7953626850118898,
      "learning_rate": 0.00017516588763346857,
      "loss": 12.81,
      "step": 5773
    },
    {
      "epoch": 1.0096612021857923,
      "grad_norm": 0.6867163545158297,
      "learning_rate": 0.0001751565470378848,
      "loss": 12.427,
      "step": 5774
    },
    {
      "epoch": 1.0098360655737706,
      "grad_norm": 0.7370398621805981,
      "learning_rate": 0.00017514720493518155,
      "loss": 12.7159,
      "step": 5775
    },
    {
      "epoch": 1.0100109289617487,
      "grad_norm": 0.757077669377326,
      "learning_rate": 0.0001751378613255461,
      "loss": 12.7321,
      "step": 5776
    },
    {
      "epoch": 1.0101857923497268,
      "grad_norm": 0.6951816140243183,
      "learning_rate": 0.0001751285162091658,
      "loss": 12.6165,
      "step": 5777
    },
    {
      "epoch": 1.010360655737705,
      "grad_norm": 0.7432646430549879,
      "learning_rate": 0.0001751191695862281,
      "loss": 12.8042,
      "step": 5778
    },
    {
      "epoch": 1.010535519125683,
      "grad_norm": 0.7262679160031349,
      "learning_rate": 0.0001751098214569204,
      "loss": 12.6526,
      "step": 5779
    },
    {
      "epoch": 1.0107103825136612,
      "grad_norm": 0.7271579533554039,
      "learning_rate": 0.00017510047182143015,
      "loss": 12.69,
      "step": 5780
    },
    {
      "epoch": 1.0108852459016393,
      "grad_norm": 0.8096724654529184,
      "learning_rate": 0.00017509112067994487,
      "loss": 12.4398,
      "step": 5781
    },
    {
      "epoch": 1.0110601092896174,
      "grad_norm": 0.7500287045365266,
      "learning_rate": 0.00017508176803265202,
      "loss": 12.6571,
      "step": 5782
    },
    {
      "epoch": 1.0112349726775955,
      "grad_norm": 0.8598559554173458,
      "learning_rate": 0.00017507241387973923,
      "loss": 12.9108,
      "step": 5783
    },
    {
      "epoch": 1.0114098360655737,
      "grad_norm": 0.7969138012351803,
      "learning_rate": 0.00017506305822139403,
      "loss": 12.8565,
      "step": 5784
    },
    {
      "epoch": 1.011584699453552,
      "grad_norm": 0.7988933368988116,
      "learning_rate": 0.00017505370105780403,
      "loss": 12.5605,
      "step": 5785
    },
    {
      "epoch": 1.0117595628415301,
      "grad_norm": 0.8558508859987637,
      "learning_rate": 0.0001750443423891569,
      "loss": 12.7193,
      "step": 5786
    },
    {
      "epoch": 1.0119344262295082,
      "grad_norm": 0.8186659582516154,
      "learning_rate": 0.00017503498221564025,
      "loss": 12.6245,
      "step": 5787
    },
    {
      "epoch": 1.0121092896174864,
      "grad_norm": 0.8918485890147584,
      "learning_rate": 0.00017502562053744184,
      "loss": 12.648,
      "step": 5788
    },
    {
      "epoch": 1.0122841530054645,
      "grad_norm": 0.7795282658071632,
      "learning_rate": 0.0001750162573547494,
      "loss": 12.8235,
      "step": 5789
    },
    {
      "epoch": 1.0124590163934426,
      "grad_norm": 0.7284346334812167,
      "learning_rate": 0.00017500689266775063,
      "loss": 12.5046,
      "step": 5790
    },
    {
      "epoch": 1.0126338797814207,
      "grad_norm": 0.723645118091975,
      "learning_rate": 0.0001749975264766334,
      "loss": 12.7058,
      "step": 5791
    },
    {
      "epoch": 1.0128087431693988,
      "grad_norm": 0.8096522202380577,
      "learning_rate": 0.00017498815878158545,
      "loss": 12.634,
      "step": 5792
    },
    {
      "epoch": 1.012983606557377,
      "grad_norm": 0.7903736869668194,
      "learning_rate": 0.0001749787895827947,
      "loss": 12.7715,
      "step": 5793
    },
    {
      "epoch": 1.0131584699453553,
      "grad_norm": 0.8058263393919567,
      "learning_rate": 0.000174969418880449,
      "loss": 12.6695,
      "step": 5794
    },
    {
      "epoch": 1.0133333333333334,
      "grad_norm": 0.8554569001115018,
      "learning_rate": 0.00017496004667473628,
      "loss": 12.8112,
      "step": 5795
    },
    {
      "epoch": 1.0135081967213115,
      "grad_norm": 0.776695446913879,
      "learning_rate": 0.00017495067296584442,
      "loss": 12.7142,
      "step": 5796
    },
    {
      "epoch": 1.0136830601092897,
      "grad_norm": 0.7774654181460082,
      "learning_rate": 0.00017494129775396148,
      "loss": 12.5731,
      "step": 5797
    },
    {
      "epoch": 1.0138579234972678,
      "grad_norm": 0.759380060222598,
      "learning_rate": 0.00017493192103927542,
      "loss": 12.6121,
      "step": 5798
    },
    {
      "epoch": 1.014032786885246,
      "grad_norm": 0.7930956936277214,
      "learning_rate": 0.00017492254282197424,
      "loss": 12.8214,
      "step": 5799
    },
    {
      "epoch": 1.014207650273224,
      "grad_norm": 0.7721594593064356,
      "learning_rate": 0.00017491316310224605,
      "loss": 12.7394,
      "step": 5800
    },
    {
      "epoch": 1.0143825136612021,
      "grad_norm": 0.8063512322832163,
      "learning_rate": 0.00017490378188027893,
      "loss": 12.7063,
      "step": 5801
    },
    {
      "epoch": 1.0145573770491803,
      "grad_norm": 0.8510965028446239,
      "learning_rate": 0.000174894399156261,
      "loss": 12.5114,
      "step": 5802
    },
    {
      "epoch": 1.0147322404371584,
      "grad_norm": 0.8319845206264648,
      "learning_rate": 0.00017488501493038042,
      "loss": 12.415,
      "step": 5803
    },
    {
      "epoch": 1.0149071038251367,
      "grad_norm": 0.7188809005538768,
      "learning_rate": 0.00017487562920282535,
      "loss": 12.7064,
      "step": 5804
    },
    {
      "epoch": 1.0150819672131148,
      "grad_norm": 0.8436561683658019,
      "learning_rate": 0.00017486624197378403,
      "loss": 12.6442,
      "step": 5805
    },
    {
      "epoch": 1.015256830601093,
      "grad_norm": 0.7604567873606987,
      "learning_rate": 0.00017485685324344468,
      "loss": 12.4662,
      "step": 5806
    },
    {
      "epoch": 1.015431693989071,
      "grad_norm": 0.8498975755825149,
      "learning_rate": 0.00017484746301199558,
      "loss": 12.7857,
      "step": 5807
    },
    {
      "epoch": 1.0156065573770492,
      "grad_norm": 0.745590887153168,
      "learning_rate": 0.00017483807127962502,
      "loss": 12.7933,
      "step": 5808
    },
    {
      "epoch": 1.0157814207650273,
      "grad_norm": 0.8742947091134752,
      "learning_rate": 0.00017482867804652136,
      "loss": 12.46,
      "step": 5809
    },
    {
      "epoch": 1.0159562841530054,
      "grad_norm": 0.7381554951810624,
      "learning_rate": 0.00017481928331287297,
      "loss": 12.6079,
      "step": 5810
    },
    {
      "epoch": 1.0161311475409835,
      "grad_norm": 0.8685501608067978,
      "learning_rate": 0.00017480988707886818,
      "loss": 12.5509,
      "step": 5811
    },
    {
      "epoch": 1.0163060109289617,
      "grad_norm": 0.7931538408509115,
      "learning_rate": 0.0001748004893446955,
      "loss": 12.7415,
      "step": 5812
    },
    {
      "epoch": 1.0164808743169398,
      "grad_norm": 0.9145736341819912,
      "learning_rate": 0.00017479109011054334,
      "loss": 12.6227,
      "step": 5813
    },
    {
      "epoch": 1.0166557377049181,
      "grad_norm": 0.8260300162576291,
      "learning_rate": 0.00017478168937660015,
      "loss": 12.1917,
      "step": 5814
    },
    {
      "epoch": 1.0168306010928962,
      "grad_norm": 0.8279907732276787,
      "learning_rate": 0.00017477228714305452,
      "loss": 12.4044,
      "step": 5815
    },
    {
      "epoch": 1.0170054644808744,
      "grad_norm": 0.9216640737308172,
      "learning_rate": 0.00017476288341009493,
      "loss": 12.4147,
      "step": 5816
    },
    {
      "epoch": 1.0171803278688525,
      "grad_norm": 0.8352084894640492,
      "learning_rate": 0.00017475347817790996,
      "loss": 12.6735,
      "step": 5817
    },
    {
      "epoch": 1.0173551912568306,
      "grad_norm": 1.0087289306528828,
      "learning_rate": 0.00017474407144668825,
      "loss": 12.8274,
      "step": 5818
    },
    {
      "epoch": 1.0175300546448087,
      "grad_norm": 0.754719669493696,
      "learning_rate": 0.0001747346632166184,
      "loss": 12.4074,
      "step": 5819
    },
    {
      "epoch": 1.0177049180327868,
      "grad_norm": 0.9620532077723984,
      "learning_rate": 0.0001747252534878891,
      "loss": 12.7374,
      "step": 5820
    },
    {
      "epoch": 1.017879781420765,
      "grad_norm": 0.80597208024182,
      "learning_rate": 0.00017471584226068902,
      "loss": 12.8139,
      "step": 5821
    },
    {
      "epoch": 1.018054644808743,
      "grad_norm": 0.8770888089160962,
      "learning_rate": 0.0001747064295352069,
      "loss": 12.8702,
      "step": 5822
    },
    {
      "epoch": 1.0182295081967214,
      "grad_norm": 0.760367815155666,
      "learning_rate": 0.0001746970153116315,
      "loss": 12.6591,
      "step": 5823
    },
    {
      "epoch": 1.0184043715846995,
      "grad_norm": 0.9582168840636849,
      "learning_rate": 0.00017468759959015157,
      "loss": 12.6807,
      "step": 5824
    },
    {
      "epoch": 1.0185792349726777,
      "grad_norm": 0.7970198227842394,
      "learning_rate": 0.00017467818237095598,
      "loss": 12.6477,
      "step": 5825
    },
    {
      "epoch": 1.0187540983606558,
      "grad_norm": 0.800691732710943,
      "learning_rate": 0.0001746687636542335,
      "loss": 12.7098,
      "step": 5826
    },
    {
      "epoch": 1.018928961748634,
      "grad_norm": 0.7621960435969929,
      "learning_rate": 0.0001746593434401731,
      "loss": 12.5049,
      "step": 5827
    },
    {
      "epoch": 1.019103825136612,
      "grad_norm": 0.8589497271118892,
      "learning_rate": 0.00017464992172896358,
      "loss": 12.6172,
      "step": 5828
    },
    {
      "epoch": 1.0192786885245901,
      "grad_norm": 0.743033281740385,
      "learning_rate": 0.00017464049852079395,
      "loss": 12.8587,
      "step": 5829
    },
    {
      "epoch": 1.0194535519125683,
      "grad_norm": 0.8716552391858612,
      "learning_rate": 0.00017463107381585313,
      "loss": 12.5668,
      "step": 5830
    },
    {
      "epoch": 1.0196284153005464,
      "grad_norm": 0.8014389423824879,
      "learning_rate": 0.00017462164761433013,
      "loss": 12.7877,
      "step": 5831
    },
    {
      "epoch": 1.0198032786885245,
      "grad_norm": 0.8548033920751519,
      "learning_rate": 0.000174612219916414,
      "loss": 12.6113,
      "step": 5832
    },
    {
      "epoch": 1.0199781420765028,
      "grad_norm": 0.8476090508765258,
      "learning_rate": 0.00017460279072229376,
      "loss": 12.6829,
      "step": 5833
    },
    {
      "epoch": 1.020153005464481,
      "grad_norm": 0.8147009074632564,
      "learning_rate": 0.0001745933600321585,
      "loss": 12.8237,
      "step": 5834
    },
    {
      "epoch": 1.020327868852459,
      "grad_norm": 0.8881181114739797,
      "learning_rate": 0.00017458392784619735,
      "loss": 12.5451,
      "step": 5835
    },
    {
      "epoch": 1.0205027322404372,
      "grad_norm": 0.8840735482716507,
      "learning_rate": 0.00017457449416459944,
      "loss": 12.645,
      "step": 5836
    },
    {
      "epoch": 1.0206775956284153,
      "grad_norm": 0.7973815565083106,
      "learning_rate": 0.00017456505898755394,
      "loss": 12.702,
      "step": 5837
    },
    {
      "epoch": 1.0208524590163934,
      "grad_norm": 0.9369752887421529,
      "learning_rate": 0.00017455562231525007,
      "loss": 12.7851,
      "step": 5838
    },
    {
      "epoch": 1.0210273224043716,
      "grad_norm": 0.816378067358125,
      "learning_rate": 0.00017454618414787707,
      "loss": 12.7989,
      "step": 5839
    },
    {
      "epoch": 1.0212021857923497,
      "grad_norm": 0.7727110794343124,
      "learning_rate": 0.00017453674448562415,
      "loss": 12.6029,
      "step": 5840
    },
    {
      "epoch": 1.0213770491803278,
      "grad_norm": 0.7247691171147901,
      "learning_rate": 0.00017452730332868067,
      "loss": 12.3305,
      "step": 5841
    },
    {
      "epoch": 1.021551912568306,
      "grad_norm": 0.8597219308510134,
      "learning_rate": 0.00017451786067723595,
      "loss": 12.2948,
      "step": 5842
    },
    {
      "epoch": 1.0217267759562842,
      "grad_norm": 0.7638910627069836,
      "learning_rate": 0.0001745084165314793,
      "loss": 12.605,
      "step": 5843
    },
    {
      "epoch": 1.0219016393442624,
      "grad_norm": 0.9122052261237733,
      "learning_rate": 0.00017449897089160014,
      "loss": 12.4096,
      "step": 5844
    },
    {
      "epoch": 1.0220765027322405,
      "grad_norm": 0.7392894055540361,
      "learning_rate": 0.00017448952375778788,
      "loss": 12.7838,
      "step": 5845
    },
    {
      "epoch": 1.0222513661202186,
      "grad_norm": 0.9315789299965389,
      "learning_rate": 0.00017448007513023192,
      "loss": 12.5635,
      "step": 5846
    },
    {
      "epoch": 1.0224262295081967,
      "grad_norm": 0.7640954078295716,
      "learning_rate": 0.0001744706250091218,
      "loss": 12.5622,
      "step": 5847
    },
    {
      "epoch": 1.0226010928961748,
      "grad_norm": 0.8610244830889358,
      "learning_rate": 0.00017446117339464697,
      "loss": 12.6503,
      "step": 5848
    },
    {
      "epoch": 1.022775956284153,
      "grad_norm": 0.7889011189194142,
      "learning_rate": 0.000174451720286997,
      "loss": 12.585,
      "step": 5849
    },
    {
      "epoch": 1.022950819672131,
      "grad_norm": 0.8063925674661245,
      "learning_rate": 0.00017444226568636144,
      "loss": 12.486,
      "step": 5850
    },
    {
      "epoch": 1.0231256830601092,
      "grad_norm": 0.865913676512027,
      "learning_rate": 0.00017443280959292989,
      "loss": 12.7232,
      "step": 5851
    },
    {
      "epoch": 1.0233005464480873,
      "grad_norm": 0.7775925817097361,
      "learning_rate": 0.00017442335200689197,
      "loss": 12.5536,
      "step": 5852
    },
    {
      "epoch": 1.0234754098360657,
      "grad_norm": 0.9555282844583275,
      "learning_rate": 0.00017441389292843733,
      "loss": 12.6565,
      "step": 5853
    },
    {
      "epoch": 1.0236502732240438,
      "grad_norm": 0.7970750945225628,
      "learning_rate": 0.00017440443235775566,
      "loss": 12.4448,
      "step": 5854
    },
    {
      "epoch": 1.023825136612022,
      "grad_norm": 0.9714324426337491,
      "learning_rate": 0.00017439497029503667,
      "loss": 12.6848,
      "step": 5855
    },
    {
      "epoch": 1.024,
      "grad_norm": 0.7915079823084379,
      "learning_rate": 0.0001743855067404701,
      "loss": 12.5861,
      "step": 5856
    },
    {
      "epoch": 1.0241748633879781,
      "grad_norm": 1.013180601092864,
      "learning_rate": 0.0001743760416942457,
      "loss": 12.6476,
      "step": 5857
    },
    {
      "epoch": 1.0243497267759563,
      "grad_norm": 0.7251985772131303,
      "learning_rate": 0.00017436657515655334,
      "loss": 12.6241,
      "step": 5858
    },
    {
      "epoch": 1.0245245901639344,
      "grad_norm": 0.9748552912289068,
      "learning_rate": 0.0001743571071275828,
      "loss": 12.8528,
      "step": 5859
    },
    {
      "epoch": 1.0246994535519125,
      "grad_norm": 0.7560378708310893,
      "learning_rate": 0.00017434763760752395,
      "loss": 12.6369,
      "step": 5860
    },
    {
      "epoch": 1.0248743169398906,
      "grad_norm": 0.9320042106665081,
      "learning_rate": 0.0001743381665965667,
      "loss": 12.6404,
      "step": 5861
    },
    {
      "epoch": 1.025049180327869,
      "grad_norm": 0.8051016109560906,
      "learning_rate": 0.00017432869409490093,
      "loss": 12.5983,
      "step": 5862
    },
    {
      "epoch": 1.025224043715847,
      "grad_norm": 0.8252287628025674,
      "learning_rate": 0.00017431922010271666,
      "loss": 12.3526,
      "step": 5863
    },
    {
      "epoch": 1.0253989071038252,
      "grad_norm": 0.8170841178603923,
      "learning_rate": 0.0001743097446202038,
      "loss": 12.3033,
      "step": 5864
    },
    {
      "epoch": 1.0255737704918033,
      "grad_norm": 0.8713040793737963,
      "learning_rate": 0.00017430026764755243,
      "loss": 12.5739,
      "step": 5865
    },
    {
      "epoch": 1.0257486338797814,
      "grad_norm": 0.7891570933198264,
      "learning_rate": 0.00017429078918495256,
      "loss": 12.4459,
      "step": 5866
    },
    {
      "epoch": 1.0259234972677596,
      "grad_norm": 0.7858909139399005,
      "learning_rate": 0.00017428130923259426,
      "loss": 12.615,
      "step": 5867
    },
    {
      "epoch": 1.0260983606557377,
      "grad_norm": 1.0335925926968974,
      "learning_rate": 0.00017427182779066765,
      "loss": 12.7806,
      "step": 5868
    },
    {
      "epoch": 1.0262732240437158,
      "grad_norm": 0.8359610073590106,
      "learning_rate": 0.00017426234485936283,
      "loss": 12.715,
      "step": 5869
    },
    {
      "epoch": 1.026448087431694,
      "grad_norm": 0.9140179269580125,
      "learning_rate": 0.00017425286043886999,
      "loss": 12.5212,
      "step": 5870
    },
    {
      "epoch": 1.026622950819672,
      "grad_norm": 0.7973184501235759,
      "learning_rate": 0.0001742433745293793,
      "loss": 12.6947,
      "step": 5871
    },
    {
      "epoch": 1.0267978142076504,
      "grad_norm": 0.8017834629473668,
      "learning_rate": 0.000174233887131081,
      "loss": 12.3485,
      "step": 5872
    },
    {
      "epoch": 1.0269726775956285,
      "grad_norm": 0.7467343614590819,
      "learning_rate": 0.00017422439824416535,
      "loss": 12.7989,
      "step": 5873
    },
    {
      "epoch": 1.0271475409836066,
      "grad_norm": 0.7292195981584098,
      "learning_rate": 0.00017421490786882258,
      "loss": 12.5237,
      "step": 5874
    },
    {
      "epoch": 1.0273224043715847,
      "grad_norm": 0.7586097753104919,
      "learning_rate": 0.00017420541600524306,
      "loss": 12.5551,
      "step": 5875
    },
    {
      "epoch": 1.0274972677595628,
      "grad_norm": 0.8117792818021391,
      "learning_rate": 0.00017419592265361712,
      "loss": 12.6648,
      "step": 5876
    },
    {
      "epoch": 1.027672131147541,
      "grad_norm": 0.7838967490835042,
      "learning_rate": 0.0001741864278141351,
      "loss": 12.5323,
      "step": 5877
    },
    {
      "epoch": 1.027846994535519,
      "grad_norm": 0.7215911457821771,
      "learning_rate": 0.00017417693148698743,
      "loss": 12.4909,
      "step": 5878
    },
    {
      "epoch": 1.0280218579234972,
      "grad_norm": 0.7162560045232093,
      "learning_rate": 0.00017416743367236451,
      "loss": 12.5049,
      "step": 5879
    },
    {
      "epoch": 1.0281967213114753,
      "grad_norm": 0.7067983573051885,
      "learning_rate": 0.00017415793437045685,
      "loss": 12.4221,
      "step": 5880
    },
    {
      "epoch": 1.0283715846994534,
      "grad_norm": 0.7060959117431808,
      "learning_rate": 0.0001741484335814549,
      "loss": 12.5157,
      "step": 5881
    },
    {
      "epoch": 1.0285464480874318,
      "grad_norm": 0.7827374155939155,
      "learning_rate": 0.0001741389313055492,
      "loss": 12.7573,
      "step": 5882
    },
    {
      "epoch": 1.02872131147541,
      "grad_norm": 0.7513326249972749,
      "learning_rate": 0.00017412942754293028,
      "loss": 12.236,
      "step": 5883
    },
    {
      "epoch": 1.028896174863388,
      "grad_norm": 0.8470551099512879,
      "learning_rate": 0.00017411992229378874,
      "loss": 12.6242,
      "step": 5884
    },
    {
      "epoch": 1.0290710382513661,
      "grad_norm": 0.7610202659641703,
      "learning_rate": 0.00017411041555831516,
      "loss": 12.6176,
      "step": 5885
    },
    {
      "epoch": 1.0292459016393443,
      "grad_norm": 0.7272533607410921,
      "learning_rate": 0.00017410090733670022,
      "loss": 12.3516,
      "step": 5886
    },
    {
      "epoch": 1.0294207650273224,
      "grad_norm": 0.8665349079254955,
      "learning_rate": 0.00017409139762913456,
      "loss": 12.8115,
      "step": 5887
    },
    {
      "epoch": 1.0295956284153005,
      "grad_norm": 0.781033347849837,
      "learning_rate": 0.00017408188643580887,
      "loss": 12.6506,
      "step": 5888
    },
    {
      "epoch": 1.0297704918032786,
      "grad_norm": 0.7881294417108621,
      "learning_rate": 0.00017407237375691392,
      "loss": 12.6093,
      "step": 5889
    },
    {
      "epoch": 1.0299453551912567,
      "grad_norm": 0.7315173071613379,
      "learning_rate": 0.00017406285959264045,
      "loss": 12.3135,
      "step": 5890
    },
    {
      "epoch": 1.030120218579235,
      "grad_norm": 0.7200424779792685,
      "learning_rate": 0.00017405334394317922,
      "loss": 12.6754,
      "step": 5891
    },
    {
      "epoch": 1.0302950819672132,
      "grad_norm": 0.7486084287341519,
      "learning_rate": 0.00017404382680872107,
      "loss": 12.6741,
      "step": 5892
    },
    {
      "epoch": 1.0304699453551913,
      "grad_norm": 0.7131011120218554,
      "learning_rate": 0.00017403430818945689,
      "loss": 12.6047,
      "step": 5893
    },
    {
      "epoch": 1.0306448087431694,
      "grad_norm": 0.7614683129250748,
      "learning_rate": 0.00017402478808557748,
      "loss": 12.6842,
      "step": 5894
    },
    {
      "epoch": 1.0308196721311476,
      "grad_norm": 0.7679970185486343,
      "learning_rate": 0.00017401526649727376,
      "loss": 12.5204,
      "step": 5895
    },
    {
      "epoch": 1.0309945355191257,
      "grad_norm": 0.7765908753573436,
      "learning_rate": 0.00017400574342473673,
      "loss": 12.4423,
      "step": 5896
    },
    {
      "epoch": 1.0311693989071038,
      "grad_norm": 0.7278127822081543,
      "learning_rate": 0.0001739962188681573,
      "loss": 12.6957,
      "step": 5897
    },
    {
      "epoch": 1.031344262295082,
      "grad_norm": 0.7620991606306775,
      "learning_rate": 0.00017398669282772645,
      "loss": 12.6175,
      "step": 5898
    },
    {
      "epoch": 1.03151912568306,
      "grad_norm": 0.8292783347155848,
      "learning_rate": 0.0001739771653036353,
      "loss": 12.5814,
      "step": 5899
    },
    {
      "epoch": 1.0316939890710382,
      "grad_norm": 0.7188015824129543,
      "learning_rate": 0.00017396763629607483,
      "loss": 12.3196,
      "step": 5900
    },
    {
      "epoch": 1.0318688524590165,
      "grad_norm": 0.8215785091729181,
      "learning_rate": 0.00017395810580523612,
      "loss": 12.8058,
      "step": 5901
    },
    {
      "epoch": 1.0320437158469946,
      "grad_norm": 0.8046374400201151,
      "learning_rate": 0.00017394857383131034,
      "loss": 12.6458,
      "step": 5902
    },
    {
      "epoch": 1.0322185792349727,
      "grad_norm": 0.7181869812783378,
      "learning_rate": 0.00017393904037448857,
      "loss": 12.4234,
      "step": 5903
    },
    {
      "epoch": 1.0323934426229509,
      "grad_norm": 0.8420842740121928,
      "learning_rate": 0.00017392950543496203,
      "loss": 12.5566,
      "step": 5904
    },
    {
      "epoch": 1.032568306010929,
      "grad_norm": 0.7596097077855617,
      "learning_rate": 0.0001739199690129219,
      "loss": 12.4921,
      "step": 5905
    },
    {
      "epoch": 1.032743169398907,
      "grad_norm": 0.7750142731721754,
      "learning_rate": 0.00017391043110855943,
      "loss": 12.697,
      "step": 5906
    },
    {
      "epoch": 1.0329180327868852,
      "grad_norm": 0.7817317344445401,
      "learning_rate": 0.00017390089172206592,
      "loss": 12.6092,
      "step": 5907
    },
    {
      "epoch": 1.0330928961748633,
      "grad_norm": 1.0208041617274888,
      "learning_rate": 0.0001738913508536326,
      "loss": 12.5752,
      "step": 5908
    },
    {
      "epoch": 1.0332677595628414,
      "grad_norm": 0.7733601689841524,
      "learning_rate": 0.00017388180850345082,
      "loss": 12.746,
      "step": 5909
    },
    {
      "epoch": 1.0334426229508196,
      "grad_norm": 0.8664900732214877,
      "learning_rate": 0.00017387226467171192,
      "loss": 12.6128,
      "step": 5910
    },
    {
      "epoch": 1.033617486338798,
      "grad_norm": 0.8117414749899752,
      "learning_rate": 0.00017386271935860732,
      "loss": 12.7345,
      "step": 5911
    },
    {
      "epoch": 1.033792349726776,
      "grad_norm": 0.7958431690983019,
      "learning_rate": 0.00017385317256432838,
      "loss": 12.7281,
      "step": 5912
    },
    {
      "epoch": 1.0339672131147541,
      "grad_norm": 0.7866756434429243,
      "learning_rate": 0.0001738436242890666,
      "loss": 12.4834,
      "step": 5913
    },
    {
      "epoch": 1.0341420765027323,
      "grad_norm": 0.7972057462113011,
      "learning_rate": 0.0001738340745330134,
      "loss": 12.6817,
      "step": 5914
    },
    {
      "epoch": 1.0343169398907104,
      "grad_norm": 0.7756917123502859,
      "learning_rate": 0.0001738245232963603,
      "loss": 12.6552,
      "step": 5915
    },
    {
      "epoch": 1.0344918032786885,
      "grad_norm": 0.9937191112201141,
      "learning_rate": 0.00017381497057929884,
      "loss": 12.7473,
      "step": 5916
    },
    {
      "epoch": 1.0346666666666666,
      "grad_norm": 0.8395205632168702,
      "learning_rate": 0.00017380541638202063,
      "loss": 12.879,
      "step": 5917
    },
    {
      "epoch": 1.0348415300546447,
      "grad_norm": 0.7474225649627796,
      "learning_rate": 0.00017379586070471716,
      "loss": 12.2019,
      "step": 5918
    },
    {
      "epoch": 1.0350163934426229,
      "grad_norm": 0.7505799828973203,
      "learning_rate": 0.0001737863035475801,
      "loss": 12.7025,
      "step": 5919
    },
    {
      "epoch": 1.035191256830601,
      "grad_norm": 0.7822718547678064,
      "learning_rate": 0.00017377674491080112,
      "loss": 12.6051,
      "step": 5920
    },
    {
      "epoch": 1.0353661202185793,
      "grad_norm": 0.8447305419560106,
      "learning_rate": 0.00017376718479457186,
      "loss": 12.6805,
      "step": 5921
    },
    {
      "epoch": 1.0355409836065574,
      "grad_norm": 0.8699168035518614,
      "learning_rate": 0.00017375762319908405,
      "loss": 12.5884,
      "step": 5922
    },
    {
      "epoch": 1.0357158469945356,
      "grad_norm": 0.8491971377997631,
      "learning_rate": 0.00017374806012452945,
      "loss": 12.6497,
      "step": 5923
    },
    {
      "epoch": 1.0358907103825137,
      "grad_norm": 1.0753298909299964,
      "learning_rate": 0.0001737384955710998,
      "loss": 12.7606,
      "step": 5924
    },
    {
      "epoch": 1.0360655737704918,
      "grad_norm": 0.7821540914467118,
      "learning_rate": 0.00017372892953898688,
      "loss": 12.5523,
      "step": 5925
    },
    {
      "epoch": 1.03624043715847,
      "grad_norm": 1.0807056028434936,
      "learning_rate": 0.00017371936202838258,
      "loss": 12.6617,
      "step": 5926
    },
    {
      "epoch": 1.036415300546448,
      "grad_norm": 0.8686006179045991,
      "learning_rate": 0.0001737097930394787,
      "loss": 12.5906,
      "step": 5927
    },
    {
      "epoch": 1.0365901639344262,
      "grad_norm": 0.8524318183539497,
      "learning_rate": 0.00017370022257246715,
      "loss": 12.7549,
      "step": 5928
    },
    {
      "epoch": 1.0367650273224043,
      "grad_norm": 0.894145890664334,
      "learning_rate": 0.00017369065062753988,
      "loss": 12.1317,
      "step": 5929
    },
    {
      "epoch": 1.0369398907103826,
      "grad_norm": 0.8092371371161828,
      "learning_rate": 0.00017368107720488877,
      "loss": 12.4371,
      "step": 5930
    },
    {
      "epoch": 1.0371147540983607,
      "grad_norm": 0.7939891340226407,
      "learning_rate": 0.00017367150230470586,
      "loss": 12.6587,
      "step": 5931
    },
    {
      "epoch": 1.0372896174863389,
      "grad_norm": 0.746172774318381,
      "learning_rate": 0.0001736619259271831,
      "loss": 12.4886,
      "step": 5932
    },
    {
      "epoch": 1.037464480874317,
      "grad_norm": 0.8138313190757037,
      "learning_rate": 0.00017365234807251257,
      "loss": 12.6145,
      "step": 5933
    },
    {
      "epoch": 1.037639344262295,
      "grad_norm": 0.7747202266509673,
      "learning_rate": 0.00017364276874088633,
      "loss": 12.6101,
      "step": 5934
    },
    {
      "epoch": 1.0378142076502732,
      "grad_norm": 0.7582537136880332,
      "learning_rate": 0.00017363318793249644,
      "loss": 12.6917,
      "step": 5935
    },
    {
      "epoch": 1.0379890710382513,
      "grad_norm": 0.7480525495123891,
      "learning_rate": 0.00017362360564753505,
      "loss": 12.7625,
      "step": 5936
    },
    {
      "epoch": 1.0381639344262295,
      "grad_norm": 0.7828227529994237,
      "learning_rate": 0.00017361402188619432,
      "loss": 12.7594,
      "step": 5937
    },
    {
      "epoch": 1.0383387978142076,
      "grad_norm": 0.6896216134843727,
      "learning_rate": 0.00017360443664866644,
      "loss": 12.6063,
      "step": 5938
    },
    {
      "epoch": 1.0385136612021857,
      "grad_norm": 0.7560747538624719,
      "learning_rate": 0.0001735948499351436,
      "loss": 12.7494,
      "step": 5939
    },
    {
      "epoch": 1.038688524590164,
      "grad_norm": 0.8106404757761243,
      "learning_rate": 0.00017358526174581805,
      "loss": 12.3574,
      "step": 5940
    },
    {
      "epoch": 1.0388633879781421,
      "grad_norm": 0.7150528777729361,
      "learning_rate": 0.00017357567208088205,
      "loss": 12.4321,
      "step": 5941
    },
    {
      "epoch": 1.0390382513661203,
      "grad_norm": 0.789402058609001,
      "learning_rate": 0.00017356608094052796,
      "loss": 12.7588,
      "step": 5942
    },
    {
      "epoch": 1.0392131147540984,
      "grad_norm": 0.7739366407701811,
      "learning_rate": 0.00017355648832494803,
      "loss": 12.6402,
      "step": 5943
    },
    {
      "epoch": 1.0393879781420765,
      "grad_norm": 0.7228968386630885,
      "learning_rate": 0.00017354689423433465,
      "loss": 12.5385,
      "step": 5944
    },
    {
      "epoch": 1.0395628415300546,
      "grad_norm": 0.7740345598150197,
      "learning_rate": 0.00017353729866888027,
      "loss": 12.6869,
      "step": 5945
    },
    {
      "epoch": 1.0397377049180327,
      "grad_norm": 0.7620585148472794,
      "learning_rate": 0.00017352770162877722,
      "loss": 12.6679,
      "step": 5946
    },
    {
      "epoch": 1.0399125683060109,
      "grad_norm": 0.9095699353349319,
      "learning_rate": 0.00017351810311421804,
      "loss": 12.762,
      "step": 5947
    },
    {
      "epoch": 1.040087431693989,
      "grad_norm": 0.8101355649216578,
      "learning_rate": 0.00017350850312539514,
      "loss": 12.55,
      "step": 5948
    },
    {
      "epoch": 1.040262295081967,
      "grad_norm": 0.867361365123829,
      "learning_rate": 0.000173498901662501,
      "loss": 12.6911,
      "step": 5949
    },
    {
      "epoch": 1.0404371584699454,
      "grad_norm": 0.9233963616227859,
      "learning_rate": 0.00017348929872572827,
      "loss": 12.4963,
      "step": 5950
    },
    {
      "epoch": 1.0406120218579236,
      "grad_norm": 0.7105580242642258,
      "learning_rate": 0.00017347969431526946,
      "loss": 12.4252,
      "step": 5951
    },
    {
      "epoch": 1.0407868852459017,
      "grad_norm": 0.8183745897858132,
      "learning_rate": 0.00017347008843131712,
      "loss": 12.612,
      "step": 5952
    },
    {
      "epoch": 1.0409617486338798,
      "grad_norm": 0.7006514102467257,
      "learning_rate": 0.00017346048107406395,
      "loss": 12.7703,
      "step": 5953
    },
    {
      "epoch": 1.041136612021858,
      "grad_norm": 0.7547333375037817,
      "learning_rate": 0.0001734508722437026,
      "loss": 12.7783,
      "step": 5954
    },
    {
      "epoch": 1.041311475409836,
      "grad_norm": 0.7478191344484312,
      "learning_rate": 0.0001734412619404257,
      "loss": 12.563,
      "step": 5955
    },
    {
      "epoch": 1.0414863387978142,
      "grad_norm": 0.80139759780516,
      "learning_rate": 0.00017343165016442605,
      "loss": 12.4791,
      "step": 5956
    },
    {
      "epoch": 1.0416612021857923,
      "grad_norm": 0.7829321014144238,
      "learning_rate": 0.00017342203691589632,
      "loss": 12.8179,
      "step": 5957
    },
    {
      "epoch": 1.0418360655737704,
      "grad_norm": 1.129935771625616,
      "learning_rate": 0.00017341242219502933,
      "loss": 12.7517,
      "step": 5958
    },
    {
      "epoch": 1.0420109289617487,
      "grad_norm": 0.7682509008154781,
      "learning_rate": 0.00017340280600201786,
      "loss": 12.4527,
      "step": 5959
    },
    {
      "epoch": 1.0421857923497269,
      "grad_norm": 0.8792199532073989,
      "learning_rate": 0.00017339318833705476,
      "loss": 12.5638,
      "step": 5960
    },
    {
      "epoch": 1.042360655737705,
      "grad_norm": 0.7302852652329549,
      "learning_rate": 0.0001733835692003329,
      "loss": 12.6072,
      "step": 5961
    },
    {
      "epoch": 1.042535519125683,
      "grad_norm": 1.1244014319445335,
      "learning_rate": 0.00017337394859204513,
      "loss": 12.7405,
      "step": 5962
    },
    {
      "epoch": 1.0427103825136612,
      "grad_norm": 0.7734639734361338,
      "learning_rate": 0.00017336432651238442,
      "loss": 12.4167,
      "step": 5963
    },
    {
      "epoch": 1.0428852459016393,
      "grad_norm": 1.178810753057323,
      "learning_rate": 0.00017335470296154368,
      "loss": 12.713,
      "step": 5964
    },
    {
      "epoch": 1.0430601092896175,
      "grad_norm": 0.8289531582301636,
      "learning_rate": 0.00017334507793971592,
      "loss": 12.5241,
      "step": 5965
    },
    {
      "epoch": 1.0432349726775956,
      "grad_norm": 1.2785167293479704,
      "learning_rate": 0.00017333545144709416,
      "loss": 12.732,
      "step": 5966
    },
    {
      "epoch": 1.0434098360655737,
      "grad_norm": 1.2380331771720035,
      "learning_rate": 0.00017332582348387145,
      "loss": 12.7669,
      "step": 5967
    },
    {
      "epoch": 1.0435846994535518,
      "grad_norm": 0.9000429709550213,
      "learning_rate": 0.00017331619405024085,
      "loss": 12.7973,
      "step": 5968
    },
    {
      "epoch": 1.0437595628415302,
      "grad_norm": 1.0577383052840155,
      "learning_rate": 0.0001733065631463954,
      "loss": 12.4005,
      "step": 5969
    },
    {
      "epoch": 1.0439344262295083,
      "grad_norm": 0.9083187332405555,
      "learning_rate": 0.0001732969307725283,
      "loss": 12.746,
      "step": 5970
    },
    {
      "epoch": 1.0441092896174864,
      "grad_norm": 1.0724825884653808,
      "learning_rate": 0.0001732872969288327,
      "loss": 12.8135,
      "step": 5971
    },
    {
      "epoch": 1.0442841530054645,
      "grad_norm": 0.8164783464321109,
      "learning_rate": 0.00017327766161550175,
      "loss": 12.6062,
      "step": 5972
    },
    {
      "epoch": 1.0444590163934426,
      "grad_norm": 1.0163588259384617,
      "learning_rate": 0.0001732680248327287,
      "loss": 12.5188,
      "step": 5973
    },
    {
      "epoch": 1.0446338797814207,
      "grad_norm": 0.8381093410890034,
      "learning_rate": 0.00017325838658070682,
      "loss": 12.5087,
      "step": 5974
    },
    {
      "epoch": 1.0448087431693989,
      "grad_norm": 0.957396571127119,
      "learning_rate": 0.00017324874685962932,
      "loss": 12.5488,
      "step": 5975
    },
    {
      "epoch": 1.044983606557377,
      "grad_norm": 0.7838124469806494,
      "learning_rate": 0.00017323910566968955,
      "loss": 12.9149,
      "step": 5976
    },
    {
      "epoch": 1.045158469945355,
      "grad_norm": 0.9292769328951308,
      "learning_rate": 0.00017322946301108084,
      "loss": 12.8313,
      "step": 5977
    },
    {
      "epoch": 1.0453333333333332,
      "grad_norm": 0.804851023775395,
      "learning_rate": 0.00017321981888399654,
      "loss": 12.6801,
      "step": 5978
    },
    {
      "epoch": 1.0455081967213116,
      "grad_norm": 0.7873368896882823,
      "learning_rate": 0.00017321017328863009,
      "loss": 12.4425,
      "step": 5979
    },
    {
      "epoch": 1.0456830601092897,
      "grad_norm": 0.7739009496640414,
      "learning_rate": 0.00017320052622517485,
      "loss": 12.9232,
      "step": 5980
    },
    {
      "epoch": 1.0458579234972678,
      "grad_norm": 0.8877993882968974,
      "learning_rate": 0.00017319087769382434,
      "loss": 12.5269,
      "step": 5981
    },
    {
      "epoch": 1.046032786885246,
      "grad_norm": 0.7440048917504085,
      "learning_rate": 0.00017318122769477195,
      "loss": 12.5806,
      "step": 5982
    },
    {
      "epoch": 1.046207650273224,
      "grad_norm": 0.775398103890312,
      "learning_rate": 0.0001731715762282113,
      "loss": 12.4517,
      "step": 5983
    },
    {
      "epoch": 1.0463825136612022,
      "grad_norm": 0.7354336120222023,
      "learning_rate": 0.00017316192329433586,
      "loss": 12.4957,
      "step": 5984
    },
    {
      "epoch": 1.0465573770491803,
      "grad_norm": 0.8017285354193804,
      "learning_rate": 0.00017315226889333922,
      "loss": 12.7382,
      "step": 5985
    },
    {
      "epoch": 1.0467322404371584,
      "grad_norm": 0.7101777309795791,
      "learning_rate": 0.00017314261302541498,
      "loss": 12.6431,
      "step": 5986
    },
    {
      "epoch": 1.0469071038251365,
      "grad_norm": 0.839937418971482,
      "learning_rate": 0.00017313295569075678,
      "loss": 12.7159,
      "step": 5987
    },
    {
      "epoch": 1.0470819672131149,
      "grad_norm": 0.8064669213192817,
      "learning_rate": 0.00017312329688955828,
      "loss": 12.4697,
      "step": 5988
    },
    {
      "epoch": 1.047256830601093,
      "grad_norm": 0.8889157748152028,
      "learning_rate": 0.00017311363662201313,
      "loss": 12.6498,
      "step": 5989
    },
    {
      "epoch": 1.047431693989071,
      "grad_norm": 0.8294319454963162,
      "learning_rate": 0.00017310397488831512,
      "loss": 12.4629,
      "step": 5990
    },
    {
      "epoch": 1.0476065573770492,
      "grad_norm": 0.8520124643333549,
      "learning_rate": 0.0001730943116886579,
      "loss": 12.3668,
      "step": 5991
    },
    {
      "epoch": 1.0477814207650273,
      "grad_norm": 0.7729910369296554,
      "learning_rate": 0.00017308464702323533,
      "loss": 12.6841,
      "step": 5992
    },
    {
      "epoch": 1.0479562841530055,
      "grad_norm": 0.8957695582770295,
      "learning_rate": 0.00017307498089224121,
      "loss": 12.6648,
      "step": 5993
    },
    {
      "epoch": 1.0481311475409836,
      "grad_norm": 0.7167207611618981,
      "learning_rate": 0.00017306531329586933,
      "loss": 12.2126,
      "step": 5994
    },
    {
      "epoch": 1.0483060109289617,
      "grad_norm": 0.8177641560677711,
      "learning_rate": 0.00017305564423431358,
      "loss": 12.5549,
      "step": 5995
    },
    {
      "epoch": 1.0484808743169398,
      "grad_norm": 0.8944523798979359,
      "learning_rate": 0.00017304597370776785,
      "loss": 12.6645,
      "step": 5996
    },
    {
      "epoch": 1.048655737704918,
      "grad_norm": 0.7656855145860425,
      "learning_rate": 0.00017303630171642607,
      "loss": 12.6089,
      "step": 5997
    },
    {
      "epoch": 1.0488306010928963,
      "grad_norm": 0.9696920028275341,
      "learning_rate": 0.00017302662826048216,
      "loss": 12.8506,
      "step": 5998
    },
    {
      "epoch": 1.0490054644808744,
      "grad_norm": 0.7912570311349967,
      "learning_rate": 0.00017301695334013018,
      "loss": 12.5412,
      "step": 5999
    },
    {
      "epoch": 1.0491803278688525,
      "grad_norm": 0.867984118683231,
      "learning_rate": 0.0001730072769555641,
      "loss": 12.6833,
      "step": 6000
    },
    {
      "epoch": 1.0493551912568306,
      "grad_norm": 0.803323090985723,
      "learning_rate": 0.00017299759910697793,
      "loss": 12.9154,
      "step": 6001
    },
    {
      "epoch": 1.0495300546448088,
      "grad_norm": 0.8322436721255382,
      "learning_rate": 0.00017298791979456575,
      "loss": 12.4403,
      "step": 6002
    },
    {
      "epoch": 1.0497049180327869,
      "grad_norm": 0.7502106790044338,
      "learning_rate": 0.00017297823901852167,
      "loss": 12.6619,
      "step": 6003
    },
    {
      "epoch": 1.049879781420765,
      "grad_norm": 0.9276821846574839,
      "learning_rate": 0.00017296855677903988,
      "loss": 12.6873,
      "step": 6004
    },
    {
      "epoch": 1.050054644808743,
      "grad_norm": 0.8328579958868184,
      "learning_rate": 0.00017295887307631444,
      "loss": 12.6379,
      "step": 6005
    },
    {
      "epoch": 1.0502295081967212,
      "grad_norm": 0.9275275643872539,
      "learning_rate": 0.0001729491879105396,
      "loss": 12.3249,
      "step": 6006
    },
    {
      "epoch": 1.0504043715846993,
      "grad_norm": 0.7568747482568889,
      "learning_rate": 0.00017293950128190956,
      "loss": 12.554,
      "step": 6007
    },
    {
      "epoch": 1.0505792349726777,
      "grad_norm": 0.9479416972118583,
      "learning_rate": 0.00017292981319061857,
      "loss": 12.8938,
      "step": 6008
    },
    {
      "epoch": 1.0507540983606558,
      "grad_norm": 0.7463551043046807,
      "learning_rate": 0.00017292012363686086,
      "loss": 12.5222,
      "step": 6009
    },
    {
      "epoch": 1.050928961748634,
      "grad_norm": 0.9254196520756778,
      "learning_rate": 0.0001729104326208308,
      "loss": 12.5062,
      "step": 6010
    },
    {
      "epoch": 1.051103825136612,
      "grad_norm": 0.7478848280833819,
      "learning_rate": 0.0001729007401427227,
      "loss": 12.4979,
      "step": 6011
    },
    {
      "epoch": 1.0512786885245902,
      "grad_norm": 0.9025019678263212,
      "learning_rate": 0.00017289104620273093,
      "loss": 12.6367,
      "step": 6012
    },
    {
      "epoch": 1.0514535519125683,
      "grad_norm": 0.7022919471549935,
      "learning_rate": 0.0001728813508010499,
      "loss": 12.6131,
      "step": 6013
    },
    {
      "epoch": 1.0516284153005464,
      "grad_norm": 0.7607869241127307,
      "learning_rate": 0.000172871653937874,
      "loss": 12.6253,
      "step": 6014
    },
    {
      "epoch": 1.0518032786885245,
      "grad_norm": 0.7561737836830493,
      "learning_rate": 0.0001728619556133977,
      "loss": 12.6638,
      "step": 6015
    },
    {
      "epoch": 1.0519781420765026,
      "grad_norm": 0.7865533139319223,
      "learning_rate": 0.00017285225582781543,
      "loss": 12.8428,
      "step": 6016
    },
    {
      "epoch": 1.052153005464481,
      "grad_norm": 0.7913477013125824,
      "learning_rate": 0.0001728425545813218,
      "loss": 12.5618,
      "step": 6017
    },
    {
      "epoch": 1.052327868852459,
      "grad_norm": 0.7809610881376658,
      "learning_rate": 0.00017283285187411128,
      "loss": 12.6946,
      "step": 6018
    },
    {
      "epoch": 1.0525027322404372,
      "grad_norm": 0.7545421157917338,
      "learning_rate": 0.00017282314770637841,
      "loss": 12.8292,
      "step": 6019
    },
    {
      "epoch": 1.0526775956284153,
      "grad_norm": 0.7331653735890966,
      "learning_rate": 0.00017281344207831787,
      "loss": 12.6565,
      "step": 6020
    },
    {
      "epoch": 1.0528524590163935,
      "grad_norm": 0.8143315343413883,
      "learning_rate": 0.00017280373499012423,
      "loss": 12.7444,
      "step": 6021
    },
    {
      "epoch": 1.0530273224043716,
      "grad_norm": 0.7924265641157247,
      "learning_rate": 0.0001727940264419922,
      "loss": 12.7094,
      "step": 6022
    },
    {
      "epoch": 1.0532021857923497,
      "grad_norm": 0.739044452018072,
      "learning_rate": 0.00017278431643411642,
      "loss": 12.7601,
      "step": 6023
    },
    {
      "epoch": 1.0533770491803278,
      "grad_norm": 0.806728247658124,
      "learning_rate": 0.0001727746049666916,
      "loss": 12.8032,
      "step": 6024
    },
    {
      "epoch": 1.053551912568306,
      "grad_norm": 0.7118863527096837,
      "learning_rate": 0.00017276489203991253,
      "loss": 12.6451,
      "step": 6025
    },
    {
      "epoch": 1.053726775956284,
      "grad_norm": 0.7260848262425821,
      "learning_rate": 0.00017275517765397394,
      "loss": 12.2555,
      "step": 6026
    },
    {
      "epoch": 1.0539016393442624,
      "grad_norm": 0.764010898454278,
      "learning_rate": 0.00017274546180907064,
      "loss": 12.6779,
      "step": 6027
    },
    {
      "epoch": 1.0540765027322405,
      "grad_norm": 0.6817449351928703,
      "learning_rate": 0.0001727357445053975,
      "loss": 12.4343,
      "step": 6028
    },
    {
      "epoch": 1.0542513661202186,
      "grad_norm": 0.8139160382232709,
      "learning_rate": 0.00017272602574314934,
      "loss": 12.7687,
      "step": 6029
    },
    {
      "epoch": 1.0544262295081968,
      "grad_norm": 0.7417439436625765,
      "learning_rate": 0.00017271630552252108,
      "loss": 12.6325,
      "step": 6030
    },
    {
      "epoch": 1.0546010928961749,
      "grad_norm": 0.7190866089014272,
      "learning_rate": 0.0001727065838437076,
      "loss": 12.5053,
      "step": 6031
    },
    {
      "epoch": 1.054775956284153,
      "grad_norm": 0.798722043655384,
      "learning_rate": 0.0001726968607069039,
      "loss": 12.7245,
      "step": 6032
    },
    {
      "epoch": 1.0549508196721311,
      "grad_norm": 0.8332047841684633,
      "learning_rate": 0.0001726871361123049,
      "loss": 12.5558,
      "step": 6033
    },
    {
      "epoch": 1.0551256830601092,
      "grad_norm": 0.6939790270044193,
      "learning_rate": 0.00017267741006010568,
      "loss": 12.7213,
      "step": 6034
    },
    {
      "epoch": 1.0553005464480874,
      "grad_norm": 0.8461871388486355,
      "learning_rate": 0.00017266768255050122,
      "loss": 12.6355,
      "step": 6035
    },
    {
      "epoch": 1.0554754098360655,
      "grad_norm": 0.7654422389031298,
      "learning_rate": 0.0001726579535836866,
      "loss": 12.5995,
      "step": 6036
    },
    {
      "epoch": 1.0556502732240438,
      "grad_norm": 0.8570626097052791,
      "learning_rate": 0.00017264822315985693,
      "loss": 12.6731,
      "step": 6037
    },
    {
      "epoch": 1.055825136612022,
      "grad_norm": 0.8551692967779896,
      "learning_rate": 0.00017263849127920733,
      "loss": 12.7079,
      "step": 6038
    },
    {
      "epoch": 1.056,
      "grad_norm": 0.757651922864764,
      "learning_rate": 0.00017262875794193296,
      "loss": 12.5634,
      "step": 6039
    },
    {
      "epoch": 1.0561748633879782,
      "grad_norm": 0.6976412698231663,
      "learning_rate": 0.000172619023148229,
      "loss": 12.4783,
      "step": 6040
    },
    {
      "epoch": 1.0563497267759563,
      "grad_norm": 0.9514148047925292,
      "learning_rate": 0.0001726092868982906,
      "loss": 12.5713,
      "step": 6041
    },
    {
      "epoch": 1.0565245901639344,
      "grad_norm": 0.8060886282988843,
      "learning_rate": 0.0001725995491923131,
      "loss": 12.8393,
      "step": 6042
    },
    {
      "epoch": 1.0566994535519125,
      "grad_norm": 0.8390949846530046,
      "learning_rate": 0.00017258981003049172,
      "loss": 12.6878,
      "step": 6043
    },
    {
      "epoch": 1.0568743169398906,
      "grad_norm": 0.9027288458072739,
      "learning_rate": 0.00017258006941302178,
      "loss": 12.5715,
      "step": 6044
    },
    {
      "epoch": 1.0570491803278688,
      "grad_norm": 0.8671828162221903,
      "learning_rate": 0.00017257032734009856,
      "loss": 12.5052,
      "step": 6045
    },
    {
      "epoch": 1.057224043715847,
      "grad_norm": 0.7851089854172464,
      "learning_rate": 0.0001725605838119175,
      "loss": 12.586,
      "step": 6046
    },
    {
      "epoch": 1.0573989071038252,
      "grad_norm": 0.7521787231361685,
      "learning_rate": 0.00017255083882867394,
      "loss": 12.4959,
      "step": 6047
    },
    {
      "epoch": 1.0575737704918033,
      "grad_norm": 0.7286467732984031,
      "learning_rate": 0.0001725410923905633,
      "loss": 12.7101,
      "step": 6048
    },
    {
      "epoch": 1.0577486338797815,
      "grad_norm": 0.7991301259461332,
      "learning_rate": 0.00017253134449778102,
      "loss": 12.5915,
      "step": 6049
    },
    {
      "epoch": 1.0579234972677596,
      "grad_norm": 0.7593486195367525,
      "learning_rate": 0.00017252159515052258,
      "loss": 12.5627,
      "step": 6050
    },
    {
      "epoch": 1.0580983606557377,
      "grad_norm": 0.7953216531130032,
      "learning_rate": 0.00017251184434898347,
      "loss": 12.7835,
      "step": 6051
    },
    {
      "epoch": 1.0582732240437158,
      "grad_norm": 0.7986765988449378,
      "learning_rate": 0.00017250209209335927,
      "loss": 12.8262,
      "step": 6052
    },
    {
      "epoch": 1.058448087431694,
      "grad_norm": 0.7621611186106874,
      "learning_rate": 0.00017249233838384552,
      "loss": 12.7068,
      "step": 6053
    },
    {
      "epoch": 1.058622950819672,
      "grad_norm": 0.9089468919997632,
      "learning_rate": 0.00017248258322063778,
      "loss": 12.5909,
      "step": 6054
    },
    {
      "epoch": 1.0587978142076502,
      "grad_norm": 0.7869663589803355,
      "learning_rate": 0.00017247282660393172,
      "loss": 12.6519,
      "step": 6055
    },
    {
      "epoch": 1.0589726775956285,
      "grad_norm": 0.8493936640877907,
      "learning_rate": 0.00017246306853392293,
      "loss": 12.5625,
      "step": 6056
    },
    {
      "epoch": 1.0591475409836066,
      "grad_norm": 0.7812381089699388,
      "learning_rate": 0.00017245330901080713,
      "loss": 12.591,
      "step": 6057
    },
    {
      "epoch": 1.0593224043715848,
      "grad_norm": 0.7976495165236764,
      "learning_rate": 0.00017244354803478003,
      "loss": 12.598,
      "step": 6058
    },
    {
      "epoch": 1.0594972677595629,
      "grad_norm": 0.7664420236723211,
      "learning_rate": 0.00017243378560603738,
      "loss": 12.6834,
      "step": 6059
    },
    {
      "epoch": 1.059672131147541,
      "grad_norm": 0.7976272672529537,
      "learning_rate": 0.0001724240217247749,
      "loss": 12.6304,
      "step": 6060
    },
    {
      "epoch": 1.0598469945355191,
      "grad_norm": 0.8030962537957116,
      "learning_rate": 0.00017241425639118844,
      "loss": 12.5254,
      "step": 6061
    },
    {
      "epoch": 1.0600218579234972,
      "grad_norm": 0.7722240057853494,
      "learning_rate": 0.00017240448960547378,
      "loss": 12.4613,
      "step": 6062
    },
    {
      "epoch": 1.0601967213114754,
      "grad_norm": 0.964189844365222,
      "learning_rate": 0.0001723947213678268,
      "loss": 12.6963,
      "step": 6063
    },
    {
      "epoch": 1.0603715846994535,
      "grad_norm": 0.796637021038577,
      "learning_rate": 0.0001723849516784434,
      "loss": 12.5494,
      "step": 6064
    },
    {
      "epoch": 1.0605464480874316,
      "grad_norm": 0.8361021388018479,
      "learning_rate": 0.00017237518053751943,
      "loss": 12.4607,
      "step": 6065
    },
    {
      "epoch": 1.06072131147541,
      "grad_norm": 0.8293736891998584,
      "learning_rate": 0.00017236540794525088,
      "loss": 12.7355,
      "step": 6066
    },
    {
      "epoch": 1.060896174863388,
      "grad_norm": 0.878655160373509,
      "learning_rate": 0.00017235563390183373,
      "loss": 12.3959,
      "step": 6067
    },
    {
      "epoch": 1.0610710382513662,
      "grad_norm": 1.0186468472772165,
      "learning_rate": 0.00017234585840746394,
      "loss": 12.547,
      "step": 6068
    },
    {
      "epoch": 1.0612459016393443,
      "grad_norm": 0.7670070853247887,
      "learning_rate": 0.00017233608146233754,
      "loss": 12.5278,
      "step": 6069
    },
    {
      "epoch": 1.0614207650273224,
      "grad_norm": 0.9407480246678788,
      "learning_rate": 0.00017232630306665065,
      "loss": 12.6973,
      "step": 6070
    },
    {
      "epoch": 1.0615956284153005,
      "grad_norm": 0.8820337057327913,
      "learning_rate": 0.00017231652322059925,
      "loss": 12.4684,
      "step": 6071
    },
    {
      "epoch": 1.0617704918032786,
      "grad_norm": 1.1129119510009204,
      "learning_rate": 0.00017230674192437957,
      "loss": 12.6457,
      "step": 6072
    },
    {
      "epoch": 1.0619453551912568,
      "grad_norm": 0.7717027052864379,
      "learning_rate": 0.0001722969591781877,
      "loss": 12.6267,
      "step": 6073
    },
    {
      "epoch": 1.0621202185792349,
      "grad_norm": 0.9025176416015681,
      "learning_rate": 0.0001722871749822198,
      "loss": 12.7244,
      "step": 6074
    },
    {
      "epoch": 1.0622950819672132,
      "grad_norm": 0.781697374389323,
      "learning_rate": 0.0001722773893366721,
      "loss": 12.7209,
      "step": 6075
    },
    {
      "epoch": 1.0624699453551913,
      "grad_norm": 0.9308170267770212,
      "learning_rate": 0.00017226760224174083,
      "loss": 12.5074,
      "step": 6076
    },
    {
      "epoch": 1.0626448087431695,
      "grad_norm": 0.8126164339794636,
      "learning_rate": 0.00017225781369762224,
      "loss": 12.5197,
      "step": 6077
    },
    {
      "epoch": 1.0628196721311476,
      "grad_norm": 1.0842798819445665,
      "learning_rate": 0.00017224802370451262,
      "loss": 12.6344,
      "step": 6078
    },
    {
      "epoch": 1.0629945355191257,
      "grad_norm": 0.9115214591646155,
      "learning_rate": 0.0001722382322626083,
      "loss": 12.6297,
      "step": 6079
    },
    {
      "epoch": 1.0631693989071038,
      "grad_norm": 0.8992384163035303,
      "learning_rate": 0.0001722284393721056,
      "loss": 12.681,
      "step": 6080
    },
    {
      "epoch": 1.063344262295082,
      "grad_norm": 0.8041632473641261,
      "learning_rate": 0.00017221864503320092,
      "loss": 12.295,
      "step": 6081
    },
    {
      "epoch": 1.06351912568306,
      "grad_norm": 0.9216555800909283,
      "learning_rate": 0.00017220884924609072,
      "loss": 12.8409,
      "step": 6082
    },
    {
      "epoch": 1.0636939890710382,
      "grad_norm": 0.8571364141414619,
      "learning_rate": 0.00017219905201097132,
      "loss": 12.6293,
      "step": 6083
    },
    {
      "epoch": 1.0638688524590163,
      "grad_norm": 1.0022153912267606,
      "learning_rate": 0.0001721892533280393,
      "loss": 12.8599,
      "step": 6084
    },
    {
      "epoch": 1.0640437158469944,
      "grad_norm": 0.807462165055078,
      "learning_rate": 0.00017217945319749107,
      "loss": 12.4202,
      "step": 6085
    },
    {
      "epoch": 1.0642185792349728,
      "grad_norm": 0.9024132088702954,
      "learning_rate": 0.0001721696516195232,
      "loss": 12.8878,
      "step": 6086
    },
    {
      "epoch": 1.0643934426229509,
      "grad_norm": 0.8406078460146561,
      "learning_rate": 0.0001721598485943322,
      "loss": 12.8202,
      "step": 6087
    },
    {
      "epoch": 1.064568306010929,
      "grad_norm": 0.8605161318182017,
      "learning_rate": 0.0001721500441221147,
      "loss": 12.4462,
      "step": 6088
    },
    {
      "epoch": 1.0647431693989071,
      "grad_norm": 0.8028440997893354,
      "learning_rate": 0.00017214023820306726,
      "loss": 12.8677,
      "step": 6089
    },
    {
      "epoch": 1.0649180327868852,
      "grad_norm": 0.7665338750127463,
      "learning_rate": 0.00017213043083738657,
      "loss": 12.5696,
      "step": 6090
    },
    {
      "epoch": 1.0650928961748634,
      "grad_norm": 0.9494257096852179,
      "learning_rate": 0.00017212062202526924,
      "loss": 12.1288,
      "step": 6091
    },
    {
      "epoch": 1.0652677595628415,
      "grad_norm": 0.818315555066339,
      "learning_rate": 0.00017211081176691205,
      "loss": 12.7524,
      "step": 6092
    },
    {
      "epoch": 1.0654426229508196,
      "grad_norm": 1.018686085132641,
      "learning_rate": 0.00017210100006251163,
      "loss": 12.7465,
      "step": 6093
    },
    {
      "epoch": 1.0656174863387977,
      "grad_norm": 0.8214382187055487,
      "learning_rate": 0.00017209118691226478,
      "loss": 12.5575,
      "step": 6094
    },
    {
      "epoch": 1.065792349726776,
      "grad_norm": 1.035555778578378,
      "learning_rate": 0.00017208137231636828,
      "loss": 12.7439,
      "step": 6095
    },
    {
      "epoch": 1.0659672131147542,
      "grad_norm": 0.8787234404298476,
      "learning_rate": 0.00017207155627501898,
      "loss": 12.4566,
      "step": 6096
    },
    {
      "epoch": 1.0661420765027323,
      "grad_norm": 0.8424691630811654,
      "learning_rate": 0.00017206173878841368,
      "loss": 12.493,
      "step": 6097
    },
    {
      "epoch": 1.0663169398907104,
      "grad_norm": 0.8257190655994745,
      "learning_rate": 0.00017205191985674925,
      "loss": 12.4583,
      "step": 6098
    },
    {
      "epoch": 1.0664918032786885,
      "grad_norm": 0.8023029107062174,
      "learning_rate": 0.00017204209948022258,
      "loss": 12.7496,
      "step": 6099
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 0.6762602118507681,
      "learning_rate": 0.00017203227765903064,
      "loss": 12.5261,
      "step": 6100
    },
    {
      "epoch": 1.0668415300546448,
      "grad_norm": 0.7996224763856449,
      "learning_rate": 0.00017202245439337037,
      "loss": 12.8736,
      "step": 6101
    },
    {
      "epoch": 1.067016393442623,
      "grad_norm": 0.7051791691967144,
      "learning_rate": 0.00017201262968343875,
      "loss": 12.7317,
      "step": 6102
    },
    {
      "epoch": 1.067191256830601,
      "grad_norm": 0.7762902696046527,
      "learning_rate": 0.00017200280352943278,
      "loss": 12.6752,
      "step": 6103
    },
    {
      "epoch": 1.0673661202185794,
      "grad_norm": 0.799612786729505,
      "learning_rate": 0.00017199297593154954,
      "loss": 12.6335,
      "step": 6104
    },
    {
      "epoch": 1.0675409836065575,
      "grad_norm": 0.813788770920797,
      "learning_rate": 0.00017198314688998608,
      "loss": 12.6333,
      "step": 6105
    },
    {
      "epoch": 1.0677158469945356,
      "grad_norm": 0.7853444937778644,
      "learning_rate": 0.0001719733164049395,
      "loss": 12.8236,
      "step": 6106
    },
    {
      "epoch": 1.0678907103825137,
      "grad_norm": 0.7861153460979368,
      "learning_rate": 0.00017196348447660694,
      "loss": 12.6456,
      "step": 6107
    },
    {
      "epoch": 1.0680655737704918,
      "grad_norm": 0.7629419576167761,
      "learning_rate": 0.00017195365110518555,
      "loss": 12.6639,
      "step": 6108
    },
    {
      "epoch": 1.06824043715847,
      "grad_norm": 0.8724342091716489,
      "learning_rate": 0.00017194381629087255,
      "loss": 12.7182,
      "step": 6109
    },
    {
      "epoch": 1.068415300546448,
      "grad_norm": 0.7803498310623336,
      "learning_rate": 0.0001719339800338651,
      "loss": 12.763,
      "step": 6110
    },
    {
      "epoch": 1.0685901639344262,
      "grad_norm": 0.7770803198697711,
      "learning_rate": 0.00017192414233436053,
      "loss": 12.6107,
      "step": 6111
    },
    {
      "epoch": 1.0687650273224043,
      "grad_norm": 0.7440630804752765,
      "learning_rate": 0.00017191430319255606,
      "loss": 12.6057,
      "step": 6112
    },
    {
      "epoch": 1.0689398907103824,
      "grad_norm": 0.889310979012124,
      "learning_rate": 0.00017190446260864902,
      "loss": 12.6581,
      "step": 6113
    },
    {
      "epoch": 1.0691147540983605,
      "grad_norm": 0.7393632917410988,
      "learning_rate": 0.00017189462058283668,
      "loss": 12.7394,
      "step": 6114
    },
    {
      "epoch": 1.0692896174863389,
      "grad_norm": 0.8205075107024712,
      "learning_rate": 0.00017188477711531648,
      "loss": 12.6586,
      "step": 6115
    },
    {
      "epoch": 1.069464480874317,
      "grad_norm": 0.7523289272536923,
      "learning_rate": 0.00017187493220628577,
      "loss": 12.3419,
      "step": 6116
    },
    {
      "epoch": 1.0696393442622951,
      "grad_norm": 0.760257486157633,
      "learning_rate": 0.00017186508585594198,
      "loss": 12.5339,
      "step": 6117
    },
    {
      "epoch": 1.0698142076502732,
      "grad_norm": 0.7171561208022669,
      "learning_rate": 0.0001718552380644826,
      "loss": 12.7116,
      "step": 6118
    },
    {
      "epoch": 1.0699890710382514,
      "grad_norm": 0.7827802200538033,
      "learning_rate": 0.00017184538883210503,
      "loss": 12.6546,
      "step": 6119
    },
    {
      "epoch": 1.0701639344262295,
      "grad_norm": 0.7675081997420211,
      "learning_rate": 0.00017183553815900687,
      "loss": 12.4935,
      "step": 6120
    },
    {
      "epoch": 1.0703387978142076,
      "grad_norm": 0.8716187086626114,
      "learning_rate": 0.00017182568604538556,
      "loss": 12.7497,
      "step": 6121
    },
    {
      "epoch": 1.0705136612021857,
      "grad_norm": 1.057124169726611,
      "learning_rate": 0.00017181583249143873,
      "loss": 12.7824,
      "step": 6122
    },
    {
      "epoch": 1.0706885245901638,
      "grad_norm": 0.7941660965481139,
      "learning_rate": 0.00017180597749736395,
      "loss": 12.7568,
      "step": 6123
    },
    {
      "epoch": 1.0708633879781422,
      "grad_norm": 0.821436677478493,
      "learning_rate": 0.00017179612106335885,
      "loss": 12.6466,
      "step": 6124
    },
    {
      "epoch": 1.0710382513661203,
      "grad_norm": 0.8515039431413173,
      "learning_rate": 0.00017178626318962108,
      "loss": 12.7171,
      "step": 6125
    },
    {
      "epoch": 1.0712131147540984,
      "grad_norm": 0.8718018892756297,
      "learning_rate": 0.00017177640387634833,
      "loss": 12.4911,
      "step": 6126
    },
    {
      "epoch": 1.0713879781420765,
      "grad_norm": 0.8125399688701933,
      "learning_rate": 0.0001717665431237383,
      "loss": 12.8667,
      "step": 6127
    },
    {
      "epoch": 1.0715628415300547,
      "grad_norm": 0.9113079599205288,
      "learning_rate": 0.0001717566809319887,
      "loss": 12.6399,
      "step": 6128
    },
    {
      "epoch": 1.0717377049180328,
      "grad_norm": 0.7833409413247208,
      "learning_rate": 0.00017174681730129735,
      "loss": 12.4723,
      "step": 6129
    },
    {
      "epoch": 1.071912568306011,
      "grad_norm": 0.9182001382471269,
      "learning_rate": 0.00017173695223186202,
      "loss": 12.4097,
      "step": 6130
    },
    {
      "epoch": 1.072087431693989,
      "grad_norm": 0.7887285064088468,
      "learning_rate": 0.00017172708572388054,
      "loss": 12.6295,
      "step": 6131
    },
    {
      "epoch": 1.0722622950819671,
      "grad_norm": 0.8125558436443374,
      "learning_rate": 0.00017171721777755074,
      "loss": 12.5288,
      "step": 6132
    },
    {
      "epoch": 1.0724371584699455,
      "grad_norm": 0.712338087219595,
      "learning_rate": 0.00017170734839307055,
      "loss": 12.622,
      "step": 6133
    },
    {
      "epoch": 1.0726120218579236,
      "grad_norm": 0.770686074915243,
      "learning_rate": 0.00017169747757063783,
      "loss": 12.5589,
      "step": 6134
    },
    {
      "epoch": 1.0727868852459017,
      "grad_norm": 0.7808183586298658,
      "learning_rate": 0.00017168760531045054,
      "loss": 12.2105,
      "step": 6135
    },
    {
      "epoch": 1.0729617486338798,
      "grad_norm": 0.7809836170192396,
      "learning_rate": 0.00017167773161270669,
      "loss": 12.7145,
      "step": 6136
    },
    {
      "epoch": 1.073136612021858,
      "grad_norm": 0.7125358975476187,
      "learning_rate": 0.0001716678564776042,
      "loss": 12.3557,
      "step": 6137
    },
    {
      "epoch": 1.073311475409836,
      "grad_norm": 0.8502775489850004,
      "learning_rate": 0.00017165797990534117,
      "loss": 12.451,
      "step": 6138
    },
    {
      "epoch": 1.0734863387978142,
      "grad_norm": 0.7386080990743585,
      "learning_rate": 0.0001716481018961156,
      "loss": 12.5922,
      "step": 6139
    },
    {
      "epoch": 1.0736612021857923,
      "grad_norm": 0.7896046295736716,
      "learning_rate": 0.00017163822245012562,
      "loss": 12.6535,
      "step": 6140
    },
    {
      "epoch": 1.0738360655737704,
      "grad_norm": 0.8016451162126167,
      "learning_rate": 0.0001716283415675693,
      "loss": 12.5019,
      "step": 6141
    },
    {
      "epoch": 1.0740109289617485,
      "grad_norm": 0.9504104292175827,
      "learning_rate": 0.00017161845924864482,
      "loss": 12.3096,
      "step": 6142
    },
    {
      "epoch": 1.0741857923497267,
      "grad_norm": 0.7353593475865089,
      "learning_rate": 0.00017160857549355032,
      "loss": 12.6609,
      "step": 6143
    },
    {
      "epoch": 1.074360655737705,
      "grad_norm": 0.9978232804960242,
      "learning_rate": 0.00017159869030248403,
      "loss": 12.5789,
      "step": 6144
    },
    {
      "epoch": 1.0745355191256831,
      "grad_norm": 0.7486947369074262,
      "learning_rate": 0.00017158880367564415,
      "loss": 12.5158,
      "step": 6145
    },
    {
      "epoch": 1.0747103825136612,
      "grad_norm": 0.9402379109913206,
      "learning_rate": 0.00017157891561322896,
      "loss": 12.6804,
      "step": 6146
    },
    {
      "epoch": 1.0748852459016394,
      "grad_norm": 0.68403902391226,
      "learning_rate": 0.00017156902611543671,
      "loss": 12.5462,
      "step": 6147
    },
    {
      "epoch": 1.0750601092896175,
      "grad_norm": 1.056524170794483,
      "learning_rate": 0.00017155913518246578,
      "loss": 12.7416,
      "step": 6148
    },
    {
      "epoch": 1.0752349726775956,
      "grad_norm": 0.8767165437866238,
      "learning_rate": 0.00017154924281451446,
      "loss": 12.5703,
      "step": 6149
    },
    {
      "epoch": 1.0754098360655737,
      "grad_norm": 1.0645308434812721,
      "learning_rate": 0.00017153934901178113,
      "loss": 12.909,
      "step": 6150
    },
    {
      "epoch": 1.0755846994535518,
      "grad_norm": 1.0431773724842435,
      "learning_rate": 0.00017152945377446417,
      "loss": 12.6292,
      "step": 6151
    },
    {
      "epoch": 1.07575956284153,
      "grad_norm": 0.8992763858295116,
      "learning_rate": 0.00017151955710276208,
      "loss": 12.5576,
      "step": 6152
    },
    {
      "epoch": 1.0759344262295083,
      "grad_norm": 1.1514258044753398,
      "learning_rate": 0.00017150965899687327,
      "loss": 12.3654,
      "step": 6153
    },
    {
      "epoch": 1.0761092896174864,
      "grad_norm": 0.7917954760275178,
      "learning_rate": 0.00017149975945699624,
      "loss": 12.4825,
      "step": 6154
    },
    {
      "epoch": 1.0762841530054645,
      "grad_norm": 0.8908159557697658,
      "learning_rate": 0.0001714898584833295,
      "loss": 12.9146,
      "step": 6155
    },
    {
      "epoch": 1.0764590163934427,
      "grad_norm": 0.7488383297103837,
      "learning_rate": 0.00017147995607607155,
      "loss": 12.5414,
      "step": 6156
    },
    {
      "epoch": 1.0766338797814208,
      "grad_norm": 0.9682852108296679,
      "learning_rate": 0.00017147005223542105,
      "loss": 12.8032,
      "step": 6157
    },
    {
      "epoch": 1.076808743169399,
      "grad_norm": 0.8259354815150927,
      "learning_rate": 0.00017146014696157655,
      "loss": 12.5444,
      "step": 6158
    },
    {
      "epoch": 1.076983606557377,
      "grad_norm": 0.9298987587516858,
      "learning_rate": 0.0001714502402547367,
      "loss": 12.6852,
      "step": 6159
    },
    {
      "epoch": 1.0771584699453551,
      "grad_norm": 0.8612877580060256,
      "learning_rate": 0.00017144033211510012,
      "loss": 12.7797,
      "step": 6160
    },
    {
      "epoch": 1.0773333333333333,
      "grad_norm": 0.8463320611155756,
      "learning_rate": 0.00017143042254286556,
      "loss": 12.4946,
      "step": 6161
    },
    {
      "epoch": 1.0775081967213114,
      "grad_norm": 0.7614790106137209,
      "learning_rate": 0.00017142051153823173,
      "loss": 12.6541,
      "step": 6162
    },
    {
      "epoch": 1.0776830601092897,
      "grad_norm": 0.7817714914084056,
      "learning_rate": 0.00017141059910139733,
      "loss": 12.5919,
      "step": 6163
    },
    {
      "epoch": 1.0778579234972678,
      "grad_norm": 0.7633411339238559,
      "learning_rate": 0.00017140068523256114,
      "loss": 12.6636,
      "step": 6164
    },
    {
      "epoch": 1.078032786885246,
      "grad_norm": 0.8137653051666267,
      "learning_rate": 0.000171390769931922,
      "loss": 12.3531,
      "step": 6165
    },
    {
      "epoch": 1.078207650273224,
      "grad_norm": 0.8040073379603582,
      "learning_rate": 0.00017138085319967874,
      "loss": 12.7364,
      "step": 6166
    },
    {
      "epoch": 1.0783825136612022,
      "grad_norm": 0.7436863312234607,
      "learning_rate": 0.0001713709350360302,
      "loss": 12.73,
      "step": 6167
    },
    {
      "epoch": 1.0785573770491803,
      "grad_norm": 0.7273384460117952,
      "learning_rate": 0.00017136101544117525,
      "loss": 12.5532,
      "step": 6168
    },
    {
      "epoch": 1.0787322404371584,
      "grad_norm": 0.8030578878588542,
      "learning_rate": 0.00017135109441531284,
      "loss": 12.5587,
      "step": 6169
    },
    {
      "epoch": 1.0789071038251365,
      "grad_norm": 0.8313772968001402,
      "learning_rate": 0.00017134117195864192,
      "loss": 12.5721,
      "step": 6170
    },
    {
      "epoch": 1.0790819672131147,
      "grad_norm": 0.7584326708844772,
      "learning_rate": 0.00017133124807136146,
      "loss": 12.4664,
      "step": 6171
    },
    {
      "epoch": 1.0792568306010928,
      "grad_norm": 0.743138668674203,
      "learning_rate": 0.00017132132275367044,
      "loss": 12.5015,
      "step": 6172
    },
    {
      "epoch": 1.0794316939890711,
      "grad_norm": 0.7596427781757454,
      "learning_rate": 0.00017131139600576792,
      "loss": 12.7065,
      "step": 6173
    },
    {
      "epoch": 1.0796065573770492,
      "grad_norm": 0.7604324336883578,
      "learning_rate": 0.00017130146782785296,
      "loss": 12.6451,
      "step": 6174
    },
    {
      "epoch": 1.0797814207650274,
      "grad_norm": 0.7531901938692246,
      "learning_rate": 0.00017129153822012465,
      "loss": 12.7001,
      "step": 6175
    },
    {
      "epoch": 1.0799562841530055,
      "grad_norm": 0.8080575931232901,
      "learning_rate": 0.00017128160718278212,
      "loss": 12.6408,
      "step": 6176
    },
    {
      "epoch": 1.0801311475409836,
      "grad_norm": 0.7732363704395199,
      "learning_rate": 0.00017127167471602447,
      "loss": 12.5868,
      "step": 6177
    },
    {
      "epoch": 1.0803060109289617,
      "grad_norm": 0.850939303194953,
      "learning_rate": 0.00017126174082005092,
      "loss": 12.4278,
      "step": 6178
    },
    {
      "epoch": 1.0804808743169398,
      "grad_norm": 0.7636462006524541,
      "learning_rate": 0.00017125180549506068,
      "loss": 12.6975,
      "step": 6179
    },
    {
      "epoch": 1.080655737704918,
      "grad_norm": 0.8535684844554825,
      "learning_rate": 0.00017124186874125296,
      "loss": 12.6153,
      "step": 6180
    },
    {
      "epoch": 1.080830601092896,
      "grad_norm": 0.7187667194733582,
      "learning_rate": 0.00017123193055882703,
      "loss": 12.4434,
      "step": 6181
    },
    {
      "epoch": 1.0810054644808744,
      "grad_norm": 0.7975305890801674,
      "learning_rate": 0.00017122199094798217,
      "loss": 12.4905,
      "step": 6182
    },
    {
      "epoch": 1.0811803278688525,
      "grad_norm": 0.8472682330248438,
      "learning_rate": 0.00017121204990891772,
      "loss": 12.6155,
      "step": 6183
    },
    {
      "epoch": 1.0813551912568307,
      "grad_norm": 0.7210871260484712,
      "learning_rate": 0.00017120210744183304,
      "loss": 12.7039,
      "step": 6184
    },
    {
      "epoch": 1.0815300546448088,
      "grad_norm": 0.7823626192464677,
      "learning_rate": 0.00017119216354692744,
      "loss": 12.7109,
      "step": 6185
    },
    {
      "epoch": 1.081704918032787,
      "grad_norm": 0.8325685481024424,
      "learning_rate": 0.0001711822182244004,
      "loss": 12.6449,
      "step": 6186
    },
    {
      "epoch": 1.081879781420765,
      "grad_norm": 0.8723218036927666,
      "learning_rate": 0.00017117227147445135,
      "loss": 12.757,
      "step": 6187
    },
    {
      "epoch": 1.0820546448087431,
      "grad_norm": 0.8173753626681238,
      "learning_rate": 0.0001711623232972797,
      "loss": 12.6253,
      "step": 6188
    },
    {
      "epoch": 1.0822295081967213,
      "grad_norm": 0.734422014358751,
      "learning_rate": 0.00017115237369308496,
      "loss": 12.6696,
      "step": 6189
    },
    {
      "epoch": 1.0824043715846994,
      "grad_norm": 0.8475377890826846,
      "learning_rate": 0.0001711424226620667,
      "loss": 12.5514,
      "step": 6190
    },
    {
      "epoch": 1.0825792349726775,
      "grad_norm": 0.7550645718237451,
      "learning_rate": 0.0001711324702044244,
      "loss": 12.6314,
      "step": 6191
    },
    {
      "epoch": 1.0827540983606558,
      "grad_norm": 0.8425345694472243,
      "learning_rate": 0.00017112251632035765,
      "loss": 12.4486,
      "step": 6192
    },
    {
      "epoch": 1.082928961748634,
      "grad_norm": 0.8468894512954014,
      "learning_rate": 0.00017111256101006613,
      "loss": 12.6742,
      "step": 6193
    },
    {
      "epoch": 1.083103825136612,
      "grad_norm": 0.9158111213388731,
      "learning_rate": 0.00017110260427374935,
      "loss": 12.7262,
      "step": 6194
    },
    {
      "epoch": 1.0832786885245902,
      "grad_norm": 0.8073328685286743,
      "learning_rate": 0.00017109264611160708,
      "loss": 12.5518,
      "step": 6195
    },
    {
      "epoch": 1.0834535519125683,
      "grad_norm": 0.8942962526400379,
      "learning_rate": 0.00017108268652383897,
      "loss": 12.4236,
      "step": 6196
    },
    {
      "epoch": 1.0836284153005464,
      "grad_norm": 0.8276693631099297,
      "learning_rate": 0.00017107272551064473,
      "loss": 12.6428,
      "step": 6197
    },
    {
      "epoch": 1.0838032786885246,
      "grad_norm": 0.8841502856036743,
      "learning_rate": 0.00017106276307222412,
      "loss": 12.4084,
      "step": 6198
    },
    {
      "epoch": 1.0839781420765027,
      "grad_norm": 0.8242851158068593,
      "learning_rate": 0.00017105279920877695,
      "loss": 12.6581,
      "step": 6199
    },
    {
      "epoch": 1.0841530054644808,
      "grad_norm": 0.8749115365761241,
      "learning_rate": 0.00017104283392050298,
      "loss": 12.6114,
      "step": 6200
    },
    {
      "epoch": 1.084327868852459,
      "grad_norm": 0.7531158814957576,
      "learning_rate": 0.00017103286720760206,
      "loss": 12.4996,
      "step": 6201
    },
    {
      "epoch": 1.0845027322404373,
      "grad_norm": 0.8397813159770044,
      "learning_rate": 0.00017102289907027402,
      "loss": 12.465,
      "step": 6202
    },
    {
      "epoch": 1.0846775956284154,
      "grad_norm": 0.8806433595232617,
      "learning_rate": 0.00017101292950871884,
      "loss": 12.6377,
      "step": 6203
    },
    {
      "epoch": 1.0848524590163935,
      "grad_norm": 1.0564982606519133,
      "learning_rate": 0.00017100295852313634,
      "loss": 12.6496,
      "step": 6204
    },
    {
      "epoch": 1.0850273224043716,
      "grad_norm": 0.9058489338693523,
      "learning_rate": 0.00017099298611372653,
      "loss": 12.6245,
      "step": 6205
    },
    {
      "epoch": 1.0852021857923497,
      "grad_norm": 0.9502644070054173,
      "learning_rate": 0.0001709830122806894,
      "loss": 12.4427,
      "step": 6206
    },
    {
      "epoch": 1.0853770491803278,
      "grad_norm": 0.7498771341594751,
      "learning_rate": 0.0001709730370242249,
      "loss": 12.6414,
      "step": 6207
    },
    {
      "epoch": 1.085551912568306,
      "grad_norm": 0.8874796384671042,
      "learning_rate": 0.00017096306034453308,
      "loss": 12.5956,
      "step": 6208
    },
    {
      "epoch": 1.085726775956284,
      "grad_norm": 0.8277558294028131,
      "learning_rate": 0.000170953082241814,
      "loss": 12.8224,
      "step": 6209
    },
    {
      "epoch": 1.0859016393442622,
      "grad_norm": 0.8884432068343481,
      "learning_rate": 0.00017094310271626782,
      "loss": 12.6109,
      "step": 6210
    },
    {
      "epoch": 1.0860765027322405,
      "grad_norm": 0.8683113350492967,
      "learning_rate": 0.00017093312176809456,
      "loss": 12.4876,
      "step": 6211
    },
    {
      "epoch": 1.0862513661202187,
      "grad_norm": 1.018988788706396,
      "learning_rate": 0.00017092313939749446,
      "loss": 12.8608,
      "step": 6212
    },
    {
      "epoch": 1.0864262295081968,
      "grad_norm": 0.7420176759143335,
      "learning_rate": 0.0001709131556046676,
      "loss": 12.7511,
      "step": 6213
    },
    {
      "epoch": 1.086601092896175,
      "grad_norm": 0.8226123227674581,
      "learning_rate": 0.00017090317038981426,
      "loss": 12.4622,
      "step": 6214
    },
    {
      "epoch": 1.086775956284153,
      "grad_norm": 0.7121446514646717,
      "learning_rate": 0.00017089318375313468,
      "loss": 12.5474,
      "step": 6215
    },
    {
      "epoch": 1.0869508196721311,
      "grad_norm": 0.9330848190088026,
      "learning_rate": 0.00017088319569482906,
      "loss": 12.4344,
      "step": 6216
    },
    {
      "epoch": 1.0871256830601093,
      "grad_norm": 0.7547823059480829,
      "learning_rate": 0.00017087320621509775,
      "loss": 12.567,
      "step": 6217
    },
    {
      "epoch": 1.0873005464480874,
      "grad_norm": 0.8859194519779217,
      "learning_rate": 0.00017086321531414105,
      "loss": 12.5319,
      "step": 6218
    },
    {
      "epoch": 1.0874754098360655,
      "grad_norm": 0.8863679787721523,
      "learning_rate": 0.00017085322299215927,
      "loss": 12.6651,
      "step": 6219
    },
    {
      "epoch": 1.0876502732240436,
      "grad_norm": 0.9429528594414062,
      "learning_rate": 0.00017084322924935282,
      "loss": 12.5777,
      "step": 6220
    },
    {
      "epoch": 1.087825136612022,
      "grad_norm": 0.8717989844323086,
      "learning_rate": 0.00017083323408592212,
      "loss": 12.5027,
      "step": 6221
    },
    {
      "epoch": 1.088,
      "grad_norm": 0.7943379987788272,
      "learning_rate": 0.00017082323750206761,
      "loss": 12.6557,
      "step": 6222
    },
    {
      "epoch": 1.0881748633879782,
      "grad_norm": 0.752747944632122,
      "learning_rate": 0.00017081323949798974,
      "loss": 12.5983,
      "step": 6223
    },
    {
      "epoch": 1.0883497267759563,
      "grad_norm": 0.7729656923933481,
      "learning_rate": 0.00017080324007388896,
      "loss": 12.5828,
      "step": 6224
    },
    {
      "epoch": 1.0885245901639344,
      "grad_norm": 0.7886127400689481,
      "learning_rate": 0.00017079323922996584,
      "loss": 12.7654,
      "step": 6225
    },
    {
      "epoch": 1.0886994535519126,
      "grad_norm": 0.8140573301854904,
      "learning_rate": 0.0001707832369664209,
      "loss": 12.6181,
      "step": 6226
    },
    {
      "epoch": 1.0888743169398907,
      "grad_norm": 0.7229008468665304,
      "learning_rate": 0.00017077323328345474,
      "loss": 12.4273,
      "step": 6227
    },
    {
      "epoch": 1.0890491803278688,
      "grad_norm": 0.8288499647787608,
      "learning_rate": 0.00017076322818126793,
      "loss": 12.7263,
      "step": 6228
    },
    {
      "epoch": 1.089224043715847,
      "grad_norm": 0.7834186445624345,
      "learning_rate": 0.00017075322166006112,
      "loss": 12.49,
      "step": 6229
    },
    {
      "epoch": 1.089398907103825,
      "grad_norm": 0.7356848755322825,
      "learning_rate": 0.000170743213720035,
      "loss": 12.6833,
      "step": 6230
    },
    {
      "epoch": 1.0895737704918034,
      "grad_norm": 0.7696113997668413,
      "learning_rate": 0.00017073320436139023,
      "loss": 12.6629,
      "step": 6231
    },
    {
      "epoch": 1.0897486338797815,
      "grad_norm": 0.7703974192857206,
      "learning_rate": 0.0001707231935843275,
      "loss": 12.6076,
      "step": 6232
    },
    {
      "epoch": 1.0899234972677596,
      "grad_norm": 0.873455046595189,
      "learning_rate": 0.00017071318138904763,
      "loss": 12.3941,
      "step": 6233
    },
    {
      "epoch": 1.0900983606557377,
      "grad_norm": 0.7499086657820414,
      "learning_rate": 0.00017070316777575135,
      "loss": 12.5164,
      "step": 6234
    },
    {
      "epoch": 1.0902732240437158,
      "grad_norm": 0.7863250947170518,
      "learning_rate": 0.00017069315274463947,
      "loss": 12.6913,
      "step": 6235
    },
    {
      "epoch": 1.090448087431694,
      "grad_norm": 0.7480196672104042,
      "learning_rate": 0.0001706831362959128,
      "loss": 12.9384,
      "step": 6236
    },
    {
      "epoch": 1.090622950819672,
      "grad_norm": 0.7635019102349443,
      "learning_rate": 0.00017067311842977224,
      "loss": 12.6515,
      "step": 6237
    },
    {
      "epoch": 1.0907978142076502,
      "grad_norm": 0.7781985325153374,
      "learning_rate": 0.00017066309914641866,
      "loss": 12.6632,
      "step": 6238
    },
    {
      "epoch": 1.0909726775956283,
      "grad_norm": 0.7811301661946326,
      "learning_rate": 0.00017065307844605295,
      "loss": 12.7762,
      "step": 6239
    },
    {
      "epoch": 1.0911475409836067,
      "grad_norm": 0.7076421720212477,
      "learning_rate": 0.0001706430563288761,
      "loss": 12.5826,
      "step": 6240
    },
    {
      "epoch": 1.0913224043715848,
      "grad_norm": 0.7450329064817542,
      "learning_rate": 0.00017063303279508908,
      "loss": 12.7352,
      "step": 6241
    },
    {
      "epoch": 1.091497267759563,
      "grad_norm": 0.9503980116575117,
      "learning_rate": 0.00017062300784489288,
      "loss": 12.5826,
      "step": 6242
    },
    {
      "epoch": 1.091672131147541,
      "grad_norm": 0.7287243305210619,
      "learning_rate": 0.00017061298147848851,
      "loss": 12.7018,
      "step": 6243
    },
    {
      "epoch": 1.0918469945355191,
      "grad_norm": 0.9644282906079215,
      "learning_rate": 0.00017060295369607708,
      "loss": 12.4453,
      "step": 6244
    },
    {
      "epoch": 1.0920218579234973,
      "grad_norm": 0.7640015485089603,
      "learning_rate": 0.00017059292449785965,
      "loss": 12.5288,
      "step": 6245
    },
    {
      "epoch": 1.0921967213114754,
      "grad_norm": 0.8077180808286887,
      "learning_rate": 0.0001705828938840373,
      "loss": 12.7805,
      "step": 6246
    },
    {
      "epoch": 1.0923715846994535,
      "grad_norm": 0.7980506471422146,
      "learning_rate": 0.00017057286185481122,
      "loss": 12.6869,
      "step": 6247
    },
    {
      "epoch": 1.0925464480874316,
      "grad_norm": 1.021937644912783,
      "learning_rate": 0.00017056282841038258,
      "loss": 12.4994,
      "step": 6248
    },
    {
      "epoch": 1.0927213114754097,
      "grad_norm": 0.8601859488150475,
      "learning_rate": 0.0001705527935509526,
      "loss": 12.2831,
      "step": 6249
    },
    {
      "epoch": 1.092896174863388,
      "grad_norm": 0.9767377283944921,
      "learning_rate": 0.00017054275727672243,
      "loss": 12.7261,
      "step": 6250
    },
    {
      "epoch": 1.0930710382513662,
      "grad_norm": 1.0038113213641318,
      "learning_rate": 0.00017053271958789344,
      "loss": 12.6269,
      "step": 6251
    },
    {
      "epoch": 1.0932459016393443,
      "grad_norm": 0.7810432567557722,
      "learning_rate": 0.00017052268048466682,
      "loss": 12.5065,
      "step": 6252
    },
    {
      "epoch": 1.0934207650273224,
      "grad_norm": 0.7691067822250133,
      "learning_rate": 0.0001705126399672439,
      "loss": 12.4399,
      "step": 6253
    },
    {
      "epoch": 1.0935956284153006,
      "grad_norm": 0.7890559484284673,
      "learning_rate": 0.00017050259803582608,
      "loss": 12.5054,
      "step": 6254
    },
    {
      "epoch": 1.0937704918032787,
      "grad_norm": 0.7005088809489387,
      "learning_rate": 0.00017049255469061474,
      "loss": 12.2438,
      "step": 6255
    },
    {
      "epoch": 1.0939453551912568,
      "grad_norm": 0.8275390570656269,
      "learning_rate": 0.0001704825099318112,
      "loss": 12.489,
      "step": 6256
    },
    {
      "epoch": 1.094120218579235,
      "grad_norm": 0.7869281658809184,
      "learning_rate": 0.0001704724637596169,
      "loss": 12.4313,
      "step": 6257
    },
    {
      "epoch": 1.094295081967213,
      "grad_norm": 0.8098406784015997,
      "learning_rate": 0.00017046241617423336,
      "loss": 12.8131,
      "step": 6258
    },
    {
      "epoch": 1.0944699453551912,
      "grad_norm": 0.9077426878416726,
      "learning_rate": 0.00017045236717586198,
      "loss": 12.7295,
      "step": 6259
    },
    {
      "epoch": 1.0946448087431695,
      "grad_norm": 0.9175640389241956,
      "learning_rate": 0.00017044231676470437,
      "loss": 12.7131,
      "step": 6260
    },
    {
      "epoch": 1.0948196721311476,
      "grad_norm": 0.8072446431850941,
      "learning_rate": 0.00017043226494096198,
      "loss": 12.6344,
      "step": 6261
    },
    {
      "epoch": 1.0949945355191257,
      "grad_norm": 1.0517368412897918,
      "learning_rate": 0.00017042221170483648,
      "loss": 12.6898,
      "step": 6262
    },
    {
      "epoch": 1.0951693989071039,
      "grad_norm": 0.8956562928733929,
      "learning_rate": 0.00017041215705652937,
      "loss": 12.6816,
      "step": 6263
    },
    {
      "epoch": 1.095344262295082,
      "grad_norm": 0.8364065270558468,
      "learning_rate": 0.00017040210099624235,
      "loss": 12.4186,
      "step": 6264
    },
    {
      "epoch": 1.09551912568306,
      "grad_norm": 0.8192666731840024,
      "learning_rate": 0.000170392043524177,
      "loss": 12.6741,
      "step": 6265
    },
    {
      "epoch": 1.0956939890710382,
      "grad_norm": 0.8787828195435331,
      "learning_rate": 0.00017038198464053506,
      "loss": 12.6939,
      "step": 6266
    },
    {
      "epoch": 1.0958688524590163,
      "grad_norm": 0.7750320697088632,
      "learning_rate": 0.00017037192434551823,
      "loss": 12.5748,
      "step": 6267
    },
    {
      "epoch": 1.0960437158469944,
      "grad_norm": 0.8543437814314062,
      "learning_rate": 0.00017036186263932825,
      "loss": 12.7613,
      "step": 6268
    },
    {
      "epoch": 1.0962185792349728,
      "grad_norm": 0.78314949987,
      "learning_rate": 0.00017035179952216688,
      "loss": 12.6659,
      "step": 6269
    },
    {
      "epoch": 1.096393442622951,
      "grad_norm": 0.8266469056024435,
      "learning_rate": 0.00017034173499423593,
      "loss": 12.4421,
      "step": 6270
    },
    {
      "epoch": 1.096568306010929,
      "grad_norm": 0.8448619660819021,
      "learning_rate": 0.00017033166905573718,
      "loss": 12.7862,
      "step": 6271
    },
    {
      "epoch": 1.0967431693989071,
      "grad_norm": 0.8524405640480353,
      "learning_rate": 0.00017032160170687256,
      "loss": 12.5583,
      "step": 6272
    },
    {
      "epoch": 1.0969180327868853,
      "grad_norm": 0.771697454845904,
      "learning_rate": 0.0001703115329478439,
      "loss": 12.2483,
      "step": 6273
    },
    {
      "epoch": 1.0970928961748634,
      "grad_norm": 0.8108366968888269,
      "learning_rate": 0.00017030146277885312,
      "loss": 12.4576,
      "step": 6274
    },
    {
      "epoch": 1.0972677595628415,
      "grad_norm": 0.795099926645912,
      "learning_rate": 0.00017029139120010216,
      "loss": 12.5924,
      "step": 6275
    },
    {
      "epoch": 1.0974426229508196,
      "grad_norm": 0.7711534371429437,
      "learning_rate": 0.00017028131821179295,
      "loss": 12.7268,
      "step": 6276
    },
    {
      "epoch": 1.0976174863387977,
      "grad_norm": 0.8089477193636196,
      "learning_rate": 0.00017027124381412758,
      "loss": 12.6286,
      "step": 6277
    },
    {
      "epoch": 1.0977923497267759,
      "grad_norm": 0.7792214858882812,
      "learning_rate": 0.00017026116800730796,
      "loss": 12.4776,
      "step": 6278
    },
    {
      "epoch": 1.097967213114754,
      "grad_norm": 0.7822043718093047,
      "learning_rate": 0.00017025109079153622,
      "loss": 12.5025,
      "step": 6279
    },
    {
      "epoch": 1.0981420765027323,
      "grad_norm": 0.8157324664372249,
      "learning_rate": 0.0001702410121670144,
      "loss": 12.5992,
      "step": 6280
    },
    {
      "epoch": 1.0983169398907104,
      "grad_norm": 0.7455504857946346,
      "learning_rate": 0.0001702309321339446,
      "loss": 12.6642,
      "step": 6281
    },
    {
      "epoch": 1.0984918032786886,
      "grad_norm": 0.735228550433342,
      "learning_rate": 0.00017022085069252899,
      "loss": 12.8098,
      "step": 6282
    },
    {
      "epoch": 1.0986666666666667,
      "grad_norm": 0.8343220529350444,
      "learning_rate": 0.00017021076784296972,
      "loss": 12.5455,
      "step": 6283
    },
    {
      "epoch": 1.0988415300546448,
      "grad_norm": 0.8597237339736526,
      "learning_rate": 0.00017020068358546898,
      "loss": 12.5609,
      "step": 6284
    },
    {
      "epoch": 1.099016393442623,
      "grad_norm": 0.6923394618584547,
      "learning_rate": 0.000170190597920229,
      "loss": 12.7218,
      "step": 6285
    },
    {
      "epoch": 1.099191256830601,
      "grad_norm": 0.7249796128644751,
      "learning_rate": 0.000170180510847452,
      "loss": 12.4438,
      "step": 6286
    },
    {
      "epoch": 1.0993661202185792,
      "grad_norm": 0.7966429999534472,
      "learning_rate": 0.0001701704223673403,
      "loss": 12.5356,
      "step": 6287
    },
    {
      "epoch": 1.0995409836065573,
      "grad_norm": 0.8201504068927638,
      "learning_rate": 0.00017016033248009612,
      "loss": 12.7571,
      "step": 6288
    },
    {
      "epoch": 1.0997158469945356,
      "grad_norm": 0.7816656566922332,
      "learning_rate": 0.00017015024118592192,
      "loss": 12.528,
      "step": 6289
    },
    {
      "epoch": 1.0998907103825137,
      "grad_norm": 0.900298754831554,
      "learning_rate": 0.00017014014848501993,
      "loss": 12.5988,
      "step": 6290
    },
    {
      "epoch": 1.1000655737704919,
      "grad_norm": 0.7533012059586806,
      "learning_rate": 0.00017013005437759268,
      "loss": 12.6491,
      "step": 6291
    },
    {
      "epoch": 1.10024043715847,
      "grad_norm": 0.8766722677353542,
      "learning_rate": 0.00017011995886384245,
      "loss": 12.4992,
      "step": 6292
    },
    {
      "epoch": 1.100415300546448,
      "grad_norm": 0.7041125298079819,
      "learning_rate": 0.00017010986194397178,
      "loss": 12.5285,
      "step": 6293
    },
    {
      "epoch": 1.1005901639344262,
      "grad_norm": 1.0751186373345851,
      "learning_rate": 0.0001700997636181831,
      "loss": 12.738,
      "step": 6294
    },
    {
      "epoch": 1.1007650273224043,
      "grad_norm": 0.8041526372721635,
      "learning_rate": 0.00017008966388667893,
      "loss": 12.6371,
      "step": 6295
    },
    {
      "epoch": 1.1009398907103825,
      "grad_norm": 0.8976625915333339,
      "learning_rate": 0.00017007956274966182,
      "loss": 12.6157,
      "step": 6296
    },
    {
      "epoch": 1.1011147540983606,
      "grad_norm": 0.7544916606000638,
      "learning_rate": 0.00017006946020733425,
      "loss": 12.3806,
      "step": 6297
    },
    {
      "epoch": 1.101289617486339,
      "grad_norm": 0.8037997702864093,
      "learning_rate": 0.0001700593562598989,
      "loss": 12.4057,
      "step": 6298
    },
    {
      "epoch": 1.101464480874317,
      "grad_norm": 0.7172206069627545,
      "learning_rate": 0.00017004925090755835,
      "loss": 12.6381,
      "step": 6299
    },
    {
      "epoch": 1.1016393442622952,
      "grad_norm": 0.7940095845550814,
      "learning_rate": 0.00017003914415051523,
      "loss": 12.7398,
      "step": 6300
    },
    {
      "epoch": 1.1018142076502733,
      "grad_norm": 0.7697058650921598,
      "learning_rate": 0.0001700290359889722,
      "loss": 12.6747,
      "step": 6301
    },
    {
      "epoch": 1.1019890710382514,
      "grad_norm": 0.7763512888038497,
      "learning_rate": 0.000170018926423132,
      "loss": 12.7149,
      "step": 6302
    },
    {
      "epoch": 1.1021639344262295,
      "grad_norm": 0.7586899281408,
      "learning_rate": 0.00017000881545319735,
      "loss": 12.576,
      "step": 6303
    },
    {
      "epoch": 1.1023387978142076,
      "grad_norm": 0.7985355838309931,
      "learning_rate": 0.000169998703079371,
      "loss": 12.6648,
      "step": 6304
    },
    {
      "epoch": 1.1025136612021857,
      "grad_norm": 0.7568762205131269,
      "learning_rate": 0.00016998858930185567,
      "loss": 12.2083,
      "step": 6305
    },
    {
      "epoch": 1.1026885245901639,
      "grad_norm": 0.7744901888431241,
      "learning_rate": 0.00016997847412085428,
      "loss": 12.6429,
      "step": 6306
    },
    {
      "epoch": 1.102863387978142,
      "grad_norm": 0.8216083306292603,
      "learning_rate": 0.00016996835753656963,
      "loss": 12.4608,
      "step": 6307
    },
    {
      "epoch": 1.10303825136612,
      "grad_norm": 0.7974216978332064,
      "learning_rate": 0.00016995823954920456,
      "loss": 12.5077,
      "step": 6308
    },
    {
      "epoch": 1.1032131147540984,
      "grad_norm": 0.8414563528994494,
      "learning_rate": 0.00016994812015896197,
      "loss": 12.6892,
      "step": 6309
    },
    {
      "epoch": 1.1033879781420766,
      "grad_norm": 0.834778103272237,
      "learning_rate": 0.00016993799936604483,
      "loss": 12.397,
      "step": 6310
    },
    {
      "epoch": 1.1035628415300547,
      "grad_norm": 0.754953057338108,
      "learning_rate": 0.00016992787717065607,
      "loss": 12.5747,
      "step": 6311
    },
    {
      "epoch": 1.1037377049180328,
      "grad_norm": 0.793896194485748,
      "learning_rate": 0.00016991775357299866,
      "loss": 12.6756,
      "step": 6312
    },
    {
      "epoch": 1.103912568306011,
      "grad_norm": 0.7648368806389843,
      "learning_rate": 0.0001699076285732756,
      "loss": 12.6235,
      "step": 6313
    },
    {
      "epoch": 1.104087431693989,
      "grad_norm": 0.8518743504118866,
      "learning_rate": 0.00016989750217168996,
      "loss": 12.5773,
      "step": 6314
    },
    {
      "epoch": 1.1042622950819672,
      "grad_norm": 0.7668780687419436,
      "learning_rate": 0.0001698873743684448,
      "loss": 12.6463,
      "step": 6315
    },
    {
      "epoch": 1.1044371584699453,
      "grad_norm": 0.7588921919227086,
      "learning_rate": 0.00016987724516374317,
      "loss": 12.5475,
      "step": 6316
    },
    {
      "epoch": 1.1046120218579234,
      "grad_norm": 0.8520699187476705,
      "learning_rate": 0.00016986711455778823,
      "loss": 12.4322,
      "step": 6317
    },
    {
      "epoch": 1.1047868852459017,
      "grad_norm": 0.7775065625434481,
      "learning_rate": 0.00016985698255078316,
      "loss": 12.5195,
      "step": 6318
    },
    {
      "epoch": 1.1049617486338799,
      "grad_norm": 0.8108734205371726,
      "learning_rate": 0.00016984684914293102,
      "loss": 12.4915,
      "step": 6319
    },
    {
      "epoch": 1.105136612021858,
      "grad_norm": 0.8436990763011579,
      "learning_rate": 0.00016983671433443517,
      "loss": 12.5191,
      "step": 6320
    },
    {
      "epoch": 1.105311475409836,
      "grad_norm": 0.8991066754569588,
      "learning_rate": 0.00016982657812549874,
      "loss": 12.775,
      "step": 6321
    },
    {
      "epoch": 1.1054863387978142,
      "grad_norm": 0.9150213862652256,
      "learning_rate": 0.00016981644051632504,
      "loss": 12.5563,
      "step": 6322
    },
    {
      "epoch": 1.1056612021857923,
      "grad_norm": 0.8869330217722272,
      "learning_rate": 0.0001698063015071173,
      "loss": 12.6449,
      "step": 6323
    },
    {
      "epoch": 1.1058360655737705,
      "grad_norm": 0.9439017828913575,
      "learning_rate": 0.00016979616109807894,
      "loss": 12.632,
      "step": 6324
    },
    {
      "epoch": 1.1060109289617486,
      "grad_norm": 0.8369463008533296,
      "learning_rate": 0.00016978601928941322,
      "loss": 12.7437,
      "step": 6325
    },
    {
      "epoch": 1.1061857923497267,
      "grad_norm": 0.913341373235093,
      "learning_rate": 0.00016977587608132355,
      "loss": 12.427,
      "step": 6326
    },
    {
      "epoch": 1.106360655737705,
      "grad_norm": 0.7831282726453802,
      "learning_rate": 0.0001697657314740133,
      "loss": 12.4004,
      "step": 6327
    },
    {
      "epoch": 1.1065355191256832,
      "grad_norm": 0.8882988233405704,
      "learning_rate": 0.00016975558546768595,
      "loss": 12.6569,
      "step": 6328
    },
    {
      "epoch": 1.1067103825136613,
      "grad_norm": 0.8656263033920035,
      "learning_rate": 0.00016974543806254496,
      "loss": 12.6916,
      "step": 6329
    },
    {
      "epoch": 1.1068852459016394,
      "grad_norm": 0.9904469163998955,
      "learning_rate": 0.00016973528925879374,
      "loss": 12.5506,
      "step": 6330
    },
    {
      "epoch": 1.1070601092896175,
      "grad_norm": 0.794139871381246,
      "learning_rate": 0.00016972513905663585,
      "loss": 12.5956,
      "step": 6331
    },
    {
      "epoch": 1.1072349726775956,
      "grad_norm": 0.9299454294365919,
      "learning_rate": 0.00016971498745627485,
      "loss": 12.859,
      "step": 6332
    },
    {
      "epoch": 1.1074098360655737,
      "grad_norm": 0.8018584431181177,
      "learning_rate": 0.00016970483445791432,
      "loss": 12.5514,
      "step": 6333
    },
    {
      "epoch": 1.1075846994535519,
      "grad_norm": 0.9857875788537801,
      "learning_rate": 0.00016969468006175782,
      "loss": 12.7225,
      "step": 6334
    },
    {
      "epoch": 1.10775956284153,
      "grad_norm": 0.8457272119552978,
      "learning_rate": 0.000169684524268009,
      "loss": 12.5433,
      "step": 6335
    },
    {
      "epoch": 1.107934426229508,
      "grad_norm": 0.8795257513626157,
      "learning_rate": 0.00016967436707687153,
      "loss": 12.5797,
      "step": 6336
    },
    {
      "epoch": 1.1081092896174862,
      "grad_norm": 0.8400473856144058,
      "learning_rate": 0.00016966420848854905,
      "loss": 12.5488,
      "step": 6337
    },
    {
      "epoch": 1.1082841530054646,
      "grad_norm": 1.0087437153832757,
      "learning_rate": 0.00016965404850324527,
      "loss": 12.5583,
      "step": 6338
    },
    {
      "epoch": 1.1084590163934427,
      "grad_norm": 0.9278348517770841,
      "learning_rate": 0.000169643887121164,
      "loss": 12.7345,
      "step": 6339
    },
    {
      "epoch": 1.1086338797814208,
      "grad_norm": 0.8090342775824336,
      "learning_rate": 0.0001696337243425089,
      "loss": 12.6043,
      "step": 6340
    },
    {
      "epoch": 1.108808743169399,
      "grad_norm": 0.787827948057993,
      "learning_rate": 0.00016962356016748386,
      "loss": 12.5809,
      "step": 6341
    },
    {
      "epoch": 1.108983606557377,
      "grad_norm": 0.7288279497141759,
      "learning_rate": 0.0001696133945962927,
      "loss": 12.6184,
      "step": 6342
    },
    {
      "epoch": 1.1091584699453552,
      "grad_norm": 0.90000965083011,
      "learning_rate": 0.00016960322762913917,
      "loss": 12.5772,
      "step": 6343
    },
    {
      "epoch": 1.1093333333333333,
      "grad_norm": 0.796223597805332,
      "learning_rate": 0.00016959305926622723,
      "loss": 12.7915,
      "step": 6344
    },
    {
      "epoch": 1.1095081967213114,
      "grad_norm": 0.8276528767332769,
      "learning_rate": 0.0001695828895077608,
      "loss": 12.4187,
      "step": 6345
    },
    {
      "epoch": 1.1096830601092895,
      "grad_norm": 0.7539139738116349,
      "learning_rate": 0.00016957271835394376,
      "loss": 12.0669,
      "step": 6346
    },
    {
      "epoch": 1.1098579234972679,
      "grad_norm": 0.800122830649583,
      "learning_rate": 0.00016956254580498015,
      "loss": 12.7322,
      "step": 6347
    },
    {
      "epoch": 1.110032786885246,
      "grad_norm": 0.7659935767617642,
      "learning_rate": 0.00016955237186107387,
      "loss": 12.6923,
      "step": 6348
    },
    {
      "epoch": 1.110207650273224,
      "grad_norm": 0.7598018329529252,
      "learning_rate": 0.00016954219652242898,
      "loss": 12.7118,
      "step": 6349
    },
    {
      "epoch": 1.1103825136612022,
      "grad_norm": 0.785700459712739,
      "learning_rate": 0.00016953201978924952,
      "loss": 12.7673,
      "step": 6350
    },
    {
      "epoch": 1.1105573770491803,
      "grad_norm": 0.7881180155440416,
      "learning_rate": 0.00016952184166173958,
      "loss": 12.6055,
      "step": 6351
    },
    {
      "epoch": 1.1107322404371585,
      "grad_norm": 0.8322107288234681,
      "learning_rate": 0.0001695116621401033,
      "loss": 12.9533,
      "step": 6352
    },
    {
      "epoch": 1.1109071038251366,
      "grad_norm": 0.8574677360642556,
      "learning_rate": 0.0001695014812245447,
      "loss": 12.6236,
      "step": 6353
    },
    {
      "epoch": 1.1110819672131147,
      "grad_norm": 0.7437658531984094,
      "learning_rate": 0.00016949129891526805,
      "loss": 12.5152,
      "step": 6354
    },
    {
      "epoch": 1.1112568306010928,
      "grad_norm": 0.8345891299646357,
      "learning_rate": 0.00016948111521247747,
      "loss": 12.4485,
      "step": 6355
    },
    {
      "epoch": 1.111431693989071,
      "grad_norm": 0.7562366990911045,
      "learning_rate": 0.0001694709301163772,
      "loss": 12.7017,
      "step": 6356
    },
    {
      "epoch": 1.1116065573770493,
      "grad_norm": 0.7896015822983359,
      "learning_rate": 0.00016946074362717147,
      "loss": 12.1584,
      "step": 6357
    },
    {
      "epoch": 1.1117814207650274,
      "grad_norm": 0.7936772849873831,
      "learning_rate": 0.00016945055574506458,
      "loss": 12.7805,
      "step": 6358
    },
    {
      "epoch": 1.1119562841530055,
      "grad_norm": 0.7866746894688234,
      "learning_rate": 0.00016944036647026077,
      "loss": 12.6564,
      "step": 6359
    },
    {
      "epoch": 1.1121311475409836,
      "grad_norm": 0.7440412323475706,
      "learning_rate": 0.0001694301758029644,
      "loss": 12.6238,
      "step": 6360
    },
    {
      "epoch": 1.1123060109289618,
      "grad_norm": 0.7444326557935479,
      "learning_rate": 0.00016941998374337986,
      "loss": 12.5006,
      "step": 6361
    },
    {
      "epoch": 1.1124808743169399,
      "grad_norm": 0.7518816651599005,
      "learning_rate": 0.00016940979029171147,
      "loss": 12.7373,
      "step": 6362
    },
    {
      "epoch": 1.112655737704918,
      "grad_norm": 0.7704744103557604,
      "learning_rate": 0.0001693995954481637,
      "loss": 12.6863,
      "step": 6363
    },
    {
      "epoch": 1.112830601092896,
      "grad_norm": 0.7483106468548055,
      "learning_rate": 0.00016938939921294096,
      "loss": 12.6182,
      "step": 6364
    },
    {
      "epoch": 1.1130054644808742,
      "grad_norm": 0.7243403096888227,
      "learning_rate": 0.00016937920158624766,
      "loss": 12.677,
      "step": 6365
    },
    {
      "epoch": 1.1131803278688523,
      "grad_norm": 0.751652310844734,
      "learning_rate": 0.00016936900256828838,
      "loss": 12.5867,
      "step": 6366
    },
    {
      "epoch": 1.1133551912568307,
      "grad_norm": 0.7171758056874281,
      "learning_rate": 0.0001693588021592676,
      "loss": 12.5217,
      "step": 6367
    },
    {
      "epoch": 1.1135300546448088,
      "grad_norm": 0.8398519096662206,
      "learning_rate": 0.0001693486003593899,
      "loss": 12.6441,
      "step": 6368
    },
    {
      "epoch": 1.113704918032787,
      "grad_norm": 0.7514654633111593,
      "learning_rate": 0.0001693383971688598,
      "loss": 12.5971,
      "step": 6369
    },
    {
      "epoch": 1.113879781420765,
      "grad_norm": 0.8343983874639702,
      "learning_rate": 0.00016932819258788198,
      "loss": 12.3804,
      "step": 6370
    },
    {
      "epoch": 1.1140546448087432,
      "grad_norm": 0.9089950830419647,
      "learning_rate": 0.000169317986616661,
      "loss": 12.6589,
      "step": 6371
    },
    {
      "epoch": 1.1142295081967213,
      "grad_norm": 0.6815254864695721,
      "learning_rate": 0.00016930777925540155,
      "loss": 12.5261,
      "step": 6372
    },
    {
      "epoch": 1.1144043715846994,
      "grad_norm": 0.7266587135188353,
      "learning_rate": 0.00016929757050430837,
      "loss": 12.3862,
      "step": 6373
    },
    {
      "epoch": 1.1145792349726775,
      "grad_norm": 0.8044937487314244,
      "learning_rate": 0.00016928736036358607,
      "loss": 12.456,
      "step": 6374
    },
    {
      "epoch": 1.1147540983606556,
      "grad_norm": 0.8395374310912568,
      "learning_rate": 0.00016927714883343948,
      "loss": 12.7138,
      "step": 6375
    },
    {
      "epoch": 1.114928961748634,
      "grad_norm": 0.7946474510848658,
      "learning_rate": 0.00016926693591407338,
      "loss": 12.7212,
      "step": 6376
    },
    {
      "epoch": 1.115103825136612,
      "grad_norm": 0.8163439223895476,
      "learning_rate": 0.00016925672160569247,
      "loss": 12.7224,
      "step": 6377
    },
    {
      "epoch": 1.1152786885245902,
      "grad_norm": 0.7431404750699427,
      "learning_rate": 0.0001692465059085017,
      "loss": 12.5486,
      "step": 6378
    },
    {
      "epoch": 1.1154535519125683,
      "grad_norm": 0.9713228077545666,
      "learning_rate": 0.00016923628882270587,
      "loss": 12.6094,
      "step": 6379
    },
    {
      "epoch": 1.1156284153005465,
      "grad_norm": 0.7935721824522929,
      "learning_rate": 0.00016922607034850984,
      "loss": 12.3354,
      "step": 6380
    },
    {
      "epoch": 1.1158032786885246,
      "grad_norm": 0.8920492212658973,
      "learning_rate": 0.00016921585048611855,
      "loss": 12.3645,
      "step": 6381
    },
    {
      "epoch": 1.1159781420765027,
      "grad_norm": 0.7522609463750023,
      "learning_rate": 0.00016920562923573694,
      "loss": 12.6699,
      "step": 6382
    },
    {
      "epoch": 1.1161530054644808,
      "grad_norm": 0.8503848693281131,
      "learning_rate": 0.00016919540659757,
      "loss": 12.7936,
      "step": 6383
    },
    {
      "epoch": 1.116327868852459,
      "grad_norm": 0.9080494981785995,
      "learning_rate": 0.00016918518257182265,
      "loss": 12.7428,
      "step": 6384
    },
    {
      "epoch": 1.116502732240437,
      "grad_norm": 0.8830787448559269,
      "learning_rate": 0.00016917495715870003,
      "loss": 12.7869,
      "step": 6385
    },
    {
      "epoch": 1.1166775956284154,
      "grad_norm": 0.7854897837479694,
      "learning_rate": 0.00016916473035840705,
      "loss": 12.6251,
      "step": 6386
    },
    {
      "epoch": 1.1168524590163935,
      "grad_norm": 0.8218148318994377,
      "learning_rate": 0.00016915450217114893,
      "loss": 12.6093,
      "step": 6387
    },
    {
      "epoch": 1.1170273224043716,
      "grad_norm": 0.8072831280442245,
      "learning_rate": 0.00016914427259713066,
      "loss": 12.3249,
      "step": 6388
    },
    {
      "epoch": 1.1172021857923498,
      "grad_norm": 0.735636015785619,
      "learning_rate": 0.00016913404163655744,
      "loss": 12.6013,
      "step": 6389
    },
    {
      "epoch": 1.1173770491803279,
      "grad_norm": 0.7383686850433955,
      "learning_rate": 0.00016912380928963444,
      "loss": 12.6666,
      "step": 6390
    },
    {
      "epoch": 1.117551912568306,
      "grad_norm": 0.9216562776930328,
      "learning_rate": 0.0001691135755565668,
      "loss": 12.5739,
      "step": 6391
    },
    {
      "epoch": 1.1177267759562841,
      "grad_norm": 0.8141624672758676,
      "learning_rate": 0.00016910334043755974,
      "loss": 12.796,
      "step": 6392
    },
    {
      "epoch": 1.1179016393442622,
      "grad_norm": 0.892478338170509,
      "learning_rate": 0.00016909310393281856,
      "loss": 12.6946,
      "step": 6393
    },
    {
      "epoch": 1.1180765027322404,
      "grad_norm": 0.7829848988216724,
      "learning_rate": 0.00016908286604254853,
      "loss": 12.6934,
      "step": 6394
    },
    {
      "epoch": 1.1182513661202185,
      "grad_norm": 0.913247769171478,
      "learning_rate": 0.0001690726267669549,
      "loss": 12.7608,
      "step": 6395
    },
    {
      "epoch": 1.1184262295081968,
      "grad_norm": 0.7404737197798551,
      "learning_rate": 0.000169062386106243,
      "loss": 12.4289,
      "step": 6396
    },
    {
      "epoch": 1.118601092896175,
      "grad_norm": 0.8573587897902935,
      "learning_rate": 0.00016905214406061822,
      "loss": 12.2849,
      "step": 6397
    },
    {
      "epoch": 1.118775956284153,
      "grad_norm": 0.7795224118336564,
      "learning_rate": 0.00016904190063028594,
      "loss": 12.571,
      "step": 6398
    },
    {
      "epoch": 1.1189508196721312,
      "grad_norm": 0.9211049945754373,
      "learning_rate": 0.00016903165581545157,
      "loss": 12.568,
      "step": 6399
    },
    {
      "epoch": 1.1191256830601093,
      "grad_norm": 0.767592173498413,
      "learning_rate": 0.00016902140961632054,
      "loss": 12.5745,
      "step": 6400
    },
    {
      "epoch": 1.1193005464480874,
      "grad_norm": 0.9496723498878955,
      "learning_rate": 0.0001690111620330983,
      "loss": 12.649,
      "step": 6401
    },
    {
      "epoch": 1.1194754098360655,
      "grad_norm": 0.7214619324048824,
      "learning_rate": 0.00016900091306599042,
      "loss": 12.5074,
      "step": 6402
    },
    {
      "epoch": 1.1196502732240436,
      "grad_norm": 0.8716379856190004,
      "learning_rate": 0.00016899066271520234,
      "loss": 12.6367,
      "step": 6403
    },
    {
      "epoch": 1.1198251366120218,
      "grad_norm": 0.7563794846861577,
      "learning_rate": 0.00016898041098093965,
      "loss": 12.6742,
      "step": 6404
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8614119767465409,
      "learning_rate": 0.0001689701578634079,
      "loss": 12.414,
      "step": 6405
    },
    {
      "epoch": 1.1201748633879782,
      "grad_norm": 0.8030585049879986,
      "learning_rate": 0.00016895990336281274,
      "loss": 12.3176,
      "step": 6406
    },
    {
      "epoch": 1.1203497267759563,
      "grad_norm": 0.8068542557320841,
      "learning_rate": 0.00016894964747935979,
      "loss": 12.3804,
      "step": 6407
    },
    {
      "epoch": 1.1205245901639345,
      "grad_norm": 0.7470519566851763,
      "learning_rate": 0.0001689393902132547,
      "loss": 12.6443,
      "step": 6408
    },
    {
      "epoch": 1.1206994535519126,
      "grad_norm": 0.8273280938545774,
      "learning_rate": 0.00016892913156470315,
      "loss": 12.6925,
      "step": 6409
    },
    {
      "epoch": 1.1208743169398907,
      "grad_norm": 0.7908104467286774,
      "learning_rate": 0.0001689188715339109,
      "loss": 12.2079,
      "step": 6410
    },
    {
      "epoch": 1.1210491803278688,
      "grad_norm": 0.7346671517619078,
      "learning_rate": 0.00016890861012108365,
      "loss": 12.454,
      "step": 6411
    },
    {
      "epoch": 1.121224043715847,
      "grad_norm": 0.7943849517998749,
      "learning_rate": 0.00016889834732642718,
      "loss": 12.7212,
      "step": 6412
    },
    {
      "epoch": 1.121398907103825,
      "grad_norm": 0.8726272588322322,
      "learning_rate": 0.00016888808315014732,
      "loss": 12.5975,
      "step": 6413
    },
    {
      "epoch": 1.1215737704918032,
      "grad_norm": 0.6930453244715361,
      "learning_rate": 0.00016887781759244988,
      "loss": 12.5552,
      "step": 6414
    },
    {
      "epoch": 1.1217486338797815,
      "grad_norm": 0.8770968786741312,
      "learning_rate": 0.0001688675506535407,
      "loss": 12.5783,
      "step": 6415
    },
    {
      "epoch": 1.1219234972677596,
      "grad_norm": 0.7496131348172069,
      "learning_rate": 0.00016885728233362568,
      "loss": 12.7561,
      "step": 6416
    },
    {
      "epoch": 1.1220983606557378,
      "grad_norm": 0.7879779751766784,
      "learning_rate": 0.0001688470126329107,
      "loss": 12.1653,
      "step": 6417
    },
    {
      "epoch": 1.1222732240437159,
      "grad_norm": 0.8283567791737131,
      "learning_rate": 0.0001688367415516018,
      "loss": 12.7996,
      "step": 6418
    },
    {
      "epoch": 1.122448087431694,
      "grad_norm": 0.8154169882232789,
      "learning_rate": 0.0001688264690899048,
      "loss": 12.5969,
      "step": 6419
    },
    {
      "epoch": 1.1226229508196721,
      "grad_norm": 0.8339519633126297,
      "learning_rate": 0.00016881619524802583,
      "loss": 12.6939,
      "step": 6420
    },
    {
      "epoch": 1.1227978142076502,
      "grad_norm": 0.793045969980434,
      "learning_rate": 0.00016880592002617082,
      "loss": 12.6866,
      "step": 6421
    },
    {
      "epoch": 1.1229726775956284,
      "grad_norm": 0.87616083006216,
      "learning_rate": 0.00016879564342454586,
      "loss": 12.7159,
      "step": 6422
    },
    {
      "epoch": 1.1231475409836065,
      "grad_norm": 0.7975137388025002,
      "learning_rate": 0.00016878536544335702,
      "loss": 12.4835,
      "step": 6423
    },
    {
      "epoch": 1.1233224043715846,
      "grad_norm": 0.858850098071165,
      "learning_rate": 0.00016877508608281042,
      "loss": 12.6828,
      "step": 6424
    },
    {
      "epoch": 1.123497267759563,
      "grad_norm": 0.8112336309395874,
      "learning_rate": 0.00016876480534311214,
      "loss": 12.5369,
      "step": 6425
    },
    {
      "epoch": 1.123672131147541,
      "grad_norm": 0.8935397811166328,
      "learning_rate": 0.0001687545232244684,
      "loss": 12.8323,
      "step": 6426
    },
    {
      "epoch": 1.1238469945355192,
      "grad_norm": 0.9069455741727492,
      "learning_rate": 0.00016874423972708535,
      "loss": 12.6043,
      "step": 6427
    },
    {
      "epoch": 1.1240218579234973,
      "grad_norm": 0.7322935186965697,
      "learning_rate": 0.00016873395485116923,
      "loss": 12.547,
      "step": 6428
    },
    {
      "epoch": 1.1241967213114754,
      "grad_norm": 0.7088941934116679,
      "learning_rate": 0.00016872366859692627,
      "loss": 12.4917,
      "step": 6429
    },
    {
      "epoch": 1.1243715846994535,
      "grad_norm": 0.8448407826415997,
      "learning_rate": 0.00016871338096456276,
      "loss": 12.4894,
      "step": 6430
    },
    {
      "epoch": 1.1245464480874316,
      "grad_norm": 0.7797594810458165,
      "learning_rate": 0.000168703091954285,
      "loss": 12.7957,
      "step": 6431
    },
    {
      "epoch": 1.1247213114754098,
      "grad_norm": 0.712655082705891,
      "learning_rate": 0.00016869280156629925,
      "loss": 12.5876,
      "step": 6432
    },
    {
      "epoch": 1.1248961748633879,
      "grad_norm": 0.8180404982138776,
      "learning_rate": 0.00016868250980081194,
      "loss": 12.7,
      "step": 6433
    },
    {
      "epoch": 1.1250710382513662,
      "grad_norm": 0.7922571939204136,
      "learning_rate": 0.0001686722166580294,
      "loss": 12.3812,
      "step": 6434
    },
    {
      "epoch": 1.1252459016393443,
      "grad_norm": 0.8498850781217364,
      "learning_rate": 0.0001686619221381581,
      "loss": 12.465,
      "step": 6435
    },
    {
      "epoch": 1.1254207650273225,
      "grad_norm": 0.8886908268312965,
      "learning_rate": 0.00016865162624140446,
      "loss": 12.3511,
      "step": 6436
    },
    {
      "epoch": 1.1255956284153006,
      "grad_norm": 0.7710483375169835,
      "learning_rate": 0.00016864132896797488,
      "loss": 12.7493,
      "step": 6437
    },
    {
      "epoch": 1.1257704918032787,
      "grad_norm": 0.8183350606428932,
      "learning_rate": 0.0001686310303180759,
      "loss": 12.6896,
      "step": 6438
    },
    {
      "epoch": 1.1259453551912568,
      "grad_norm": 0.8143337496059834,
      "learning_rate": 0.00016862073029191404,
      "loss": 12.7432,
      "step": 6439
    },
    {
      "epoch": 1.126120218579235,
      "grad_norm": 0.7971624056633229,
      "learning_rate": 0.0001686104288896959,
      "loss": 12.7726,
      "step": 6440
    },
    {
      "epoch": 1.126295081967213,
      "grad_norm": 0.821937504191945,
      "learning_rate": 0.00016860012611162795,
      "loss": 12.5462,
      "step": 6441
    },
    {
      "epoch": 1.1264699453551912,
      "grad_norm": 0.7854869444131788,
      "learning_rate": 0.00016858982195791685,
      "loss": 12.4569,
      "step": 6442
    },
    {
      "epoch": 1.1266448087431695,
      "grad_norm": 0.746916567180434,
      "learning_rate": 0.00016857951642876921,
      "loss": 12.7717,
      "step": 6443
    },
    {
      "epoch": 1.1268196721311474,
      "grad_norm": 0.8189367128945005,
      "learning_rate": 0.0001685692095243917,
      "loss": 12.4233,
      "step": 6444
    },
    {
      "epoch": 1.1269945355191258,
      "grad_norm": 0.786790087391437,
      "learning_rate": 0.000168558901244991,
      "loss": 12.407,
      "step": 6445
    },
    {
      "epoch": 1.1271693989071039,
      "grad_norm": 0.7294539550790667,
      "learning_rate": 0.00016854859159077385,
      "loss": 12.6512,
      "step": 6446
    },
    {
      "epoch": 1.127344262295082,
      "grad_norm": 0.929951015721952,
      "learning_rate": 0.00016853828056194697,
      "loss": 12.7415,
      "step": 6447
    },
    {
      "epoch": 1.1275191256830601,
      "grad_norm": 0.8432347977688273,
      "learning_rate": 0.0001685279681587171,
      "loss": 12.7147,
      "step": 6448
    },
    {
      "epoch": 1.1276939890710382,
      "grad_norm": 0.754989215969738,
      "learning_rate": 0.00016851765438129107,
      "loss": 12.6801,
      "step": 6449
    },
    {
      "epoch": 1.1278688524590164,
      "grad_norm": 0.7640039416905777,
      "learning_rate": 0.00016850733922987567,
      "loss": 12.6288,
      "step": 6450
    },
    {
      "epoch": 1.1280437158469945,
      "grad_norm": 0.8199755370977974,
      "learning_rate": 0.0001684970227046778,
      "loss": 12.6032,
      "step": 6451
    },
    {
      "epoch": 1.1282185792349726,
      "grad_norm": 0.8056983927648668,
      "learning_rate": 0.0001684867048059043,
      "loss": 12.7121,
      "step": 6452
    },
    {
      "epoch": 1.1283934426229507,
      "grad_norm": 0.8935697635268629,
      "learning_rate": 0.0001684763855337621,
      "loss": 12.5425,
      "step": 6453
    },
    {
      "epoch": 1.128568306010929,
      "grad_norm": 0.9103865693072618,
      "learning_rate": 0.00016846606488845811,
      "loss": 12.5558,
      "step": 6454
    },
    {
      "epoch": 1.1287431693989072,
      "grad_norm": 0.8575877697669506,
      "learning_rate": 0.0001684557428701993,
      "loss": 12.3084,
      "step": 6455
    },
    {
      "epoch": 1.1289180327868853,
      "grad_norm": 0.9221992412145987,
      "learning_rate": 0.00016844541947919268,
      "loss": 12.7838,
      "step": 6456
    },
    {
      "epoch": 1.1290928961748634,
      "grad_norm": 0.6866890237924835,
      "learning_rate": 0.00016843509471564522,
      "loss": 12.5997,
      "step": 6457
    },
    {
      "epoch": 1.1292677595628415,
      "grad_norm": 0.8630887772645043,
      "learning_rate": 0.00016842476857976396,
      "loss": 12.5267,
      "step": 6458
    },
    {
      "epoch": 1.1294426229508197,
      "grad_norm": 0.7528217409889154,
      "learning_rate": 0.00016841444107175605,
      "loss": 12.551,
      "step": 6459
    },
    {
      "epoch": 1.1296174863387978,
      "grad_norm": 0.8999207723877078,
      "learning_rate": 0.00016840411219182854,
      "loss": 12.7527,
      "step": 6460
    },
    {
      "epoch": 1.129792349726776,
      "grad_norm": 0.7300053879265211,
      "learning_rate": 0.00016839378194018851,
      "loss": 12.2961,
      "step": 6461
    },
    {
      "epoch": 1.129967213114754,
      "grad_norm": 0.8202230461523423,
      "learning_rate": 0.00016838345031704318,
      "loss": 12.6424,
      "step": 6462
    },
    {
      "epoch": 1.1301420765027324,
      "grad_norm": 0.7116972750068653,
      "learning_rate": 0.0001683731173225997,
      "loss": 12.4038,
      "step": 6463
    },
    {
      "epoch": 1.1303169398907105,
      "grad_norm": 0.8774837107754009,
      "learning_rate": 0.00016836278295706528,
      "loss": 12.5563,
      "step": 6464
    },
    {
      "epoch": 1.1304918032786886,
      "grad_norm": 0.7427966946787968,
      "learning_rate": 0.00016835244722064716,
      "loss": 12.2695,
      "step": 6465
    },
    {
      "epoch": 1.1306666666666667,
      "grad_norm": 0.8412293096662699,
      "learning_rate": 0.00016834211011355264,
      "loss": 12.4202,
      "step": 6466
    },
    {
      "epoch": 1.1308415300546448,
      "grad_norm": 0.6817147231704372,
      "learning_rate": 0.00016833177163598894,
      "loss": 12.5717,
      "step": 6467
    },
    {
      "epoch": 1.131016393442623,
      "grad_norm": 0.8193338413314025,
      "learning_rate": 0.00016832143178816341,
      "loss": 12.5521,
      "step": 6468
    },
    {
      "epoch": 1.131191256830601,
      "grad_norm": 0.7827924426133301,
      "learning_rate": 0.00016831109057028343,
      "loss": 12.6755,
      "step": 6469
    },
    {
      "epoch": 1.1313661202185792,
      "grad_norm": 0.6959142486410195,
      "learning_rate": 0.0001683007479825563,
      "loss": 12.2523,
      "step": 6470
    },
    {
      "epoch": 1.1315409836065573,
      "grad_norm": 0.7724987008330649,
      "learning_rate": 0.0001682904040251895,
      "loss": 12.7984,
      "step": 6471
    },
    {
      "epoch": 1.1317158469945354,
      "grad_norm": 0.755602674461817,
      "learning_rate": 0.0001682800586983904,
      "loss": 12.4343,
      "step": 6472
    },
    {
      "epoch": 1.1318907103825135,
      "grad_norm": 0.7386393927381969,
      "learning_rate": 0.00016826971200236646,
      "loss": 12.5046,
      "step": 6473
    },
    {
      "epoch": 1.1320655737704919,
      "grad_norm": 0.7176082836231903,
      "learning_rate": 0.0001682593639373252,
      "loss": 12.5345,
      "step": 6474
    },
    {
      "epoch": 1.13224043715847,
      "grad_norm": 0.7437679270056837,
      "learning_rate": 0.0001682490145034741,
      "loss": 12.7889,
      "step": 6475
    },
    {
      "epoch": 1.1324153005464481,
      "grad_norm": 0.7409490089143662,
      "learning_rate": 0.0001682386637010207,
      "loss": 12.7459,
      "step": 6476
    },
    {
      "epoch": 1.1325901639344262,
      "grad_norm": 0.7939945445479538,
      "learning_rate": 0.0001682283115301726,
      "loss": 12.6858,
      "step": 6477
    },
    {
      "epoch": 1.1327650273224044,
      "grad_norm": 0.7321884506492226,
      "learning_rate": 0.00016821795799113734,
      "loss": 12.6161,
      "step": 6478
    },
    {
      "epoch": 1.1329398907103825,
      "grad_norm": 0.7821992655681181,
      "learning_rate": 0.00016820760308412257,
      "loss": 12.4963,
      "step": 6479
    },
    {
      "epoch": 1.1331147540983606,
      "grad_norm": 0.731272058604909,
      "learning_rate": 0.00016819724680933594,
      "loss": 12.6809,
      "step": 6480
    },
    {
      "epoch": 1.1332896174863387,
      "grad_norm": 0.6951845991925447,
      "learning_rate": 0.00016818688916698512,
      "loss": 12.6527,
      "step": 6481
    },
    {
      "epoch": 1.1334644808743168,
      "grad_norm": 0.765638826084974,
      "learning_rate": 0.0001681765301572778,
      "loss": 12.6231,
      "step": 6482
    },
    {
      "epoch": 1.1336393442622952,
      "grad_norm": 0.7601076184920201,
      "learning_rate": 0.00016816616978042174,
      "loss": 12.5216,
      "step": 6483
    },
    {
      "epoch": 1.1338142076502733,
      "grad_norm": 0.678448848273167,
      "learning_rate": 0.00016815580803662466,
      "loss": 12.4796,
      "step": 6484
    },
    {
      "epoch": 1.1339890710382514,
      "grad_norm": 0.7390126456921224,
      "learning_rate": 0.00016814544492609436,
      "loss": 12.7429,
      "step": 6485
    },
    {
      "epoch": 1.1341639344262295,
      "grad_norm": 0.7304814027036731,
      "learning_rate": 0.0001681350804490387,
      "loss": 12.7188,
      "step": 6486
    },
    {
      "epoch": 1.1343387978142077,
      "grad_norm": 0.6951999741437037,
      "learning_rate": 0.0001681247146056654,
      "loss": 12.2996,
      "step": 6487
    },
    {
      "epoch": 1.1345136612021858,
      "grad_norm": 0.6991068309051767,
      "learning_rate": 0.00016811434739618246,
      "loss": 12.5921,
      "step": 6488
    },
    {
      "epoch": 1.134688524590164,
      "grad_norm": 0.7153129748092334,
      "learning_rate": 0.0001681039788207977,
      "loss": 12.7404,
      "step": 6489
    },
    {
      "epoch": 1.134863387978142,
      "grad_norm": 0.7879862804994858,
      "learning_rate": 0.00016809360887971907,
      "loss": 12.6783,
      "step": 6490
    },
    {
      "epoch": 1.1350382513661201,
      "grad_norm": 0.9312422490965783,
      "learning_rate": 0.0001680832375731545,
      "loss": 12.7006,
      "step": 6491
    },
    {
      "epoch": 1.1352131147540985,
      "grad_norm": 0.718834380166545,
      "learning_rate": 0.00016807286490131196,
      "loss": 12.6573,
      "step": 6492
    },
    {
      "epoch": 1.1353879781420766,
      "grad_norm": 0.7595603817615143,
      "learning_rate": 0.00016806249086439947,
      "loss": 12.1107,
      "step": 6493
    },
    {
      "epoch": 1.1355628415300547,
      "grad_norm": 0.7261684094255989,
      "learning_rate": 0.00016805211546262505,
      "loss": 12.481,
      "step": 6494
    },
    {
      "epoch": 1.1357377049180328,
      "grad_norm": 0.8886044838357391,
      "learning_rate": 0.00016804173869619678,
      "loss": 12.7746,
      "step": 6495
    },
    {
      "epoch": 1.135912568306011,
      "grad_norm": 0.718949885201903,
      "learning_rate": 0.00016803136056532276,
      "loss": 12.4697,
      "step": 6496
    },
    {
      "epoch": 1.136087431693989,
      "grad_norm": 0.7951806809175435,
      "learning_rate": 0.00016802098107021105,
      "loss": 12.4783,
      "step": 6497
    },
    {
      "epoch": 1.1362622950819672,
      "grad_norm": 1.389633955273701,
      "learning_rate": 0.00016801060021106982,
      "loss": 12.5606,
      "step": 6498
    },
    {
      "epoch": 1.1364371584699453,
      "grad_norm": 0.8139634256366597,
      "learning_rate": 0.0001680002179881072,
      "loss": 12.3185,
      "step": 6499
    },
    {
      "epoch": 1.1366120218579234,
      "grad_norm": 0.7076849573639439,
      "learning_rate": 0.00016798983440153146,
      "loss": 12.5925,
      "step": 6500
    },
    {
      "epoch": 1.1367868852459015,
      "grad_norm": 1.2605796262893993,
      "learning_rate": 0.0001679794494515508,
      "loss": 12.7763,
      "step": 6501
    },
    {
      "epoch": 1.1369617486338797,
      "grad_norm": 0.7665688279709483,
      "learning_rate": 0.00016796906313837343,
      "loss": 12.5368,
      "step": 6502
    },
    {
      "epoch": 1.137136612021858,
      "grad_norm": 0.833487617442899,
      "learning_rate": 0.00016795867546220766,
      "loss": 12.7132,
      "step": 6503
    },
    {
      "epoch": 1.1373114754098361,
      "grad_norm": 0.8649795385952614,
      "learning_rate": 0.00016794828642326178,
      "loss": 12.5761,
      "step": 6504
    },
    {
      "epoch": 1.1374863387978142,
      "grad_norm": 0.7420331058753002,
      "learning_rate": 0.00016793789602174413,
      "loss": 12.6094,
      "step": 6505
    },
    {
      "epoch": 1.1376612021857924,
      "grad_norm": 0.7700587970954932,
      "learning_rate": 0.00016792750425786307,
      "loss": 12.7782,
      "step": 6506
    },
    {
      "epoch": 1.1378360655737705,
      "grad_norm": 0.7837671060556963,
      "learning_rate": 0.000167917111131827,
      "loss": 12.5798,
      "step": 6507
    },
    {
      "epoch": 1.1380109289617486,
      "grad_norm": 0.7908168551842519,
      "learning_rate": 0.00016790671664384429,
      "loss": 12.5097,
      "step": 6508
    },
    {
      "epoch": 1.1381857923497267,
      "grad_norm": 0.7839023537210744,
      "learning_rate": 0.0001678963207941234,
      "loss": 12.4267,
      "step": 6509
    },
    {
      "epoch": 1.1383606557377048,
      "grad_norm": 0.731458553428524,
      "learning_rate": 0.00016788592358287286,
      "loss": 12.4614,
      "step": 6510
    },
    {
      "epoch": 1.138535519125683,
      "grad_norm": 0.7679165200757111,
      "learning_rate": 0.00016787552501030105,
      "loss": 12.2684,
      "step": 6511
    },
    {
      "epoch": 1.1387103825136613,
      "grad_norm": 0.6915891341116619,
      "learning_rate": 0.00016786512507661663,
      "loss": 12.3496,
      "step": 6512
    },
    {
      "epoch": 1.1388852459016394,
      "grad_norm": 0.797301056501388,
      "learning_rate": 0.00016785472378202804,
      "loss": 12.3784,
      "step": 6513
    },
    {
      "epoch": 1.1390601092896175,
      "grad_norm": 0.7338027405504123,
      "learning_rate": 0.00016784432112674388,
      "loss": 12.5636,
      "step": 6514
    },
    {
      "epoch": 1.1392349726775957,
      "grad_norm": 0.8573871836363107,
      "learning_rate": 0.0001678339171109728,
      "loss": 12.433,
      "step": 6515
    },
    {
      "epoch": 1.1394098360655738,
      "grad_norm": 0.7304380650092231,
      "learning_rate": 0.00016782351173492342,
      "loss": 12.7416,
      "step": 6516
    },
    {
      "epoch": 1.139584699453552,
      "grad_norm": 0.9458156537959527,
      "learning_rate": 0.00016781310499880436,
      "loss": 12.641,
      "step": 6517
    },
    {
      "epoch": 1.13975956284153,
      "grad_norm": 0.6942342688414059,
      "learning_rate": 0.0001678026969028243,
      "loss": 12.6247,
      "step": 6518
    },
    {
      "epoch": 1.1399344262295081,
      "grad_norm": 0.9597964490108087,
      "learning_rate": 0.00016779228744719205,
      "loss": 12.6123,
      "step": 6519
    },
    {
      "epoch": 1.1401092896174863,
      "grad_norm": 0.7194258605424871,
      "learning_rate": 0.00016778187663211625,
      "loss": 12.6608,
      "step": 6520
    },
    {
      "epoch": 1.1402841530054646,
      "grad_norm": 1.0201620974046681,
      "learning_rate": 0.00016777146445780573,
      "loss": 12.6648,
      "step": 6521
    },
    {
      "epoch": 1.1404590163934427,
      "grad_norm": 0.7268511172409204,
      "learning_rate": 0.00016776105092446923,
      "loss": 12.7371,
      "step": 6522
    },
    {
      "epoch": 1.1406338797814208,
      "grad_norm": 1.0397886808974197,
      "learning_rate": 0.0001677506360323156,
      "loss": 12.7055,
      "step": 6523
    },
    {
      "epoch": 1.140808743169399,
      "grad_norm": 0.8294122913947918,
      "learning_rate": 0.00016774021978155373,
      "loss": 12.5694,
      "step": 6524
    },
    {
      "epoch": 1.140983606557377,
      "grad_norm": 0.7356414434090225,
      "learning_rate": 0.00016772980217239246,
      "loss": 12.6602,
      "step": 6525
    },
    {
      "epoch": 1.1411584699453552,
      "grad_norm": 0.734274959298731,
      "learning_rate": 0.00016771938320504065,
      "loss": 12.6968,
      "step": 6526
    },
    {
      "epoch": 1.1413333333333333,
      "grad_norm": 0.8676577281562885,
      "learning_rate": 0.00016770896287970734,
      "loss": 12.4293,
      "step": 6527
    },
    {
      "epoch": 1.1415081967213114,
      "grad_norm": 0.7998977980059346,
      "learning_rate": 0.0001676985411966014,
      "loss": 12.8252,
      "step": 6528
    },
    {
      "epoch": 1.1416830601092895,
      "grad_norm": 0.7358788512582095,
      "learning_rate": 0.00016768811815593185,
      "loss": 12.2461,
      "step": 6529
    },
    {
      "epoch": 1.1418579234972677,
      "grad_norm": 0.7728928984546111,
      "learning_rate": 0.0001676776937579077,
      "loss": 12.6967,
      "step": 6530
    },
    {
      "epoch": 1.1420327868852458,
      "grad_norm": 0.9008215680027145,
      "learning_rate": 0.000167667268002738,
      "loss": 12.4871,
      "step": 6531
    },
    {
      "epoch": 1.1422076502732241,
      "grad_norm": 0.7978050392021111,
      "learning_rate": 0.00016765684089063183,
      "loss": 12.4694,
      "step": 6532
    },
    {
      "epoch": 1.1423825136612022,
      "grad_norm": 0.818791706300247,
      "learning_rate": 0.00016764641242179824,
      "loss": 12.532,
      "step": 6533
    },
    {
      "epoch": 1.1425573770491804,
      "grad_norm": 0.7729254221065542,
      "learning_rate": 0.00016763598259644636,
      "loss": 12.3018,
      "step": 6534
    },
    {
      "epoch": 1.1427322404371585,
      "grad_norm": 0.7862621677595681,
      "learning_rate": 0.0001676255514147854,
      "loss": 12.7815,
      "step": 6535
    },
    {
      "epoch": 1.1429071038251366,
      "grad_norm": 0.735138241868105,
      "learning_rate": 0.0001676151188770245,
      "loss": 12.5733,
      "step": 6536
    },
    {
      "epoch": 1.1430819672131147,
      "grad_norm": 0.7736622281585251,
      "learning_rate": 0.00016760468498337283,
      "loss": 12.6,
      "step": 6537
    },
    {
      "epoch": 1.1432568306010928,
      "grad_norm": 0.755408862828583,
      "learning_rate": 0.00016759424973403965,
      "loss": 12.5802,
      "step": 6538
    },
    {
      "epoch": 1.143431693989071,
      "grad_norm": 0.7896004478747218,
      "learning_rate": 0.00016758381312923426,
      "loss": 12.7005,
      "step": 6539
    },
    {
      "epoch": 1.143606557377049,
      "grad_norm": 0.7912918003505135,
      "learning_rate": 0.00016757337516916588,
      "loss": 12.5667,
      "step": 6540
    },
    {
      "epoch": 1.1437814207650274,
      "grad_norm": 0.7008796708925994,
      "learning_rate": 0.00016756293585404384,
      "loss": 12.6598,
      "step": 6541
    },
    {
      "epoch": 1.1439562841530055,
      "grad_norm": 0.7720726922956546,
      "learning_rate": 0.00016755249518407748,
      "loss": 12.5534,
      "step": 6542
    },
    {
      "epoch": 1.1441311475409837,
      "grad_norm": 0.730147695999511,
      "learning_rate": 0.0001675420531594762,
      "loss": 12.4366,
      "step": 6543
    },
    {
      "epoch": 1.1443060109289618,
      "grad_norm": 0.6782006860617723,
      "learning_rate": 0.0001675316097804494,
      "loss": 12.6277,
      "step": 6544
    },
    {
      "epoch": 1.14448087431694,
      "grad_norm": 0.7161068234948703,
      "learning_rate": 0.00016752116504720644,
      "loss": 12.557,
      "step": 6545
    },
    {
      "epoch": 1.144655737704918,
      "grad_norm": 0.7054844358765994,
      "learning_rate": 0.00016751071895995684,
      "loss": 12.4706,
      "step": 6546
    },
    {
      "epoch": 1.1448306010928961,
      "grad_norm": 0.7607152281697297,
      "learning_rate": 0.00016750027151890998,
      "loss": 12.6565,
      "step": 6547
    },
    {
      "epoch": 1.1450054644808743,
      "grad_norm": 0.7217470142350776,
      "learning_rate": 0.00016748982272427547,
      "loss": 12.5376,
      "step": 6548
    },
    {
      "epoch": 1.1451803278688524,
      "grad_norm": 0.7504026203345956,
      "learning_rate": 0.00016747937257626276,
      "loss": 12.6499,
      "step": 6549
    },
    {
      "epoch": 1.1453551912568307,
      "grad_norm": 0.7996659639248487,
      "learning_rate": 0.00016746892107508146,
      "loss": 12.6755,
      "step": 6550
    },
    {
      "epoch": 1.1455300546448088,
      "grad_norm": 0.8169980851409648,
      "learning_rate": 0.00016745846822094115,
      "loss": 12.4743,
      "step": 6551
    },
    {
      "epoch": 1.145704918032787,
      "grad_norm": 0.887407531881815,
      "learning_rate": 0.0001674480140140514,
      "loss": 12.646,
      "step": 6552
    },
    {
      "epoch": 1.145879781420765,
      "grad_norm": 0.7933069930407232,
      "learning_rate": 0.0001674375584546219,
      "loss": 12.6292,
      "step": 6553
    },
    {
      "epoch": 1.1460546448087432,
      "grad_norm": 0.7472675273834078,
      "learning_rate": 0.00016742710154286228,
      "loss": 12.6555,
      "step": 6554
    },
    {
      "epoch": 1.1462295081967213,
      "grad_norm": 0.7842625483791824,
      "learning_rate": 0.00016741664327898223,
      "loss": 12.5425,
      "step": 6555
    },
    {
      "epoch": 1.1464043715846994,
      "grad_norm": 0.7294970623972886,
      "learning_rate": 0.00016740618366319152,
      "loss": 12.2107,
      "step": 6556
    },
    {
      "epoch": 1.1465792349726776,
      "grad_norm": 0.7162586178383142,
      "learning_rate": 0.00016739572269569984,
      "loss": 12.3168,
      "step": 6557
    },
    {
      "epoch": 1.1467540983606557,
      "grad_norm": 0.7798382063081112,
      "learning_rate": 0.000167385260376717,
      "loss": 12.6653,
      "step": 6558
    },
    {
      "epoch": 1.1469289617486338,
      "grad_norm": 0.7756466661676883,
      "learning_rate": 0.00016737479670645276,
      "loss": 12.6987,
      "step": 6559
    },
    {
      "epoch": 1.147103825136612,
      "grad_norm": 0.7244099057178074,
      "learning_rate": 0.000167364331685117,
      "loss": 12.477,
      "step": 6560
    },
    {
      "epoch": 1.1472786885245903,
      "grad_norm": 0.8275667239654266,
      "learning_rate": 0.0001673538653129196,
      "loss": 12.718,
      "step": 6561
    },
    {
      "epoch": 1.1474535519125684,
      "grad_norm": 0.8091457303533968,
      "learning_rate": 0.00016734339759007035,
      "loss": 12.4505,
      "step": 6562
    },
    {
      "epoch": 1.1476284153005465,
      "grad_norm": 0.7491290225188032,
      "learning_rate": 0.0001673329285167792,
      "loss": 12.3864,
      "step": 6563
    },
    {
      "epoch": 1.1478032786885246,
      "grad_norm": 0.8141932170179703,
      "learning_rate": 0.0001673224580932561,
      "loss": 12.6648,
      "step": 6564
    },
    {
      "epoch": 1.1479781420765027,
      "grad_norm": 0.778055154062543,
      "learning_rate": 0.000167311986319711,
      "loss": 12.6395,
      "step": 6565
    },
    {
      "epoch": 1.1481530054644808,
      "grad_norm": 0.7595027420845599,
      "learning_rate": 0.00016730151319635392,
      "loss": 12.1466,
      "step": 6566
    },
    {
      "epoch": 1.148327868852459,
      "grad_norm": 0.723236361910834,
      "learning_rate": 0.00016729103872339485,
      "loss": 12.5227,
      "step": 6567
    },
    {
      "epoch": 1.148502732240437,
      "grad_norm": 0.8048899099748694,
      "learning_rate": 0.00016728056290104385,
      "loss": 12.7775,
      "step": 6568
    },
    {
      "epoch": 1.1486775956284152,
      "grad_norm": 0.7459649571249921,
      "learning_rate": 0.00016727008572951096,
      "loss": 12.6744,
      "step": 6569
    },
    {
      "epoch": 1.1488524590163935,
      "grad_norm": 0.9784941754552238,
      "learning_rate": 0.00016725960720900633,
      "loss": 12.5378,
      "step": 6570
    },
    {
      "epoch": 1.1490273224043717,
      "grad_norm": 0.7624667046415647,
      "learning_rate": 0.00016724912733974005,
      "loss": 12.673,
      "step": 6571
    },
    {
      "epoch": 1.1492021857923498,
      "grad_norm": 0.8507514344026383,
      "learning_rate": 0.0001672386461219223,
      "loss": 12.6727,
      "step": 6572
    },
    {
      "epoch": 1.149377049180328,
      "grad_norm": 0.8144686800359582,
      "learning_rate": 0.00016722816355576323,
      "loss": 12.6122,
      "step": 6573
    },
    {
      "epoch": 1.149551912568306,
      "grad_norm": 0.8290332763987065,
      "learning_rate": 0.00016721767964147306,
      "loss": 12.4603,
      "step": 6574
    },
    {
      "epoch": 1.1497267759562841,
      "grad_norm": 0.7265141134025633,
      "learning_rate": 0.00016720719437926204,
      "loss": 12.6224,
      "step": 6575
    },
    {
      "epoch": 1.1499016393442623,
      "grad_norm": 0.7775682919479004,
      "learning_rate": 0.0001671967077693404,
      "loss": 12.7887,
      "step": 6576
    },
    {
      "epoch": 1.1500765027322404,
      "grad_norm": 0.7962694502134159,
      "learning_rate": 0.00016718621981191847,
      "loss": 12.5668,
      "step": 6577
    },
    {
      "epoch": 1.1502513661202185,
      "grad_norm": 0.7437240794759157,
      "learning_rate": 0.00016717573050720655,
      "loss": 12.548,
      "step": 6578
    },
    {
      "epoch": 1.1504262295081968,
      "grad_norm": 0.7789910256973,
      "learning_rate": 0.00016716523985541498,
      "loss": 12.6029,
      "step": 6579
    },
    {
      "epoch": 1.150601092896175,
      "grad_norm": 0.7420854573154421,
      "learning_rate": 0.0001671547478567541,
      "loss": 12.4122,
      "step": 6580
    },
    {
      "epoch": 1.150775956284153,
      "grad_norm": 0.7209336336230402,
      "learning_rate": 0.00016714425451143435,
      "loss": 12.3421,
      "step": 6581
    },
    {
      "epoch": 1.1509508196721312,
      "grad_norm": 0.7826399396916159,
      "learning_rate": 0.00016713375981966612,
      "loss": 12.5192,
      "step": 6582
    },
    {
      "epoch": 1.1511256830601093,
      "grad_norm": 0.7777731836916923,
      "learning_rate": 0.00016712326378165985,
      "loss": 12.5697,
      "step": 6583
    },
    {
      "epoch": 1.1513005464480874,
      "grad_norm": 0.7183941209530051,
      "learning_rate": 0.0001671127663976261,
      "loss": 12.764,
      "step": 6584
    },
    {
      "epoch": 1.1514754098360656,
      "grad_norm": 0.7788570691524629,
      "learning_rate": 0.00016710226766777531,
      "loss": 12.6398,
      "step": 6585
    },
    {
      "epoch": 1.1516502732240437,
      "grad_norm": 0.7464834833937546,
      "learning_rate": 0.00016709176759231804,
      "loss": 12.6728,
      "step": 6586
    },
    {
      "epoch": 1.1518251366120218,
      "grad_norm": 0.7072382708534496,
      "learning_rate": 0.0001670812661714648,
      "loss": 12.2797,
      "step": 6587
    },
    {
      "epoch": 1.152,
      "grad_norm": 0.9448577021154826,
      "learning_rate": 0.0001670707634054262,
      "loss": 12.7963,
      "step": 6588
    },
    {
      "epoch": 1.152174863387978,
      "grad_norm": 0.8815796586535959,
      "learning_rate": 0.00016706025929441286,
      "loss": 12.5475,
      "step": 6589
    },
    {
      "epoch": 1.1523497267759564,
      "grad_norm": 0.9733717820812143,
      "learning_rate": 0.00016704975383863543,
      "loss": 12.2946,
      "step": 6590
    },
    {
      "epoch": 1.1525245901639345,
      "grad_norm": 0.839011822539395,
      "learning_rate": 0.0001670392470383046,
      "loss": 12.7978,
      "step": 6591
    },
    {
      "epoch": 1.1526994535519126,
      "grad_norm": 0.7942736400620423,
      "learning_rate": 0.000167028738893631,
      "loss": 12.5105,
      "step": 6592
    },
    {
      "epoch": 1.1528743169398907,
      "grad_norm": 0.8767862663903352,
      "learning_rate": 0.00016701822940482536,
      "loss": 12.5715,
      "step": 6593
    },
    {
      "epoch": 1.1530491803278689,
      "grad_norm": 0.7788158530379239,
      "learning_rate": 0.0001670077185720985,
      "loss": 12.5317,
      "step": 6594
    },
    {
      "epoch": 1.153224043715847,
      "grad_norm": 0.7779679250055542,
      "learning_rate": 0.0001669972063956611,
      "loss": 12.5641,
      "step": 6595
    },
    {
      "epoch": 1.153398907103825,
      "grad_norm": 0.7449997632229146,
      "learning_rate": 0.000166986692875724,
      "loss": 12.6276,
      "step": 6596
    },
    {
      "epoch": 1.1535737704918032,
      "grad_norm": 0.7959296824015794,
      "learning_rate": 0.00016697617801249802,
      "loss": 12.5063,
      "step": 6597
    },
    {
      "epoch": 1.1537486338797813,
      "grad_norm": 0.8784150172306877,
      "learning_rate": 0.00016696566180619406,
      "loss": 12.393,
      "step": 6598
    },
    {
      "epoch": 1.1539234972677597,
      "grad_norm": 0.7893345916104823,
      "learning_rate": 0.00016695514425702296,
      "loss": 12.5258,
      "step": 6599
    },
    {
      "epoch": 1.1540983606557378,
      "grad_norm": 0.7259972284732186,
      "learning_rate": 0.00016694462536519561,
      "loss": 12.4831,
      "step": 6600
    },
    {
      "epoch": 1.154273224043716,
      "grad_norm": 0.8181670079600345,
      "learning_rate": 0.00016693410513092305,
      "loss": 12.7775,
      "step": 6601
    },
    {
      "epoch": 1.154448087431694,
      "grad_norm": 0.7175217933237257,
      "learning_rate": 0.00016692358355441608,
      "loss": 12.5409,
      "step": 6602
    },
    {
      "epoch": 1.1546229508196721,
      "grad_norm": 0.7675527019564145,
      "learning_rate": 0.00016691306063588583,
      "loss": 12.5785,
      "step": 6603
    },
    {
      "epoch": 1.1547978142076503,
      "grad_norm": 0.7148190883088889,
      "learning_rate": 0.00016690253637554326,
      "loss": 12.3198,
      "step": 6604
    },
    {
      "epoch": 1.1549726775956284,
      "grad_norm": 0.8224387161511276,
      "learning_rate": 0.0001668920107735994,
      "loss": 12.4214,
      "step": 6605
    },
    {
      "epoch": 1.1551475409836065,
      "grad_norm": 0.7101331823392619,
      "learning_rate": 0.00016688148383026534,
      "loss": 12.6583,
      "step": 6606
    },
    {
      "epoch": 1.1553224043715846,
      "grad_norm": 0.7610855872459723,
      "learning_rate": 0.00016687095554575218,
      "loss": 12.3744,
      "step": 6607
    },
    {
      "epoch": 1.155497267759563,
      "grad_norm": 0.8166752663641077,
      "learning_rate": 0.00016686042592027107,
      "loss": 12.7572,
      "step": 6608
    },
    {
      "epoch": 1.1556721311475409,
      "grad_norm": 0.7921468689489822,
      "learning_rate": 0.00016684989495403308,
      "loss": 12.6346,
      "step": 6609
    },
    {
      "epoch": 1.1558469945355192,
      "grad_norm": 0.7867037689046026,
      "learning_rate": 0.00016683936264724948,
      "loss": 12.6358,
      "step": 6610
    },
    {
      "epoch": 1.1560218579234973,
      "grad_norm": 0.8310502060027246,
      "learning_rate": 0.00016682882900013142,
      "loss": 12.6307,
      "step": 6611
    },
    {
      "epoch": 1.1561967213114754,
      "grad_norm": 0.8416262531683225,
      "learning_rate": 0.00016681829401289013,
      "loss": 12.4241,
      "step": 6612
    },
    {
      "epoch": 1.1563715846994536,
      "grad_norm": 0.8542042322205327,
      "learning_rate": 0.0001668077576857369,
      "loss": 12.5873,
      "step": 6613
    },
    {
      "epoch": 1.1565464480874317,
      "grad_norm": 0.7403744926697752,
      "learning_rate": 0.00016679722001888302,
      "loss": 12.6925,
      "step": 6614
    },
    {
      "epoch": 1.1567213114754098,
      "grad_norm": 0.8850520069647634,
      "learning_rate": 0.00016678668101253975,
      "loss": 12.6727,
      "step": 6615
    },
    {
      "epoch": 1.156896174863388,
      "grad_norm": 0.759426615231939,
      "learning_rate": 0.0001667761406669185,
      "loss": 12.4445,
      "step": 6616
    },
    {
      "epoch": 1.157071038251366,
      "grad_norm": 0.7361352625535043,
      "learning_rate": 0.00016676559898223058,
      "loss": 12.718,
      "step": 6617
    },
    {
      "epoch": 1.1572459016393442,
      "grad_norm": 0.786120617849801,
      "learning_rate": 0.0001667550559586874,
      "loss": 12.508,
      "step": 6618
    },
    {
      "epoch": 1.1574207650273225,
      "grad_norm": 0.7933897145365678,
      "learning_rate": 0.0001667445115965004,
      "loss": 12.5744,
      "step": 6619
    },
    {
      "epoch": 1.1575956284153006,
      "grad_norm": 0.7346842296771033,
      "learning_rate": 0.00016673396589588098,
      "loss": 12.5002,
      "step": 6620
    },
    {
      "epoch": 1.1577704918032787,
      "grad_norm": 0.7866127769445299,
      "learning_rate": 0.00016672341885704065,
      "loss": 12.4855,
      "step": 6621
    },
    {
      "epoch": 1.1579453551912569,
      "grad_norm": 0.7500263559841777,
      "learning_rate": 0.00016671287048019093,
      "loss": 12.4189,
      "step": 6622
    },
    {
      "epoch": 1.158120218579235,
      "grad_norm": 0.8312607633965228,
      "learning_rate": 0.00016670232076554332,
      "loss": 12.6439,
      "step": 6623
    },
    {
      "epoch": 1.158295081967213,
      "grad_norm": 0.7356004206245202,
      "learning_rate": 0.00016669176971330937,
      "loss": 12.4062,
      "step": 6624
    },
    {
      "epoch": 1.1584699453551912,
      "grad_norm": 0.7838760284630857,
      "learning_rate": 0.00016668121732370066,
      "loss": 12.5583,
      "step": 6625
    },
    {
      "epoch": 1.1586448087431693,
      "grad_norm": 0.7391102827825146,
      "learning_rate": 0.0001666706635969288,
      "loss": 12.67,
      "step": 6626
    },
    {
      "epoch": 1.1588196721311474,
      "grad_norm": 0.7495469285933236,
      "learning_rate": 0.00016666010853320543,
      "loss": 12.4348,
      "step": 6627
    },
    {
      "epoch": 1.1589945355191258,
      "grad_norm": 0.7636633232086357,
      "learning_rate": 0.00016664955213274222,
      "loss": 12.4322,
      "step": 6628
    },
    {
      "epoch": 1.159169398907104,
      "grad_norm": 0.7575863034258117,
      "learning_rate": 0.00016663899439575084,
      "loss": 12.5155,
      "step": 6629
    },
    {
      "epoch": 1.159344262295082,
      "grad_norm": 0.7328720987841617,
      "learning_rate": 0.00016662843532244304,
      "loss": 12.6101,
      "step": 6630
    },
    {
      "epoch": 1.1595191256830601,
      "grad_norm": 0.7775822654791267,
      "learning_rate": 0.0001666178749130305,
      "loss": 12.6503,
      "step": 6631
    },
    {
      "epoch": 1.1596939890710383,
      "grad_norm": 0.8033072713075379,
      "learning_rate": 0.00016660731316772505,
      "loss": 12.6749,
      "step": 6632
    },
    {
      "epoch": 1.1598688524590164,
      "grad_norm": 0.7449207734377786,
      "learning_rate": 0.00016659675008673844,
      "loss": 12.6887,
      "step": 6633
    },
    {
      "epoch": 1.1600437158469945,
      "grad_norm": 0.7541183646329539,
      "learning_rate": 0.00016658618567028251,
      "loss": 12.5524,
      "step": 6634
    },
    {
      "epoch": 1.1602185792349726,
      "grad_norm": 0.7374391515528688,
      "learning_rate": 0.0001665756199185691,
      "loss": 12.6406,
      "step": 6635
    },
    {
      "epoch": 1.1603934426229507,
      "grad_norm": 0.7464105973994318,
      "learning_rate": 0.0001665650528318101,
      "loss": 12.5831,
      "step": 6636
    },
    {
      "epoch": 1.160568306010929,
      "grad_norm": 0.8628753599056969,
      "learning_rate": 0.00016655448441021747,
      "loss": 12.5568,
      "step": 6637
    },
    {
      "epoch": 1.160743169398907,
      "grad_norm": 0.7793449636256001,
      "learning_rate": 0.00016654391465400302,
      "loss": 12.6484,
      "step": 6638
    },
    {
      "epoch": 1.1609180327868853,
      "grad_norm": 0.8336236327842049,
      "learning_rate": 0.00016653334356337877,
      "loss": 12.5058,
      "step": 6639
    },
    {
      "epoch": 1.1610928961748634,
      "grad_norm": 0.7471397276212298,
      "learning_rate": 0.0001665227711385567,
      "loss": 12.7608,
      "step": 6640
    },
    {
      "epoch": 1.1612677595628416,
      "grad_norm": 0.8089814052657076,
      "learning_rate": 0.0001665121973797488,
      "loss": 12.3772,
      "step": 6641
    },
    {
      "epoch": 1.1614426229508197,
      "grad_norm": 0.6990413640732843,
      "learning_rate": 0.0001665016222871672,
      "loss": 12.5446,
      "step": 6642
    },
    {
      "epoch": 1.1616174863387978,
      "grad_norm": 0.8242125198533232,
      "learning_rate": 0.00016649104586102383,
      "loss": 12.5381,
      "step": 6643
    },
    {
      "epoch": 1.161792349726776,
      "grad_norm": 0.7542419325497656,
      "learning_rate": 0.0001664804681015308,
      "loss": 12.3949,
      "step": 6644
    },
    {
      "epoch": 1.161967213114754,
      "grad_norm": 0.8579161579542309,
      "learning_rate": 0.0001664698890089003,
      "loss": 12.5079,
      "step": 6645
    },
    {
      "epoch": 1.1621420765027322,
      "grad_norm": 0.8237912920240895,
      "learning_rate": 0.00016645930858334444,
      "loss": 12.6423,
      "step": 6646
    },
    {
      "epoch": 1.1623169398907103,
      "grad_norm": 0.8877457961527883,
      "learning_rate": 0.0001664487268250754,
      "loss": 12.6912,
      "step": 6647
    },
    {
      "epoch": 1.1624918032786886,
      "grad_norm": 0.7821407345459604,
      "learning_rate": 0.00016643814373430534,
      "loss": 12.6398,
      "step": 6648
    },
    {
      "epoch": 1.1626666666666667,
      "grad_norm": 0.727381715313304,
      "learning_rate": 0.0001664275593112465,
      "loss": 12.5976,
      "step": 6649
    },
    {
      "epoch": 1.1628415300546449,
      "grad_norm": 0.8215180218617805,
      "learning_rate": 0.00016641697355611115,
      "loss": 12.4227,
      "step": 6650
    },
    {
      "epoch": 1.163016393442623,
      "grad_norm": 0.747466543931289,
      "learning_rate": 0.0001664063864691116,
      "loss": 12.7462,
      "step": 6651
    },
    {
      "epoch": 1.163191256830601,
      "grad_norm": 0.7608745149481226,
      "learning_rate": 0.00016639579805046,
      "loss": 12.5785,
      "step": 6652
    },
    {
      "epoch": 1.1633661202185792,
      "grad_norm": 0.6993507053193281,
      "learning_rate": 0.00016638520830036888,
      "loss": 12.3582,
      "step": 6653
    },
    {
      "epoch": 1.1635409836065573,
      "grad_norm": 0.7372715274446564,
      "learning_rate": 0.00016637461721905045,
      "loss": 12.5034,
      "step": 6654
    },
    {
      "epoch": 1.1637158469945355,
      "grad_norm": 0.7359670470185271,
      "learning_rate": 0.00016636402480671718,
      "loss": 12.4162,
      "step": 6655
    },
    {
      "epoch": 1.1638907103825136,
      "grad_norm": 0.9679092309724232,
      "learning_rate": 0.00016635343106358142,
      "loss": 12.6406,
      "step": 6656
    },
    {
      "epoch": 1.164065573770492,
      "grad_norm": 0.7701829318776026,
      "learning_rate": 0.00016634283598985564,
      "loss": 12.6004,
      "step": 6657
    },
    {
      "epoch": 1.16424043715847,
      "grad_norm": 0.8293739180851394,
      "learning_rate": 0.00016633223958575233,
      "loss": 12.5927,
      "step": 6658
    },
    {
      "epoch": 1.1644153005464482,
      "grad_norm": 0.7670730972529329,
      "learning_rate": 0.00016632164185148395,
      "loss": 12.6388,
      "step": 6659
    },
    {
      "epoch": 1.1645901639344263,
      "grad_norm": 0.8578632888690321,
      "learning_rate": 0.00016631104278726298,
      "loss": 12.3778,
      "step": 6660
    },
    {
      "epoch": 1.1647650273224044,
      "grad_norm": 0.7540262623662533,
      "learning_rate": 0.00016630044239330204,
      "loss": 12.4946,
      "step": 6661
    },
    {
      "epoch": 1.1649398907103825,
      "grad_norm": 0.8684940757396902,
      "learning_rate": 0.00016628984066981362,
      "loss": 12.6145,
      "step": 6662
    },
    {
      "epoch": 1.1651147540983606,
      "grad_norm": 0.7184757416013907,
      "learning_rate": 0.00016627923761701038,
      "loss": 12.3847,
      "step": 6663
    },
    {
      "epoch": 1.1652896174863387,
      "grad_norm": 0.9776304550404636,
      "learning_rate": 0.0001662686332351049,
      "loss": 12.7061,
      "step": 6664
    },
    {
      "epoch": 1.1654644808743169,
      "grad_norm": 0.7857394209675919,
      "learning_rate": 0.0001662580275243099,
      "loss": 12.5728,
      "step": 6665
    },
    {
      "epoch": 1.165639344262295,
      "grad_norm": 0.7911742637002724,
      "learning_rate": 0.00016624742048483796,
      "loss": 12.4439,
      "step": 6666
    },
    {
      "epoch": 1.165814207650273,
      "grad_norm": 0.7297229822146715,
      "learning_rate": 0.00016623681211690183,
      "loss": 12.5804,
      "step": 6667
    },
    {
      "epoch": 1.1659890710382514,
      "grad_norm": 0.7684062116415836,
      "learning_rate": 0.00016622620242071428,
      "loss": 12.6409,
      "step": 6668
    },
    {
      "epoch": 1.1661639344262296,
      "grad_norm": 0.7974067972966541,
      "learning_rate": 0.000166215591396488,
      "loss": 12.8465,
      "step": 6669
    },
    {
      "epoch": 1.1663387978142077,
      "grad_norm": 0.8013812427979227,
      "learning_rate": 0.00016620497904443584,
      "loss": 12.7145,
      "step": 6670
    },
    {
      "epoch": 1.1665136612021858,
      "grad_norm": 0.7193931398389192,
      "learning_rate": 0.00016619436536477054,
      "loss": 12.3947,
      "step": 6671
    },
    {
      "epoch": 1.166688524590164,
      "grad_norm": 0.7908956703521515,
      "learning_rate": 0.00016618375035770498,
      "loss": 12.5116,
      "step": 6672
    },
    {
      "epoch": 1.166863387978142,
      "grad_norm": 0.8287109200126582,
      "learning_rate": 0.00016617313402345201,
      "loss": 12.5645,
      "step": 6673
    },
    {
      "epoch": 1.1670382513661202,
      "grad_norm": 0.7410032677547891,
      "learning_rate": 0.00016616251636222454,
      "loss": 12.6241,
      "step": 6674
    },
    {
      "epoch": 1.1672131147540983,
      "grad_norm": 0.7989820761370418,
      "learning_rate": 0.00016615189737423548,
      "loss": 12.6,
      "step": 6675
    },
    {
      "epoch": 1.1673879781420764,
      "grad_norm": 0.7365299093655368,
      "learning_rate": 0.00016614127705969777,
      "loss": 12.6569,
      "step": 6676
    },
    {
      "epoch": 1.1675628415300547,
      "grad_norm": 0.7406803270293363,
      "learning_rate": 0.00016613065541882437,
      "loss": 12.5293,
      "step": 6677
    },
    {
      "epoch": 1.1677377049180329,
      "grad_norm": 0.7097329147441515,
      "learning_rate": 0.00016612003245182827,
      "loss": 12.508,
      "step": 6678
    },
    {
      "epoch": 1.167912568306011,
      "grad_norm": 0.8107915707724732,
      "learning_rate": 0.00016610940815892246,
      "loss": 12.6687,
      "step": 6679
    },
    {
      "epoch": 1.168087431693989,
      "grad_norm": 0.7788743325688543,
      "learning_rate": 0.0001660987825403201,
      "loss": 12.6125,
      "step": 6680
    },
    {
      "epoch": 1.1682622950819672,
      "grad_norm": 0.8056572510264555,
      "learning_rate": 0.00016608815559623414,
      "loss": 12.8028,
      "step": 6681
    },
    {
      "epoch": 1.1684371584699453,
      "grad_norm": 0.7577911798140388,
      "learning_rate": 0.0001660775273268778,
      "loss": 12.6544,
      "step": 6682
    },
    {
      "epoch": 1.1686120218579235,
      "grad_norm": 0.7561953358505743,
      "learning_rate": 0.00016606689773246414,
      "loss": 12.7822,
      "step": 6683
    },
    {
      "epoch": 1.1687868852459016,
      "grad_norm": 0.7428702914547753,
      "learning_rate": 0.0001660562668132063,
      "loss": 12.4034,
      "step": 6684
    },
    {
      "epoch": 1.1689617486338797,
      "grad_norm": 0.7363894342985561,
      "learning_rate": 0.00016604563456931753,
      "loss": 12.3921,
      "step": 6685
    },
    {
      "epoch": 1.169136612021858,
      "grad_norm": 0.7902500311574121,
      "learning_rate": 0.00016603500100101094,
      "loss": 12.6343,
      "step": 6686
    },
    {
      "epoch": 1.1693114754098362,
      "grad_norm": 0.7109428832485059,
      "learning_rate": 0.00016602436610849987,
      "loss": 12.4812,
      "step": 6687
    },
    {
      "epoch": 1.1694863387978143,
      "grad_norm": 0.7559449225508796,
      "learning_rate": 0.00016601372989199755,
      "loss": 12.5856,
      "step": 6688
    },
    {
      "epoch": 1.1696612021857924,
      "grad_norm": 0.7153967663599965,
      "learning_rate": 0.00016600309235171725,
      "loss": 12.4716,
      "step": 6689
    },
    {
      "epoch": 1.1698360655737705,
      "grad_norm": 0.7727282013601982,
      "learning_rate": 0.0001659924534878723,
      "loss": 12.7372,
      "step": 6690
    },
    {
      "epoch": 1.1700109289617486,
      "grad_norm": 0.7480800633141959,
      "learning_rate": 0.000165981813300676,
      "loss": 12.5735,
      "step": 6691
    },
    {
      "epoch": 1.1701857923497267,
      "grad_norm": 0.8627877601020502,
      "learning_rate": 0.00016597117179034177,
      "loss": 12.6492,
      "step": 6692
    },
    {
      "epoch": 1.1703606557377049,
      "grad_norm": 0.6717948469136243,
      "learning_rate": 0.00016596052895708296,
      "loss": 12.2646,
      "step": 6693
    },
    {
      "epoch": 1.170535519125683,
      "grad_norm": 0.7639352196005473,
      "learning_rate": 0.00016594988480111305,
      "loss": 12.4369,
      "step": 6694
    },
    {
      "epoch": 1.170710382513661,
      "grad_norm": 0.7679314909426914,
      "learning_rate": 0.00016593923932264545,
      "loss": 12.7665,
      "step": 6695
    },
    {
      "epoch": 1.1708852459016392,
      "grad_norm": 0.775201660945774,
      "learning_rate": 0.00016592859252189364,
      "loss": 12.5403,
      "step": 6696
    },
    {
      "epoch": 1.1710601092896176,
      "grad_norm": 0.738655452495324,
      "learning_rate": 0.00016591794439907114,
      "loss": 12.6013,
      "step": 6697
    },
    {
      "epoch": 1.1712349726775957,
      "grad_norm": 0.8635051078994095,
      "learning_rate": 0.00016590729495439145,
      "loss": 12.5731,
      "step": 6698
    },
    {
      "epoch": 1.1714098360655738,
      "grad_norm": 0.7946146980994491,
      "learning_rate": 0.00016589664418806814,
      "loss": 12.5735,
      "step": 6699
    },
    {
      "epoch": 1.171584699453552,
      "grad_norm": 0.7522390609952232,
      "learning_rate": 0.00016588599210031477,
      "loss": 12.5859,
      "step": 6700
    },
    {
      "epoch": 1.17175956284153,
      "grad_norm": 0.7800170296643114,
      "learning_rate": 0.00016587533869134497,
      "loss": 12.6156,
      "step": 6701
    },
    {
      "epoch": 1.1719344262295082,
      "grad_norm": 0.7803496492697006,
      "learning_rate": 0.00016586468396137237,
      "loss": 12.3409,
      "step": 6702
    },
    {
      "epoch": 1.1721092896174863,
      "grad_norm": 0.7774388643850654,
      "learning_rate": 0.0001658540279106106,
      "loss": 12.7658,
      "step": 6703
    },
    {
      "epoch": 1.1722841530054644,
      "grad_norm": 0.7386680560678839,
      "learning_rate": 0.00016584337053927338,
      "loss": 12.4243,
      "step": 6704
    },
    {
      "epoch": 1.1724590163934425,
      "grad_norm": 0.7912227432374296,
      "learning_rate": 0.00016583271184757444,
      "loss": 12.7201,
      "step": 6705
    },
    {
      "epoch": 1.1726338797814209,
      "grad_norm": 0.7066663176235323,
      "learning_rate": 0.00016582205183572748,
      "loss": 12.5217,
      "step": 6706
    },
    {
      "epoch": 1.172808743169399,
      "grad_norm": 0.7761813708660358,
      "learning_rate": 0.00016581139050394627,
      "loss": 12.5727,
      "step": 6707
    },
    {
      "epoch": 1.172983606557377,
      "grad_norm": 0.7414318594289562,
      "learning_rate": 0.00016580072785244462,
      "loss": 12.5636,
      "step": 6708
    },
    {
      "epoch": 1.1731584699453552,
      "grad_norm": 0.7522705892906096,
      "learning_rate": 0.00016579006388143635,
      "loss": 12.4163,
      "step": 6709
    },
    {
      "epoch": 1.1733333333333333,
      "grad_norm": 0.7852021360285071,
      "learning_rate": 0.0001657793985911353,
      "loss": 12.6911,
      "step": 6710
    },
    {
      "epoch": 1.1735081967213115,
      "grad_norm": 0.7387602731885761,
      "learning_rate": 0.00016576873198175532,
      "loss": 12.5849,
      "step": 6711
    },
    {
      "epoch": 1.1736830601092896,
      "grad_norm": 0.7796364584409373,
      "learning_rate": 0.00016575806405351036,
      "loss": 12.6904,
      "step": 6712
    },
    {
      "epoch": 1.1738579234972677,
      "grad_norm": 0.7486167697889411,
      "learning_rate": 0.00016574739480661426,
      "loss": 12.6223,
      "step": 6713
    },
    {
      "epoch": 1.1740327868852458,
      "grad_norm": 0.7870896321520787,
      "learning_rate": 0.00016573672424128106,
      "loss": 12.7633,
      "step": 6714
    },
    {
      "epoch": 1.1742076502732242,
      "grad_norm": 0.7073982344471331,
      "learning_rate": 0.0001657260523577247,
      "loss": 12.1957,
      "step": 6715
    },
    {
      "epoch": 1.1743825136612023,
      "grad_norm": 0.7922688268557291,
      "learning_rate": 0.0001657153791561592,
      "loss": 12.6287,
      "step": 6716
    },
    {
      "epoch": 1.1745573770491804,
      "grad_norm": 0.8075170107677491,
      "learning_rate": 0.00016570470463679856,
      "loss": 12.7563,
      "step": 6717
    },
    {
      "epoch": 1.1747322404371585,
      "grad_norm": 0.7524174216414993,
      "learning_rate": 0.00016569402879985682,
      "loss": 12.5681,
      "step": 6718
    },
    {
      "epoch": 1.1749071038251366,
      "grad_norm": 0.7860186461137487,
      "learning_rate": 0.00016568335164554812,
      "loss": 12.4771,
      "step": 6719
    },
    {
      "epoch": 1.1750819672131148,
      "grad_norm": 0.8068091287093835,
      "learning_rate": 0.00016567267317408655,
      "loss": 12.6243,
      "step": 6720
    },
    {
      "epoch": 1.1752568306010929,
      "grad_norm": 0.7446110549364935,
      "learning_rate": 0.00016566199338568624,
      "loss": 12.5707,
      "step": 6721
    },
    {
      "epoch": 1.175431693989071,
      "grad_norm": 0.8098009626081671,
      "learning_rate": 0.00016565131228056133,
      "loss": 12.341,
      "step": 6722
    },
    {
      "epoch": 1.175606557377049,
      "grad_norm": 0.6906208849693269,
      "learning_rate": 0.00016564062985892608,
      "loss": 12.5231,
      "step": 6723
    },
    {
      "epoch": 1.1757814207650272,
      "grad_norm": 0.8508344122982832,
      "learning_rate": 0.00016562994612099466,
      "loss": 12.6449,
      "step": 6724
    },
    {
      "epoch": 1.1759562841530053,
      "grad_norm": 0.7835037546030017,
      "learning_rate": 0.00016561926106698125,
      "loss": 12.5534,
      "step": 6725
    },
    {
      "epoch": 1.1761311475409837,
      "grad_norm": 0.8362776955660876,
      "learning_rate": 0.00016560857469710022,
      "loss": 12.4993,
      "step": 6726
    },
    {
      "epoch": 1.1763060109289618,
      "grad_norm": 0.7547002534902095,
      "learning_rate": 0.0001655978870115658,
      "loss": 12.2041,
      "step": 6727
    },
    {
      "epoch": 1.17648087431694,
      "grad_norm": 0.9516846728041717,
      "learning_rate": 0.00016558719801059235,
      "loss": 12.587,
      "step": 6728
    },
    {
      "epoch": 1.176655737704918,
      "grad_norm": 0.7395511113962021,
      "learning_rate": 0.0001655765076943942,
      "loss": 12.4735,
      "step": 6729
    },
    {
      "epoch": 1.1768306010928962,
      "grad_norm": 0.9284589201749388,
      "learning_rate": 0.0001655658160631857,
      "loss": 12.5992,
      "step": 6730
    },
    {
      "epoch": 1.1770054644808743,
      "grad_norm": 0.8186134545138438,
      "learning_rate": 0.00016555512311718134,
      "loss": 12.7076,
      "step": 6731
    },
    {
      "epoch": 1.1771803278688524,
      "grad_norm": 0.79606669030355,
      "learning_rate": 0.0001655444288565954,
      "loss": 12.3727,
      "step": 6732
    },
    {
      "epoch": 1.1773551912568305,
      "grad_norm": 0.7065039176399033,
      "learning_rate": 0.00016553373328164245,
      "loss": 12.497,
      "step": 6733
    },
    {
      "epoch": 1.1775300546448086,
      "grad_norm": 0.9204974587238829,
      "learning_rate": 0.00016552303639253695,
      "loss": 12.3272,
      "step": 6734
    },
    {
      "epoch": 1.177704918032787,
      "grad_norm": 0.7766621233529146,
      "learning_rate": 0.00016551233818949337,
      "loss": 12.3952,
      "step": 6735
    },
    {
      "epoch": 1.177879781420765,
      "grad_norm": 0.7884237791697191,
      "learning_rate": 0.00016550163867272625,
      "loss": 12.2401,
      "step": 6736
    },
    {
      "epoch": 1.1780546448087432,
      "grad_norm": 0.8543056293965783,
      "learning_rate": 0.00016549093784245017,
      "loss": 12.7348,
      "step": 6737
    },
    {
      "epoch": 1.1782295081967213,
      "grad_norm": 0.849823704277141,
      "learning_rate": 0.0001654802356988797,
      "loss": 12.6292,
      "step": 6738
    },
    {
      "epoch": 1.1784043715846995,
      "grad_norm": 0.7887752183208424,
      "learning_rate": 0.00016546953224222944,
      "loss": 12.6744,
      "step": 6739
    },
    {
      "epoch": 1.1785792349726776,
      "grad_norm": 0.8771746504969974,
      "learning_rate": 0.00016545882747271407,
      "loss": 12.5206,
      "step": 6740
    },
    {
      "epoch": 1.1787540983606557,
      "grad_norm": 0.7673725901205289,
      "learning_rate": 0.00016544812139054823,
      "loss": 12.4052,
      "step": 6741
    },
    {
      "epoch": 1.1789289617486338,
      "grad_norm": 0.7988972541352887,
      "learning_rate": 0.00016543741399594657,
      "loss": 12.5286,
      "step": 6742
    },
    {
      "epoch": 1.179103825136612,
      "grad_norm": 0.754751079650605,
      "learning_rate": 0.00016542670528912385,
      "loss": 12.6961,
      "step": 6743
    },
    {
      "epoch": 1.1792786885245903,
      "grad_norm": 0.728737699754667,
      "learning_rate": 0.0001654159952702948,
      "loss": 12.5568,
      "step": 6744
    },
    {
      "epoch": 1.1794535519125684,
      "grad_norm": 0.7241019552352239,
      "learning_rate": 0.00016540528393967421,
      "loss": 12.6095,
      "step": 6745
    },
    {
      "epoch": 1.1796284153005465,
      "grad_norm": 0.7190793845390427,
      "learning_rate": 0.00016539457129747685,
      "loss": 12.5934,
      "step": 6746
    },
    {
      "epoch": 1.1798032786885246,
      "grad_norm": 0.7313827516914707,
      "learning_rate": 0.00016538385734391755,
      "loss": 12.4988,
      "step": 6747
    },
    {
      "epoch": 1.1799781420765028,
      "grad_norm": 0.833747767918475,
      "learning_rate": 0.00016537314207921115,
      "loss": 12.5909,
      "step": 6748
    },
    {
      "epoch": 1.1801530054644809,
      "grad_norm": 0.9602741279977989,
      "learning_rate": 0.00016536242550357251,
      "loss": 12.4447,
      "step": 6749
    },
    {
      "epoch": 1.180327868852459,
      "grad_norm": 0.723292011612479,
      "learning_rate": 0.00016535170761721658,
      "loss": 12.4321,
      "step": 6750
    },
    {
      "epoch": 1.1805027322404371,
      "grad_norm": 0.8938703525848496,
      "learning_rate": 0.00016534098842035824,
      "loss": 12.6838,
      "step": 6751
    },
    {
      "epoch": 1.1806775956284152,
      "grad_norm": 0.7584827100162905,
      "learning_rate": 0.00016533026791321247,
      "loss": 12.6396,
      "step": 6752
    },
    {
      "epoch": 1.1808524590163934,
      "grad_norm": 0.8732055414946762,
      "learning_rate": 0.0001653195460959942,
      "loss": 12.5731,
      "step": 6753
    },
    {
      "epoch": 1.1810273224043715,
      "grad_norm": 0.8335081667620271,
      "learning_rate": 0.00016530882296891851,
      "loss": 12.6193,
      "step": 6754
    },
    {
      "epoch": 1.1812021857923498,
      "grad_norm": 1.1392564308595432,
      "learning_rate": 0.00016529809853220039,
      "loss": 12.7417,
      "step": 6755
    },
    {
      "epoch": 1.181377049180328,
      "grad_norm": 0.8189750224921435,
      "learning_rate": 0.00016528737278605492,
      "loss": 12.5429,
      "step": 6756
    },
    {
      "epoch": 1.181551912568306,
      "grad_norm": 1.0777257434427638,
      "learning_rate": 0.00016527664573069717,
      "loss": 12.4864,
      "step": 6757
    },
    {
      "epoch": 1.1817267759562842,
      "grad_norm": 0.7430600787472587,
      "learning_rate": 0.0001652659173663422,
      "loss": 12.6544,
      "step": 6758
    },
    {
      "epoch": 1.1819016393442623,
      "grad_norm": 0.9149457958124525,
      "learning_rate": 0.00016525518769320522,
      "loss": 12.4155,
      "step": 6759
    },
    {
      "epoch": 1.1820765027322404,
      "grad_norm": 0.7706473723716063,
      "learning_rate": 0.00016524445671150137,
      "loss": 12.6457,
      "step": 6760
    },
    {
      "epoch": 1.1822513661202185,
      "grad_norm": 0.7422002458941647,
      "learning_rate": 0.00016523372442144583,
      "loss": 12.6462,
      "step": 6761
    },
    {
      "epoch": 1.1824262295081966,
      "grad_norm": 0.8159010803882148,
      "learning_rate": 0.00016522299082325382,
      "loss": 12.6008,
      "step": 6762
    },
    {
      "epoch": 1.1826010928961748,
      "grad_norm": 0.7485950693327065,
      "learning_rate": 0.0001652122559171406,
      "loss": 12.643,
      "step": 6763
    },
    {
      "epoch": 1.182775956284153,
      "grad_norm": 0.769628969429702,
      "learning_rate": 0.0001652015197033214,
      "loss": 12.5593,
      "step": 6764
    },
    {
      "epoch": 1.1829508196721312,
      "grad_norm": 0.7045003524064075,
      "learning_rate": 0.00016519078218201155,
      "loss": 12.5067,
      "step": 6765
    },
    {
      "epoch": 1.1831256830601093,
      "grad_norm": 0.7124316859934231,
      "learning_rate": 0.00016518004335342633,
      "loss": 12.7514,
      "step": 6766
    },
    {
      "epoch": 1.1833005464480875,
      "grad_norm": 0.7487477971356362,
      "learning_rate": 0.00016516930321778116,
      "loss": 12.4484,
      "step": 6767
    },
    {
      "epoch": 1.1834754098360656,
      "grad_norm": 0.7672115376952974,
      "learning_rate": 0.00016515856177529132,
      "loss": 12.6344,
      "step": 6768
    },
    {
      "epoch": 1.1836502732240437,
      "grad_norm": 0.7286820495468814,
      "learning_rate": 0.00016514781902617228,
      "loss": 12.6278,
      "step": 6769
    },
    {
      "epoch": 1.1838251366120218,
      "grad_norm": 0.7352433493833095,
      "learning_rate": 0.0001651370749706394,
      "loss": 12.672,
      "step": 6770
    },
    {
      "epoch": 1.184,
      "grad_norm": 0.740481429421829,
      "learning_rate": 0.00016512632960890823,
      "loss": 12.4395,
      "step": 6771
    },
    {
      "epoch": 1.184174863387978,
      "grad_norm": 0.808948396493167,
      "learning_rate": 0.00016511558294119417,
      "loss": 12.7074,
      "step": 6772
    },
    {
      "epoch": 1.1843497267759564,
      "grad_norm": 0.798160047426565,
      "learning_rate": 0.0001651048349677127,
      "loss": 12.5821,
      "step": 6773
    },
    {
      "epoch": 1.1845245901639345,
      "grad_norm": 0.7492405181217981,
      "learning_rate": 0.00016509408568867942,
      "loss": 12.6462,
      "step": 6774
    },
    {
      "epoch": 1.1846994535519126,
      "grad_norm": 0.9519924869556979,
      "learning_rate": 0.00016508333510430987,
      "loss": 12.3305,
      "step": 6775
    },
    {
      "epoch": 1.1848743169398908,
      "grad_norm": 0.720751264761372,
      "learning_rate": 0.00016507258321481959,
      "loss": 12.3454,
      "step": 6776
    },
    {
      "epoch": 1.1850491803278689,
      "grad_norm": 0.902588479209206,
      "learning_rate": 0.0001650618300204242,
      "loss": 12.7141,
      "step": 6777
    },
    {
      "epoch": 1.185224043715847,
      "grad_norm": 0.8298051308481366,
      "learning_rate": 0.00016505107552133942,
      "loss": 12.7915,
      "step": 6778
    },
    {
      "epoch": 1.1853989071038251,
      "grad_norm": 0.6958618781554229,
      "learning_rate": 0.0001650403197177808,
      "loss": 12.6517,
      "step": 6779
    },
    {
      "epoch": 1.1855737704918032,
      "grad_norm": 0.7819625009981525,
      "learning_rate": 0.00016502956260996404,
      "loss": 12.6166,
      "step": 6780
    },
    {
      "epoch": 1.1857486338797814,
      "grad_norm": 0.7411297950093689,
      "learning_rate": 0.00016501880419810493,
      "loss": 12.5838,
      "step": 6781
    },
    {
      "epoch": 1.1859234972677595,
      "grad_norm": 0.7566065489480303,
      "learning_rate": 0.00016500804448241915,
      "loss": 12.3981,
      "step": 6782
    },
    {
      "epoch": 1.1860983606557376,
      "grad_norm": 0.881191862667335,
      "learning_rate": 0.00016499728346312247,
      "loss": 12.5862,
      "step": 6783
    },
    {
      "epoch": 1.186273224043716,
      "grad_norm": 0.7336905516245288,
      "learning_rate": 0.00016498652114043072,
      "loss": 12.5505,
      "step": 6784
    },
    {
      "epoch": 1.186448087431694,
      "grad_norm": 0.8471940371667123,
      "learning_rate": 0.00016497575751455964,
      "loss": 12.7525,
      "step": 6785
    },
    {
      "epoch": 1.1866229508196722,
      "grad_norm": 0.8030042610402931,
      "learning_rate": 0.00016496499258572515,
      "loss": 12.4463,
      "step": 6786
    },
    {
      "epoch": 1.1867978142076503,
      "grad_norm": 0.803394477620318,
      "learning_rate": 0.0001649542263541431,
      "loss": 12.6836,
      "step": 6787
    },
    {
      "epoch": 1.1869726775956284,
      "grad_norm": 0.8465956361245639,
      "learning_rate": 0.00016494345882002933,
      "loss": 12.5359,
      "step": 6788
    },
    {
      "epoch": 1.1871475409836065,
      "grad_norm": 0.7094078034530955,
      "learning_rate": 0.00016493268998359986,
      "loss": 12.7887,
      "step": 6789
    },
    {
      "epoch": 1.1873224043715846,
      "grad_norm": 0.8111075764162519,
      "learning_rate": 0.00016492191984507056,
      "loss": 12.5622,
      "step": 6790
    },
    {
      "epoch": 1.1874972677595628,
      "grad_norm": 0.8109703677771031,
      "learning_rate": 0.0001649111484046574,
      "loss": 12.6267,
      "step": 6791
    },
    {
      "epoch": 1.1876721311475409,
      "grad_norm": 0.8727648920886851,
      "learning_rate": 0.00016490037566257646,
      "loss": 12.6982,
      "step": 6792
    },
    {
      "epoch": 1.1878469945355192,
      "grad_norm": 0.7131773547746563,
      "learning_rate": 0.00016488960161904369,
      "loss": 12.6396,
      "step": 6793
    },
    {
      "epoch": 1.1880218579234973,
      "grad_norm": 0.8075725986300467,
      "learning_rate": 0.00016487882627427519,
      "loss": 12.4312,
      "step": 6794
    },
    {
      "epoch": 1.1881967213114755,
      "grad_norm": 0.7118981228311386,
      "learning_rate": 0.000164868049628487,
      "loss": 12.5719,
      "step": 6795
    },
    {
      "epoch": 1.1883715846994536,
      "grad_norm": 0.7404230460898807,
      "learning_rate": 0.0001648572716818953,
      "loss": 12.4555,
      "step": 6796
    },
    {
      "epoch": 1.1885464480874317,
      "grad_norm": 0.7645840927731752,
      "learning_rate": 0.0001648464924347161,
      "loss": 12.605,
      "step": 6797
    },
    {
      "epoch": 1.1887213114754098,
      "grad_norm": 0.6859131978861891,
      "learning_rate": 0.00016483571188716562,
      "loss": 12.4606,
      "step": 6798
    },
    {
      "epoch": 1.188896174863388,
      "grad_norm": 0.8020008107635005,
      "learning_rate": 0.00016482493003946007,
      "loss": 12.6458,
      "step": 6799
    },
    {
      "epoch": 1.189071038251366,
      "grad_norm": 0.717585457281477,
      "learning_rate": 0.00016481414689181563,
      "loss": 12.4747,
      "step": 6800
    },
    {
      "epoch": 1.1892459016393442,
      "grad_norm": 0.7522246380476684,
      "learning_rate": 0.00016480336244444855,
      "loss": 12.5484,
      "step": 6801
    },
    {
      "epoch": 1.1894207650273225,
      "grad_norm": 0.7554325048946869,
      "learning_rate": 0.00016479257669757503,
      "loss": 12.4359,
      "step": 6802
    },
    {
      "epoch": 1.1895956284153006,
      "grad_norm": 0.7204014419510576,
      "learning_rate": 0.0001647817896514115,
      "loss": 12.8528,
      "step": 6803
    },
    {
      "epoch": 1.1897704918032788,
      "grad_norm": 0.7870854267459398,
      "learning_rate": 0.0001647710013061741,
      "loss": 12.6569,
      "step": 6804
    },
    {
      "epoch": 1.1899453551912569,
      "grad_norm": 0.7346474351447713,
      "learning_rate": 0.00016476021166207926,
      "loss": 12.6422,
      "step": 6805
    },
    {
      "epoch": 1.190120218579235,
      "grad_norm": 0.7502315916158157,
      "learning_rate": 0.00016474942071934337,
      "loss": 12.4204,
      "step": 6806
    },
    {
      "epoch": 1.1902950819672131,
      "grad_norm": 0.8057005662603591,
      "learning_rate": 0.00016473862847818277,
      "loss": 12.5502,
      "step": 6807
    },
    {
      "epoch": 1.1904699453551912,
      "grad_norm": 0.7674071163508234,
      "learning_rate": 0.0001647278349388139,
      "loss": 12.4275,
      "step": 6808
    },
    {
      "epoch": 1.1906448087431694,
      "grad_norm": 0.8102745937439434,
      "learning_rate": 0.00016471704010145318,
      "loss": 12.5759,
      "step": 6809
    },
    {
      "epoch": 1.1908196721311475,
      "grad_norm": 0.8003413850816012,
      "learning_rate": 0.00016470624396631712,
      "loss": 12.773,
      "step": 6810
    },
    {
      "epoch": 1.1909945355191256,
      "grad_norm": 0.7356653915356693,
      "learning_rate": 0.00016469544653362222,
      "loss": 12.7436,
      "step": 6811
    },
    {
      "epoch": 1.1911693989071037,
      "grad_norm": 0.8735464511238309,
      "learning_rate": 0.00016468464780358493,
      "loss": 12.7339,
      "step": 6812
    },
    {
      "epoch": 1.191344262295082,
      "grad_norm": 0.7824108815352638,
      "learning_rate": 0.00016467384777642185,
      "loss": 12.3919,
      "step": 6813
    },
    {
      "epoch": 1.1915191256830602,
      "grad_norm": 0.7751413196583746,
      "learning_rate": 0.00016466304645234957,
      "loss": 12.5105,
      "step": 6814
    },
    {
      "epoch": 1.1916939890710383,
      "grad_norm": 0.7477706933475566,
      "learning_rate": 0.00016465224383158464,
      "loss": 12.5873,
      "step": 6815
    },
    {
      "epoch": 1.1918688524590164,
      "grad_norm": 0.7955654232681544,
      "learning_rate": 0.00016464143991434373,
      "loss": 12.5204,
      "step": 6816
    },
    {
      "epoch": 1.1920437158469945,
      "grad_norm": 0.8053828985530173,
      "learning_rate": 0.00016463063470084348,
      "loss": 12.6626,
      "step": 6817
    },
    {
      "epoch": 1.1922185792349727,
      "grad_norm": 0.6481802066752363,
      "learning_rate": 0.00016461982819130055,
      "loss": 12.5472,
      "step": 6818
    },
    {
      "epoch": 1.1923934426229508,
      "grad_norm": 0.8302920403406188,
      "learning_rate": 0.00016460902038593165,
      "loss": 12.6405,
      "step": 6819
    },
    {
      "epoch": 1.192568306010929,
      "grad_norm": 0.8239505529169676,
      "learning_rate": 0.00016459821128495355,
      "loss": 12.803,
      "step": 6820
    },
    {
      "epoch": 1.192743169398907,
      "grad_norm": 0.7477126831906483,
      "learning_rate": 0.00016458740088858294,
      "loss": 12.5755,
      "step": 6821
    },
    {
      "epoch": 1.1929180327868854,
      "grad_norm": 0.846917893852084,
      "learning_rate": 0.00016457658919703664,
      "loss": 12.7355,
      "step": 6822
    },
    {
      "epoch": 1.1930928961748635,
      "grad_norm": 0.7321321732616772,
      "learning_rate": 0.00016456577621053143,
      "loss": 12.6554,
      "step": 6823
    },
    {
      "epoch": 1.1932677595628416,
      "grad_norm": 1.0078183581269562,
      "learning_rate": 0.0001645549619292842,
      "loss": 12.6366,
      "step": 6824
    },
    {
      "epoch": 1.1934426229508197,
      "grad_norm": 0.7956808673547938,
      "learning_rate": 0.00016454414635351175,
      "loss": 12.3727,
      "step": 6825
    },
    {
      "epoch": 1.1936174863387978,
      "grad_norm": 1.0840665158336613,
      "learning_rate": 0.00016453332948343098,
      "loss": 12.5705,
      "step": 6826
    },
    {
      "epoch": 1.193792349726776,
      "grad_norm": 0.9115703395926932,
      "learning_rate": 0.00016452251131925883,
      "loss": 12.6209,
      "step": 6827
    },
    {
      "epoch": 1.193967213114754,
      "grad_norm": 0.7085554447137105,
      "learning_rate": 0.00016451169186121223,
      "loss": 12.6756,
      "step": 6828
    },
    {
      "epoch": 1.1941420765027322,
      "grad_norm": 0.7723707570361394,
      "learning_rate": 0.0001645008711095081,
      "loss": 12.582,
      "step": 6829
    },
    {
      "epoch": 1.1943169398907103,
      "grad_norm": 0.7244886236389011,
      "learning_rate": 0.00016449004906436346,
      "loss": 12.5007,
      "step": 6830
    },
    {
      "epoch": 1.1944918032786886,
      "grad_norm": 0.7296913815003998,
      "learning_rate": 0.00016447922572599536,
      "loss": 12.4535,
      "step": 6831
    },
    {
      "epoch": 1.1946666666666665,
      "grad_norm": 0.8106861261676924,
      "learning_rate": 0.00016446840109462079,
      "loss": 12.6728,
      "step": 6832
    },
    {
      "epoch": 1.1948415300546449,
      "grad_norm": 0.7756025818191061,
      "learning_rate": 0.00016445757517045682,
      "loss": 12.3546,
      "step": 6833
    },
    {
      "epoch": 1.195016393442623,
      "grad_norm": 0.7726198293738804,
      "learning_rate": 0.00016444674795372058,
      "loss": 12.7169,
      "step": 6834
    },
    {
      "epoch": 1.1951912568306011,
      "grad_norm": 0.8610850138473646,
      "learning_rate": 0.00016443591944462915,
      "loss": 12.4764,
      "step": 6835
    },
    {
      "epoch": 1.1953661202185792,
      "grad_norm": 0.827260168927397,
      "learning_rate": 0.00016442508964339972,
      "loss": 12.39,
      "step": 6836
    },
    {
      "epoch": 1.1955409836065574,
      "grad_norm": 0.7900539296523496,
      "learning_rate": 0.0001644142585502494,
      "loss": 12.7001,
      "step": 6837
    },
    {
      "epoch": 1.1957158469945355,
      "grad_norm": 0.7585234415373696,
      "learning_rate": 0.0001644034261653954,
      "loss": 12.4009,
      "step": 6838
    },
    {
      "epoch": 1.1958907103825136,
      "grad_norm": 0.8950686122068143,
      "learning_rate": 0.00016439259248905502,
      "loss": 12.6899,
      "step": 6839
    },
    {
      "epoch": 1.1960655737704917,
      "grad_norm": 0.7616304824696624,
      "learning_rate": 0.00016438175752144538,
      "loss": 12.6044,
      "step": 6840
    },
    {
      "epoch": 1.1962404371584698,
      "grad_norm": 0.7471189992146859,
      "learning_rate": 0.0001643709212627839,
      "loss": 12.3828,
      "step": 6841
    },
    {
      "epoch": 1.1964153005464482,
      "grad_norm": 0.7516816948208743,
      "learning_rate": 0.00016436008371328774,
      "loss": 12.3703,
      "step": 6842
    },
    {
      "epoch": 1.1965901639344263,
      "grad_norm": 0.8403769282827472,
      "learning_rate": 0.0001643492448731743,
      "loss": 12.7075,
      "step": 6843
    },
    {
      "epoch": 1.1967650273224044,
      "grad_norm": 0.8350589852076157,
      "learning_rate": 0.00016433840474266095,
      "loss": 12.7478,
      "step": 6844
    },
    {
      "epoch": 1.1969398907103825,
      "grad_norm": 0.7744731269484328,
      "learning_rate": 0.00016432756332196503,
      "loss": 12.4755,
      "step": 6845
    },
    {
      "epoch": 1.1971147540983607,
      "grad_norm": 0.8015319578686272,
      "learning_rate": 0.00016431672061130396,
      "loss": 12.477,
      "step": 6846
    },
    {
      "epoch": 1.1972896174863388,
      "grad_norm": 0.7293690837161964,
      "learning_rate": 0.00016430587661089513,
      "loss": 12.5076,
      "step": 6847
    },
    {
      "epoch": 1.197464480874317,
      "grad_norm": 0.8879042660833416,
      "learning_rate": 0.00016429503132095607,
      "loss": 12.6647,
      "step": 6848
    },
    {
      "epoch": 1.197639344262295,
      "grad_norm": 0.8355659539461953,
      "learning_rate": 0.0001642841847417042,
      "loss": 12.523,
      "step": 6849
    },
    {
      "epoch": 1.1978142076502731,
      "grad_norm": 0.8403326404793676,
      "learning_rate": 0.00016427333687335705,
      "loss": 12.6821,
      "step": 6850
    },
    {
      "epoch": 1.1979890710382515,
      "grad_norm": 0.907575944905921,
      "learning_rate": 0.00016426248771613216,
      "loss": 12.3189,
      "step": 6851
    },
    {
      "epoch": 1.1981639344262296,
      "grad_norm": 0.7965174422032555,
      "learning_rate": 0.00016425163727024707,
      "loss": 12.4008,
      "step": 6852
    },
    {
      "epoch": 1.1983387978142077,
      "grad_norm": 0.9632475234148332,
      "learning_rate": 0.00016424078553591937,
      "loss": 12.5025,
      "step": 6853
    },
    {
      "epoch": 1.1985136612021858,
      "grad_norm": 0.892144113504193,
      "learning_rate": 0.00016422993251336666,
      "loss": 12.8352,
      "step": 6854
    },
    {
      "epoch": 1.198688524590164,
      "grad_norm": 0.8042363555935247,
      "learning_rate": 0.00016421907820280662,
      "loss": 12.3767,
      "step": 6855
    },
    {
      "epoch": 1.198863387978142,
      "grad_norm": 0.8377899608853466,
      "learning_rate": 0.0001642082226044569,
      "loss": 12.6004,
      "step": 6856
    },
    {
      "epoch": 1.1990382513661202,
      "grad_norm": 0.822173221532045,
      "learning_rate": 0.00016419736571853511,
      "loss": 12.7009,
      "step": 6857
    },
    {
      "epoch": 1.1992131147540983,
      "grad_norm": 0.7355910674838352,
      "learning_rate": 0.0001641865075452591,
      "loss": 12.8225,
      "step": 6858
    },
    {
      "epoch": 1.1993879781420764,
      "grad_norm": 0.7416352644979104,
      "learning_rate": 0.00016417564808484647,
      "loss": 12.4951,
      "step": 6859
    },
    {
      "epoch": 1.1995628415300548,
      "grad_norm": 0.8564383225372019,
      "learning_rate": 0.00016416478733751507,
      "loss": 12.7091,
      "step": 6860
    },
    {
      "epoch": 1.1997377049180327,
      "grad_norm": 0.7238008902598803,
      "learning_rate": 0.0001641539253034827,
      "loss": 12.4208,
      "step": 6861
    },
    {
      "epoch": 1.199912568306011,
      "grad_norm": 0.782048436924902,
      "learning_rate": 0.00016414306198296713,
      "loss": 12.7168,
      "step": 6862
    },
    {
      "epoch": 1.2000874316939891,
      "grad_norm": 0.7730721491235446,
      "learning_rate": 0.00016413219737618624,
      "loss": 12.396,
      "step": 6863
    },
    {
      "epoch": 1.2002622950819672,
      "grad_norm": 0.7616627678274527,
      "learning_rate": 0.00016412133148335784,
      "loss": 12.7246,
      "step": 6864
    },
    {
      "epoch": 1.2004371584699454,
      "grad_norm": 0.7334314751945954,
      "learning_rate": 0.00016411046430469988,
      "loss": 12.358,
      "step": 6865
    },
    {
      "epoch": 1.2006120218579235,
      "grad_norm": 0.7215179129006156,
      "learning_rate": 0.0001640995958404303,
      "loss": 12.214,
      "step": 6866
    },
    {
      "epoch": 1.2007868852459016,
      "grad_norm": 0.7411200181158588,
      "learning_rate": 0.00016408872609076697,
      "loss": 12.6663,
      "step": 6867
    },
    {
      "epoch": 1.2009617486338797,
      "grad_norm": 0.7789095045285889,
      "learning_rate": 0.0001640778550559279,
      "loss": 12.6753,
      "step": 6868
    },
    {
      "epoch": 1.2011366120218578,
      "grad_norm": 0.8525787070850778,
      "learning_rate": 0.00016406698273613114,
      "loss": 12.6563,
      "step": 6869
    },
    {
      "epoch": 1.201311475409836,
      "grad_norm": 0.7657188845089552,
      "learning_rate": 0.00016405610913159465,
      "loss": 12.2932,
      "step": 6870
    },
    {
      "epoch": 1.2014863387978143,
      "grad_norm": 0.9783194078019152,
      "learning_rate": 0.00016404523424253645,
      "loss": 12.5045,
      "step": 6871
    },
    {
      "epoch": 1.2016612021857924,
      "grad_norm": 0.9450368448836088,
      "learning_rate": 0.00016403435806917467,
      "loss": 12.4873,
      "step": 6872
    },
    {
      "epoch": 1.2018360655737705,
      "grad_norm": 0.7720957844818204,
      "learning_rate": 0.00016402348061172743,
      "loss": 12.5812,
      "step": 6873
    },
    {
      "epoch": 1.2020109289617487,
      "grad_norm": 0.9591794907363764,
      "learning_rate": 0.00016401260187041283,
      "loss": 12.4987,
      "step": 6874
    },
    {
      "epoch": 1.2021857923497268,
      "grad_norm": 0.6942284261307105,
      "learning_rate": 0.000164001721845449,
      "loss": 12.5942,
      "step": 6875
    },
    {
      "epoch": 1.202360655737705,
      "grad_norm": 0.8328116108235867,
      "learning_rate": 0.00016399084053705414,
      "loss": 12.5297,
      "step": 6876
    },
    {
      "epoch": 1.202535519125683,
      "grad_norm": 0.7741782850602205,
      "learning_rate": 0.00016397995794544646,
      "loss": 12.6768,
      "step": 6877
    },
    {
      "epoch": 1.2027103825136611,
      "grad_norm": 0.8021280827196494,
      "learning_rate": 0.00016396907407084413,
      "loss": 12.4351,
      "step": 6878
    },
    {
      "epoch": 1.2028852459016393,
      "grad_norm": 0.7818606788136916,
      "learning_rate": 0.0001639581889134655,
      "loss": 12.5272,
      "step": 6879
    },
    {
      "epoch": 1.2030601092896176,
      "grad_norm": 0.7137090798862048,
      "learning_rate": 0.00016394730247352882,
      "loss": 12.47,
      "step": 6880
    },
    {
      "epoch": 1.2032349726775957,
      "grad_norm": 0.8495453948574365,
      "learning_rate": 0.00016393641475125233,
      "loss": 12.4552,
      "step": 6881
    },
    {
      "epoch": 1.2034098360655738,
      "grad_norm": 0.8154036152216891,
      "learning_rate": 0.00016392552574685446,
      "loss": 12.506,
      "step": 6882
    },
    {
      "epoch": 1.203584699453552,
      "grad_norm": 0.8639736904316815,
      "learning_rate": 0.00016391463546055348,
      "loss": 12.3036,
      "step": 6883
    },
    {
      "epoch": 1.20375956284153,
      "grad_norm": 0.7809473538145065,
      "learning_rate": 0.00016390374389256787,
      "loss": 12.5477,
      "step": 6884
    },
    {
      "epoch": 1.2039344262295082,
      "grad_norm": 0.8349227697804332,
      "learning_rate": 0.00016389285104311593,
      "loss": 12.5188,
      "step": 6885
    },
    {
      "epoch": 1.2041092896174863,
      "grad_norm": 0.7940053087771974,
      "learning_rate": 0.00016388195691241617,
      "loss": 12.3081,
      "step": 6886
    },
    {
      "epoch": 1.2042841530054644,
      "grad_norm": 0.8823914636973051,
      "learning_rate": 0.00016387106150068703,
      "loss": 12.6467,
      "step": 6887
    },
    {
      "epoch": 1.2044590163934425,
      "grad_norm": 0.798930434316724,
      "learning_rate": 0.000163860164808147,
      "loss": 12.3369,
      "step": 6888
    },
    {
      "epoch": 1.2046338797814207,
      "grad_norm": 0.7877830218232417,
      "learning_rate": 0.00016384926683501457,
      "loss": 12.5494,
      "step": 6889
    },
    {
      "epoch": 1.2048087431693988,
      "grad_norm": 0.7544754433969295,
      "learning_rate": 0.00016383836758150833,
      "loss": 12.3175,
      "step": 6890
    },
    {
      "epoch": 1.2049836065573771,
      "grad_norm": 0.8724692280835987,
      "learning_rate": 0.00016382746704784678,
      "loss": 12.7339,
      "step": 6891
    },
    {
      "epoch": 1.2051584699453552,
      "grad_norm": 0.7518470804554764,
      "learning_rate": 0.00016381656523424857,
      "loss": 12.5083,
      "step": 6892
    },
    {
      "epoch": 1.2053333333333334,
      "grad_norm": 0.8445053145789299,
      "learning_rate": 0.00016380566214093225,
      "loss": 12.667,
      "step": 6893
    },
    {
      "epoch": 1.2055081967213115,
      "grad_norm": 0.7496999473344479,
      "learning_rate": 0.00016379475776811652,
      "loss": 12.7063,
      "step": 6894
    },
    {
      "epoch": 1.2056830601092896,
      "grad_norm": 0.8303107310897047,
      "learning_rate": 0.00016378385211602002,
      "loss": 12.4052,
      "step": 6895
    },
    {
      "epoch": 1.2058579234972677,
      "grad_norm": 0.7013304318715168,
      "learning_rate": 0.00016377294518486144,
      "loss": 12.6627,
      "step": 6896
    },
    {
      "epoch": 1.2060327868852458,
      "grad_norm": 0.8673601250312786,
      "learning_rate": 0.0001637620369748595,
      "loss": 12.6725,
      "step": 6897
    },
    {
      "epoch": 1.206207650273224,
      "grad_norm": 0.7475890826661165,
      "learning_rate": 0.00016375112748623293,
      "loss": 12.812,
      "step": 6898
    },
    {
      "epoch": 1.206382513661202,
      "grad_norm": 0.7895854130826252,
      "learning_rate": 0.00016374021671920056,
      "loss": 12.5051,
      "step": 6899
    },
    {
      "epoch": 1.2065573770491804,
      "grad_norm": 0.7521699786545495,
      "learning_rate": 0.00016372930467398108,
      "loss": 12.644,
      "step": 6900
    },
    {
      "epoch": 1.2067322404371585,
      "grad_norm": 0.805026684395379,
      "learning_rate": 0.00016371839135079339,
      "loss": 12.7561,
      "step": 6901
    },
    {
      "epoch": 1.2069071038251367,
      "grad_norm": 0.7222421318786102,
      "learning_rate": 0.0001637074767498563,
      "loss": 12.216,
      "step": 6902
    },
    {
      "epoch": 1.2070819672131148,
      "grad_norm": 0.746034117185879,
      "learning_rate": 0.0001636965608713887,
      "loss": 12.6167,
      "step": 6903
    },
    {
      "epoch": 1.207256830601093,
      "grad_norm": 0.7918730797263225,
      "learning_rate": 0.0001636856437156095,
      "loss": 12.2461,
      "step": 6904
    },
    {
      "epoch": 1.207431693989071,
      "grad_norm": 0.7735397839812431,
      "learning_rate": 0.00016367472528273756,
      "loss": 12.446,
      "step": 6905
    },
    {
      "epoch": 1.2076065573770491,
      "grad_norm": 0.7445575996247001,
      "learning_rate": 0.0001636638055729919,
      "loss": 12.6411,
      "step": 6906
    },
    {
      "epoch": 1.2077814207650273,
      "grad_norm": 0.7353297524293085,
      "learning_rate": 0.00016365288458659147,
      "loss": 12.51,
      "step": 6907
    },
    {
      "epoch": 1.2079562841530054,
      "grad_norm": 0.7659456903004915,
      "learning_rate": 0.00016364196232375526,
      "loss": 12.4337,
      "step": 6908
    },
    {
      "epoch": 1.2081311475409837,
      "grad_norm": 0.746597575932967,
      "learning_rate": 0.0001636310387847023,
      "loss": 12.4664,
      "step": 6909
    },
    {
      "epoch": 1.2083060109289618,
      "grad_norm": 0.8919571645002236,
      "learning_rate": 0.00016362011396965165,
      "loss": 12.4686,
      "step": 6910
    },
    {
      "epoch": 1.20848087431694,
      "grad_norm": 0.7760165293882518,
      "learning_rate": 0.00016360918787882236,
      "loss": 12.6906,
      "step": 6911
    },
    {
      "epoch": 1.208655737704918,
      "grad_norm": 0.7264014000041337,
      "learning_rate": 0.00016359826051243353,
      "loss": 12.4447,
      "step": 6912
    },
    {
      "epoch": 1.2088306010928962,
      "grad_norm": 0.7212530372268663,
      "learning_rate": 0.00016358733187070432,
      "loss": 12.7556,
      "step": 6913
    },
    {
      "epoch": 1.2090054644808743,
      "grad_norm": 0.7630877390001315,
      "learning_rate": 0.00016357640195385388,
      "loss": 12.5774,
      "step": 6914
    },
    {
      "epoch": 1.2091803278688524,
      "grad_norm": 0.8092255232950745,
      "learning_rate": 0.00016356547076210135,
      "loss": 12.6416,
      "step": 6915
    },
    {
      "epoch": 1.2093551912568306,
      "grad_norm": 0.7825923668847764,
      "learning_rate": 0.000163554538295666,
      "loss": 12.6872,
      "step": 6916
    },
    {
      "epoch": 1.2095300546448087,
      "grad_norm": 0.7260531343793575,
      "learning_rate": 0.00016354360455476697,
      "loss": 12.4952,
      "step": 6917
    },
    {
      "epoch": 1.2097049180327868,
      "grad_norm": 0.8295670035985462,
      "learning_rate": 0.0001635326695396236,
      "loss": 12.4913,
      "step": 6918
    },
    {
      "epoch": 1.209879781420765,
      "grad_norm": 0.7506325271436001,
      "learning_rate": 0.0001635217332504551,
      "loss": 12.6422,
      "step": 6919
    },
    {
      "epoch": 1.2100546448087433,
      "grad_norm": 0.8126760334421166,
      "learning_rate": 0.00016351079568748084,
      "loss": 12.5818,
      "step": 6920
    },
    {
      "epoch": 1.2102295081967214,
      "grad_norm": 0.797819673134614,
      "learning_rate": 0.0001634998568509201,
      "loss": 12.6328,
      "step": 6921
    },
    {
      "epoch": 1.2104043715846995,
      "grad_norm": 0.8276126376583591,
      "learning_rate": 0.0001634889167409923,
      "loss": 12.4695,
      "step": 6922
    },
    {
      "epoch": 1.2105792349726776,
      "grad_norm": 0.8345485949407808,
      "learning_rate": 0.00016347797535791675,
      "loss": 12.3522,
      "step": 6923
    },
    {
      "epoch": 1.2107540983606557,
      "grad_norm": 0.7984174685648479,
      "learning_rate": 0.0001634670327019129,
      "loss": 12.2977,
      "step": 6924
    },
    {
      "epoch": 1.2109289617486338,
      "grad_norm": 0.8545988056682128,
      "learning_rate": 0.0001634560887732002,
      "loss": 12.6193,
      "step": 6925
    },
    {
      "epoch": 1.211103825136612,
      "grad_norm": 0.7903422844422339,
      "learning_rate": 0.00016344514357199803,
      "loss": 12.2946,
      "step": 6926
    },
    {
      "epoch": 1.21127868852459,
      "grad_norm": 1.0676759043389887,
      "learning_rate": 0.00016343419709852596,
      "loss": 12.2295,
      "step": 6927
    },
    {
      "epoch": 1.2114535519125682,
      "grad_norm": 0.8698988024092448,
      "learning_rate": 0.00016342324935300347,
      "loss": 12.6159,
      "step": 6928
    },
    {
      "epoch": 1.2116284153005465,
      "grad_norm": 0.7694701312035399,
      "learning_rate": 0.00016341230033565008,
      "loss": 12.5771,
      "step": 6929
    },
    {
      "epoch": 1.2118032786885247,
      "grad_norm": 0.9010019868747443,
      "learning_rate": 0.0001634013500466854,
      "loss": 12.471,
      "step": 6930
    },
    {
      "epoch": 1.2119781420765028,
      "grad_norm": 0.7317313967671167,
      "learning_rate": 0.00016339039848632894,
      "loss": 12.3584,
      "step": 6931
    },
    {
      "epoch": 1.212153005464481,
      "grad_norm": 0.8891431369355299,
      "learning_rate": 0.0001633794456548004,
      "loss": 12.5,
      "step": 6932
    },
    {
      "epoch": 1.212327868852459,
      "grad_norm": 0.7409209559170912,
      "learning_rate": 0.00016336849155231935,
      "loss": 12.7004,
      "step": 6933
    },
    {
      "epoch": 1.2125027322404371,
      "grad_norm": 0.914316321166712,
      "learning_rate": 0.0001633575361791055,
      "loss": 12.4428,
      "step": 6934
    },
    {
      "epoch": 1.2126775956284153,
      "grad_norm": 0.6855652548434951,
      "learning_rate": 0.00016334657953537849,
      "loss": 12.6031,
      "step": 6935
    },
    {
      "epoch": 1.2128524590163934,
      "grad_norm": 0.7660408322890905,
      "learning_rate": 0.00016333562162135806,
      "loss": 12.5056,
      "step": 6936
    },
    {
      "epoch": 1.2130273224043715,
      "grad_norm": 0.8358479817430273,
      "learning_rate": 0.00016332466243726396,
      "loss": 12.4877,
      "step": 6937
    },
    {
      "epoch": 1.2132021857923498,
      "grad_norm": 0.7688260116949344,
      "learning_rate": 0.00016331370198331593,
      "loss": 12.4707,
      "step": 6938
    },
    {
      "epoch": 1.213377049180328,
      "grad_norm": 1.730315992728134,
      "learning_rate": 0.0001633027402597338,
      "loss": 12.4933,
      "step": 6939
    },
    {
      "epoch": 1.213551912568306,
      "grad_norm": 0.7170939146563041,
      "learning_rate": 0.00016329177726673736,
      "loss": 12.2836,
      "step": 6940
    },
    {
      "epoch": 1.2137267759562842,
      "grad_norm": 0.6904014150698392,
      "learning_rate": 0.00016328081300454644,
      "loss": 12.3845,
      "step": 6941
    },
    {
      "epoch": 1.2139016393442623,
      "grad_norm": 0.7219133132653607,
      "learning_rate": 0.00016326984747338095,
      "loss": 12.4703,
      "step": 6942
    },
    {
      "epoch": 1.2140765027322404,
      "grad_norm": 0.836838963780377,
      "learning_rate": 0.00016325888067346073,
      "loss": 12.6869,
      "step": 6943
    },
    {
      "epoch": 1.2142513661202186,
      "grad_norm": 0.8515735055930094,
      "learning_rate": 0.00016324791260500572,
      "loss": 12.676,
      "step": 6944
    },
    {
      "epoch": 1.2144262295081967,
      "grad_norm": 0.7212106491498761,
      "learning_rate": 0.00016323694326823585,
      "loss": 12.585,
      "step": 6945
    },
    {
      "epoch": 1.2146010928961748,
      "grad_norm": 0.7570866224569873,
      "learning_rate": 0.00016322597266337115,
      "loss": 12.4033,
      "step": 6946
    },
    {
      "epoch": 1.214775956284153,
      "grad_norm": 0.7600900657174142,
      "learning_rate": 0.00016321500079063153,
      "loss": 12.6729,
      "step": 6947
    },
    {
      "epoch": 1.214950819672131,
      "grad_norm": 0.9769959147601479,
      "learning_rate": 0.00016320402765023707,
      "loss": 12.7565,
      "step": 6948
    },
    {
      "epoch": 1.2151256830601094,
      "grad_norm": 0.7410592083413408,
      "learning_rate": 0.0001631930532424078,
      "loss": 12.68,
      "step": 6949
    },
    {
      "epoch": 1.2153005464480875,
      "grad_norm": 0.8115406466963712,
      "learning_rate": 0.00016318207756736377,
      "loss": 12.4441,
      "step": 6950
    },
    {
      "epoch": 1.2154754098360656,
      "grad_norm": 0.7791059412114217,
      "learning_rate": 0.0001631711006253251,
      "loss": 12.8156,
      "step": 6951
    },
    {
      "epoch": 1.2156502732240437,
      "grad_norm": 0.8165248313952801,
      "learning_rate": 0.0001631601224165119,
      "loss": 12.3872,
      "step": 6952
    },
    {
      "epoch": 1.2158251366120219,
      "grad_norm": 0.8172617447423912,
      "learning_rate": 0.0001631491429411443,
      "loss": 12.5949,
      "step": 6953
    },
    {
      "epoch": 1.216,
      "grad_norm": 0.8832828553497144,
      "learning_rate": 0.00016313816219944253,
      "loss": 12.5027,
      "step": 6954
    },
    {
      "epoch": 1.216174863387978,
      "grad_norm": 0.7620836918283695,
      "learning_rate": 0.00016312718019162673,
      "loss": 12.5709,
      "step": 6955
    },
    {
      "epoch": 1.2163497267759562,
      "grad_norm": 0.8186979305778159,
      "learning_rate": 0.00016311619691791713,
      "loss": 12.8444,
      "step": 6956
    },
    {
      "epoch": 1.2165245901639343,
      "grad_norm": 0.7228740715650727,
      "learning_rate": 0.00016310521237853398,
      "loss": 12.6385,
      "step": 6957
    },
    {
      "epoch": 1.2166994535519127,
      "grad_norm": 0.8055179500079568,
      "learning_rate": 0.00016309422657369757,
      "loss": 12.3231,
      "step": 6958
    },
    {
      "epoch": 1.2168743169398908,
      "grad_norm": 0.8477899687193309,
      "learning_rate": 0.00016308323950362823,
      "loss": 12.6747,
      "step": 6959
    },
    {
      "epoch": 1.217049180327869,
      "grad_norm": 0.8797932618434634,
      "learning_rate": 0.00016307225116854622,
      "loss": 12.6174,
      "step": 6960
    },
    {
      "epoch": 1.217224043715847,
      "grad_norm": 0.74742989660789,
      "learning_rate": 0.00016306126156867188,
      "loss": 12.2083,
      "step": 6961
    },
    {
      "epoch": 1.2173989071038251,
      "grad_norm": 0.87439729277674,
      "learning_rate": 0.00016305027070422567,
      "loss": 12.4984,
      "step": 6962
    },
    {
      "epoch": 1.2175737704918033,
      "grad_norm": 0.694161731036791,
      "learning_rate": 0.0001630392785754279,
      "loss": 12.5651,
      "step": 6963
    },
    {
      "epoch": 1.2177486338797814,
      "grad_norm": 0.7899767746477329,
      "learning_rate": 0.00016302828518249907,
      "loss": 12.4782,
      "step": 6964
    },
    {
      "epoch": 1.2179234972677595,
      "grad_norm": 0.832910123675902,
      "learning_rate": 0.00016301729052565963,
      "loss": 12.5559,
      "step": 6965
    },
    {
      "epoch": 1.2180983606557376,
      "grad_norm": 0.7794628135200076,
      "learning_rate": 0.00016300629460512997,
      "loss": 12.3131,
      "step": 6966
    },
    {
      "epoch": 1.218273224043716,
      "grad_norm": 0.8613205401912252,
      "learning_rate": 0.00016299529742113067,
      "loss": 12.6859,
      "step": 6967
    },
    {
      "epoch": 1.218448087431694,
      "grad_norm": 0.7600168247280316,
      "learning_rate": 0.00016298429897388224,
      "loss": 12.5415,
      "step": 6968
    },
    {
      "epoch": 1.2186229508196722,
      "grad_norm": 0.7310944800476659,
      "learning_rate": 0.00016297329926360523,
      "loss": 12.6462,
      "step": 6969
    },
    {
      "epoch": 1.2187978142076503,
      "grad_norm": 0.7823654937045122,
      "learning_rate": 0.00016296229829052022,
      "loss": 12.5243,
      "step": 6970
    },
    {
      "epoch": 1.2189726775956284,
      "grad_norm": 0.8133484145434556,
      "learning_rate": 0.0001629512960548478,
      "loss": 12.7197,
      "step": 6971
    },
    {
      "epoch": 1.2191475409836066,
      "grad_norm": 0.7580435388800542,
      "learning_rate": 0.0001629402925568086,
      "loss": 12.3705,
      "step": 6972
    },
    {
      "epoch": 1.2193224043715847,
      "grad_norm": 0.8056331133978397,
      "learning_rate": 0.0001629292877966233,
      "loss": 12.576,
      "step": 6973
    },
    {
      "epoch": 1.2194972677595628,
      "grad_norm": 0.7833729623250342,
      "learning_rate": 0.00016291828177451256,
      "loss": 12.6081,
      "step": 6974
    },
    {
      "epoch": 1.219672131147541,
      "grad_norm": 0.8315459982065949,
      "learning_rate": 0.00016290727449069708,
      "loss": 12.6409,
      "step": 6975
    },
    {
      "epoch": 1.219846994535519,
      "grad_norm": 0.8208664045469855,
      "learning_rate": 0.0001628962659453976,
      "loss": 12.5445,
      "step": 6976
    },
    {
      "epoch": 1.2200218579234972,
      "grad_norm": 0.7678923407071166,
      "learning_rate": 0.00016288525613883484,
      "loss": 12.5674,
      "step": 6977
    },
    {
      "epoch": 1.2201967213114755,
      "grad_norm": 0.7219127598927572,
      "learning_rate": 0.00016287424507122964,
      "loss": 12.8008,
      "step": 6978
    },
    {
      "epoch": 1.2203715846994536,
      "grad_norm": 0.8958145041815215,
      "learning_rate": 0.00016286323274280278,
      "loss": 12.4797,
      "step": 6979
    },
    {
      "epoch": 1.2205464480874317,
      "grad_norm": 0.8058936218836877,
      "learning_rate": 0.00016285221915377508,
      "loss": 12.6115,
      "step": 6980
    },
    {
      "epoch": 1.2207213114754099,
      "grad_norm": 0.7316726387709895,
      "learning_rate": 0.0001628412043043674,
      "loss": 12.5329,
      "step": 6981
    },
    {
      "epoch": 1.220896174863388,
      "grad_norm": 0.8372422262362524,
      "learning_rate": 0.00016283018819480066,
      "loss": 12.6292,
      "step": 6982
    },
    {
      "epoch": 1.221071038251366,
      "grad_norm": 0.7975054102859515,
      "learning_rate": 0.0001628191708252957,
      "loss": 12.4162,
      "step": 6983
    },
    {
      "epoch": 1.2212459016393442,
      "grad_norm": 0.7332149412732429,
      "learning_rate": 0.00016280815219607347,
      "loss": 12.6782,
      "step": 6984
    },
    {
      "epoch": 1.2214207650273223,
      "grad_norm": 0.7769827330938975,
      "learning_rate": 0.000162797132307355,
      "loss": 12.5849,
      "step": 6985
    },
    {
      "epoch": 1.2215956284153004,
      "grad_norm": 0.6923910310156558,
      "learning_rate": 0.00016278611115936114,
      "loss": 12.4976,
      "step": 6986
    },
    {
      "epoch": 1.2217704918032788,
      "grad_norm": 0.8436444705782278,
      "learning_rate": 0.00016277508875231302,
      "loss": 12.6993,
      "step": 6987
    },
    {
      "epoch": 1.221945355191257,
      "grad_norm": 0.7670376778970933,
      "learning_rate": 0.00016276406508643163,
      "loss": 12.5874,
      "step": 6988
    },
    {
      "epoch": 1.222120218579235,
      "grad_norm": 0.7147622832039332,
      "learning_rate": 0.00016275304016193798,
      "loss": 12.7534,
      "step": 6989
    },
    {
      "epoch": 1.2222950819672131,
      "grad_norm": 0.7244581916832565,
      "learning_rate": 0.00016274201397905325,
      "loss": 12.6035,
      "step": 6990
    },
    {
      "epoch": 1.2224699453551913,
      "grad_norm": 0.8860141578735242,
      "learning_rate": 0.00016273098653799846,
      "loss": 12.5296,
      "step": 6991
    },
    {
      "epoch": 1.2226448087431694,
      "grad_norm": 0.7396495543262613,
      "learning_rate": 0.0001627199578389948,
      "loss": 12.7252,
      "step": 6992
    },
    {
      "epoch": 1.2228196721311475,
      "grad_norm": 0.7383765323794887,
      "learning_rate": 0.0001627089278822634,
      "loss": 12.5846,
      "step": 6993
    },
    {
      "epoch": 1.2229945355191256,
      "grad_norm": 0.821592573967514,
      "learning_rate": 0.00016269789666802545,
      "loss": 12.643,
      "step": 6994
    },
    {
      "epoch": 1.2231693989071037,
      "grad_norm": 0.8026184644119265,
      "learning_rate": 0.0001626868641965022,
      "loss": 12.4815,
      "step": 6995
    },
    {
      "epoch": 1.223344262295082,
      "grad_norm": 0.7040613598462461,
      "learning_rate": 0.0001626758304679148,
      "loss": 12.4569,
      "step": 6996
    },
    {
      "epoch": 1.2235191256830602,
      "grad_norm": 0.7159101562827256,
      "learning_rate": 0.00016266479548248457,
      "loss": 12.4727,
      "step": 6997
    },
    {
      "epoch": 1.2236939890710383,
      "grad_norm": 0.708382505681996,
      "learning_rate": 0.0001626537592404328,
      "loss": 12.5454,
      "step": 6998
    },
    {
      "epoch": 1.2238688524590164,
      "grad_norm": 0.7565777951181166,
      "learning_rate": 0.00016264272174198077,
      "loss": 12.6859,
      "step": 6999
    },
    {
      "epoch": 1.2240437158469946,
      "grad_norm": 0.7336159123566333,
      "learning_rate": 0.00016263168298734986,
      "loss": 12.326,
      "step": 7000
    },
    {
      "epoch": 1.2242185792349727,
      "grad_norm": 0.7196667560640992,
      "learning_rate": 0.00016262064297676136,
      "loss": 12.3883,
      "step": 7001
    },
    {
      "epoch": 1.2243934426229508,
      "grad_norm": 0.7404298487891124,
      "learning_rate": 0.00016260960171043674,
      "loss": 12.6027,
      "step": 7002
    },
    {
      "epoch": 1.224568306010929,
      "grad_norm": 0.8073991875200149,
      "learning_rate": 0.00016259855918859733,
      "loss": 12.5484,
      "step": 7003
    },
    {
      "epoch": 1.224743169398907,
      "grad_norm": 0.7236926017811367,
      "learning_rate": 0.00016258751541146463,
      "loss": 12.5992,
      "step": 7004
    },
    {
      "epoch": 1.2249180327868852,
      "grad_norm": 0.7913729769754295,
      "learning_rate": 0.00016257647037926006,
      "loss": 12.3895,
      "step": 7005
    },
    {
      "epoch": 1.2250928961748633,
      "grad_norm": 0.7087457971477431,
      "learning_rate": 0.00016256542409220513,
      "loss": 12.6386,
      "step": 7006
    },
    {
      "epoch": 1.2252677595628416,
      "grad_norm": 0.8401602660809767,
      "learning_rate": 0.00016255437655052137,
      "loss": 12.5655,
      "step": 7007
    },
    {
      "epoch": 1.2254426229508197,
      "grad_norm": 0.7188931374694798,
      "learning_rate": 0.00016254332775443028,
      "loss": 12.7193,
      "step": 7008
    },
    {
      "epoch": 1.2256174863387979,
      "grad_norm": 0.8171706643207577,
      "learning_rate": 0.0001625322777041534,
      "loss": 12.6113,
      "step": 7009
    },
    {
      "epoch": 1.225792349726776,
      "grad_norm": 0.7406629099951381,
      "learning_rate": 0.00016252122639991238,
      "loss": 12.422,
      "step": 7010
    },
    {
      "epoch": 1.225967213114754,
      "grad_norm": 0.8023016129666594,
      "learning_rate": 0.00016251017384192885,
      "loss": 12.488,
      "step": 7011
    },
    {
      "epoch": 1.2261420765027322,
      "grad_norm": 0.7349375712463421,
      "learning_rate": 0.00016249912003042435,
      "loss": 12.4087,
      "step": 7012
    },
    {
      "epoch": 1.2263169398907103,
      "grad_norm": 0.8229652491906553,
      "learning_rate": 0.00016248806496562058,
      "loss": 12.608,
      "step": 7013
    },
    {
      "epoch": 1.2264918032786885,
      "grad_norm": 0.6648873781983933,
      "learning_rate": 0.00016247700864773927,
      "loss": 12.5654,
      "step": 7014
    },
    {
      "epoch": 1.2266666666666666,
      "grad_norm": 0.7948120389411081,
      "learning_rate": 0.0001624659510770021,
      "loss": 12.6479,
      "step": 7015
    },
    {
      "epoch": 1.226841530054645,
      "grad_norm": 0.8350206558990195,
      "learning_rate": 0.00016245489225363082,
      "loss": 12.2006,
      "step": 7016
    },
    {
      "epoch": 1.227016393442623,
      "grad_norm": 0.9087867591727196,
      "learning_rate": 0.0001624438321778472,
      "loss": 12.552,
      "step": 7017
    },
    {
      "epoch": 1.2271912568306012,
      "grad_norm": 0.7396320165999897,
      "learning_rate": 0.000162432770849873,
      "loss": 12.2365,
      "step": 7018
    },
    {
      "epoch": 1.2273661202185793,
      "grad_norm": 0.8950765364537876,
      "learning_rate": 0.00016242170826993006,
      "loss": 12.5054,
      "step": 7019
    },
    {
      "epoch": 1.2275409836065574,
      "grad_norm": 0.7874696209409806,
      "learning_rate": 0.00016241064443824022,
      "loss": 12.6117,
      "step": 7020
    },
    {
      "epoch": 1.2277158469945355,
      "grad_norm": 1.0644505098409096,
      "learning_rate": 0.00016239957935502533,
      "loss": 12.5636,
      "step": 7021
    },
    {
      "epoch": 1.2278907103825136,
      "grad_norm": 0.8574205661347285,
      "learning_rate": 0.00016238851302050725,
      "loss": 12.6361,
      "step": 7022
    },
    {
      "epoch": 1.2280655737704917,
      "grad_norm": 1.0974445678307863,
      "learning_rate": 0.00016237744543490796,
      "loss": 12.3853,
      "step": 7023
    },
    {
      "epoch": 1.2282404371584699,
      "grad_norm": 0.6877334486367672,
      "learning_rate": 0.00016236637659844936,
      "loss": 12.3677,
      "step": 7024
    },
    {
      "epoch": 1.2284153005464482,
      "grad_norm": 0.9505889709302144,
      "learning_rate": 0.0001623553065113534,
      "loss": 12.402,
      "step": 7025
    },
    {
      "epoch": 1.228590163934426,
      "grad_norm": 0.8459973252325217,
      "learning_rate": 0.0001623442351738421,
      "loss": 12.6179,
      "step": 7026
    },
    {
      "epoch": 1.2287650273224044,
      "grad_norm": 1.2068777525642413,
      "learning_rate": 0.00016233316258613747,
      "loss": 12.6797,
      "step": 7027
    },
    {
      "epoch": 1.2289398907103826,
      "grad_norm": 0.801967628888413,
      "learning_rate": 0.00016232208874846155,
      "loss": 12.657,
      "step": 7028
    },
    {
      "epoch": 1.2291147540983607,
      "grad_norm": 1.0458345778526126,
      "learning_rate": 0.00016231101366103636,
      "loss": 12.3734,
      "step": 7029
    },
    {
      "epoch": 1.2292896174863388,
      "grad_norm": 0.8693564923280999,
      "learning_rate": 0.00016229993732408407,
      "loss": 12.0882,
      "step": 7030
    },
    {
      "epoch": 1.229464480874317,
      "grad_norm": 1.040767429034885,
      "learning_rate": 0.0001622888597378267,
      "loss": 12.5363,
      "step": 7031
    },
    {
      "epoch": 1.229639344262295,
      "grad_norm": 0.8172462131226357,
      "learning_rate": 0.00016227778090248648,
      "loss": 12.28,
      "step": 7032
    },
    {
      "epoch": 1.2298142076502732,
      "grad_norm": 0.9004105812174006,
      "learning_rate": 0.00016226670081828553,
      "loss": 12.7388,
      "step": 7033
    },
    {
      "epoch": 1.2299890710382513,
      "grad_norm": 0.8027530530759805,
      "learning_rate": 0.00016225561948544602,
      "loss": 12.4011,
      "step": 7034
    },
    {
      "epoch": 1.2301639344262294,
      "grad_norm": 0.7545068235060388,
      "learning_rate": 0.00016224453690419024,
      "loss": 12.7747,
      "step": 7035
    },
    {
      "epoch": 1.2303387978142077,
      "grad_norm": 0.9061041491285948,
      "learning_rate": 0.00016223345307474034,
      "loss": 12.4389,
      "step": 7036
    },
    {
      "epoch": 1.2305136612021859,
      "grad_norm": 0.7542333053806306,
      "learning_rate": 0.00016222236799731863,
      "loss": 12.6579,
      "step": 7037
    },
    {
      "epoch": 1.230688524590164,
      "grad_norm": 0.8343339373434729,
      "learning_rate": 0.0001622112816721474,
      "loss": 12.6393,
      "step": 7038
    },
    {
      "epoch": 1.230863387978142,
      "grad_norm": 0.7724145684968712,
      "learning_rate": 0.00016220019409944896,
      "loss": 12.6685,
      "step": 7039
    },
    {
      "epoch": 1.2310382513661202,
      "grad_norm": 0.8197027052096079,
      "learning_rate": 0.00016218910527944564,
      "loss": 12.5991,
      "step": 7040
    },
    {
      "epoch": 1.2312131147540983,
      "grad_norm": 0.8262666439308465,
      "learning_rate": 0.0001621780152123598,
      "loss": 12.5317,
      "step": 7041
    },
    {
      "epoch": 1.2313879781420765,
      "grad_norm": 0.869885968889681,
      "learning_rate": 0.00016216692389841387,
      "loss": 12.2734,
      "step": 7042
    },
    {
      "epoch": 1.2315628415300546,
      "grad_norm": 0.7712568013590788,
      "learning_rate": 0.00016215583133783024,
      "loss": 12.5913,
      "step": 7043
    },
    {
      "epoch": 1.2317377049180327,
      "grad_norm": 0.8421675610230681,
      "learning_rate": 0.00016214473753083133,
      "loss": 12.6189,
      "step": 7044
    },
    {
      "epoch": 1.231912568306011,
      "grad_norm": 0.893156804797699,
      "learning_rate": 0.00016213364247763963,
      "loss": 12.4856,
      "step": 7045
    },
    {
      "epoch": 1.2320874316939892,
      "grad_norm": 0.9098233464328009,
      "learning_rate": 0.0001621225461784776,
      "loss": 12.2258,
      "step": 7046
    },
    {
      "epoch": 1.2322622950819673,
      "grad_norm": 0.7854347812680745,
      "learning_rate": 0.00016211144863356784,
      "loss": 12.5458,
      "step": 7047
    },
    {
      "epoch": 1.2324371584699454,
      "grad_norm": 0.866811217652998,
      "learning_rate": 0.00016210034984313277,
      "loss": 12.5948,
      "step": 7048
    },
    {
      "epoch": 1.2326120218579235,
      "grad_norm": 0.8406158732222117,
      "learning_rate": 0.00016208924980739504,
      "loss": 12.3801,
      "step": 7049
    },
    {
      "epoch": 1.2327868852459016,
      "grad_norm": 0.7617374384189095,
      "learning_rate": 0.0001620781485265772,
      "loss": 12.6315,
      "step": 7050
    },
    {
      "epoch": 1.2329617486338798,
      "grad_norm": 0.8815681677827182,
      "learning_rate": 0.00016206704600090188,
      "loss": 12.6236,
      "step": 7051
    },
    {
      "epoch": 1.2331366120218579,
      "grad_norm": 0.8019327864764778,
      "learning_rate": 0.00016205594223059173,
      "loss": 12.6606,
      "step": 7052
    },
    {
      "epoch": 1.233311475409836,
      "grad_norm": 0.8828151250096304,
      "learning_rate": 0.00016204483721586938,
      "loss": 12.2872,
      "step": 7053
    },
    {
      "epoch": 1.2334863387978143,
      "grad_norm": 0.7405546266198009,
      "learning_rate": 0.00016203373095695754,
      "loss": 12.5153,
      "step": 7054
    },
    {
      "epoch": 1.2336612021857922,
      "grad_norm": 0.9330408641646303,
      "learning_rate": 0.00016202262345407895,
      "loss": 12.5457,
      "step": 7055
    },
    {
      "epoch": 1.2338360655737706,
      "grad_norm": 0.698227490329256,
      "learning_rate": 0.0001620115147074563,
      "loss": 12.6413,
      "step": 7056
    },
    {
      "epoch": 1.2340109289617487,
      "grad_norm": 0.858737714195457,
      "learning_rate": 0.0001620004047173124,
      "loss": 12.3409,
      "step": 7057
    },
    {
      "epoch": 1.2341857923497268,
      "grad_norm": 0.6942581328719927,
      "learning_rate": 0.00016198929348387,
      "loss": 12.6004,
      "step": 7058
    },
    {
      "epoch": 1.234360655737705,
      "grad_norm": 0.8020936386384637,
      "learning_rate": 0.00016197818100735197,
      "loss": 12.6831,
      "step": 7059
    },
    {
      "epoch": 1.234535519125683,
      "grad_norm": 0.7403225306558797,
      "learning_rate": 0.00016196706728798107,
      "loss": 12.6726,
      "step": 7060
    },
    {
      "epoch": 1.2347103825136612,
      "grad_norm": 0.9168515839696265,
      "learning_rate": 0.00016195595232598023,
      "loss": 12.7335,
      "step": 7061
    },
    {
      "epoch": 1.2348852459016393,
      "grad_norm": 0.7681452747705523,
      "learning_rate": 0.0001619448361215723,
      "loss": 12.3556,
      "step": 7062
    },
    {
      "epoch": 1.2350601092896174,
      "grad_norm": 0.8776890020199174,
      "learning_rate": 0.00016193371867498021,
      "loss": 12.6278,
      "step": 7063
    },
    {
      "epoch": 1.2352349726775955,
      "grad_norm": 0.8530881392759752,
      "learning_rate": 0.00016192259998642694,
      "loss": 12.7003,
      "step": 7064
    },
    {
      "epoch": 1.2354098360655739,
      "grad_norm": 0.7616271417429681,
      "learning_rate": 0.00016191148005613538,
      "loss": 12.6206,
      "step": 7065
    },
    {
      "epoch": 1.235584699453552,
      "grad_norm": 0.7482621078436664,
      "learning_rate": 0.00016190035888432856,
      "loss": 12.6719,
      "step": 7066
    },
    {
      "epoch": 1.23575956284153,
      "grad_norm": 0.8065889084279174,
      "learning_rate": 0.00016188923647122947,
      "loss": 12.3801,
      "step": 7067
    },
    {
      "epoch": 1.2359344262295082,
      "grad_norm": 0.709812700220681,
      "learning_rate": 0.00016187811281706115,
      "loss": 12.6247,
      "step": 7068
    },
    {
      "epoch": 1.2361092896174863,
      "grad_norm": 0.7802405740485806,
      "learning_rate": 0.00016186698792204673,
      "loss": 12.356,
      "step": 7069
    },
    {
      "epoch": 1.2362841530054645,
      "grad_norm": 0.7045835046723326,
      "learning_rate": 0.0001618558617864092,
      "loss": 12.6586,
      "step": 7070
    },
    {
      "epoch": 1.2364590163934426,
      "grad_norm": 0.7309423959449727,
      "learning_rate": 0.00016184473441037173,
      "loss": 12.4449,
      "step": 7071
    },
    {
      "epoch": 1.2366338797814207,
      "grad_norm": 0.7237026814158428,
      "learning_rate": 0.00016183360579415744,
      "loss": 12.6899,
      "step": 7072
    },
    {
      "epoch": 1.2368087431693988,
      "grad_norm": 0.7696301596805818,
      "learning_rate": 0.0001618224759379895,
      "loss": 12.5455,
      "step": 7073
    },
    {
      "epoch": 1.2369836065573772,
      "grad_norm": 0.6889723889723243,
      "learning_rate": 0.00016181134484209112,
      "loss": 12.2473,
      "step": 7074
    },
    {
      "epoch": 1.2371584699453553,
      "grad_norm": 0.8345405005205553,
      "learning_rate": 0.00016180021250668548,
      "loss": 12.6044,
      "step": 7075
    },
    {
      "epoch": 1.2373333333333334,
      "grad_norm": 0.7991955322587555,
      "learning_rate": 0.00016178907893199583,
      "loss": 12.443,
      "step": 7076
    },
    {
      "epoch": 1.2375081967213115,
      "grad_norm": 0.7536336468292554,
      "learning_rate": 0.00016177794411824544,
      "loss": 12.548,
      "step": 7077
    },
    {
      "epoch": 1.2376830601092896,
      "grad_norm": 0.7338321621027276,
      "learning_rate": 0.00016176680806565758,
      "loss": 12.6141,
      "step": 7078
    },
    {
      "epoch": 1.2378579234972678,
      "grad_norm": 0.7532243720241047,
      "learning_rate": 0.00016175567077445554,
      "loss": 12.5083,
      "step": 7079
    },
    {
      "epoch": 1.2380327868852459,
      "grad_norm": 0.7361972007849574,
      "learning_rate": 0.00016174453224486273,
      "loss": 12.4096,
      "step": 7080
    },
    {
      "epoch": 1.238207650273224,
      "grad_norm": 0.7104127657937033,
      "learning_rate": 0.00016173339247710244,
      "loss": 12.0144,
      "step": 7081
    },
    {
      "epoch": 1.2383825136612021,
      "grad_norm": 0.7327353565843296,
      "learning_rate": 0.0001617222514713981,
      "loss": 12.379,
      "step": 7082
    },
    {
      "epoch": 1.2385573770491805,
      "grad_norm": 0.7223602186389023,
      "learning_rate": 0.00016171110922797309,
      "loss": 12.5292,
      "step": 7083
    },
    {
      "epoch": 1.2387322404371583,
      "grad_norm": 0.8044868563269483,
      "learning_rate": 0.00016169996574705086,
      "loss": 12.7205,
      "step": 7084
    },
    {
      "epoch": 1.2389071038251367,
      "grad_norm": 0.8527548677450713,
      "learning_rate": 0.00016168882102885488,
      "loss": 12.4226,
      "step": 7085
    },
    {
      "epoch": 1.2390819672131148,
      "grad_norm": 0.9043999830189169,
      "learning_rate": 0.00016167767507360866,
      "loss": 12.5085,
      "step": 7086
    },
    {
      "epoch": 1.239256830601093,
      "grad_norm": 0.7641945188997853,
      "learning_rate": 0.00016166652788153564,
      "loss": 12.5862,
      "step": 7087
    },
    {
      "epoch": 1.239431693989071,
      "grad_norm": 0.9482007199233694,
      "learning_rate": 0.00016165537945285938,
      "loss": 12.622,
      "step": 7088
    },
    {
      "epoch": 1.2396065573770492,
      "grad_norm": 0.8236236874016091,
      "learning_rate": 0.0001616442297878035,
      "loss": 12.6466,
      "step": 7089
    },
    {
      "epoch": 1.2397814207650273,
      "grad_norm": 0.8242152248532868,
      "learning_rate": 0.0001616330788865915,
      "loss": 12.5562,
      "step": 7090
    },
    {
      "epoch": 1.2399562841530054,
      "grad_norm": 0.72194328653184,
      "learning_rate": 0.00016162192674944707,
      "loss": 12.1475,
      "step": 7091
    },
    {
      "epoch": 1.2401311475409835,
      "grad_norm": 0.8331922020287955,
      "learning_rate": 0.00016161077337659377,
      "loss": 12.3919,
      "step": 7092
    },
    {
      "epoch": 1.2403060109289616,
      "grad_norm": 0.6811122857687366,
      "learning_rate": 0.0001615996187682553,
      "loss": 12.3316,
      "step": 7093
    },
    {
      "epoch": 1.24048087431694,
      "grad_norm": 0.8907342963246874,
      "learning_rate": 0.00016158846292465534,
      "loss": 12.7007,
      "step": 7094
    },
    {
      "epoch": 1.240655737704918,
      "grad_norm": 0.7082100752039622,
      "learning_rate": 0.00016157730584601764,
      "loss": 12.5067,
      "step": 7095
    },
    {
      "epoch": 1.2408306010928962,
      "grad_norm": 0.957386022321036,
      "learning_rate": 0.0001615661475325658,
      "loss": 12.7173,
      "step": 7096
    },
    {
      "epoch": 1.2410054644808743,
      "grad_norm": 0.8383288824825696,
      "learning_rate": 0.00016155498798452376,
      "loss": 12.5399,
      "step": 7097
    },
    {
      "epoch": 1.2411803278688525,
      "grad_norm": 0.8193056585761395,
      "learning_rate": 0.00016154382720211518,
      "loss": 12.5988,
      "step": 7098
    },
    {
      "epoch": 1.2413551912568306,
      "grad_norm": 0.7898108208409178,
      "learning_rate": 0.0001615326651855639,
      "loss": 12.7083,
      "step": 7099
    },
    {
      "epoch": 1.2415300546448087,
      "grad_norm": 0.9138894484569295,
      "learning_rate": 0.00016152150193509375,
      "loss": 12.7077,
      "step": 7100
    },
    {
      "epoch": 1.2417049180327868,
      "grad_norm": 0.7360247699054234,
      "learning_rate": 0.00016151033745092857,
      "loss": 12.5565,
      "step": 7101
    },
    {
      "epoch": 1.241879781420765,
      "grad_norm": 0.822032937668546,
      "learning_rate": 0.0001614991717332923,
      "loss": 12.5248,
      "step": 7102
    },
    {
      "epoch": 1.2420546448087433,
      "grad_norm": 0.827776887727701,
      "learning_rate": 0.00016148800478240879,
      "loss": 12.4135,
      "step": 7103
    },
    {
      "epoch": 1.2422295081967214,
      "grad_norm": 0.9262616819429595,
      "learning_rate": 0.00016147683659850198,
      "loss": 12.6736,
      "step": 7104
    },
    {
      "epoch": 1.2424043715846995,
      "grad_norm": 0.9554875963825119,
      "learning_rate": 0.00016146566718179585,
      "loss": 12.6198,
      "step": 7105
    },
    {
      "epoch": 1.2425792349726776,
      "grad_norm": 0.9770581259040927,
      "learning_rate": 0.00016145449653251435,
      "loss": 12.5272,
      "step": 7106
    },
    {
      "epoch": 1.2427540983606558,
      "grad_norm": 0.8543547223738966,
      "learning_rate": 0.00016144332465088156,
      "loss": 12.6756,
      "step": 7107
    },
    {
      "epoch": 1.2429289617486339,
      "grad_norm": 0.9111403187542721,
      "learning_rate": 0.0001614321515371214,
      "loss": 12.5744,
      "step": 7108
    },
    {
      "epoch": 1.243103825136612,
      "grad_norm": 0.8250588344570228,
      "learning_rate": 0.00016142097719145802,
      "loss": 12.331,
      "step": 7109
    },
    {
      "epoch": 1.2432786885245901,
      "grad_norm": 0.7744016789643006,
      "learning_rate": 0.00016140980161411542,
      "loss": 12.7059,
      "step": 7110
    },
    {
      "epoch": 1.2434535519125682,
      "grad_norm": 0.9721947260018949,
      "learning_rate": 0.00016139862480531777,
      "loss": 12.7057,
      "step": 7111
    },
    {
      "epoch": 1.2436284153005464,
      "grad_norm": 0.7448024375848992,
      "learning_rate": 0.00016138744676528916,
      "loss": 12.487,
      "step": 7112
    },
    {
      "epoch": 1.2438032786885245,
      "grad_norm": 1.019830528560477,
      "learning_rate": 0.00016137626749425377,
      "loss": 12.6868,
      "step": 7113
    },
    {
      "epoch": 1.2439781420765028,
      "grad_norm": 0.8806312584625907,
      "learning_rate": 0.00016136508699243575,
      "loss": 12.4803,
      "step": 7114
    },
    {
      "epoch": 1.244153005464481,
      "grad_norm": 0.8569286326028452,
      "learning_rate": 0.00016135390526005936,
      "loss": 12.5245,
      "step": 7115
    },
    {
      "epoch": 1.244327868852459,
      "grad_norm": 0.8683507605941146,
      "learning_rate": 0.00016134272229734875,
      "loss": 12.3929,
      "step": 7116
    },
    {
      "epoch": 1.2445027322404372,
      "grad_norm": 0.9015248798779353,
      "learning_rate": 0.0001613315381045282,
      "loss": 12.6431,
      "step": 7117
    },
    {
      "epoch": 1.2446775956284153,
      "grad_norm": 0.9658761978900028,
      "learning_rate": 0.000161320352681822,
      "loss": 12.3605,
      "step": 7118
    },
    {
      "epoch": 1.2448524590163934,
      "grad_norm": 0.7102342736022151,
      "learning_rate": 0.0001613091660294545,
      "loss": 12.5964,
      "step": 7119
    },
    {
      "epoch": 1.2450273224043715,
      "grad_norm": 0.858195647044979,
      "learning_rate": 0.00016129797814764995,
      "loss": 12.5825,
      "step": 7120
    },
    {
      "epoch": 1.2452021857923496,
      "grad_norm": 0.913401196951368,
      "learning_rate": 0.00016128678903663273,
      "loss": 12.7239,
      "step": 7121
    },
    {
      "epoch": 1.2453770491803278,
      "grad_norm": 0.9188904170624138,
      "learning_rate": 0.00016127559869662722,
      "loss": 12.5807,
      "step": 7122
    },
    {
      "epoch": 1.245551912568306,
      "grad_norm": 0.8032600941249657,
      "learning_rate": 0.0001612644071278578,
      "loss": 12.6866,
      "step": 7123
    },
    {
      "epoch": 1.2457267759562842,
      "grad_norm": 0.8011797444584716,
      "learning_rate": 0.00016125321433054892,
      "loss": 12.5215,
      "step": 7124
    },
    {
      "epoch": 1.2459016393442623,
      "grad_norm": 0.8280274383453493,
      "learning_rate": 0.000161242020304925,
      "loss": 12.5278,
      "step": 7125
    },
    {
      "epoch": 1.2460765027322405,
      "grad_norm": 0.7515024043782041,
      "learning_rate": 0.00016123082505121056,
      "loss": 12.5791,
      "step": 7126
    },
    {
      "epoch": 1.2462513661202186,
      "grad_norm": 0.6847088113823623,
      "learning_rate": 0.0001612196285696301,
      "loss": 12.34,
      "step": 7127
    },
    {
      "epoch": 1.2464262295081967,
      "grad_norm": 0.8111212820564981,
      "learning_rate": 0.00016120843086040804,
      "loss": 12.4669,
      "step": 7128
    },
    {
      "epoch": 1.2466010928961748,
      "grad_norm": 0.7392388023041068,
      "learning_rate": 0.00016119723192376908,
      "loss": 12.5778,
      "step": 7129
    },
    {
      "epoch": 1.246775956284153,
      "grad_norm": 0.863024012482543,
      "learning_rate": 0.00016118603175993768,
      "loss": 12.3202,
      "step": 7130
    },
    {
      "epoch": 1.246950819672131,
      "grad_norm": 0.7952097219241533,
      "learning_rate": 0.0001611748303691385,
      "loss": 12.6675,
      "step": 7131
    },
    {
      "epoch": 1.2471256830601094,
      "grad_norm": 0.9807079955162792,
      "learning_rate": 0.00016116362775159612,
      "loss": 12.6513,
      "step": 7132
    },
    {
      "epoch": 1.2473005464480875,
      "grad_norm": 0.678462696047411,
      "learning_rate": 0.00016115242390753523,
      "loss": 12.5179,
      "step": 7133
    },
    {
      "epoch": 1.2474754098360656,
      "grad_norm": 0.7827166164078605,
      "learning_rate": 0.00016114121883718045,
      "loss": 12.7336,
      "step": 7134
    },
    {
      "epoch": 1.2476502732240438,
      "grad_norm": 0.769796400505054,
      "learning_rate": 0.00016113001254075652,
      "loss": 12.4593,
      "step": 7135
    },
    {
      "epoch": 1.2478251366120219,
      "grad_norm": 0.8031253271998263,
      "learning_rate": 0.00016111880501848815,
      "loss": 12.5066,
      "step": 7136
    },
    {
      "epoch": 1.248,
      "grad_norm": 0.715694493906316,
      "learning_rate": 0.00016110759627060005,
      "loss": 12.6001,
      "step": 7137
    },
    {
      "epoch": 1.2481748633879781,
      "grad_norm": 0.8130720799784472,
      "learning_rate": 0.00016109638629731705,
      "loss": 12.529,
      "step": 7138
    },
    {
      "epoch": 1.2483497267759562,
      "grad_norm": 0.7616161065765653,
      "learning_rate": 0.00016108517509886392,
      "loss": 12.661,
      "step": 7139
    },
    {
      "epoch": 1.2485245901639344,
      "grad_norm": 0.7470407558600133,
      "learning_rate": 0.00016107396267546546,
      "loss": 12.6033,
      "step": 7140
    },
    {
      "epoch": 1.2486994535519125,
      "grad_norm": 0.7976385380569646,
      "learning_rate": 0.0001610627490273465,
      "loss": 12.6251,
      "step": 7141
    },
    {
      "epoch": 1.2488743169398906,
      "grad_norm": 0.8606793332660675,
      "learning_rate": 0.00016105153415473197,
      "loss": 12.5289,
      "step": 7142
    },
    {
      "epoch": 1.249049180327869,
      "grad_norm": 0.7656518414753195,
      "learning_rate": 0.0001610403180578467,
      "loss": 12.4671,
      "step": 7143
    },
    {
      "epoch": 1.249224043715847,
      "grad_norm": 0.7918846843433602,
      "learning_rate": 0.0001610291007369157,
      "loss": 12.494,
      "step": 7144
    },
    {
      "epoch": 1.2493989071038252,
      "grad_norm": 0.709367400464751,
      "learning_rate": 0.00016101788219216378,
      "loss": 12.5297,
      "step": 7145
    },
    {
      "epoch": 1.2495737704918033,
      "grad_norm": 0.8586573819087572,
      "learning_rate": 0.000161006662423816,
      "loss": 12.4644,
      "step": 7146
    },
    {
      "epoch": 1.2497486338797814,
      "grad_norm": 0.7040532248410825,
      "learning_rate": 0.0001609954414320973,
      "loss": 12.5277,
      "step": 7147
    },
    {
      "epoch": 1.2499234972677595,
      "grad_norm": 0.7611795566500034,
      "learning_rate": 0.00016098421921723275,
      "loss": 12.1106,
      "step": 7148
    },
    {
      "epoch": 1.2500983606557377,
      "grad_norm": 0.7435993278246723,
      "learning_rate": 0.00016097299577944735,
      "loss": 12.4903,
      "step": 7149
    },
    {
      "epoch": 1.2502732240437158,
      "grad_norm": 0.733951699231899,
      "learning_rate": 0.00016096177111896615,
      "loss": 12.3461,
      "step": 7150
    },
    {
      "epoch": 1.2504480874316939,
      "grad_norm": 0.7292035358662148,
      "learning_rate": 0.00016095054523601428,
      "loss": 12.6516,
      "step": 7151
    },
    {
      "epoch": 1.2506229508196722,
      "grad_norm": 0.6930601952249976,
      "learning_rate": 0.00016093931813081683,
      "loss": 12.5538,
      "step": 7152
    },
    {
      "epoch": 1.2507978142076503,
      "grad_norm": 0.7471036374469242,
      "learning_rate": 0.00016092808980359895,
      "loss": 12.4337,
      "step": 7153
    },
    {
      "epoch": 1.2509726775956285,
      "grad_norm": 0.6947502352101764,
      "learning_rate": 0.00016091686025458576,
      "loss": 12.4807,
      "step": 7154
    },
    {
      "epoch": 1.2511475409836066,
      "grad_norm": 0.8297075755061739,
      "learning_rate": 0.00016090562948400255,
      "loss": 12.7447,
      "step": 7155
    },
    {
      "epoch": 1.2513224043715847,
      "grad_norm": 0.7329366579253103,
      "learning_rate": 0.00016089439749207443,
      "loss": 12.4546,
      "step": 7156
    },
    {
      "epoch": 1.2514972677595628,
      "grad_norm": 0.9035725976458332,
      "learning_rate": 0.00016088316427902665,
      "loss": 12.8136,
      "step": 7157
    },
    {
      "epoch": 1.251672131147541,
      "grad_norm": 0.7957704304356237,
      "learning_rate": 0.00016087192984508451,
      "loss": 12.6831,
      "step": 7158
    },
    {
      "epoch": 1.251846994535519,
      "grad_norm": 0.7047177312957762,
      "learning_rate": 0.00016086069419047328,
      "loss": 12.3889,
      "step": 7159
    },
    {
      "epoch": 1.2520218579234972,
      "grad_norm": 0.7666628761630223,
      "learning_rate": 0.00016084945731541827,
      "loss": 12.4899,
      "step": 7160
    },
    {
      "epoch": 1.2521967213114755,
      "grad_norm": 0.7976716385878727,
      "learning_rate": 0.00016083821922014478,
      "loss": 12.4899,
      "step": 7161
    },
    {
      "epoch": 1.2523715846994534,
      "grad_norm": 0.7622694225055232,
      "learning_rate": 0.0001608269799048782,
      "loss": 12.4525,
      "step": 7162
    },
    {
      "epoch": 1.2525464480874318,
      "grad_norm": 0.7729260275742179,
      "learning_rate": 0.00016081573936984397,
      "loss": 12.5817,
      "step": 7163
    },
    {
      "epoch": 1.2527213114754099,
      "grad_norm": 0.779494613518286,
      "learning_rate": 0.00016080449761526734,
      "loss": 12.2626,
      "step": 7164
    },
    {
      "epoch": 1.252896174863388,
      "grad_norm": 0.746470157320161,
      "learning_rate": 0.0001607932546413739,
      "loss": 12.5983,
      "step": 7165
    },
    {
      "epoch": 1.2530710382513661,
      "grad_norm": 0.7386730940397416,
      "learning_rate": 0.00016078201044838903,
      "loss": 12.3225,
      "step": 7166
    },
    {
      "epoch": 1.2532459016393442,
      "grad_norm": 0.7192889544907634,
      "learning_rate": 0.00016077076503653826,
      "loss": 12.5306,
      "step": 7167
    },
    {
      "epoch": 1.2534207650273224,
      "grad_norm": 0.7163274803702788,
      "learning_rate": 0.000160759518406047,
      "loss": 12.3052,
      "step": 7168
    },
    {
      "epoch": 1.2535956284153005,
      "grad_norm": 0.7424130643436331,
      "learning_rate": 0.00016074827055714085,
      "loss": 12.4447,
      "step": 7169
    },
    {
      "epoch": 1.2537704918032788,
      "grad_norm": 0.7722917766811107,
      "learning_rate": 0.00016073702149004538,
      "loss": 12.5041,
      "step": 7170
    },
    {
      "epoch": 1.2539453551912567,
      "grad_norm": 0.7367012559331085,
      "learning_rate": 0.00016072577120498615,
      "loss": 12.6818,
      "step": 7171
    },
    {
      "epoch": 1.254120218579235,
      "grad_norm": 0.747271668626147,
      "learning_rate": 0.00016071451970218873,
      "loss": 12.5438,
      "step": 7172
    },
    {
      "epoch": 1.2542950819672132,
      "grad_norm": 0.7500945983043659,
      "learning_rate": 0.0001607032669818788,
      "loss": 12.5871,
      "step": 7173
    },
    {
      "epoch": 1.2544699453551913,
      "grad_norm": 0.7792808054088941,
      "learning_rate": 0.00016069201304428195,
      "loss": 12.5893,
      "step": 7174
    },
    {
      "epoch": 1.2546448087431694,
      "grad_norm": 0.8764376623306592,
      "learning_rate": 0.0001606807578896239,
      "loss": 12.7096,
      "step": 7175
    },
    {
      "epoch": 1.2548196721311475,
      "grad_norm": 0.7941760689012094,
      "learning_rate": 0.00016066950151813033,
      "loss": 12.5896,
      "step": 7176
    },
    {
      "epoch": 1.2549945355191257,
      "grad_norm": 0.720095584017881,
      "learning_rate": 0.000160658243930027,
      "loss": 12.3906,
      "step": 7177
    },
    {
      "epoch": 1.2551693989071038,
      "grad_norm": 0.7956566273523129,
      "learning_rate": 0.00016064698512553962,
      "loss": 12.7211,
      "step": 7178
    },
    {
      "epoch": 1.255344262295082,
      "grad_norm": 0.7509337580346264,
      "learning_rate": 0.000160635725104894,
      "loss": 12.4424,
      "step": 7179
    },
    {
      "epoch": 1.25551912568306,
      "grad_norm": 0.7536384578847598,
      "learning_rate": 0.00016062446386831588,
      "loss": 12.6179,
      "step": 7180
    },
    {
      "epoch": 1.2556939890710384,
      "grad_norm": 0.9382398654650066,
      "learning_rate": 0.00016061320141603114,
      "loss": 12.7717,
      "step": 7181
    },
    {
      "epoch": 1.2558688524590165,
      "grad_norm": 0.7025877997761366,
      "learning_rate": 0.00016060193774826558,
      "loss": 12.568,
      "step": 7182
    },
    {
      "epoch": 1.2560437158469946,
      "grad_norm": 0.8324242552672124,
      "learning_rate": 0.0001605906728652451,
      "loss": 12.4302,
      "step": 7183
    },
    {
      "epoch": 1.2562185792349727,
      "grad_norm": 0.8277592631185987,
      "learning_rate": 0.00016057940676719563,
      "loss": 12.5115,
      "step": 7184
    },
    {
      "epoch": 1.2563934426229508,
      "grad_norm": 0.754235377143343,
      "learning_rate": 0.00016056813945434302,
      "loss": 12.6111,
      "step": 7185
    },
    {
      "epoch": 1.256568306010929,
      "grad_norm": 0.7234418676231684,
      "learning_rate": 0.00016055687092691326,
      "loss": 12.477,
      "step": 7186
    },
    {
      "epoch": 1.256743169398907,
      "grad_norm": 0.9142604087821218,
      "learning_rate": 0.00016054560118513228,
      "loss": 12.3093,
      "step": 7187
    },
    {
      "epoch": 1.2569180327868852,
      "grad_norm": 0.7682567734387605,
      "learning_rate": 0.00016053433022922612,
      "loss": 12.5482,
      "step": 7188
    },
    {
      "epoch": 1.2570928961748633,
      "grad_norm": 0.8166652627354167,
      "learning_rate": 0.00016052305805942075,
      "loss": 12.6564,
      "step": 7189
    },
    {
      "epoch": 1.2572677595628416,
      "grad_norm": 0.7246360574980955,
      "learning_rate": 0.00016051178467594226,
      "loss": 12.4755,
      "step": 7190
    },
    {
      "epoch": 1.2574426229508195,
      "grad_norm": 0.8350731999641625,
      "learning_rate": 0.00016050051007901667,
      "loss": 12.6261,
      "step": 7191
    },
    {
      "epoch": 1.2576174863387979,
      "grad_norm": 0.7594841062115782,
      "learning_rate": 0.00016048923426887013,
      "loss": 12.4648,
      "step": 7192
    },
    {
      "epoch": 1.257792349726776,
      "grad_norm": 0.8361263923133123,
      "learning_rate": 0.00016047795724572866,
      "loss": 12.4765,
      "step": 7193
    },
    {
      "epoch": 1.2579672131147541,
      "grad_norm": 0.7430983444690351,
      "learning_rate": 0.0001604666790098185,
      "loss": 12.4357,
      "step": 7194
    },
    {
      "epoch": 1.2581420765027322,
      "grad_norm": 0.8411877752456935,
      "learning_rate": 0.00016045539956136576,
      "loss": 12.6373,
      "step": 7195
    },
    {
      "epoch": 1.2583169398907104,
      "grad_norm": 0.7546446629991604,
      "learning_rate": 0.00016044411890059665,
      "loss": 12.4708,
      "step": 7196
    },
    {
      "epoch": 1.2584918032786885,
      "grad_norm": 0.8158895991140461,
      "learning_rate": 0.00016043283702773734,
      "loss": 12.6985,
      "step": 7197
    },
    {
      "epoch": 1.2586666666666666,
      "grad_norm": 0.8145548633362942,
      "learning_rate": 0.0001604215539430141,
      "loss": 12.577,
      "step": 7198
    },
    {
      "epoch": 1.258841530054645,
      "grad_norm": 0.7580115548404948,
      "learning_rate": 0.0001604102696466532,
      "loss": 12.5739,
      "step": 7199
    },
    {
      "epoch": 1.2590163934426228,
      "grad_norm": 0.8305657796030627,
      "learning_rate": 0.0001603989841388809,
      "loss": 12.5243,
      "step": 7200
    },
    {
      "epoch": 1.2591912568306012,
      "grad_norm": 0.8380545310209214,
      "learning_rate": 0.00016038769741992352,
      "loss": 12.6703,
      "step": 7201
    },
    {
      "epoch": 1.2593661202185793,
      "grad_norm": 0.7543826372941559,
      "learning_rate": 0.00016037640949000738,
      "loss": 12.5216,
      "step": 7202
    },
    {
      "epoch": 1.2595409836065574,
      "grad_norm": 0.7466905603318494,
      "learning_rate": 0.00016036512034935886,
      "loss": 12.4482,
      "step": 7203
    },
    {
      "epoch": 1.2597158469945355,
      "grad_norm": 0.7461358007388673,
      "learning_rate": 0.00016035382999820435,
      "loss": 12.5286,
      "step": 7204
    },
    {
      "epoch": 1.2598907103825137,
      "grad_norm": 0.779151256761305,
      "learning_rate": 0.0001603425384367702,
      "loss": 12.6184,
      "step": 7205
    },
    {
      "epoch": 1.2600655737704918,
      "grad_norm": 0.749148401679659,
      "learning_rate": 0.0001603312456652829,
      "loss": 12.658,
      "step": 7206
    },
    {
      "epoch": 1.26024043715847,
      "grad_norm": 0.7370124334355982,
      "learning_rate": 0.0001603199516839689,
      "loss": 12.5809,
      "step": 7207
    },
    {
      "epoch": 1.260415300546448,
      "grad_norm": 0.7601757043227334,
      "learning_rate": 0.00016030865649305462,
      "loss": 12.5164,
      "step": 7208
    },
    {
      "epoch": 1.2605901639344261,
      "grad_norm": 0.7208946288810064,
      "learning_rate": 0.00016029736009276664,
      "loss": 12.4327,
      "step": 7209
    },
    {
      "epoch": 1.2607650273224045,
      "grad_norm": 0.7330930563492875,
      "learning_rate": 0.00016028606248333143,
      "loss": 12.6436,
      "step": 7210
    },
    {
      "epoch": 1.2609398907103826,
      "grad_norm": 0.710482307424686,
      "learning_rate": 0.0001602747636649756,
      "loss": 12.4784,
      "step": 7211
    },
    {
      "epoch": 1.2611147540983607,
      "grad_norm": 0.7544123020854285,
      "learning_rate": 0.00016026346363792567,
      "loss": 12.4736,
      "step": 7212
    },
    {
      "epoch": 1.2612896174863388,
      "grad_norm": 0.8698166636278017,
      "learning_rate": 0.00016025216240240825,
      "loss": 12.4635,
      "step": 7213
    },
    {
      "epoch": 1.261464480874317,
      "grad_norm": 0.7161859876161651,
      "learning_rate": 0.00016024085995864997,
      "loss": 12.474,
      "step": 7214
    },
    {
      "epoch": 1.261639344262295,
      "grad_norm": 0.7858352876394048,
      "learning_rate": 0.0001602295563068775,
      "loss": 12.5954,
      "step": 7215
    },
    {
      "epoch": 1.2618142076502732,
      "grad_norm": 0.7381906236419725,
      "learning_rate": 0.00016021825144731748,
      "loss": 12.4384,
      "step": 7216
    },
    {
      "epoch": 1.2619890710382513,
      "grad_norm": 0.7554776174190828,
      "learning_rate": 0.00016020694538019666,
      "loss": 12.5763,
      "step": 7217
    },
    {
      "epoch": 1.2621639344262294,
      "grad_norm": 0.7601705287406731,
      "learning_rate": 0.0001601956381057417,
      "loss": 12.4967,
      "step": 7218
    },
    {
      "epoch": 1.2623387978142078,
      "grad_norm": 0.7271643419890101,
      "learning_rate": 0.00016018432962417936,
      "loss": 12.4485,
      "step": 7219
    },
    {
      "epoch": 1.2625136612021857,
      "grad_norm": 0.7829064403403359,
      "learning_rate": 0.00016017301993573644,
      "loss": 12.5658,
      "step": 7220
    },
    {
      "epoch": 1.262688524590164,
      "grad_norm": 0.7201338173033327,
      "learning_rate": 0.0001601617090406397,
      "loss": 12.4927,
      "step": 7221
    },
    {
      "epoch": 1.2628633879781421,
      "grad_norm": 0.727220611791474,
      "learning_rate": 0.000160150396939116,
      "loss": 12.6744,
      "step": 7222
    },
    {
      "epoch": 1.2630382513661202,
      "grad_norm": 0.6805243172827659,
      "learning_rate": 0.00016013908363139216,
      "loss": 12.582,
      "step": 7223
    },
    {
      "epoch": 1.2632131147540984,
      "grad_norm": 0.7499664495934255,
      "learning_rate": 0.000160127769117695,
      "loss": 12.4231,
      "step": 7224
    },
    {
      "epoch": 1.2633879781420765,
      "grad_norm": 0.7178449101065147,
      "learning_rate": 0.00016011645339825148,
      "loss": 12.3258,
      "step": 7225
    },
    {
      "epoch": 1.2635628415300546,
      "grad_norm": 0.7027106365385635,
      "learning_rate": 0.0001601051364732885,
      "loss": 12.6887,
      "step": 7226
    },
    {
      "epoch": 1.2637377049180327,
      "grad_norm": 0.8170048657210034,
      "learning_rate": 0.00016009381834303295,
      "loss": 12.5463,
      "step": 7227
    },
    {
      "epoch": 1.263912568306011,
      "grad_norm": 0.6749534454054568,
      "learning_rate": 0.00016008249900771184,
      "loss": 12.1726,
      "step": 7228
    },
    {
      "epoch": 1.264087431693989,
      "grad_norm": 0.8388898449438451,
      "learning_rate": 0.00016007117846755214,
      "loss": 12.509,
      "step": 7229
    },
    {
      "epoch": 1.2642622950819673,
      "grad_norm": 0.7708219093483464,
      "learning_rate": 0.00016005985672278093,
      "loss": 12.5193,
      "step": 7230
    },
    {
      "epoch": 1.2644371584699454,
      "grad_norm": 0.8193989065171823,
      "learning_rate": 0.0001600485337736251,
      "loss": 12.5413,
      "step": 7231
    },
    {
      "epoch": 1.2646120218579235,
      "grad_norm": 0.7620133133731792,
      "learning_rate": 0.00016003720962031186,
      "loss": 12.3717,
      "step": 7232
    },
    {
      "epoch": 1.2647868852459017,
      "grad_norm": 0.8031885314424789,
      "learning_rate": 0.0001600258842630682,
      "loss": 12.4033,
      "step": 7233
    },
    {
      "epoch": 1.2649617486338798,
      "grad_norm": 0.747583539572594,
      "learning_rate": 0.00016001455770212126,
      "loss": 12.4026,
      "step": 7234
    },
    {
      "epoch": 1.265136612021858,
      "grad_norm": 0.7905309710325068,
      "learning_rate": 0.00016000322993769817,
      "loss": 12.4156,
      "step": 7235
    },
    {
      "epoch": 1.265311475409836,
      "grad_norm": 0.765126480332747,
      "learning_rate": 0.00015999190097002608,
      "loss": 12.5928,
      "step": 7236
    },
    {
      "epoch": 1.2654863387978141,
      "grad_norm": 0.8031244759876178,
      "learning_rate": 0.00015998057079933217,
      "loss": 12.2388,
      "step": 7237
    },
    {
      "epoch": 1.2656612021857923,
      "grad_norm": 0.7842233441917086,
      "learning_rate": 0.00015996923942584367,
      "loss": 12.6431,
      "step": 7238
    },
    {
      "epoch": 1.2658360655737706,
      "grad_norm": 0.8045504607991935,
      "learning_rate": 0.0001599579068497878,
      "loss": 12.44,
      "step": 7239
    },
    {
      "epoch": 1.2660109289617487,
      "grad_norm": 0.8492550473564314,
      "learning_rate": 0.00015994657307139178,
      "loss": 12.4782,
      "step": 7240
    },
    {
      "epoch": 1.2661857923497268,
      "grad_norm": 0.8539612248128234,
      "learning_rate": 0.0001599352380908829,
      "loss": 12.5217,
      "step": 7241
    },
    {
      "epoch": 1.266360655737705,
      "grad_norm": 0.7592954239099104,
      "learning_rate": 0.0001599239019084885,
      "loss": 12.4527,
      "step": 7242
    },
    {
      "epoch": 1.266535519125683,
      "grad_norm": 0.7658810306589223,
      "learning_rate": 0.00015991256452443583,
      "loss": 12.3566,
      "step": 7243
    },
    {
      "epoch": 1.2667103825136612,
      "grad_norm": 0.7423321940679389,
      "learning_rate": 0.00015990122593895233,
      "loss": 12.6705,
      "step": 7244
    },
    {
      "epoch": 1.2668852459016393,
      "grad_norm": 0.7864220416780269,
      "learning_rate": 0.0001598898861522653,
      "loss": 12.7493,
      "step": 7245
    },
    {
      "epoch": 1.2670601092896174,
      "grad_norm": 0.7332017144478513,
      "learning_rate": 0.0001598785451646022,
      "loss": 12.6778,
      "step": 7246
    },
    {
      "epoch": 1.2672349726775956,
      "grad_norm": 0.8067384686301384,
      "learning_rate": 0.0001598672029761904,
      "loss": 12.4163,
      "step": 7247
    },
    {
      "epoch": 1.267409836065574,
      "grad_norm": 0.8707282860279013,
      "learning_rate": 0.00015985585958725736,
      "loss": 12.4933,
      "step": 7248
    },
    {
      "epoch": 1.2675846994535518,
      "grad_norm": 0.8427909661802252,
      "learning_rate": 0.00015984451499803055,
      "loss": 12.5548,
      "step": 7249
    },
    {
      "epoch": 1.2677595628415301,
      "grad_norm": 0.7632734615154657,
      "learning_rate": 0.0001598331692087375,
      "loss": 12.5133,
      "step": 7250
    },
    {
      "epoch": 1.2679344262295082,
      "grad_norm": 0.7907010556229366,
      "learning_rate": 0.00015982182221960565,
      "loss": 12.4112,
      "step": 7251
    },
    {
      "epoch": 1.2681092896174864,
      "grad_norm": 0.8152718413148308,
      "learning_rate": 0.00015981047403086263,
      "loss": 12.4592,
      "step": 7252
    },
    {
      "epoch": 1.2682841530054645,
      "grad_norm": 0.7135355738849913,
      "learning_rate": 0.00015979912464273593,
      "loss": 12.317,
      "step": 7253
    },
    {
      "epoch": 1.2684590163934426,
      "grad_norm": 0.8805598554514275,
      "learning_rate": 0.0001597877740554532,
      "loss": 12.6573,
      "step": 7254
    },
    {
      "epoch": 1.2686338797814207,
      "grad_norm": 0.7343110367828473,
      "learning_rate": 0.000159776422269242,
      "loss": 12.4031,
      "step": 7255
    },
    {
      "epoch": 1.2688087431693988,
      "grad_norm": 0.7939552633336454,
      "learning_rate": 0.00015976506928433,
      "loss": 12.5384,
      "step": 7256
    },
    {
      "epoch": 1.2689836065573772,
      "grad_norm": 0.7719084217275171,
      "learning_rate": 0.00015975371510094485,
      "loss": 12.548,
      "step": 7257
    },
    {
      "epoch": 1.269158469945355,
      "grad_norm": 0.9170874399608314,
      "learning_rate": 0.00015974235971931428,
      "loss": 11.8944,
      "step": 7258
    },
    {
      "epoch": 1.2693333333333334,
      "grad_norm": 0.9037408481955516,
      "learning_rate": 0.00015973100313966597,
      "loss": 12.592,
      "step": 7259
    },
    {
      "epoch": 1.2695081967213115,
      "grad_norm": 0.9855888061760495,
      "learning_rate": 0.0001597196453622276,
      "loss": 12.7492,
      "step": 7260
    },
    {
      "epoch": 1.2696830601092897,
      "grad_norm": 0.8141285001961859,
      "learning_rate": 0.00015970828638722703,
      "loss": 12.5305,
      "step": 7261
    },
    {
      "epoch": 1.2698579234972678,
      "grad_norm": 0.7937887635542372,
      "learning_rate": 0.00015969692621489192,
      "loss": 12.5119,
      "step": 7262
    },
    {
      "epoch": 1.270032786885246,
      "grad_norm": 0.9830693618633536,
      "learning_rate": 0.0001596855648454502,
      "loss": 12.5662,
      "step": 7263
    },
    {
      "epoch": 1.270207650273224,
      "grad_norm": 0.9510149243978265,
      "learning_rate": 0.00015967420227912965,
      "loss": 12.246,
      "step": 7264
    },
    {
      "epoch": 1.2703825136612021,
      "grad_norm": 0.7989227218857904,
      "learning_rate": 0.0001596628385161581,
      "loss": 12.3986,
      "step": 7265
    },
    {
      "epoch": 1.2705573770491803,
      "grad_norm": 0.8130102362545164,
      "learning_rate": 0.00015965147355676343,
      "loss": 12.5251,
      "step": 7266
    },
    {
      "epoch": 1.2707322404371584,
      "grad_norm": 0.779205656333399,
      "learning_rate": 0.00015964010740117358,
      "loss": 12.4786,
      "step": 7267
    },
    {
      "epoch": 1.2709071038251367,
      "grad_norm": 0.7420161542127446,
      "learning_rate": 0.00015962874004961646,
      "loss": 12.4211,
      "step": 7268
    },
    {
      "epoch": 1.2710819672131146,
      "grad_norm": 0.7056551346066374,
      "learning_rate": 0.00015961737150232,
      "loss": 12.312,
      "step": 7269
    },
    {
      "epoch": 1.271256830601093,
      "grad_norm": 0.7438426837165221,
      "learning_rate": 0.00015960600175951223,
      "loss": 12.5871,
      "step": 7270
    },
    {
      "epoch": 1.271431693989071,
      "grad_norm": 0.8076994387031039,
      "learning_rate": 0.00015959463082142113,
      "loss": 12.5734,
      "step": 7271
    },
    {
      "epoch": 1.2716065573770492,
      "grad_norm": 0.7060698519202274,
      "learning_rate": 0.0001595832586882746,
      "loss": 12.5185,
      "step": 7272
    },
    {
      "epoch": 1.2717814207650273,
      "grad_norm": 0.9146522631335191,
      "learning_rate": 0.00015957188536030087,
      "loss": 12.4311,
      "step": 7273
    },
    {
      "epoch": 1.2719562841530054,
      "grad_norm": 0.6918321283836462,
      "learning_rate": 0.0001595605108377279,
      "loss": 12.528,
      "step": 7274
    },
    {
      "epoch": 1.2721311475409836,
      "grad_norm": 0.8518003815528844,
      "learning_rate": 0.00015954913512078385,
      "loss": 11.9143,
      "step": 7275
    },
    {
      "epoch": 1.2723060109289617,
      "grad_norm": 0.8026807687742217,
      "learning_rate": 0.00015953775820969676,
      "loss": 12.4734,
      "step": 7276
    },
    {
      "epoch": 1.27248087431694,
      "grad_norm": 0.8142929221257567,
      "learning_rate": 0.00015952638010469482,
      "loss": 12.5365,
      "step": 7277
    },
    {
      "epoch": 1.272655737704918,
      "grad_norm": 0.6929866413669404,
      "learning_rate": 0.00015951500080600624,
      "loss": 12.4512,
      "step": 7278
    },
    {
      "epoch": 1.2728306010928963,
      "grad_norm": 0.7897175822792794,
      "learning_rate": 0.00015950362031385908,
      "loss": 12.5606,
      "step": 7279
    },
    {
      "epoch": 1.2730054644808744,
      "grad_norm": 0.7473347218091817,
      "learning_rate": 0.00015949223862848168,
      "loss": 12.7605,
      "step": 7280
    },
    {
      "epoch": 1.2731803278688525,
      "grad_norm": 0.7213618801779323,
      "learning_rate": 0.00015948085575010222,
      "loss": 12.5407,
      "step": 7281
    },
    {
      "epoch": 1.2733551912568306,
      "grad_norm": 0.7033679051786862,
      "learning_rate": 0.00015946947167894896,
      "loss": 12.6699,
      "step": 7282
    },
    {
      "epoch": 1.2735300546448087,
      "grad_norm": 0.7386168541868597,
      "learning_rate": 0.00015945808641525022,
      "loss": 12.7122,
      "step": 7283
    },
    {
      "epoch": 1.2737049180327868,
      "grad_norm": 0.7362984006608895,
      "learning_rate": 0.00015944669995923426,
      "loss": 12.3777,
      "step": 7284
    },
    {
      "epoch": 1.273879781420765,
      "grad_norm": 0.701853702556391,
      "learning_rate": 0.00015943531231112944,
      "loss": 12.527,
      "step": 7285
    },
    {
      "epoch": 1.274054644808743,
      "grad_norm": 0.6927524742496594,
      "learning_rate": 0.00015942392347116413,
      "loss": 12.4545,
      "step": 7286
    },
    {
      "epoch": 1.2742295081967212,
      "grad_norm": 0.6877349113558899,
      "learning_rate": 0.00015941253343956671,
      "loss": 12.476,
      "step": 7287
    },
    {
      "epoch": 1.2744043715846995,
      "grad_norm": 0.7322438755818764,
      "learning_rate": 0.00015940114221656558,
      "loss": 12.5866,
      "step": 7288
    },
    {
      "epoch": 1.2745792349726777,
      "grad_norm": 0.721851208731775,
      "learning_rate": 0.00015938974980238914,
      "loss": 12.4311,
      "step": 7289
    },
    {
      "epoch": 1.2747540983606558,
      "grad_norm": 0.8505333911037221,
      "learning_rate": 0.0001593783561972659,
      "loss": 12.7019,
      "step": 7290
    },
    {
      "epoch": 1.274928961748634,
      "grad_norm": 0.7539867282219941,
      "learning_rate": 0.00015936696140142424,
      "loss": 12.6051,
      "step": 7291
    },
    {
      "epoch": 1.275103825136612,
      "grad_norm": 0.7308250761843161,
      "learning_rate": 0.00015935556541509278,
      "loss": 12.4347,
      "step": 7292
    },
    {
      "epoch": 1.2752786885245901,
      "grad_norm": 0.7047132479580122,
      "learning_rate": 0.00015934416823849997,
      "loss": 12.5914,
      "step": 7293
    },
    {
      "epoch": 1.2754535519125683,
      "grad_norm": 0.7651739250219585,
      "learning_rate": 0.00015933276987187438,
      "loss": 12.6737,
      "step": 7294
    },
    {
      "epoch": 1.2756284153005464,
      "grad_norm": 0.7634184643537034,
      "learning_rate": 0.0001593213703154446,
      "loss": 12.6302,
      "step": 7295
    },
    {
      "epoch": 1.2758032786885245,
      "grad_norm": 0.7450573448090796,
      "learning_rate": 0.00015930996956943914,
      "loss": 12.6705,
      "step": 7296
    },
    {
      "epoch": 1.2759781420765028,
      "grad_norm": 0.7392487643998229,
      "learning_rate": 0.00015929856763408673,
      "loss": 12.5474,
      "step": 7297
    },
    {
      "epoch": 1.2761530054644807,
      "grad_norm": 0.7648108995012366,
      "learning_rate": 0.00015928716450961593,
      "loss": 12.4463,
      "step": 7298
    },
    {
      "epoch": 1.276327868852459,
      "grad_norm": 0.7107956000711183,
      "learning_rate": 0.0001592757601962555,
      "loss": 12.6148,
      "step": 7299
    },
    {
      "epoch": 1.2765027322404372,
      "grad_norm": 0.9117402530619,
      "learning_rate": 0.00015926435469423405,
      "loss": 12.5838,
      "step": 7300
    },
    {
      "epoch": 1.2766775956284153,
      "grad_norm": 0.9009295487230434,
      "learning_rate": 0.00015925294800378032,
      "loss": 12.3407,
      "step": 7301
    },
    {
      "epoch": 1.2768524590163934,
      "grad_norm": 0.8280821291491222,
      "learning_rate": 0.00015924154012512308,
      "loss": 12.4036,
      "step": 7302
    },
    {
      "epoch": 1.2770273224043716,
      "grad_norm": 0.8171576378491809,
      "learning_rate": 0.00015923013105849102,
      "loss": 12.5353,
      "step": 7303
    },
    {
      "epoch": 1.2772021857923497,
      "grad_norm": 0.8184644882993845,
      "learning_rate": 0.00015921872080411302,
      "loss": 12.3949,
      "step": 7304
    },
    {
      "epoch": 1.2773770491803278,
      "grad_norm": 0.8605507962623897,
      "learning_rate": 0.0001592073093622178,
      "loss": 12.5772,
      "step": 7305
    },
    {
      "epoch": 1.2775519125683061,
      "grad_norm": 0.7659242062333016,
      "learning_rate": 0.00015919589673303426,
      "loss": 12.3429,
      "step": 7306
    },
    {
      "epoch": 1.277726775956284,
      "grad_norm": 0.8652181809907562,
      "learning_rate": 0.00015918448291679123,
      "loss": 12.4032,
      "step": 7307
    },
    {
      "epoch": 1.2779016393442624,
      "grad_norm": 0.8072783538860001,
      "learning_rate": 0.0001591730679137176,
      "loss": 12.5204,
      "step": 7308
    },
    {
      "epoch": 1.2780765027322405,
      "grad_norm": 0.9038543020713911,
      "learning_rate": 0.00015916165172404228,
      "loss": 12.5754,
      "step": 7309
    },
    {
      "epoch": 1.2782513661202186,
      "grad_norm": 0.9782799042414179,
      "learning_rate": 0.00015915023434799415,
      "loss": 12.7337,
      "step": 7310
    },
    {
      "epoch": 1.2784262295081967,
      "grad_norm": 0.955136634637519,
      "learning_rate": 0.00015913881578580227,
      "loss": 12.5659,
      "step": 7311
    },
    {
      "epoch": 1.2786010928961749,
      "grad_norm": 0.8689358039440901,
      "learning_rate": 0.00015912739603769554,
      "loss": 12.4822,
      "step": 7312
    },
    {
      "epoch": 1.278775956284153,
      "grad_norm": 0.7807092691162008,
      "learning_rate": 0.00015911597510390296,
      "loss": 12.4368,
      "step": 7313
    },
    {
      "epoch": 1.278950819672131,
      "grad_norm": 1.1023119635678322,
      "learning_rate": 0.00015910455298465356,
      "loss": 12.699,
      "step": 7314
    },
    {
      "epoch": 1.2791256830601092,
      "grad_norm": 0.8611033142562232,
      "learning_rate": 0.0001590931296801764,
      "loss": 12.5091,
      "step": 7315
    },
    {
      "epoch": 1.2793005464480873,
      "grad_norm": 0.9836645981023397,
      "learning_rate": 0.00015908170519070056,
      "loss": 12.6091,
      "step": 7316
    },
    {
      "epoch": 1.2794754098360657,
      "grad_norm": 0.8784655272607123,
      "learning_rate": 0.0001590702795164551,
      "loss": 12.5121,
      "step": 7317
    },
    {
      "epoch": 1.2796502732240438,
      "grad_norm": 0.9446672019571026,
      "learning_rate": 0.0001590588526576692,
      "loss": 12.3729,
      "step": 7318
    },
    {
      "epoch": 1.279825136612022,
      "grad_norm": 0.7983396129547927,
      "learning_rate": 0.000159047424614572,
      "loss": 12.4427,
      "step": 7319
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9273997180254754,
      "learning_rate": 0.00015903599538739254,
      "loss": 12.6851,
      "step": 7320
    },
    {
      "epoch": 1.2801748633879781,
      "grad_norm": 0.7785604766125052,
      "learning_rate": 0.0001590245649763602,
      "loss": 12.6689,
      "step": 7321
    },
    {
      "epoch": 1.2803497267759563,
      "grad_norm": 0.8594204393394415,
      "learning_rate": 0.00015901313338170407,
      "loss": 12.5116,
      "step": 7322
    },
    {
      "epoch": 1.2805245901639344,
      "grad_norm": 0.8036780113765342,
      "learning_rate": 0.0001590017006036534,
      "loss": 12.3007,
      "step": 7323
    },
    {
      "epoch": 1.2806994535519125,
      "grad_norm": 0.8169783763319587,
      "learning_rate": 0.0001589902666424375,
      "loss": 12.7162,
      "step": 7324
    },
    {
      "epoch": 1.2808743169398906,
      "grad_norm": 0.7970357395190248,
      "learning_rate": 0.00015897883149828564,
      "loss": 12.5144,
      "step": 7325
    },
    {
      "epoch": 1.281049180327869,
      "grad_norm": 0.7685634388394175,
      "learning_rate": 0.0001589673951714271,
      "loss": 12.5016,
      "step": 7326
    },
    {
      "epoch": 1.2812240437158469,
      "grad_norm": 0.790060197198123,
      "learning_rate": 0.00015895595766209121,
      "loss": 12.2686,
      "step": 7327
    },
    {
      "epoch": 1.2813989071038252,
      "grad_norm": 0.688498233354961,
      "learning_rate": 0.00015894451897050738,
      "loss": 12.5389,
      "step": 7328
    },
    {
      "epoch": 1.2815737704918033,
      "grad_norm": 0.75974870761688,
      "learning_rate": 0.00015893307909690493,
      "loss": 12.6164,
      "step": 7329
    },
    {
      "epoch": 1.2817486338797814,
      "grad_norm": 0.7250530915229292,
      "learning_rate": 0.00015892163804151333,
      "loss": 12.7295,
      "step": 7330
    },
    {
      "epoch": 1.2819234972677596,
      "grad_norm": 0.7609023096716873,
      "learning_rate": 0.000158910195804562,
      "loss": 12.5332,
      "step": 7331
    },
    {
      "epoch": 1.2820983606557377,
      "grad_norm": 0.7351550656497019,
      "learning_rate": 0.00015889875238628032,
      "loss": 11.9798,
      "step": 7332
    },
    {
      "epoch": 1.2822732240437158,
      "grad_norm": 0.7307864861795976,
      "learning_rate": 0.00015888730778689783,
      "loss": 12.7272,
      "step": 7333
    },
    {
      "epoch": 1.282448087431694,
      "grad_norm": 0.7482394524038956,
      "learning_rate": 0.000158875862006644,
      "loss": 12.8301,
      "step": 7334
    },
    {
      "epoch": 1.2826229508196723,
      "grad_norm": 0.7482021170274455,
      "learning_rate": 0.00015886441504574838,
      "loss": 12.4418,
      "step": 7335
    },
    {
      "epoch": 1.2827978142076502,
      "grad_norm": 0.7103271979926342,
      "learning_rate": 0.00015885296690444048,
      "loss": 12.3904,
      "step": 7336
    },
    {
      "epoch": 1.2829726775956285,
      "grad_norm": 0.6684199324295161,
      "learning_rate": 0.0001588415175829499,
      "loss": 12.4279,
      "step": 7337
    },
    {
      "epoch": 1.2831475409836066,
      "grad_norm": 0.7644819909913678,
      "learning_rate": 0.00015883006708150623,
      "loss": 12.6221,
      "step": 7338
    },
    {
      "epoch": 1.2833224043715847,
      "grad_norm": 0.6971642709123079,
      "learning_rate": 0.00015881861540033906,
      "loss": 12.3456,
      "step": 7339
    },
    {
      "epoch": 1.2834972677595629,
      "grad_norm": 0.7027983071554995,
      "learning_rate": 0.0001588071625396781,
      "loss": 12.4131,
      "step": 7340
    },
    {
      "epoch": 1.283672131147541,
      "grad_norm": 0.7084847374027196,
      "learning_rate": 0.0001587957084997529,
      "loss": 12.4863,
      "step": 7341
    },
    {
      "epoch": 1.283846994535519,
      "grad_norm": 0.757304599221211,
      "learning_rate": 0.0001587842532807933,
      "loss": 12.681,
      "step": 7342
    },
    {
      "epoch": 1.2840218579234972,
      "grad_norm": 0.8442851269191362,
      "learning_rate": 0.0001587727968830289,
      "loss": 12.8642,
      "step": 7343
    },
    {
      "epoch": 1.2841967213114753,
      "grad_norm": 0.7272478279302748,
      "learning_rate": 0.00015876133930668943,
      "loss": 12.6233,
      "step": 7344
    },
    {
      "epoch": 1.2843715846994535,
      "grad_norm": 0.7403397859184443,
      "learning_rate": 0.00015874988055200473,
      "loss": 12.6058,
      "step": 7345
    },
    {
      "epoch": 1.2845464480874318,
      "grad_norm": 0.7323400959714164,
      "learning_rate": 0.00015873842061920452,
      "loss": 12.5884,
      "step": 7346
    },
    {
      "epoch": 1.28472131147541,
      "grad_norm": 0.7301396723235416,
      "learning_rate": 0.0001587269595085186,
      "loss": 12.3697,
      "step": 7347
    },
    {
      "epoch": 1.284896174863388,
      "grad_norm": 0.7483368114481923,
      "learning_rate": 0.0001587154972201769,
      "loss": 11.933,
      "step": 7348
    },
    {
      "epoch": 1.2850710382513661,
      "grad_norm": 0.6962278559215439,
      "learning_rate": 0.00015870403375440914,
      "loss": 12.4581,
      "step": 7349
    },
    {
      "epoch": 1.2852459016393443,
      "grad_norm": 0.8312612490874659,
      "learning_rate": 0.00015869256911144527,
      "loss": 12.484,
      "step": 7350
    },
    {
      "epoch": 1.2854207650273224,
      "grad_norm": 0.8652624876943301,
      "learning_rate": 0.0001586811032915152,
      "loss": 12.7748,
      "step": 7351
    },
    {
      "epoch": 1.2855956284153005,
      "grad_norm": 0.8540150263437252,
      "learning_rate": 0.00015866963629484882,
      "loss": 12.6262,
      "step": 7352
    },
    {
      "epoch": 1.2857704918032786,
      "grad_norm": 0.739175856805589,
      "learning_rate": 0.00015865816812167606,
      "loss": 12.7132,
      "step": 7353
    },
    {
      "epoch": 1.2859453551912567,
      "grad_norm": 0.7079393584322651,
      "learning_rate": 0.00015864669877222696,
      "loss": 12.4449,
      "step": 7354
    },
    {
      "epoch": 1.286120218579235,
      "grad_norm": 0.7440723310711553,
      "learning_rate": 0.00015863522824673148,
      "loss": 12.6624,
      "step": 7355
    },
    {
      "epoch": 1.286295081967213,
      "grad_norm": 0.7770632376012865,
      "learning_rate": 0.00015862375654541964,
      "loss": 12.4032,
      "step": 7356
    },
    {
      "epoch": 1.2864699453551913,
      "grad_norm": 0.7713714402922095,
      "learning_rate": 0.00015861228366852148,
      "loss": 12.6731,
      "step": 7357
    },
    {
      "epoch": 1.2866448087431694,
      "grad_norm": 0.6936882753671801,
      "learning_rate": 0.00015860080961626705,
      "loss": 12.5597,
      "step": 7358
    },
    {
      "epoch": 1.2868196721311476,
      "grad_norm": 0.7463602059588558,
      "learning_rate": 0.0001585893343888865,
      "loss": 12.3517,
      "step": 7359
    },
    {
      "epoch": 1.2869945355191257,
      "grad_norm": 0.864184416372597,
      "learning_rate": 0.00015857785798660987,
      "loss": 12.5045,
      "step": 7360
    },
    {
      "epoch": 1.2871693989071038,
      "grad_norm": 0.8960529239260182,
      "learning_rate": 0.00015856638040966735,
      "loss": 12.6392,
      "step": 7361
    },
    {
      "epoch": 1.287344262295082,
      "grad_norm": 0.7519460589143714,
      "learning_rate": 0.00015855490165828906,
      "loss": 12.625,
      "step": 7362
    },
    {
      "epoch": 1.28751912568306,
      "grad_norm": 0.7423037896520833,
      "learning_rate": 0.0001585434217327052,
      "loss": 12.5463,
      "step": 7363
    },
    {
      "epoch": 1.2876939890710384,
      "grad_norm": 0.7987024088479284,
      "learning_rate": 0.00015853194063314603,
      "loss": 12.3806,
      "step": 7364
    },
    {
      "epoch": 1.2878688524590163,
      "grad_norm": 0.7274503739919713,
      "learning_rate": 0.0001585204583598417,
      "loss": 12.4423,
      "step": 7365
    },
    {
      "epoch": 1.2880437158469946,
      "grad_norm": 0.8248330366971167,
      "learning_rate": 0.00015850897491302248,
      "loss": 12.5452,
      "step": 7366
    },
    {
      "epoch": 1.2882185792349727,
      "grad_norm": 0.8555694148472643,
      "learning_rate": 0.0001584974902929187,
      "loss": 12.5179,
      "step": 7367
    },
    {
      "epoch": 1.2883934426229509,
      "grad_norm": 0.7762112008217406,
      "learning_rate": 0.00015848600449976062,
      "loss": 12.1469,
      "step": 7368
    },
    {
      "epoch": 1.288568306010929,
      "grad_norm": 0.8455434278685395,
      "learning_rate": 0.00015847451753377856,
      "loss": 12.2672,
      "step": 7369
    },
    {
      "epoch": 1.288743169398907,
      "grad_norm": 0.7158360544854498,
      "learning_rate": 0.00015846302939520287,
      "loss": 12.5256,
      "step": 7370
    },
    {
      "epoch": 1.2889180327868852,
      "grad_norm": 0.9045111971269753,
      "learning_rate": 0.00015845154008426398,
      "loss": 12.5153,
      "step": 7371
    },
    {
      "epoch": 1.2890928961748633,
      "grad_norm": 0.74605047473326,
      "learning_rate": 0.00015844004960119222,
      "loss": 12.5174,
      "step": 7372
    },
    {
      "epoch": 1.2892677595628415,
      "grad_norm": 0.7461969325117249,
      "learning_rate": 0.00015842855794621802,
      "loss": 12.5471,
      "step": 7373
    },
    {
      "epoch": 1.2894426229508196,
      "grad_norm": 0.760135484205816,
      "learning_rate": 0.00015841706511957184,
      "loss": 12.3183,
      "step": 7374
    },
    {
      "epoch": 1.289617486338798,
      "grad_norm": 0.7132720466894927,
      "learning_rate": 0.00015840557112148414,
      "loss": 12.4532,
      "step": 7375
    },
    {
      "epoch": 1.289792349726776,
      "grad_norm": 0.8353443242222354,
      "learning_rate": 0.0001583940759521854,
      "loss": 12.3962,
      "step": 7376
    },
    {
      "epoch": 1.2899672131147542,
      "grad_norm": 0.7732196988384569,
      "learning_rate": 0.00015838257961190618,
      "loss": 12.7333,
      "step": 7377
    },
    {
      "epoch": 1.2901420765027323,
      "grad_norm": 0.8232822837978113,
      "learning_rate": 0.00015837108210087694,
      "loss": 12.5796,
      "step": 7378
    },
    {
      "epoch": 1.2903169398907104,
      "grad_norm": 0.7475371450279573,
      "learning_rate": 0.00015835958341932832,
      "loss": 12.6404,
      "step": 7379
    },
    {
      "epoch": 1.2904918032786885,
      "grad_norm": 0.8012915037124031,
      "learning_rate": 0.00015834808356749081,
      "loss": 12.4897,
      "step": 7380
    },
    {
      "epoch": 1.2906666666666666,
      "grad_norm": 0.7896245442315973,
      "learning_rate": 0.00015833658254559515,
      "loss": 12.515,
      "step": 7381
    },
    {
      "epoch": 1.2908415300546447,
      "grad_norm": 0.7023848467245446,
      "learning_rate": 0.00015832508035387184,
      "loss": 12.5955,
      "step": 7382
    },
    {
      "epoch": 1.2910163934426229,
      "grad_norm": 0.7124806439885275,
      "learning_rate": 0.00015831357699255157,
      "loss": 12.4426,
      "step": 7383
    },
    {
      "epoch": 1.2911912568306012,
      "grad_norm": 0.7762614629007872,
      "learning_rate": 0.00015830207246186507,
      "loss": 12.2791,
      "step": 7384
    },
    {
      "epoch": 1.291366120218579,
      "grad_norm": 0.7320781621449153,
      "learning_rate": 0.000158290566762043,
      "loss": 12.3636,
      "step": 7385
    },
    {
      "epoch": 1.2915409836065574,
      "grad_norm": 0.7927373729442702,
      "learning_rate": 0.0001582790598933161,
      "loss": 12.4433,
      "step": 7386
    },
    {
      "epoch": 1.2917158469945356,
      "grad_norm": 0.7236833421100132,
      "learning_rate": 0.0001582675518559151,
      "loss": 12.574,
      "step": 7387
    },
    {
      "epoch": 1.2918907103825137,
      "grad_norm": 0.7825302639754042,
      "learning_rate": 0.00015825604265007077,
      "loss": 12.793,
      "step": 7388
    },
    {
      "epoch": 1.2920655737704918,
      "grad_norm": 0.694175232286597,
      "learning_rate": 0.0001582445322760139,
      "loss": 12.5576,
      "step": 7389
    },
    {
      "epoch": 1.29224043715847,
      "grad_norm": 0.795706809219799,
      "learning_rate": 0.00015823302073397536,
      "loss": 12.5407,
      "step": 7390
    },
    {
      "epoch": 1.292415300546448,
      "grad_norm": 0.6924469841576856,
      "learning_rate": 0.00015822150802418595,
      "loss": 12.4232,
      "step": 7391
    },
    {
      "epoch": 1.2925901639344262,
      "grad_norm": 0.7488456314699502,
      "learning_rate": 0.00015820999414687656,
      "loss": 12.4491,
      "step": 7392
    },
    {
      "epoch": 1.2927650273224045,
      "grad_norm": 0.711430979509328,
      "learning_rate": 0.00015819847910227802,
      "loss": 12.5049,
      "step": 7393
    },
    {
      "epoch": 1.2929398907103824,
      "grad_norm": 0.724812515656928,
      "learning_rate": 0.0001581869628906213,
      "loss": 12.4668,
      "step": 7394
    },
    {
      "epoch": 1.2931147540983607,
      "grad_norm": 0.6982383202385761,
      "learning_rate": 0.0001581754455121373,
      "loss": 12.6962,
      "step": 7395
    },
    {
      "epoch": 1.2932896174863389,
      "grad_norm": 0.7402531781261953,
      "learning_rate": 0.00015816392696705702,
      "loss": 12.3617,
      "step": 7396
    },
    {
      "epoch": 1.293464480874317,
      "grad_norm": 0.7106135583343746,
      "learning_rate": 0.0001581524072556114,
      "loss": 12.4636,
      "step": 7397
    },
    {
      "epoch": 1.293639344262295,
      "grad_norm": 0.807319854817941,
      "learning_rate": 0.00015814088637803147,
      "loss": 12.301,
      "step": 7398
    },
    {
      "epoch": 1.2938142076502732,
      "grad_norm": 0.760472868710861,
      "learning_rate": 0.00015812936433454822,
      "loss": 12.6782,
      "step": 7399
    },
    {
      "epoch": 1.2939890710382513,
      "grad_norm": 0.8270544640000678,
      "learning_rate": 0.0001581178411253928,
      "loss": 12.698,
      "step": 7400
    },
    {
      "epoch": 1.2941639344262295,
      "grad_norm": 0.7938192454006074,
      "learning_rate": 0.00015810631675079617,
      "loss": 12.6258,
      "step": 7401
    },
    {
      "epoch": 1.2943387978142076,
      "grad_norm": 0.7489705155483388,
      "learning_rate": 0.0001580947912109895,
      "loss": 12.3795,
      "step": 7402
    },
    {
      "epoch": 1.2945136612021857,
      "grad_norm": 0.9131754879681117,
      "learning_rate": 0.0001580832645062039,
      "loss": 12.7173,
      "step": 7403
    },
    {
      "epoch": 1.294688524590164,
      "grad_norm": 0.7383820594412474,
      "learning_rate": 0.00015807173663667047,
      "loss": 12.6039,
      "step": 7404
    },
    {
      "epoch": 1.2948633879781422,
      "grad_norm": 0.8034060241530026,
      "learning_rate": 0.00015806020760262044,
      "loss": 12.599,
      "step": 7405
    },
    {
      "epoch": 1.2950382513661203,
      "grad_norm": 0.7765289564603265,
      "learning_rate": 0.000158048677404285,
      "loss": 12.4363,
      "step": 7406
    },
    {
      "epoch": 1.2952131147540984,
      "grad_norm": 0.6994422906563026,
      "learning_rate": 0.00015803714604189532,
      "loss": 12.6809,
      "step": 7407
    },
    {
      "epoch": 1.2953879781420765,
      "grad_norm": 0.697819371280455,
      "learning_rate": 0.00015802561351568265,
      "loss": 12.3486,
      "step": 7408
    },
    {
      "epoch": 1.2955628415300546,
      "grad_norm": 0.7573545729687653,
      "learning_rate": 0.0001580140798258783,
      "loss": 12.6547,
      "step": 7409
    },
    {
      "epoch": 1.2957377049180328,
      "grad_norm": 0.799890164569057,
      "learning_rate": 0.00015800254497271352,
      "loss": 12.6882,
      "step": 7410
    },
    {
      "epoch": 1.2959125683060109,
      "grad_norm": 0.7415677625594831,
      "learning_rate": 0.00015799100895641958,
      "loss": 12.4253,
      "step": 7411
    },
    {
      "epoch": 1.296087431693989,
      "grad_norm": 0.8666151942924372,
      "learning_rate": 0.00015797947177722792,
      "loss": 12.4059,
      "step": 7412
    },
    {
      "epoch": 1.2962622950819673,
      "grad_norm": 0.75055821366375,
      "learning_rate": 0.00015796793343536978,
      "loss": 12.7602,
      "step": 7413
    },
    {
      "epoch": 1.2964371584699452,
      "grad_norm": 0.751412215248254,
      "learning_rate": 0.0001579563939310766,
      "loss": 12.533,
      "step": 7414
    },
    {
      "epoch": 1.2966120218579236,
      "grad_norm": 0.6930705047187617,
      "learning_rate": 0.0001579448532645798,
      "loss": 12.3247,
      "step": 7415
    },
    {
      "epoch": 1.2967868852459017,
      "grad_norm": 0.7377737587631807,
      "learning_rate": 0.00015793331143611075,
      "loss": 12.3745,
      "step": 7416
    },
    {
      "epoch": 1.2969617486338798,
      "grad_norm": 0.7467789311357745,
      "learning_rate": 0.00015792176844590095,
      "loss": 12.665,
      "step": 7417
    },
    {
      "epoch": 1.297136612021858,
      "grad_norm": 0.782777534011278,
      "learning_rate": 0.00015791022429418182,
      "loss": 12.6665,
      "step": 7418
    },
    {
      "epoch": 1.297311475409836,
      "grad_norm": 0.7475285963737395,
      "learning_rate": 0.0001578986789811849,
      "loss": 12.5075,
      "step": 7419
    },
    {
      "epoch": 1.2974863387978142,
      "grad_norm": 0.7259633917388728,
      "learning_rate": 0.00015788713250714168,
      "loss": 12.4568,
      "step": 7420
    },
    {
      "epoch": 1.2976612021857923,
      "grad_norm": 0.7448866844845708,
      "learning_rate": 0.00015787558487228372,
      "loss": 12.5357,
      "step": 7421
    },
    {
      "epoch": 1.2978360655737706,
      "grad_norm": 0.7614045410312523,
      "learning_rate": 0.0001578640360768426,
      "loss": 12.4796,
      "step": 7422
    },
    {
      "epoch": 1.2980109289617485,
      "grad_norm": 0.7610006317176045,
      "learning_rate": 0.00015785248612104985,
      "loss": 12.5222,
      "step": 7423
    },
    {
      "epoch": 1.2981857923497269,
      "grad_norm": 0.7344467521958259,
      "learning_rate": 0.00015784093500513716,
      "loss": 12.5748,
      "step": 7424
    },
    {
      "epoch": 1.298360655737705,
      "grad_norm": 0.6787647090122007,
      "learning_rate": 0.00015782938272933613,
      "loss": 12.501,
      "step": 7425
    },
    {
      "epoch": 1.298535519125683,
      "grad_norm": 0.7625508403125029,
      "learning_rate": 0.00015781782929387842,
      "loss": 12.7117,
      "step": 7426
    },
    {
      "epoch": 1.2987103825136612,
      "grad_norm": 0.6808513958315724,
      "learning_rate": 0.0001578062746989957,
      "loss": 12.5437,
      "step": 7427
    },
    {
      "epoch": 1.2988852459016393,
      "grad_norm": 0.7923731730247541,
      "learning_rate": 0.00015779471894491966,
      "loss": 12.6698,
      "step": 7428
    },
    {
      "epoch": 1.2990601092896175,
      "grad_norm": 0.6887762594502383,
      "learning_rate": 0.00015778316203188207,
      "loss": 12.3928,
      "step": 7429
    },
    {
      "epoch": 1.2992349726775956,
      "grad_norm": 0.8551760051579186,
      "learning_rate": 0.00015777160396011466,
      "loss": 12.5893,
      "step": 7430
    },
    {
      "epoch": 1.2994098360655737,
      "grad_norm": 0.7118195641635698,
      "learning_rate": 0.00015776004472984925,
      "loss": 12.4549,
      "step": 7431
    },
    {
      "epoch": 1.2995846994535518,
      "grad_norm": 0.7930488207671504,
      "learning_rate": 0.00015774848434131755,
      "loss": 12.4892,
      "step": 7432
    },
    {
      "epoch": 1.2997595628415302,
      "grad_norm": 0.7112145326334668,
      "learning_rate": 0.00015773692279475143,
      "loss": 12.112,
      "step": 7433
    },
    {
      "epoch": 1.2999344262295083,
      "grad_norm": 0.8819280483683921,
      "learning_rate": 0.00015772536009038275,
      "loss": 12.2806,
      "step": 7434
    },
    {
      "epoch": 1.3001092896174864,
      "grad_norm": 0.6531291103677764,
      "learning_rate": 0.00015771379622844337,
      "loss": 12.4254,
      "step": 7435
    },
    {
      "epoch": 1.3002841530054645,
      "grad_norm": 0.8114295023923647,
      "learning_rate": 0.0001577022312091651,
      "loss": 12.3844,
      "step": 7436
    },
    {
      "epoch": 1.3004590163934426,
      "grad_norm": 0.7143025810521424,
      "learning_rate": 0.00015769066503277997,
      "loss": 12.3099,
      "step": 7437
    },
    {
      "epoch": 1.3006338797814208,
      "grad_norm": 0.7178801314302433,
      "learning_rate": 0.00015767909769951992,
      "loss": 12.4282,
      "step": 7438
    },
    {
      "epoch": 1.3008087431693989,
      "grad_norm": 0.7502872799835854,
      "learning_rate": 0.00015766752920961684,
      "loss": 12.0401,
      "step": 7439
    },
    {
      "epoch": 1.300983606557377,
      "grad_norm": 0.6953225523091721,
      "learning_rate": 0.0001576559595633027,
      "loss": 12.7172,
      "step": 7440
    },
    {
      "epoch": 1.3011584699453551,
      "grad_norm": 0.707352302859239,
      "learning_rate": 0.0001576443887608096,
      "loss": 12.708,
      "step": 7441
    },
    {
      "epoch": 1.3013333333333335,
      "grad_norm": 0.7293167534550906,
      "learning_rate": 0.00015763281680236942,
      "loss": 12.5996,
      "step": 7442
    },
    {
      "epoch": 1.3015081967213114,
      "grad_norm": 0.7688714694734313,
      "learning_rate": 0.00015762124368821437,
      "loss": 12.5315,
      "step": 7443
    },
    {
      "epoch": 1.3016830601092897,
      "grad_norm": 0.7360810113568886,
      "learning_rate": 0.00015760966941857647,
      "loss": 12.6952,
      "step": 7444
    },
    {
      "epoch": 1.3018579234972678,
      "grad_norm": 0.7716541274406735,
      "learning_rate": 0.0001575980939936878,
      "loss": 11.8587,
      "step": 7445
    },
    {
      "epoch": 1.302032786885246,
      "grad_norm": 0.7396825707981359,
      "learning_rate": 0.0001575865174137805,
      "loss": 12.4913,
      "step": 7446
    },
    {
      "epoch": 1.302207650273224,
      "grad_norm": 0.7841100489043075,
      "learning_rate": 0.0001575749396790867,
      "loss": 12.5742,
      "step": 7447
    },
    {
      "epoch": 1.3023825136612022,
      "grad_norm": 0.7673086104004693,
      "learning_rate": 0.00015756336078983862,
      "loss": 12.4885,
      "step": 7448
    },
    {
      "epoch": 1.3025573770491803,
      "grad_norm": 0.7870525588948015,
      "learning_rate": 0.00015755178074626836,
      "loss": 12.5815,
      "step": 7449
    },
    {
      "epoch": 1.3027322404371584,
      "grad_norm": 0.793585567011192,
      "learning_rate": 0.00015754019954860823,
      "loss": 12.5319,
      "step": 7450
    },
    {
      "epoch": 1.3029071038251367,
      "grad_norm": 0.7641537888699701,
      "learning_rate": 0.0001575286171970904,
      "loss": 12.5431,
      "step": 7451
    },
    {
      "epoch": 1.3030819672131146,
      "grad_norm": 0.8086138945350453,
      "learning_rate": 0.00015751703369194716,
      "loss": 12.1439,
      "step": 7452
    },
    {
      "epoch": 1.303256830601093,
      "grad_norm": 0.792348599369136,
      "learning_rate": 0.00015750544903341083,
      "loss": 12.4773,
      "step": 7453
    },
    {
      "epoch": 1.303431693989071,
      "grad_norm": 0.8399341365220676,
      "learning_rate": 0.00015749386322171365,
      "loss": 12.465,
      "step": 7454
    },
    {
      "epoch": 1.3036065573770492,
      "grad_norm": 0.7431468528866597,
      "learning_rate": 0.00015748227625708797,
      "loss": 12.5605,
      "step": 7455
    },
    {
      "epoch": 1.3037814207650273,
      "grad_norm": 0.767666382945982,
      "learning_rate": 0.0001574706881397662,
      "loss": 12.5123,
      "step": 7456
    },
    {
      "epoch": 1.3039562841530055,
      "grad_norm": 0.7859041971161834,
      "learning_rate": 0.00015745909886998063,
      "loss": 12.5467,
      "step": 7457
    },
    {
      "epoch": 1.3041311475409836,
      "grad_norm": 0.8211249514917254,
      "learning_rate": 0.00015744750844796373,
      "loss": 12.5859,
      "step": 7458
    },
    {
      "epoch": 1.3043060109289617,
      "grad_norm": 0.7187390269849734,
      "learning_rate": 0.0001574359168739479,
      "loss": 12.5645,
      "step": 7459
    },
    {
      "epoch": 1.3044808743169398,
      "grad_norm": 0.7425261099965974,
      "learning_rate": 0.00015742432414816556,
      "loss": 12.3169,
      "step": 7460
    },
    {
      "epoch": 1.304655737704918,
      "grad_norm": 0.7430972844353688,
      "learning_rate": 0.00015741273027084925,
      "loss": 12.4714,
      "step": 7461
    },
    {
      "epoch": 1.3048306010928963,
      "grad_norm": 0.756384375094487,
      "learning_rate": 0.00015740113524223136,
      "loss": 12.4902,
      "step": 7462
    },
    {
      "epoch": 1.3050054644808744,
      "grad_norm": 0.7321428850211102,
      "learning_rate": 0.00015738953906254451,
      "loss": 12.6591,
      "step": 7463
    },
    {
      "epoch": 1.3051803278688525,
      "grad_norm": 0.7385453300383238,
      "learning_rate": 0.0001573779417320212,
      "loss": 12.4574,
      "step": 7464
    },
    {
      "epoch": 1.3053551912568306,
      "grad_norm": 0.7597936516550692,
      "learning_rate": 0.00015736634325089397,
      "loss": 12.1889,
      "step": 7465
    },
    {
      "epoch": 1.3055300546448088,
      "grad_norm": 0.9006720784274249,
      "learning_rate": 0.0001573547436193954,
      "loss": 12.4941,
      "step": 7466
    },
    {
      "epoch": 1.3057049180327869,
      "grad_norm": 0.8214467148272786,
      "learning_rate": 0.00015734314283775812,
      "loss": 12.5782,
      "step": 7467
    },
    {
      "epoch": 1.305879781420765,
      "grad_norm": 0.8048340144980389,
      "learning_rate": 0.00015733154090621479,
      "loss": 12.4254,
      "step": 7468
    },
    {
      "epoch": 1.3060546448087431,
      "grad_norm": 0.8165570001446237,
      "learning_rate": 0.000157319937824998,
      "loss": 12.639,
      "step": 7469
    },
    {
      "epoch": 1.3062295081967212,
      "grad_norm": 0.755344586342109,
      "learning_rate": 0.00015730833359434048,
      "loss": 12.3325,
      "step": 7470
    },
    {
      "epoch": 1.3064043715846996,
      "grad_norm": 0.7740737536024247,
      "learning_rate": 0.0001572967282144749,
      "loss": 12.5466,
      "step": 7471
    },
    {
      "epoch": 1.3065792349726775,
      "grad_norm": 0.7950067900044221,
      "learning_rate": 0.00015728512168563402,
      "loss": 12.6836,
      "step": 7472
    },
    {
      "epoch": 1.3067540983606558,
      "grad_norm": 0.7796217818637123,
      "learning_rate": 0.00015727351400805052,
      "loss": 12.3656,
      "step": 7473
    },
    {
      "epoch": 1.306928961748634,
      "grad_norm": 0.8168121227340371,
      "learning_rate": 0.00015726190518195726,
      "loss": 12.6302,
      "step": 7474
    },
    {
      "epoch": 1.307103825136612,
      "grad_norm": 0.8370320255675143,
      "learning_rate": 0.00015725029520758695,
      "loss": 12.5197,
      "step": 7475
    },
    {
      "epoch": 1.3072786885245902,
      "grad_norm": 0.729178786989414,
      "learning_rate": 0.00015723868408517243,
      "loss": 12.6662,
      "step": 7476
    },
    {
      "epoch": 1.3074535519125683,
      "grad_norm": 0.8601048414444373,
      "learning_rate": 0.00015722707181494659,
      "loss": 12.5965,
      "step": 7477
    },
    {
      "epoch": 1.3076284153005464,
      "grad_norm": 0.8813343880426869,
      "learning_rate": 0.00015721545839714223,
      "loss": 12.5912,
      "step": 7478
    },
    {
      "epoch": 1.3078032786885245,
      "grad_norm": 0.8028306113619248,
      "learning_rate": 0.00015720384383199226,
      "loss": 12.5104,
      "step": 7479
    },
    {
      "epoch": 1.3079781420765029,
      "grad_norm": 0.9862092003358545,
      "learning_rate": 0.00015719222811972958,
      "loss": 12.3559,
      "step": 7480
    },
    {
      "epoch": 1.3081530054644808,
      "grad_norm": 0.8537014012205086,
      "learning_rate": 0.0001571806112605871,
      "loss": 12.4427,
      "step": 7481
    },
    {
      "epoch": 1.308327868852459,
      "grad_norm": 1.009160412549557,
      "learning_rate": 0.0001571689932547978,
      "loss": 12.6448,
      "step": 7482
    },
    {
      "epoch": 1.3085027322404372,
      "grad_norm": 0.7046452548426237,
      "learning_rate": 0.0001571573741025947,
      "loss": 12.3208,
      "step": 7483
    },
    {
      "epoch": 1.3086775956284153,
      "grad_norm": 0.7535862975959641,
      "learning_rate": 0.0001571457538042107,
      "loss": 11.9424,
      "step": 7484
    },
    {
      "epoch": 1.3088524590163935,
      "grad_norm": 0.8734035657818733,
      "learning_rate": 0.0001571341323598789,
      "loss": 12.5607,
      "step": 7485
    },
    {
      "epoch": 1.3090273224043716,
      "grad_norm": 0.6932555396255604,
      "learning_rate": 0.00015712250976983235,
      "loss": 12.5491,
      "step": 7486
    },
    {
      "epoch": 1.3092021857923497,
      "grad_norm": 0.7621241240240134,
      "learning_rate": 0.00015711088603430405,
      "loss": 12.5322,
      "step": 7487
    },
    {
      "epoch": 1.3093770491803278,
      "grad_norm": 0.7156832153337609,
      "learning_rate": 0.00015709926115352714,
      "loss": 12.4659,
      "step": 7488
    },
    {
      "epoch": 1.309551912568306,
      "grad_norm": 0.7336442404810809,
      "learning_rate": 0.0001570876351277347,
      "loss": 12.4423,
      "step": 7489
    },
    {
      "epoch": 1.309726775956284,
      "grad_norm": 0.7191403537899029,
      "learning_rate": 0.0001570760079571599,
      "loss": 12.6344,
      "step": 7490
    },
    {
      "epoch": 1.3099016393442624,
      "grad_norm": 0.6782849578269783,
      "learning_rate": 0.00015706437964203596,
      "loss": 12.0903,
      "step": 7491
    },
    {
      "epoch": 1.3100765027322403,
      "grad_norm": 0.6868664876975671,
      "learning_rate": 0.00015705275018259592,
      "loss": 12.5278,
      "step": 7492
    },
    {
      "epoch": 1.3102513661202186,
      "grad_norm": 0.7952082562480408,
      "learning_rate": 0.00015704111957907311,
      "loss": 12.5808,
      "step": 7493
    },
    {
      "epoch": 1.3104262295081968,
      "grad_norm": 0.8040403963633803,
      "learning_rate": 0.0001570294878317007,
      "loss": 11.832,
      "step": 7494
    },
    {
      "epoch": 1.3106010928961749,
      "grad_norm": 0.7442335315047957,
      "learning_rate": 0.00015701785494071194,
      "loss": 12.2479,
      "step": 7495
    },
    {
      "epoch": 1.310775956284153,
      "grad_norm": 0.900706976016305,
      "learning_rate": 0.00015700622090634013,
      "loss": 12.3811,
      "step": 7496
    },
    {
      "epoch": 1.3109508196721311,
      "grad_norm": 0.791640019054681,
      "learning_rate": 0.00015699458572881856,
      "loss": 12.6089,
      "step": 7497
    },
    {
      "epoch": 1.3111256830601092,
      "grad_norm": 0.8044841436658383,
      "learning_rate": 0.0001569829494083806,
      "loss": 12.4716,
      "step": 7498
    },
    {
      "epoch": 1.3113005464480874,
      "grad_norm": 0.744089344059799,
      "learning_rate": 0.00015697131194525947,
      "loss": 12.5418,
      "step": 7499
    },
    {
      "epoch": 1.3114754098360657,
      "grad_norm": 0.7013059009287839,
      "learning_rate": 0.0001569596733396886,
      "loss": 12.2728,
      "step": 7500
    },
    {
      "epoch": 1.3116502732240436,
      "grad_norm": 0.7452021645584098,
      "learning_rate": 0.00015694803359190147,
      "loss": 12.446,
      "step": 7501
    },
    {
      "epoch": 1.311825136612022,
      "grad_norm": 0.6913540218964753,
      "learning_rate": 0.00015693639270213136,
      "loss": 12.5845,
      "step": 7502
    },
    {
      "epoch": 1.312,
      "grad_norm": 0.7184859254119218,
      "learning_rate": 0.0001569247506706118,
      "loss": 12.6075,
      "step": 7503
    },
    {
      "epoch": 1.3121748633879782,
      "grad_norm": 0.7126055768711426,
      "learning_rate": 0.00015691310749757617,
      "loss": 12.5146,
      "step": 7504
    },
    {
      "epoch": 1.3123497267759563,
      "grad_norm": 0.7806933031626103,
      "learning_rate": 0.00015690146318325798,
      "loss": 12.343,
      "step": 7505
    },
    {
      "epoch": 1.3125245901639344,
      "grad_norm": 0.6876883159366104,
      "learning_rate": 0.00015688981772789075,
      "loss": 12.018,
      "step": 7506
    },
    {
      "epoch": 1.3126994535519125,
      "grad_norm": 0.7684565738874906,
      "learning_rate": 0.000156878171131708,
      "loss": 12.2828,
      "step": 7507
    },
    {
      "epoch": 1.3128743169398907,
      "grad_norm": 0.7491184259381872,
      "learning_rate": 0.0001568665233949433,
      "loss": 12.5304,
      "step": 7508
    },
    {
      "epoch": 1.3130491803278688,
      "grad_norm": 0.7949228502843448,
      "learning_rate": 0.00015685487451783017,
      "loss": 12.4539,
      "step": 7509
    },
    {
      "epoch": 1.313224043715847,
      "grad_norm": 0.8881938904534096,
      "learning_rate": 0.00015684322450060226,
      "loss": 12.608,
      "step": 7510
    },
    {
      "epoch": 1.3133989071038252,
      "grad_norm": 0.8232458902257412,
      "learning_rate": 0.0001568315733434932,
      "loss": 12.4269,
      "step": 7511
    },
    {
      "epoch": 1.3135737704918033,
      "grad_norm": 0.8495279887211953,
      "learning_rate": 0.00015681992104673654,
      "loss": 12.5483,
      "step": 7512
    },
    {
      "epoch": 1.3137486338797815,
      "grad_norm": 0.9039607113175563,
      "learning_rate": 0.00015680826761056604,
      "loss": 12.3919,
      "step": 7513
    },
    {
      "epoch": 1.3139234972677596,
      "grad_norm": 0.737486582296234,
      "learning_rate": 0.00015679661303521535,
      "loss": 12.6232,
      "step": 7514
    },
    {
      "epoch": 1.3140983606557377,
      "grad_norm": 0.808891776983148,
      "learning_rate": 0.00015678495732091813,
      "loss": 12.3099,
      "step": 7515
    },
    {
      "epoch": 1.3142732240437158,
      "grad_norm": 0.73685863744283,
      "learning_rate": 0.0001567733004679082,
      "loss": 12.6387,
      "step": 7516
    },
    {
      "epoch": 1.314448087431694,
      "grad_norm": 0.7713443892645835,
      "learning_rate": 0.0001567616424764193,
      "loss": 12.4166,
      "step": 7517
    },
    {
      "epoch": 1.314622950819672,
      "grad_norm": 0.8323827974707182,
      "learning_rate": 0.0001567499833466852,
      "loss": 12.6388,
      "step": 7518
    },
    {
      "epoch": 1.3147978142076502,
      "grad_norm": 0.8716895887094601,
      "learning_rate": 0.00015673832307893962,
      "loss": 12.4046,
      "step": 7519
    },
    {
      "epoch": 1.3149726775956285,
      "grad_norm": 0.9053338974597184,
      "learning_rate": 0.0001567266616734165,
      "loss": 12.3683,
      "step": 7520
    },
    {
      "epoch": 1.3151475409836064,
      "grad_norm": 0.9209786671983413,
      "learning_rate": 0.00015671499913034963,
      "loss": 12.8304,
      "step": 7521
    },
    {
      "epoch": 1.3153224043715848,
      "grad_norm": 0.7992464312275305,
      "learning_rate": 0.00015670333544997287,
      "loss": 12.5812,
      "step": 7522
    },
    {
      "epoch": 1.3154972677595629,
      "grad_norm": 0.7384109547626192,
      "learning_rate": 0.00015669167063252015,
      "loss": 12.5914,
      "step": 7523
    },
    {
      "epoch": 1.315672131147541,
      "grad_norm": 0.7259562620287761,
      "learning_rate": 0.00015668000467822536,
      "loss": 12.3606,
      "step": 7524
    },
    {
      "epoch": 1.3158469945355191,
      "grad_norm": 0.7309116149244171,
      "learning_rate": 0.00015666833758732246,
      "loss": 12.4821,
      "step": 7525
    },
    {
      "epoch": 1.3160218579234972,
      "grad_norm": 0.7649253738063326,
      "learning_rate": 0.0001566566693600454,
      "loss": 12.489,
      "step": 7526
    },
    {
      "epoch": 1.3161967213114754,
      "grad_norm": 0.746055207282059,
      "learning_rate": 0.00015664499999662815,
      "loss": 12.5982,
      "step": 7527
    },
    {
      "epoch": 1.3163715846994535,
      "grad_norm": 0.9019385667824366,
      "learning_rate": 0.00015663332949730473,
      "loss": 12.4314,
      "step": 7528
    },
    {
      "epoch": 1.3165464480874318,
      "grad_norm": 0.7427930612761048,
      "learning_rate": 0.00015662165786230916,
      "loss": 12.2486,
      "step": 7529
    },
    {
      "epoch": 1.3167213114754097,
      "grad_norm": 0.688037914654294,
      "learning_rate": 0.0001566099850918755,
      "loss": 12.4597,
      "step": 7530
    },
    {
      "epoch": 1.316896174863388,
      "grad_norm": 0.7092515415398154,
      "learning_rate": 0.0001565983111862378,
      "loss": 12.6527,
      "step": 7531
    },
    {
      "epoch": 1.3170710382513662,
      "grad_norm": 0.7196187293031846,
      "learning_rate": 0.00015658663614563024,
      "loss": 12.4765,
      "step": 7532
    },
    {
      "epoch": 1.3172459016393443,
      "grad_norm": 0.6870288585126252,
      "learning_rate": 0.0001565749599702868,
      "loss": 12.6582,
      "step": 7533
    },
    {
      "epoch": 1.3174207650273224,
      "grad_norm": 0.7720591017731988,
      "learning_rate": 0.0001565632826604418,
      "loss": 12.6746,
      "step": 7534
    },
    {
      "epoch": 1.3175956284153005,
      "grad_norm": 0.6508119356563553,
      "learning_rate": 0.00015655160421632926,
      "loss": 12.5056,
      "step": 7535
    },
    {
      "epoch": 1.3177704918032787,
      "grad_norm": 0.7047463691846375,
      "learning_rate": 0.0001565399246381834,
      "loss": 12.6777,
      "step": 7536
    },
    {
      "epoch": 1.3179453551912568,
      "grad_norm": 0.7337254424939238,
      "learning_rate": 0.00015652824392623852,
      "loss": 12.5226,
      "step": 7537
    },
    {
      "epoch": 1.318120218579235,
      "grad_norm": 0.6828691629408178,
      "learning_rate": 0.00015651656208072869,
      "loss": 12.4427,
      "step": 7538
    },
    {
      "epoch": 1.318295081967213,
      "grad_norm": 0.7185928400526881,
      "learning_rate": 0.0001565048791018883,
      "loss": 12.7042,
      "step": 7539
    },
    {
      "epoch": 1.3184699453551914,
      "grad_norm": 0.7809849715933789,
      "learning_rate": 0.00015649319498995163,
      "loss": 12.6957,
      "step": 7540
    },
    {
      "epoch": 1.3186448087431695,
      "grad_norm": 0.6407759769068655,
      "learning_rate": 0.0001564815097451529,
      "loss": 12.5282,
      "step": 7541
    },
    {
      "epoch": 1.3188196721311476,
      "grad_norm": 0.7952616190435214,
      "learning_rate": 0.00015646982336772652,
      "loss": 12.6309,
      "step": 7542
    },
    {
      "epoch": 1.3189945355191257,
      "grad_norm": 0.7527531177455805,
      "learning_rate": 0.00015645813585790674,
      "loss": 12.1736,
      "step": 7543
    },
    {
      "epoch": 1.3191693989071038,
      "grad_norm": 0.7482955067889946,
      "learning_rate": 0.00015644644721592803,
      "loss": 12.4051,
      "step": 7544
    },
    {
      "epoch": 1.319344262295082,
      "grad_norm": 0.6605864579194104,
      "learning_rate": 0.0001564347574420247,
      "loss": 12.1498,
      "step": 7545
    },
    {
      "epoch": 1.31951912568306,
      "grad_norm": 0.8547226359920497,
      "learning_rate": 0.00015642306653643125,
      "loss": 12.4759,
      "step": 7546
    },
    {
      "epoch": 1.3196939890710382,
      "grad_norm": 0.7207243473129435,
      "learning_rate": 0.00015641137449938207,
      "loss": 12.5141,
      "step": 7547
    },
    {
      "epoch": 1.3198688524590163,
      "grad_norm": 0.8889855547552467,
      "learning_rate": 0.00015639968133111158,
      "loss": 12.599,
      "step": 7548
    },
    {
      "epoch": 1.3200437158469946,
      "grad_norm": 0.7828872775032115,
      "learning_rate": 0.00015638798703185435,
      "loss": 12.5953,
      "step": 7549
    },
    {
      "epoch": 1.3202185792349725,
      "grad_norm": 0.7589547521705737,
      "learning_rate": 0.00015637629160184484,
      "loss": 12.5979,
      "step": 7550
    },
    {
      "epoch": 1.3203934426229509,
      "grad_norm": 0.8026579011531026,
      "learning_rate": 0.0001563645950413176,
      "loss": 12.5856,
      "step": 7551
    },
    {
      "epoch": 1.320568306010929,
      "grad_norm": 0.7268265498325592,
      "learning_rate": 0.00015635289735050714,
      "loss": 12.3349,
      "step": 7552
    },
    {
      "epoch": 1.3207431693989071,
      "grad_norm": 0.7559951488763204,
      "learning_rate": 0.00015634119852964807,
      "loss": 12.5298,
      "step": 7553
    },
    {
      "epoch": 1.3209180327868852,
      "grad_norm": 0.7108358485155135,
      "learning_rate": 0.00015632949857897498,
      "loss": 12.2731,
      "step": 7554
    },
    {
      "epoch": 1.3210928961748634,
      "grad_norm": 0.7983557533457235,
      "learning_rate": 0.0001563177974987225,
      "loss": 12.4822,
      "step": 7555
    },
    {
      "epoch": 1.3212677595628415,
      "grad_norm": 0.6969962754075758,
      "learning_rate": 0.00015630609528912524,
      "loss": 12.4736,
      "step": 7556
    },
    {
      "epoch": 1.3214426229508196,
      "grad_norm": 0.8705550164759562,
      "learning_rate": 0.0001562943919504179,
      "loss": 12.4334,
      "step": 7557
    },
    {
      "epoch": 1.321617486338798,
      "grad_norm": 0.7631093130630485,
      "learning_rate": 0.00015628268748283516,
      "loss": 12.6297,
      "step": 7558
    },
    {
      "epoch": 1.3217923497267758,
      "grad_norm": 0.995571607539056,
      "learning_rate": 0.00015627098188661172,
      "loss": 12.682,
      "step": 7559
    },
    {
      "epoch": 1.3219672131147542,
      "grad_norm": 0.8423088706787234,
      "learning_rate": 0.00015625927516198232,
      "loss": 12.1054,
      "step": 7560
    },
    {
      "epoch": 1.3221420765027323,
      "grad_norm": 0.8672604509817747,
      "learning_rate": 0.0001562475673091817,
      "loss": 12.3158,
      "step": 7561
    },
    {
      "epoch": 1.3223169398907104,
      "grad_norm": 0.838824249732686,
      "learning_rate": 0.00015623585832844468,
      "loss": 12.7195,
      "step": 7562
    },
    {
      "epoch": 1.3224918032786885,
      "grad_norm": 0.9164209539976841,
      "learning_rate": 0.000156224148220006,
      "loss": 12.372,
      "step": 7563
    },
    {
      "epoch": 1.3226666666666667,
      "grad_norm": 0.78019483514585,
      "learning_rate": 0.00015621243698410057,
      "loss": 12.699,
      "step": 7564
    },
    {
      "epoch": 1.3228415300546448,
      "grad_norm": 1.098529083910837,
      "learning_rate": 0.00015620072462096314,
      "loss": 12.4475,
      "step": 7565
    },
    {
      "epoch": 1.323016393442623,
      "grad_norm": 0.6867854462082807,
      "learning_rate": 0.00015618901113082862,
      "loss": 12.5052,
      "step": 7566
    },
    {
      "epoch": 1.323191256830601,
      "grad_norm": 1.2155956616728714,
      "learning_rate": 0.00015617729651393192,
      "loss": 12.6223,
      "step": 7567
    },
    {
      "epoch": 1.3233661202185791,
      "grad_norm": 0.6685471946301585,
      "learning_rate": 0.00015616558077050793,
      "loss": 12.1469,
      "step": 7568
    },
    {
      "epoch": 1.3235409836065575,
      "grad_norm": 0.941442312337805,
      "learning_rate": 0.0001561538639007916,
      "loss": 12.5664,
      "step": 7569
    },
    {
      "epoch": 1.3237158469945356,
      "grad_norm": 0.6749095518191613,
      "learning_rate": 0.0001561421459050179,
      "loss": 12.2431,
      "step": 7570
    },
    {
      "epoch": 1.3238907103825137,
      "grad_norm": 0.895753681459132,
      "learning_rate": 0.0001561304267834218,
      "loss": 12.4193,
      "step": 7571
    },
    {
      "epoch": 1.3240655737704918,
      "grad_norm": 0.8628434809510137,
      "learning_rate": 0.00015611870653623825,
      "loss": 12.538,
      "step": 7572
    },
    {
      "epoch": 1.32424043715847,
      "grad_norm": 0.8161005274802616,
      "learning_rate": 0.00015610698516370234,
      "loss": 12.389,
      "step": 7573
    },
    {
      "epoch": 1.324415300546448,
      "grad_norm": 0.8403375112699867,
      "learning_rate": 0.00015609526266604912,
      "loss": 12.4697,
      "step": 7574
    },
    {
      "epoch": 1.3245901639344262,
      "grad_norm": 0.7908903361008618,
      "learning_rate": 0.00015608353904351366,
      "loss": 12.4877,
      "step": 7575
    },
    {
      "epoch": 1.3247650273224043,
      "grad_norm": 0.879547719135353,
      "learning_rate": 0.00015607181429633105,
      "loss": 12.5652,
      "step": 7576
    },
    {
      "epoch": 1.3249398907103824,
      "grad_norm": 0.8094571239845578,
      "learning_rate": 0.00015606008842473638,
      "loss": 12.5359,
      "step": 7577
    },
    {
      "epoch": 1.3251147540983608,
      "grad_norm": 0.9588362573236666,
      "learning_rate": 0.00015604836142896482,
      "loss": 12.4008,
      "step": 7578
    },
    {
      "epoch": 1.3252896174863387,
      "grad_norm": 0.8683731727304879,
      "learning_rate": 0.00015603663330925153,
      "loss": 12.6454,
      "step": 7579
    },
    {
      "epoch": 1.325464480874317,
      "grad_norm": 0.865803600307534,
      "learning_rate": 0.00015602490406583166,
      "loss": 12.7619,
      "step": 7580
    },
    {
      "epoch": 1.3256393442622951,
      "grad_norm": 0.7137603956226555,
      "learning_rate": 0.00015601317369894044,
      "loss": 12.5718,
      "step": 7581
    },
    {
      "epoch": 1.3258142076502732,
      "grad_norm": 0.9386270633976143,
      "learning_rate": 0.00015600144220881312,
      "loss": 12.651,
      "step": 7582
    },
    {
      "epoch": 1.3259890710382514,
      "grad_norm": 0.7714749954494409,
      "learning_rate": 0.00015598970959568496,
      "loss": 12.5723,
      "step": 7583
    },
    {
      "epoch": 1.3261639344262295,
      "grad_norm": 0.892976623555502,
      "learning_rate": 0.00015597797585979118,
      "loss": 12.5681,
      "step": 7584
    },
    {
      "epoch": 1.3263387978142076,
      "grad_norm": 0.7290567295151692,
      "learning_rate": 0.00015596624100136711,
      "loss": 12.2854,
      "step": 7585
    },
    {
      "epoch": 1.3265136612021857,
      "grad_norm": 0.8286602511608799,
      "learning_rate": 0.00015595450502064808,
      "loss": 12.5285,
      "step": 7586
    },
    {
      "epoch": 1.326688524590164,
      "grad_norm": 0.7167596851366635,
      "learning_rate": 0.0001559427679178694,
      "loss": 12.6201,
      "step": 7587
    },
    {
      "epoch": 1.326863387978142,
      "grad_norm": 0.8397580294708942,
      "learning_rate": 0.0001559310296932665,
      "loss": 12.4065,
      "step": 7588
    },
    {
      "epoch": 1.3270382513661203,
      "grad_norm": 0.6852704949952302,
      "learning_rate": 0.0001559192903470747,
      "loss": 12.5719,
      "step": 7589
    },
    {
      "epoch": 1.3272131147540984,
      "grad_norm": 0.804876499038312,
      "learning_rate": 0.00015590754987952944,
      "loss": 12.7783,
      "step": 7590
    },
    {
      "epoch": 1.3273879781420765,
      "grad_norm": 0.7027456483056498,
      "learning_rate": 0.00015589580829086612,
      "loss": 12.2241,
      "step": 7591
    },
    {
      "epoch": 1.3275628415300547,
      "grad_norm": 0.6944481306777851,
      "learning_rate": 0.00015588406558132025,
      "loss": 12.5081,
      "step": 7592
    },
    {
      "epoch": 1.3277377049180328,
      "grad_norm": 0.9042625786385372,
      "learning_rate": 0.00015587232175112728,
      "loss": 12.6544,
      "step": 7593
    },
    {
      "epoch": 1.327912568306011,
      "grad_norm": 0.7340093564483537,
      "learning_rate": 0.00015586057680052267,
      "loss": 12.4254,
      "step": 7594
    },
    {
      "epoch": 1.328087431693989,
      "grad_norm": 0.8565653825914021,
      "learning_rate": 0.000155848830729742,
      "loss": 12.5118,
      "step": 7595
    },
    {
      "epoch": 1.3282622950819671,
      "grad_norm": 0.676059138286586,
      "learning_rate": 0.0001558370835390208,
      "loss": 12.2005,
      "step": 7596
    },
    {
      "epoch": 1.3284371584699453,
      "grad_norm": 0.8522757819989022,
      "learning_rate": 0.00015582533522859467,
      "loss": 12.4769,
      "step": 7597
    },
    {
      "epoch": 1.3286120218579236,
      "grad_norm": 0.7467344007900784,
      "learning_rate": 0.0001558135857986991,
      "loss": 12.5472,
      "step": 7598
    },
    {
      "epoch": 1.3287868852459017,
      "grad_norm": 0.7876912339851961,
      "learning_rate": 0.00015580183524956982,
      "loss": 12.6063,
      "step": 7599
    },
    {
      "epoch": 1.3289617486338798,
      "grad_norm": 0.6686601434042401,
      "learning_rate": 0.0001557900835814424,
      "loss": 12.5539,
      "step": 7600
    },
    {
      "epoch": 1.329136612021858,
      "grad_norm": 0.9211390008722802,
      "learning_rate": 0.00015577833079455247,
      "loss": 12.5556,
      "step": 7601
    },
    {
      "epoch": 1.329311475409836,
      "grad_norm": 0.8283428385748306,
      "learning_rate": 0.00015576657688913577,
      "loss": 12.6898,
      "step": 7602
    },
    {
      "epoch": 1.3294863387978142,
      "grad_norm": 0.7253403930889194,
      "learning_rate": 0.00015575482186542795,
      "loss": 12.2731,
      "step": 7603
    },
    {
      "epoch": 1.3296612021857923,
      "grad_norm": 0.8052272964378282,
      "learning_rate": 0.00015574306572366478,
      "loss": 12.4754,
      "step": 7604
    },
    {
      "epoch": 1.3298360655737704,
      "grad_norm": 0.8475917760159256,
      "learning_rate": 0.00015573130846408199,
      "loss": 12.4353,
      "step": 7605
    },
    {
      "epoch": 1.3300109289617486,
      "grad_norm": 0.6988935626939157,
      "learning_rate": 0.00015571955008691535,
      "loss": 12.4494,
      "step": 7606
    },
    {
      "epoch": 1.330185792349727,
      "grad_norm": 0.8353763669405591,
      "learning_rate": 0.00015570779059240065,
      "loss": 12.5697,
      "step": 7607
    },
    {
      "epoch": 1.3303606557377048,
      "grad_norm": 0.731306631840937,
      "learning_rate": 0.0001556960299807737,
      "loss": 12.6605,
      "step": 7608
    },
    {
      "epoch": 1.3305355191256831,
      "grad_norm": 0.7395452032559311,
      "learning_rate": 0.00015568426825227037,
      "loss": 12.6518,
      "step": 7609
    },
    {
      "epoch": 1.3307103825136612,
      "grad_norm": 0.7934970911706157,
      "learning_rate": 0.00015567250540712645,
      "loss": 12.4848,
      "step": 7610
    },
    {
      "epoch": 1.3308852459016394,
      "grad_norm": 0.7991696538375842,
      "learning_rate": 0.00015566074144557788,
      "loss": 12.1943,
      "step": 7611
    },
    {
      "epoch": 1.3310601092896175,
      "grad_norm": 0.7402881852640091,
      "learning_rate": 0.00015564897636786054,
      "loss": 12.5533,
      "step": 7612
    },
    {
      "epoch": 1.3312349726775956,
      "grad_norm": 0.8262758315655772,
      "learning_rate": 0.00015563721017421034,
      "loss": 12.6063,
      "step": 7613
    },
    {
      "epoch": 1.3314098360655737,
      "grad_norm": 0.7075538048152428,
      "learning_rate": 0.00015562544286486327,
      "loss": 12.6366,
      "step": 7614
    },
    {
      "epoch": 1.3315846994535518,
      "grad_norm": 0.7103837307265395,
      "learning_rate": 0.00015561367444005524,
      "loss": 12.3726,
      "step": 7615
    },
    {
      "epoch": 1.3317595628415302,
      "grad_norm": 0.7168834599738758,
      "learning_rate": 0.00015560190490002232,
      "loss": 12.7035,
      "step": 7616
    },
    {
      "epoch": 1.331934426229508,
      "grad_norm": 0.7720059578882789,
      "learning_rate": 0.00015559013424500047,
      "loss": 12.2757,
      "step": 7617
    },
    {
      "epoch": 1.3321092896174864,
      "grad_norm": 0.7301032646243937,
      "learning_rate": 0.00015557836247522575,
      "loss": 12.6512,
      "step": 7618
    },
    {
      "epoch": 1.3322841530054645,
      "grad_norm": 0.6870508369986342,
      "learning_rate": 0.00015556658959093423,
      "loss": 12.5292,
      "step": 7619
    },
    {
      "epoch": 1.3324590163934427,
      "grad_norm": 0.7939419124021644,
      "learning_rate": 0.00015555481559236193,
      "loss": 12.5664,
      "step": 7620
    },
    {
      "epoch": 1.3326338797814208,
      "grad_norm": 0.753485257830007,
      "learning_rate": 0.00015554304047974506,
      "loss": 12.4859,
      "step": 7621
    },
    {
      "epoch": 1.332808743169399,
      "grad_norm": 0.7516965125309483,
      "learning_rate": 0.00015553126425331962,
      "loss": 12.2092,
      "step": 7622
    },
    {
      "epoch": 1.332983606557377,
      "grad_norm": 0.8078964352040386,
      "learning_rate": 0.0001555194869133219,
      "loss": 12.5466,
      "step": 7623
    },
    {
      "epoch": 1.3331584699453551,
      "grad_norm": 0.7713566198868026,
      "learning_rate": 0.00015550770845998798,
      "loss": 12.466,
      "step": 7624
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.7263240767155166,
      "learning_rate": 0.00015549592889355408,
      "loss": 12.5169,
      "step": 7625
    },
    {
      "epoch": 1.3335081967213114,
      "grad_norm": 0.8369590103603229,
      "learning_rate": 0.00015548414821425638,
      "loss": 12.5512,
      "step": 7626
    },
    {
      "epoch": 1.3336830601092897,
      "grad_norm": 0.7662427793923433,
      "learning_rate": 0.00015547236642233115,
      "loss": 12.5893,
      "step": 7627
    },
    {
      "epoch": 1.3338579234972678,
      "grad_norm": 0.7814023419036072,
      "learning_rate": 0.00015546058351801465,
      "loss": 12.7222,
      "step": 7628
    },
    {
      "epoch": 1.334032786885246,
      "grad_norm": 0.7580495966165756,
      "learning_rate": 0.00015544879950154318,
      "loss": 12.443,
      "step": 7629
    },
    {
      "epoch": 1.334207650273224,
      "grad_norm": 0.8069346647190928,
      "learning_rate": 0.00015543701437315303,
      "loss": 12.4393,
      "step": 7630
    },
    {
      "epoch": 1.3343825136612022,
      "grad_norm": 0.7648231816180342,
      "learning_rate": 0.00015542522813308054,
      "loss": 12.5307,
      "step": 7631
    },
    {
      "epoch": 1.3345573770491803,
      "grad_norm": 0.7233344604326319,
      "learning_rate": 0.00015541344078156205,
      "loss": 12.3611,
      "step": 7632
    },
    {
      "epoch": 1.3347322404371584,
      "grad_norm": 0.7526650976529554,
      "learning_rate": 0.0001554016523188339,
      "loss": 12.5372,
      "step": 7633
    },
    {
      "epoch": 1.3349071038251366,
      "grad_norm": 0.7705663888118791,
      "learning_rate": 0.00015538986274513252,
      "loss": 12.2558,
      "step": 7634
    },
    {
      "epoch": 1.3350819672131147,
      "grad_norm": 1.0015985178828124,
      "learning_rate": 0.00015537807206069434,
      "loss": 12.6932,
      "step": 7635
    },
    {
      "epoch": 1.335256830601093,
      "grad_norm": 0.7288989542867258,
      "learning_rate": 0.0001553662802657558,
      "loss": 12.497,
      "step": 7636
    },
    {
      "epoch": 1.335431693989071,
      "grad_norm": 0.9556916174768328,
      "learning_rate": 0.0001553544873605533,
      "loss": 12.5913,
      "step": 7637
    },
    {
      "epoch": 1.3356065573770493,
      "grad_norm": 0.6676264244063198,
      "learning_rate": 0.0001553426933453234,
      "loss": 12.3631,
      "step": 7638
    },
    {
      "epoch": 1.3357814207650274,
      "grad_norm": 0.7553321688596156,
      "learning_rate": 0.00015533089822030254,
      "loss": 12.4256,
      "step": 7639
    },
    {
      "epoch": 1.3359562841530055,
      "grad_norm": 0.7684031398349177,
      "learning_rate": 0.00015531910198572733,
      "loss": 12.4292,
      "step": 7640
    },
    {
      "epoch": 1.3361311475409836,
      "grad_norm": 0.7955784377681939,
      "learning_rate": 0.0001553073046418342,
      "loss": 12.7062,
      "step": 7641
    },
    {
      "epoch": 1.3363060109289617,
      "grad_norm": 0.7238560471459348,
      "learning_rate": 0.00015529550618885985,
      "loss": 12.5698,
      "step": 7642
    },
    {
      "epoch": 1.3364808743169398,
      "grad_norm": 0.813909763450226,
      "learning_rate": 0.00015528370662704083,
      "loss": 12.6662,
      "step": 7643
    },
    {
      "epoch": 1.336655737704918,
      "grad_norm": 0.7017178964458235,
      "learning_rate": 0.00015527190595661375,
      "loss": 12.5572,
      "step": 7644
    },
    {
      "epoch": 1.3368306010928963,
      "grad_norm": 0.8125842585211546,
      "learning_rate": 0.00015526010417781522,
      "loss": 12.5889,
      "step": 7645
    },
    {
      "epoch": 1.3370054644808742,
      "grad_norm": 0.8100088949894918,
      "learning_rate": 0.00015524830129088195,
      "loss": 12.4485,
      "step": 7646
    },
    {
      "epoch": 1.3371803278688525,
      "grad_norm": 0.7900034683873798,
      "learning_rate": 0.0001552364972960506,
      "loss": 12.288,
      "step": 7647
    },
    {
      "epoch": 1.3373551912568307,
      "grad_norm": 0.6904247069287175,
      "learning_rate": 0.00015522469219355787,
      "loss": 12.4218,
      "step": 7648
    },
    {
      "epoch": 1.3375300546448088,
      "grad_norm": 0.8293757998630535,
      "learning_rate": 0.0001552128859836405,
      "loss": 12.5027,
      "step": 7649
    },
    {
      "epoch": 1.337704918032787,
      "grad_norm": 0.6993359114026737,
      "learning_rate": 0.00015520107866653525,
      "loss": 12.0899,
      "step": 7650
    },
    {
      "epoch": 1.337879781420765,
      "grad_norm": 0.8243652132264195,
      "learning_rate": 0.0001551892702424789,
      "loss": 12.2504,
      "step": 7651
    },
    {
      "epoch": 1.3380546448087431,
      "grad_norm": 0.7017840412129098,
      "learning_rate": 0.0001551774607117082,
      "loss": 12.5585,
      "step": 7652
    },
    {
      "epoch": 1.3382295081967213,
      "grad_norm": 0.8331143601490688,
      "learning_rate": 0.00015516565007446,
      "loss": 12.5717,
      "step": 7653
    },
    {
      "epoch": 1.3384043715846994,
      "grad_norm": 0.6944101326526508,
      "learning_rate": 0.00015515383833097115,
      "loss": 12.3942,
      "step": 7654
    },
    {
      "epoch": 1.3385792349726775,
      "grad_norm": 0.8054274698754381,
      "learning_rate": 0.00015514202548147846,
      "loss": 12.5625,
      "step": 7655
    },
    {
      "epoch": 1.3387540983606558,
      "grad_norm": 0.7178683550581284,
      "learning_rate": 0.0001551302115262189,
      "loss": 12.4051,
      "step": 7656
    },
    {
      "epoch": 1.338928961748634,
      "grad_norm": 0.8463752902255072,
      "learning_rate": 0.0001551183964654293,
      "loss": 12.5484,
      "step": 7657
    },
    {
      "epoch": 1.339103825136612,
      "grad_norm": 0.7780231302706018,
      "learning_rate": 0.00015510658029934662,
      "loss": 12.5252,
      "step": 7658
    },
    {
      "epoch": 1.3392786885245902,
      "grad_norm": 0.6763274496987192,
      "learning_rate": 0.0001550947630282078,
      "loss": 12.287,
      "step": 7659
    },
    {
      "epoch": 1.3394535519125683,
      "grad_norm": 0.7474028301235691,
      "learning_rate": 0.00015508294465224983,
      "loss": 12.4688,
      "step": 7660
    },
    {
      "epoch": 1.3396284153005464,
      "grad_norm": 0.6830128826104909,
      "learning_rate": 0.00015507112517170967,
      "loss": 12.1557,
      "step": 7661
    },
    {
      "epoch": 1.3398032786885246,
      "grad_norm": 0.7878577935038898,
      "learning_rate": 0.0001550593045868244,
      "loss": 12.3966,
      "step": 7662
    },
    {
      "epoch": 1.3399781420765027,
      "grad_norm": 0.7591961347529289,
      "learning_rate": 0.000155047482897831,
      "loss": 12.4661,
      "step": 7663
    },
    {
      "epoch": 1.3401530054644808,
      "grad_norm": 0.6843375857598195,
      "learning_rate": 0.00015503566010496655,
      "loss": 12.3691,
      "step": 7664
    },
    {
      "epoch": 1.3403278688524591,
      "grad_norm": 0.7707450528611476,
      "learning_rate": 0.00015502383620846812,
      "loss": 12.6962,
      "step": 7665
    },
    {
      "epoch": 1.340502732240437,
      "grad_norm": 0.733676096590673,
      "learning_rate": 0.00015501201120857283,
      "loss": 12.4191,
      "step": 7666
    },
    {
      "epoch": 1.3406775956284154,
      "grad_norm": 0.7669580568322668,
      "learning_rate": 0.00015500018510551783,
      "loss": 12.5355,
      "step": 7667
    },
    {
      "epoch": 1.3408524590163935,
      "grad_norm": 0.6614938744629651,
      "learning_rate": 0.00015498835789954022,
      "loss": 12.229,
      "step": 7668
    },
    {
      "epoch": 1.3410273224043716,
      "grad_norm": 0.8695370463196314,
      "learning_rate": 0.0001549765295908772,
      "loss": 12.6221,
      "step": 7669
    },
    {
      "epoch": 1.3412021857923497,
      "grad_norm": 0.7579154091086364,
      "learning_rate": 0.00015496470017976598,
      "loss": 12.4681,
      "step": 7670
    },
    {
      "epoch": 1.3413770491803279,
      "grad_norm": 0.7808205873475662,
      "learning_rate": 0.00015495286966644373,
      "loss": 12.5899,
      "step": 7671
    },
    {
      "epoch": 1.341551912568306,
      "grad_norm": 0.8050005219544824,
      "learning_rate": 0.00015494103805114776,
      "loss": 12.664,
      "step": 7672
    },
    {
      "epoch": 1.341726775956284,
      "grad_norm": 0.7243330942428632,
      "learning_rate": 0.00015492920533411524,
      "loss": 12.2927,
      "step": 7673
    },
    {
      "epoch": 1.3419016393442624,
      "grad_norm": 0.7378168388173264,
      "learning_rate": 0.0001549173715155835,
      "loss": 12.4296,
      "step": 7674
    },
    {
      "epoch": 1.3420765027322403,
      "grad_norm": 0.8748253376765376,
      "learning_rate": 0.00015490553659578988,
      "loss": 12.6497,
      "step": 7675
    },
    {
      "epoch": 1.3422513661202187,
      "grad_norm": 0.7502271983233809,
      "learning_rate": 0.00015489370057497165,
      "loss": 12.4867,
      "step": 7676
    },
    {
      "epoch": 1.3424262295081968,
      "grad_norm": 0.7469657241052583,
      "learning_rate": 0.00015488186345336615,
      "loss": 12.531,
      "step": 7677
    },
    {
      "epoch": 1.342601092896175,
      "grad_norm": 0.7261153081571415,
      "learning_rate": 0.00015487002523121082,
      "loss": 12.5573,
      "step": 7678
    },
    {
      "epoch": 1.342775956284153,
      "grad_norm": 0.7224321720705722,
      "learning_rate": 0.000154858185908743,
      "loss": 12.4262,
      "step": 7679
    },
    {
      "epoch": 1.3429508196721311,
      "grad_norm": 0.8295365799572476,
      "learning_rate": 0.0001548463454862001,
      "loss": 12.4742,
      "step": 7680
    },
    {
      "epoch": 1.3431256830601093,
      "grad_norm": 0.7035252358635933,
      "learning_rate": 0.00015483450396381958,
      "loss": 12.4688,
      "step": 7681
    },
    {
      "epoch": 1.3433005464480874,
      "grad_norm": 0.6870676624121099,
      "learning_rate": 0.0001548226613418389,
      "loss": 12.4984,
      "step": 7682
    },
    {
      "epoch": 1.3434754098360655,
      "grad_norm": 0.7789026563549957,
      "learning_rate": 0.00015481081762049552,
      "loss": 12.6274,
      "step": 7683
    },
    {
      "epoch": 1.3436502732240436,
      "grad_norm": 0.7415353665560214,
      "learning_rate": 0.00015479897280002697,
      "loss": 12.4912,
      "step": 7684
    },
    {
      "epoch": 1.343825136612022,
      "grad_norm": 0.6904111991363866,
      "learning_rate": 0.00015478712688067075,
      "loss": 12.3779,
      "step": 7685
    },
    {
      "epoch": 1.3439999999999999,
      "grad_norm": 0.7830156636348197,
      "learning_rate": 0.00015477527986266444,
      "loss": 12.3158,
      "step": 7686
    },
    {
      "epoch": 1.3441748633879782,
      "grad_norm": 0.7090260000801651,
      "learning_rate": 0.00015476343174624553,
      "loss": 12.2671,
      "step": 7687
    },
    {
      "epoch": 1.3443497267759563,
      "grad_norm": 0.6598234087770672,
      "learning_rate": 0.00015475158253165172,
      "loss": 12.2787,
      "step": 7688
    },
    {
      "epoch": 1.3445245901639344,
      "grad_norm": 0.6771047351130807,
      "learning_rate": 0.00015473973221912055,
      "loss": 12.4177,
      "step": 7689
    },
    {
      "epoch": 1.3446994535519126,
      "grad_norm": 0.7563486440436876,
      "learning_rate": 0.00015472788080888968,
      "loss": 12.3039,
      "step": 7690
    },
    {
      "epoch": 1.3448743169398907,
      "grad_norm": 0.6984029444442509,
      "learning_rate": 0.00015471602830119675,
      "loss": 12.2753,
      "step": 7691
    },
    {
      "epoch": 1.3450491803278688,
      "grad_norm": 0.7515919444136429,
      "learning_rate": 0.0001547041746962795,
      "loss": 12.4028,
      "step": 7692
    },
    {
      "epoch": 1.345224043715847,
      "grad_norm": 0.6993992476914326,
      "learning_rate": 0.00015469231999437554,
      "loss": 12.3403,
      "step": 7693
    },
    {
      "epoch": 1.3453989071038253,
      "grad_norm": 0.7500544666898052,
      "learning_rate": 0.00015468046419572263,
      "loss": 12.7287,
      "step": 7694
    },
    {
      "epoch": 1.3455737704918032,
      "grad_norm": 0.7463223000490754,
      "learning_rate": 0.00015466860730055853,
      "loss": 12.6094,
      "step": 7695
    },
    {
      "epoch": 1.3457486338797815,
      "grad_norm": 0.7082110266651189,
      "learning_rate": 0.00015465674930912105,
      "loss": 12.5323,
      "step": 7696
    },
    {
      "epoch": 1.3459234972677596,
      "grad_norm": 0.7602980071969875,
      "learning_rate": 0.00015464489022164787,
      "loss": 12.4958,
      "step": 7697
    },
    {
      "epoch": 1.3460983606557377,
      "grad_norm": 0.7364471385694363,
      "learning_rate": 0.0001546330300383769,
      "loss": 12.6524,
      "step": 7698
    },
    {
      "epoch": 1.3462732240437159,
      "grad_norm": 0.7494669516239626,
      "learning_rate": 0.00015462116875954588,
      "loss": 12.4863,
      "step": 7699
    },
    {
      "epoch": 1.346448087431694,
      "grad_norm": 0.7448273262605641,
      "learning_rate": 0.00015460930638539275,
      "loss": 12.5339,
      "step": 7700
    },
    {
      "epoch": 1.346622950819672,
      "grad_norm": 0.6779415454074084,
      "learning_rate": 0.00015459744291615533,
      "loss": 12.5107,
      "step": 7701
    },
    {
      "epoch": 1.3467978142076502,
      "grad_norm": 0.7738607040356149,
      "learning_rate": 0.00015458557835207158,
      "loss": 12.53,
      "step": 7702
    },
    {
      "epoch": 1.3469726775956286,
      "grad_norm": 0.7935239993820897,
      "learning_rate": 0.00015457371269337938,
      "loss": 12.6454,
      "step": 7703
    },
    {
      "epoch": 1.3471475409836065,
      "grad_norm": 0.767368977957214,
      "learning_rate": 0.00015456184594031663,
      "loss": 12.6744,
      "step": 7704
    },
    {
      "epoch": 1.3473224043715848,
      "grad_norm": 0.6932514294094039,
      "learning_rate": 0.0001545499780931214,
      "loss": 12.4724,
      "step": 7705
    },
    {
      "epoch": 1.347497267759563,
      "grad_norm": 0.795194473853508,
      "learning_rate": 0.0001545381091520316,
      "loss": 12.6135,
      "step": 7706
    },
    {
      "epoch": 1.347672131147541,
      "grad_norm": 0.825553745757009,
      "learning_rate": 0.00015452623911728523,
      "loss": 12.5495,
      "step": 7707
    },
    {
      "epoch": 1.3478469945355191,
      "grad_norm": 0.6806468307634869,
      "learning_rate": 0.00015451436798912037,
      "loss": 12.612,
      "step": 7708
    },
    {
      "epoch": 1.3480218579234973,
      "grad_norm": 0.7854250880223445,
      "learning_rate": 0.00015450249576777505,
      "loss": 12.5836,
      "step": 7709
    },
    {
      "epoch": 1.3481967213114754,
      "grad_norm": 0.7611894830613785,
      "learning_rate": 0.00015449062245348735,
      "loss": 12.6002,
      "step": 7710
    },
    {
      "epoch": 1.3483715846994535,
      "grad_norm": 0.7217055671804242,
      "learning_rate": 0.00015447874804649535,
      "loss": 12.5734,
      "step": 7711
    },
    {
      "epoch": 1.3485464480874316,
      "grad_norm": 0.7022090660715363,
      "learning_rate": 0.0001544668725470372,
      "loss": 12.214,
      "step": 7712
    },
    {
      "epoch": 1.3487213114754097,
      "grad_norm": 0.6838988965901085,
      "learning_rate": 0.000154454995955351,
      "loss": 12.6028,
      "step": 7713
    },
    {
      "epoch": 1.348896174863388,
      "grad_norm": 0.7785073405824353,
      "learning_rate": 0.00015444311827167494,
      "loss": 12.6599,
      "step": 7714
    },
    {
      "epoch": 1.349071038251366,
      "grad_norm": 0.7233902663678173,
      "learning_rate": 0.00015443123949624717,
      "loss": 12.5004,
      "step": 7715
    },
    {
      "epoch": 1.3492459016393443,
      "grad_norm": 0.996017639691201,
      "learning_rate": 0.00015441935962930598,
      "loss": 12.4807,
      "step": 7716
    },
    {
      "epoch": 1.3494207650273224,
      "grad_norm": 0.7196305768550688,
      "learning_rate": 0.0001544074786710895,
      "loss": 12.2955,
      "step": 7717
    },
    {
      "epoch": 1.3495956284153006,
      "grad_norm": 0.9605664118410286,
      "learning_rate": 0.00015439559662183602,
      "loss": 12.5487,
      "step": 7718
    },
    {
      "epoch": 1.3497704918032787,
      "grad_norm": 0.8250782540664063,
      "learning_rate": 0.00015438371348178382,
      "loss": 12.5098,
      "step": 7719
    },
    {
      "epoch": 1.3499453551912568,
      "grad_norm": 1.1906173218988048,
      "learning_rate": 0.0001543718292511712,
      "loss": 12.5121,
      "step": 7720
    },
    {
      "epoch": 1.350120218579235,
      "grad_norm": 0.6862937825499649,
      "learning_rate": 0.00015435994393023643,
      "loss": 12.4846,
      "step": 7721
    },
    {
      "epoch": 1.350295081967213,
      "grad_norm": 1.164061998456655,
      "learning_rate": 0.0001543480575192179,
      "loss": 12.4413,
      "step": 7722
    },
    {
      "epoch": 1.3504699453551914,
      "grad_norm": 0.662697394644408,
      "learning_rate": 0.00015433617001835395,
      "loss": 12.3982,
      "step": 7723
    },
    {
      "epoch": 1.3506448087431693,
      "grad_norm": 0.9796051684963837,
      "learning_rate": 0.00015432428142788293,
      "loss": 12.4883,
      "step": 7724
    },
    {
      "epoch": 1.3508196721311476,
      "grad_norm": 0.7795828194758,
      "learning_rate": 0.00015431239174804328,
      "loss": 12.514,
      "step": 7725
    },
    {
      "epoch": 1.3509945355191257,
      "grad_norm": 0.9582231940323693,
      "learning_rate": 0.00015430050097907342,
      "loss": 12.4963,
      "step": 7726
    },
    {
      "epoch": 1.3511693989071039,
      "grad_norm": 0.8036059704605648,
      "learning_rate": 0.00015428860912121181,
      "loss": 12.5927,
      "step": 7727
    },
    {
      "epoch": 1.351344262295082,
      "grad_norm": 0.9044031918720239,
      "learning_rate": 0.00015427671617469687,
      "loss": 12.4422,
      "step": 7728
    },
    {
      "epoch": 1.35151912568306,
      "grad_norm": 0.783376832622353,
      "learning_rate": 0.00015426482213976714,
      "loss": 12.1971,
      "step": 7729
    },
    {
      "epoch": 1.3516939890710382,
      "grad_norm": 0.7551743058290009,
      "learning_rate": 0.0001542529270166611,
      "loss": 12.6771,
      "step": 7730
    },
    {
      "epoch": 1.3518688524590163,
      "grad_norm": 0.8833076432194885,
      "learning_rate": 0.00015424103080561728,
      "loss": 12.4676,
      "step": 7731
    },
    {
      "epoch": 1.3520437158469945,
      "grad_norm": 0.755996738315401,
      "learning_rate": 0.00015422913350687426,
      "loss": 12.4419,
      "step": 7732
    },
    {
      "epoch": 1.3522185792349726,
      "grad_norm": 1.000896641581868,
      "learning_rate": 0.00015421723512067063,
      "loss": 12.235,
      "step": 7733
    },
    {
      "epoch": 1.352393442622951,
      "grad_norm": 0.7871432918491915,
      "learning_rate": 0.00015420533564724495,
      "loss": 12.563,
      "step": 7734
    },
    {
      "epoch": 1.352568306010929,
      "grad_norm": 0.8524688992514529,
      "learning_rate": 0.00015419343508683584,
      "loss": 12.273,
      "step": 7735
    },
    {
      "epoch": 1.3527431693989072,
      "grad_norm": 0.7280247703963232,
      "learning_rate": 0.000154181533439682,
      "loss": 12.3729,
      "step": 7736
    },
    {
      "epoch": 1.3529180327868853,
      "grad_norm": 0.8622723165524072,
      "learning_rate": 0.00015416963070602203,
      "loss": 12.5542,
      "step": 7737
    },
    {
      "epoch": 1.3530928961748634,
      "grad_norm": 0.7507625766407905,
      "learning_rate": 0.00015415772688609464,
      "loss": 12.4162,
      "step": 7738
    },
    {
      "epoch": 1.3532677595628415,
      "grad_norm": 0.759626029975704,
      "learning_rate": 0.00015414582198013856,
      "loss": 12.524,
      "step": 7739
    },
    {
      "epoch": 1.3534426229508196,
      "grad_norm": 0.68318151238217,
      "learning_rate": 0.00015413391598839246,
      "loss": 12.5575,
      "step": 7740
    },
    {
      "epoch": 1.3536174863387977,
      "grad_norm": 0.7870143940994687,
      "learning_rate": 0.00015412200891109517,
      "loss": 11.979,
      "step": 7741
    },
    {
      "epoch": 1.3537923497267759,
      "grad_norm": 0.696719632999948,
      "learning_rate": 0.0001541101007484854,
      "loss": 12.3785,
      "step": 7742
    },
    {
      "epoch": 1.3539672131147542,
      "grad_norm": 0.7252246712897571,
      "learning_rate": 0.000154098191500802,
      "loss": 12.4553,
      "step": 7743
    },
    {
      "epoch": 1.354142076502732,
      "grad_norm": 0.8645707779426883,
      "learning_rate": 0.00015408628116828375,
      "loss": 12.5446,
      "step": 7744
    },
    {
      "epoch": 1.3543169398907104,
      "grad_norm": 0.7411302754462874,
      "learning_rate": 0.00015407436975116946,
      "loss": 12.3517,
      "step": 7745
    },
    {
      "epoch": 1.3544918032786886,
      "grad_norm": 0.7754646214635753,
      "learning_rate": 0.00015406245724969807,
      "loss": 12.5407,
      "step": 7746
    },
    {
      "epoch": 1.3546666666666667,
      "grad_norm": 0.6806895374274907,
      "learning_rate": 0.0001540505436641084,
      "loss": 12.4186,
      "step": 7747
    },
    {
      "epoch": 1.3548415300546448,
      "grad_norm": 0.7671528573906107,
      "learning_rate": 0.0001540386289946394,
      "loss": 12.4671,
      "step": 7748
    },
    {
      "epoch": 1.355016393442623,
      "grad_norm": 0.7199911277290412,
      "learning_rate": 0.00015402671324152992,
      "loss": 12.4281,
      "step": 7749
    },
    {
      "epoch": 1.355191256830601,
      "grad_norm": 0.8837182547261195,
      "learning_rate": 0.00015401479640501898,
      "loss": 12.7719,
      "step": 7750
    },
    {
      "epoch": 1.3553661202185792,
      "grad_norm": 0.7210971052989811,
      "learning_rate": 0.0001540028784853455,
      "loss": 12.4999,
      "step": 7751
    },
    {
      "epoch": 1.3555409836065575,
      "grad_norm": 0.7044823098808818,
      "learning_rate": 0.00015399095948274852,
      "loss": 12.412,
      "step": 7752
    },
    {
      "epoch": 1.3557158469945354,
      "grad_norm": 0.7655452883680968,
      "learning_rate": 0.00015397903939746706,
      "loss": 12.4575,
      "step": 7753
    },
    {
      "epoch": 1.3558907103825137,
      "grad_norm": 0.7304532572715203,
      "learning_rate": 0.00015396711822974007,
      "loss": 12.4352,
      "step": 7754
    },
    {
      "epoch": 1.3560655737704919,
      "grad_norm": 0.8317114552239134,
      "learning_rate": 0.00015395519597980668,
      "loss": 12.3975,
      "step": 7755
    },
    {
      "epoch": 1.35624043715847,
      "grad_norm": 0.8021964800076848,
      "learning_rate": 0.00015394327264790593,
      "loss": 12.3444,
      "step": 7756
    },
    {
      "epoch": 1.356415300546448,
      "grad_norm": 0.7263941596593707,
      "learning_rate": 0.00015393134823427697,
      "loss": 12.4709,
      "step": 7757
    },
    {
      "epoch": 1.3565901639344262,
      "grad_norm": 0.7759375010964897,
      "learning_rate": 0.00015391942273915884,
      "loss": 12.711,
      "step": 7758
    },
    {
      "epoch": 1.3567650273224043,
      "grad_norm": 0.9312267276224939,
      "learning_rate": 0.0001539074961627908,
      "loss": 12.6227,
      "step": 7759
    },
    {
      "epoch": 1.3569398907103825,
      "grad_norm": 0.8653217784173126,
      "learning_rate": 0.00015389556850541188,
      "loss": 12.5875,
      "step": 7760
    },
    {
      "epoch": 1.3571147540983606,
      "grad_norm": 0.8242863461920653,
      "learning_rate": 0.00015388363976726133,
      "loss": 12.6454,
      "step": 7761
    },
    {
      "epoch": 1.3572896174863387,
      "grad_norm": 0.8632040175464132,
      "learning_rate": 0.0001538717099485784,
      "loss": 12.7405,
      "step": 7762
    },
    {
      "epoch": 1.357464480874317,
      "grad_norm": 0.7248051909686523,
      "learning_rate": 0.00015385977904960226,
      "loss": 12.3176,
      "step": 7763
    },
    {
      "epoch": 1.3576393442622952,
      "grad_norm": 0.790379169689711,
      "learning_rate": 0.00015384784707057217,
      "loss": 12.5751,
      "step": 7764
    },
    {
      "epoch": 1.3578142076502733,
      "grad_norm": 0.8458465528655416,
      "learning_rate": 0.00015383591401172742,
      "loss": 12.6183,
      "step": 7765
    },
    {
      "epoch": 1.3579890710382514,
      "grad_norm": 0.8404661278623171,
      "learning_rate": 0.0001538239798733073,
      "loss": 12.3303,
      "step": 7766
    },
    {
      "epoch": 1.3581639344262295,
      "grad_norm": 0.7800298399944477,
      "learning_rate": 0.0001538120446555511,
      "loss": 12.0375,
      "step": 7767
    },
    {
      "epoch": 1.3583387978142076,
      "grad_norm": 0.8134173861456527,
      "learning_rate": 0.0001538001083586982,
      "loss": 12.2594,
      "step": 7768
    },
    {
      "epoch": 1.3585136612021858,
      "grad_norm": 0.8473471823383766,
      "learning_rate": 0.00015378817098298794,
      "loss": 12.6582,
      "step": 7769
    },
    {
      "epoch": 1.3586885245901639,
      "grad_norm": 0.7638395179337202,
      "learning_rate": 0.00015377623252865968,
      "loss": 12.3275,
      "step": 7770
    },
    {
      "epoch": 1.358863387978142,
      "grad_norm": 0.8200735217673826,
      "learning_rate": 0.00015376429299595286,
      "loss": 12.4973,
      "step": 7771
    },
    {
      "epoch": 1.3590382513661203,
      "grad_norm": 0.740526635177118,
      "learning_rate": 0.0001537523523851069,
      "loss": 12.4969,
      "step": 7772
    },
    {
      "epoch": 1.3592131147540982,
      "grad_norm": 0.7743638763495878,
      "learning_rate": 0.00015374041069636123,
      "loss": 12.402,
      "step": 7773
    },
    {
      "epoch": 1.3593879781420766,
      "grad_norm": 0.7405389715212177,
      "learning_rate": 0.00015372846792995528,
      "loss": 12.3752,
      "step": 7774
    },
    {
      "epoch": 1.3595628415300547,
      "grad_norm": 0.7550816537735217,
      "learning_rate": 0.0001537165240861286,
      "loss": 12.5873,
      "step": 7775
    },
    {
      "epoch": 1.3597377049180328,
      "grad_norm": 0.7903552684706511,
      "learning_rate": 0.0001537045791651207,
      "loss": 12.617,
      "step": 7776
    },
    {
      "epoch": 1.359912568306011,
      "grad_norm": 0.781221422811572,
      "learning_rate": 0.0001536926331671711,
      "loss": 12.5615,
      "step": 7777
    },
    {
      "epoch": 1.360087431693989,
      "grad_norm": 0.6565041523174304,
      "learning_rate": 0.00015368068609251934,
      "loss": 12.6213,
      "step": 7778
    },
    {
      "epoch": 1.3602622950819672,
      "grad_norm": 0.8362337547794777,
      "learning_rate": 0.00015366873794140498,
      "loss": 12.5145,
      "step": 7779
    },
    {
      "epoch": 1.3604371584699453,
      "grad_norm": 0.7317629904184457,
      "learning_rate": 0.00015365678871406767,
      "loss": 12.3436,
      "step": 7780
    },
    {
      "epoch": 1.3606120218579236,
      "grad_norm": 0.7481266353123536,
      "learning_rate": 0.00015364483841074695,
      "loss": 12.4864,
      "step": 7781
    },
    {
      "epoch": 1.3607868852459015,
      "grad_norm": 0.7316594689489495,
      "learning_rate": 0.00015363288703168257,
      "loss": 12.5684,
      "step": 7782
    },
    {
      "epoch": 1.3609617486338799,
      "grad_norm": 0.7282863182497828,
      "learning_rate": 0.00015362093457711409,
      "loss": 12.4166,
      "step": 7783
    },
    {
      "epoch": 1.361136612021858,
      "grad_norm": 0.6575225994002074,
      "learning_rate": 0.00015360898104728124,
      "loss": 12.5128,
      "step": 7784
    },
    {
      "epoch": 1.361311475409836,
      "grad_norm": 0.7388119563613899,
      "learning_rate": 0.00015359702644242372,
      "loss": 12.0837,
      "step": 7785
    },
    {
      "epoch": 1.3614863387978142,
      "grad_norm": 0.7053616571375582,
      "learning_rate": 0.00015358507076278127,
      "loss": 12.5026,
      "step": 7786
    },
    {
      "epoch": 1.3616612021857923,
      "grad_norm": 0.7623079526201214,
      "learning_rate": 0.0001535731140085936,
      "loss": 12.5549,
      "step": 7787
    },
    {
      "epoch": 1.3618360655737705,
      "grad_norm": 0.7222867554738126,
      "learning_rate": 0.0001535611561801005,
      "loss": 12.2012,
      "step": 7788
    },
    {
      "epoch": 1.3620109289617486,
      "grad_norm": 0.7170838974697145,
      "learning_rate": 0.0001535491972775418,
      "loss": 12.3902,
      "step": 7789
    },
    {
      "epoch": 1.3621857923497267,
      "grad_norm": 0.8212971326924883,
      "learning_rate": 0.00015353723730115724,
      "loss": 12.6258,
      "step": 7790
    },
    {
      "epoch": 1.3623606557377048,
      "grad_norm": 0.6879426596716464,
      "learning_rate": 0.00015352527625118673,
      "loss": 12.5757,
      "step": 7791
    },
    {
      "epoch": 1.3625355191256832,
      "grad_norm": 0.6858164870721489,
      "learning_rate": 0.00015351331412787004,
      "loss": 12.6565,
      "step": 7792
    },
    {
      "epoch": 1.3627103825136613,
      "grad_norm": 0.8017111854578822,
      "learning_rate": 0.00015350135093144713,
      "loss": 12.6313,
      "step": 7793
    },
    {
      "epoch": 1.3628852459016394,
      "grad_norm": 0.6781675194729923,
      "learning_rate": 0.00015348938666215787,
      "loss": 12.3911,
      "step": 7794
    },
    {
      "epoch": 1.3630601092896175,
      "grad_norm": 0.7949160654878068,
      "learning_rate": 0.00015347742132024213,
      "loss": 12.6145,
      "step": 7795
    },
    {
      "epoch": 1.3632349726775956,
      "grad_norm": 0.6840762888655417,
      "learning_rate": 0.00015346545490593994,
      "loss": 12.5133,
      "step": 7796
    },
    {
      "epoch": 1.3634098360655738,
      "grad_norm": 0.7183595291216394,
      "learning_rate": 0.00015345348741949117,
      "loss": 12.5396,
      "step": 7797
    },
    {
      "epoch": 1.3635846994535519,
      "grad_norm": 0.7535220484238674,
      "learning_rate": 0.00015344151886113592,
      "loss": 12.5672,
      "step": 7798
    },
    {
      "epoch": 1.36375956284153,
      "grad_norm": 0.7266629772706034,
      "learning_rate": 0.00015342954923111408,
      "loss": 12.4862,
      "step": 7799
    },
    {
      "epoch": 1.3639344262295081,
      "grad_norm": 0.782277660751163,
      "learning_rate": 0.00015341757852966573,
      "loss": 12.4706,
      "step": 7800
    },
    {
      "epoch": 1.3641092896174865,
      "grad_norm": 0.7435711731742758,
      "learning_rate": 0.00015340560675703092,
      "loss": 12.5346,
      "step": 7801
    },
    {
      "epoch": 1.3642841530054644,
      "grad_norm": 0.7193748190103348,
      "learning_rate": 0.00015339363391344971,
      "loss": 12.726,
      "step": 7802
    },
    {
      "epoch": 1.3644590163934427,
      "grad_norm": 0.6965798037451021,
      "learning_rate": 0.00015338165999916224,
      "loss": 12.2929,
      "step": 7803
    },
    {
      "epoch": 1.3646338797814208,
      "grad_norm": 0.6807469853635498,
      "learning_rate": 0.00015336968501440855,
      "loss": 12.4205,
      "step": 7804
    },
    {
      "epoch": 1.364808743169399,
      "grad_norm": 0.6982223904317957,
      "learning_rate": 0.00015335770895942883,
      "loss": 12.5386,
      "step": 7805
    },
    {
      "epoch": 1.364983606557377,
      "grad_norm": 0.6921620320699744,
      "learning_rate": 0.0001533457318344632,
      "loss": 12.4456,
      "step": 7806
    },
    {
      "epoch": 1.3651584699453552,
      "grad_norm": 0.8259709201707106,
      "learning_rate": 0.00015333375363975183,
      "loss": 12.3953,
      "step": 7807
    },
    {
      "epoch": 1.3653333333333333,
      "grad_norm": 0.7599621757346903,
      "learning_rate": 0.00015332177437553499,
      "loss": 12.6654,
      "step": 7808
    },
    {
      "epoch": 1.3655081967213114,
      "grad_norm": 0.7184684407493962,
      "learning_rate": 0.00015330979404205283,
      "loss": 12.5355,
      "step": 7809
    },
    {
      "epoch": 1.3656830601092897,
      "grad_norm": 0.696653296178167,
      "learning_rate": 0.00015329781263954563,
      "loss": 12.2956,
      "step": 7810
    },
    {
      "epoch": 1.3658579234972676,
      "grad_norm": 0.6955715089849888,
      "learning_rate": 0.00015328583016825363,
      "loss": 12.5044,
      "step": 7811
    },
    {
      "epoch": 1.366032786885246,
      "grad_norm": 0.74823226042871,
      "learning_rate": 0.00015327384662841713,
      "loss": 12.4028,
      "step": 7812
    },
    {
      "epoch": 1.366207650273224,
      "grad_norm": 0.7211283760924447,
      "learning_rate": 0.00015326186202027642,
      "loss": 12.6265,
      "step": 7813
    },
    {
      "epoch": 1.3663825136612022,
      "grad_norm": 0.7706740779757133,
      "learning_rate": 0.00015324987634407185,
      "loss": 12.5643,
      "step": 7814
    },
    {
      "epoch": 1.3665573770491803,
      "grad_norm": 0.7616978189736637,
      "learning_rate": 0.00015323788960004377,
      "loss": 12.3583,
      "step": 7815
    },
    {
      "epoch": 1.3667322404371585,
      "grad_norm": 0.8120800754006064,
      "learning_rate": 0.00015322590178843255,
      "loss": 12.6782,
      "step": 7816
    },
    {
      "epoch": 1.3669071038251366,
      "grad_norm": 0.6779726807756993,
      "learning_rate": 0.00015321391290947856,
      "loss": 12.6129,
      "step": 7817
    },
    {
      "epoch": 1.3670819672131147,
      "grad_norm": 0.6888497179478671,
      "learning_rate": 0.0001532019229634222,
      "loss": 12.4303,
      "step": 7818
    },
    {
      "epoch": 1.3672568306010928,
      "grad_norm": 0.7724891802879418,
      "learning_rate": 0.00015318993195050396,
      "loss": 12.5295,
      "step": 7819
    },
    {
      "epoch": 1.367431693989071,
      "grad_norm": 0.710354539200616,
      "learning_rate": 0.0001531779398709643,
      "loss": 12.2315,
      "step": 7820
    },
    {
      "epoch": 1.3676065573770493,
      "grad_norm": 0.6700549411090552,
      "learning_rate": 0.0001531659467250436,
      "loss": 12.3997,
      "step": 7821
    },
    {
      "epoch": 1.3677814207650274,
      "grad_norm": 0.6517212522801608,
      "learning_rate": 0.00015315395251298247,
      "loss": 12.282,
      "step": 7822
    },
    {
      "epoch": 1.3679562841530055,
      "grad_norm": 0.7123844354117358,
      "learning_rate": 0.0001531419572350214,
      "loss": 12.3084,
      "step": 7823
    },
    {
      "epoch": 1.3681311475409836,
      "grad_norm": 0.7007465991595322,
      "learning_rate": 0.00015312996089140088,
      "loss": 12.4764,
      "step": 7824
    },
    {
      "epoch": 1.3683060109289618,
      "grad_norm": 0.7703520697758005,
      "learning_rate": 0.00015311796348236155,
      "loss": 12.6483,
      "step": 7825
    },
    {
      "epoch": 1.3684808743169399,
      "grad_norm": 0.7749029814293705,
      "learning_rate": 0.00015310596500814393,
      "loss": 12.4802,
      "step": 7826
    },
    {
      "epoch": 1.368655737704918,
      "grad_norm": 0.763026415509345,
      "learning_rate": 0.0001530939654689887,
      "loss": 12.4314,
      "step": 7827
    },
    {
      "epoch": 1.3688306010928961,
      "grad_norm": 0.7160729616987797,
      "learning_rate": 0.00015308196486513642,
      "loss": 12.6663,
      "step": 7828
    },
    {
      "epoch": 1.3690054644808742,
      "grad_norm": 0.7328849470825051,
      "learning_rate": 0.00015306996319682772,
      "loss": 12.4309,
      "step": 7829
    },
    {
      "epoch": 1.3691803278688526,
      "grad_norm": 0.7765974533857188,
      "learning_rate": 0.0001530579604643034,
      "loss": 12.5808,
      "step": 7830
    },
    {
      "epoch": 1.3693551912568305,
      "grad_norm": 0.8787197796226064,
      "learning_rate": 0.00015304595666780396,
      "loss": 12.5659,
      "step": 7831
    },
    {
      "epoch": 1.3695300546448088,
      "grad_norm": 0.7014726387964635,
      "learning_rate": 0.00015303395180757027,
      "loss": 12.5017,
      "step": 7832
    },
    {
      "epoch": 1.369704918032787,
      "grad_norm": 0.7038816453778238,
      "learning_rate": 0.00015302194588384302,
      "loss": 12.2714,
      "step": 7833
    },
    {
      "epoch": 1.369879781420765,
      "grad_norm": 0.7312667754536248,
      "learning_rate": 0.00015300993889686294,
      "loss": 12.6732,
      "step": 7834
    },
    {
      "epoch": 1.3700546448087432,
      "grad_norm": 0.7090410879219178,
      "learning_rate": 0.00015299793084687084,
      "loss": 12.3953,
      "step": 7835
    },
    {
      "epoch": 1.3702295081967213,
      "grad_norm": 0.6692841642865952,
      "learning_rate": 0.00015298592173410746,
      "loss": 12.1108,
      "step": 7836
    },
    {
      "epoch": 1.3704043715846994,
      "grad_norm": 0.7070406018290613,
      "learning_rate": 0.00015297391155881372,
      "loss": 12.206,
      "step": 7837
    },
    {
      "epoch": 1.3705792349726775,
      "grad_norm": 0.7646844352724976,
      "learning_rate": 0.00015296190032123034,
      "loss": 12.2954,
      "step": 7838
    },
    {
      "epoch": 1.3707540983606559,
      "grad_norm": 0.7071533705837948,
      "learning_rate": 0.0001529498880215983,
      "loss": 12.5098,
      "step": 7839
    },
    {
      "epoch": 1.3709289617486338,
      "grad_norm": 0.7267868479624943,
      "learning_rate": 0.00015293787466015842,
      "loss": 12.6577,
      "step": 7840
    },
    {
      "epoch": 1.371103825136612,
      "grad_norm": 0.6929381045167851,
      "learning_rate": 0.00015292586023715155,
      "loss": 12.7305,
      "step": 7841
    },
    {
      "epoch": 1.3712786885245902,
      "grad_norm": 0.7147558846759451,
      "learning_rate": 0.00015291384475281877,
      "loss": 12.2822,
      "step": 7842
    },
    {
      "epoch": 1.3714535519125683,
      "grad_norm": 0.669978606275904,
      "learning_rate": 0.00015290182820740085,
      "loss": 12.3694,
      "step": 7843
    },
    {
      "epoch": 1.3716284153005465,
      "grad_norm": 0.740130326973818,
      "learning_rate": 0.00015288981060113888,
      "loss": 12.7995,
      "step": 7844
    },
    {
      "epoch": 1.3718032786885246,
      "grad_norm": 0.7454822677086242,
      "learning_rate": 0.00015287779193427382,
      "loss": 12.5408,
      "step": 7845
    },
    {
      "epoch": 1.3719781420765027,
      "grad_norm": 0.7748639995674861,
      "learning_rate": 0.0001528657722070467,
      "loss": 12.5856,
      "step": 7846
    },
    {
      "epoch": 1.3721530054644808,
      "grad_norm": 0.7816250303240068,
      "learning_rate": 0.00015285375141969849,
      "loss": 12.607,
      "step": 7847
    },
    {
      "epoch": 1.372327868852459,
      "grad_norm": 0.9107696769299728,
      "learning_rate": 0.0001528417295724703,
      "loss": 12.4905,
      "step": 7848
    },
    {
      "epoch": 1.372502732240437,
      "grad_norm": 0.7555885194797288,
      "learning_rate": 0.00015282970666560317,
      "loss": 12.557,
      "step": 7849
    },
    {
      "epoch": 1.3726775956284154,
      "grad_norm": 0.8207904611664093,
      "learning_rate": 0.0001528176826993382,
      "loss": 12.4132,
      "step": 7850
    },
    {
      "epoch": 1.3728524590163935,
      "grad_norm": 0.7601831805427562,
      "learning_rate": 0.00015280565767391657,
      "loss": 12.7487,
      "step": 7851
    },
    {
      "epoch": 1.3730273224043716,
      "grad_norm": 0.9803678107948909,
      "learning_rate": 0.00015279363158957935,
      "loss": 12.4278,
      "step": 7852
    },
    {
      "epoch": 1.3732021857923498,
      "grad_norm": 0.8540970124681418,
      "learning_rate": 0.0001527816044465677,
      "loss": 12.5835,
      "step": 7853
    },
    {
      "epoch": 1.3733770491803279,
      "grad_norm": 1.1075836141120214,
      "learning_rate": 0.00015276957624512285,
      "loss": 12.3809,
      "step": 7854
    },
    {
      "epoch": 1.373551912568306,
      "grad_norm": 0.8839227304483409,
      "learning_rate": 0.00015275754698548593,
      "loss": 12.3502,
      "step": 7855
    },
    {
      "epoch": 1.3737267759562841,
      "grad_norm": 1.0777310847806043,
      "learning_rate": 0.00015274551666789825,
      "loss": 12.4159,
      "step": 7856
    },
    {
      "epoch": 1.3739016393442622,
      "grad_norm": 0.8342188939876082,
      "learning_rate": 0.00015273348529260098,
      "loss": 12.4624,
      "step": 7857
    },
    {
      "epoch": 1.3740765027322404,
      "grad_norm": 0.9091860521205853,
      "learning_rate": 0.00015272145285983544,
      "loss": 12.3051,
      "step": 7858
    },
    {
      "epoch": 1.3742513661202187,
      "grad_norm": 0.8735477607235402,
      "learning_rate": 0.00015270941936984287,
      "loss": 12.4323,
      "step": 7859
    },
    {
      "epoch": 1.3744262295081966,
      "grad_norm": 0.8383153228344971,
      "learning_rate": 0.0001526973848228646,
      "loss": 12.3792,
      "step": 7860
    },
    {
      "epoch": 1.374601092896175,
      "grad_norm": 0.8058766627249766,
      "learning_rate": 0.00015268534921914197,
      "loss": 12.4859,
      "step": 7861
    },
    {
      "epoch": 1.374775956284153,
      "grad_norm": 0.7832194807009019,
      "learning_rate": 0.00015267331255891634,
      "loss": 12.4464,
      "step": 7862
    },
    {
      "epoch": 1.3749508196721312,
      "grad_norm": 0.7998244173121103,
      "learning_rate": 0.00015266127484242904,
      "loss": 12.4966,
      "step": 7863
    },
    {
      "epoch": 1.3751256830601093,
      "grad_norm": 0.76359826202014,
      "learning_rate": 0.0001526492360699215,
      "loss": 12.5862,
      "step": 7864
    },
    {
      "epoch": 1.3753005464480874,
      "grad_norm": 0.7252711970005525,
      "learning_rate": 0.00015263719624163512,
      "loss": 12.4535,
      "step": 7865
    },
    {
      "epoch": 1.3754754098360655,
      "grad_norm": 0.7893861600266904,
      "learning_rate": 0.00015262515535781134,
      "loss": 12.5336,
      "step": 7866
    },
    {
      "epoch": 1.3756502732240437,
      "grad_norm": 0.7672235388296663,
      "learning_rate": 0.00015261311341869163,
      "loss": 12.2219,
      "step": 7867
    },
    {
      "epoch": 1.375825136612022,
      "grad_norm": 0.7974438114635073,
      "learning_rate": 0.00015260107042451747,
      "loss": 12.3313,
      "step": 7868
    },
    {
      "epoch": 1.376,
      "grad_norm": 0.7485987398865425,
      "learning_rate": 0.0001525890263755303,
      "loss": 12.546,
      "step": 7869
    },
    {
      "epoch": 1.3761748633879782,
      "grad_norm": 0.746612143009083,
      "learning_rate": 0.0001525769812719717,
      "loss": 12.3785,
      "step": 7870
    },
    {
      "epoch": 1.3763497267759564,
      "grad_norm": 0.8093553649042446,
      "learning_rate": 0.00015256493511408317,
      "loss": 12.6096,
      "step": 7871
    },
    {
      "epoch": 1.3765245901639345,
      "grad_norm": 0.8380362693127836,
      "learning_rate": 0.0001525528879021063,
      "loss": 12.4452,
      "step": 7872
    },
    {
      "epoch": 1.3766994535519126,
      "grad_norm": 0.7464754325835069,
      "learning_rate": 0.00015254083963628267,
      "loss": 12.1118,
      "step": 7873
    },
    {
      "epoch": 1.3768743169398907,
      "grad_norm": 0.8068928243183873,
      "learning_rate": 0.0001525287903168539,
      "loss": 12.656,
      "step": 7874
    },
    {
      "epoch": 1.3770491803278688,
      "grad_norm": 0.7015488090003359,
      "learning_rate": 0.0001525167399440616,
      "loss": 12.3828,
      "step": 7875
    },
    {
      "epoch": 1.377224043715847,
      "grad_norm": 0.7118367759266877,
      "learning_rate": 0.00015250468851814742,
      "loss": 12.2279,
      "step": 7876
    },
    {
      "epoch": 1.377398907103825,
      "grad_norm": 0.7204449193396809,
      "learning_rate": 0.000152492636039353,
      "loss": 12.5221,
      "step": 7877
    },
    {
      "epoch": 1.3775737704918032,
      "grad_norm": 0.7525813981571878,
      "learning_rate": 0.00015248058250792008,
      "loss": 12.4988,
      "step": 7878
    },
    {
      "epoch": 1.3777486338797815,
      "grad_norm": 0.7527260161896764,
      "learning_rate": 0.00015246852792409033,
      "loss": 12.4621,
      "step": 7879
    },
    {
      "epoch": 1.3779234972677596,
      "grad_norm": 0.7742561326413003,
      "learning_rate": 0.0001524564722881055,
      "loss": 12.4509,
      "step": 7880
    },
    {
      "epoch": 1.3780983606557378,
      "grad_norm": 0.7137363977327891,
      "learning_rate": 0.0001524444156002073,
      "loss": 12.4063,
      "step": 7881
    },
    {
      "epoch": 1.3782732240437159,
      "grad_norm": 0.7302642919901429,
      "learning_rate": 0.00015243235786063761,
      "loss": 12.5081,
      "step": 7882
    },
    {
      "epoch": 1.378448087431694,
      "grad_norm": 0.7221296198444274,
      "learning_rate": 0.00015242029906963812,
      "loss": 12.4353,
      "step": 7883
    },
    {
      "epoch": 1.3786229508196721,
      "grad_norm": 0.7128437146822192,
      "learning_rate": 0.00015240823922745066,
      "loss": 12.5317,
      "step": 7884
    },
    {
      "epoch": 1.3787978142076502,
      "grad_norm": 0.8008083900003433,
      "learning_rate": 0.00015239617833431713,
      "loss": 12.4255,
      "step": 7885
    },
    {
      "epoch": 1.3789726775956284,
      "grad_norm": 0.7370568649819038,
      "learning_rate": 0.00015238411639047935,
      "loss": 12.5164,
      "step": 7886
    },
    {
      "epoch": 1.3791475409836065,
      "grad_norm": 0.709095896690852,
      "learning_rate": 0.00015237205339617917,
      "loss": 12.1563,
      "step": 7887
    },
    {
      "epoch": 1.3793224043715848,
      "grad_norm": 0.7481524416464936,
      "learning_rate": 0.0001523599893516585,
      "loss": 12.4064,
      "step": 7888
    },
    {
      "epoch": 1.3794972677595627,
      "grad_norm": 0.7063021319471664,
      "learning_rate": 0.0001523479242571593,
      "loss": 12.3219,
      "step": 7889
    },
    {
      "epoch": 1.379672131147541,
      "grad_norm": 0.7793129662209154,
      "learning_rate": 0.00015233585811292352,
      "loss": 12.4049,
      "step": 7890
    },
    {
      "epoch": 1.3798469945355192,
      "grad_norm": 0.7007533865966116,
      "learning_rate": 0.00015232379091919304,
      "loss": 12.4726,
      "step": 7891
    },
    {
      "epoch": 1.3800218579234973,
      "grad_norm": 0.7165745226650707,
      "learning_rate": 0.00015231172267620994,
      "loss": 12.4891,
      "step": 7892
    },
    {
      "epoch": 1.3801967213114754,
      "grad_norm": 0.713960070562325,
      "learning_rate": 0.00015229965338421614,
      "loss": 12.376,
      "step": 7893
    },
    {
      "epoch": 1.3803715846994535,
      "grad_norm": 0.8978772193556341,
      "learning_rate": 0.00015228758304345377,
      "loss": 12.4935,
      "step": 7894
    },
    {
      "epoch": 1.3805464480874317,
      "grad_norm": 0.7648303772192179,
      "learning_rate": 0.00015227551165416476,
      "loss": 12.6631,
      "step": 7895
    },
    {
      "epoch": 1.3807213114754098,
      "grad_norm": 0.686499352084862,
      "learning_rate": 0.00015226343921659124,
      "loss": 12.5415,
      "step": 7896
    },
    {
      "epoch": 1.3808961748633881,
      "grad_norm": 0.7227439143234876,
      "learning_rate": 0.00015225136573097532,
      "loss": 12.5692,
      "step": 7897
    },
    {
      "epoch": 1.381071038251366,
      "grad_norm": 0.7855166454405158,
      "learning_rate": 0.00015223929119755907,
      "loss": 12.4058,
      "step": 7898
    },
    {
      "epoch": 1.3812459016393444,
      "grad_norm": 0.8187512449336327,
      "learning_rate": 0.00015222721561658462,
      "loss": 12.623,
      "step": 7899
    },
    {
      "epoch": 1.3814207650273225,
      "grad_norm": 0.8331640857596728,
      "learning_rate": 0.00015221513898829416,
      "loss": 12.405,
      "step": 7900
    },
    {
      "epoch": 1.3815956284153006,
      "grad_norm": 0.808001368670777,
      "learning_rate": 0.00015220306131292986,
      "loss": 12.4105,
      "step": 7901
    },
    {
      "epoch": 1.3817704918032787,
      "grad_norm": 0.7760309980614793,
      "learning_rate": 0.00015219098259073387,
      "loss": 12.4023,
      "step": 7902
    },
    {
      "epoch": 1.3819453551912568,
      "grad_norm": 0.8798387830213746,
      "learning_rate": 0.00015217890282194843,
      "loss": 12.1795,
      "step": 7903
    },
    {
      "epoch": 1.382120218579235,
      "grad_norm": 0.7456530231400476,
      "learning_rate": 0.00015216682200681574,
      "loss": 12.7203,
      "step": 7904
    },
    {
      "epoch": 1.382295081967213,
      "grad_norm": 0.783228330744672,
      "learning_rate": 0.00015215474014557815,
      "loss": 12.3753,
      "step": 7905
    },
    {
      "epoch": 1.3824699453551912,
      "grad_norm": 0.796469712005767,
      "learning_rate": 0.00015214265723847788,
      "loss": 12.5035,
      "step": 7906
    },
    {
      "epoch": 1.3826448087431693,
      "grad_norm": 0.817032104353189,
      "learning_rate": 0.00015213057328575724,
      "loss": 12.564,
      "step": 7907
    },
    {
      "epoch": 1.3828196721311476,
      "grad_norm": 0.7843416631542485,
      "learning_rate": 0.0001521184882876585,
      "loss": 12.665,
      "step": 7908
    },
    {
      "epoch": 1.3829945355191255,
      "grad_norm": 0.7082846458160101,
      "learning_rate": 0.0001521064022444241,
      "loss": 12.194,
      "step": 7909
    },
    {
      "epoch": 1.3831693989071039,
      "grad_norm": 0.7206360084836695,
      "learning_rate": 0.0001520943151562963,
      "loss": 12.2642,
      "step": 7910
    },
    {
      "epoch": 1.383344262295082,
      "grad_norm": 0.6998625479409247,
      "learning_rate": 0.00015208222702351755,
      "loss": 12.369,
      "step": 7911
    },
    {
      "epoch": 1.3835191256830601,
      "grad_norm": 0.7156858476088755,
      "learning_rate": 0.00015207013784633024,
      "loss": 12.36,
      "step": 7912
    },
    {
      "epoch": 1.3836939890710382,
      "grad_norm": 0.7171534149866874,
      "learning_rate": 0.0001520580476249768,
      "loss": 12.4185,
      "step": 7913
    },
    {
      "epoch": 1.3838688524590164,
      "grad_norm": 0.7073234455442273,
      "learning_rate": 0.00015204595635969964,
      "loss": 12.4017,
      "step": 7914
    },
    {
      "epoch": 1.3840437158469945,
      "grad_norm": 0.6775623642318007,
      "learning_rate": 0.00015203386405074126,
      "loss": 12.4695,
      "step": 7915
    },
    {
      "epoch": 1.3842185792349726,
      "grad_norm": 0.6884662691445722,
      "learning_rate": 0.00015202177069834416,
      "loss": 12.1643,
      "step": 7916
    },
    {
      "epoch": 1.384393442622951,
      "grad_norm": 0.7374173046661984,
      "learning_rate": 0.0001520096763027508,
      "loss": 12.2762,
      "step": 7917
    },
    {
      "epoch": 1.3845683060109288,
      "grad_norm": 0.7369444999248275,
      "learning_rate": 0.00015199758086420376,
      "loss": 12.6965,
      "step": 7918
    },
    {
      "epoch": 1.3847431693989072,
      "grad_norm": 0.776341580428632,
      "learning_rate": 0.00015198548438294558,
      "loss": 12.6764,
      "step": 7919
    },
    {
      "epoch": 1.3849180327868853,
      "grad_norm": 0.6749901309560515,
      "learning_rate": 0.0001519733868592188,
      "loss": 12.2534,
      "step": 7920
    },
    {
      "epoch": 1.3850928961748634,
      "grad_norm": 0.7520924568495975,
      "learning_rate": 0.00015196128829326604,
      "loss": 12.5718,
      "step": 7921
    },
    {
      "epoch": 1.3852677595628415,
      "grad_norm": 0.727886583499679,
      "learning_rate": 0.0001519491886853299,
      "loss": 12.4198,
      "step": 7922
    },
    {
      "epoch": 1.3854426229508197,
      "grad_norm": 0.6576207897105049,
      "learning_rate": 0.00015193708803565303,
      "loss": 12.6051,
      "step": 7923
    },
    {
      "epoch": 1.3856174863387978,
      "grad_norm": 0.7751054065246223,
      "learning_rate": 0.0001519249863444781,
      "loss": 12.4921,
      "step": 7924
    },
    {
      "epoch": 1.385792349726776,
      "grad_norm": 0.7514558150206301,
      "learning_rate": 0.0001519128836120477,
      "loss": 12.3752,
      "step": 7925
    },
    {
      "epoch": 1.385967213114754,
      "grad_norm": 0.7595666890564347,
      "learning_rate": 0.00015190077983860467,
      "loss": 12.559,
      "step": 7926
    },
    {
      "epoch": 1.3861420765027321,
      "grad_norm": 0.7300901592193251,
      "learning_rate": 0.0001518886750243916,
      "loss": 12.5078,
      "step": 7927
    },
    {
      "epoch": 1.3863169398907105,
      "grad_norm": 0.732099444813577,
      "learning_rate": 0.0001518765691696513,
      "loss": 12.1939,
      "step": 7928
    },
    {
      "epoch": 1.3864918032786886,
      "grad_norm": 0.7461699478698076,
      "learning_rate": 0.00015186446227462646,
      "loss": 12.4448,
      "step": 7929
    },
    {
      "epoch": 1.3866666666666667,
      "grad_norm": 0.7352685405235759,
      "learning_rate": 0.00015185235433955993,
      "loss": 12.5743,
      "step": 7930
    },
    {
      "epoch": 1.3868415300546448,
      "grad_norm": 0.7264726672237529,
      "learning_rate": 0.00015184024536469453,
      "loss": 12.0125,
      "step": 7931
    },
    {
      "epoch": 1.387016393442623,
      "grad_norm": 0.7266369337274731,
      "learning_rate": 0.000151828135350273,
      "loss": 11.9475,
      "step": 7932
    },
    {
      "epoch": 1.387191256830601,
      "grad_norm": 0.6851572252409671,
      "learning_rate": 0.00015181602429653824,
      "loss": 12.6227,
      "step": 7933
    },
    {
      "epoch": 1.3873661202185792,
      "grad_norm": 0.8769348031931381,
      "learning_rate": 0.00015180391220373304,
      "loss": 12.3328,
      "step": 7934
    },
    {
      "epoch": 1.3875409836065573,
      "grad_norm": 0.8556908848339349,
      "learning_rate": 0.00015179179907210042,
      "loss": 12.5319,
      "step": 7935
    },
    {
      "epoch": 1.3877158469945354,
      "grad_norm": 0.7924736488719207,
      "learning_rate": 0.0001517796849018832,
      "loss": 12.6976,
      "step": 7936
    },
    {
      "epoch": 1.3878907103825138,
      "grad_norm": 0.740884267369453,
      "learning_rate": 0.00015176756969332425,
      "loss": 12.4358,
      "step": 7937
    },
    {
      "epoch": 1.3880655737704917,
      "grad_norm": 0.7399012729822488,
      "learning_rate": 0.0001517554534466666,
      "loss": 12.4448,
      "step": 7938
    },
    {
      "epoch": 1.38824043715847,
      "grad_norm": 0.8311119787285014,
      "learning_rate": 0.00015174333616215324,
      "loss": 12.606,
      "step": 7939
    },
    {
      "epoch": 1.3884153005464481,
      "grad_norm": 0.6665969149733194,
      "learning_rate": 0.0001517312178400271,
      "loss": 12.4641,
      "step": 7940
    },
    {
      "epoch": 1.3885901639344262,
      "grad_norm": 0.6858534922068817,
      "learning_rate": 0.00015171909848053119,
      "loss": 12.3779,
      "step": 7941
    },
    {
      "epoch": 1.3887650273224044,
      "grad_norm": 0.7593524629247193,
      "learning_rate": 0.00015170697808390856,
      "loss": 12.4591,
      "step": 7942
    },
    {
      "epoch": 1.3889398907103825,
      "grad_norm": 0.6959895473298555,
      "learning_rate": 0.00015169485665040227,
      "loss": 12.3693,
      "step": 7943
    },
    {
      "epoch": 1.3891147540983606,
      "grad_norm": 0.7619628489241673,
      "learning_rate": 0.00015168273418025537,
      "loss": 12.46,
      "step": 7944
    },
    {
      "epoch": 1.3892896174863387,
      "grad_norm": 0.7254659979922883,
      "learning_rate": 0.00015167061067371097,
      "loss": 12.3923,
      "step": 7945
    },
    {
      "epoch": 1.389464480874317,
      "grad_norm": 0.8384383790368886,
      "learning_rate": 0.0001516584861310122,
      "loss": 12.4642,
      "step": 7946
    },
    {
      "epoch": 1.389639344262295,
      "grad_norm": 0.6643127178296852,
      "learning_rate": 0.0001516463605524021,
      "loss": 12.238,
      "step": 7947
    },
    {
      "epoch": 1.3898142076502733,
      "grad_norm": 0.8373488149388592,
      "learning_rate": 0.00015163423393812394,
      "loss": 12.5651,
      "step": 7948
    },
    {
      "epoch": 1.3899890710382514,
      "grad_norm": 0.7341693079366252,
      "learning_rate": 0.00015162210628842087,
      "loss": 12.6922,
      "step": 7949
    },
    {
      "epoch": 1.3901639344262295,
      "grad_norm": 0.795026838646223,
      "learning_rate": 0.00015160997760353605,
      "loss": 12.4881,
      "step": 7950
    },
    {
      "epoch": 1.3903387978142077,
      "grad_norm": 0.7901454753357857,
      "learning_rate": 0.0001515978478837127,
      "loss": 12.4871,
      "step": 7951
    },
    {
      "epoch": 1.3905136612021858,
      "grad_norm": 0.828196835844852,
      "learning_rate": 0.00015158571712919411,
      "loss": 12.8306,
      "step": 7952
    },
    {
      "epoch": 1.390688524590164,
      "grad_norm": 0.7969233985117073,
      "learning_rate": 0.00015157358534022348,
      "loss": 12.3918,
      "step": 7953
    },
    {
      "epoch": 1.390863387978142,
      "grad_norm": 0.7260953239598106,
      "learning_rate": 0.00015156145251704413,
      "loss": 12.4541,
      "step": 7954
    },
    {
      "epoch": 1.3910382513661201,
      "grad_norm": 0.7651628708343071,
      "learning_rate": 0.00015154931865989928,
      "loss": 12.3763,
      "step": 7955
    },
    {
      "epoch": 1.3912131147540983,
      "grad_norm": 0.7514192581918877,
      "learning_rate": 0.0001515371837690324,
      "loss": 12.612,
      "step": 7956
    },
    {
      "epoch": 1.3913879781420766,
      "grad_norm": 0.7955586696441415,
      "learning_rate": 0.00015152504784468672,
      "loss": 12.2244,
      "step": 7957
    },
    {
      "epoch": 1.3915628415300547,
      "grad_norm": 0.7454109305011077,
      "learning_rate": 0.00015151291088710557,
      "loss": 12.4488,
      "step": 7958
    },
    {
      "epoch": 1.3917377049180328,
      "grad_norm": 0.843008809103038,
      "learning_rate": 0.00015150077289653244,
      "loss": 12.5718,
      "step": 7959
    },
    {
      "epoch": 1.391912568306011,
      "grad_norm": 0.7153943152697663,
      "learning_rate": 0.0001514886338732107,
      "loss": 12.535,
      "step": 7960
    },
    {
      "epoch": 1.392087431693989,
      "grad_norm": 1.0497067682252061,
      "learning_rate": 0.00015147649381738374,
      "loss": 12.1205,
      "step": 7961
    },
    {
      "epoch": 1.3922622950819672,
      "grad_norm": 0.7843406371978685,
      "learning_rate": 0.00015146435272929502,
      "loss": 12.5637,
      "step": 7962
    },
    {
      "epoch": 1.3924371584699453,
      "grad_norm": 1.0349955345036348,
      "learning_rate": 0.00015145221060918802,
      "loss": 12.4002,
      "step": 7963
    },
    {
      "epoch": 1.3926120218579234,
      "grad_norm": 0.7356668991237412,
      "learning_rate": 0.0001514400674573062,
      "loss": 12.3649,
      "step": 7964
    },
    {
      "epoch": 1.3927868852459016,
      "grad_norm": 1.0624868640452407,
      "learning_rate": 0.00015142792327389311,
      "loss": 12.5373,
      "step": 7965
    },
    {
      "epoch": 1.39296174863388,
      "grad_norm": 0.7198129793234938,
      "learning_rate": 0.00015141577805919226,
      "loss": 12.176,
      "step": 7966
    },
    {
      "epoch": 1.3931366120218578,
      "grad_norm": 0.9452904512562164,
      "learning_rate": 0.0001514036318134472,
      "loss": 12.4027,
      "step": 7967
    },
    {
      "epoch": 1.3933114754098361,
      "grad_norm": 0.7496312975075633,
      "learning_rate": 0.00015139148453690145,
      "loss": 12.448,
      "step": 7968
    },
    {
      "epoch": 1.3934863387978143,
      "grad_norm": 0.9099466117626761,
      "learning_rate": 0.00015137933622979866,
      "loss": 12.3484,
      "step": 7969
    },
    {
      "epoch": 1.3936612021857924,
      "grad_norm": 1.070613669606374,
      "learning_rate": 0.00015136718689238242,
      "loss": 12.3587,
      "step": 7970
    },
    {
      "epoch": 1.3938360655737705,
      "grad_norm": 0.8858199733831821,
      "learning_rate": 0.00015135503652489635,
      "loss": 12.3529,
      "step": 7971
    },
    {
      "epoch": 1.3940109289617486,
      "grad_norm": 1.5684022478021893,
      "learning_rate": 0.00015134288512758416,
      "loss": 12.4978,
      "step": 7972
    },
    {
      "epoch": 1.3941857923497267,
      "grad_norm": 0.7293030948226776,
      "learning_rate": 0.00015133073270068946,
      "loss": 12.4094,
      "step": 7973
    },
    {
      "epoch": 1.3943606557377048,
      "grad_norm": 0.9474692830804002,
      "learning_rate": 0.00015131857924445598,
      "loss": 12.3162,
      "step": 7974
    },
    {
      "epoch": 1.3945355191256832,
      "grad_norm": 0.8336913579676852,
      "learning_rate": 0.00015130642475912737,
      "loss": 12.5831,
      "step": 7975
    },
    {
      "epoch": 1.394710382513661,
      "grad_norm": 0.8884791560764446,
      "learning_rate": 0.00015129426924494744,
      "loss": 12.1776,
      "step": 7976
    },
    {
      "epoch": 1.3948852459016394,
      "grad_norm": 0.7114915271084162,
      "learning_rate": 0.00015128211270215992,
      "loss": 12.3914,
      "step": 7977
    },
    {
      "epoch": 1.3950601092896175,
      "grad_norm": 0.883539204398382,
      "learning_rate": 0.00015126995513100856,
      "loss": 12.5011,
      "step": 7978
    },
    {
      "epoch": 1.3952349726775957,
      "grad_norm": 0.755264728909199,
      "learning_rate": 0.00015125779653173723,
      "loss": 12.4444,
      "step": 7979
    },
    {
      "epoch": 1.3954098360655738,
      "grad_norm": 0.8392471227813506,
      "learning_rate": 0.00015124563690458966,
      "loss": 12.606,
      "step": 7980
    },
    {
      "epoch": 1.395584699453552,
      "grad_norm": 0.7355166817853592,
      "learning_rate": 0.00015123347624980974,
      "loss": 12.6497,
      "step": 7981
    },
    {
      "epoch": 1.39575956284153,
      "grad_norm": 0.8941227209924585,
      "learning_rate": 0.0001512213145676413,
      "loss": 12.2608,
      "step": 7982
    },
    {
      "epoch": 1.3959344262295081,
      "grad_norm": 0.7366354894266127,
      "learning_rate": 0.00015120915185832823,
      "loss": 12.4642,
      "step": 7983
    },
    {
      "epoch": 1.3961092896174863,
      "grad_norm": 0.9487369021879799,
      "learning_rate": 0.00015119698812211447,
      "loss": 12.6286,
      "step": 7984
    },
    {
      "epoch": 1.3962841530054644,
      "grad_norm": 0.786209508361558,
      "learning_rate": 0.00015118482335924388,
      "loss": 12.4389,
      "step": 7985
    },
    {
      "epoch": 1.3964590163934427,
      "grad_norm": 0.7881278933270847,
      "learning_rate": 0.0001511726575699604,
      "loss": 12.2946,
      "step": 7986
    },
    {
      "epoch": 1.3966338797814208,
      "grad_norm": 0.8400456466858033,
      "learning_rate": 0.00015116049075450803,
      "loss": 12.6432,
      "step": 7987
    },
    {
      "epoch": 1.396808743169399,
      "grad_norm": 0.7566973095959149,
      "learning_rate": 0.00015114832291313076,
      "loss": 12.3738,
      "step": 7988
    },
    {
      "epoch": 1.396983606557377,
      "grad_norm": 0.7492198173848632,
      "learning_rate": 0.00015113615404607254,
      "loss": 12.4889,
      "step": 7989
    },
    {
      "epoch": 1.3971584699453552,
      "grad_norm": 0.7364614999233813,
      "learning_rate": 0.00015112398415357745,
      "loss": 12.588,
      "step": 7990
    },
    {
      "epoch": 1.3973333333333333,
      "grad_norm": 0.7062257837050198,
      "learning_rate": 0.0001511118132358895,
      "loss": 12.4253,
      "step": 7991
    },
    {
      "epoch": 1.3975081967213114,
      "grad_norm": 0.8159206677689265,
      "learning_rate": 0.00015109964129325274,
      "loss": 12.4282,
      "step": 7992
    },
    {
      "epoch": 1.3976830601092896,
      "grad_norm": 0.7028217473208135,
      "learning_rate": 0.00015108746832591128,
      "loss": 12.2397,
      "step": 7993
    },
    {
      "epoch": 1.3978579234972677,
      "grad_norm": 0.7263811262712458,
      "learning_rate": 0.00015107529433410925,
      "loss": 12.4883,
      "step": 7994
    },
    {
      "epoch": 1.398032786885246,
      "grad_norm": 0.7165768411028748,
      "learning_rate": 0.0001510631193180907,
      "loss": 12.5054,
      "step": 7995
    },
    {
      "epoch": 1.398207650273224,
      "grad_norm": 0.7515396410653903,
      "learning_rate": 0.00015105094327809985,
      "loss": 12.5297,
      "step": 7996
    },
    {
      "epoch": 1.3983825136612023,
      "grad_norm": 0.7835688016007183,
      "learning_rate": 0.00015103876621438086,
      "loss": 12.5785,
      "step": 7997
    },
    {
      "epoch": 1.3985573770491804,
      "grad_norm": 0.7595767322581966,
      "learning_rate": 0.00015102658812717786,
      "loss": 12.6722,
      "step": 7998
    },
    {
      "epoch": 1.3987322404371585,
      "grad_norm": 0.858167268146442,
      "learning_rate": 0.0001510144090167351,
      "loss": 12.5164,
      "step": 7999
    },
    {
      "epoch": 1.3989071038251366,
      "grad_norm": 0.7057267955774116,
      "learning_rate": 0.00015100222888329683,
      "loss": 12.5422,
      "step": 8000
    },
    {
      "epoch": 1.3990819672131147,
      "grad_norm": 0.6982416727500168,
      "learning_rate": 0.00015099004772710725,
      "loss": 12.4705,
      "step": 8001
    },
    {
      "epoch": 1.3992568306010928,
      "grad_norm": 0.7304775287692773,
      "learning_rate": 0.00015097786554841067,
      "loss": 12.5202,
      "step": 8002
    },
    {
      "epoch": 1.399431693989071,
      "grad_norm": 0.7795368183161974,
      "learning_rate": 0.00015096568234745136,
      "loss": 12.6155,
      "step": 8003
    },
    {
      "epoch": 1.3996065573770493,
      "grad_norm": 0.7237072560199784,
      "learning_rate": 0.0001509534981244736,
      "loss": 12.7796,
      "step": 8004
    },
    {
      "epoch": 1.3997814207650272,
      "grad_norm": 0.6725981499821543,
      "learning_rate": 0.0001509413128797218,
      "loss": 12.3915,
      "step": 8005
    },
    {
      "epoch": 1.3999562841530055,
      "grad_norm": 0.7615232452487072,
      "learning_rate": 0.00015092912661344028,
      "loss": 12.2578,
      "step": 8006
    },
    {
      "epoch": 1.4001311475409837,
      "grad_norm": 0.7172028019107828,
      "learning_rate": 0.00015091693932587337,
      "loss": 12.3813,
      "step": 8007
    },
    {
      "epoch": 1.4003060109289618,
      "grad_norm": 0.751606695259355,
      "learning_rate": 0.00015090475101726546,
      "loss": 12.577,
      "step": 8008
    },
    {
      "epoch": 1.40048087431694,
      "grad_norm": 0.8707650378165689,
      "learning_rate": 0.00015089256168786104,
      "loss": 12.3358,
      "step": 8009
    },
    {
      "epoch": 1.400655737704918,
      "grad_norm": 0.7332951659035251,
      "learning_rate": 0.00015088037133790448,
      "loss": 12.5534,
      "step": 8010
    },
    {
      "epoch": 1.4008306010928961,
      "grad_norm": 0.7247426817667776,
      "learning_rate": 0.00015086817996764026,
      "loss": 12.39,
      "step": 8011
    },
    {
      "epoch": 1.4010054644808743,
      "grad_norm": 0.671435160005309,
      "learning_rate": 0.00015085598757731282,
      "loss": 12.4024,
      "step": 8012
    },
    {
      "epoch": 1.4011803278688524,
      "grad_norm": 0.7599835371991469,
      "learning_rate": 0.0001508437941671667,
      "loss": 12.3659,
      "step": 8013
    },
    {
      "epoch": 1.4013551912568305,
      "grad_norm": 0.7936660392157586,
      "learning_rate": 0.0001508315997374464,
      "loss": 12.2585,
      "step": 8014
    },
    {
      "epoch": 1.4015300546448088,
      "grad_norm": 0.7297923365688652,
      "learning_rate": 0.00015081940428839647,
      "loss": 12.5311,
      "step": 8015
    },
    {
      "epoch": 1.401704918032787,
      "grad_norm": 0.7107967231694663,
      "learning_rate": 0.00015080720782026141,
      "loss": 12.2841,
      "step": 8016
    },
    {
      "epoch": 1.401879781420765,
      "grad_norm": 0.7331756723615701,
      "learning_rate": 0.00015079501033328586,
      "loss": 12.5108,
      "step": 8017
    },
    {
      "epoch": 1.4020546448087432,
      "grad_norm": 0.7114634706663003,
      "learning_rate": 0.00015078281182771437,
      "loss": 12.4316,
      "step": 8018
    },
    {
      "epoch": 1.4022295081967213,
      "grad_norm": 0.7203393783145455,
      "learning_rate": 0.0001507706123037916,
      "loss": 12.368,
      "step": 8019
    },
    {
      "epoch": 1.4024043715846994,
      "grad_norm": 0.743754336851893,
      "learning_rate": 0.0001507584117617622,
      "loss": 12.436,
      "step": 8020
    },
    {
      "epoch": 1.4025792349726776,
      "grad_norm": 0.694664995315629,
      "learning_rate": 0.00015074621020187072,
      "loss": 12.2934,
      "step": 8021
    },
    {
      "epoch": 1.4027540983606557,
      "grad_norm": 0.6939993585640046,
      "learning_rate": 0.00015073400762436197,
      "loss": 12.3833,
      "step": 8022
    },
    {
      "epoch": 1.4029289617486338,
      "grad_norm": 0.719332115135954,
      "learning_rate": 0.00015072180402948054,
      "loss": 12.2872,
      "step": 8023
    },
    {
      "epoch": 1.4031038251366121,
      "grad_norm": 0.7079071707955068,
      "learning_rate": 0.00015070959941747124,
      "loss": 12.6138,
      "step": 8024
    },
    {
      "epoch": 1.40327868852459,
      "grad_norm": 0.6829611431072377,
      "learning_rate": 0.00015069739378857877,
      "loss": 12.137,
      "step": 8025
    },
    {
      "epoch": 1.4034535519125684,
      "grad_norm": 0.692359638403121,
      "learning_rate": 0.00015068518714304787,
      "loss": 12.6085,
      "step": 8026
    },
    {
      "epoch": 1.4036284153005465,
      "grad_norm": 0.674004565406666,
      "learning_rate": 0.00015067297948112338,
      "loss": 12.5263,
      "step": 8027
    },
    {
      "epoch": 1.4038032786885246,
      "grad_norm": 0.7576748685524763,
      "learning_rate": 0.00015066077080305002,
      "loss": 12.5711,
      "step": 8028
    },
    {
      "epoch": 1.4039781420765027,
      "grad_norm": 0.741181003016534,
      "learning_rate": 0.0001506485611090727,
      "loss": 12.1289,
      "step": 8029
    },
    {
      "epoch": 1.4041530054644809,
      "grad_norm": 0.8139363750290226,
      "learning_rate": 0.0001506363503994362,
      "loss": 12.4319,
      "step": 8030
    },
    {
      "epoch": 1.404327868852459,
      "grad_norm": 0.6978691459139812,
      "learning_rate": 0.0001506241386743854,
      "loss": 12.3674,
      "step": 8031
    },
    {
      "epoch": 1.404502732240437,
      "grad_norm": 0.8982082430076718,
      "learning_rate": 0.0001506119259341652,
      "loss": 12.4677,
      "step": 8032
    },
    {
      "epoch": 1.4046775956284154,
      "grad_norm": 0.8217800299779592,
      "learning_rate": 0.0001505997121790205,
      "loss": 12.3667,
      "step": 8033
    },
    {
      "epoch": 1.4048524590163933,
      "grad_norm": 0.843628579572895,
      "learning_rate": 0.00015058749740919618,
      "loss": 12.4066,
      "step": 8034
    },
    {
      "epoch": 1.4050273224043717,
      "grad_norm": 0.7623234733328967,
      "learning_rate": 0.0001505752816249372,
      "loss": 12.7617,
      "step": 8035
    },
    {
      "epoch": 1.4052021857923498,
      "grad_norm": 0.8545769796096679,
      "learning_rate": 0.00015056306482648854,
      "loss": 12.386,
      "step": 8036
    },
    {
      "epoch": 1.405377049180328,
      "grad_norm": 0.697559508989017,
      "learning_rate": 0.00015055084701409517,
      "loss": 12.3186,
      "step": 8037
    },
    {
      "epoch": 1.405551912568306,
      "grad_norm": 0.8114640181877288,
      "learning_rate": 0.00015053862818800216,
      "loss": 12.1646,
      "step": 8038
    },
    {
      "epoch": 1.4057267759562841,
      "grad_norm": 0.657750191343373,
      "learning_rate": 0.00015052640834845447,
      "loss": 12.0662,
      "step": 8039
    },
    {
      "epoch": 1.4059016393442623,
      "grad_norm": 0.7290661806422221,
      "learning_rate": 0.0001505141874956971,
      "loss": 12.3417,
      "step": 8040
    },
    {
      "epoch": 1.4060765027322404,
      "grad_norm": 0.7778444848613678,
      "learning_rate": 0.00015050196562997522,
      "loss": 12.4367,
      "step": 8041
    },
    {
      "epoch": 1.4062513661202185,
      "grad_norm": 0.6396545485554549,
      "learning_rate": 0.00015048974275153385,
      "loss": 12.2117,
      "step": 8042
    },
    {
      "epoch": 1.4064262295081966,
      "grad_norm": 0.7156105916726041,
      "learning_rate": 0.00015047751886061814,
      "loss": 12.4752,
      "step": 8043
    },
    {
      "epoch": 1.406601092896175,
      "grad_norm": 0.752083960177357,
      "learning_rate": 0.00015046529395747315,
      "loss": 12.2552,
      "step": 8044
    },
    {
      "epoch": 1.406775956284153,
      "grad_norm": 0.7732041596515198,
      "learning_rate": 0.0001504530680423441,
      "loss": 12.5799,
      "step": 8045
    },
    {
      "epoch": 1.4069508196721312,
      "grad_norm": 0.8233482475541765,
      "learning_rate": 0.00015044084111547613,
      "loss": 12.5662,
      "step": 8046
    },
    {
      "epoch": 1.4071256830601093,
      "grad_norm": 0.6626283487424042,
      "learning_rate": 0.0001504286131771144,
      "loss": 12.5238,
      "step": 8047
    },
    {
      "epoch": 1.4073005464480874,
      "grad_norm": 0.7810540720096324,
      "learning_rate": 0.0001504163842275041,
      "loss": 12.3883,
      "step": 8048
    },
    {
      "epoch": 1.4074754098360656,
      "grad_norm": 0.6595206071764598,
      "learning_rate": 0.00015040415426689055,
      "loss": 12.4982,
      "step": 8049
    },
    {
      "epoch": 1.4076502732240437,
      "grad_norm": 0.7990514429561053,
      "learning_rate": 0.00015039192329551894,
      "loss": 12.3821,
      "step": 8050
    },
    {
      "epoch": 1.4078251366120218,
      "grad_norm": 0.7322479083161354,
      "learning_rate": 0.00015037969131363453,
      "loss": 12.5085,
      "step": 8051
    },
    {
      "epoch": 1.408,
      "grad_norm": 0.7758670555423084,
      "learning_rate": 0.00015036745832148265,
      "loss": 12.4408,
      "step": 8052
    },
    {
      "epoch": 1.4081748633879783,
      "grad_norm": 0.6972884161527769,
      "learning_rate": 0.00015035522431930856,
      "loss": 12.5619,
      "step": 8053
    },
    {
      "epoch": 1.4083497267759562,
      "grad_norm": 0.8276520674055393,
      "learning_rate": 0.0001503429893073576,
      "loss": 12.6135,
      "step": 8054
    },
    {
      "epoch": 1.4085245901639345,
      "grad_norm": 0.9190636506083758,
      "learning_rate": 0.00015033075328587517,
      "loss": 12.6127,
      "step": 8055
    },
    {
      "epoch": 1.4086994535519126,
      "grad_norm": 0.8087272509620158,
      "learning_rate": 0.0001503185162551066,
      "loss": 12.3984,
      "step": 8056
    },
    {
      "epoch": 1.4088743169398907,
      "grad_norm": 0.7119311159833399,
      "learning_rate": 0.00015030627821529727,
      "loss": 12.4669,
      "step": 8057
    },
    {
      "epoch": 1.4090491803278689,
      "grad_norm": 0.8220287055317805,
      "learning_rate": 0.00015029403916669258,
      "loss": 12.3517,
      "step": 8058
    },
    {
      "epoch": 1.409224043715847,
      "grad_norm": 0.763539473374422,
      "learning_rate": 0.000150281799109538,
      "loss": 12.2932,
      "step": 8059
    },
    {
      "epoch": 1.409398907103825,
      "grad_norm": 0.7795469287753232,
      "learning_rate": 0.00015026955804407893,
      "loss": 12.3589,
      "step": 8060
    },
    {
      "epoch": 1.4095737704918032,
      "grad_norm": 0.7715812304883946,
      "learning_rate": 0.0001502573159705609,
      "loss": 12.3152,
      "step": 8061
    },
    {
      "epoch": 1.4097486338797816,
      "grad_norm": 0.7786282141617518,
      "learning_rate": 0.00015024507288922937,
      "loss": 12.6565,
      "step": 8062
    },
    {
      "epoch": 1.4099234972677595,
      "grad_norm": 0.7946215981462795,
      "learning_rate": 0.0001502328288003299,
      "loss": 12.4717,
      "step": 8063
    },
    {
      "epoch": 1.4100983606557378,
      "grad_norm": 0.7065787732521283,
      "learning_rate": 0.00015022058370410792,
      "loss": 12.4435,
      "step": 8064
    },
    {
      "epoch": 1.410273224043716,
      "grad_norm": 0.7470095326901384,
      "learning_rate": 0.00015020833760080905,
      "loss": 12.4275,
      "step": 8065
    },
    {
      "epoch": 1.410448087431694,
      "grad_norm": 0.7833517052351703,
      "learning_rate": 0.00015019609049067887,
      "loss": 12.6622,
      "step": 8066
    },
    {
      "epoch": 1.4106229508196722,
      "grad_norm": 0.7399238191849835,
      "learning_rate": 0.00015018384237396292,
      "loss": 12.5244,
      "step": 8067
    },
    {
      "epoch": 1.4107978142076503,
      "grad_norm": 0.7355182152403059,
      "learning_rate": 0.0001501715932509069,
      "loss": 12.6028,
      "step": 8068
    },
    {
      "epoch": 1.4109726775956284,
      "grad_norm": 0.771777765569022,
      "learning_rate": 0.00015015934312175632,
      "loss": 12.3527,
      "step": 8069
    },
    {
      "epoch": 1.4111475409836065,
      "grad_norm": 0.7158258188955345,
      "learning_rate": 0.00015014709198675697,
      "loss": 12.5128,
      "step": 8070
    },
    {
      "epoch": 1.4113224043715846,
      "grad_norm": 0.7512591858733823,
      "learning_rate": 0.0001501348398461544,
      "loss": 12.6478,
      "step": 8071
    },
    {
      "epoch": 1.4114972677595627,
      "grad_norm": 0.6782766663285104,
      "learning_rate": 0.00015012258670019438,
      "loss": 12.6196,
      "step": 8072
    },
    {
      "epoch": 1.411672131147541,
      "grad_norm": 0.7615511622181743,
      "learning_rate": 0.00015011033254912258,
      "loss": 12.5283,
      "step": 8073
    },
    {
      "epoch": 1.4118469945355192,
      "grad_norm": 0.6864357474075719,
      "learning_rate": 0.00015009807739318478,
      "loss": 12.579,
      "step": 8074
    },
    {
      "epoch": 1.4120218579234973,
      "grad_norm": 0.75867773755457,
      "learning_rate": 0.0001500858212326267,
      "loss": 12.5524,
      "step": 8075
    },
    {
      "epoch": 1.4121967213114754,
      "grad_norm": 0.7944701465335932,
      "learning_rate": 0.0001500735640676941,
      "loss": 12.511,
      "step": 8076
    },
    {
      "epoch": 1.4123715846994536,
      "grad_norm": 0.6914960916466044,
      "learning_rate": 0.00015006130589863284,
      "loss": 12.5111,
      "step": 8077
    },
    {
      "epoch": 1.4125464480874317,
      "grad_norm": 0.6888742698703753,
      "learning_rate": 0.00015004904672568864,
      "loss": 12.463,
      "step": 8078
    },
    {
      "epoch": 1.4127213114754098,
      "grad_norm": 0.7016178513675122,
      "learning_rate": 0.00015003678654910743,
      "loss": 12.5222,
      "step": 8079
    },
    {
      "epoch": 1.412896174863388,
      "grad_norm": 0.684598197159483,
      "learning_rate": 0.00015002452536913498,
      "loss": 12.4979,
      "step": 8080
    },
    {
      "epoch": 1.413071038251366,
      "grad_norm": 0.7228446674179841,
      "learning_rate": 0.0001500122631860172,
      "loss": 12.2244,
      "step": 8081
    },
    {
      "epoch": 1.4132459016393444,
      "grad_norm": 0.7005754031143885,
      "learning_rate": 0.00015000000000000001,
      "loss": 12.279,
      "step": 8082
    },
    {
      "epoch": 1.4134207650273223,
      "grad_norm": 0.7495790445480615,
      "learning_rate": 0.00014998773581132926,
      "loss": 12.4304,
      "step": 8083
    },
    {
      "epoch": 1.4135956284153006,
      "grad_norm": 0.715752049909187,
      "learning_rate": 0.000149975470620251,
      "loss": 12.4275,
      "step": 8084
    },
    {
      "epoch": 1.4137704918032787,
      "grad_norm": 0.773897047768189,
      "learning_rate": 0.00014996320442701102,
      "loss": 12.4871,
      "step": 8085
    },
    {
      "epoch": 1.4139453551912569,
      "grad_norm": 1.1928444870672623,
      "learning_rate": 0.00014995093723185545,
      "loss": 12.3233,
      "step": 8086
    },
    {
      "epoch": 1.414120218579235,
      "grad_norm": 0.716785975744775,
      "learning_rate": 0.0001499386690350302,
      "loss": 12.4737,
      "step": 8087
    },
    {
      "epoch": 1.414295081967213,
      "grad_norm": 0.9440052563745499,
      "learning_rate": 0.0001499263998367813,
      "loss": 12.5452,
      "step": 8088
    },
    {
      "epoch": 1.4144699453551912,
      "grad_norm": 1.0032979108105562,
      "learning_rate": 0.0001499141296373548,
      "loss": 12.5405,
      "step": 8089
    },
    {
      "epoch": 1.4146448087431693,
      "grad_norm": 0.7492641890497801,
      "learning_rate": 0.0001499018584369967,
      "loss": 12.4702,
      "step": 8090
    },
    {
      "epoch": 1.4148196721311477,
      "grad_norm": 0.8455185480575657,
      "learning_rate": 0.00014988958623595315,
      "loss": 12.5707,
      "step": 8091
    },
    {
      "epoch": 1.4149945355191256,
      "grad_norm": 0.7763651474164269,
      "learning_rate": 0.0001498773130344702,
      "loss": 12.2633,
      "step": 8092
    },
    {
      "epoch": 1.415169398907104,
      "grad_norm": 0.7164307096109964,
      "learning_rate": 0.00014986503883279403,
      "loss": 12.3567,
      "step": 8093
    },
    {
      "epoch": 1.415344262295082,
      "grad_norm": 0.7702062555088066,
      "learning_rate": 0.00014985276363117065,
      "loss": 12.5251,
      "step": 8094
    },
    {
      "epoch": 1.4155191256830602,
      "grad_norm": 0.9052096973209101,
      "learning_rate": 0.00014984048742984637,
      "loss": 12.5995,
      "step": 8095
    },
    {
      "epoch": 1.4156939890710383,
      "grad_norm": 0.8931189568158744,
      "learning_rate": 0.00014982821022906724,
      "loss": 12.2481,
      "step": 8096
    },
    {
      "epoch": 1.4158688524590164,
      "grad_norm": 0.7711605857865212,
      "learning_rate": 0.0001498159320290795,
      "loss": 12.2019,
      "step": 8097
    },
    {
      "epoch": 1.4160437158469945,
      "grad_norm": 0.709663840439138,
      "learning_rate": 0.0001498036528301294,
      "loss": 12.4989,
      "step": 8098
    },
    {
      "epoch": 1.4162185792349726,
      "grad_norm": 0.7070028602488675,
      "learning_rate": 0.0001497913726324631,
      "loss": 12.512,
      "step": 8099
    },
    {
      "epoch": 1.4163934426229507,
      "grad_norm": 0.757877463878274,
      "learning_rate": 0.00014977909143632694,
      "loss": 12.2574,
      "step": 8100
    },
    {
      "epoch": 1.4165683060109289,
      "grad_norm": 0.7851962264588013,
      "learning_rate": 0.0001497668092419671,
      "loss": 12.4723,
      "step": 8101
    },
    {
      "epoch": 1.4167431693989072,
      "grad_norm": 0.7109510397745691,
      "learning_rate": 0.00014975452604962998,
      "loss": 12.4304,
      "step": 8102
    },
    {
      "epoch": 1.4169180327868853,
      "grad_norm": 0.8632540342480458,
      "learning_rate": 0.00014974224185956186,
      "loss": 12.5626,
      "step": 8103
    },
    {
      "epoch": 1.4170928961748634,
      "grad_norm": 0.83328370600807,
      "learning_rate": 0.00014972995667200903,
      "loss": 12.4671,
      "step": 8104
    },
    {
      "epoch": 1.4172677595628416,
      "grad_norm": 0.8127881754211385,
      "learning_rate": 0.00014971767048721786,
      "loss": 12.6866,
      "step": 8105
    },
    {
      "epoch": 1.4174426229508197,
      "grad_norm": 0.8610168223089317,
      "learning_rate": 0.0001497053833054348,
      "loss": 12.6761,
      "step": 8106
    },
    {
      "epoch": 1.4176174863387978,
      "grad_norm": 0.8249257585765051,
      "learning_rate": 0.0001496930951269061,
      "loss": 12.7514,
      "step": 8107
    },
    {
      "epoch": 1.417792349726776,
      "grad_norm": 0.7505656645522075,
      "learning_rate": 0.00014968080595187833,
      "loss": 12.2721,
      "step": 8108
    },
    {
      "epoch": 1.417967213114754,
      "grad_norm": 0.8102721827503117,
      "learning_rate": 0.00014966851578059786,
      "loss": 12.3016,
      "step": 8109
    },
    {
      "epoch": 1.4181420765027322,
      "grad_norm": 0.6869129659468547,
      "learning_rate": 0.00014965622461331115,
      "loss": 12.2468,
      "step": 8110
    },
    {
      "epoch": 1.4183169398907105,
      "grad_norm": 0.7158546603801399,
      "learning_rate": 0.00014964393245026466,
      "loss": 12.4136,
      "step": 8111
    },
    {
      "epoch": 1.4184918032786884,
      "grad_norm": 0.7300729582914477,
      "learning_rate": 0.0001496316392917049,
      "loss": 12.459,
      "step": 8112
    },
    {
      "epoch": 1.4186666666666667,
      "grad_norm": 0.6932734312713101,
      "learning_rate": 0.00014961934513787838,
      "loss": 12.462,
      "step": 8113
    },
    {
      "epoch": 1.4188415300546449,
      "grad_norm": 0.7807665925639229,
      "learning_rate": 0.00014960704998903164,
      "loss": 12.3621,
      "step": 8114
    },
    {
      "epoch": 1.419016393442623,
      "grad_norm": 0.6713970673293862,
      "learning_rate": 0.00014959475384541122,
      "loss": 12.466,
      "step": 8115
    },
    {
      "epoch": 1.419191256830601,
      "grad_norm": 0.6844077864142205,
      "learning_rate": 0.00014958245670726371,
      "loss": 12.4394,
      "step": 8116
    },
    {
      "epoch": 1.4193661202185792,
      "grad_norm": 0.8555206873151281,
      "learning_rate": 0.0001495701585748357,
      "loss": 12.6993,
      "step": 8117
    },
    {
      "epoch": 1.4195409836065573,
      "grad_norm": 0.8408994300713404,
      "learning_rate": 0.00014955785944837386,
      "loss": 12.1862,
      "step": 8118
    },
    {
      "epoch": 1.4197158469945355,
      "grad_norm": 0.7210818549693598,
      "learning_rate": 0.00014954555932812473,
      "loss": 12.6144,
      "step": 8119
    },
    {
      "epoch": 1.4198907103825138,
      "grad_norm": 0.6960408835192655,
      "learning_rate": 0.00014953325821433504,
      "loss": 12.2578,
      "step": 8120
    },
    {
      "epoch": 1.4200655737704917,
      "grad_norm": 0.8015199088846641,
      "learning_rate": 0.00014952095610725139,
      "loss": 12.0378,
      "step": 8121
    },
    {
      "epoch": 1.42024043715847,
      "grad_norm": 0.750674121957495,
      "learning_rate": 0.0001495086530071206,
      "loss": 12.4985,
      "step": 8122
    },
    {
      "epoch": 1.4204153005464482,
      "grad_norm": 0.7745223594924857,
      "learning_rate": 0.00014949634891418923,
      "loss": 12.4739,
      "step": 8123
    },
    {
      "epoch": 1.4205901639344263,
      "grad_norm": 0.713550627134828,
      "learning_rate": 0.0001494840438287041,
      "loss": 12.595,
      "step": 8124
    },
    {
      "epoch": 1.4207650273224044,
      "grad_norm": 0.731894601246426,
      "learning_rate": 0.000149471737750912,
      "loss": 12.5906,
      "step": 8125
    },
    {
      "epoch": 1.4209398907103825,
      "grad_norm": 0.700918593360723,
      "learning_rate": 0.0001494594306810596,
      "loss": 12.5044,
      "step": 8126
    },
    {
      "epoch": 1.4211147540983606,
      "grad_norm": 0.7602107829034339,
      "learning_rate": 0.00014944712261939379,
      "loss": 12.4193,
      "step": 8127
    },
    {
      "epoch": 1.4212896174863388,
      "grad_norm": 0.681748799764201,
      "learning_rate": 0.00014943481356616135,
      "loss": 12.2798,
      "step": 8128
    },
    {
      "epoch": 1.4214644808743169,
      "grad_norm": 0.7983691043719461,
      "learning_rate": 0.0001494225035216091,
      "loss": 12.5177,
      "step": 8129
    },
    {
      "epoch": 1.421639344262295,
      "grad_norm": 0.678316343983982,
      "learning_rate": 0.0001494101924859839,
      "loss": 12.5155,
      "step": 8130
    },
    {
      "epoch": 1.4218142076502733,
      "grad_norm": 0.694845211363677,
      "learning_rate": 0.00014939788045953268,
      "loss": 12.5229,
      "step": 8131
    },
    {
      "epoch": 1.4219890710382512,
      "grad_norm": 0.7007440220093701,
      "learning_rate": 0.00014938556744250223,
      "loss": 12.4264,
      "step": 8132
    },
    {
      "epoch": 1.4221639344262296,
      "grad_norm": 0.7973428231347942,
      "learning_rate": 0.00014937325343513953,
      "loss": 12.4139,
      "step": 8133
    },
    {
      "epoch": 1.4223387978142077,
      "grad_norm": 0.6695028663156162,
      "learning_rate": 0.00014936093843769154,
      "loss": 12.294,
      "step": 8134
    },
    {
      "epoch": 1.4225136612021858,
      "grad_norm": 0.8551014662424024,
      "learning_rate": 0.00014934862245040512,
      "loss": 12.6135,
      "step": 8135
    },
    {
      "epoch": 1.422688524590164,
      "grad_norm": 0.7503216372083706,
      "learning_rate": 0.00014933630547352735,
      "loss": 12.5421,
      "step": 8136
    },
    {
      "epoch": 1.422863387978142,
      "grad_norm": 0.7502091782301824,
      "learning_rate": 0.00014932398750730513,
      "loss": 12.5842,
      "step": 8137
    },
    {
      "epoch": 1.4230382513661202,
      "grad_norm": 0.7254256642312877,
      "learning_rate": 0.00014931166855198556,
      "loss": 12.4075,
      "step": 8138
    },
    {
      "epoch": 1.4232131147540983,
      "grad_norm": 0.7547905037874232,
      "learning_rate": 0.0001492993486078156,
      "loss": 12.5056,
      "step": 8139
    },
    {
      "epoch": 1.4233879781420766,
      "grad_norm": 0.7797055538296304,
      "learning_rate": 0.00014928702767504233,
      "loss": 12.5419,
      "step": 8140
    },
    {
      "epoch": 1.4235628415300545,
      "grad_norm": 0.8527677693096717,
      "learning_rate": 0.00014927470575391282,
      "loss": 12.3817,
      "step": 8141
    },
    {
      "epoch": 1.4237377049180329,
      "grad_norm": 0.7136164587638834,
      "learning_rate": 0.0001492623828446742,
      "loss": 12.4416,
      "step": 8142
    },
    {
      "epoch": 1.423912568306011,
      "grad_norm": 0.9192852009246197,
      "learning_rate": 0.00014925005894757352,
      "loss": 12.5735,
      "step": 8143
    },
    {
      "epoch": 1.424087431693989,
      "grad_norm": 0.7168418208694654,
      "learning_rate": 0.00014923773406285795,
      "loss": 12.4345,
      "step": 8144
    },
    {
      "epoch": 1.4242622950819672,
      "grad_norm": 0.8058484970961761,
      "learning_rate": 0.00014922540819077465,
      "loss": 12.6943,
      "step": 8145
    },
    {
      "epoch": 1.4244371584699453,
      "grad_norm": 0.9256630571766031,
      "learning_rate": 0.00014921308133157072,
      "loss": 12.5903,
      "step": 8146
    },
    {
      "epoch": 1.4246120218579235,
      "grad_norm": 0.8893803826887087,
      "learning_rate": 0.00014920075348549345,
      "loss": 12.4013,
      "step": 8147
    },
    {
      "epoch": 1.4247868852459016,
      "grad_norm": 0.7838509244956113,
      "learning_rate": 0.00014918842465279,
      "loss": 11.9267,
      "step": 8148
    },
    {
      "epoch": 1.4249617486338797,
      "grad_norm": 0.8215589882882204,
      "learning_rate": 0.00014917609483370761,
      "loss": 12.429,
      "step": 8149
    },
    {
      "epoch": 1.4251366120218578,
      "grad_norm": 0.7413251037746693,
      "learning_rate": 0.00014916376402849353,
      "loss": 12.533,
      "step": 8150
    },
    {
      "epoch": 1.4253114754098362,
      "grad_norm": 0.7093890046118985,
      "learning_rate": 0.000149151432237395,
      "loss": 12.3888,
      "step": 8151
    },
    {
      "epoch": 1.4254863387978143,
      "grad_norm": 0.7420120681980609,
      "learning_rate": 0.00014913909946065936,
      "loss": 12.3875,
      "step": 8152
    },
    {
      "epoch": 1.4256612021857924,
      "grad_norm": 0.7560413920969292,
      "learning_rate": 0.0001491267656985339,
      "loss": 12.4148,
      "step": 8153
    },
    {
      "epoch": 1.4258360655737705,
      "grad_norm": 0.7812079271419133,
      "learning_rate": 0.00014911443095126593,
      "loss": 12.3969,
      "step": 8154
    },
    {
      "epoch": 1.4260109289617486,
      "grad_norm": 0.7185512448931468,
      "learning_rate": 0.00014910209521910286,
      "loss": 12.5018,
      "step": 8155
    },
    {
      "epoch": 1.4261857923497268,
      "grad_norm": 0.7354027285581253,
      "learning_rate": 0.000149089758502292,
      "loss": 12.2373,
      "step": 8156
    },
    {
      "epoch": 1.4263606557377049,
      "grad_norm": 0.6983641802762396,
      "learning_rate": 0.00014907742080108073,
      "loss": 12.521,
      "step": 8157
    },
    {
      "epoch": 1.426535519125683,
      "grad_norm": 0.7318673222751899,
      "learning_rate": 0.0001490650821157165,
      "loss": 12.2417,
      "step": 8158
    },
    {
      "epoch": 1.4267103825136611,
      "grad_norm": 0.7240039514143026,
      "learning_rate": 0.00014905274244644677,
      "loss": 12.4165,
      "step": 8159
    },
    {
      "epoch": 1.4268852459016395,
      "grad_norm": 0.8427147206360993,
      "learning_rate": 0.00014904040179351886,
      "loss": 12.5601,
      "step": 8160
    },
    {
      "epoch": 1.4270601092896174,
      "grad_norm": 0.7444205842476086,
      "learning_rate": 0.0001490280601571804,
      "loss": 12.4847,
      "step": 8161
    },
    {
      "epoch": 1.4272349726775957,
      "grad_norm": 0.7477626603766597,
      "learning_rate": 0.00014901571753767872,
      "loss": 12.5772,
      "step": 8162
    },
    {
      "epoch": 1.4274098360655738,
      "grad_norm": 0.7699129393458319,
      "learning_rate": 0.00014900337393526145,
      "loss": 12.3185,
      "step": 8163
    },
    {
      "epoch": 1.427584699453552,
      "grad_norm": 0.7797952520578934,
      "learning_rate": 0.00014899102935017605,
      "loss": 12.3948,
      "step": 8164
    },
    {
      "epoch": 1.42775956284153,
      "grad_norm": 0.6973527102165944,
      "learning_rate": 0.00014897868378267013,
      "loss": 12.4672,
      "step": 8165
    },
    {
      "epoch": 1.4279344262295082,
      "grad_norm": 0.7484874565000758,
      "learning_rate": 0.0001489663372329912,
      "loss": 12.4382,
      "step": 8166
    },
    {
      "epoch": 1.4281092896174863,
      "grad_norm": 0.7906634883155205,
      "learning_rate": 0.00014895398970138683,
      "loss": 12.545,
      "step": 8167
    },
    {
      "epoch": 1.4282841530054644,
      "grad_norm": 0.7153001924537568,
      "learning_rate": 0.0001489416411881047,
      "loss": 12.4904,
      "step": 8168
    },
    {
      "epoch": 1.4284590163934427,
      "grad_norm": 0.8099954473326949,
      "learning_rate": 0.00014892929169339235,
      "loss": 12.3743,
      "step": 8169
    },
    {
      "epoch": 1.4286338797814206,
      "grad_norm": 0.6663943378317175,
      "learning_rate": 0.00014891694121749752,
      "loss": 12.2485,
      "step": 8170
    },
    {
      "epoch": 1.428808743169399,
      "grad_norm": 0.8017474658633686,
      "learning_rate": 0.0001489045897606678,
      "loss": 12.5317,
      "step": 8171
    },
    {
      "epoch": 1.428983606557377,
      "grad_norm": 0.8271185564750967,
      "learning_rate": 0.00014889223732315087,
      "loss": 12.3677,
      "step": 8172
    },
    {
      "epoch": 1.4291584699453552,
      "grad_norm": 0.8502074188797617,
      "learning_rate": 0.00014887988390519452,
      "loss": 11.9557,
      "step": 8173
    },
    {
      "epoch": 1.4293333333333333,
      "grad_norm": 0.7109484194280143,
      "learning_rate": 0.00014886752950704637,
      "loss": 12.3399,
      "step": 8174
    },
    {
      "epoch": 1.4295081967213115,
      "grad_norm": 0.7050781739183777,
      "learning_rate": 0.00014885517412895424,
      "loss": 12.495,
      "step": 8175
    },
    {
      "epoch": 1.4296830601092896,
      "grad_norm": 0.6953923475522584,
      "learning_rate": 0.00014884281777116583,
      "loss": 12.3548,
      "step": 8176
    },
    {
      "epoch": 1.4298579234972677,
      "grad_norm": 0.7767545097384236,
      "learning_rate": 0.00014883046043392902,
      "loss": 12.3567,
      "step": 8177
    },
    {
      "epoch": 1.4300327868852458,
      "grad_norm": 0.7295129804216309,
      "learning_rate": 0.0001488181021174915,
      "loss": 12.5433,
      "step": 8178
    },
    {
      "epoch": 1.430207650273224,
      "grad_norm": 0.7953353427478229,
      "learning_rate": 0.00014880574282210115,
      "loss": 12.4524,
      "step": 8179
    },
    {
      "epoch": 1.4303825136612023,
      "grad_norm": 0.708704512986367,
      "learning_rate": 0.0001487933825480058,
      "loss": 12.3948,
      "step": 8180
    },
    {
      "epoch": 1.4305573770491804,
      "grad_norm": 0.8290686187449756,
      "learning_rate": 0.00014878102129545329,
      "loss": 12.3716,
      "step": 8181
    },
    {
      "epoch": 1.4307322404371585,
      "grad_norm": 0.6736605860476605,
      "learning_rate": 0.00014876865906469154,
      "loss": 12.4628,
      "step": 8182
    },
    {
      "epoch": 1.4309071038251366,
      "grad_norm": 0.7823557489122538,
      "learning_rate": 0.00014875629585596846,
      "loss": 12.2318,
      "step": 8183
    },
    {
      "epoch": 1.4310819672131148,
      "grad_norm": 0.830512388060757,
      "learning_rate": 0.00014874393166953192,
      "loss": 12.4101,
      "step": 8184
    },
    {
      "epoch": 1.4312568306010929,
      "grad_norm": 1.034184491297834,
      "learning_rate": 0.0001487315665056299,
      "loss": 12.5627,
      "step": 8185
    },
    {
      "epoch": 1.431431693989071,
      "grad_norm": 0.71897898684248,
      "learning_rate": 0.00014871920036451037,
      "loss": 12.6086,
      "step": 8186
    },
    {
      "epoch": 1.4316065573770491,
      "grad_norm": 0.913111122147829,
      "learning_rate": 0.00014870683324642124,
      "loss": 12.4291,
      "step": 8187
    },
    {
      "epoch": 1.4317814207650272,
      "grad_norm": 0.9077494146050484,
      "learning_rate": 0.00014869446515161058,
      "loss": 12.5226,
      "step": 8188
    },
    {
      "epoch": 1.4319562841530056,
      "grad_norm": 0.9370330941452724,
      "learning_rate": 0.0001486820960803264,
      "loss": 12.5106,
      "step": 8189
    },
    {
      "epoch": 1.4321311475409835,
      "grad_norm": 0.7059357602237014,
      "learning_rate": 0.0001486697260328167,
      "loss": 12.2609,
      "step": 8190
    },
    {
      "epoch": 1.4323060109289618,
      "grad_norm": 0.8098284434856978,
      "learning_rate": 0.00014865735500932953,
      "loss": 12.4639,
      "step": 8191
    },
    {
      "epoch": 1.43248087431694,
      "grad_norm": 0.832862568232886,
      "learning_rate": 0.00014864498301011303,
      "loss": 12.6359,
      "step": 8192
    },
    {
      "epoch": 1.432655737704918,
      "grad_norm": 0.7775201686266944,
      "learning_rate": 0.00014863261003541525,
      "loss": 12.2518,
      "step": 8193
    },
    {
      "epoch": 1.4328306010928962,
      "grad_norm": 0.9909055526988179,
      "learning_rate": 0.0001486202360854843,
      "loss": 12.3284,
      "step": 8194
    },
    {
      "epoch": 1.4330054644808743,
      "grad_norm": 0.7732890385396826,
      "learning_rate": 0.00014860786116056835,
      "loss": 12.4222,
      "step": 8195
    },
    {
      "epoch": 1.4331803278688524,
      "grad_norm": 1.0238342328597336,
      "learning_rate": 0.00014859548526091553,
      "loss": 12.3614,
      "step": 8196
    },
    {
      "epoch": 1.4333551912568305,
      "grad_norm": 0.8158915271363458,
      "learning_rate": 0.00014858310838677404,
      "loss": 12.6612,
      "step": 8197
    },
    {
      "epoch": 1.4335300546448089,
      "grad_norm": 0.9151266251925323,
      "learning_rate": 0.00014857073053839206,
      "loss": 12.2024,
      "step": 8198
    },
    {
      "epoch": 1.4337049180327868,
      "grad_norm": 0.7950889568165446,
      "learning_rate": 0.00014855835171601777,
      "loss": 12.6239,
      "step": 8199
    },
    {
      "epoch": 1.433879781420765,
      "grad_norm": 0.8039991869890486,
      "learning_rate": 0.00014854597191989947,
      "loss": 12.2163,
      "step": 8200
    },
    {
      "epoch": 1.4340546448087432,
      "grad_norm": 0.8647891434727425,
      "learning_rate": 0.00014853359115028536,
      "loss": 12.6058,
      "step": 8201
    },
    {
      "epoch": 1.4342295081967213,
      "grad_norm": 0.7831302040901646,
      "learning_rate": 0.0001485212094074237,
      "loss": 12.5579,
      "step": 8202
    },
    {
      "epoch": 1.4344043715846995,
      "grad_norm": 0.8969201899142887,
      "learning_rate": 0.00014850882669156286,
      "loss": 11.8885,
      "step": 8203
    },
    {
      "epoch": 1.4345792349726776,
      "grad_norm": 0.6983012335164351,
      "learning_rate": 0.0001484964430029511,
      "loss": 12.3793,
      "step": 8204
    },
    {
      "epoch": 1.4347540983606557,
      "grad_norm": 0.9174643167155834,
      "learning_rate": 0.00014848405834183673,
      "loss": 12.3089,
      "step": 8205
    },
    {
      "epoch": 1.4349289617486338,
      "grad_norm": 0.7609084600231926,
      "learning_rate": 0.00014847167270846812,
      "loss": 12.3053,
      "step": 8206
    },
    {
      "epoch": 1.435103825136612,
      "grad_norm": 0.8516097768538103,
      "learning_rate": 0.0001484592861030937,
      "loss": 12.529,
      "step": 8207
    },
    {
      "epoch": 1.43527868852459,
      "grad_norm": 0.8270772032756416,
      "learning_rate": 0.00014844689852596172,
      "loss": 12.6454,
      "step": 8208
    },
    {
      "epoch": 1.4354535519125684,
      "grad_norm": 0.6989261106370743,
      "learning_rate": 0.00014843450997732073,
      "loss": 12.5091,
      "step": 8209
    },
    {
      "epoch": 1.4356284153005465,
      "grad_norm": 0.7930979388579509,
      "learning_rate": 0.00014842212045741907,
      "loss": 12.3082,
      "step": 8210
    },
    {
      "epoch": 1.4358032786885246,
      "grad_norm": 0.6681908341325832,
      "learning_rate": 0.00014840972996650525,
      "loss": 12.4971,
      "step": 8211
    },
    {
      "epoch": 1.4359781420765028,
      "grad_norm": 0.7824998600211968,
      "learning_rate": 0.0001483973385048277,
      "loss": 12.2807,
      "step": 8212
    },
    {
      "epoch": 1.4361530054644809,
      "grad_norm": 0.6982093814098702,
      "learning_rate": 0.0001483849460726349,
      "loss": 12.4772,
      "step": 8213
    },
    {
      "epoch": 1.436327868852459,
      "grad_norm": 0.7687220605661028,
      "learning_rate": 0.00014837255267017538,
      "loss": 12.4588,
      "step": 8214
    },
    {
      "epoch": 1.4365027322404371,
      "grad_norm": 0.6782995270766029,
      "learning_rate": 0.00014836015829769762,
      "loss": 12.3071,
      "step": 8215
    },
    {
      "epoch": 1.4366775956284152,
      "grad_norm": 0.7264749447468372,
      "learning_rate": 0.00014834776295545026,
      "loss": 12.4866,
      "step": 8216
    },
    {
      "epoch": 1.4368524590163934,
      "grad_norm": 0.7827470870282798,
      "learning_rate": 0.00014833536664368176,
      "loss": 12.3551,
      "step": 8217
    },
    {
      "epoch": 1.4370273224043717,
      "grad_norm": 0.7349361481039859,
      "learning_rate": 0.0001483229693626408,
      "loss": 12.5024,
      "step": 8218
    },
    {
      "epoch": 1.4372021857923496,
      "grad_norm": 0.7729892176904856,
      "learning_rate": 0.0001483105711125759,
      "loss": 12.0445,
      "step": 8219
    },
    {
      "epoch": 1.437377049180328,
      "grad_norm": 0.7776969116849813,
      "learning_rate": 0.0001482981718937357,
      "loss": 12.5322,
      "step": 8220
    },
    {
      "epoch": 1.437551912568306,
      "grad_norm": 0.7527859451254313,
      "learning_rate": 0.0001482857717063689,
      "loss": 12.4328,
      "step": 8221
    },
    {
      "epoch": 1.4377267759562842,
      "grad_norm": 0.6325530117814507,
      "learning_rate": 0.00014827337055072406,
      "loss": 12.3983,
      "step": 8222
    },
    {
      "epoch": 1.4379016393442623,
      "grad_norm": 0.7861873887996337,
      "learning_rate": 0.00014826096842705,
      "loss": 12.3588,
      "step": 8223
    },
    {
      "epoch": 1.4380765027322404,
      "grad_norm": 0.6925121442109003,
      "learning_rate": 0.00014824856533559525,
      "loss": 12.3928,
      "step": 8224
    },
    {
      "epoch": 1.4382513661202185,
      "grad_norm": 0.874139271927852,
      "learning_rate": 0.0001482361612766087,
      "loss": 12.3544,
      "step": 8225
    },
    {
      "epoch": 1.4384262295081967,
      "grad_norm": 0.6688349592418539,
      "learning_rate": 0.000148223756250339,
      "loss": 12.5293,
      "step": 8226
    },
    {
      "epoch": 1.438601092896175,
      "grad_norm": 0.8292430863969547,
      "learning_rate": 0.0001482113502570349,
      "loss": 12.2771,
      "step": 8227
    },
    {
      "epoch": 1.438775956284153,
      "grad_norm": 0.7608608630144043,
      "learning_rate": 0.00014819894329694521,
      "loss": 12.6216,
      "step": 8228
    },
    {
      "epoch": 1.4389508196721312,
      "grad_norm": 0.7020280844862082,
      "learning_rate": 0.00014818653537031868,
      "loss": 12.2468,
      "step": 8229
    },
    {
      "epoch": 1.4391256830601094,
      "grad_norm": 0.8997674193854259,
      "learning_rate": 0.00014817412647740423,
      "loss": 12.5267,
      "step": 8230
    },
    {
      "epoch": 1.4393005464480875,
      "grad_norm": 0.7224541299464903,
      "learning_rate": 0.0001481617166184506,
      "loss": 12.3174,
      "step": 8231
    },
    {
      "epoch": 1.4394754098360656,
      "grad_norm": 0.8425666772679484,
      "learning_rate": 0.00014814930579370666,
      "loss": 12.4836,
      "step": 8232
    },
    {
      "epoch": 1.4396502732240437,
      "grad_norm": 0.7501628889636928,
      "learning_rate": 0.00014813689400342133,
      "loss": 12.7054,
      "step": 8233
    },
    {
      "epoch": 1.4398251366120218,
      "grad_norm": 0.7552270435699706,
      "learning_rate": 0.00014812448124784344,
      "loss": 12.6052,
      "step": 8234
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.698580252808284,
      "learning_rate": 0.00014811206752722195,
      "loss": 12.0959,
      "step": 8235
    },
    {
      "epoch": 1.440174863387978,
      "grad_norm": 0.7846894545956976,
      "learning_rate": 0.0001480996528418058,
      "loss": 12.295,
      "step": 8236
    },
    {
      "epoch": 1.4403497267759562,
      "grad_norm": 0.7694521407126753,
      "learning_rate": 0.00014808723719184392,
      "loss": 12.4913,
      "step": 8237
    },
    {
      "epoch": 1.4405245901639345,
      "grad_norm": 0.7782376707284696,
      "learning_rate": 0.00014807482057758528,
      "loss": 12.4494,
      "step": 8238
    },
    {
      "epoch": 1.4406994535519126,
      "grad_norm": 0.8425619581754218,
      "learning_rate": 0.0001480624029992789,
      "loss": 12.4952,
      "step": 8239
    },
    {
      "epoch": 1.4408743169398908,
      "grad_norm": 0.7569841613879533,
      "learning_rate": 0.00014804998445717372,
      "loss": 12.4137,
      "step": 8240
    },
    {
      "epoch": 1.4410491803278689,
      "grad_norm": 0.7567307799058884,
      "learning_rate": 0.00014803756495151887,
      "loss": 12.3707,
      "step": 8241
    },
    {
      "epoch": 1.441224043715847,
      "grad_norm": 0.7234110080640577,
      "learning_rate": 0.00014802514448256332,
      "loss": 12.2236,
      "step": 8242
    },
    {
      "epoch": 1.4413989071038251,
      "grad_norm": 0.6999213472022237,
      "learning_rate": 0.0001480127230505562,
      "loss": 12.183,
      "step": 8243
    },
    {
      "epoch": 1.4415737704918032,
      "grad_norm": 0.7700994957418291,
      "learning_rate": 0.0001480003006557465,
      "loss": 12.2783,
      "step": 8244
    },
    {
      "epoch": 1.4417486338797814,
      "grad_norm": 0.6892864474802852,
      "learning_rate": 0.00014798787729838344,
      "loss": 12.3869,
      "step": 8245
    },
    {
      "epoch": 1.4419234972677595,
      "grad_norm": 0.6881090774914712,
      "learning_rate": 0.0001479754529787161,
      "loss": 12.4815,
      "step": 8246
    },
    {
      "epoch": 1.4420983606557378,
      "grad_norm": 0.7234759682505821,
      "learning_rate": 0.0001479630276969936,
      "loss": 12.3589,
      "step": 8247
    },
    {
      "epoch": 1.4422732240437157,
      "grad_norm": 0.8601882983547631,
      "learning_rate": 0.00014795060145346512,
      "loss": 12.2778,
      "step": 8248
    },
    {
      "epoch": 1.442448087431694,
      "grad_norm": 0.7140055173383172,
      "learning_rate": 0.00014793817424837988,
      "loss": 12.5546,
      "step": 8249
    },
    {
      "epoch": 1.4426229508196722,
      "grad_norm": 0.7637871307119863,
      "learning_rate": 0.00014792574608198705,
      "loss": 12.5254,
      "step": 8250
    },
    {
      "epoch": 1.4427978142076503,
      "grad_norm": 0.7131490520156605,
      "learning_rate": 0.00014791331695453586,
      "loss": 12.6127,
      "step": 8251
    },
    {
      "epoch": 1.4429726775956284,
      "grad_norm": 0.7436542480318943,
      "learning_rate": 0.00014790088686627556,
      "loss": 12.4898,
      "step": 8252
    },
    {
      "epoch": 1.4431475409836065,
      "grad_norm": 0.7234134564622844,
      "learning_rate": 0.00014788845581745539,
      "loss": 12.53,
      "step": 8253
    },
    {
      "epoch": 1.4433224043715847,
      "grad_norm": 0.8096287384829127,
      "learning_rate": 0.00014787602380832463,
      "loss": 12.6796,
      "step": 8254
    },
    {
      "epoch": 1.4434972677595628,
      "grad_norm": 0.758063305087995,
      "learning_rate": 0.00014786359083913267,
      "loss": 12.1008,
      "step": 8255
    },
    {
      "epoch": 1.4436721311475411,
      "grad_norm": 0.7633822729145145,
      "learning_rate": 0.00014785115691012864,
      "loss": 12.3781,
      "step": 8256
    },
    {
      "epoch": 1.443846994535519,
      "grad_norm": 0.7828996718495551,
      "learning_rate": 0.00014783872202156208,
      "loss": 12.384,
      "step": 8257
    },
    {
      "epoch": 1.4440218579234974,
      "grad_norm": 0.7623552000424735,
      "learning_rate": 0.0001478262861736822,
      "loss": 12.3421,
      "step": 8258
    },
    {
      "epoch": 1.4441967213114755,
      "grad_norm": 0.722511117344341,
      "learning_rate": 0.00014781384936673848,
      "loss": 12.6799,
      "step": 8259
    },
    {
      "epoch": 1.4443715846994536,
      "grad_norm": 0.733154408218735,
      "learning_rate": 0.00014780141160098024,
      "loss": 12.5815,
      "step": 8260
    },
    {
      "epoch": 1.4445464480874317,
      "grad_norm": 0.7350404891378168,
      "learning_rate": 0.00014778897287665696,
      "loss": 12.2891,
      "step": 8261
    },
    {
      "epoch": 1.4447213114754098,
      "grad_norm": 0.729626855407926,
      "learning_rate": 0.00014777653319401803,
      "loss": 12.3156,
      "step": 8262
    },
    {
      "epoch": 1.444896174863388,
      "grad_norm": 0.6978112540481872,
      "learning_rate": 0.0001477640925533129,
      "loss": 12.4454,
      "step": 8263
    },
    {
      "epoch": 1.445071038251366,
      "grad_norm": 0.6896630528768763,
      "learning_rate": 0.00014775165095479108,
      "loss": 12.4555,
      "step": 8264
    },
    {
      "epoch": 1.4452459016393442,
      "grad_norm": 0.6963343791898244,
      "learning_rate": 0.000147739208398702,
      "loss": 12.6111,
      "step": 8265
    },
    {
      "epoch": 1.4454207650273223,
      "grad_norm": 0.6965605470769377,
      "learning_rate": 0.00014772676488529528,
      "loss": 12.1889,
      "step": 8266
    },
    {
      "epoch": 1.4455956284153006,
      "grad_norm": 0.8041481432572245,
      "learning_rate": 0.00014771432041482037,
      "loss": 12.0528,
      "step": 8267
    },
    {
      "epoch": 1.4457704918032788,
      "grad_norm": 0.7263365792997305,
      "learning_rate": 0.0001477018749875268,
      "loss": 12.4754,
      "step": 8268
    },
    {
      "epoch": 1.4459453551912569,
      "grad_norm": 0.6905311248414553,
      "learning_rate": 0.0001476894286036642,
      "loss": 12.4074,
      "step": 8269
    },
    {
      "epoch": 1.446120218579235,
      "grad_norm": 0.6631718845562201,
      "learning_rate": 0.00014767698126348208,
      "loss": 12.2036,
      "step": 8270
    },
    {
      "epoch": 1.4462950819672131,
      "grad_norm": 0.7021721862098379,
      "learning_rate": 0.00014766453296723015,
      "loss": 12.4271,
      "step": 8271
    },
    {
      "epoch": 1.4464699453551912,
      "grad_norm": 0.7684186790846318,
      "learning_rate": 0.00014765208371515795,
      "loss": 12.443,
      "step": 8272
    },
    {
      "epoch": 1.4466448087431694,
      "grad_norm": 0.6331124622024055,
      "learning_rate": 0.0001476396335075152,
      "loss": 12.4752,
      "step": 8273
    },
    {
      "epoch": 1.4468196721311475,
      "grad_norm": 0.8272469912885155,
      "learning_rate": 0.00014762718234455151,
      "loss": 12.6539,
      "step": 8274
    },
    {
      "epoch": 1.4469945355191256,
      "grad_norm": 0.729544054500534,
      "learning_rate": 0.00014761473022651658,
      "loss": 12.6427,
      "step": 8275
    },
    {
      "epoch": 1.447169398907104,
      "grad_norm": 0.919127814231035,
      "learning_rate": 0.0001476022771536601,
      "loss": 12.485,
      "step": 8276
    },
    {
      "epoch": 1.4473442622950818,
      "grad_norm": 0.7809717106272073,
      "learning_rate": 0.00014758982312623178,
      "loss": 12.2059,
      "step": 8277
    },
    {
      "epoch": 1.4475191256830602,
      "grad_norm": 0.8955662467158664,
      "learning_rate": 0.00014757736814448145,
      "loss": 12.4335,
      "step": 8278
    },
    {
      "epoch": 1.4476939890710383,
      "grad_norm": 0.8598953841340936,
      "learning_rate": 0.00014756491220865877,
      "loss": 12.7427,
      "step": 8279
    },
    {
      "epoch": 1.4478688524590164,
      "grad_norm": 0.823814663474808,
      "learning_rate": 0.00014755245531901357,
      "loss": 12.2444,
      "step": 8280
    },
    {
      "epoch": 1.4480437158469945,
      "grad_norm": 0.8399224199188301,
      "learning_rate": 0.00014753999747579562,
      "loss": 12.4877,
      "step": 8281
    },
    {
      "epoch": 1.4482185792349727,
      "grad_norm": 0.7772339999357233,
      "learning_rate": 0.00014752753867925474,
      "loss": 12.4591,
      "step": 8282
    },
    {
      "epoch": 1.4483934426229508,
      "grad_norm": 0.7687283394921194,
      "learning_rate": 0.00014751507892964082,
      "loss": 12.4175,
      "step": 8283
    },
    {
      "epoch": 1.448568306010929,
      "grad_norm": 0.7111972842784601,
      "learning_rate": 0.00014750261822720365,
      "loss": 12.2578,
      "step": 8284
    },
    {
      "epoch": 1.4487431693989072,
      "grad_norm": 0.827847049214714,
      "learning_rate": 0.00014749015657219313,
      "loss": 12.3048,
      "step": 8285
    },
    {
      "epoch": 1.4489180327868851,
      "grad_norm": 0.7934601379356692,
      "learning_rate": 0.00014747769396485918,
      "loss": 12.5638,
      "step": 8286
    },
    {
      "epoch": 1.4490928961748635,
      "grad_norm": 0.8478458475631252,
      "learning_rate": 0.0001474652304054517,
      "loss": 12.5503,
      "step": 8287
    },
    {
      "epoch": 1.4492677595628416,
      "grad_norm": 0.6853497227564483,
      "learning_rate": 0.00014745276589422056,
      "loss": 12.3623,
      "step": 8288
    },
    {
      "epoch": 1.4494426229508197,
      "grad_norm": 0.7361874575262713,
      "learning_rate": 0.0001474403004314158,
      "loss": 12.5224,
      "step": 8289
    },
    {
      "epoch": 1.4496174863387978,
      "grad_norm": 0.8486471705924656,
      "learning_rate": 0.00014742783401728736,
      "loss": 12.5152,
      "step": 8290
    },
    {
      "epoch": 1.449792349726776,
      "grad_norm": 0.8745761717712361,
      "learning_rate": 0.0001474153666520852,
      "loss": 12.4179,
      "step": 8291
    },
    {
      "epoch": 1.449967213114754,
      "grad_norm": 0.7460091985577004,
      "learning_rate": 0.00014740289833605939,
      "loss": 12.6919,
      "step": 8292
    },
    {
      "epoch": 1.4501420765027322,
      "grad_norm": 0.728209500593976,
      "learning_rate": 0.0001473904290694599,
      "loss": 12.573,
      "step": 8293
    },
    {
      "epoch": 1.4503169398907103,
      "grad_norm": 0.8983321955336183,
      "learning_rate": 0.0001473779588525368,
      "loss": 12.4086,
      "step": 8294
    },
    {
      "epoch": 1.4504918032786884,
      "grad_norm": 0.7683343360309531,
      "learning_rate": 0.00014736548768554014,
      "loss": 12.4749,
      "step": 8295
    },
    {
      "epoch": 1.4506666666666668,
      "grad_norm": 1.0264744927122358,
      "learning_rate": 0.00014735301556872006,
      "loss": 12.2779,
      "step": 8296
    },
    {
      "epoch": 1.450841530054645,
      "grad_norm": 0.6290419395766558,
      "learning_rate": 0.0001473405425023266,
      "loss": 12.3858,
      "step": 8297
    },
    {
      "epoch": 1.451016393442623,
      "grad_norm": 0.9416962790035341,
      "learning_rate": 0.00014732806848660995,
      "loss": 12.5507,
      "step": 8298
    },
    {
      "epoch": 1.4511912568306011,
      "grad_norm": 0.6941550514547838,
      "learning_rate": 0.00014731559352182015,
      "loss": 12.427,
      "step": 8299
    },
    {
      "epoch": 1.4513661202185792,
      "grad_norm": 0.9281715159266692,
      "learning_rate": 0.00014730311760820743,
      "loss": 12.59,
      "step": 8300
    },
    {
      "epoch": 1.4515409836065574,
      "grad_norm": 0.7030399926264218,
      "learning_rate": 0.00014729064074602198,
      "loss": 12.596,
      "step": 8301
    },
    {
      "epoch": 1.4517158469945355,
      "grad_norm": 0.8491863882896221,
      "learning_rate": 0.00014727816293551397,
      "loss": 12.48,
      "step": 8302
    },
    {
      "epoch": 1.4518907103825136,
      "grad_norm": 0.7396896778418226,
      "learning_rate": 0.00014726568417693366,
      "loss": 12.6715,
      "step": 8303
    },
    {
      "epoch": 1.4520655737704917,
      "grad_norm": 0.7740638835746779,
      "learning_rate": 0.0001472532044705312,
      "loss": 12.4455,
      "step": 8304
    },
    {
      "epoch": 1.45224043715847,
      "grad_norm": 0.8665677193979682,
      "learning_rate": 0.000147240723816557,
      "loss": 12.5074,
      "step": 8305
    },
    {
      "epoch": 1.452415300546448,
      "grad_norm": 0.8403240529984426,
      "learning_rate": 0.00014722824221526115,
      "loss": 12.5067,
      "step": 8306
    },
    {
      "epoch": 1.4525901639344263,
      "grad_norm": 0.8457062372112437,
      "learning_rate": 0.00014721575966689408,
      "loss": 12.5371,
      "step": 8307
    },
    {
      "epoch": 1.4527650273224044,
      "grad_norm": 0.7673214176884281,
      "learning_rate": 0.00014720327617170602,
      "loss": 12.6396,
      "step": 8308
    },
    {
      "epoch": 1.4529398907103825,
      "grad_norm": 0.9049751596999962,
      "learning_rate": 0.00014719079172994737,
      "loss": 12.547,
      "step": 8309
    },
    {
      "epoch": 1.4531147540983607,
      "grad_norm": 0.7362811204032785,
      "learning_rate": 0.00014717830634186844,
      "loss": 12.4115,
      "step": 8310
    },
    {
      "epoch": 1.4532896174863388,
      "grad_norm": 0.7065553136759806,
      "learning_rate": 0.0001471658200077196,
      "loss": 12.5046,
      "step": 8311
    },
    {
      "epoch": 1.453464480874317,
      "grad_norm": 0.8889024564752763,
      "learning_rate": 0.00014715333272775127,
      "loss": 12.3026,
      "step": 8312
    },
    {
      "epoch": 1.453639344262295,
      "grad_norm": 0.6879562877297328,
      "learning_rate": 0.00014714084450221382,
      "loss": 12.1966,
      "step": 8313
    },
    {
      "epoch": 1.4538142076502734,
      "grad_norm": 0.9543818124794075,
      "learning_rate": 0.00014712835533135774,
      "loss": 12.3018,
      "step": 8314
    },
    {
      "epoch": 1.4539890710382513,
      "grad_norm": 0.7035326348906781,
      "learning_rate": 0.0001471158652154334,
      "loss": 12.2097,
      "step": 8315
    },
    {
      "epoch": 1.4541639344262296,
      "grad_norm": 1.0861878613998797,
      "learning_rate": 0.00014710337415469127,
      "loss": 12.6539,
      "step": 8316
    },
    {
      "epoch": 1.4543387978142077,
      "grad_norm": 0.6768083939397704,
      "learning_rate": 0.00014709088214938192,
      "loss": 12.4326,
      "step": 8317
    },
    {
      "epoch": 1.4545136612021858,
      "grad_norm": 1.033361636131365,
      "learning_rate": 0.00014707838919975573,
      "loss": 12.4683,
      "step": 8318
    },
    {
      "epoch": 1.454688524590164,
      "grad_norm": 0.7365535449213046,
      "learning_rate": 0.00014706589530606335,
      "loss": 12.5441,
      "step": 8319
    },
    {
      "epoch": 1.454863387978142,
      "grad_norm": 0.8755044962624231,
      "learning_rate": 0.00014705340046855523,
      "loss": 12.3053,
      "step": 8320
    },
    {
      "epoch": 1.4550382513661202,
      "grad_norm": 0.7582849102402618,
      "learning_rate": 0.00014704090468748197,
      "loss": 12.4878,
      "step": 8321
    },
    {
      "epoch": 1.4552131147540983,
      "grad_norm": 0.7714164822832267,
      "learning_rate": 0.0001470284079630941,
      "loss": 12.4536,
      "step": 8322
    },
    {
      "epoch": 1.4553879781420764,
      "grad_norm": 0.8351427748614799,
      "learning_rate": 0.0001470159102956423,
      "loss": 12.5284,
      "step": 8323
    },
    {
      "epoch": 1.4555628415300546,
      "grad_norm": 0.6944251184908053,
      "learning_rate": 0.00014700341168537713,
      "loss": 12.3948,
      "step": 8324
    },
    {
      "epoch": 1.455737704918033,
      "grad_norm": 0.6855312212416179,
      "learning_rate": 0.00014699091213254918,
      "loss": 12.4125,
      "step": 8325
    },
    {
      "epoch": 1.4559125683060108,
      "grad_norm": 0.7126853291178151,
      "learning_rate": 0.00014697841163740925,
      "loss": 12.409,
      "step": 8326
    },
    {
      "epoch": 1.4560874316939891,
      "grad_norm": 0.7119608144873739,
      "learning_rate": 0.00014696591020020786,
      "loss": 12.3649,
      "step": 8327
    },
    {
      "epoch": 1.4562622950819673,
      "grad_norm": 0.7164265998423649,
      "learning_rate": 0.0001469534078211958,
      "loss": 12.2829,
      "step": 8328
    },
    {
      "epoch": 1.4564371584699454,
      "grad_norm": 0.7060183139508998,
      "learning_rate": 0.00014694090450062373,
      "loss": 12.4869,
      "step": 8329
    },
    {
      "epoch": 1.4566120218579235,
      "grad_norm": 0.7840820504852246,
      "learning_rate": 0.0001469284002387424,
      "loss": 12.6099,
      "step": 8330
    },
    {
      "epoch": 1.4567868852459016,
      "grad_norm": 0.7298730224468889,
      "learning_rate": 0.00014691589503580257,
      "loss": 12.5569,
      "step": 8331
    },
    {
      "epoch": 1.4569617486338797,
      "grad_norm": 0.7070838265338791,
      "learning_rate": 0.00014690338889205497,
      "loss": 12.5063,
      "step": 8332
    },
    {
      "epoch": 1.4571366120218578,
      "grad_norm": 0.7123657341873079,
      "learning_rate": 0.0001468908818077504,
      "loss": 12.1194,
      "step": 8333
    },
    {
      "epoch": 1.4573114754098362,
      "grad_norm": 0.7588125188953142,
      "learning_rate": 0.00014687837378313972,
      "loss": 12.4777,
      "step": 8334
    },
    {
      "epoch": 1.457486338797814,
      "grad_norm": 0.707201039625655,
      "learning_rate": 0.00014686586481847368,
      "loss": 12.2261,
      "step": 8335
    },
    {
      "epoch": 1.4576612021857924,
      "grad_norm": 0.75509886307921,
      "learning_rate": 0.00014685335491400317,
      "loss": 12.5249,
      "step": 8336
    },
    {
      "epoch": 1.4578360655737705,
      "grad_norm": 0.7181688465889018,
      "learning_rate": 0.00014684084406997903,
      "loss": 12.4454,
      "step": 8337
    },
    {
      "epoch": 1.4580109289617487,
      "grad_norm": 0.7439740758507227,
      "learning_rate": 0.00014682833228665213,
      "loss": 12.0971,
      "step": 8338
    },
    {
      "epoch": 1.4581857923497268,
      "grad_norm": 0.6347144361732873,
      "learning_rate": 0.0001468158195642734,
      "loss": 12.4745,
      "step": 8339
    },
    {
      "epoch": 1.458360655737705,
      "grad_norm": 0.6635098091546797,
      "learning_rate": 0.00014680330590309375,
      "loss": 12.2751,
      "step": 8340
    },
    {
      "epoch": 1.458535519125683,
      "grad_norm": 0.7717229079564649,
      "learning_rate": 0.0001467907913033641,
      "loss": 12.5346,
      "step": 8341
    },
    {
      "epoch": 1.4587103825136611,
      "grad_norm": 0.7671118519400363,
      "learning_rate": 0.00014677827576533542,
      "loss": 12.6308,
      "step": 8342
    },
    {
      "epoch": 1.4588852459016395,
      "grad_norm": 0.7311138484913283,
      "learning_rate": 0.00014676575928925867,
      "loss": 12.5011,
      "step": 8343
    },
    {
      "epoch": 1.4590601092896174,
      "grad_norm": 0.6762989025350702,
      "learning_rate": 0.00014675324187538488,
      "loss": 12.5388,
      "step": 8344
    },
    {
      "epoch": 1.4592349726775957,
      "grad_norm": 0.85230811438951,
      "learning_rate": 0.00014674072352396505,
      "loss": 12.3893,
      "step": 8345
    },
    {
      "epoch": 1.4594098360655738,
      "grad_norm": 0.6592354761337909,
      "learning_rate": 0.0001467282042352502,
      "loss": 12.6115,
      "step": 8346
    },
    {
      "epoch": 1.459584699453552,
      "grad_norm": 0.7067799638634326,
      "learning_rate": 0.00014671568400949133,
      "loss": 12.2512,
      "step": 8347
    },
    {
      "epoch": 1.45975956284153,
      "grad_norm": 0.7479758614128809,
      "learning_rate": 0.00014670316284693958,
      "loss": 12.2432,
      "step": 8348
    },
    {
      "epoch": 1.4599344262295082,
      "grad_norm": 0.8226270513868837,
      "learning_rate": 0.00014669064074784605,
      "loss": 12.3023,
      "step": 8349
    },
    {
      "epoch": 1.4601092896174863,
      "grad_norm": 0.8278098851941172,
      "learning_rate": 0.0001466781177124618,
      "loss": 12.377,
      "step": 8350
    },
    {
      "epoch": 1.4602841530054644,
      "grad_norm": 0.8482648771122984,
      "learning_rate": 0.00014666559374103798,
      "loss": 12.2437,
      "step": 8351
    },
    {
      "epoch": 1.4604590163934426,
      "grad_norm": 0.7707926840284167,
      "learning_rate": 0.0001466530688338257,
      "loss": 12.4251,
      "step": 8352
    },
    {
      "epoch": 1.4606338797814207,
      "grad_norm": 0.7015745913518009,
      "learning_rate": 0.00014664054299107614,
      "loss": 12.3667,
      "step": 8353
    },
    {
      "epoch": 1.460808743169399,
      "grad_norm": 0.8334781991384903,
      "learning_rate": 0.00014662801621304048,
      "loss": 12.4988,
      "step": 8354
    },
    {
      "epoch": 1.460983606557377,
      "grad_norm": 0.7437207611228253,
      "learning_rate": 0.00014661548849996997,
      "loss": 12.3882,
      "step": 8355
    },
    {
      "epoch": 1.4611584699453553,
      "grad_norm": 0.8222081907379601,
      "learning_rate": 0.00014660295985211575,
      "loss": 12.5325,
      "step": 8356
    },
    {
      "epoch": 1.4613333333333334,
      "grad_norm": 0.7023096055352556,
      "learning_rate": 0.0001465904302697291,
      "loss": 12.4418,
      "step": 8357
    },
    {
      "epoch": 1.4615081967213115,
      "grad_norm": 0.7474207903679628,
      "learning_rate": 0.0001465778997530613,
      "loss": 12.6346,
      "step": 8358
    },
    {
      "epoch": 1.4616830601092896,
      "grad_norm": 0.8548333750926215,
      "learning_rate": 0.00014656536830236357,
      "loss": 12.5503,
      "step": 8359
    },
    {
      "epoch": 1.4618579234972677,
      "grad_norm": 0.7633928964235028,
      "learning_rate": 0.00014655283591788725,
      "loss": 12.555,
      "step": 8360
    },
    {
      "epoch": 1.4620327868852458,
      "grad_norm": 0.9574208884432602,
      "learning_rate": 0.00014654030259988362,
      "loss": 12.3652,
      "step": 8361
    },
    {
      "epoch": 1.462207650273224,
      "grad_norm": 0.7267499431913299,
      "learning_rate": 0.00014652776834860404,
      "loss": 12.5323,
      "step": 8362
    },
    {
      "epoch": 1.4623825136612023,
      "grad_norm": 0.7894660291511071,
      "learning_rate": 0.00014651523316429983,
      "loss": 12.549,
      "step": 8363
    },
    {
      "epoch": 1.4625573770491802,
      "grad_norm": 0.6626918204184332,
      "learning_rate": 0.00014650269704722237,
      "loss": 12.5477,
      "step": 8364
    },
    {
      "epoch": 1.4627322404371585,
      "grad_norm": 0.7596932892616057,
      "learning_rate": 0.00014649015999762305,
      "loss": 12.1509,
      "step": 8365
    },
    {
      "epoch": 1.4629071038251367,
      "grad_norm": 0.7856435168565991,
      "learning_rate": 0.00014647762201575328,
      "loss": 12.5058,
      "step": 8366
    },
    {
      "epoch": 1.4630819672131148,
      "grad_norm": 0.7754692088294235,
      "learning_rate": 0.0001464650831018645,
      "loss": 12.4179,
      "step": 8367
    },
    {
      "epoch": 1.463256830601093,
      "grad_norm": 0.9726815860769019,
      "learning_rate": 0.00014645254325620813,
      "loss": 12.5244,
      "step": 8368
    },
    {
      "epoch": 1.463431693989071,
      "grad_norm": 0.647818372300074,
      "learning_rate": 0.00014644000247903565,
      "loss": 12.2839,
      "step": 8369
    },
    {
      "epoch": 1.4636065573770491,
      "grad_norm": 0.9411435460488029,
      "learning_rate": 0.00014642746077059851,
      "loss": 12.5514,
      "step": 8370
    },
    {
      "epoch": 1.4637814207650273,
      "grad_norm": 0.7586663988951369,
      "learning_rate": 0.00014641491813114824,
      "loss": 12.2946,
      "step": 8371
    },
    {
      "epoch": 1.4639562841530054,
      "grad_norm": 0.9042824595265342,
      "learning_rate": 0.00014640237456093634,
      "loss": 12.5065,
      "step": 8372
    },
    {
      "epoch": 1.4641311475409835,
      "grad_norm": 0.8740561456975936,
      "learning_rate": 0.00014638983006021438,
      "loss": 12.3784,
      "step": 8373
    },
    {
      "epoch": 1.4643060109289618,
      "grad_norm": 0.8513701930196689,
      "learning_rate": 0.0001463772846292339,
      "loss": 12.1001,
      "step": 8374
    },
    {
      "epoch": 1.46448087431694,
      "grad_norm": 0.9409292589170658,
      "learning_rate": 0.00014636473826824643,
      "loss": 12.4679,
      "step": 8375
    },
    {
      "epoch": 1.464655737704918,
      "grad_norm": 0.7251520411106496,
      "learning_rate": 0.00014635219097750362,
      "loss": 12.6383,
      "step": 8376
    },
    {
      "epoch": 1.4648306010928962,
      "grad_norm": 0.8734707640890678,
      "learning_rate": 0.00014633964275725706,
      "loss": 12.3763,
      "step": 8377
    },
    {
      "epoch": 1.4650054644808743,
      "grad_norm": 0.7123343581459424,
      "learning_rate": 0.00014632709360775837,
      "loss": 12.24,
      "step": 8378
    },
    {
      "epoch": 1.4651803278688524,
      "grad_norm": 0.8912753659942193,
      "learning_rate": 0.0001463145435292592,
      "loss": 12.3715,
      "step": 8379
    },
    {
      "epoch": 1.4653551912568306,
      "grad_norm": 0.6845664740561612,
      "learning_rate": 0.00014630199252201124,
      "loss": 12.5022,
      "step": 8380
    },
    {
      "epoch": 1.4655300546448087,
      "grad_norm": 0.6855700979151715,
      "learning_rate": 0.00014628944058626617,
      "loss": 12.1763,
      "step": 8381
    },
    {
      "epoch": 1.4657049180327868,
      "grad_norm": 0.7805391482675874,
      "learning_rate": 0.0001462768877222757,
      "loss": 12.346,
      "step": 8382
    },
    {
      "epoch": 1.4658797814207651,
      "grad_norm": 0.7305057534856904,
      "learning_rate": 0.00014626433393029152,
      "loss": 12.3499,
      "step": 8383
    },
    {
      "epoch": 1.466054644808743,
      "grad_norm": 0.7502567560114694,
      "learning_rate": 0.00014625177921056542,
      "loss": 12.2601,
      "step": 8384
    },
    {
      "epoch": 1.4662295081967214,
      "grad_norm": 0.7034775283023209,
      "learning_rate": 0.0001462392235633491,
      "loss": 12.5706,
      "step": 8385
    },
    {
      "epoch": 1.4664043715846995,
      "grad_norm": 0.8291907484238147,
      "learning_rate": 0.0001462266669888944,
      "loss": 12.5639,
      "step": 8386
    },
    {
      "epoch": 1.4665792349726776,
      "grad_norm": 0.7089410762712086,
      "learning_rate": 0.0001462141094874531,
      "loss": 12.3944,
      "step": 8387
    },
    {
      "epoch": 1.4667540983606557,
      "grad_norm": 0.711485108265513,
      "learning_rate": 0.00014620155105927698,
      "loss": 12.2555,
      "step": 8388
    },
    {
      "epoch": 1.4669289617486339,
      "grad_norm": 0.7486369896335061,
      "learning_rate": 0.00014618899170461792,
      "loss": 12.4432,
      "step": 8389
    },
    {
      "epoch": 1.467103825136612,
      "grad_norm": 0.7514970874011148,
      "learning_rate": 0.00014617643142372775,
      "loss": 12.5052,
      "step": 8390
    },
    {
      "epoch": 1.46727868852459,
      "grad_norm": 0.7944904417385185,
      "learning_rate": 0.00014616387021685836,
      "loss": 12.2111,
      "step": 8391
    },
    {
      "epoch": 1.4674535519125684,
      "grad_norm": 0.6811746440901465,
      "learning_rate": 0.00014615130808426162,
      "loss": 12.4473,
      "step": 8392
    },
    {
      "epoch": 1.4676284153005463,
      "grad_norm": 0.8903527921365355,
      "learning_rate": 0.00014613874502618948,
      "loss": 12.5307,
      "step": 8393
    },
    {
      "epoch": 1.4678032786885247,
      "grad_norm": 0.7115486173202471,
      "learning_rate": 0.00014612618104289381,
      "loss": 12.603,
      "step": 8394
    },
    {
      "epoch": 1.4679781420765028,
      "grad_norm": 0.7278592715784614,
      "learning_rate": 0.00014611361613462657,
      "loss": 12.483,
      "step": 8395
    },
    {
      "epoch": 1.468153005464481,
      "grad_norm": 0.7225194440786538,
      "learning_rate": 0.00014610105030163977,
      "loss": 12.5235,
      "step": 8396
    },
    {
      "epoch": 1.468327868852459,
      "grad_norm": 0.7274804569032788,
      "learning_rate": 0.00014608848354418535,
      "loss": 12.335,
      "step": 8397
    },
    {
      "epoch": 1.4685027322404371,
      "grad_norm": 0.7276615901302586,
      "learning_rate": 0.0001460759158625153,
      "loss": 12.5886,
      "step": 8398
    },
    {
      "epoch": 1.4686775956284153,
      "grad_norm": 0.7081057751024884,
      "learning_rate": 0.0001460633472568817,
      "loss": 12.4177,
      "step": 8399
    },
    {
      "epoch": 1.4688524590163934,
      "grad_norm": 0.7418289967511255,
      "learning_rate": 0.00014605077772753656,
      "loss": 12.467,
      "step": 8400
    },
    {
      "epoch": 1.4690273224043715,
      "grad_norm": 1.1923379711711062,
      "learning_rate": 0.0001460382072747319,
      "loss": 12.2252,
      "step": 8401
    },
    {
      "epoch": 1.4692021857923496,
      "grad_norm": 0.7448873008704315,
      "learning_rate": 0.00014602563589871984,
      "loss": 12.4632,
      "step": 8402
    },
    {
      "epoch": 1.469377049180328,
      "grad_norm": 0.7609119615029327,
      "learning_rate": 0.00014601306359975248,
      "loss": 12.6912,
      "step": 8403
    },
    {
      "epoch": 1.469551912568306,
      "grad_norm": 0.8412125392774594,
      "learning_rate": 0.0001460004903780819,
      "loss": 12.392,
      "step": 8404
    },
    {
      "epoch": 1.4697267759562842,
      "grad_norm": 0.71170907327602,
      "learning_rate": 0.00014598791623396024,
      "loss": 12.4191,
      "step": 8405
    },
    {
      "epoch": 1.4699016393442623,
      "grad_norm": 0.6944833612944731,
      "learning_rate": 0.0001459753411676397,
      "loss": 12.5098,
      "step": 8406
    },
    {
      "epoch": 1.4700765027322404,
      "grad_norm": 0.8396363205295921,
      "learning_rate": 0.00014596276517937236,
      "loss": 12.6206,
      "step": 8407
    },
    {
      "epoch": 1.4702513661202186,
      "grad_norm": 0.8017109014501855,
      "learning_rate": 0.0001459501882694105,
      "loss": 12.551,
      "step": 8408
    },
    {
      "epoch": 1.4704262295081967,
      "grad_norm": 0.7421712001074767,
      "learning_rate": 0.00014593761043800622,
      "loss": 12.2021,
      "step": 8409
    },
    {
      "epoch": 1.4706010928961748,
      "grad_norm": 0.7223210498381862,
      "learning_rate": 0.00014592503168541185,
      "loss": 12.552,
      "step": 8410
    },
    {
      "epoch": 1.470775956284153,
      "grad_norm": 0.8431196791957363,
      "learning_rate": 0.00014591245201187958,
      "loss": 12.6752,
      "step": 8411
    },
    {
      "epoch": 1.4709508196721313,
      "grad_norm": 0.7507800761193539,
      "learning_rate": 0.00014589987141766164,
      "loss": 12.546,
      "step": 8412
    },
    {
      "epoch": 1.4711256830601092,
      "grad_norm": 0.7087407091899106,
      "learning_rate": 0.00014588728990301038,
      "loss": 12.5004,
      "step": 8413
    },
    {
      "epoch": 1.4713005464480875,
      "grad_norm": 0.7334928673534458,
      "learning_rate": 0.00014587470746817803,
      "loss": 12.3464,
      "step": 8414
    },
    {
      "epoch": 1.4714754098360656,
      "grad_norm": 0.7606554075900491,
      "learning_rate": 0.000145862124113417,
      "loss": 12.5023,
      "step": 8415
    },
    {
      "epoch": 1.4716502732240437,
      "grad_norm": 0.7393844618067031,
      "learning_rate": 0.00014584953983897951,
      "loss": 12.4358,
      "step": 8416
    },
    {
      "epoch": 1.4718251366120219,
      "grad_norm": 0.6723574782135963,
      "learning_rate": 0.00014583695464511802,
      "loss": 12.3766,
      "step": 8417
    },
    {
      "epoch": 1.472,
      "grad_norm": 0.7983913948813303,
      "learning_rate": 0.00014582436853208483,
      "loss": 12.4907,
      "step": 8418
    },
    {
      "epoch": 1.472174863387978,
      "grad_norm": 0.7390420578841986,
      "learning_rate": 0.00014581178150013235,
      "loss": 12.4051,
      "step": 8419
    },
    {
      "epoch": 1.4723497267759562,
      "grad_norm": 0.6543323991992192,
      "learning_rate": 0.00014579919354951298,
      "loss": 12.5824,
      "step": 8420
    },
    {
      "epoch": 1.4725245901639346,
      "grad_norm": 0.7216976097636544,
      "learning_rate": 0.00014578660468047918,
      "loss": 12.7019,
      "step": 8421
    },
    {
      "epoch": 1.4726994535519125,
      "grad_norm": 0.720225234481975,
      "learning_rate": 0.00014577401489328335,
      "loss": 12.4103,
      "step": 8422
    },
    {
      "epoch": 1.4728743169398908,
      "grad_norm": 0.675695962553612,
      "learning_rate": 0.000145761424188178,
      "loss": 12.3136,
      "step": 8423
    },
    {
      "epoch": 1.473049180327869,
      "grad_norm": 0.7033985361923092,
      "learning_rate": 0.00014574883256541558,
      "loss": 12.5108,
      "step": 8424
    },
    {
      "epoch": 1.473224043715847,
      "grad_norm": 0.7321353702891459,
      "learning_rate": 0.0001457362400252486,
      "loss": 12.5071,
      "step": 8425
    },
    {
      "epoch": 1.4733989071038252,
      "grad_norm": 0.6491970186516234,
      "learning_rate": 0.00014572364656792957,
      "loss": 12.2009,
      "step": 8426
    },
    {
      "epoch": 1.4735737704918033,
      "grad_norm": 0.72918165432562,
      "learning_rate": 0.00014571105219371102,
      "loss": 12.3023,
      "step": 8427
    },
    {
      "epoch": 1.4737486338797814,
      "grad_norm": 0.700522802123038,
      "learning_rate": 0.00014569845690284557,
      "loss": 12.4098,
      "step": 8428
    },
    {
      "epoch": 1.4739234972677595,
      "grad_norm": 0.7172690072984894,
      "learning_rate": 0.00014568586069558568,
      "loss": 12.3452,
      "step": 8429
    },
    {
      "epoch": 1.4740983606557376,
      "grad_norm": 0.696656888964871,
      "learning_rate": 0.00014567326357218407,
      "loss": 12.2908,
      "step": 8430
    },
    {
      "epoch": 1.4742732240437157,
      "grad_norm": 0.7854223831434743,
      "learning_rate": 0.0001456606655328933,
      "loss": 12.4644,
      "step": 8431
    },
    {
      "epoch": 1.474448087431694,
      "grad_norm": 0.7424710723474899,
      "learning_rate": 0.00014564806657796592,
      "loss": 12.3062,
      "step": 8432
    },
    {
      "epoch": 1.4746229508196722,
      "grad_norm": 0.773398731043287,
      "learning_rate": 0.0001456354667076547,
      "loss": 12.3144,
      "step": 8433
    },
    {
      "epoch": 1.4747978142076503,
      "grad_norm": 0.7781374518827335,
      "learning_rate": 0.00014562286592221223,
      "loss": 12.4162,
      "step": 8434
    },
    {
      "epoch": 1.4749726775956284,
      "grad_norm": 0.6456980654392157,
      "learning_rate": 0.00014561026422189123,
      "loss": 12.4075,
      "step": 8435
    },
    {
      "epoch": 1.4751475409836066,
      "grad_norm": 0.7867058394086262,
      "learning_rate": 0.00014559766160694436,
      "loss": 12.5526,
      "step": 8436
    },
    {
      "epoch": 1.4753224043715847,
      "grad_norm": 0.7463980169777685,
      "learning_rate": 0.0001455850580776244,
      "loss": 12.4639,
      "step": 8437
    },
    {
      "epoch": 1.4754972677595628,
      "grad_norm": 0.7777643076452303,
      "learning_rate": 0.00014557245363418407,
      "loss": 12.4326,
      "step": 8438
    },
    {
      "epoch": 1.475672131147541,
      "grad_norm": 0.730104318652341,
      "learning_rate": 0.00014555984827687612,
      "loss": 12.2868,
      "step": 8439
    },
    {
      "epoch": 1.475846994535519,
      "grad_norm": 0.6780082052494675,
      "learning_rate": 0.0001455472420059533,
      "loss": 12.1424,
      "step": 8440
    },
    {
      "epoch": 1.4760218579234974,
      "grad_norm": 0.6978462957192387,
      "learning_rate": 0.00014553463482166844,
      "loss": 12.5015,
      "step": 8441
    },
    {
      "epoch": 1.4761967213114753,
      "grad_norm": 0.6237062565781011,
      "learning_rate": 0.00014552202672427438,
      "loss": 12.4536,
      "step": 8442
    },
    {
      "epoch": 1.4763715846994536,
      "grad_norm": 0.686590802406318,
      "learning_rate": 0.00014550941771402384,
      "loss": 12.5413,
      "step": 8443
    },
    {
      "epoch": 1.4765464480874317,
      "grad_norm": 0.715071294698452,
      "learning_rate": 0.0001454968077911698,
      "loss": 12.2998,
      "step": 8444
    },
    {
      "epoch": 1.4767213114754099,
      "grad_norm": 0.7104043539735765,
      "learning_rate": 0.00014548419695596505,
      "loss": 12.3303,
      "step": 8445
    },
    {
      "epoch": 1.476896174863388,
      "grad_norm": 0.683693395961882,
      "learning_rate": 0.0001454715852086625,
      "loss": 12.2712,
      "step": 8446
    },
    {
      "epoch": 1.477071038251366,
      "grad_norm": 0.6739575663468971,
      "learning_rate": 0.00014545897254951507,
      "loss": 12.3296,
      "step": 8447
    },
    {
      "epoch": 1.4772459016393442,
      "grad_norm": 0.6864827840536496,
      "learning_rate": 0.0001454463589787756,
      "loss": 12.4047,
      "step": 8448
    },
    {
      "epoch": 1.4774207650273223,
      "grad_norm": 0.7158461526517502,
      "learning_rate": 0.00014543374449669716,
      "loss": 12.4695,
      "step": 8449
    },
    {
      "epoch": 1.4775956284153007,
      "grad_norm": 0.7270231636592961,
      "learning_rate": 0.0001454211291035326,
      "loss": 12.4858,
      "step": 8450
    },
    {
      "epoch": 1.4777704918032786,
      "grad_norm": 0.8098403271978635,
      "learning_rate": 0.00014540851279953496,
      "loss": 12.4266,
      "step": 8451
    },
    {
      "epoch": 1.477945355191257,
      "grad_norm": 0.6810361805302887,
      "learning_rate": 0.00014539589558495717,
      "loss": 12.3434,
      "step": 8452
    },
    {
      "epoch": 1.478120218579235,
      "grad_norm": 0.8882702556902031,
      "learning_rate": 0.00014538327746005237,
      "loss": 12.3558,
      "step": 8453
    },
    {
      "epoch": 1.4782950819672132,
      "grad_norm": 0.6959792592254432,
      "learning_rate": 0.0001453706584250735,
      "loss": 12.5209,
      "step": 8454
    },
    {
      "epoch": 1.4784699453551913,
      "grad_norm": 0.8910527338966964,
      "learning_rate": 0.00014535803848027357,
      "loss": 12.3257,
      "step": 8455
    },
    {
      "epoch": 1.4786448087431694,
      "grad_norm": 0.6846095444959279,
      "learning_rate": 0.00014534541762590574,
      "loss": 12.3923,
      "step": 8456
    },
    {
      "epoch": 1.4788196721311475,
      "grad_norm": 0.7118517047425332,
      "learning_rate": 0.00014533279586222302,
      "loss": 12.393,
      "step": 8457
    },
    {
      "epoch": 1.4789945355191256,
      "grad_norm": 0.9120912494801939,
      "learning_rate": 0.0001453201731894786,
      "loss": 12.2927,
      "step": 8458
    },
    {
      "epoch": 1.4791693989071037,
      "grad_norm": 0.7062771346294425,
      "learning_rate": 0.00014530754960792553,
      "loss": 12.3955,
      "step": 8459
    },
    {
      "epoch": 1.4793442622950819,
      "grad_norm": 0.8268075632043866,
      "learning_rate": 0.00014529492511781698,
      "loss": 12.2635,
      "step": 8460
    },
    {
      "epoch": 1.4795191256830602,
      "grad_norm": 0.7399821925236931,
      "learning_rate": 0.00014528229971940614,
      "loss": 12.4765,
      "step": 8461
    },
    {
      "epoch": 1.4796939890710383,
      "grad_norm": 0.9105010823724626,
      "learning_rate": 0.00014526967341294607,
      "loss": 12.7461,
      "step": 8462
    },
    {
      "epoch": 1.4798688524590164,
      "grad_norm": 0.6840087568786283,
      "learning_rate": 0.00014525704619869015,
      "loss": 12.3607,
      "step": 8463
    },
    {
      "epoch": 1.4800437158469946,
      "grad_norm": 0.8947604652317244,
      "learning_rate": 0.00014524441807689143,
      "loss": 12.5143,
      "step": 8464
    },
    {
      "epoch": 1.4802185792349727,
      "grad_norm": 0.8260842048440988,
      "learning_rate": 0.00014523178904780326,
      "loss": 12.4058,
      "step": 8465
    },
    {
      "epoch": 1.4803934426229508,
      "grad_norm": 0.7273594504830805,
      "learning_rate": 0.0001452191591116788,
      "loss": 12.4506,
      "step": 8466
    },
    {
      "epoch": 1.480568306010929,
      "grad_norm": 0.7445516150039256,
      "learning_rate": 0.00014520652826877136,
      "loss": 12.4709,
      "step": 8467
    },
    {
      "epoch": 1.480743169398907,
      "grad_norm": 0.8406195347025532,
      "learning_rate": 0.00014519389651933425,
      "loss": 12.4095,
      "step": 8468
    },
    {
      "epoch": 1.4809180327868852,
      "grad_norm": 0.8469553965814796,
      "learning_rate": 0.00014518126386362074,
      "loss": 12.2092,
      "step": 8469
    },
    {
      "epoch": 1.4810928961748635,
      "grad_norm": 0.8721510118228775,
      "learning_rate": 0.00014516863030188416,
      "loss": 12.1185,
      "step": 8470
    },
    {
      "epoch": 1.4812677595628414,
      "grad_norm": 0.6817734092787046,
      "learning_rate": 0.00014515599583437788,
      "loss": 12.4551,
      "step": 8471
    },
    {
      "epoch": 1.4814426229508197,
      "grad_norm": 0.7961843564101045,
      "learning_rate": 0.00014514336046135518,
      "loss": 12.3428,
      "step": 8472
    },
    {
      "epoch": 1.4816174863387979,
      "grad_norm": 0.7720392132195277,
      "learning_rate": 0.00014513072418306954,
      "loss": 12.408,
      "step": 8473
    },
    {
      "epoch": 1.481792349726776,
      "grad_norm": 0.69633972022606,
      "learning_rate": 0.00014511808699977432,
      "loss": 12.5564,
      "step": 8474
    },
    {
      "epoch": 1.481967213114754,
      "grad_norm": 0.6861523714932878,
      "learning_rate": 0.0001451054489117229,
      "loss": 12.201,
      "step": 8475
    },
    {
      "epoch": 1.4821420765027322,
      "grad_norm": 0.7871899039605084,
      "learning_rate": 0.00014509280991916875,
      "loss": 12.5158,
      "step": 8476
    },
    {
      "epoch": 1.4823169398907103,
      "grad_norm": 0.6798464808811303,
      "learning_rate": 0.0001450801700223653,
      "loss": 12.4018,
      "step": 8477
    },
    {
      "epoch": 1.4824918032786885,
      "grad_norm": 0.7099174534542273,
      "learning_rate": 0.000145067529221566,
      "loss": 12.5009,
      "step": 8478
    },
    {
      "epoch": 1.4826666666666668,
      "grad_norm": 0.707812289685974,
      "learning_rate": 0.00014505488751702443,
      "loss": 12.1115,
      "step": 8479
    },
    {
      "epoch": 1.4828415300546447,
      "grad_norm": 0.6919950252171619,
      "learning_rate": 0.00014504224490899398,
      "loss": 12.1512,
      "step": 8480
    },
    {
      "epoch": 1.483016393442623,
      "grad_norm": 0.7866944352014305,
      "learning_rate": 0.00014502960139772824,
      "loss": 12.4598,
      "step": 8481
    },
    {
      "epoch": 1.4831912568306012,
      "grad_norm": 0.7835920602160426,
      "learning_rate": 0.00014501695698348072,
      "loss": 12.2838,
      "step": 8482
    },
    {
      "epoch": 1.4833661202185793,
      "grad_norm": 0.7770996961157819,
      "learning_rate": 0.00014500431166650498,
      "loss": 12.1915,
      "step": 8483
    },
    {
      "epoch": 1.4835409836065574,
      "grad_norm": 0.6912230349494611,
      "learning_rate": 0.00014499166544705466,
      "loss": 12.2555,
      "step": 8484
    },
    {
      "epoch": 1.4837158469945355,
      "grad_norm": 0.7547069322506703,
      "learning_rate": 0.00014497901832538325,
      "loss": 12.4027,
      "step": 8485
    },
    {
      "epoch": 1.4838907103825136,
      "grad_norm": 0.7705826448931006,
      "learning_rate": 0.00014496637030174443,
      "loss": 12.5738,
      "step": 8486
    },
    {
      "epoch": 1.4840655737704918,
      "grad_norm": 0.8075137974552354,
      "learning_rate": 0.00014495372137639185,
      "loss": 12.6156,
      "step": 8487
    },
    {
      "epoch": 1.4842404371584699,
      "grad_norm": 0.6755084141142172,
      "learning_rate": 0.0001449410715495791,
      "loss": 12.3345,
      "step": 8488
    },
    {
      "epoch": 1.484415300546448,
      "grad_norm": 0.7806523774468864,
      "learning_rate": 0.0001449284208215599,
      "loss": 12.4182,
      "step": 8489
    },
    {
      "epoch": 1.4845901639344263,
      "grad_norm": 0.9871578200267671,
      "learning_rate": 0.00014491576919258792,
      "loss": 12.3976,
      "step": 8490
    },
    {
      "epoch": 1.4847650273224045,
      "grad_norm": 0.8458753150566662,
      "learning_rate": 0.00014490311666291684,
      "loss": 12.4941,
      "step": 8491
    },
    {
      "epoch": 1.4849398907103826,
      "grad_norm": 0.7849772562597225,
      "learning_rate": 0.0001448904632328004,
      "loss": 12.6135,
      "step": 8492
    },
    {
      "epoch": 1.4851147540983607,
      "grad_norm": 0.6952068632896112,
      "learning_rate": 0.00014487780890249234,
      "loss": 12.1953,
      "step": 8493
    },
    {
      "epoch": 1.4852896174863388,
      "grad_norm": 0.7028787540768213,
      "learning_rate": 0.0001448651536722464,
      "loss": 12.5358,
      "step": 8494
    },
    {
      "epoch": 1.485464480874317,
      "grad_norm": 0.7151024254564847,
      "learning_rate": 0.00014485249754231646,
      "loss": 12.5613,
      "step": 8495
    },
    {
      "epoch": 1.485639344262295,
      "grad_norm": 0.689549086457626,
      "learning_rate": 0.00014483984051295612,
      "loss": 12.4287,
      "step": 8496
    },
    {
      "epoch": 1.4858142076502732,
      "grad_norm": 0.7285444214954695,
      "learning_rate": 0.00014482718258441936,
      "loss": 12.5187,
      "step": 8497
    },
    {
      "epoch": 1.4859890710382513,
      "grad_norm": 0.7587204575733303,
      "learning_rate": 0.00014481452375695995,
      "loss": 12.3954,
      "step": 8498
    },
    {
      "epoch": 1.4861639344262296,
      "grad_norm": 0.7721182701770245,
      "learning_rate": 0.00014480186403083173,
      "loss": 12.4623,
      "step": 8499
    },
    {
      "epoch": 1.4863387978142075,
      "grad_norm": 0.6689835245573994,
      "learning_rate": 0.00014478920340628857,
      "loss": 12.5729,
      "step": 8500
    },
    {
      "epoch": 1.4865136612021859,
      "grad_norm": 0.7373721674213727,
      "learning_rate": 0.00014477654188358436,
      "loss": 12.3698,
      "step": 8501
    },
    {
      "epoch": 1.486688524590164,
      "grad_norm": 0.701978006199364,
      "learning_rate": 0.00014476387946297302,
      "loss": 12.3314,
      "step": 8502
    },
    {
      "epoch": 1.486863387978142,
      "grad_norm": 0.784618893490127,
      "learning_rate": 0.00014475121614470842,
      "loss": 12.4525,
      "step": 8503
    },
    {
      "epoch": 1.4870382513661202,
      "grad_norm": 0.6867502226748078,
      "learning_rate": 0.00014473855192904456,
      "loss": 12.4433,
      "step": 8504
    },
    {
      "epoch": 1.4872131147540983,
      "grad_norm": 0.6470499878178734,
      "learning_rate": 0.0001447258868162354,
      "loss": 12.5349,
      "step": 8505
    },
    {
      "epoch": 1.4873879781420765,
      "grad_norm": 0.7692879346137519,
      "learning_rate": 0.00014471322080653485,
      "loss": 12.4082,
      "step": 8506
    },
    {
      "epoch": 1.4875628415300546,
      "grad_norm": 0.7244895703716018,
      "learning_rate": 0.00014470055390019692,
      "loss": 12.4939,
      "step": 8507
    },
    {
      "epoch": 1.487737704918033,
      "grad_norm": 0.6956530125776694,
      "learning_rate": 0.00014468788609747565,
      "loss": 12.3134,
      "step": 8508
    },
    {
      "epoch": 1.4879125683060108,
      "grad_norm": 0.7472285430027881,
      "learning_rate": 0.00014467521739862504,
      "loss": 12.3958,
      "step": 8509
    },
    {
      "epoch": 1.4880874316939892,
      "grad_norm": 0.7239655301127983,
      "learning_rate": 0.00014466254780389913,
      "loss": 12.3635,
      "step": 8510
    },
    {
      "epoch": 1.4882622950819673,
      "grad_norm": 0.721642651841516,
      "learning_rate": 0.00014464987731355205,
      "loss": 11.9735,
      "step": 8511
    },
    {
      "epoch": 1.4884371584699454,
      "grad_norm": 0.7172048548385397,
      "learning_rate": 0.00014463720592783783,
      "loss": 12.2762,
      "step": 8512
    },
    {
      "epoch": 1.4886120218579235,
      "grad_norm": 0.7791140819984651,
      "learning_rate": 0.00014462453364701058,
      "loss": 12.5597,
      "step": 8513
    },
    {
      "epoch": 1.4887868852459016,
      "grad_norm": 0.6973014845135678,
      "learning_rate": 0.0001446118604713244,
      "loss": 12.0909,
      "step": 8514
    },
    {
      "epoch": 1.4889617486338798,
      "grad_norm": 0.6671203778514653,
      "learning_rate": 0.00014459918640103345,
      "loss": 12.3968,
      "step": 8515
    },
    {
      "epoch": 1.4891366120218579,
      "grad_norm": 0.7138636765814423,
      "learning_rate": 0.00014458651143639187,
      "loss": 12.3917,
      "step": 8516
    },
    {
      "epoch": 1.489311475409836,
      "grad_norm": 0.7405518849891618,
      "learning_rate": 0.00014457383557765386,
      "loss": 12.3932,
      "step": 8517
    },
    {
      "epoch": 1.4894863387978141,
      "grad_norm": 0.7428393945284218,
      "learning_rate": 0.00014456115882507356,
      "loss": 12.5323,
      "step": 8518
    },
    {
      "epoch": 1.4896612021857925,
      "grad_norm": 0.7845850567839977,
      "learning_rate": 0.0001445484811789052,
      "loss": 12.2479,
      "step": 8519
    },
    {
      "epoch": 1.4898360655737706,
      "grad_norm": 0.6940357046060932,
      "learning_rate": 0.00014453580263940302,
      "loss": 12.2304,
      "step": 8520
    },
    {
      "epoch": 1.4900109289617487,
      "grad_norm": 0.8232084942314072,
      "learning_rate": 0.00014452312320682124,
      "loss": 12.4228,
      "step": 8521
    },
    {
      "epoch": 1.4901857923497268,
      "grad_norm": 0.7038458476180125,
      "learning_rate": 0.00014451044288141414,
      "loss": 12.5546,
      "step": 8522
    },
    {
      "epoch": 1.490360655737705,
      "grad_norm": 0.8079672780975996,
      "learning_rate": 0.00014449776166343604,
      "loss": 11.9782,
      "step": 8523
    },
    {
      "epoch": 1.490535519125683,
      "grad_norm": 0.7434284523166228,
      "learning_rate": 0.00014448507955314114,
      "loss": 12.3925,
      "step": 8524
    },
    {
      "epoch": 1.4907103825136612,
      "grad_norm": 0.8329346780290712,
      "learning_rate": 0.00014447239655078384,
      "loss": 12.4255,
      "step": 8525
    },
    {
      "epoch": 1.4908852459016393,
      "grad_norm": 0.7338881645622168,
      "learning_rate": 0.00014445971265661842,
      "loss": 12.0577,
      "step": 8526
    },
    {
      "epoch": 1.4910601092896174,
      "grad_norm": 0.846938450496644,
      "learning_rate": 0.00014444702787089924,
      "loss": 12.5616,
      "step": 8527
    },
    {
      "epoch": 1.4912349726775957,
      "grad_norm": 0.6898530591313494,
      "learning_rate": 0.00014443434219388071,
      "loss": 12.5236,
      "step": 8528
    },
    {
      "epoch": 1.4914098360655736,
      "grad_norm": 0.7726022978039098,
      "learning_rate": 0.00014442165562581716,
      "loss": 12.4213,
      "step": 8529
    },
    {
      "epoch": 1.491584699453552,
      "grad_norm": 0.7106350603406405,
      "learning_rate": 0.00014440896816696305,
      "loss": 12.3823,
      "step": 8530
    },
    {
      "epoch": 1.49175956284153,
      "grad_norm": 0.7183455889948288,
      "learning_rate": 0.0001443962798175728,
      "loss": 12.2724,
      "step": 8531
    },
    {
      "epoch": 1.4919344262295082,
      "grad_norm": 0.7654368171670932,
      "learning_rate": 0.00014438359057790075,
      "loss": 12.5845,
      "step": 8532
    },
    {
      "epoch": 1.4921092896174863,
      "grad_norm": 0.643164127446816,
      "learning_rate": 0.0001443709004482015,
      "loss": 12.2979,
      "step": 8533
    },
    {
      "epoch": 1.4922841530054645,
      "grad_norm": 0.8380686346403367,
      "learning_rate": 0.0001443582094287294,
      "loss": 12.3242,
      "step": 8534
    },
    {
      "epoch": 1.4924590163934426,
      "grad_norm": 0.7754677592684419,
      "learning_rate": 0.00014434551751973907,
      "loss": 12.446,
      "step": 8535
    },
    {
      "epoch": 1.4926338797814207,
      "grad_norm": 0.7901989147962184,
      "learning_rate": 0.0001443328247214849,
      "loss": 12.2617,
      "step": 8536
    },
    {
      "epoch": 1.492808743169399,
      "grad_norm": 0.765826590755664,
      "learning_rate": 0.0001443201310342215,
      "loss": 12.3801,
      "step": 8537
    },
    {
      "epoch": 1.492983606557377,
      "grad_norm": 0.6919781932638752,
      "learning_rate": 0.0001443074364582034,
      "loss": 12.3324,
      "step": 8538
    },
    {
      "epoch": 1.4931584699453553,
      "grad_norm": 0.7391714121264673,
      "learning_rate": 0.00014429474099368514,
      "loss": 12.4488,
      "step": 8539
    },
    {
      "epoch": 1.4933333333333334,
      "grad_norm": 0.6691480913206931,
      "learning_rate": 0.00014428204464092136,
      "loss": 12.4183,
      "step": 8540
    },
    {
      "epoch": 1.4935081967213115,
      "grad_norm": 0.7954137332345447,
      "learning_rate": 0.00014426934740016656,
      "loss": 12.514,
      "step": 8541
    },
    {
      "epoch": 1.4936830601092896,
      "grad_norm": 0.7951826829674499,
      "learning_rate": 0.00014425664927167543,
      "loss": 12.4843,
      "step": 8542
    },
    {
      "epoch": 1.4938579234972678,
      "grad_norm": 0.6927358086283443,
      "learning_rate": 0.00014424395025570267,
      "loss": 12.3005,
      "step": 8543
    },
    {
      "epoch": 1.4940327868852459,
      "grad_norm": 0.7896496277903887,
      "learning_rate": 0.00014423125035250276,
      "loss": 12.4667,
      "step": 8544
    },
    {
      "epoch": 1.494207650273224,
      "grad_norm": 0.7463314872648819,
      "learning_rate": 0.00014421854956233053,
      "loss": 12.3072,
      "step": 8545
    },
    {
      "epoch": 1.4943825136612021,
      "grad_norm": 0.6475266327858569,
      "learning_rate": 0.00014420584788544057,
      "loss": 12.3945,
      "step": 8546
    },
    {
      "epoch": 1.4945573770491802,
      "grad_norm": 0.7188200996544892,
      "learning_rate": 0.0001441931453220877,
      "loss": 12.4547,
      "step": 8547
    },
    {
      "epoch": 1.4947322404371586,
      "grad_norm": 0.744094270633061,
      "learning_rate": 0.00014418044187252652,
      "loss": 12.6192,
      "step": 8548
    },
    {
      "epoch": 1.4949071038251365,
      "grad_norm": 0.8728200005898233,
      "learning_rate": 0.00014416773753701183,
      "loss": 12.4804,
      "step": 8549
    },
    {
      "epoch": 1.4950819672131148,
      "grad_norm": 0.7292253192017336,
      "learning_rate": 0.00014415503231579843,
      "loss": 12.3951,
      "step": 8550
    },
    {
      "epoch": 1.495256830601093,
      "grad_norm": 0.7534536211988222,
      "learning_rate": 0.000144142326209141,
      "loss": 12.5286,
      "step": 8551
    },
    {
      "epoch": 1.495431693989071,
      "grad_norm": 0.7920573458366194,
      "learning_rate": 0.00014412961921729444,
      "loss": 12.7286,
      "step": 8552
    },
    {
      "epoch": 1.4956065573770492,
      "grad_norm": 0.6765779523894306,
      "learning_rate": 0.00014411691134051348,
      "loss": 12.6085,
      "step": 8553
    },
    {
      "epoch": 1.4957814207650273,
      "grad_norm": 0.7486773747687827,
      "learning_rate": 0.00014410420257905305,
      "loss": 12.3063,
      "step": 8554
    },
    {
      "epoch": 1.4959562841530054,
      "grad_norm": 0.7125274419275688,
      "learning_rate": 0.00014409149293316792,
      "loss": 12.3349,
      "step": 8555
    },
    {
      "epoch": 1.4961311475409835,
      "grad_norm": 0.7413115393377436,
      "learning_rate": 0.00014407878240311293,
      "loss": 12.3567,
      "step": 8556
    },
    {
      "epoch": 1.4963060109289619,
      "grad_norm": 0.7462203526598042,
      "learning_rate": 0.00014406607098914307,
      "loss": 12.4951,
      "step": 8557
    },
    {
      "epoch": 1.4964808743169398,
      "grad_norm": 0.6591507885310234,
      "learning_rate": 0.00014405335869151315,
      "loss": 12.3396,
      "step": 8558
    },
    {
      "epoch": 1.496655737704918,
      "grad_norm": 0.7342590686829786,
      "learning_rate": 0.00014404064551047817,
      "loss": 12.5549,
      "step": 8559
    },
    {
      "epoch": 1.4968306010928962,
      "grad_norm": 0.6944285554857396,
      "learning_rate": 0.00014402793144629295,
      "loss": 12.5853,
      "step": 8560
    },
    {
      "epoch": 1.4970054644808743,
      "grad_norm": 0.7299460602606274,
      "learning_rate": 0.0001440152164992126,
      "loss": 12.4713,
      "step": 8561
    },
    {
      "epoch": 1.4971803278688525,
      "grad_norm": 0.6744718257535404,
      "learning_rate": 0.00014400250066949196,
      "loss": 12.6001,
      "step": 8562
    },
    {
      "epoch": 1.4973551912568306,
      "grad_norm": 0.7693845760150777,
      "learning_rate": 0.00014398978395738612,
      "loss": 12.4324,
      "step": 8563
    },
    {
      "epoch": 1.4975300546448087,
      "grad_norm": 0.70977392187945,
      "learning_rate": 0.00014397706636315,
      "loss": 12.5128,
      "step": 8564
    },
    {
      "epoch": 1.4977049180327868,
      "grad_norm": 0.6738880970644274,
      "learning_rate": 0.0001439643478870387,
      "loss": 12.4163,
      "step": 8565
    },
    {
      "epoch": 1.497879781420765,
      "grad_norm": 0.7094077147617118,
      "learning_rate": 0.00014395162852930723,
      "loss": 12.4332,
      "step": 8566
    },
    {
      "epoch": 1.498054644808743,
      "grad_norm": 0.7201138215045257,
      "learning_rate": 0.00014393890829021065,
      "loss": 12.4371,
      "step": 8567
    },
    {
      "epoch": 1.4982295081967214,
      "grad_norm": 0.7649101572326481,
      "learning_rate": 0.00014392618717000408,
      "loss": 12.5941,
      "step": 8568
    },
    {
      "epoch": 1.4984043715846995,
      "grad_norm": 0.8317486772053252,
      "learning_rate": 0.00014391346516894255,
      "loss": 12.6077,
      "step": 8569
    },
    {
      "epoch": 1.4985792349726776,
      "grad_norm": 0.708394838221976,
      "learning_rate": 0.00014390074228728123,
      "loss": 12.3864,
      "step": 8570
    },
    {
      "epoch": 1.4987540983606558,
      "grad_norm": 0.7244934144898657,
      "learning_rate": 0.00014388801852527526,
      "loss": 12.4568,
      "step": 8571
    },
    {
      "epoch": 1.4989289617486339,
      "grad_norm": 0.6806469045827788,
      "learning_rate": 0.00014387529388317971,
      "loss": 12.4393,
      "step": 8572
    },
    {
      "epoch": 1.499103825136612,
      "grad_norm": 0.8141938695163711,
      "learning_rate": 0.00014386256836124983,
      "loss": 12.2559,
      "step": 8573
    },
    {
      "epoch": 1.4992786885245901,
      "grad_norm": 0.6716281050037625,
      "learning_rate": 0.00014384984195974076,
      "loss": 12.4828,
      "step": 8574
    },
    {
      "epoch": 1.4994535519125682,
      "grad_norm": 0.7187709927840578,
      "learning_rate": 0.00014383711467890774,
      "loss": 12.5178,
      "step": 8575
    },
    {
      "epoch": 1.4996284153005464,
      "grad_norm": 0.7852829034046802,
      "learning_rate": 0.00014382438651900598,
      "loss": 12.5003,
      "step": 8576
    },
    {
      "epoch": 1.4998032786885247,
      "grad_norm": 0.7387409593615787,
      "learning_rate": 0.0001438116574802907,
      "loss": 12.451,
      "step": 8577
    },
    {
      "epoch": 1.4999781420765026,
      "grad_norm": 0.7860553747532942,
      "learning_rate": 0.0001437989275630172,
      "loss": 12.4646,
      "step": 8578
    },
    {
      "epoch": 1.500153005464481,
      "grad_norm": 0.7474992274612391,
      "learning_rate": 0.0001437861967674407,
      "loss": 12.3107,
      "step": 8579
    },
    {
      "epoch": 1.500327868852459,
      "grad_norm": 0.8303228035351569,
      "learning_rate": 0.00014377346509381647,
      "loss": 12.5859,
      "step": 8580
    },
    {
      "epoch": 1.5005027322404372,
      "grad_norm": 0.7176534562974519,
      "learning_rate": 0.00014376073254239991,
      "loss": 12.3323,
      "step": 8581
    },
    {
      "epoch": 1.5006775956284153,
      "grad_norm": 0.9015906707121063,
      "learning_rate": 0.00014374799911344628,
      "loss": 12.2309,
      "step": 8582
    },
    {
      "epoch": 1.5008524590163934,
      "grad_norm": 0.6943095546357465,
      "learning_rate": 0.00014373526480721097,
      "loss": 12.3824,
      "step": 8583
    },
    {
      "epoch": 1.5010273224043715,
      "grad_norm": 1.1294946970872872,
      "learning_rate": 0.0001437225296239493,
      "loss": 12.6094,
      "step": 8584
    },
    {
      "epoch": 1.5012021857923497,
      "grad_norm": 0.69868260941596,
      "learning_rate": 0.00014370979356391665,
      "loss": 12.3635,
      "step": 8585
    },
    {
      "epoch": 1.501377049180328,
      "grad_norm": 1.0559746898657463,
      "learning_rate": 0.00014369705662736846,
      "loss": 12.5251,
      "step": 8586
    },
    {
      "epoch": 1.501551912568306,
      "grad_norm": 0.7068613040595868,
      "learning_rate": 0.00014368431881456008,
      "loss": 12.4769,
      "step": 8587
    },
    {
      "epoch": 1.5017267759562842,
      "grad_norm": 0.849467063396524,
      "learning_rate": 0.00014367158012574703,
      "loss": 12.4348,
      "step": 8588
    },
    {
      "epoch": 1.5019016393442621,
      "grad_norm": 0.8227887550305196,
      "learning_rate": 0.00014365884056118466,
      "loss": 12.5263,
      "step": 8589
    },
    {
      "epoch": 1.5020765027322405,
      "grad_norm": 0.8717042866549747,
      "learning_rate": 0.00014364610012112852,
      "loss": 12.4317,
      "step": 8590
    },
    {
      "epoch": 1.5022513661202186,
      "grad_norm": 0.8357408886318227,
      "learning_rate": 0.00014363335880583406,
      "loss": 12.5366,
      "step": 8591
    },
    {
      "epoch": 1.5024262295081967,
      "grad_norm": 0.7173132553064718,
      "learning_rate": 0.00014362061661555675,
      "loss": 12.1222,
      "step": 8592
    },
    {
      "epoch": 1.5026010928961748,
      "grad_norm": 0.8938012090826917,
      "learning_rate": 0.00014360787355055218,
      "loss": 12.5936,
      "step": 8593
    },
    {
      "epoch": 1.502775956284153,
      "grad_norm": 0.7948559431303991,
      "learning_rate": 0.00014359512961107583,
      "loss": 12.5493,
      "step": 8594
    },
    {
      "epoch": 1.5029508196721313,
      "grad_norm": 0.8776066807696219,
      "learning_rate": 0.0001435823847973833,
      "loss": 12.2791,
      "step": 8595
    },
    {
      "epoch": 1.5031256830601092,
      "grad_norm": 0.6818341114952607,
      "learning_rate": 0.00014356963910973008,
      "loss": 12.2699,
      "step": 8596
    },
    {
      "epoch": 1.5033005464480875,
      "grad_norm": 0.9603308047145057,
      "learning_rate": 0.00014355689254837191,
      "loss": 12.2244,
      "step": 8597
    },
    {
      "epoch": 1.5034754098360654,
      "grad_norm": 0.7195284392755729,
      "learning_rate": 0.00014354414511356427,
      "loss": 12.5697,
      "step": 8598
    },
    {
      "epoch": 1.5036502732240438,
      "grad_norm": 0.9239298400234233,
      "learning_rate": 0.0001435313968055628,
      "loss": 12.4353,
      "step": 8599
    },
    {
      "epoch": 1.5038251366120219,
      "grad_norm": 0.8035458369328247,
      "learning_rate": 0.00014351864762462318,
      "loss": 12.6109,
      "step": 8600
    },
    {
      "epoch": 1.504,
      "grad_norm": 0.9099118992508765,
      "learning_rate": 0.00014350589757100105,
      "loss": 12.4081,
      "step": 8601
    },
    {
      "epoch": 1.5041748633879781,
      "grad_norm": 0.8910614905334869,
      "learning_rate": 0.0001434931466449521,
      "loss": 12.7144,
      "step": 8602
    },
    {
      "epoch": 1.5043497267759562,
      "grad_norm": 0.8954366523434834,
      "learning_rate": 0.00014348039484673203,
      "loss": 12.4807,
      "step": 8603
    },
    {
      "epoch": 1.5045245901639346,
      "grad_norm": 0.8729572198204777,
      "learning_rate": 0.00014346764217659653,
      "loss": 12.3443,
      "step": 8604
    },
    {
      "epoch": 1.5046994535519125,
      "grad_norm": 0.709007081028022,
      "learning_rate": 0.00014345488863480132,
      "loss": 12.4293,
      "step": 8605
    },
    {
      "epoch": 1.5048743169398908,
      "grad_norm": 0.9802856757202079,
      "learning_rate": 0.00014344213422160218,
      "loss": 12.2393,
      "step": 8606
    },
    {
      "epoch": 1.5050491803278687,
      "grad_norm": 0.6780323866771147,
      "learning_rate": 0.00014342937893725488,
      "loss": 12.414,
      "step": 8607
    },
    {
      "epoch": 1.505224043715847,
      "grad_norm": 1.0015367228787468,
      "learning_rate": 0.00014341662278201516,
      "loss": 12.2753,
      "step": 8608
    },
    {
      "epoch": 1.5053989071038252,
      "grad_norm": 0.7194063473207285,
      "learning_rate": 0.00014340386575613885,
      "loss": 12.4614,
      "step": 8609
    },
    {
      "epoch": 1.5055737704918033,
      "grad_norm": 0.83689275360185,
      "learning_rate": 0.0001433911078598818,
      "loss": 12.4543,
      "step": 8610
    },
    {
      "epoch": 1.5057486338797814,
      "grad_norm": 0.8235477471554673,
      "learning_rate": 0.0001433783490934998,
      "loss": 12.1476,
      "step": 8611
    },
    {
      "epoch": 1.5059234972677595,
      "grad_norm": 0.7723941970350322,
      "learning_rate": 0.0001433655894572487,
      "loss": 12.4262,
      "step": 8612
    },
    {
      "epoch": 1.5060983606557377,
      "grad_norm": 0.7900742021718773,
      "learning_rate": 0.0001433528289513844,
      "loss": 12.2622,
      "step": 8613
    },
    {
      "epoch": 1.5062732240437158,
      "grad_norm": 0.7138804104617363,
      "learning_rate": 0.00014334006757616273,
      "loss": 12.4612,
      "step": 8614
    },
    {
      "epoch": 1.5064480874316941,
      "grad_norm": 0.7747591117601695,
      "learning_rate": 0.00014332730533183965,
      "loss": 12.5992,
      "step": 8615
    },
    {
      "epoch": 1.506622950819672,
      "grad_norm": 0.7396095474294392,
      "learning_rate": 0.00014331454221867108,
      "loss": 12.5601,
      "step": 8616
    },
    {
      "epoch": 1.5067978142076504,
      "grad_norm": 0.8114336516935697,
      "learning_rate": 0.0001433017782369129,
      "loss": 12.5619,
      "step": 8617
    },
    {
      "epoch": 1.5069726775956283,
      "grad_norm": 0.6743824957682675,
      "learning_rate": 0.00014328901338682117,
      "loss": 12.1802,
      "step": 8618
    },
    {
      "epoch": 1.5071475409836066,
      "grad_norm": 0.7835022681614501,
      "learning_rate": 0.00014327624766865177,
      "loss": 12.2062,
      "step": 8619
    },
    {
      "epoch": 1.5073224043715847,
      "grad_norm": 0.6772254493111032,
      "learning_rate": 0.00014326348108266074,
      "loss": 12.1031,
      "step": 8620
    },
    {
      "epoch": 1.5074972677595628,
      "grad_norm": 0.68398669310681,
      "learning_rate": 0.0001432507136291041,
      "loss": 12.3974,
      "step": 8621
    },
    {
      "epoch": 1.507672131147541,
      "grad_norm": 0.7607747180659197,
      "learning_rate": 0.00014323794530823783,
      "loss": 12.0742,
      "step": 8622
    },
    {
      "epoch": 1.507846994535519,
      "grad_norm": 0.6568736564698842,
      "learning_rate": 0.000143225176120318,
      "loss": 12.4757,
      "step": 8623
    },
    {
      "epoch": 1.5080218579234974,
      "grad_norm": 0.7439718715644471,
      "learning_rate": 0.00014321240606560067,
      "loss": 12.1289,
      "step": 8624
    },
    {
      "epoch": 1.5081967213114753,
      "grad_norm": 0.7334790484764746,
      "learning_rate": 0.0001431996351443419,
      "loss": 12.178,
      "step": 8625
    },
    {
      "epoch": 1.5083715846994536,
      "grad_norm": 0.6797118083915806,
      "learning_rate": 0.00014318686335679784,
      "loss": 12.246,
      "step": 8626
    },
    {
      "epoch": 1.5085464480874315,
      "grad_norm": 0.6939784051590551,
      "learning_rate": 0.0001431740907032246,
      "loss": 12.175,
      "step": 8627
    },
    {
      "epoch": 1.5087213114754099,
      "grad_norm": 0.6648541292605484,
      "learning_rate": 0.0001431613171838782,
      "loss": 12.2781,
      "step": 8628
    },
    {
      "epoch": 1.508896174863388,
      "grad_norm": 0.6775499898391442,
      "learning_rate": 0.00014314854279901492,
      "loss": 12.3273,
      "step": 8629
    },
    {
      "epoch": 1.5090710382513661,
      "grad_norm": 0.776797619260395,
      "learning_rate": 0.00014313576754889084,
      "loss": 12.3511,
      "step": 8630
    },
    {
      "epoch": 1.5092459016393442,
      "grad_norm": 0.6720542886861456,
      "learning_rate": 0.00014312299143376218,
      "loss": 12.2115,
      "step": 8631
    },
    {
      "epoch": 1.5094207650273224,
      "grad_norm": 0.8407001928857512,
      "learning_rate": 0.00014311021445388516,
      "loss": 12.2628,
      "step": 8632
    },
    {
      "epoch": 1.5095956284153007,
      "grad_norm": 0.6705016082594599,
      "learning_rate": 0.00014309743660951595,
      "loss": 12.5841,
      "step": 8633
    },
    {
      "epoch": 1.5097704918032786,
      "grad_norm": 0.8297944428666308,
      "learning_rate": 0.00014308465790091086,
      "loss": 12.4509,
      "step": 8634
    },
    {
      "epoch": 1.509945355191257,
      "grad_norm": 0.7278037311789509,
      "learning_rate": 0.00014307187832832602,
      "loss": 12.5452,
      "step": 8635
    },
    {
      "epoch": 1.5101202185792348,
      "grad_norm": 0.7570508336799749,
      "learning_rate": 0.00014305909789201783,
      "loss": 12.1967,
      "step": 8636
    },
    {
      "epoch": 1.5102950819672132,
      "grad_norm": 0.6524088371707079,
      "learning_rate": 0.0001430463165922425,
      "loss": 12.4142,
      "step": 8637
    },
    {
      "epoch": 1.5104699453551913,
      "grad_norm": 0.81217477758824,
      "learning_rate": 0.00014303353442925634,
      "loss": 12.5492,
      "step": 8638
    },
    {
      "epoch": 1.5106448087431694,
      "grad_norm": 0.7698524156009723,
      "learning_rate": 0.0001430207514033157,
      "loss": 12.1042,
      "step": 8639
    },
    {
      "epoch": 1.5108196721311475,
      "grad_norm": 0.8313949895230518,
      "learning_rate": 0.0001430079675146769,
      "loss": 12.1343,
      "step": 8640
    },
    {
      "epoch": 1.5109945355191257,
      "grad_norm": 0.7488123085736137,
      "learning_rate": 0.00014299518276359629,
      "loss": 12.4966,
      "step": 8641
    },
    {
      "epoch": 1.5111693989071038,
      "grad_norm": 0.9485658445233025,
      "learning_rate": 0.00014298239715033025,
      "loss": 12.5542,
      "step": 8642
    },
    {
      "epoch": 1.511344262295082,
      "grad_norm": 0.7129748035017917,
      "learning_rate": 0.0001429696106751352,
      "loss": 12.4081,
      "step": 8643
    },
    {
      "epoch": 1.5115191256830602,
      "grad_norm": 0.8435083118983463,
      "learning_rate": 0.00014295682333826748,
      "loss": 12.5637,
      "step": 8644
    },
    {
      "epoch": 1.5116939890710381,
      "grad_norm": 0.6850238649110068,
      "learning_rate": 0.0001429440351399836,
      "loss": 12.3854,
      "step": 8645
    },
    {
      "epoch": 1.5118688524590165,
      "grad_norm": 0.7994914648959337,
      "learning_rate": 0.00014293124608053993,
      "loss": 12.31,
      "step": 8646
    },
    {
      "epoch": 1.5120437158469944,
      "grad_norm": 0.7232215443160885,
      "learning_rate": 0.00014291845616019297,
      "loss": 12.3321,
      "step": 8647
    },
    {
      "epoch": 1.5122185792349727,
      "grad_norm": 0.6827369733653533,
      "learning_rate": 0.0001429056653791992,
      "loss": 11.9341,
      "step": 8648
    },
    {
      "epoch": 1.5123934426229508,
      "grad_norm": 0.7149714110633723,
      "learning_rate": 0.00014289287373781505,
      "loss": 12.1835,
      "step": 8649
    },
    {
      "epoch": 1.512568306010929,
      "grad_norm": 0.6928223861718932,
      "learning_rate": 0.00014288008123629716,
      "loss": 12.2746,
      "step": 8650
    },
    {
      "epoch": 1.512743169398907,
      "grad_norm": 0.648425834566239,
      "learning_rate": 0.00014286728787490195,
      "loss": 12.4676,
      "step": 8651
    },
    {
      "epoch": 1.5129180327868852,
      "grad_norm": 0.726661493654488,
      "learning_rate": 0.00014285449365388598,
      "loss": 12.3822,
      "step": 8652
    },
    {
      "epoch": 1.5130928961748635,
      "grad_norm": 0.7809738183547706,
      "learning_rate": 0.00014284169857350584,
      "loss": 12.5593,
      "step": 8653
    },
    {
      "epoch": 1.5132677595628414,
      "grad_norm": 0.7012249171018916,
      "learning_rate": 0.0001428289026340181,
      "loss": 12.3936,
      "step": 8654
    },
    {
      "epoch": 1.5134426229508198,
      "grad_norm": 0.7613464297420518,
      "learning_rate": 0.0001428161058356794,
      "loss": 11.9889,
      "step": 8655
    },
    {
      "epoch": 1.5136174863387977,
      "grad_norm": 0.7161753634422697,
      "learning_rate": 0.00014280330817874624,
      "loss": 12.1875,
      "step": 8656
    },
    {
      "epoch": 1.513792349726776,
      "grad_norm": 0.7347109489689484,
      "learning_rate": 0.0001427905096634754,
      "loss": 12.4132,
      "step": 8657
    },
    {
      "epoch": 1.5139672131147541,
      "grad_norm": 0.7448228108904705,
      "learning_rate": 0.00014277771029012342,
      "loss": 12.2084,
      "step": 8658
    },
    {
      "epoch": 1.5141420765027322,
      "grad_norm": 0.6532737203186086,
      "learning_rate": 0.00014276491005894704,
      "loss": 12.4748,
      "step": 8659
    },
    {
      "epoch": 1.5143169398907104,
      "grad_norm": 0.6880222768020735,
      "learning_rate": 0.00014275210897020285,
      "loss": 12.5447,
      "step": 8660
    },
    {
      "epoch": 1.5144918032786885,
      "grad_norm": 0.7603615053251965,
      "learning_rate": 0.00014273930702414766,
      "loss": 12.4931,
      "step": 8661
    },
    {
      "epoch": 1.5146666666666668,
      "grad_norm": 0.7024416843346596,
      "learning_rate": 0.0001427265042210381,
      "loss": 12.0654,
      "step": 8662
    },
    {
      "epoch": 1.5148415300546447,
      "grad_norm": 0.7343758801607286,
      "learning_rate": 0.00014271370056113096,
      "loss": 12.3623,
      "step": 8663
    },
    {
      "epoch": 1.515016393442623,
      "grad_norm": 0.7513196075864526,
      "learning_rate": 0.00014270089604468298,
      "loss": 12.3715,
      "step": 8664
    },
    {
      "epoch": 1.515191256830601,
      "grad_norm": 0.7092362305549932,
      "learning_rate": 0.0001426880906719509,
      "loss": 12.3283,
      "step": 8665
    },
    {
      "epoch": 1.5153661202185793,
      "grad_norm": 0.7416701306510982,
      "learning_rate": 0.00014267528444319156,
      "loss": 12.3647,
      "step": 8666
    },
    {
      "epoch": 1.5155409836065574,
      "grad_norm": 0.7061719069636712,
      "learning_rate": 0.00014266247735866174,
      "loss": 12.2785,
      "step": 8667
    },
    {
      "epoch": 1.5157158469945355,
      "grad_norm": 0.7254989768574497,
      "learning_rate": 0.00014264966941861823,
      "loss": 12.1142,
      "step": 8668
    },
    {
      "epoch": 1.5158907103825137,
      "grad_norm": 0.8811526462449485,
      "learning_rate": 0.0001426368606233179,
      "loss": 12.3653,
      "step": 8669
    },
    {
      "epoch": 1.5160655737704918,
      "grad_norm": 0.7007263516957922,
      "learning_rate": 0.00014262405097301763,
      "loss": 12.3953,
      "step": 8670
    },
    {
      "epoch": 1.51624043715847,
      "grad_norm": 0.7615288326992136,
      "learning_rate": 0.00014261124046797426,
      "loss": 12.4182,
      "step": 8671
    },
    {
      "epoch": 1.516415300546448,
      "grad_norm": 0.7320611983027832,
      "learning_rate": 0.00014259842910844468,
      "loss": 12.4016,
      "step": 8672
    },
    {
      "epoch": 1.5165901639344264,
      "grad_norm": 0.6933711133580963,
      "learning_rate": 0.00014258561689468578,
      "loss": 12.5168,
      "step": 8673
    },
    {
      "epoch": 1.5167650273224043,
      "grad_norm": 0.69577666336641,
      "learning_rate": 0.00014257280382695453,
      "loss": 12.085,
      "step": 8674
    },
    {
      "epoch": 1.5169398907103826,
      "grad_norm": 0.7951023143515363,
      "learning_rate": 0.00014255998990550785,
      "loss": 12.7319,
      "step": 8675
    },
    {
      "epoch": 1.5171147540983605,
      "grad_norm": 0.6973288177306002,
      "learning_rate": 0.00014254717513060268,
      "loss": 12.3111,
      "step": 8676
    },
    {
      "epoch": 1.5172896174863388,
      "grad_norm": 0.76815387296111,
      "learning_rate": 0.000142534359502496,
      "loss": 12.3363,
      "step": 8677
    },
    {
      "epoch": 1.517464480874317,
      "grad_norm": 0.6787485033466083,
      "learning_rate": 0.00014252154302144484,
      "loss": 12.3626,
      "step": 8678
    },
    {
      "epoch": 1.517639344262295,
      "grad_norm": 0.7477811969522015,
      "learning_rate": 0.0001425087256877062,
      "loss": 12.6015,
      "step": 8679
    },
    {
      "epoch": 1.5178142076502732,
      "grad_norm": 0.7245745597069226,
      "learning_rate": 0.00014249590750153705,
      "loss": 12.4173,
      "step": 8680
    },
    {
      "epoch": 1.5179890710382513,
      "grad_norm": 0.9408326776057057,
      "learning_rate": 0.0001424830884631945,
      "loss": 12.1765,
      "step": 8681
    },
    {
      "epoch": 1.5181639344262297,
      "grad_norm": 0.7732112737092659,
      "learning_rate": 0.0001424702685729356,
      "loss": 12.6496,
      "step": 8682
    },
    {
      "epoch": 1.5183387978142076,
      "grad_norm": 0.9391030836911627,
      "learning_rate": 0.00014245744783101735,
      "loss": 12.1585,
      "step": 8683
    },
    {
      "epoch": 1.518513661202186,
      "grad_norm": 1.190439276631646,
      "learning_rate": 0.000142444626237697,
      "loss": 12.3935,
      "step": 8684
    },
    {
      "epoch": 1.5186885245901638,
      "grad_norm": 0.86088952233062,
      "learning_rate": 0.00014243180379323148,
      "loss": 12.3429,
      "step": 8685
    },
    {
      "epoch": 1.5188633879781421,
      "grad_norm": 0.72762617497036,
      "learning_rate": 0.00014241898049787807,
      "loss": 12.414,
      "step": 8686
    },
    {
      "epoch": 1.5190382513661203,
      "grad_norm": 0.918730324873082,
      "learning_rate": 0.00014240615635189384,
      "loss": 12.5981,
      "step": 8687
    },
    {
      "epoch": 1.5192131147540984,
      "grad_norm": 0.8517225302196497,
      "learning_rate": 0.00014239333135553596,
      "loss": 12.2059,
      "step": 8688
    },
    {
      "epoch": 1.5193879781420765,
      "grad_norm": 0.8686738701959841,
      "learning_rate": 0.00014238050550906167,
      "loss": 12.4785,
      "step": 8689
    },
    {
      "epoch": 1.5195628415300546,
      "grad_norm": 0.7506526935411307,
      "learning_rate": 0.00014236767881272805,
      "loss": 12.4493,
      "step": 8690
    },
    {
      "epoch": 1.519737704918033,
      "grad_norm": 0.9453357532324703,
      "learning_rate": 0.00014235485126679243,
      "loss": 12.3157,
      "step": 8691
    },
    {
      "epoch": 1.5199125683060108,
      "grad_norm": 0.7319339095903916,
      "learning_rate": 0.000142342022871512,
      "loss": 12.4201,
      "step": 8692
    },
    {
      "epoch": 1.5200874316939892,
      "grad_norm": 0.803593416357443,
      "learning_rate": 0.00014232919362714398,
      "loss": 12.4537,
      "step": 8693
    },
    {
      "epoch": 1.520262295081967,
      "grad_norm": 0.890472795488744,
      "learning_rate": 0.00014231636353394567,
      "loss": 12.3901,
      "step": 8694
    },
    {
      "epoch": 1.5204371584699454,
      "grad_norm": 1.387448218929877,
      "learning_rate": 0.00014230353259217434,
      "loss": 12.4847,
      "step": 8695
    },
    {
      "epoch": 1.5206120218579235,
      "grad_norm": 0.9345185947545814,
      "learning_rate": 0.0001422907008020873,
      "loss": 12.6826,
      "step": 8696
    },
    {
      "epoch": 1.5207868852459017,
      "grad_norm": 2.2780866022724635,
      "learning_rate": 0.00014227786816394184,
      "loss": 12.5512,
      "step": 8697
    },
    {
      "epoch": 1.5209617486338798,
      "grad_norm": 0.8120139014637997,
      "learning_rate": 0.00014226503467799535,
      "loss": 12.5113,
      "step": 8698
    },
    {
      "epoch": 1.521136612021858,
      "grad_norm": 0.7481250318399169,
      "learning_rate": 0.00014225220034450514,
      "loss": 12.3096,
      "step": 8699
    },
    {
      "epoch": 1.521311475409836,
      "grad_norm": 1.3436546063506296,
      "learning_rate": 0.00014223936516372852,
      "loss": 12.6201,
      "step": 8700
    },
    {
      "epoch": 1.5214863387978141,
      "grad_norm": 0.7481679718681379,
      "learning_rate": 0.000142226529135923,
      "loss": 12.3151,
      "step": 8701
    },
    {
      "epoch": 1.5216612021857925,
      "grad_norm": 0.8265269765621395,
      "learning_rate": 0.0001422136922613459,
      "loss": 12.3046,
      "step": 8702
    },
    {
      "epoch": 1.5218360655737704,
      "grad_norm": 0.7926633411690696,
      "learning_rate": 0.00014220085454025465,
      "loss": 12.5425,
      "step": 8703
    },
    {
      "epoch": 1.5220109289617487,
      "grad_norm": 0.9391283665482648,
      "learning_rate": 0.0001421880159729067,
      "loss": 12.3986,
      "step": 8704
    },
    {
      "epoch": 1.5221857923497266,
      "grad_norm": 0.7141009011870654,
      "learning_rate": 0.0001421751765595595,
      "loss": 12.4343,
      "step": 8705
    },
    {
      "epoch": 1.522360655737705,
      "grad_norm": 0.8071961052671892,
      "learning_rate": 0.0001421623363004705,
      "loss": 12.3372,
      "step": 8706
    },
    {
      "epoch": 1.522535519125683,
      "grad_norm": 0.7463185404849796,
      "learning_rate": 0.00014214949519589722,
      "loss": 12.1593,
      "step": 8707
    },
    {
      "epoch": 1.5227103825136612,
      "grad_norm": 1.164425408886971,
      "learning_rate": 0.00014213665324609712,
      "loss": 12.2913,
      "step": 8708
    },
    {
      "epoch": 1.5228852459016393,
      "grad_norm": 0.8185337583486795,
      "learning_rate": 0.00014212381045132776,
      "loss": 12.0839,
      "step": 8709
    },
    {
      "epoch": 1.5230601092896174,
      "grad_norm": 0.727973534181438,
      "learning_rate": 0.00014211096681184668,
      "loss": 12.4466,
      "step": 8710
    },
    {
      "epoch": 1.5232349726775958,
      "grad_norm": 0.8178192404174303,
      "learning_rate": 0.0001420981223279114,
      "loss": 12.3054,
      "step": 8711
    },
    {
      "epoch": 1.5234098360655737,
      "grad_norm": 0.773700077168283,
      "learning_rate": 0.00014208527699977952,
      "loss": 12.5518,
      "step": 8712
    },
    {
      "epoch": 1.523584699453552,
      "grad_norm": 0.7211452549916576,
      "learning_rate": 0.0001420724308277086,
      "loss": 12.5153,
      "step": 8713
    },
    {
      "epoch": 1.52375956284153,
      "grad_norm": 0.9338227663488367,
      "learning_rate": 0.0001420595838119563,
      "loss": 12.3676,
      "step": 8714
    },
    {
      "epoch": 1.5239344262295083,
      "grad_norm": 0.725325376841725,
      "learning_rate": 0.00014204673595278016,
      "loss": 12.4505,
      "step": 8715
    },
    {
      "epoch": 1.5241092896174864,
      "grad_norm": 0.9092701537621393,
      "learning_rate": 0.00014203388725043787,
      "loss": 12.7538,
      "step": 8716
    },
    {
      "epoch": 1.5242841530054645,
      "grad_norm": 0.7304142129142672,
      "learning_rate": 0.00014202103770518708,
      "loss": 12.4353,
      "step": 8717
    },
    {
      "epoch": 1.5244590163934426,
      "grad_norm": 0.8660168286819174,
      "learning_rate": 0.00014200818731728546,
      "loss": 12.2421,
      "step": 8718
    },
    {
      "epoch": 1.5246338797814207,
      "grad_norm": 0.76831029997075,
      "learning_rate": 0.00014199533608699074,
      "loss": 12.4876,
      "step": 8719
    },
    {
      "epoch": 1.524808743169399,
      "grad_norm": 0.8234460243690443,
      "learning_rate": 0.00014198248401456055,
      "loss": 12.1903,
      "step": 8720
    },
    {
      "epoch": 1.524983606557377,
      "grad_norm": 0.7545529184586603,
      "learning_rate": 0.0001419696311002527,
      "loss": 12.4584,
      "step": 8721
    },
    {
      "epoch": 1.5251584699453553,
      "grad_norm": 0.7572825699831286,
      "learning_rate": 0.0001419567773443248,
      "loss": 12.2464,
      "step": 8722
    },
    {
      "epoch": 1.5253333333333332,
      "grad_norm": 0.7603865383076578,
      "learning_rate": 0.0001419439227470348,
      "loss": 12.4143,
      "step": 8723
    },
    {
      "epoch": 1.5255081967213115,
      "grad_norm": 0.7839088144801742,
      "learning_rate": 0.00014193106730864025,
      "loss": 12.1748,
      "step": 8724
    },
    {
      "epoch": 1.5256830601092897,
      "grad_norm": 0.8711760487205966,
      "learning_rate": 0.0001419182110293991,
      "loss": 12.2885,
      "step": 8725
    },
    {
      "epoch": 1.5258579234972678,
      "grad_norm": 0.7529051592256356,
      "learning_rate": 0.00014190535390956914,
      "loss": 12.4555,
      "step": 8726
    },
    {
      "epoch": 1.526032786885246,
      "grad_norm": 0.9049360344016694,
      "learning_rate": 0.00014189249594940813,
      "loss": 12.4552,
      "step": 8727
    },
    {
      "epoch": 1.526207650273224,
      "grad_norm": 0.7136729201250183,
      "learning_rate": 0.00014187963714917397,
      "loss": 12.2861,
      "step": 8728
    },
    {
      "epoch": 1.5263825136612021,
      "grad_norm": 0.7930793896943825,
      "learning_rate": 0.0001418667775091245,
      "loss": 12.2192,
      "step": 8729
    },
    {
      "epoch": 1.5265573770491803,
      "grad_norm": 0.741751574723391,
      "learning_rate": 0.00014185391702951761,
      "loss": 12.4338,
      "step": 8730
    },
    {
      "epoch": 1.5267322404371586,
      "grad_norm": 0.8939355613313263,
      "learning_rate": 0.00014184105571061115,
      "loss": 12.5638,
      "step": 8731
    },
    {
      "epoch": 1.5269071038251365,
      "grad_norm": 0.7541337877922308,
      "learning_rate": 0.00014182819355266308,
      "loss": 12.3413,
      "step": 8732
    },
    {
      "epoch": 1.5270819672131148,
      "grad_norm": 0.8198131108273838,
      "learning_rate": 0.00014181533055593123,
      "loss": 12.2498,
      "step": 8733
    },
    {
      "epoch": 1.5272568306010927,
      "grad_norm": 0.6972658456439088,
      "learning_rate": 0.00014180246672067367,
      "loss": 12.4985,
      "step": 8734
    },
    {
      "epoch": 1.527431693989071,
      "grad_norm": 0.8187762535173145,
      "learning_rate": 0.00014178960204714832,
      "loss": 12.4669,
      "step": 8735
    },
    {
      "epoch": 1.5276065573770492,
      "grad_norm": 0.7159876697147765,
      "learning_rate": 0.0001417767365356131,
      "loss": 12.1454,
      "step": 8736
    },
    {
      "epoch": 1.5277814207650273,
      "grad_norm": 0.6897015035342913,
      "learning_rate": 0.00014176387018632606,
      "loss": 12.416,
      "step": 8737
    },
    {
      "epoch": 1.5279562841530054,
      "grad_norm": 0.7641519596971487,
      "learning_rate": 0.00014175100299954516,
      "loss": 12.3622,
      "step": 8738
    },
    {
      "epoch": 1.5281311475409836,
      "grad_norm": 0.7339680769759358,
      "learning_rate": 0.0001417381349755285,
      "loss": 12.2092,
      "step": 8739
    },
    {
      "epoch": 1.528306010928962,
      "grad_norm": 0.7002308679660273,
      "learning_rate": 0.00014172526611453404,
      "loss": 12.3688,
      "step": 8740
    },
    {
      "epoch": 1.5284808743169398,
      "grad_norm": 0.7259140414376449,
      "learning_rate": 0.00014171239641681992,
      "loss": 12.577,
      "step": 8741
    },
    {
      "epoch": 1.5286557377049181,
      "grad_norm": 0.830206219686388,
      "learning_rate": 0.00014169952588264417,
      "loss": 12.4236,
      "step": 8742
    },
    {
      "epoch": 1.528830601092896,
      "grad_norm": 0.6937026448378201,
      "learning_rate": 0.00014168665451226486,
      "loss": 12.3442,
      "step": 8743
    },
    {
      "epoch": 1.5290054644808744,
      "grad_norm": 0.7667126285263227,
      "learning_rate": 0.00014167378230594017,
      "loss": 12.5202,
      "step": 8744
    },
    {
      "epoch": 1.5291803278688525,
      "grad_norm": 0.8016421229041675,
      "learning_rate": 0.00014166090926392814,
      "loss": 12.467,
      "step": 8745
    },
    {
      "epoch": 1.5293551912568306,
      "grad_norm": 0.7539703907491692,
      "learning_rate": 0.00014164803538648703,
      "loss": 12.3824,
      "step": 8746
    },
    {
      "epoch": 1.5295300546448087,
      "grad_norm": 0.7178033807443245,
      "learning_rate": 0.0001416351606738749,
      "loss": 12.074,
      "step": 8747
    },
    {
      "epoch": 1.5297049180327869,
      "grad_norm": 0.765252996733438,
      "learning_rate": 0.00014162228512634996,
      "loss": 12.27,
      "step": 8748
    },
    {
      "epoch": 1.529879781420765,
      "grad_norm": 0.7644140236081117,
      "learning_rate": 0.0001416094087441704,
      "loss": 12.419,
      "step": 8749
    },
    {
      "epoch": 1.530054644808743,
      "grad_norm": 0.7376430349243948,
      "learning_rate": 0.00014159653152759444,
      "loss": 12.4843,
      "step": 8750
    },
    {
      "epoch": 1.5302295081967214,
      "grad_norm": 0.8072843156454633,
      "learning_rate": 0.00014158365347688033,
      "loss": 12.5477,
      "step": 8751
    },
    {
      "epoch": 1.5304043715846993,
      "grad_norm": 0.6859584065856095,
      "learning_rate": 0.00014157077459228624,
      "loss": 12.4566,
      "step": 8752
    },
    {
      "epoch": 1.5305792349726777,
      "grad_norm": 0.718620386723654,
      "learning_rate": 0.00014155789487407052,
      "loss": 12.3754,
      "step": 8753
    },
    {
      "epoch": 1.5307540983606558,
      "grad_norm": 0.717833417696987,
      "learning_rate": 0.0001415450143224914,
      "loss": 12.6211,
      "step": 8754
    },
    {
      "epoch": 1.530928961748634,
      "grad_norm": 0.6746208348719975,
      "learning_rate": 0.00014153213293780718,
      "loss": 12.3838,
      "step": 8755
    },
    {
      "epoch": 1.531103825136612,
      "grad_norm": 0.7033217559655516,
      "learning_rate": 0.0001415192507202762,
      "loss": 12.3944,
      "step": 8756
    },
    {
      "epoch": 1.5312786885245901,
      "grad_norm": 0.7569701431711254,
      "learning_rate": 0.00014150636767015673,
      "loss": 12.5627,
      "step": 8757
    },
    {
      "epoch": 1.5314535519125683,
      "grad_norm": 0.7945992903534763,
      "learning_rate": 0.00014149348378770716,
      "loss": 12.3404,
      "step": 8758
    },
    {
      "epoch": 1.5316284153005464,
      "grad_norm": 0.7574401757904765,
      "learning_rate": 0.00014148059907318581,
      "loss": 12.5891,
      "step": 8759
    },
    {
      "epoch": 1.5318032786885247,
      "grad_norm": 0.7483156058320481,
      "learning_rate": 0.00014146771352685112,
      "loss": 12.4497,
      "step": 8760
    },
    {
      "epoch": 1.5319781420765026,
      "grad_norm": 0.655418522912103,
      "learning_rate": 0.00014145482714896144,
      "loss": 12.2649,
      "step": 8761
    },
    {
      "epoch": 1.532153005464481,
      "grad_norm": 0.7268964171324787,
      "learning_rate": 0.0001414419399397752,
      "loss": 12.4543,
      "step": 8762
    },
    {
      "epoch": 1.5323278688524589,
      "grad_norm": 0.652631038585297,
      "learning_rate": 0.0001414290518995508,
      "loss": 12.4117,
      "step": 8763
    },
    {
      "epoch": 1.5325027322404372,
      "grad_norm": 0.7901948524326445,
      "learning_rate": 0.00014141616302854672,
      "loss": 12.2468,
      "step": 8764
    },
    {
      "epoch": 1.5326775956284153,
      "grad_norm": 0.9738625437692656,
      "learning_rate": 0.0001414032733270214,
      "loss": 12.4529,
      "step": 8765
    },
    {
      "epoch": 1.5328524590163934,
      "grad_norm": 0.6635858390393765,
      "learning_rate": 0.00014139038279523334,
      "loss": 12.3504,
      "step": 8766
    },
    {
      "epoch": 1.5330273224043716,
      "grad_norm": 0.7566804172596565,
      "learning_rate": 0.00014137749143344102,
      "loss": 12.4489,
      "step": 8767
    },
    {
      "epoch": 1.5332021857923497,
      "grad_norm": 0.761632300899466,
      "learning_rate": 0.00014136459924190292,
      "loss": 12.382,
      "step": 8768
    },
    {
      "epoch": 1.533377049180328,
      "grad_norm": 0.8046232904239887,
      "learning_rate": 0.00014135170622087763,
      "loss": 12.3215,
      "step": 8769
    },
    {
      "epoch": 1.533551912568306,
      "grad_norm": 0.7077730215416997,
      "learning_rate": 0.00014133881237062365,
      "loss": 12.4003,
      "step": 8770
    },
    {
      "epoch": 1.5337267759562843,
      "grad_norm": 0.7129009560664722,
      "learning_rate": 0.00014132591769139958,
      "loss": 12.4631,
      "step": 8771
    },
    {
      "epoch": 1.5339016393442622,
      "grad_norm": 0.751401203560394,
      "learning_rate": 0.00014131302218346395,
      "loss": 12.0494,
      "step": 8772
    },
    {
      "epoch": 1.5340765027322405,
      "grad_norm": 0.7531564706431229,
      "learning_rate": 0.0001413001258470754,
      "loss": 12.4995,
      "step": 8773
    },
    {
      "epoch": 1.5342513661202186,
      "grad_norm": 0.7211760626240379,
      "learning_rate": 0.00014128722868249246,
      "loss": 12.5931,
      "step": 8774
    },
    {
      "epoch": 1.5344262295081967,
      "grad_norm": 0.7992122493035614,
      "learning_rate": 0.00014127433068997387,
      "loss": 12.5255,
      "step": 8775
    },
    {
      "epoch": 1.5346010928961749,
      "grad_norm": 0.7375561732393189,
      "learning_rate": 0.0001412614318697782,
      "loss": 12.5069,
      "step": 8776
    },
    {
      "epoch": 1.534775956284153,
      "grad_norm": 0.8301540979418984,
      "learning_rate": 0.00014124853222216414,
      "loss": 12.4026,
      "step": 8777
    },
    {
      "epoch": 1.534950819672131,
      "grad_norm": 0.7206814242858254,
      "learning_rate": 0.00014123563174739037,
      "loss": 12.5144,
      "step": 8778
    },
    {
      "epoch": 1.5351256830601092,
      "grad_norm": 0.6833893486810377,
      "learning_rate": 0.00014122273044571553,
      "loss": 12.2865,
      "step": 8779
    },
    {
      "epoch": 1.5353005464480876,
      "grad_norm": 0.7561120911330393,
      "learning_rate": 0.00014120982831739845,
      "loss": 12.3638,
      "step": 8780
    },
    {
      "epoch": 1.5354754098360655,
      "grad_norm": 0.7739371430191138,
      "learning_rate": 0.00014119692536269772,
      "loss": 12.3232,
      "step": 8781
    },
    {
      "epoch": 1.5356502732240438,
      "grad_norm": 0.794404752499583,
      "learning_rate": 0.00014118402158187215,
      "loss": 12.465,
      "step": 8782
    },
    {
      "epoch": 1.535825136612022,
      "grad_norm": 0.6394535468212249,
      "learning_rate": 0.00014117111697518054,
      "loss": 12.1404,
      "step": 8783
    },
    {
      "epoch": 1.536,
      "grad_norm": 0.775569863181222,
      "learning_rate": 0.00014115821154288155,
      "loss": 12.5202,
      "step": 8784
    },
    {
      "epoch": 1.5361748633879782,
      "grad_norm": 0.6458586483194421,
      "learning_rate": 0.0001411453052852341,
      "loss": 12.3647,
      "step": 8785
    },
    {
      "epoch": 1.5363497267759563,
      "grad_norm": 0.8816035643858622,
      "learning_rate": 0.00014113239820249694,
      "loss": 12.4256,
      "step": 8786
    },
    {
      "epoch": 1.5365245901639344,
      "grad_norm": 0.718475521766701,
      "learning_rate": 0.0001411194902949289,
      "loss": 12.4846,
      "step": 8787
    },
    {
      "epoch": 1.5366994535519125,
      "grad_norm": 0.8270136029563441,
      "learning_rate": 0.00014110658156278878,
      "loss": 12.5149,
      "step": 8788
    },
    {
      "epoch": 1.5368743169398908,
      "grad_norm": 0.745674245729135,
      "learning_rate": 0.00014109367200633557,
      "loss": 12.4271,
      "step": 8789
    },
    {
      "epoch": 1.5370491803278687,
      "grad_norm": 0.7576957719636704,
      "learning_rate": 0.00014108076162582802,
      "loss": 12.3615,
      "step": 8790
    },
    {
      "epoch": 1.537224043715847,
      "grad_norm": 0.7426152107893682,
      "learning_rate": 0.00014106785042152504,
      "loss": 12.5737,
      "step": 8791
    },
    {
      "epoch": 1.537398907103825,
      "grad_norm": 0.706714052094972,
      "learning_rate": 0.00014105493839368561,
      "loss": 12.4859,
      "step": 8792
    },
    {
      "epoch": 1.5375737704918033,
      "grad_norm": 0.7615232866156151,
      "learning_rate": 0.00014104202554256857,
      "loss": 12.5408,
      "step": 8793
    },
    {
      "epoch": 1.5377486338797814,
      "grad_norm": 0.7070305440110024,
      "learning_rate": 0.00014102911186843295,
      "loss": 12.2081,
      "step": 8794
    },
    {
      "epoch": 1.5379234972677596,
      "grad_norm": 0.7921498154998052,
      "learning_rate": 0.00014101619737153764,
      "loss": 12.5327,
      "step": 8795
    },
    {
      "epoch": 1.5380983606557377,
      "grad_norm": 0.6877459925939867,
      "learning_rate": 0.0001410032820521416,
      "loss": 12.2494,
      "step": 8796
    },
    {
      "epoch": 1.5382732240437158,
      "grad_norm": 0.7091949488656614,
      "learning_rate": 0.0001409903659105039,
      "loss": 12.0914,
      "step": 8797
    },
    {
      "epoch": 1.5384480874316941,
      "grad_norm": 0.7326796013676486,
      "learning_rate": 0.0001409774489468835,
      "loss": 12.4525,
      "step": 8798
    },
    {
      "epoch": 1.538622950819672,
      "grad_norm": 0.663760071846245,
      "learning_rate": 0.00014096453116153944,
      "loss": 12.0548,
      "step": 8799
    },
    {
      "epoch": 1.5387978142076504,
      "grad_norm": 0.6940953284526153,
      "learning_rate": 0.0001409516125547307,
      "loss": 12.3615,
      "step": 8800
    },
    {
      "epoch": 1.5389726775956283,
      "grad_norm": 0.7596309815660622,
      "learning_rate": 0.00014093869312671646,
      "loss": 12.4368,
      "step": 8801
    },
    {
      "epoch": 1.5391475409836066,
      "grad_norm": 0.6990190028090252,
      "learning_rate": 0.0001409257728777557,
      "loss": 12.4363,
      "step": 8802
    },
    {
      "epoch": 1.5393224043715847,
      "grad_norm": 0.6930291159323851,
      "learning_rate": 0.0001409128518081075,
      "loss": 12.5981,
      "step": 8803
    },
    {
      "epoch": 1.5394972677595629,
      "grad_norm": 0.7096379447038428,
      "learning_rate": 0.00014089992991803105,
      "loss": 12.5931,
      "step": 8804
    },
    {
      "epoch": 1.539672131147541,
      "grad_norm": 0.6595041674925821,
      "learning_rate": 0.00014088700720778542,
      "loss": 12.3807,
      "step": 8805
    },
    {
      "epoch": 1.539846994535519,
      "grad_norm": 0.660271680598311,
      "learning_rate": 0.00014087408367762972,
      "loss": 12.3672,
      "step": 8806
    },
    {
      "epoch": 1.5400218579234972,
      "grad_norm": 0.6665824756160801,
      "learning_rate": 0.00014086115932782314,
      "loss": 12.4658,
      "step": 8807
    },
    {
      "epoch": 1.5401967213114753,
      "grad_norm": 0.7096193682437298,
      "learning_rate": 0.0001408482341586249,
      "loss": 12.4794,
      "step": 8808
    },
    {
      "epoch": 1.5403715846994537,
      "grad_norm": 1.3779452046181273,
      "learning_rate": 0.00014083530817029412,
      "loss": 12.4801,
      "step": 8809
    },
    {
      "epoch": 1.5405464480874316,
      "grad_norm": 0.7676918015156095,
      "learning_rate": 0.00014082238136309006,
      "loss": 12.2256,
      "step": 8810
    },
    {
      "epoch": 1.54072131147541,
      "grad_norm": 0.7269560896165347,
      "learning_rate": 0.00014080945373727186,
      "loss": 12.275,
      "step": 8811
    },
    {
      "epoch": 1.5408961748633878,
      "grad_norm": 0.745477163475386,
      "learning_rate": 0.00014079652529309887,
      "loss": 12.3805,
      "step": 8812
    },
    {
      "epoch": 1.5410710382513662,
      "grad_norm": 0.669555094829224,
      "learning_rate": 0.00014078359603083022,
      "loss": 12.471,
      "step": 8813
    },
    {
      "epoch": 1.5412459016393443,
      "grad_norm": 0.6449918938411193,
      "learning_rate": 0.0001407706659507253,
      "loss": 12.4792,
      "step": 8814
    },
    {
      "epoch": 1.5414207650273224,
      "grad_norm": 0.8271359617969377,
      "learning_rate": 0.00014075773505304332,
      "loss": 12.356,
      "step": 8815
    },
    {
      "epoch": 1.5415956284153005,
      "grad_norm": 0.7415788820618834,
      "learning_rate": 0.00014074480333804362,
      "loss": 12.4164,
      "step": 8816
    },
    {
      "epoch": 1.5417704918032786,
      "grad_norm": 0.6845960936163301,
      "learning_rate": 0.00014073187080598548,
      "loss": 12.4222,
      "step": 8817
    },
    {
      "epoch": 1.541945355191257,
      "grad_norm": 0.8216955236197954,
      "learning_rate": 0.00014071893745712833,
      "loss": 12.3371,
      "step": 8818
    },
    {
      "epoch": 1.5421202185792349,
      "grad_norm": 0.8509407768126954,
      "learning_rate": 0.00014070600329173143,
      "loss": 12.2603,
      "step": 8819
    },
    {
      "epoch": 1.5422950819672132,
      "grad_norm": 0.8281789318572782,
      "learning_rate": 0.00014069306831005415,
      "loss": 12.4474,
      "step": 8820
    },
    {
      "epoch": 1.542469945355191,
      "grad_norm": 0.707103462071531,
      "learning_rate": 0.00014068013251235595,
      "loss": 12.2466,
      "step": 8821
    },
    {
      "epoch": 1.5426448087431694,
      "grad_norm": 0.838354435408726,
      "learning_rate": 0.00014066719589889614,
      "loss": 12.3144,
      "step": 8822
    },
    {
      "epoch": 1.5428196721311476,
      "grad_norm": 0.7030489617103993,
      "learning_rate": 0.00014065425846993424,
      "loss": 12.2821,
      "step": 8823
    },
    {
      "epoch": 1.5429945355191257,
      "grad_norm": 1.0046982445626387,
      "learning_rate": 0.0001406413202257296,
      "loss": 12.6055,
      "step": 8824
    },
    {
      "epoch": 1.5431693989071038,
      "grad_norm": 0.7702245146962239,
      "learning_rate": 0.0001406283811665417,
      "loss": 12.566,
      "step": 8825
    },
    {
      "epoch": 1.543344262295082,
      "grad_norm": 0.8180229843385627,
      "learning_rate": 0.00014061544129263006,
      "loss": 12.5355,
      "step": 8826
    },
    {
      "epoch": 1.5435191256830603,
      "grad_norm": 0.8521910361752953,
      "learning_rate": 0.00014060250060425407,
      "loss": 12.4155,
      "step": 8827
    },
    {
      "epoch": 1.5436939890710382,
      "grad_norm": 0.6670161235401769,
      "learning_rate": 0.00014058955910167332,
      "loss": 12.5007,
      "step": 8828
    },
    {
      "epoch": 1.5438688524590165,
      "grad_norm": 0.8245706245963701,
      "learning_rate": 0.0001405766167851472,
      "loss": 12.4765,
      "step": 8829
    },
    {
      "epoch": 1.5440437158469944,
      "grad_norm": 0.6946683808163016,
      "learning_rate": 0.00014056367365493543,
      "loss": 12.0562,
      "step": 8830
    },
    {
      "epoch": 1.5442185792349727,
      "grad_norm": 0.7341204399898464,
      "learning_rate": 0.00014055072971129738,
      "loss": 12.5281,
      "step": 8831
    },
    {
      "epoch": 1.5443934426229509,
      "grad_norm": 0.720060165420004,
      "learning_rate": 0.0001405377849544927,
      "loss": 12.5868,
      "step": 8832
    },
    {
      "epoch": 1.544568306010929,
      "grad_norm": 0.6854492539375288,
      "learning_rate": 0.000140524839384781,
      "loss": 12.3416,
      "step": 8833
    },
    {
      "epoch": 1.544743169398907,
      "grad_norm": 0.6939841121833173,
      "learning_rate": 0.0001405118930024218,
      "loss": 12.2599,
      "step": 8834
    },
    {
      "epoch": 1.5449180327868852,
      "grad_norm": 0.6946307461640027,
      "learning_rate": 0.00014049894580767482,
      "loss": 12.4589,
      "step": 8835
    },
    {
      "epoch": 1.5450928961748633,
      "grad_norm": 0.762427128002092,
      "learning_rate": 0.00014048599780079957,
      "loss": 12.463,
      "step": 8836
    },
    {
      "epoch": 1.5452677595628415,
      "grad_norm": 0.6730786848792634,
      "learning_rate": 0.00014047304898205578,
      "loss": 12.3643,
      "step": 8837
    },
    {
      "epoch": 1.5454426229508198,
      "grad_norm": 0.6962428128568647,
      "learning_rate": 0.0001404600993517031,
      "loss": 12.1986,
      "step": 8838
    },
    {
      "epoch": 1.5456174863387977,
      "grad_norm": 0.7985844539774316,
      "learning_rate": 0.00014044714891000117,
      "loss": 12.3912,
      "step": 8839
    },
    {
      "epoch": 1.545792349726776,
      "grad_norm": 0.6608556515356842,
      "learning_rate": 0.00014043419765720976,
      "loss": 12.1838,
      "step": 8840
    },
    {
      "epoch": 1.545967213114754,
      "grad_norm": 0.7195213582554657,
      "learning_rate": 0.00014042124559358846,
      "loss": 12.2911,
      "step": 8841
    },
    {
      "epoch": 1.5461420765027323,
      "grad_norm": 0.7388688892747838,
      "learning_rate": 0.00014040829271939716,
      "loss": 12.5408,
      "step": 8842
    },
    {
      "epoch": 1.5463169398907104,
      "grad_norm": 0.7272900153739329,
      "learning_rate": 0.00014039533903489548,
      "loss": 12.5413,
      "step": 8843
    },
    {
      "epoch": 1.5464918032786885,
      "grad_norm": 0.714535455748634,
      "learning_rate": 0.00014038238454034324,
      "loss": 12.6357,
      "step": 8844
    },
    {
      "epoch": 1.5466666666666666,
      "grad_norm": 0.7570125439279468,
      "learning_rate": 0.00014036942923600017,
      "loss": 12.373,
      "step": 8845
    },
    {
      "epoch": 1.5468415300546448,
      "grad_norm": 0.7479112009686687,
      "learning_rate": 0.00014035647312212612,
      "loss": 12.3857,
      "step": 8846
    },
    {
      "epoch": 1.547016393442623,
      "grad_norm": 0.7874778563853432,
      "learning_rate": 0.00014034351619898088,
      "loss": 12.294,
      "step": 8847
    },
    {
      "epoch": 1.547191256830601,
      "grad_norm": 0.6824713682334103,
      "learning_rate": 0.00014033055846682422,
      "loss": 12.2339,
      "step": 8848
    },
    {
      "epoch": 1.5473661202185793,
      "grad_norm": 0.8760180898013209,
      "learning_rate": 0.0001403175999259161,
      "loss": 12.4302,
      "step": 8849
    },
    {
      "epoch": 1.5475409836065572,
      "grad_norm": 0.6721658215509544,
      "learning_rate": 0.00014030464057651626,
      "loss": 12.4277,
      "step": 8850
    },
    {
      "epoch": 1.5477158469945356,
      "grad_norm": 0.7855734723330466,
      "learning_rate": 0.00014029168041888464,
      "loss": 12.2473,
      "step": 8851
    },
    {
      "epoch": 1.5478907103825137,
      "grad_norm": 0.7196424949114693,
      "learning_rate": 0.00014027871945328113,
      "loss": 12.1031,
      "step": 8852
    },
    {
      "epoch": 1.5480655737704918,
      "grad_norm": 0.7247260515901347,
      "learning_rate": 0.0001402657576799656,
      "loss": 12.2007,
      "step": 8853
    },
    {
      "epoch": 1.54824043715847,
      "grad_norm": 0.6701852457703538,
      "learning_rate": 0.00014025279509919799,
      "loss": 12.5599,
      "step": 8854
    },
    {
      "epoch": 1.548415300546448,
      "grad_norm": 0.7097552409197907,
      "learning_rate": 0.00014023983171123828,
      "loss": 12.339,
      "step": 8855
    },
    {
      "epoch": 1.5485901639344264,
      "grad_norm": 0.7995099420141749,
      "learning_rate": 0.00014022686751634638,
      "loss": 12.3105,
      "step": 8856
    },
    {
      "epoch": 1.5487650273224043,
      "grad_norm": 0.7417333308912782,
      "learning_rate": 0.00014021390251478226,
      "loss": 12.4378,
      "step": 8857
    },
    {
      "epoch": 1.5489398907103826,
      "grad_norm": 0.7617518392173137,
      "learning_rate": 0.00014020093670680592,
      "loss": 12.1992,
      "step": 8858
    },
    {
      "epoch": 1.5491147540983605,
      "grad_norm": 0.650258347120054,
      "learning_rate": 0.00014018797009267736,
      "loss": 12.2418,
      "step": 8859
    },
    {
      "epoch": 1.5492896174863389,
      "grad_norm": 0.7032356871039739,
      "learning_rate": 0.00014017500267265664,
      "loss": 12.3134,
      "step": 8860
    },
    {
      "epoch": 1.549464480874317,
      "grad_norm": 0.6657256870946314,
      "learning_rate": 0.00014016203444700373,
      "loss": 12.273,
      "step": 8861
    },
    {
      "epoch": 1.549639344262295,
      "grad_norm": 0.9479161984076917,
      "learning_rate": 0.00014014906541597873,
      "loss": 12.306,
      "step": 8862
    },
    {
      "epoch": 1.5498142076502732,
      "grad_norm": 0.6875637143211965,
      "learning_rate": 0.00014013609557984172,
      "loss": 12.246,
      "step": 8863
    },
    {
      "epoch": 1.5499890710382513,
      "grad_norm": 0.7300240470319216,
      "learning_rate": 0.00014012312493885272,
      "loss": 12.3605,
      "step": 8864
    },
    {
      "epoch": 1.5501639344262295,
      "grad_norm": 0.6757347203499025,
      "learning_rate": 0.00014011015349327187,
      "loss": 12.1252,
      "step": 8865
    },
    {
      "epoch": 1.5503387978142076,
      "grad_norm": 0.7369149810917808,
      "learning_rate": 0.00014009718124335932,
      "loss": 12.4955,
      "step": 8866
    },
    {
      "epoch": 1.550513661202186,
      "grad_norm": 0.848778045644381,
      "learning_rate": 0.00014008420818937517,
      "loss": 12.3722,
      "step": 8867
    },
    {
      "epoch": 1.5506885245901638,
      "grad_norm": 0.9253508728386723,
      "learning_rate": 0.00014007123433157953,
      "loss": 12.2402,
      "step": 8868
    },
    {
      "epoch": 1.5508633879781422,
      "grad_norm": 0.8908890319825636,
      "learning_rate": 0.00014005825967023264,
      "loss": 12.4191,
      "step": 8869
    },
    {
      "epoch": 1.55103825136612,
      "grad_norm": 0.8621676876842748,
      "learning_rate": 0.00014004528420559465,
      "loss": 12.4731,
      "step": 8870
    },
    {
      "epoch": 1.5512131147540984,
      "grad_norm": 0.8137568101331167,
      "learning_rate": 0.00014003230793792573,
      "loss": 12.313,
      "step": 8871
    },
    {
      "epoch": 1.5513879781420765,
      "grad_norm": 0.7971712485933673,
      "learning_rate": 0.00014001933086748615,
      "loss": 12.5055,
      "step": 8872
    },
    {
      "epoch": 1.5515628415300546,
      "grad_norm": 0.8312606140501265,
      "learning_rate": 0.00014000635299453608,
      "loss": 12.2377,
      "step": 8873
    },
    {
      "epoch": 1.5517377049180328,
      "grad_norm": 0.7135313744999042,
      "learning_rate": 0.00013999337431933582,
      "loss": 12.394,
      "step": 8874
    },
    {
      "epoch": 1.5519125683060109,
      "grad_norm": 0.8991677642940076,
      "learning_rate": 0.0001399803948421456,
      "loss": 12.4765,
      "step": 8875
    },
    {
      "epoch": 1.5520874316939892,
      "grad_norm": 0.6819166810183775,
      "learning_rate": 0.00013996741456322574,
      "loss": 12.3739,
      "step": 8876
    },
    {
      "epoch": 1.5522622950819671,
      "grad_norm": 0.9451757839285339,
      "learning_rate": 0.00013995443348283645,
      "loss": 12.4783,
      "step": 8877
    },
    {
      "epoch": 1.5524371584699455,
      "grad_norm": 0.6950040417845984,
      "learning_rate": 0.00013994145160123816,
      "loss": 12.0788,
      "step": 8878
    },
    {
      "epoch": 1.5526120218579234,
      "grad_norm": 0.8093443520416714,
      "learning_rate": 0.00013992846891869106,
      "loss": 12.4426,
      "step": 8879
    },
    {
      "epoch": 1.5527868852459017,
      "grad_norm": 0.7233349018280693,
      "learning_rate": 0.0001399154854354556,
      "loss": 12.5764,
      "step": 8880
    },
    {
      "epoch": 1.5529617486338798,
      "grad_norm": 0.7442299321948725,
      "learning_rate": 0.0001399025011517921,
      "loss": 12.3858,
      "step": 8881
    },
    {
      "epoch": 1.553136612021858,
      "grad_norm": 0.9311867407410849,
      "learning_rate": 0.00013988951606796088,
      "loss": 12.434,
      "step": 8882
    },
    {
      "epoch": 1.553311475409836,
      "grad_norm": 0.6700749852116151,
      "learning_rate": 0.00013987653018422247,
      "loss": 12.4546,
      "step": 8883
    },
    {
      "epoch": 1.5534863387978142,
      "grad_norm": 0.7106470882455176,
      "learning_rate": 0.00013986354350083714,
      "loss": 11.9735,
      "step": 8884
    },
    {
      "epoch": 1.5536612021857925,
      "grad_norm": 0.6947333513000319,
      "learning_rate": 0.0001398505560180654,
      "loss": 12.3985,
      "step": 8885
    },
    {
      "epoch": 1.5538360655737704,
      "grad_norm": 0.6452091935826239,
      "learning_rate": 0.00013983756773616762,
      "loss": 12.4286,
      "step": 8886
    },
    {
      "epoch": 1.5540109289617487,
      "grad_norm": 0.6880491778852029,
      "learning_rate": 0.00013982457865540428,
      "loss": 12.6874,
      "step": 8887
    },
    {
      "epoch": 1.5541857923497266,
      "grad_norm": 0.6707098968027188,
      "learning_rate": 0.0001398115887760359,
      "loss": 12.4038,
      "step": 8888
    },
    {
      "epoch": 1.554360655737705,
      "grad_norm": 0.7040547523610539,
      "learning_rate": 0.00013979859809832288,
      "loss": 12.0678,
      "step": 8889
    },
    {
      "epoch": 1.554535519125683,
      "grad_norm": 0.7061258899905107,
      "learning_rate": 0.00013978560662252583,
      "loss": 12.1581,
      "step": 8890
    },
    {
      "epoch": 1.5547103825136612,
      "grad_norm": 0.7178581850324357,
      "learning_rate": 0.00013977261434890517,
      "loss": 12.6399,
      "step": 8891
    },
    {
      "epoch": 1.5548852459016393,
      "grad_norm": 0.7244657915851019,
      "learning_rate": 0.00013975962127772148,
      "loss": 12.4761,
      "step": 8892
    },
    {
      "epoch": 1.5550601092896175,
      "grad_norm": 0.7094454322502944,
      "learning_rate": 0.0001397466274092353,
      "loss": 12.4746,
      "step": 8893
    },
    {
      "epoch": 1.5552349726775956,
      "grad_norm": 0.6319763970386676,
      "learning_rate": 0.00013973363274370721,
      "loss": 12.1106,
      "step": 8894
    },
    {
      "epoch": 1.5554098360655737,
      "grad_norm": 0.6694155205097505,
      "learning_rate": 0.0001397206372813978,
      "loss": 12.0908,
      "step": 8895
    },
    {
      "epoch": 1.555584699453552,
      "grad_norm": 0.6623471574465869,
      "learning_rate": 0.00013970764102256764,
      "loss": 12.1509,
      "step": 8896
    },
    {
      "epoch": 1.55575956284153,
      "grad_norm": 0.7311070353903115,
      "learning_rate": 0.00013969464396747735,
      "loss": 12.299,
      "step": 8897
    },
    {
      "epoch": 1.5559344262295083,
      "grad_norm": 0.6930533429253549,
      "learning_rate": 0.00013968164611638759,
      "loss": 12.4073,
      "step": 8898
    },
    {
      "epoch": 1.5561092896174862,
      "grad_norm": 0.8809620974181323,
      "learning_rate": 0.00013966864746955897,
      "loss": 12.2799,
      "step": 8899
    },
    {
      "epoch": 1.5562841530054645,
      "grad_norm": 0.6724583187449991,
      "learning_rate": 0.00013965564802725217,
      "loss": 12.3022,
      "step": 8900
    },
    {
      "epoch": 1.5564590163934426,
      "grad_norm": 0.7999512847812947,
      "learning_rate": 0.00013964264778972785,
      "loss": 12.3688,
      "step": 8901
    },
    {
      "epoch": 1.5566338797814208,
      "grad_norm": 0.6836341540577534,
      "learning_rate": 0.00013962964675724677,
      "loss": 12.2664,
      "step": 8902
    },
    {
      "epoch": 1.5568087431693989,
      "grad_norm": 0.7246219964814615,
      "learning_rate": 0.00013961664493006957,
      "loss": 12.4632,
      "step": 8903
    },
    {
      "epoch": 1.556983606557377,
      "grad_norm": 0.7266161341290631,
      "learning_rate": 0.000139603642308457,
      "loss": 12.4407,
      "step": 8904
    },
    {
      "epoch": 1.5571584699453553,
      "grad_norm": 0.6627109928637501,
      "learning_rate": 0.00013959063889266982,
      "loss": 12.433,
      "step": 8905
    },
    {
      "epoch": 1.5573333333333332,
      "grad_norm": 0.6772551495750825,
      "learning_rate": 0.00013957763468296875,
      "loss": 12.5189,
      "step": 8906
    },
    {
      "epoch": 1.5575081967213116,
      "grad_norm": 0.6885142227680481,
      "learning_rate": 0.00013956462967961456,
      "loss": 12.2701,
      "step": 8907
    },
    {
      "epoch": 1.5576830601092895,
      "grad_norm": 0.6546911887610204,
      "learning_rate": 0.0001395516238828681,
      "loss": 12.363,
      "step": 8908
    },
    {
      "epoch": 1.5578579234972678,
      "grad_norm": 0.7211268794939008,
      "learning_rate": 0.00013953861729299014,
      "loss": 12.1954,
      "step": 8909
    },
    {
      "epoch": 1.558032786885246,
      "grad_norm": 0.6824613176859052,
      "learning_rate": 0.0001395256099102415,
      "loss": 12.5546,
      "step": 8910
    },
    {
      "epoch": 1.558207650273224,
      "grad_norm": 0.7269497451013175,
      "learning_rate": 0.00013951260173488303,
      "loss": 12.5254,
      "step": 8911
    },
    {
      "epoch": 1.5583825136612022,
      "grad_norm": 0.704625871484811,
      "learning_rate": 0.00013949959276717558,
      "loss": 12.2138,
      "step": 8912
    },
    {
      "epoch": 1.5585573770491803,
      "grad_norm": 0.675964470885383,
      "learning_rate": 0.00013948658300737998,
      "loss": 12.3607,
      "step": 8913
    },
    {
      "epoch": 1.5587322404371586,
      "grad_norm": 0.7088531975456459,
      "learning_rate": 0.00013947357245575718,
      "loss": 12.499,
      "step": 8914
    },
    {
      "epoch": 1.5589071038251365,
      "grad_norm": 0.7130659745629162,
      "learning_rate": 0.00013946056111256805,
      "loss": 12.3666,
      "step": 8915
    },
    {
      "epoch": 1.5590819672131149,
      "grad_norm": 0.780629682260589,
      "learning_rate": 0.0001394475489780735,
      "loss": 12.4271,
      "step": 8916
    },
    {
      "epoch": 1.5592568306010928,
      "grad_norm": 0.7289758252490632,
      "learning_rate": 0.0001394345360525345,
      "loss": 12.3418,
      "step": 8917
    },
    {
      "epoch": 1.559431693989071,
      "grad_norm": 0.7632310503679102,
      "learning_rate": 0.00013942152233621194,
      "loss": 12.5402,
      "step": 8918
    },
    {
      "epoch": 1.5596065573770492,
      "grad_norm": 0.6471698756928412,
      "learning_rate": 0.00013940850782936682,
      "loss": 12.4646,
      "step": 8919
    },
    {
      "epoch": 1.5597814207650273,
      "grad_norm": 0.673501781808649,
      "learning_rate": 0.00013939549253226012,
      "loss": 12.5106,
      "step": 8920
    },
    {
      "epoch": 1.5599562841530055,
      "grad_norm": 0.710598713086398,
      "learning_rate": 0.00013938247644515282,
      "loss": 12.5231,
      "step": 8921
    },
    {
      "epoch": 1.5601311475409836,
      "grad_norm": 0.8006966393405953,
      "learning_rate": 0.00013936945956830602,
      "loss": 12.447,
      "step": 8922
    },
    {
      "epoch": 1.5603060109289617,
      "grad_norm": 0.8097382950825921,
      "learning_rate": 0.0001393564419019806,
      "loss": 12.0527,
      "step": 8923
    },
    {
      "epoch": 1.5604808743169398,
      "grad_norm": 0.7644252723869918,
      "learning_rate": 0.00013934342344643771,
      "loss": 12.2036,
      "step": 8924
    },
    {
      "epoch": 1.5606557377049182,
      "grad_norm": 0.7452802373766435,
      "learning_rate": 0.00013933040420193837,
      "loss": 12.3895,
      "step": 8925
    },
    {
      "epoch": 1.560830601092896,
      "grad_norm": 0.7664783252530024,
      "learning_rate": 0.00013931738416874366,
      "loss": 12.5122,
      "step": 8926
    },
    {
      "epoch": 1.5610054644808744,
      "grad_norm": 0.7669827188865873,
      "learning_rate": 0.00013930436334711472,
      "loss": 12.4142,
      "step": 8927
    },
    {
      "epoch": 1.5611803278688523,
      "grad_norm": 0.6850839418339935,
      "learning_rate": 0.00013929134173731257,
      "loss": 12.0465,
      "step": 8928
    },
    {
      "epoch": 1.5613551912568306,
      "grad_norm": 0.6591017098061558,
      "learning_rate": 0.0001392783193395984,
      "loss": 12.5359,
      "step": 8929
    },
    {
      "epoch": 1.5615300546448088,
      "grad_norm": 0.7470412150284893,
      "learning_rate": 0.00013926529615423326,
      "loss": 12.4197,
      "step": 8930
    },
    {
      "epoch": 1.5617049180327869,
      "grad_norm": 0.681986353034611,
      "learning_rate": 0.00013925227218147847,
      "loss": 12.2973,
      "step": 8931
    },
    {
      "epoch": 1.561879781420765,
      "grad_norm": 0.7166895451757969,
      "learning_rate": 0.00013923924742159505,
      "loss": 12.6172,
      "step": 8932
    },
    {
      "epoch": 1.5620546448087431,
      "grad_norm": 0.8454631053139777,
      "learning_rate": 0.00013922622187484426,
      "loss": 12.4389,
      "step": 8933
    },
    {
      "epoch": 1.5622295081967215,
      "grad_norm": 0.7711229580416953,
      "learning_rate": 0.00013921319554148726,
      "loss": 12.2127,
      "step": 8934
    },
    {
      "epoch": 1.5624043715846994,
      "grad_norm": 0.7137794688128366,
      "learning_rate": 0.0001392001684217853,
      "loss": 12.4514,
      "step": 8935
    },
    {
      "epoch": 1.5625792349726777,
      "grad_norm": 0.7409775762939524,
      "learning_rate": 0.0001391871405159996,
      "loss": 12.4087,
      "step": 8936
    },
    {
      "epoch": 1.5627540983606556,
      "grad_norm": 0.798440510922124,
      "learning_rate": 0.0001391741118243914,
      "loss": 12.4643,
      "step": 8937
    },
    {
      "epoch": 1.562928961748634,
      "grad_norm": 0.7502338784685268,
      "learning_rate": 0.000139161082347222,
      "loss": 12.4363,
      "step": 8938
    },
    {
      "epoch": 1.563103825136612,
      "grad_norm": 0.7355808424881083,
      "learning_rate": 0.00013914805208475265,
      "loss": 12.3516,
      "step": 8939
    },
    {
      "epoch": 1.5632786885245902,
      "grad_norm": 0.6855711151016544,
      "learning_rate": 0.00013913502103724468,
      "loss": 12.5285,
      "step": 8940
    },
    {
      "epoch": 1.5634535519125683,
      "grad_norm": 0.7737792444343529,
      "learning_rate": 0.00013912198920495934,
      "loss": 12.0362,
      "step": 8941
    },
    {
      "epoch": 1.5636284153005464,
      "grad_norm": 0.6699009904130999,
      "learning_rate": 0.000139108956588158,
      "loss": 12.3924,
      "step": 8942
    },
    {
      "epoch": 1.5638032786885248,
      "grad_norm": 0.651024623237768,
      "learning_rate": 0.00013909592318710204,
      "loss": 12.3206,
      "step": 8943
    },
    {
      "epoch": 1.5639781420765027,
      "grad_norm": 0.6864548116233996,
      "learning_rate": 0.00013908288900205273,
      "loss": 12.228,
      "step": 8944
    },
    {
      "epoch": 1.564153005464481,
      "grad_norm": 0.7183442483062544,
      "learning_rate": 0.0001390698540332715,
      "loss": 12.583,
      "step": 8945
    },
    {
      "epoch": 1.564327868852459,
      "grad_norm": 0.7801757468634645,
      "learning_rate": 0.00013905681828101977,
      "loss": 12.427,
      "step": 8946
    },
    {
      "epoch": 1.5645027322404372,
      "grad_norm": 0.7260121673449309,
      "learning_rate": 0.0001390437817455589,
      "loss": 12.2467,
      "step": 8947
    },
    {
      "epoch": 1.5646775956284154,
      "grad_norm": 0.6765209783252017,
      "learning_rate": 0.00013903074442715032,
      "loss": 12.4556,
      "step": 8948
    },
    {
      "epoch": 1.5648524590163935,
      "grad_norm": 0.8263405960553877,
      "learning_rate": 0.00013901770632605547,
      "loss": 12.3763,
      "step": 8949
    },
    {
      "epoch": 1.5650273224043716,
      "grad_norm": 0.8840956948008707,
      "learning_rate": 0.0001390046674425358,
      "loss": 12.2235,
      "step": 8950
    },
    {
      "epoch": 1.5652021857923497,
      "grad_norm": 0.7151508027436103,
      "learning_rate": 0.00013899162777685278,
      "loss": 12.3273,
      "step": 8951
    },
    {
      "epoch": 1.5653770491803278,
      "grad_norm": 0.7796841009587784,
      "learning_rate": 0.00013897858732926793,
      "loss": 12.5088,
      "step": 8952
    },
    {
      "epoch": 1.565551912568306,
      "grad_norm": 0.6855729080151043,
      "learning_rate": 0.00013896554610004275,
      "loss": 12.5731,
      "step": 8953
    },
    {
      "epoch": 1.5657267759562843,
      "grad_norm": 0.7850989299932923,
      "learning_rate": 0.00013895250408943867,
      "loss": 12.4096,
      "step": 8954
    },
    {
      "epoch": 1.5659016393442622,
      "grad_norm": 0.6868785897272043,
      "learning_rate": 0.00013893946129771734,
      "loss": 12.3391,
      "step": 8955
    },
    {
      "epoch": 1.5660765027322405,
      "grad_norm": 0.7943221154374615,
      "learning_rate": 0.00013892641772514022,
      "loss": 12.5163,
      "step": 8956
    },
    {
      "epoch": 1.5662513661202184,
      "grad_norm": 0.778632113847875,
      "learning_rate": 0.00013891337337196887,
      "loss": 12.395,
      "step": 8957
    },
    {
      "epoch": 1.5664262295081968,
      "grad_norm": 0.8801184243329808,
      "learning_rate": 0.00013890032823846496,
      "loss": 12.1447,
      "step": 8958
    },
    {
      "epoch": 1.5666010928961749,
      "grad_norm": 0.8886465134889379,
      "learning_rate": 0.00013888728232489,
      "loss": 12.4306,
      "step": 8959
    },
    {
      "epoch": 1.566775956284153,
      "grad_norm": 0.7249152094480742,
      "learning_rate": 0.00013887423563150567,
      "loss": 12.229,
      "step": 8960
    },
    {
      "epoch": 1.5669508196721311,
      "grad_norm": 0.8636466829470011,
      "learning_rate": 0.00013886118815857352,
      "loss": 12.3164,
      "step": 8961
    },
    {
      "epoch": 1.5671256830601092,
      "grad_norm": 0.6740542025212453,
      "learning_rate": 0.0001388481399063552,
      "loss": 12.0427,
      "step": 8962
    },
    {
      "epoch": 1.5673005464480876,
      "grad_norm": 0.8773012315552674,
      "learning_rate": 0.00013883509087511247,
      "loss": 12.4731,
      "step": 8963
    },
    {
      "epoch": 1.5674754098360655,
      "grad_norm": 0.6926804352072244,
      "learning_rate": 0.0001388220410651069,
      "loss": 12.2095,
      "step": 8964
    },
    {
      "epoch": 1.5676502732240438,
      "grad_norm": 0.8765221858564017,
      "learning_rate": 0.00013880899047660018,
      "loss": 12.4629,
      "step": 8965
    },
    {
      "epoch": 1.5678251366120217,
      "grad_norm": 0.8714545086050877,
      "learning_rate": 0.00013879593910985405,
      "loss": 12.4197,
      "step": 8966
    },
    {
      "epoch": 1.568,
      "grad_norm": 0.8067086901215355,
      "learning_rate": 0.00013878288696513022,
      "loss": 12.2741,
      "step": 8967
    },
    {
      "epoch": 1.5681748633879782,
      "grad_norm": 0.7975820797896019,
      "learning_rate": 0.00013876983404269045,
      "loss": 12.2994,
      "step": 8968
    },
    {
      "epoch": 1.5683497267759563,
      "grad_norm": 0.7934787156109949,
      "learning_rate": 0.00013875678034279645,
      "loss": 12.3301,
      "step": 8969
    },
    {
      "epoch": 1.5685245901639344,
      "grad_norm": 0.7811132549868725,
      "learning_rate": 0.00013874372586571002,
      "loss": 12.5673,
      "step": 8970
    },
    {
      "epoch": 1.5686994535519125,
      "grad_norm": 0.6721126430024765,
      "learning_rate": 0.00013873067061169288,
      "loss": 12.459,
      "step": 8971
    },
    {
      "epoch": 1.5688743169398907,
      "grad_norm": 1.005543151968233,
      "learning_rate": 0.00013871761458100697,
      "loss": 12.3282,
      "step": 8972
    },
    {
      "epoch": 1.5690491803278688,
      "grad_norm": 0.7682065626096218,
      "learning_rate": 0.0001387045577739139,
      "loss": 12.4772,
      "step": 8973
    },
    {
      "epoch": 1.5692240437158471,
      "grad_norm": 0.9299822486632022,
      "learning_rate": 0.0001386915001906757,
      "loss": 12.484,
      "step": 8974
    },
    {
      "epoch": 1.569398907103825,
      "grad_norm": 0.7554622221946447,
      "learning_rate": 0.00013867844183155408,
      "loss": 12.3168,
      "step": 8975
    },
    {
      "epoch": 1.5695737704918034,
      "grad_norm": 0.7373817928341205,
      "learning_rate": 0.0001386653826968109,
      "loss": 12.2654,
      "step": 8976
    },
    {
      "epoch": 1.5697486338797815,
      "grad_norm": 0.9041193621530925,
      "learning_rate": 0.00013865232278670816,
      "loss": 12.4848,
      "step": 8977
    },
    {
      "epoch": 1.5699234972677596,
      "grad_norm": 0.7412367382124799,
      "learning_rate": 0.00013863926210150762,
      "loss": 12.403,
      "step": 8978
    },
    {
      "epoch": 1.5700983606557377,
      "grad_norm": 0.8972376823020374,
      "learning_rate": 0.00013862620064147128,
      "loss": 12.2907,
      "step": 8979
    },
    {
      "epoch": 1.5702732240437158,
      "grad_norm": 0.6993173696225413,
      "learning_rate": 0.00013861313840686097,
      "loss": 12.5156,
      "step": 8980
    },
    {
      "epoch": 1.570448087431694,
      "grad_norm": 0.81291580381691,
      "learning_rate": 0.00013860007539793871,
      "loss": 11.8188,
      "step": 8981
    },
    {
      "epoch": 1.570622950819672,
      "grad_norm": 0.7731079984010879,
      "learning_rate": 0.0001385870116149664,
      "loss": 12.4701,
      "step": 8982
    },
    {
      "epoch": 1.5707978142076504,
      "grad_norm": 0.7839467405451381,
      "learning_rate": 0.00013857394705820604,
      "loss": 12.335,
      "step": 8983
    },
    {
      "epoch": 1.5709726775956283,
      "grad_norm": 0.6839264772381679,
      "learning_rate": 0.0001385608817279196,
      "loss": 12.4059,
      "step": 8984
    },
    {
      "epoch": 1.5711475409836066,
      "grad_norm": 0.8342005032606846,
      "learning_rate": 0.00013854781562436906,
      "loss": 12.3384,
      "step": 8985
    },
    {
      "epoch": 1.5713224043715845,
      "grad_norm": 0.7767650387494464,
      "learning_rate": 0.0001385347487478165,
      "loss": 12.2352,
      "step": 8986
    },
    {
      "epoch": 1.5714972677595629,
      "grad_norm": 0.7966788503365912,
      "learning_rate": 0.0001385216810985239,
      "loss": 12.2511,
      "step": 8987
    },
    {
      "epoch": 1.571672131147541,
      "grad_norm": 0.7364711092873621,
      "learning_rate": 0.0001385086126767533,
      "loss": 12.4698,
      "step": 8988
    },
    {
      "epoch": 1.5718469945355191,
      "grad_norm": 0.6713181273971549,
      "learning_rate": 0.0001384955434827668,
      "loss": 12.2406,
      "step": 8989
    },
    {
      "epoch": 1.5720218579234972,
      "grad_norm": 0.6733888376576777,
      "learning_rate": 0.00013848247351682644,
      "loss": 12.3796,
      "step": 8990
    },
    {
      "epoch": 1.5721967213114754,
      "grad_norm": 0.7345310014501538,
      "learning_rate": 0.00013846940277919433,
      "loss": 12.5473,
      "step": 8991
    },
    {
      "epoch": 1.5723715846994537,
      "grad_norm": 0.7302424162423633,
      "learning_rate": 0.00013845633127013257,
      "loss": 12.4701,
      "step": 8992
    },
    {
      "epoch": 1.5725464480874316,
      "grad_norm": 0.6614911539293982,
      "learning_rate": 0.0001384432589899033,
      "loss": 12.3156,
      "step": 8993
    },
    {
      "epoch": 1.57272131147541,
      "grad_norm": 0.6981795734852795,
      "learning_rate": 0.00013843018593876868,
      "loss": 12.6481,
      "step": 8994
    },
    {
      "epoch": 1.5728961748633878,
      "grad_norm": 0.6620390022215182,
      "learning_rate": 0.00013841711211699082,
      "loss": 12.4886,
      "step": 8995
    },
    {
      "epoch": 1.5730710382513662,
      "grad_norm": 0.6717838137237373,
      "learning_rate": 0.00013840403752483189,
      "loss": 12.4724,
      "step": 8996
    },
    {
      "epoch": 1.5732459016393443,
      "grad_norm": 0.7578100475282094,
      "learning_rate": 0.0001383909621625541,
      "loss": 12.3076,
      "step": 8997
    },
    {
      "epoch": 1.5734207650273224,
      "grad_norm": 0.725037023801091,
      "learning_rate": 0.00013837788603041963,
      "loss": 12.3418,
      "step": 8998
    },
    {
      "epoch": 1.5735956284153005,
      "grad_norm": 0.7381196278618611,
      "learning_rate": 0.00013836480912869073,
      "loss": 12.3836,
      "step": 8999
    },
    {
      "epoch": 1.5737704918032787,
      "grad_norm": 0.7617725720607298,
      "learning_rate": 0.00013835173145762962,
      "loss": 12.4118,
      "step": 9000
    },
    {
      "epoch": 1.5739453551912568,
      "grad_norm": 0.6581200443614894,
      "learning_rate": 0.0001383386530174985,
      "loss": 12.26,
      "step": 9001
    },
    {
      "epoch": 1.574120218579235,
      "grad_norm": 0.6655808699482914,
      "learning_rate": 0.0001383255738085597,
      "loss": 12.4427,
      "step": 9002
    },
    {
      "epoch": 1.5742950819672132,
      "grad_norm": 0.6718515141161531,
      "learning_rate": 0.00013831249383107545,
      "loss": 12.4864,
      "step": 9003
    },
    {
      "epoch": 1.5744699453551911,
      "grad_norm": 0.6825771584140549,
      "learning_rate": 0.0001382994130853081,
      "loss": 12.2526,
      "step": 9004
    },
    {
      "epoch": 1.5746448087431695,
      "grad_norm": 0.6504667775554087,
      "learning_rate": 0.00013828633157151988,
      "loss": 12.3005,
      "step": 9005
    },
    {
      "epoch": 1.5748196721311474,
      "grad_norm": 0.7634734783610729,
      "learning_rate": 0.0001382732492899732,
      "loss": 12.0744,
      "step": 9006
    },
    {
      "epoch": 1.5749945355191257,
      "grad_norm": 0.6637430096424413,
      "learning_rate": 0.00013826016624093033,
      "loss": 12.4784,
      "step": 9007
    },
    {
      "epoch": 1.5751693989071038,
      "grad_norm": 0.7041141110781768,
      "learning_rate": 0.00013824708242465364,
      "loss": 12.151,
      "step": 9008
    },
    {
      "epoch": 1.575344262295082,
      "grad_norm": 0.7470774192905819,
      "learning_rate": 0.00013823399784140556,
      "loss": 12.4933,
      "step": 9009
    },
    {
      "epoch": 1.57551912568306,
      "grad_norm": 0.6278048432594486,
      "learning_rate": 0.00013822091249144838,
      "loss": 12.3771,
      "step": 9010
    },
    {
      "epoch": 1.5756939890710382,
      "grad_norm": 0.7336872202247464,
      "learning_rate": 0.0001382078263750446,
      "loss": 12.555,
      "step": 9011
    },
    {
      "epoch": 1.5758688524590165,
      "grad_norm": 0.7536149990693803,
      "learning_rate": 0.00013819473949245654,
      "loss": 12.6568,
      "step": 9012
    },
    {
      "epoch": 1.5760437158469944,
      "grad_norm": 0.7098948968193229,
      "learning_rate": 0.0001381816518439467,
      "loss": 12.4184,
      "step": 9013
    },
    {
      "epoch": 1.5762185792349728,
      "grad_norm": 0.701789578636101,
      "learning_rate": 0.0001381685634297775,
      "loss": 12.5402,
      "step": 9014
    },
    {
      "epoch": 1.5763934426229507,
      "grad_norm": 0.7148227888879223,
      "learning_rate": 0.00013815547425021144,
      "loss": 12.2838,
      "step": 9015
    },
    {
      "epoch": 1.576568306010929,
      "grad_norm": 0.6788121940764448,
      "learning_rate": 0.00013814238430551095,
      "loss": 12.3321,
      "step": 9016
    },
    {
      "epoch": 1.5767431693989071,
      "grad_norm": 0.7689870287449302,
      "learning_rate": 0.00013812929359593854,
      "loss": 12.3168,
      "step": 9017
    },
    {
      "epoch": 1.5769180327868852,
      "grad_norm": 0.7925242460258468,
      "learning_rate": 0.0001381162021217567,
      "loss": 12.2806,
      "step": 9018
    },
    {
      "epoch": 1.5770928961748634,
      "grad_norm": 0.60481185598132,
      "learning_rate": 0.000138103109883228,
      "loss": 12.409,
      "step": 9019
    },
    {
      "epoch": 1.5772677595628415,
      "grad_norm": 0.8196891879327144,
      "learning_rate": 0.00013809001688061496,
      "loss": 12.626,
      "step": 9020
    },
    {
      "epoch": 1.5774426229508198,
      "grad_norm": 0.7578354903626092,
      "learning_rate": 0.0001380769231141801,
      "loss": 12.5279,
      "step": 9021
    },
    {
      "epoch": 1.5776174863387977,
      "grad_norm": 0.772925943530664,
      "learning_rate": 0.00013806382858418608,
      "loss": 12.4561,
      "step": 9022
    },
    {
      "epoch": 1.577792349726776,
      "grad_norm": 0.7483817833460866,
      "learning_rate": 0.00013805073329089537,
      "loss": 12.4364,
      "step": 9023
    },
    {
      "epoch": 1.577967213114754,
      "grad_norm": 0.7171439905024014,
      "learning_rate": 0.00013803763723457065,
      "loss": 12.1203,
      "step": 9024
    },
    {
      "epoch": 1.5781420765027323,
      "grad_norm": 0.7446824401317776,
      "learning_rate": 0.00013802454041547451,
      "loss": 12.5273,
      "step": 9025
    },
    {
      "epoch": 1.5783169398907104,
      "grad_norm": 0.7275487256271806,
      "learning_rate": 0.00013801144283386954,
      "loss": 12.4702,
      "step": 9026
    },
    {
      "epoch": 1.5784918032786885,
      "grad_norm": 0.758317596267901,
      "learning_rate": 0.0001379983444900185,
      "loss": 12.1114,
      "step": 9027
    },
    {
      "epoch": 1.5786666666666667,
      "grad_norm": 0.6938249982813269,
      "learning_rate": 0.00013798524538418395,
      "loss": 12.4661,
      "step": 9028
    },
    {
      "epoch": 1.5788415300546448,
      "grad_norm": 0.8305670243143608,
      "learning_rate": 0.0001379721455166286,
      "loss": 12.4261,
      "step": 9029
    },
    {
      "epoch": 1.579016393442623,
      "grad_norm": 0.7129049456218152,
      "learning_rate": 0.00013795904488761516,
      "loss": 12.3071,
      "step": 9030
    },
    {
      "epoch": 1.579191256830601,
      "grad_norm": 0.7806085975621334,
      "learning_rate": 0.00013794594349740628,
      "loss": 12.3578,
      "step": 9031
    },
    {
      "epoch": 1.5793661202185794,
      "grad_norm": 0.7854121542392883,
      "learning_rate": 0.00013793284134626477,
      "loss": 12.359,
      "step": 9032
    },
    {
      "epoch": 1.5795409836065573,
      "grad_norm": 0.7312308392962676,
      "learning_rate": 0.00013791973843445328,
      "loss": 12.4543,
      "step": 9033
    },
    {
      "epoch": 1.5797158469945356,
      "grad_norm": 0.8435776690905333,
      "learning_rate": 0.00013790663476223464,
      "loss": 12.4452,
      "step": 9034
    },
    {
      "epoch": 1.5798907103825135,
      "grad_norm": 0.753514709432152,
      "learning_rate": 0.00013789353032987156,
      "loss": 12.2342,
      "step": 9035
    },
    {
      "epoch": 1.5800655737704918,
      "grad_norm": 0.7760936909252337,
      "learning_rate": 0.00013788042513762687,
      "loss": 12.3249,
      "step": 9036
    },
    {
      "epoch": 1.58024043715847,
      "grad_norm": 0.6570339906763186,
      "learning_rate": 0.0001378673191857633,
      "loss": 12.0556,
      "step": 9037
    },
    {
      "epoch": 1.580415300546448,
      "grad_norm": 0.7110618323501617,
      "learning_rate": 0.00013785421247454376,
      "loss": 12.2911,
      "step": 9038
    },
    {
      "epoch": 1.5805901639344262,
      "grad_norm": 0.6750943286125688,
      "learning_rate": 0.00013784110500423104,
      "loss": 12.537,
      "step": 9039
    },
    {
      "epoch": 1.5807650273224043,
      "grad_norm": 0.7181979685306724,
      "learning_rate": 0.00013782799677508796,
      "loss": 12.5987,
      "step": 9040
    },
    {
      "epoch": 1.5809398907103827,
      "grad_norm": 0.700916393513681,
      "learning_rate": 0.00013781488778737738,
      "loss": 12.0119,
      "step": 9041
    },
    {
      "epoch": 1.5811147540983606,
      "grad_norm": 0.7593487162336059,
      "learning_rate": 0.00013780177804136222,
      "loss": 12.3271,
      "step": 9042
    },
    {
      "epoch": 1.581289617486339,
      "grad_norm": 0.8581839933679921,
      "learning_rate": 0.00013778866753730532,
      "loss": 12.4114,
      "step": 9043
    },
    {
      "epoch": 1.5814644808743168,
      "grad_norm": 0.7995779281236969,
      "learning_rate": 0.00013777555627546962,
      "loss": 12.545,
      "step": 9044
    },
    {
      "epoch": 1.5816393442622951,
      "grad_norm": 0.8455913186717575,
      "learning_rate": 0.00013776244425611807,
      "loss": 12.3504,
      "step": 9045
    },
    {
      "epoch": 1.5818142076502733,
      "grad_norm": 0.7084264786921358,
      "learning_rate": 0.0001377493314795135,
      "loss": 12.5593,
      "step": 9046
    },
    {
      "epoch": 1.5819890710382514,
      "grad_norm": 0.9329960693467559,
      "learning_rate": 0.00013773621794591895,
      "loss": 12.316,
      "step": 9047
    },
    {
      "epoch": 1.5821639344262295,
      "grad_norm": 0.6812655088569829,
      "learning_rate": 0.0001377231036555974,
      "loss": 12.292,
      "step": 9048
    },
    {
      "epoch": 1.5823387978142076,
      "grad_norm": 0.8489267677708411,
      "learning_rate": 0.00013770998860881175,
      "loss": 12.4541,
      "step": 9049
    },
    {
      "epoch": 1.582513661202186,
      "grad_norm": 0.7636219670347544,
      "learning_rate": 0.00013769687280582507,
      "loss": 12.348,
      "step": 9050
    },
    {
      "epoch": 1.5826885245901638,
      "grad_norm": 0.7727306678016584,
      "learning_rate": 0.00013768375624690032,
      "loss": 12.1839,
      "step": 9051
    },
    {
      "epoch": 1.5828633879781422,
      "grad_norm": 0.7382115951953218,
      "learning_rate": 0.0001376706389323006,
      "loss": 12.4562,
      "step": 9052
    },
    {
      "epoch": 1.58303825136612,
      "grad_norm": 0.7102287095089465,
      "learning_rate": 0.00013765752086228886,
      "loss": 12.489,
      "step": 9053
    },
    {
      "epoch": 1.5832131147540984,
      "grad_norm": 0.6839357558448917,
      "learning_rate": 0.00013764440203712822,
      "loss": 12.4499,
      "step": 9054
    },
    {
      "epoch": 1.5833879781420765,
      "grad_norm": 0.7397749247730847,
      "learning_rate": 0.00013763128245708174,
      "loss": 12.4496,
      "step": 9055
    },
    {
      "epoch": 1.5835628415300547,
      "grad_norm": 0.6602798311426713,
      "learning_rate": 0.0001376181621224125,
      "loss": 12.2813,
      "step": 9056
    },
    {
      "epoch": 1.5837377049180328,
      "grad_norm": 0.731108811951312,
      "learning_rate": 0.0001376050410333836,
      "loss": 12.4259,
      "step": 9057
    },
    {
      "epoch": 1.583912568306011,
      "grad_norm": 0.72226255844894,
      "learning_rate": 0.00013759191919025815,
      "loss": 12.4721,
      "step": 9058
    },
    {
      "epoch": 1.584087431693989,
      "grad_norm": 0.6655594426033948,
      "learning_rate": 0.00013757879659329934,
      "loss": 12.387,
      "step": 9059
    },
    {
      "epoch": 1.5842622950819671,
      "grad_norm": 0.7177542235414656,
      "learning_rate": 0.00013756567324277024,
      "loss": 12.3288,
      "step": 9060
    },
    {
      "epoch": 1.5844371584699455,
      "grad_norm": 0.797482514738411,
      "learning_rate": 0.0001375525491389341,
      "loss": 12.5647,
      "step": 9061
    },
    {
      "epoch": 1.5846120218579234,
      "grad_norm": 0.8422269736961546,
      "learning_rate": 0.000137539424282054,
      "loss": 12.4248,
      "step": 9062
    },
    {
      "epoch": 1.5847868852459017,
      "grad_norm": 0.7014319000430114,
      "learning_rate": 0.0001375262986723932,
      "loss": 12.4627,
      "step": 9063
    },
    {
      "epoch": 1.5849617486338796,
      "grad_norm": 0.8433197012921783,
      "learning_rate": 0.00013751317231021487,
      "loss": 12.1301,
      "step": 9064
    },
    {
      "epoch": 1.585136612021858,
      "grad_norm": 0.9033392998480613,
      "learning_rate": 0.00013750004519578226,
      "loss": 12.469,
      "step": 9065
    },
    {
      "epoch": 1.585311475409836,
      "grad_norm": 0.7631975752996282,
      "learning_rate": 0.00013748691732935864,
      "loss": 12.4894,
      "step": 9066
    },
    {
      "epoch": 1.5854863387978142,
      "grad_norm": 0.6637785486181639,
      "learning_rate": 0.00013747378871120717,
      "loss": 12.2677,
      "step": 9067
    },
    {
      "epoch": 1.5856612021857923,
      "grad_norm": 0.7828105699767753,
      "learning_rate": 0.00013746065934159123,
      "loss": 12.1546,
      "step": 9068
    },
    {
      "epoch": 1.5858360655737704,
      "grad_norm": 0.7566018574405159,
      "learning_rate": 0.000137447529220774,
      "loss": 12.5584,
      "step": 9069
    },
    {
      "epoch": 1.5860109289617488,
      "grad_norm": 0.7230841381420267,
      "learning_rate": 0.0001374343983490189,
      "loss": 12.3626,
      "step": 9070
    },
    {
      "epoch": 1.5861857923497267,
      "grad_norm": 0.8302848115063179,
      "learning_rate": 0.0001374212667265891,
      "loss": 12.3384,
      "step": 9071
    },
    {
      "epoch": 1.586360655737705,
      "grad_norm": 0.7327000674175226,
      "learning_rate": 0.00013740813435374804,
      "loss": 12.4362,
      "step": 9072
    },
    {
      "epoch": 1.586535519125683,
      "grad_norm": 0.7899199417325101,
      "learning_rate": 0.00013739500123075905,
      "loss": 12.4922,
      "step": 9073
    },
    {
      "epoch": 1.5867103825136613,
      "grad_norm": 0.6421314368440778,
      "learning_rate": 0.00013738186735788542,
      "loss": 12.3247,
      "step": 9074
    },
    {
      "epoch": 1.5868852459016394,
      "grad_norm": 0.8071961949687138,
      "learning_rate": 0.00013736873273539058,
      "loss": 12.5009,
      "step": 9075
    },
    {
      "epoch": 1.5870601092896175,
      "grad_norm": 0.732863779555944,
      "learning_rate": 0.00013735559736353792,
      "loss": 12.2972,
      "step": 9076
    },
    {
      "epoch": 1.5872349726775956,
      "grad_norm": 0.7968583772316169,
      "learning_rate": 0.00013734246124259087,
      "loss": 12.3453,
      "step": 9077
    },
    {
      "epoch": 1.5874098360655737,
      "grad_norm": 0.7149988845428557,
      "learning_rate": 0.0001373293243728128,
      "loss": 12.2529,
      "step": 9078
    },
    {
      "epoch": 1.587584699453552,
      "grad_norm": 0.6910223087114964,
      "learning_rate": 0.00013731618675446713,
      "loss": 12.516,
      "step": 9079
    },
    {
      "epoch": 1.58775956284153,
      "grad_norm": 0.8943058622349572,
      "learning_rate": 0.00013730304838781735,
      "loss": 12.2665,
      "step": 9080
    },
    {
      "epoch": 1.5879344262295083,
      "grad_norm": 0.7659524107559682,
      "learning_rate": 0.00013728990927312692,
      "loss": 12.4519,
      "step": 9081
    },
    {
      "epoch": 1.5881092896174862,
      "grad_norm": 0.8164640950334829,
      "learning_rate": 0.00013727676941065932,
      "loss": 12.3621,
      "step": 9082
    },
    {
      "epoch": 1.5882841530054645,
      "grad_norm": 0.8582314589685464,
      "learning_rate": 0.00013726362880067802,
      "loss": 12.442,
      "step": 9083
    },
    {
      "epoch": 1.5884590163934427,
      "grad_norm": 0.9232723427007227,
      "learning_rate": 0.00013725048744344658,
      "loss": 12.4538,
      "step": 9084
    },
    {
      "epoch": 1.5886338797814208,
      "grad_norm": 0.8683893701883126,
      "learning_rate": 0.00013723734533922847,
      "loss": 12.2263,
      "step": 9085
    },
    {
      "epoch": 1.588808743169399,
      "grad_norm": 0.7475736999020688,
      "learning_rate": 0.00013722420248828726,
      "loss": 12.2972,
      "step": 9086
    },
    {
      "epoch": 1.588983606557377,
      "grad_norm": 0.8761106171124,
      "learning_rate": 0.0001372110588908865,
      "loss": 12.517,
      "step": 9087
    },
    {
      "epoch": 1.5891584699453551,
      "grad_norm": 0.6827647691666965,
      "learning_rate": 0.00013719791454728974,
      "loss": 12.444,
      "step": 9088
    },
    {
      "epoch": 1.5893333333333333,
      "grad_norm": 0.8594887957184526,
      "learning_rate": 0.0001371847694577606,
      "loss": 12.2869,
      "step": 9089
    },
    {
      "epoch": 1.5895081967213116,
      "grad_norm": 0.7197505793445583,
      "learning_rate": 0.00013717162362256264,
      "loss": 12.3005,
      "step": 9090
    },
    {
      "epoch": 1.5896830601092895,
      "grad_norm": 0.8733429542951817,
      "learning_rate": 0.0001371584770419595,
      "loss": 12.5226,
      "step": 9091
    },
    {
      "epoch": 1.5898579234972678,
      "grad_norm": 0.697988573161943,
      "learning_rate": 0.0001371453297162148,
      "loss": 12.4028,
      "step": 9092
    },
    {
      "epoch": 1.5900327868852457,
      "grad_norm": 0.7540907452049966,
      "learning_rate": 0.00013713218164559222,
      "loss": 12.4643,
      "step": 9093
    },
    {
      "epoch": 1.590207650273224,
      "grad_norm": 0.7568198259717848,
      "learning_rate": 0.00013711903283035538,
      "loss": 12.0716,
      "step": 9094
    },
    {
      "epoch": 1.5903825136612022,
      "grad_norm": 0.7498433807083758,
      "learning_rate": 0.0001371058832707679,
      "loss": 12.2987,
      "step": 9095
    },
    {
      "epoch": 1.5905573770491803,
      "grad_norm": 0.7268387768815858,
      "learning_rate": 0.0001370927329670936,
      "loss": 12.4335,
      "step": 9096
    },
    {
      "epoch": 1.5907322404371584,
      "grad_norm": 0.7409283969290231,
      "learning_rate": 0.00013707958191959608,
      "loss": 12.5037,
      "step": 9097
    },
    {
      "epoch": 1.5909071038251366,
      "grad_norm": 0.8282699011667957,
      "learning_rate": 0.0001370664301285391,
      "loss": 12.5316,
      "step": 9098
    },
    {
      "epoch": 1.591081967213115,
      "grad_norm": 0.7091798523159569,
      "learning_rate": 0.0001370532775941864,
      "loss": 12.2806,
      "step": 9099
    },
    {
      "epoch": 1.5912568306010928,
      "grad_norm": 0.8993693611094787,
      "learning_rate": 0.0001370401243168017,
      "loss": 12.3024,
      "step": 9100
    },
    {
      "epoch": 1.5914316939890711,
      "grad_norm": 0.7443377494246736,
      "learning_rate": 0.0001370269702966488,
      "loss": 12.2258,
      "step": 9101
    },
    {
      "epoch": 1.591606557377049,
      "grad_norm": 0.8683997235473111,
      "learning_rate": 0.00013701381553399145,
      "loss": 12.3759,
      "step": 9102
    },
    {
      "epoch": 1.5917814207650274,
      "grad_norm": 0.7503609701943327,
      "learning_rate": 0.00013700066002909344,
      "loss": 12.1966,
      "step": 9103
    },
    {
      "epoch": 1.5919562841530055,
      "grad_norm": 0.6987385562301169,
      "learning_rate": 0.0001369875037822186,
      "loss": 12.3821,
      "step": 9104
    },
    {
      "epoch": 1.5921311475409836,
      "grad_norm": 0.744983892941335,
      "learning_rate": 0.00013697434679363073,
      "loss": 12.2152,
      "step": 9105
    },
    {
      "epoch": 1.5923060109289617,
      "grad_norm": 0.7538055116632311,
      "learning_rate": 0.0001369611890635937,
      "loss": 12.5675,
      "step": 9106
    },
    {
      "epoch": 1.5924808743169399,
      "grad_norm": 0.797122465791437,
      "learning_rate": 0.00013694803059237135,
      "loss": 12.4144,
      "step": 9107
    },
    {
      "epoch": 1.5926557377049182,
      "grad_norm": 0.7373604421700088,
      "learning_rate": 0.00013693487138022748,
      "loss": 12.2242,
      "step": 9108
    },
    {
      "epoch": 1.592830601092896,
      "grad_norm": 0.6967094018489861,
      "learning_rate": 0.0001369217114274261,
      "loss": 12.2734,
      "step": 9109
    },
    {
      "epoch": 1.5930054644808744,
      "grad_norm": 0.7373169593893892,
      "learning_rate": 0.00013690855073423102,
      "loss": 12.4746,
      "step": 9110
    },
    {
      "epoch": 1.5931803278688523,
      "grad_norm": 0.6881392742404662,
      "learning_rate": 0.00013689538930090618,
      "loss": 12.6242,
      "step": 9111
    },
    {
      "epoch": 1.5933551912568307,
      "grad_norm": 0.740500341588493,
      "learning_rate": 0.0001368822271277155,
      "loss": 12.366,
      "step": 9112
    },
    {
      "epoch": 1.5935300546448088,
      "grad_norm": 0.6718609719022945,
      "learning_rate": 0.0001368690642149229,
      "loss": 12.4673,
      "step": 9113
    },
    {
      "epoch": 1.593704918032787,
      "grad_norm": 0.8020152840660294,
      "learning_rate": 0.00013685590056279238,
      "loss": 12.268,
      "step": 9114
    },
    {
      "epoch": 1.593879781420765,
      "grad_norm": 0.7358874911353265,
      "learning_rate": 0.00013684273617158788,
      "loss": 12.242,
      "step": 9115
    },
    {
      "epoch": 1.5940546448087431,
      "grad_norm": 0.7288240216578405,
      "learning_rate": 0.00013682957104157342,
      "loss": 12.4258,
      "step": 9116
    },
    {
      "epoch": 1.5942295081967213,
      "grad_norm": 0.747182161326531,
      "learning_rate": 0.00013681640517301292,
      "loss": 12.2924,
      "step": 9117
    },
    {
      "epoch": 1.5944043715846994,
      "grad_norm": 0.7412440852515978,
      "learning_rate": 0.00013680323856617052,
      "loss": 12.4493,
      "step": 9118
    },
    {
      "epoch": 1.5945792349726777,
      "grad_norm": 0.7794464846623937,
      "learning_rate": 0.00013679007122131015,
      "loss": 12.5984,
      "step": 9119
    },
    {
      "epoch": 1.5947540983606556,
      "grad_norm": 0.723928416558884,
      "learning_rate": 0.00013677690313869593,
      "loss": 12.2908,
      "step": 9120
    },
    {
      "epoch": 1.594928961748634,
      "grad_norm": 0.9415550571437385,
      "learning_rate": 0.00013676373431859186,
      "loss": 12.3328,
      "step": 9121
    },
    {
      "epoch": 1.5951038251366119,
      "grad_norm": 0.6920965755171665,
      "learning_rate": 0.000136750564761262,
      "loss": 12.4169,
      "step": 9122
    },
    {
      "epoch": 1.5952786885245902,
      "grad_norm": 1.0373878684807702,
      "learning_rate": 0.00013673739446697054,
      "loss": 12.4609,
      "step": 9123
    },
    {
      "epoch": 1.5954535519125683,
      "grad_norm": 0.7601025383409515,
      "learning_rate": 0.00013672422343598148,
      "loss": 12.6293,
      "step": 9124
    },
    {
      "epoch": 1.5956284153005464,
      "grad_norm": 0.72948901071498,
      "learning_rate": 0.00013671105166855902,
      "loss": 12.2289,
      "step": 9125
    },
    {
      "epoch": 1.5958032786885246,
      "grad_norm": 0.7369811806386852,
      "learning_rate": 0.00013669787916496722,
      "loss": 12.0115,
      "step": 9126
    },
    {
      "epoch": 1.5959781420765027,
      "grad_norm": 0.749026355216912,
      "learning_rate": 0.0001366847059254703,
      "loss": 12.4528,
      "step": 9127
    },
    {
      "epoch": 1.596153005464481,
      "grad_norm": 0.7040864516338644,
      "learning_rate": 0.00013667153195033235,
      "loss": 12.5051,
      "step": 9128
    },
    {
      "epoch": 1.596327868852459,
      "grad_norm": 0.9059560598373029,
      "learning_rate": 0.0001366583572398176,
      "loss": 12.3888,
      "step": 9129
    },
    {
      "epoch": 1.5965027322404373,
      "grad_norm": 0.6593612284551404,
      "learning_rate": 0.00013664518179419023,
      "loss": 12.0772,
      "step": 9130
    },
    {
      "epoch": 1.5966775956284152,
      "grad_norm": 0.762360842008574,
      "learning_rate": 0.00013663200561371447,
      "loss": 12.1437,
      "step": 9131
    },
    {
      "epoch": 1.5968524590163935,
      "grad_norm": 0.6707724529460406,
      "learning_rate": 0.0001366188286986545,
      "loss": 12.5357,
      "step": 9132
    },
    {
      "epoch": 1.5970273224043716,
      "grad_norm": 0.8116536422082777,
      "learning_rate": 0.0001366056510492746,
      "loss": 12.3574,
      "step": 9133
    },
    {
      "epoch": 1.5972021857923497,
      "grad_norm": 0.7198431545394149,
      "learning_rate": 0.000136592472665839,
      "loss": 12.21,
      "step": 9134
    },
    {
      "epoch": 1.5973770491803279,
      "grad_norm": 0.6899550959892472,
      "learning_rate": 0.00013657929354861194,
      "loss": 12.3726,
      "step": 9135
    },
    {
      "epoch": 1.597551912568306,
      "grad_norm": 0.850816515125804,
      "learning_rate": 0.00013656611369785776,
      "loss": 12.4477,
      "step": 9136
    },
    {
      "epoch": 1.5977267759562843,
      "grad_norm": 0.7227964022473665,
      "learning_rate": 0.00013655293311384072,
      "loss": 12.3652,
      "step": 9137
    },
    {
      "epoch": 1.5979016393442622,
      "grad_norm": 0.8243801314908922,
      "learning_rate": 0.00013653975179682515,
      "loss": 12.3879,
      "step": 9138
    },
    {
      "epoch": 1.5980765027322406,
      "grad_norm": 0.7432656686764955,
      "learning_rate": 0.00013652656974707535,
      "loss": 12.0598,
      "step": 9139
    },
    {
      "epoch": 1.5982513661202185,
      "grad_norm": 0.7953035868408561,
      "learning_rate": 0.00013651338696485567,
      "loss": 12.1991,
      "step": 9140
    },
    {
      "epoch": 1.5984262295081968,
      "grad_norm": 0.7339757037751055,
      "learning_rate": 0.00013650020345043048,
      "loss": 12.4059,
      "step": 9141
    },
    {
      "epoch": 1.598601092896175,
      "grad_norm": 0.715067315676268,
      "learning_rate": 0.00013648701920406414,
      "loss": 12.1948,
      "step": 9142
    },
    {
      "epoch": 1.598775956284153,
      "grad_norm": 0.8403772973795941,
      "learning_rate": 0.00013647383422602102,
      "loss": 12.4153,
      "step": 9143
    },
    {
      "epoch": 1.5989508196721312,
      "grad_norm": 0.7027144003228545,
      "learning_rate": 0.00013646064851656554,
      "loss": 12.2047,
      "step": 9144
    },
    {
      "epoch": 1.5991256830601093,
      "grad_norm": 0.816195400126237,
      "learning_rate": 0.00013644746207596215,
      "loss": 12.3272,
      "step": 9145
    },
    {
      "epoch": 1.5993005464480874,
      "grad_norm": 0.678535146016512,
      "learning_rate": 0.0001364342749044752,
      "loss": 12.2023,
      "step": 9146
    },
    {
      "epoch": 1.5994754098360655,
      "grad_norm": 0.8317254978601709,
      "learning_rate": 0.00013642108700236916,
      "loss": 12.5634,
      "step": 9147
    },
    {
      "epoch": 1.5996502732240439,
      "grad_norm": 0.6577172847960229,
      "learning_rate": 0.00013640789836990853,
      "loss": 12.3953,
      "step": 9148
    },
    {
      "epoch": 1.5998251366120217,
      "grad_norm": 0.8838327871161681,
      "learning_rate": 0.0001363947090073577,
      "loss": 12.0312,
      "step": 9149
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7481639381585217,
      "learning_rate": 0.00013638151891498128,
      "loss": 12.3728,
      "step": 9150
    },
    {
      "epoch": 1.600174863387978,
      "grad_norm": 0.9365070898989514,
      "learning_rate": 0.00013636832809304364,
      "loss": 12.4545,
      "step": 9151
    },
    {
      "epoch": 1.6003497267759563,
      "grad_norm": 0.6805418762112189,
      "learning_rate": 0.00013635513654180939,
      "loss": 12.3533,
      "step": 9152
    },
    {
      "epoch": 1.6005245901639344,
      "grad_norm": 0.7962287791332859,
      "learning_rate": 0.000136341944261543,
      "loss": 12.592,
      "step": 9153
    },
    {
      "epoch": 1.6006994535519126,
      "grad_norm": 0.7082962563390616,
      "learning_rate": 0.00013632875125250905,
      "loss": 11.9978,
      "step": 9154
    },
    {
      "epoch": 1.6008743169398907,
      "grad_norm": 0.7917094701473172,
      "learning_rate": 0.00013631555751497215,
      "loss": 12.1224,
      "step": 9155
    },
    {
      "epoch": 1.6010491803278688,
      "grad_norm": 0.6821064671058448,
      "learning_rate": 0.00013630236304919673,
      "loss": 12.4609,
      "step": 9156
    },
    {
      "epoch": 1.6012240437158471,
      "grad_norm": 0.7735054017266415,
      "learning_rate": 0.00013628916785544755,
      "loss": 12.1014,
      "step": 9157
    },
    {
      "epoch": 1.601398907103825,
      "grad_norm": 0.6985242422381902,
      "learning_rate": 0.00013627597193398904,
      "loss": 12.4784,
      "step": 9158
    },
    {
      "epoch": 1.6015737704918034,
      "grad_norm": 0.8138324593888885,
      "learning_rate": 0.000136262775285086,
      "loss": 12.3914,
      "step": 9159
    },
    {
      "epoch": 1.6017486338797813,
      "grad_norm": 0.7035290625427756,
      "learning_rate": 0.0001362495779090029,
      "loss": 12.3925,
      "step": 9160
    },
    {
      "epoch": 1.6019234972677596,
      "grad_norm": 0.6802615026045389,
      "learning_rate": 0.0001362363798060045,
      "loss": 12.5026,
      "step": 9161
    },
    {
      "epoch": 1.6020983606557377,
      "grad_norm": 0.6795791426458456,
      "learning_rate": 0.00013622318097635544,
      "loss": 12.3674,
      "step": 9162
    },
    {
      "epoch": 1.6022732240437159,
      "grad_norm": 0.6658562210453385,
      "learning_rate": 0.00013620998142032035,
      "loss": 12.5351,
      "step": 9163
    },
    {
      "epoch": 1.602448087431694,
      "grad_norm": 0.7054716551157442,
      "learning_rate": 0.000136196781138164,
      "loss": 12.3525,
      "step": 9164
    },
    {
      "epoch": 1.602622950819672,
      "grad_norm": 0.6907540637573135,
      "learning_rate": 0.00013618358013015098,
      "loss": 12.2736,
      "step": 9165
    },
    {
      "epoch": 1.6027978142076504,
      "grad_norm": 0.7871356484998663,
      "learning_rate": 0.00013617037839654613,
      "loss": 12.3471,
      "step": 9166
    },
    {
      "epoch": 1.6029726775956283,
      "grad_norm": 0.7152873663313037,
      "learning_rate": 0.00013615717593761412,
      "loss": 12.4043,
      "step": 9167
    },
    {
      "epoch": 1.6031475409836067,
      "grad_norm": 0.7440958409650292,
      "learning_rate": 0.00013614397275361973,
      "loss": 12.5133,
      "step": 9168
    },
    {
      "epoch": 1.6033224043715846,
      "grad_norm": 0.645917786200601,
      "learning_rate": 0.0001361307688448277,
      "loss": 12.2669,
      "step": 9169
    },
    {
      "epoch": 1.603497267759563,
      "grad_norm": 0.7889019943064539,
      "learning_rate": 0.0001361175642115028,
      "loss": 12.2156,
      "step": 9170
    },
    {
      "epoch": 1.603672131147541,
      "grad_norm": 0.6514822028432827,
      "learning_rate": 0.00013610435885390987,
      "loss": 12.3694,
      "step": 9171
    },
    {
      "epoch": 1.6038469945355192,
      "grad_norm": 0.6374350862732677,
      "learning_rate": 0.00013609115277231365,
      "loss": 12.3069,
      "step": 9172
    },
    {
      "epoch": 1.6040218579234973,
      "grad_norm": 0.6811105620891249,
      "learning_rate": 0.00013607794596697906,
      "loss": 12.2888,
      "step": 9173
    },
    {
      "epoch": 1.6041967213114754,
      "grad_norm": 0.7233894291246696,
      "learning_rate": 0.00013606473843817086,
      "loss": 12.1577,
      "step": 9174
    },
    {
      "epoch": 1.6043715846994535,
      "grad_norm": 0.7342556388655844,
      "learning_rate": 0.0001360515301861539,
      "loss": 12.3349,
      "step": 9175
    },
    {
      "epoch": 1.6045464480874316,
      "grad_norm": 0.661921757784968,
      "learning_rate": 0.0001360383212111931,
      "loss": 12.2494,
      "step": 9176
    },
    {
      "epoch": 1.60472131147541,
      "grad_norm": 0.7300826689432625,
      "learning_rate": 0.00013602511151355326,
      "loss": 12.5159,
      "step": 9177
    },
    {
      "epoch": 1.6048961748633879,
      "grad_norm": 0.6386170665573465,
      "learning_rate": 0.00013601190109349936,
      "loss": 12.5043,
      "step": 9178
    },
    {
      "epoch": 1.6050710382513662,
      "grad_norm": 0.6867871361487763,
      "learning_rate": 0.00013599868995129626,
      "loss": 12.3381,
      "step": 9179
    },
    {
      "epoch": 1.605245901639344,
      "grad_norm": 0.6713520800909558,
      "learning_rate": 0.00013598547808720888,
      "loss": 12.178,
      "step": 9180
    },
    {
      "epoch": 1.6054207650273224,
      "grad_norm": 0.7384729865688915,
      "learning_rate": 0.0001359722655015022,
      "loss": 12.228,
      "step": 9181
    },
    {
      "epoch": 1.6055956284153006,
      "grad_norm": 0.6799414865217026,
      "learning_rate": 0.00013595905219444113,
      "loss": 12.4458,
      "step": 9182
    },
    {
      "epoch": 1.6057704918032787,
      "grad_norm": 0.7071812890619764,
      "learning_rate": 0.0001359458381662907,
      "loss": 12.3481,
      "step": 9183
    },
    {
      "epoch": 1.6059453551912568,
      "grad_norm": 0.6837002879088544,
      "learning_rate": 0.00013593262341731578,
      "loss": 12.2257,
      "step": 9184
    },
    {
      "epoch": 1.606120218579235,
      "grad_norm": 0.6640058996590473,
      "learning_rate": 0.00013591940794778146,
      "loss": 12.2641,
      "step": 9185
    },
    {
      "epoch": 1.6062950819672133,
      "grad_norm": 0.6872479737901137,
      "learning_rate": 0.0001359061917579527,
      "loss": 12.3546,
      "step": 9186
    },
    {
      "epoch": 1.6064699453551912,
      "grad_norm": 0.6903043535705157,
      "learning_rate": 0.0001358929748480946,
      "loss": 12.232,
      "step": 9187
    },
    {
      "epoch": 1.6066448087431695,
      "grad_norm": 0.6734414497250087,
      "learning_rate": 0.0001358797572184721,
      "loss": 12.3054,
      "step": 9188
    },
    {
      "epoch": 1.6068196721311474,
      "grad_norm": 0.7391661480814867,
      "learning_rate": 0.00013586653886935035,
      "loss": 12.3011,
      "step": 9189
    },
    {
      "epoch": 1.6069945355191257,
      "grad_norm": 0.7522879456612898,
      "learning_rate": 0.00013585331980099435,
      "loss": 12.3802,
      "step": 9190
    },
    {
      "epoch": 1.6071693989071039,
      "grad_norm": 0.6689002299058955,
      "learning_rate": 0.00013584010001366917,
      "loss": 12.0631,
      "step": 9191
    },
    {
      "epoch": 1.607344262295082,
      "grad_norm": 0.7618337337411487,
      "learning_rate": 0.00013582687950764,
      "loss": 12.3314,
      "step": 9192
    },
    {
      "epoch": 1.60751912568306,
      "grad_norm": 0.6844312249901019,
      "learning_rate": 0.00013581365828317185,
      "loss": 12.4,
      "step": 9193
    },
    {
      "epoch": 1.6076939890710382,
      "grad_norm": 0.7199229957217662,
      "learning_rate": 0.0001358004363405299,
      "loss": 12.2728,
      "step": 9194
    },
    {
      "epoch": 1.6078688524590163,
      "grad_norm": 0.7166988161916675,
      "learning_rate": 0.0001357872136799793,
      "loss": 12.1982,
      "step": 9195
    },
    {
      "epoch": 1.6080437158469945,
      "grad_norm": 0.715094201620342,
      "learning_rate": 0.00013577399030178516,
      "loss": 12.3216,
      "step": 9196
    },
    {
      "epoch": 1.6082185792349728,
      "grad_norm": 0.6305268976840821,
      "learning_rate": 0.00013576076620621267,
      "loss": 12.4348,
      "step": 9197
    },
    {
      "epoch": 1.6083934426229507,
      "grad_norm": 0.6682945377793978,
      "learning_rate": 0.00013574754139352707,
      "loss": 12.2506,
      "step": 9198
    },
    {
      "epoch": 1.608568306010929,
      "grad_norm": 0.7002785040759905,
      "learning_rate": 0.00013573431586399342,
      "loss": 12.4276,
      "step": 9199
    },
    {
      "epoch": 1.6087431693989072,
      "grad_norm": 0.6257650158955408,
      "learning_rate": 0.00013572108961787705,
      "loss": 11.8909,
      "step": 9200
    },
    {
      "epoch": 1.6089180327868853,
      "grad_norm": 0.7460698009626695,
      "learning_rate": 0.0001357078626554432,
      "loss": 12.3565,
      "step": 9201
    },
    {
      "epoch": 1.6090928961748634,
      "grad_norm": 0.6666824714247918,
      "learning_rate": 0.00013569463497695703,
      "loss": 12.4305,
      "step": 9202
    },
    {
      "epoch": 1.6092677595628415,
      "grad_norm": 0.8379707237777517,
      "learning_rate": 0.00013568140658268385,
      "loss": 12.4453,
      "step": 9203
    },
    {
      "epoch": 1.6094426229508196,
      "grad_norm": 0.7012450043590147,
      "learning_rate": 0.00013566817747288887,
      "loss": 12.3575,
      "step": 9204
    },
    {
      "epoch": 1.6096174863387978,
      "grad_norm": 0.7263266812297121,
      "learning_rate": 0.00013565494764783745,
      "loss": 12.0295,
      "step": 9205
    },
    {
      "epoch": 1.609792349726776,
      "grad_norm": 0.687790115754992,
      "learning_rate": 0.00013564171710779484,
      "loss": 12.4574,
      "step": 9206
    },
    {
      "epoch": 1.609967213114754,
      "grad_norm": 0.7086767869238306,
      "learning_rate": 0.0001356284858530264,
      "loss": 12.5267,
      "step": 9207
    },
    {
      "epoch": 1.6101420765027323,
      "grad_norm": 0.8868136421354018,
      "learning_rate": 0.00013561525388379738,
      "loss": 12.1675,
      "step": 9208
    },
    {
      "epoch": 1.6103169398907102,
      "grad_norm": 0.6342471797579928,
      "learning_rate": 0.00013560202120037318,
      "loss": 12.4014,
      "step": 9209
    },
    {
      "epoch": 1.6104918032786886,
      "grad_norm": 0.711340937509255,
      "learning_rate": 0.00013558878780301918,
      "loss": 12.5103,
      "step": 9210
    },
    {
      "epoch": 1.6106666666666667,
      "grad_norm": 0.6587565631347638,
      "learning_rate": 0.00013557555369200064,
      "loss": 12.2431,
      "step": 9211
    },
    {
      "epoch": 1.6108415300546448,
      "grad_norm": 0.6965989615342817,
      "learning_rate": 0.0001355623188675831,
      "loss": 12.3133,
      "step": 9212
    },
    {
      "epoch": 1.611016393442623,
      "grad_norm": 0.7009946400228254,
      "learning_rate": 0.0001355490833300318,
      "loss": 12.2428,
      "step": 9213
    },
    {
      "epoch": 1.611191256830601,
      "grad_norm": 0.7187141677772168,
      "learning_rate": 0.0001355358470796123,
      "loss": 12.3425,
      "step": 9214
    },
    {
      "epoch": 1.6113661202185794,
      "grad_norm": 0.7144460883512843,
      "learning_rate": 0.0001355226101165899,
      "loss": 12.2493,
      "step": 9215
    },
    {
      "epoch": 1.6115409836065573,
      "grad_norm": 0.6135431212460772,
      "learning_rate": 0.0001355093724412301,
      "loss": 12.3992,
      "step": 9216
    },
    {
      "epoch": 1.6117158469945356,
      "grad_norm": 0.6657078965577183,
      "learning_rate": 0.0001354961340537984,
      "loss": 12.098,
      "step": 9217
    },
    {
      "epoch": 1.6118907103825135,
      "grad_norm": 0.7118094826023236,
      "learning_rate": 0.00013548289495456017,
      "loss": 12.4953,
      "step": 9218
    },
    {
      "epoch": 1.6120655737704919,
      "grad_norm": 0.5990862751688073,
      "learning_rate": 0.000135469655143781,
      "loss": 12.4275,
      "step": 9219
    },
    {
      "epoch": 1.61224043715847,
      "grad_norm": 0.7100384287655668,
      "learning_rate": 0.0001354564146217263,
      "loss": 12.2478,
      "step": 9220
    },
    {
      "epoch": 1.612415300546448,
      "grad_norm": 0.6618185121754172,
      "learning_rate": 0.00013544317338866165,
      "loss": 12.0895,
      "step": 9221
    },
    {
      "epoch": 1.6125901639344262,
      "grad_norm": 0.6539692441122905,
      "learning_rate": 0.00013542993144485254,
      "loss": 12.3174,
      "step": 9222
    },
    {
      "epoch": 1.6127650273224043,
      "grad_norm": 1.3672451718909118,
      "learning_rate": 0.00013541668879056453,
      "loss": 12.1051,
      "step": 9223
    },
    {
      "epoch": 1.6129398907103825,
      "grad_norm": 0.6839861989050253,
      "learning_rate": 0.00013540344542606314,
      "loss": 12.0282,
      "step": 9224
    },
    {
      "epoch": 1.6131147540983606,
      "grad_norm": 0.6542927298129603,
      "learning_rate": 0.00013539020135161397,
      "loss": 12.45,
      "step": 9225
    },
    {
      "epoch": 1.613289617486339,
      "grad_norm": 0.8843055962837099,
      "learning_rate": 0.00013537695656748263,
      "loss": 12.4126,
      "step": 9226
    },
    {
      "epoch": 1.6134644808743168,
      "grad_norm": 0.6766305098331098,
      "learning_rate": 0.00013536371107393465,
      "loss": 12.2387,
      "step": 9227
    },
    {
      "epoch": 1.6136393442622952,
      "grad_norm": 0.8277086736486159,
      "learning_rate": 0.0001353504648712357,
      "loss": 12.3987,
      "step": 9228
    },
    {
      "epoch": 1.613814207650273,
      "grad_norm": 0.6498168635663885,
      "learning_rate": 0.0001353372179596514,
      "loss": 12.3406,
      "step": 9229
    },
    {
      "epoch": 1.6139890710382514,
      "grad_norm": 0.8148360856964013,
      "learning_rate": 0.00013532397033944737,
      "loss": 12.5409,
      "step": 9230
    },
    {
      "epoch": 1.6141639344262295,
      "grad_norm": 0.8039222897875666,
      "learning_rate": 0.00013531072201088928,
      "loss": 12.3281,
      "step": 9231
    },
    {
      "epoch": 1.6143387978142076,
      "grad_norm": 0.7153777779828708,
      "learning_rate": 0.0001352974729742428,
      "loss": 12.3131,
      "step": 9232
    },
    {
      "epoch": 1.6145136612021858,
      "grad_norm": 0.8059825663673336,
      "learning_rate": 0.0001352842232297736,
      "loss": 12.3993,
      "step": 9233
    },
    {
      "epoch": 1.6146885245901639,
      "grad_norm": 0.7762419992633278,
      "learning_rate": 0.00013527097277774738,
      "loss": 12.1356,
      "step": 9234
    },
    {
      "epoch": 1.6148633879781422,
      "grad_norm": 0.7848074064386978,
      "learning_rate": 0.00013525772161842987,
      "loss": 12.3892,
      "step": 9235
    },
    {
      "epoch": 1.6150382513661201,
      "grad_norm": 0.647844027615527,
      "learning_rate": 0.00013524446975208678,
      "loss": 12.2935,
      "step": 9236
    },
    {
      "epoch": 1.6152131147540985,
      "grad_norm": 0.6697535527507207,
      "learning_rate": 0.00013523121717898387,
      "loss": 12.4553,
      "step": 9237
    },
    {
      "epoch": 1.6153879781420764,
      "grad_norm": 0.7036029202426595,
      "learning_rate": 0.00013521796389938685,
      "loss": 12.4028,
      "step": 9238
    },
    {
      "epoch": 1.6155628415300547,
      "grad_norm": 0.8789539701806479,
      "learning_rate": 0.00013520470991356157,
      "loss": 12.4988,
      "step": 9239
    },
    {
      "epoch": 1.6157377049180328,
      "grad_norm": 0.7005667359205262,
      "learning_rate": 0.00013519145522177372,
      "loss": 12.1821,
      "step": 9240
    },
    {
      "epoch": 1.615912568306011,
      "grad_norm": 0.7084519533747043,
      "learning_rate": 0.00013517819982428915,
      "loss": 12.3345,
      "step": 9241
    },
    {
      "epoch": 1.616087431693989,
      "grad_norm": 0.664232883813704,
      "learning_rate": 0.00013516494372137368,
      "loss": 12.3747,
      "step": 9242
    },
    {
      "epoch": 1.6162622950819672,
      "grad_norm": 0.7622611826006227,
      "learning_rate": 0.0001351516869132931,
      "loss": 12.5198,
      "step": 9243
    },
    {
      "epoch": 1.6164371584699455,
      "grad_norm": 0.746199036988587,
      "learning_rate": 0.0001351384294003133,
      "loss": 12.1486,
      "step": 9244
    },
    {
      "epoch": 1.6166120218579234,
      "grad_norm": 0.7252302083402142,
      "learning_rate": 0.00013512517118270006,
      "loss": 12.4822,
      "step": 9245
    },
    {
      "epoch": 1.6167868852459018,
      "grad_norm": 0.7307017697927349,
      "learning_rate": 0.00013511191226071932,
      "loss": 12.232,
      "step": 9246
    },
    {
      "epoch": 1.6169617486338796,
      "grad_norm": 0.69070166306939,
      "learning_rate": 0.0001350986526346369,
      "loss": 12.3216,
      "step": 9247
    },
    {
      "epoch": 1.617136612021858,
      "grad_norm": 0.7573759779851179,
      "learning_rate": 0.00013508539230471878,
      "loss": 12.2822,
      "step": 9248
    },
    {
      "epoch": 1.617311475409836,
      "grad_norm": 0.7229754310518177,
      "learning_rate": 0.00013507213127123075,
      "loss": 12.4314,
      "step": 9249
    },
    {
      "epoch": 1.6174863387978142,
      "grad_norm": 0.6843032873026473,
      "learning_rate": 0.00013505886953443886,
      "loss": 12.2749,
      "step": 9250
    },
    {
      "epoch": 1.6176612021857923,
      "grad_norm": 0.6912611543343028,
      "learning_rate": 0.00013504560709460896,
      "loss": 12.3815,
      "step": 9251
    },
    {
      "epoch": 1.6178360655737705,
      "grad_norm": 0.7119857711590205,
      "learning_rate": 0.000135032343952007,
      "loss": 12.288,
      "step": 9252
    },
    {
      "epoch": 1.6180109289617486,
      "grad_norm": 0.6747419841087487,
      "learning_rate": 0.00013501908010689906,
      "loss": 11.9833,
      "step": 9253
    },
    {
      "epoch": 1.6181857923497267,
      "grad_norm": 0.6925280942415879,
      "learning_rate": 0.000135005815559551,
      "loss": 12.2207,
      "step": 9254
    },
    {
      "epoch": 1.618360655737705,
      "grad_norm": 0.6475911148008436,
      "learning_rate": 0.00013499255031022885,
      "loss": 12.5812,
      "step": 9255
    },
    {
      "epoch": 1.618535519125683,
      "grad_norm": 0.6373120414493089,
      "learning_rate": 0.00013497928435919862,
      "loss": 12.549,
      "step": 9256
    },
    {
      "epoch": 1.6187103825136613,
      "grad_norm": 0.7792301851319943,
      "learning_rate": 0.00013496601770672637,
      "loss": 12.2948,
      "step": 9257
    },
    {
      "epoch": 1.6188852459016392,
      "grad_norm": 0.7070172437372308,
      "learning_rate": 0.00013495275035307807,
      "loss": 12.4086,
      "step": 9258
    },
    {
      "epoch": 1.6190601092896175,
      "grad_norm": 0.7732854310677312,
      "learning_rate": 0.00013493948229851982,
      "loss": 12.2598,
      "step": 9259
    },
    {
      "epoch": 1.6192349726775956,
      "grad_norm": 0.7522592411679951,
      "learning_rate": 0.0001349262135433177,
      "loss": 12.4926,
      "step": 9260
    },
    {
      "epoch": 1.6194098360655738,
      "grad_norm": 0.7702783197624965,
      "learning_rate": 0.00013491294408773773,
      "loss": 12.2799,
      "step": 9261
    },
    {
      "epoch": 1.6195846994535519,
      "grad_norm": 0.7301140382249004,
      "learning_rate": 0.00013489967393204603,
      "loss": 12.4646,
      "step": 9262
    },
    {
      "epoch": 1.61975956284153,
      "grad_norm": 0.8348275521421135,
      "learning_rate": 0.00013488640307650872,
      "loss": 12.354,
      "step": 9263
    },
    {
      "epoch": 1.6199344262295083,
      "grad_norm": 0.7178338630140045,
      "learning_rate": 0.0001348731315213919,
      "loss": 12.2378,
      "step": 9264
    },
    {
      "epoch": 1.6201092896174862,
      "grad_norm": 0.7532600500143778,
      "learning_rate": 0.00013485985926696173,
      "loss": 12.1837,
      "step": 9265
    },
    {
      "epoch": 1.6202841530054646,
      "grad_norm": 0.7250225616966632,
      "learning_rate": 0.00013484658631348435,
      "loss": 12.356,
      "step": 9266
    },
    {
      "epoch": 1.6204590163934425,
      "grad_norm": 0.7834240520770415,
      "learning_rate": 0.00013483331266122593,
      "loss": 12.3282,
      "step": 9267
    },
    {
      "epoch": 1.6206338797814208,
      "grad_norm": 0.6667236578799549,
      "learning_rate": 0.00013482003831045262,
      "loss": 12.3757,
      "step": 9268
    },
    {
      "epoch": 1.620808743169399,
      "grad_norm": 0.7667057072559277,
      "learning_rate": 0.00013480676326143066,
      "loss": 12.195,
      "step": 9269
    },
    {
      "epoch": 1.620983606557377,
      "grad_norm": 0.7260479103099341,
      "learning_rate": 0.0001347934875144262,
      "loss": 12.2571,
      "step": 9270
    },
    {
      "epoch": 1.6211584699453552,
      "grad_norm": 0.7014372418336746,
      "learning_rate": 0.0001347802110697055,
      "loss": 12.2887,
      "step": 9271
    },
    {
      "epoch": 1.6213333333333333,
      "grad_norm": 0.8930604351620083,
      "learning_rate": 0.00013476693392753476,
      "loss": 12.5018,
      "step": 9272
    },
    {
      "epoch": 1.6215081967213116,
      "grad_norm": 0.819227425249925,
      "learning_rate": 0.00013475365608818027,
      "loss": 12.4636,
      "step": 9273
    },
    {
      "epoch": 1.6216830601092895,
      "grad_norm": 0.7353977840604559,
      "learning_rate": 0.00013474037755190825,
      "loss": 12.2062,
      "step": 9274
    },
    {
      "epoch": 1.6218579234972679,
      "grad_norm": 0.6981807299762721,
      "learning_rate": 0.000134727098318985,
      "loss": 12.4901,
      "step": 9275
    },
    {
      "epoch": 1.6220327868852458,
      "grad_norm": 0.9147680908887112,
      "learning_rate": 0.0001347138183896768,
      "loss": 12.5324,
      "step": 9276
    },
    {
      "epoch": 1.622207650273224,
      "grad_norm": 0.6326235838034521,
      "learning_rate": 0.00013470053776424995,
      "loss": 12.1052,
      "step": 9277
    },
    {
      "epoch": 1.6223825136612022,
      "grad_norm": 0.7764588194787552,
      "learning_rate": 0.0001346872564429708,
      "loss": 12.2493,
      "step": 9278
    },
    {
      "epoch": 1.6225573770491803,
      "grad_norm": 0.7333870755847414,
      "learning_rate": 0.00013467397442610564,
      "loss": 12.1963,
      "step": 9279
    },
    {
      "epoch": 1.6227322404371585,
      "grad_norm": 0.7545418329558157,
      "learning_rate": 0.0001346606917139208,
      "loss": 12.3376,
      "step": 9280
    },
    {
      "epoch": 1.6229071038251366,
      "grad_norm": 0.7683790655343153,
      "learning_rate": 0.0001346474083066827,
      "loss": 12.2463,
      "step": 9281
    },
    {
      "epoch": 1.6230819672131147,
      "grad_norm": 0.751220871525631,
      "learning_rate": 0.00013463412420465767,
      "loss": 12.4246,
      "step": 9282
    },
    {
      "epoch": 1.6232568306010928,
      "grad_norm": 0.6717979730509988,
      "learning_rate": 0.0001346208394081121,
      "loss": 11.9319,
      "step": 9283
    },
    {
      "epoch": 1.6234316939890712,
      "grad_norm": 0.7509272723484194,
      "learning_rate": 0.00013460755391731243,
      "loss": 12.4545,
      "step": 9284
    },
    {
      "epoch": 1.623606557377049,
      "grad_norm": 0.7064328427984755,
      "learning_rate": 0.000134594267732525,
      "loss": 12.4862,
      "step": 9285
    },
    {
      "epoch": 1.6237814207650274,
      "grad_norm": 0.6548150466274847,
      "learning_rate": 0.00013458098085401634,
      "loss": 12.3851,
      "step": 9286
    },
    {
      "epoch": 1.6239562841530053,
      "grad_norm": 0.7590894800998019,
      "learning_rate": 0.00013456769328205282,
      "loss": 12.3748,
      "step": 9287
    },
    {
      "epoch": 1.6241311475409836,
      "grad_norm": 0.735945179976273,
      "learning_rate": 0.00013455440501690088,
      "loss": 12.4736,
      "step": 9288
    },
    {
      "epoch": 1.6243060109289618,
      "grad_norm": 0.7777846649229885,
      "learning_rate": 0.00013454111605882706,
      "loss": 12.3727,
      "step": 9289
    },
    {
      "epoch": 1.6244808743169399,
      "grad_norm": 0.7109198432331492,
      "learning_rate": 0.00013452782640809778,
      "loss": 12.3249,
      "step": 9290
    },
    {
      "epoch": 1.624655737704918,
      "grad_norm": 0.6798409659675916,
      "learning_rate": 0.00013451453606497956,
      "loss": 12.1357,
      "step": 9291
    },
    {
      "epoch": 1.6248306010928961,
      "grad_norm": 0.8305606429554442,
      "learning_rate": 0.00013450124502973896,
      "loss": 12.31,
      "step": 9292
    },
    {
      "epoch": 1.6250054644808745,
      "grad_norm": 0.6498522709456189,
      "learning_rate": 0.00013448795330264242,
      "loss": 12.4805,
      "step": 9293
    },
    {
      "epoch": 1.6251803278688524,
      "grad_norm": 0.8203945707665701,
      "learning_rate": 0.00013447466088395657,
      "loss": 12.2919,
      "step": 9294
    },
    {
      "epoch": 1.6253551912568307,
      "grad_norm": 0.7384020095006177,
      "learning_rate": 0.00013446136777394788,
      "loss": 12.3386,
      "step": 9295
    },
    {
      "epoch": 1.6255300546448086,
      "grad_norm": 0.7345274092534733,
      "learning_rate": 0.00013444807397288298,
      "loss": 12.3043,
      "step": 9296
    },
    {
      "epoch": 1.625704918032787,
      "grad_norm": 0.6787064249187027,
      "learning_rate": 0.0001344347794810284,
      "loss": 12.3455,
      "step": 9297
    },
    {
      "epoch": 1.625879781420765,
      "grad_norm": 0.8190747307702572,
      "learning_rate": 0.0001344214842986508,
      "loss": 12.1577,
      "step": 9298
    },
    {
      "epoch": 1.6260546448087432,
      "grad_norm": 0.7219770854003479,
      "learning_rate": 0.0001344081884260167,
      "loss": 12.2694,
      "step": 9299
    },
    {
      "epoch": 1.6262295081967213,
      "grad_norm": 0.7266093258261911,
      "learning_rate": 0.00013439489186339282,
      "loss": 12.3615,
      "step": 9300
    },
    {
      "epoch": 1.6264043715846994,
      "grad_norm": 0.7557533923908264,
      "learning_rate": 0.00013438159461104577,
      "loss": 12.4879,
      "step": 9301
    },
    {
      "epoch": 1.6265792349726778,
      "grad_norm": 0.6811149705444858,
      "learning_rate": 0.00013436829666924212,
      "loss": 12.3257,
      "step": 9302
    },
    {
      "epoch": 1.6267540983606557,
      "grad_norm": 0.7250336331618163,
      "learning_rate": 0.00013435499803824864,
      "loss": 12.5355,
      "step": 9303
    },
    {
      "epoch": 1.626928961748634,
      "grad_norm": 0.6999154672186187,
      "learning_rate": 0.00013434169871833192,
      "loss": 12.4186,
      "step": 9304
    },
    {
      "epoch": 1.627103825136612,
      "grad_norm": 0.8226653744385026,
      "learning_rate": 0.00013432839870975878,
      "loss": 12.336,
      "step": 9305
    },
    {
      "epoch": 1.6272786885245902,
      "grad_norm": 0.7344082263322426,
      "learning_rate": 0.00013431509801279577,
      "loss": 12.1979,
      "step": 9306
    },
    {
      "epoch": 1.6274535519125684,
      "grad_norm": 0.809778265604121,
      "learning_rate": 0.0001343017966277097,
      "loss": 12.5793,
      "step": 9307
    },
    {
      "epoch": 1.6276284153005465,
      "grad_norm": 0.7069902174631786,
      "learning_rate": 0.00013428849455476732,
      "loss": 12.2673,
      "step": 9308
    },
    {
      "epoch": 1.6278032786885246,
      "grad_norm": 0.717517705727052,
      "learning_rate": 0.00013427519179423528,
      "loss": 12.514,
      "step": 9309
    },
    {
      "epoch": 1.6279781420765027,
      "grad_norm": 0.7511479616006113,
      "learning_rate": 0.00013426188834638047,
      "loss": 12.13,
      "step": 9310
    },
    {
      "epoch": 1.6281530054644808,
      "grad_norm": 0.7861109998364936,
      "learning_rate": 0.0001342485842114696,
      "loss": 12.4251,
      "step": 9311
    },
    {
      "epoch": 1.628327868852459,
      "grad_norm": 0.7931581403648896,
      "learning_rate": 0.00013423527938976942,
      "loss": 12.4897,
      "step": 9312
    },
    {
      "epoch": 1.6285027322404373,
      "grad_norm": 0.7116133118566919,
      "learning_rate": 0.0001342219738815468,
      "loss": 12.4701,
      "step": 9313
    },
    {
      "epoch": 1.6286775956284152,
      "grad_norm": 0.9425186237038944,
      "learning_rate": 0.00013420866768706853,
      "loss": 12.2339,
      "step": 9314
    },
    {
      "epoch": 1.6288524590163935,
      "grad_norm": 0.7473541603117387,
      "learning_rate": 0.00013419536080660143,
      "loss": 12.1978,
      "step": 9315
    },
    {
      "epoch": 1.6290273224043714,
      "grad_norm": 0.9567558598873064,
      "learning_rate": 0.00013418205324041232,
      "loss": 12.5948,
      "step": 9316
    },
    {
      "epoch": 1.6292021857923498,
      "grad_norm": 0.7385247637761841,
      "learning_rate": 0.00013416874498876812,
      "loss": 12.1885,
      "step": 9317
    },
    {
      "epoch": 1.6293770491803279,
      "grad_norm": 0.7154049272735888,
      "learning_rate": 0.00013415543605193567,
      "loss": 12.2098,
      "step": 9318
    },
    {
      "epoch": 1.629551912568306,
      "grad_norm": 0.784528049745702,
      "learning_rate": 0.00013414212643018188,
      "loss": 12.5298,
      "step": 9319
    },
    {
      "epoch": 1.6297267759562841,
      "grad_norm": 0.6933533686081822,
      "learning_rate": 0.00013412881612377358,
      "loss": 12.4108,
      "step": 9320
    },
    {
      "epoch": 1.6299016393442622,
      "grad_norm": 0.8525077360042859,
      "learning_rate": 0.00013411550513297774,
      "loss": 12.3368,
      "step": 9321
    },
    {
      "epoch": 1.6300765027322406,
      "grad_norm": 0.7897011976825655,
      "learning_rate": 0.0001341021934580613,
      "loss": 12.5569,
      "step": 9322
    },
    {
      "epoch": 1.6302513661202185,
      "grad_norm": 0.8011790390833848,
      "learning_rate": 0.00013408888109929114,
      "loss": 12.1023,
      "step": 9323
    },
    {
      "epoch": 1.6304262295081968,
      "grad_norm": 0.8178288902643469,
      "learning_rate": 0.00013407556805693428,
      "loss": 12.5051,
      "step": 9324
    },
    {
      "epoch": 1.6306010928961747,
      "grad_norm": 0.7774020845195551,
      "learning_rate": 0.00013406225433125764,
      "loss": 12.493,
      "step": 9325
    },
    {
      "epoch": 1.630775956284153,
      "grad_norm": 0.7137767979129253,
      "learning_rate": 0.0001340489399225282,
      "loss": 12.3628,
      "step": 9326
    },
    {
      "epoch": 1.6309508196721312,
      "grad_norm": 0.7403708604939525,
      "learning_rate": 0.00013403562483101298,
      "loss": 12.3142,
      "step": 9327
    },
    {
      "epoch": 1.6311256830601093,
      "grad_norm": 0.7635716911616972,
      "learning_rate": 0.00013402230905697898,
      "loss": 12.4321,
      "step": 9328
    },
    {
      "epoch": 1.6313005464480874,
      "grad_norm": 0.6561575273083431,
      "learning_rate": 0.00013400899260069323,
      "loss": 12.4234,
      "step": 9329
    },
    {
      "epoch": 1.6314754098360655,
      "grad_norm": 0.722696664278108,
      "learning_rate": 0.00013399567546242276,
      "loss": 12.4301,
      "step": 9330
    },
    {
      "epoch": 1.6316502732240439,
      "grad_norm": 0.7429964141375114,
      "learning_rate": 0.00013398235764243462,
      "loss": 12.4501,
      "step": 9331
    },
    {
      "epoch": 1.6318251366120218,
      "grad_norm": 0.7191049848817289,
      "learning_rate": 0.00013396903914099584,
      "loss": 12.305,
      "step": 9332
    },
    {
      "epoch": 1.6320000000000001,
      "grad_norm": 0.7606225794489083,
      "learning_rate": 0.00013395571995837355,
      "loss": 12.4311,
      "step": 9333
    },
    {
      "epoch": 1.632174863387978,
      "grad_norm": 0.7122043442686359,
      "learning_rate": 0.0001339424000948348,
      "loss": 12.358,
      "step": 9334
    },
    {
      "epoch": 1.6323497267759564,
      "grad_norm": 0.6484385109412694,
      "learning_rate": 0.00013392907955064675,
      "loss": 12.3174,
      "step": 9335
    },
    {
      "epoch": 1.6325245901639345,
      "grad_norm": 0.6658055598774796,
      "learning_rate": 0.00013391575832607643,
      "loss": 12.1688,
      "step": 9336
    },
    {
      "epoch": 1.6326994535519126,
      "grad_norm": 0.7640200446797005,
      "learning_rate": 0.00013390243642139103,
      "loss": 12.3955,
      "step": 9337
    },
    {
      "epoch": 1.6328743169398907,
      "grad_norm": 0.6303983484715081,
      "learning_rate": 0.00013388911383685768,
      "loss": 12.3944,
      "step": 9338
    },
    {
      "epoch": 1.6330491803278688,
      "grad_norm": 0.6457394858558191,
      "learning_rate": 0.0001338757905727436,
      "loss": 12.34,
      "step": 9339
    },
    {
      "epoch": 1.633224043715847,
      "grad_norm": 0.7260188381729117,
      "learning_rate": 0.00013386246662931586,
      "loss": 12.3288,
      "step": 9340
    },
    {
      "epoch": 1.633398907103825,
      "grad_norm": 0.7815285348809766,
      "learning_rate": 0.00013384914200684168,
      "loss": 12.3526,
      "step": 9341
    },
    {
      "epoch": 1.6335737704918034,
      "grad_norm": 0.7637736002025464,
      "learning_rate": 0.00013383581670558834,
      "loss": 12.5051,
      "step": 9342
    },
    {
      "epoch": 1.6337486338797813,
      "grad_norm": 0.6954166630217316,
      "learning_rate": 0.00013382249072582293,
      "loss": 12.4365,
      "step": 9343
    },
    {
      "epoch": 1.6339234972677597,
      "grad_norm": 0.6771609883136821,
      "learning_rate": 0.00013380916406781275,
      "loss": 11.7291,
      "step": 9344
    },
    {
      "epoch": 1.6340983606557375,
      "grad_norm": 0.6905481418431315,
      "learning_rate": 0.000133795836731825,
      "loss": 12.3648,
      "step": 9345
    },
    {
      "epoch": 1.6342732240437159,
      "grad_norm": 0.7200756331667844,
      "learning_rate": 0.00013378250871812694,
      "loss": 12.4297,
      "step": 9346
    },
    {
      "epoch": 1.634448087431694,
      "grad_norm": 0.7302268990534272,
      "learning_rate": 0.00013376918002698594,
      "loss": 12.15,
      "step": 9347
    },
    {
      "epoch": 1.6346229508196721,
      "grad_norm": 0.706898358305757,
      "learning_rate": 0.0001337558506586691,
      "loss": 12.294,
      "step": 9348
    },
    {
      "epoch": 1.6347978142076502,
      "grad_norm": 0.6777039511334669,
      "learning_rate": 0.00013374252061344388,
      "loss": 12.3756,
      "step": 9349
    },
    {
      "epoch": 1.6349726775956284,
      "grad_norm": 0.7051510928178848,
      "learning_rate": 0.00013372918989157744,
      "loss": 12.4766,
      "step": 9350
    },
    {
      "epoch": 1.6351475409836067,
      "grad_norm": 0.7308982749419706,
      "learning_rate": 0.00013371585849333728,
      "loss": 12.3539,
      "step": 9351
    },
    {
      "epoch": 1.6353224043715846,
      "grad_norm": 0.7712510479111819,
      "learning_rate": 0.00013370252641899054,
      "loss": 12.3391,
      "step": 9352
    },
    {
      "epoch": 1.635497267759563,
      "grad_norm": 0.6992798476448344,
      "learning_rate": 0.00013368919366880475,
      "loss": 12.4427,
      "step": 9353
    },
    {
      "epoch": 1.6356721311475408,
      "grad_norm": 0.6467425985974674,
      "learning_rate": 0.00013367586024304714,
      "loss": 12.3323,
      "step": 9354
    },
    {
      "epoch": 1.6358469945355192,
      "grad_norm": 0.7119978628274716,
      "learning_rate": 0.00013366252614198515,
      "loss": 12.1489,
      "step": 9355
    },
    {
      "epoch": 1.6360218579234973,
      "grad_norm": 0.6626938307851463,
      "learning_rate": 0.00013364919136588614,
      "loss": 12.5043,
      "step": 9356
    },
    {
      "epoch": 1.6361967213114754,
      "grad_norm": 0.6754820294016619,
      "learning_rate": 0.0001336358559150175,
      "loss": 12.6407,
      "step": 9357
    },
    {
      "epoch": 1.6363715846994535,
      "grad_norm": 0.7014443701692517,
      "learning_rate": 0.00013362251978964675,
      "loss": 12.4337,
      "step": 9358
    },
    {
      "epoch": 1.6365464480874317,
      "grad_norm": 0.7696951881259078,
      "learning_rate": 0.00013360918299004117,
      "loss": 12.4254,
      "step": 9359
    },
    {
      "epoch": 1.63672131147541,
      "grad_norm": 0.7445107861408166,
      "learning_rate": 0.0001335958455164683,
      "loss": 12.3959,
      "step": 9360
    },
    {
      "epoch": 1.636896174863388,
      "grad_norm": 0.6599456648552341,
      "learning_rate": 0.0001335825073691956,
      "loss": 12.0756,
      "step": 9361
    },
    {
      "epoch": 1.6370710382513662,
      "grad_norm": 0.7493448942455229,
      "learning_rate": 0.00013356916854849048,
      "loss": 12.1084,
      "step": 9362
    },
    {
      "epoch": 1.6372459016393441,
      "grad_norm": 0.7485354089282357,
      "learning_rate": 0.0001335558290546205,
      "loss": 12.35,
      "step": 9363
    },
    {
      "epoch": 1.6374207650273225,
      "grad_norm": 0.7389241871095071,
      "learning_rate": 0.00013354248888785306,
      "loss": 12.3527,
      "step": 9364
    },
    {
      "epoch": 1.6375956284153006,
      "grad_norm": 0.7463214799261843,
      "learning_rate": 0.00013352914804845578,
      "loss": 12.24,
      "step": 9365
    },
    {
      "epoch": 1.6377704918032787,
      "grad_norm": 0.727893037781575,
      "learning_rate": 0.0001335158065366961,
      "loss": 12.3104,
      "step": 9366
    },
    {
      "epoch": 1.6379453551912568,
      "grad_norm": 0.6872715068708503,
      "learning_rate": 0.00013350246435284163,
      "loss": 12.3428,
      "step": 9367
    },
    {
      "epoch": 1.638120218579235,
      "grad_norm": 0.8037703603636663,
      "learning_rate": 0.00013348912149715986,
      "loss": 12.5113,
      "step": 9368
    },
    {
      "epoch": 1.638295081967213,
      "grad_norm": 0.6716155798918182,
      "learning_rate": 0.00013347577796991836,
      "loss": 12.3148,
      "step": 9369
    },
    {
      "epoch": 1.6384699453551912,
      "grad_norm": 0.7697078131034928,
      "learning_rate": 0.00013346243377138477,
      "loss": 12.3712,
      "step": 9370
    },
    {
      "epoch": 1.6386448087431695,
      "grad_norm": 0.6497020702557814,
      "learning_rate": 0.0001334490889018266,
      "loss": 12.378,
      "step": 9371
    },
    {
      "epoch": 1.6388196721311474,
      "grad_norm": 0.8534287011795941,
      "learning_rate": 0.00013343574336151153,
      "loss": 12.5363,
      "step": 9372
    },
    {
      "epoch": 1.6389945355191258,
      "grad_norm": 0.673380321640647,
      "learning_rate": 0.00013342239715070713,
      "loss": 12.5551,
      "step": 9373
    },
    {
      "epoch": 1.6391693989071037,
      "grad_norm": 0.8036774725927274,
      "learning_rate": 0.00013340905026968102,
      "loss": 12.0793,
      "step": 9374
    },
    {
      "epoch": 1.639344262295082,
      "grad_norm": 0.6861252482694395,
      "learning_rate": 0.0001333957027187009,
      "loss": 12.2864,
      "step": 9375
    },
    {
      "epoch": 1.6395191256830601,
      "grad_norm": 0.6805243638359856,
      "learning_rate": 0.0001333823544980344,
      "loss": 12.0595,
      "step": 9376
    },
    {
      "epoch": 1.6396939890710382,
      "grad_norm": 0.7376271228895982,
      "learning_rate": 0.0001333690056079492,
      "loss": 12.2406,
      "step": 9377
    },
    {
      "epoch": 1.6398688524590164,
      "grad_norm": 0.7036965678333029,
      "learning_rate": 0.00013335565604871298,
      "loss": 12.5799,
      "step": 9378
    },
    {
      "epoch": 1.6400437158469945,
      "grad_norm": 0.6521696487115575,
      "learning_rate": 0.00013334230582059343,
      "loss": 12.2596,
      "step": 9379
    },
    {
      "epoch": 1.6402185792349728,
      "grad_norm": 0.6744536730091534,
      "learning_rate": 0.0001333289549238583,
      "loss": 12.5877,
      "step": 9380
    },
    {
      "epoch": 1.6403934426229507,
      "grad_norm": 0.7335429022482077,
      "learning_rate": 0.00013331560335877525,
      "loss": 12.3354,
      "step": 9381
    },
    {
      "epoch": 1.640568306010929,
      "grad_norm": 0.7309528799403356,
      "learning_rate": 0.00013330225112561208,
      "loss": 12.4385,
      "step": 9382
    },
    {
      "epoch": 1.640743169398907,
      "grad_norm": 0.6336190076444417,
      "learning_rate": 0.00013328889822463656,
      "loss": 12.4811,
      "step": 9383
    },
    {
      "epoch": 1.6409180327868853,
      "grad_norm": 0.6918922994208697,
      "learning_rate": 0.00013327554465611632,
      "loss": 12.4971,
      "step": 9384
    },
    {
      "epoch": 1.6410928961748634,
      "grad_norm": 0.6622092468091525,
      "learning_rate": 0.00013326219042031934,
      "loss": 12.2943,
      "step": 9385
    },
    {
      "epoch": 1.6412677595628415,
      "grad_norm": 0.6847791375539084,
      "learning_rate": 0.00013324883551751326,
      "loss": 12.4098,
      "step": 9386
    },
    {
      "epoch": 1.6414426229508197,
      "grad_norm": 0.7142991373908064,
      "learning_rate": 0.00013323547994796597,
      "loss": 12.4813,
      "step": 9387
    },
    {
      "epoch": 1.6416174863387978,
      "grad_norm": 0.7234856304325088,
      "learning_rate": 0.00013322212371194524,
      "loss": 12.4014,
      "step": 9388
    },
    {
      "epoch": 1.641792349726776,
      "grad_norm": 0.7802861816523589,
      "learning_rate": 0.00013320876680971894,
      "loss": 12.4723,
      "step": 9389
    },
    {
      "epoch": 1.641967213114754,
      "grad_norm": 0.6775381607817208,
      "learning_rate": 0.0001331954092415549,
      "loss": 12.1459,
      "step": 9390
    },
    {
      "epoch": 1.6421420765027324,
      "grad_norm": 0.7300316766443696,
      "learning_rate": 0.000133182051007721,
      "loss": 12.4868,
      "step": 9391
    },
    {
      "epoch": 1.6423169398907103,
      "grad_norm": 0.7090587615044772,
      "learning_rate": 0.00013316869210848505,
      "loss": 12.2453,
      "step": 9392
    },
    {
      "epoch": 1.6424918032786886,
      "grad_norm": 0.7520762746515511,
      "learning_rate": 0.000133155332544115,
      "loss": 12.379,
      "step": 9393
    },
    {
      "epoch": 1.6426666666666667,
      "grad_norm": 0.6898696784360846,
      "learning_rate": 0.00013314197231487876,
      "loss": 12.5015,
      "step": 9394
    },
    {
      "epoch": 1.6428415300546448,
      "grad_norm": 0.6460549287633015,
      "learning_rate": 0.00013312861142104418,
      "loss": 12.1957,
      "step": 9395
    },
    {
      "epoch": 1.643016393442623,
      "grad_norm": 0.6445721779927691,
      "learning_rate": 0.00013311524986287922,
      "loss": 12.032,
      "step": 9396
    },
    {
      "epoch": 1.643191256830601,
      "grad_norm": 0.756111458013511,
      "learning_rate": 0.00013310188764065187,
      "loss": 12.4559,
      "step": 9397
    },
    {
      "epoch": 1.6433661202185792,
      "grad_norm": 0.6676338237533868,
      "learning_rate": 0.00013308852475462998,
      "loss": 12.507,
      "step": 9398
    },
    {
      "epoch": 1.6435409836065573,
      "grad_norm": 0.7175476689149763,
      "learning_rate": 0.00013307516120508161,
      "loss": 12.2425,
      "step": 9399
    },
    {
      "epoch": 1.6437158469945357,
      "grad_norm": 0.6973837687914143,
      "learning_rate": 0.00013306179699227469,
      "loss": 12.3943,
      "step": 9400
    },
    {
      "epoch": 1.6438907103825136,
      "grad_norm": 0.7620544850868054,
      "learning_rate": 0.00013304843211647723,
      "loss": 12.3608,
      "step": 9401
    },
    {
      "epoch": 1.644065573770492,
      "grad_norm": 0.715388528360205,
      "learning_rate": 0.00013303506657795725,
      "loss": 12.2823,
      "step": 9402
    },
    {
      "epoch": 1.6442404371584698,
      "grad_norm": 0.6761200381058583,
      "learning_rate": 0.0001330217003769827,
      "loss": 12.1671,
      "step": 9403
    },
    {
      "epoch": 1.6444153005464481,
      "grad_norm": 0.726420766428865,
      "learning_rate": 0.00013300833351382173,
      "loss": 12.3124,
      "step": 9404
    },
    {
      "epoch": 1.6445901639344263,
      "grad_norm": 0.7816314609343952,
      "learning_rate": 0.00013299496598874225,
      "loss": 12.4554,
      "step": 9405
    },
    {
      "epoch": 1.6447650273224044,
      "grad_norm": 0.7685615170704658,
      "learning_rate": 0.0001329815978020125,
      "loss": 12.3919,
      "step": 9406
    },
    {
      "epoch": 1.6449398907103825,
      "grad_norm": 0.7233900548431315,
      "learning_rate": 0.00013296822895390037,
      "loss": 12.3942,
      "step": 9407
    },
    {
      "epoch": 1.6451147540983606,
      "grad_norm": 0.7172833947604268,
      "learning_rate": 0.00013295485944467405,
      "loss": 12.2543,
      "step": 9408
    },
    {
      "epoch": 1.645289617486339,
      "grad_norm": 0.8305010680718978,
      "learning_rate": 0.00013294148927460163,
      "loss": 12.3838,
      "step": 9409
    },
    {
      "epoch": 1.6454644808743168,
      "grad_norm": 0.7506098461001561,
      "learning_rate": 0.00013292811844395117,
      "loss": 12.5861,
      "step": 9410
    },
    {
      "epoch": 1.6456393442622952,
      "grad_norm": 0.8592291561931594,
      "learning_rate": 0.00013291474695299084,
      "loss": 12.2479,
      "step": 9411
    },
    {
      "epoch": 1.645814207650273,
      "grad_norm": 0.7105526648622588,
      "learning_rate": 0.00013290137480198875,
      "loss": 12.3591,
      "step": 9412
    },
    {
      "epoch": 1.6459890710382514,
      "grad_norm": 0.9429456644958571,
      "learning_rate": 0.0001328880019912131,
      "loss": 12.3577,
      "step": 9413
    },
    {
      "epoch": 1.6461639344262295,
      "grad_norm": 0.6152843871832377,
      "learning_rate": 0.00013287462852093205,
      "loss": 12.2249,
      "step": 9414
    },
    {
      "epoch": 1.6463387978142077,
      "grad_norm": 0.7405415516916536,
      "learning_rate": 0.00013286125439141374,
      "loss": 12.4417,
      "step": 9415
    },
    {
      "epoch": 1.6465136612021858,
      "grad_norm": 0.6761405182428033,
      "learning_rate": 0.0001328478796029264,
      "loss": 12.232,
      "step": 9416
    },
    {
      "epoch": 1.646688524590164,
      "grad_norm": 0.6467811333234194,
      "learning_rate": 0.0001328345041557382,
      "loss": 12.2308,
      "step": 9417
    },
    {
      "epoch": 1.646863387978142,
      "grad_norm": 0.6786538168812363,
      "learning_rate": 0.0001328211280501174,
      "loss": 12.3168,
      "step": 9418
    },
    {
      "epoch": 1.6470382513661201,
      "grad_norm": 0.7365245605742375,
      "learning_rate": 0.00013280775128633217,
      "loss": 12.6172,
      "step": 9419
    },
    {
      "epoch": 1.6472131147540985,
      "grad_norm": 0.7155244680872676,
      "learning_rate": 0.00013279437386465082,
      "loss": 12.2757,
      "step": 9420
    },
    {
      "epoch": 1.6473879781420764,
      "grad_norm": 0.6932833206739735,
      "learning_rate": 0.00013278099578534157,
      "loss": 12.4518,
      "step": 9421
    },
    {
      "epoch": 1.6475628415300547,
      "grad_norm": 0.7819576964064833,
      "learning_rate": 0.00013276761704867268,
      "loss": 12.1461,
      "step": 9422
    },
    {
      "epoch": 1.6477377049180328,
      "grad_norm": 0.7707407189609671,
      "learning_rate": 0.0001327542376549125,
      "loss": 12.1828,
      "step": 9423
    },
    {
      "epoch": 1.647912568306011,
      "grad_norm": 0.7609179562363098,
      "learning_rate": 0.00013274085760432927,
      "loss": 12.3621,
      "step": 9424
    },
    {
      "epoch": 1.648087431693989,
      "grad_norm": 0.7306161428433663,
      "learning_rate": 0.00013272747689719132,
      "loss": 12.4113,
      "step": 9425
    },
    {
      "epoch": 1.6482622950819672,
      "grad_norm": 0.7463037529905396,
      "learning_rate": 0.000132714095533767,
      "loss": 12.3957,
      "step": 9426
    },
    {
      "epoch": 1.6484371584699453,
      "grad_norm": 0.7551648023307133,
      "learning_rate": 0.00013270071351432456,
      "loss": 12.5058,
      "step": 9427
    },
    {
      "epoch": 1.6486120218579234,
      "grad_norm": 0.6524038538664922,
      "learning_rate": 0.00013268733083913244,
      "loss": 12.4702,
      "step": 9428
    },
    {
      "epoch": 1.6487868852459018,
      "grad_norm": 0.7221305375953369,
      "learning_rate": 0.000132673947508459,
      "loss": 12.4634,
      "step": 9429
    },
    {
      "epoch": 1.6489617486338797,
      "grad_norm": 0.7203134292058209,
      "learning_rate": 0.00013266056352257254,
      "loss": 12.3106,
      "step": 9430
    },
    {
      "epoch": 1.649136612021858,
      "grad_norm": 0.6838841486515551,
      "learning_rate": 0.00013264717888174156,
      "loss": 12.275,
      "step": 9431
    },
    {
      "epoch": 1.649311475409836,
      "grad_norm": 0.6945268981576918,
      "learning_rate": 0.00013263379358623433,
      "loss": 12.5234,
      "step": 9432
    },
    {
      "epoch": 1.6494863387978143,
      "grad_norm": 0.7029593700709332,
      "learning_rate": 0.00013262040763631938,
      "loss": 12.2725,
      "step": 9433
    },
    {
      "epoch": 1.6496612021857924,
      "grad_norm": 0.729833691851587,
      "learning_rate": 0.0001326070210322651,
      "loss": 12.3114,
      "step": 9434
    },
    {
      "epoch": 1.6498360655737705,
      "grad_norm": 0.6851119162032083,
      "learning_rate": 0.00013259363377433994,
      "loss": 12.2179,
      "step": 9435
    },
    {
      "epoch": 1.6500109289617486,
      "grad_norm": 0.6299110086664272,
      "learning_rate": 0.00013258024586281232,
      "loss": 12.4028,
      "step": 9436
    },
    {
      "epoch": 1.6501857923497267,
      "grad_norm": 0.650608646352871,
      "learning_rate": 0.00013256685729795076,
      "loss": 12.5007,
      "step": 9437
    },
    {
      "epoch": 1.650360655737705,
      "grad_norm": 0.6413614214277997,
      "learning_rate": 0.00013255346808002373,
      "loss": 12.0012,
      "step": 9438
    },
    {
      "epoch": 1.650535519125683,
      "grad_norm": 0.7841570881516678,
      "learning_rate": 0.00013254007820929966,
      "loss": 12.5329,
      "step": 9439
    },
    {
      "epoch": 1.6507103825136613,
      "grad_norm": 0.6775621550414808,
      "learning_rate": 0.00013252668768604716,
      "loss": 12.4124,
      "step": 9440
    },
    {
      "epoch": 1.6508852459016392,
      "grad_norm": 0.7323842408434574,
      "learning_rate": 0.00013251329651053467,
      "loss": 12.5139,
      "step": 9441
    },
    {
      "epoch": 1.6510601092896176,
      "grad_norm": 0.6518296079985147,
      "learning_rate": 0.00013249990468303078,
      "loss": 12.4329,
      "step": 9442
    },
    {
      "epoch": 1.6512349726775957,
      "grad_norm": 0.6738669461977607,
      "learning_rate": 0.00013248651220380402,
      "loss": 12.3969,
      "step": 9443
    },
    {
      "epoch": 1.6514098360655738,
      "grad_norm": 0.6803195918102454,
      "learning_rate": 0.0001324731190731229,
      "loss": 12.0736,
      "step": 9444
    },
    {
      "epoch": 1.651584699453552,
      "grad_norm": 0.619806090541969,
      "learning_rate": 0.00013245972529125606,
      "loss": 12.2307,
      "step": 9445
    },
    {
      "epoch": 1.65175956284153,
      "grad_norm": 0.6750327964071625,
      "learning_rate": 0.00013244633085847206,
      "loss": 12.5236,
      "step": 9446
    },
    {
      "epoch": 1.6519344262295081,
      "grad_norm": 0.7078788343932977,
      "learning_rate": 0.00013243293577503955,
      "loss": 12.0691,
      "step": 9447
    },
    {
      "epoch": 1.6521092896174863,
      "grad_norm": 0.6405382829145689,
      "learning_rate": 0.00013241954004122704,
      "loss": 12.2483,
      "step": 9448
    },
    {
      "epoch": 1.6522841530054646,
      "grad_norm": 0.7231238641464633,
      "learning_rate": 0.00013240614365730324,
      "loss": 12.3733,
      "step": 9449
    },
    {
      "epoch": 1.6524590163934425,
      "grad_norm": 0.6724414099974235,
      "learning_rate": 0.00013239274662353673,
      "loss": 12.5056,
      "step": 9450
    },
    {
      "epoch": 1.6526338797814208,
      "grad_norm": 0.6942606674747377,
      "learning_rate": 0.0001323793489401962,
      "loss": 12.4134,
      "step": 9451
    },
    {
      "epoch": 1.6528087431693987,
      "grad_norm": 0.6825767670863628,
      "learning_rate": 0.00013236595060755032,
      "loss": 12.5401,
      "step": 9452
    },
    {
      "epoch": 1.652983606557377,
      "grad_norm": 0.633651996942118,
      "learning_rate": 0.00013235255162586773,
      "loss": 12.3328,
      "step": 9453
    },
    {
      "epoch": 1.6531584699453552,
      "grad_norm": 0.7165957910683497,
      "learning_rate": 0.0001323391519954172,
      "loss": 12.4128,
      "step": 9454
    },
    {
      "epoch": 1.6533333333333333,
      "grad_norm": 0.6692409701104088,
      "learning_rate": 0.00013232575171646732,
      "loss": 12.3026,
      "step": 9455
    },
    {
      "epoch": 1.6535081967213114,
      "grad_norm": 0.6661822229193596,
      "learning_rate": 0.0001323123507892869,
      "loss": 12.2738,
      "step": 9456
    },
    {
      "epoch": 1.6536830601092896,
      "grad_norm": 0.6991175136289223,
      "learning_rate": 0.0001322989492141446,
      "loss": 12.2875,
      "step": 9457
    },
    {
      "epoch": 1.653857923497268,
      "grad_norm": 0.7564756911269844,
      "learning_rate": 0.00013228554699130923,
      "loss": 12.0856,
      "step": 9458
    },
    {
      "epoch": 1.6540327868852458,
      "grad_norm": 0.7011284344939338,
      "learning_rate": 0.00013227214412104952,
      "loss": 12.1993,
      "step": 9459
    },
    {
      "epoch": 1.6542076502732241,
      "grad_norm": 0.6897940526911903,
      "learning_rate": 0.0001322587406036342,
      "loss": 12.0982,
      "step": 9460
    },
    {
      "epoch": 1.654382513661202,
      "grad_norm": 0.7033065154444236,
      "learning_rate": 0.00013224533643933211,
      "loss": 12.553,
      "step": 9461
    },
    {
      "epoch": 1.6545573770491804,
      "grad_norm": 0.695249301035729,
      "learning_rate": 0.000132231931628412,
      "loss": 12.401,
      "step": 9462
    },
    {
      "epoch": 1.6547322404371585,
      "grad_norm": 0.6930546711606156,
      "learning_rate": 0.0001322185261711427,
      "loss": 12.377,
      "step": 9463
    },
    {
      "epoch": 1.6549071038251366,
      "grad_norm": 0.6624895184192925,
      "learning_rate": 0.00013220512006779301,
      "loss": 12.4943,
      "step": 9464
    },
    {
      "epoch": 1.6550819672131147,
      "grad_norm": 0.7890195566435696,
      "learning_rate": 0.0001321917133186318,
      "loss": 12.4887,
      "step": 9465
    },
    {
      "epoch": 1.6552568306010929,
      "grad_norm": 0.7187858788963734,
      "learning_rate": 0.0001321783059239279,
      "loss": 12.3886,
      "step": 9466
    },
    {
      "epoch": 1.6554316939890712,
      "grad_norm": 0.7137128760963942,
      "learning_rate": 0.00013216489788395018,
      "loss": 12.2815,
      "step": 9467
    },
    {
      "epoch": 1.655606557377049,
      "grad_norm": 0.7006734615640932,
      "learning_rate": 0.00013215148919896744,
      "loss": 12.4373,
      "step": 9468
    },
    {
      "epoch": 1.6557814207650274,
      "grad_norm": 0.7270089755874004,
      "learning_rate": 0.00013213807986924868,
      "loss": 12.6517,
      "step": 9469
    },
    {
      "epoch": 1.6559562841530053,
      "grad_norm": 0.7645405945363263,
      "learning_rate": 0.00013212466989506272,
      "loss": 12.6264,
      "step": 9470
    },
    {
      "epoch": 1.6561311475409837,
      "grad_norm": 0.7857032624764512,
      "learning_rate": 0.0001321112592766785,
      "loss": 12.3742,
      "step": 9471
    },
    {
      "epoch": 1.6563060109289618,
      "grad_norm": 0.7016969432247565,
      "learning_rate": 0.00013209784801436494,
      "loss": 12.3464,
      "step": 9472
    },
    {
      "epoch": 1.65648087431694,
      "grad_norm": 0.7516212365550267,
      "learning_rate": 0.00013208443610839097,
      "loss": 12.52,
      "step": 9473
    },
    {
      "epoch": 1.656655737704918,
      "grad_norm": 0.6737665163985201,
      "learning_rate": 0.00013207102355902552,
      "loss": 12.3565,
      "step": 9474
    },
    {
      "epoch": 1.6568306010928961,
      "grad_norm": 0.7261717183524924,
      "learning_rate": 0.0001320576103665376,
      "loss": 12.4626,
      "step": 9475
    },
    {
      "epoch": 1.6570054644808743,
      "grad_norm": 0.7418470370351075,
      "learning_rate": 0.00013204419653119615,
      "loss": 12.325,
      "step": 9476
    },
    {
      "epoch": 1.6571803278688524,
      "grad_norm": 0.7312843180495634,
      "learning_rate": 0.0001320307820532702,
      "loss": 12.4893,
      "step": 9477
    },
    {
      "epoch": 1.6573551912568307,
      "grad_norm": 0.7623549064033674,
      "learning_rate": 0.00013201736693302867,
      "loss": 12.4961,
      "step": 9478
    },
    {
      "epoch": 1.6575300546448086,
      "grad_norm": 0.6981886322891968,
      "learning_rate": 0.00013200395117074073,
      "loss": 12.2353,
      "step": 9479
    },
    {
      "epoch": 1.657704918032787,
      "grad_norm": 0.7596277682727284,
      "learning_rate": 0.00013199053476667518,
      "loss": 12.3061,
      "step": 9480
    },
    {
      "epoch": 1.6578797814207649,
      "grad_norm": 0.7113748290420892,
      "learning_rate": 0.00013197711772110126,
      "loss": 12.0979,
      "step": 9481
    },
    {
      "epoch": 1.6580546448087432,
      "grad_norm": 0.7060434306840653,
      "learning_rate": 0.0001319637000342879,
      "loss": 12.4627,
      "step": 9482
    },
    {
      "epoch": 1.6582295081967213,
      "grad_norm": 0.7433498775057095,
      "learning_rate": 0.00013195028170650424,
      "loss": 12.332,
      "step": 9483
    },
    {
      "epoch": 1.6584043715846994,
      "grad_norm": 0.6679645560209946,
      "learning_rate": 0.00013193686273801932,
      "loss": 12.3524,
      "step": 9484
    },
    {
      "epoch": 1.6585792349726776,
      "grad_norm": 0.7090077492278847,
      "learning_rate": 0.00013192344312910228,
      "loss": 12.3065,
      "step": 9485
    },
    {
      "epoch": 1.6587540983606557,
      "grad_norm": 0.6261377092316809,
      "learning_rate": 0.00013191002288002218,
      "loss": 12.3016,
      "step": 9486
    },
    {
      "epoch": 1.658928961748634,
      "grad_norm": 0.703258192992267,
      "learning_rate": 0.0001318966019910481,
      "loss": 12.1638,
      "step": 9487
    },
    {
      "epoch": 1.659103825136612,
      "grad_norm": 0.7670029163833981,
      "learning_rate": 0.00013188318046244925,
      "loss": 12.0511,
      "step": 9488
    },
    {
      "epoch": 1.6592786885245903,
      "grad_norm": 0.6563464848316377,
      "learning_rate": 0.0001318697582944947,
      "loss": 12.1272,
      "step": 9489
    },
    {
      "epoch": 1.6594535519125682,
      "grad_norm": 0.6962182683140973,
      "learning_rate": 0.00013185633548745367,
      "loss": 12.3813,
      "step": 9490
    },
    {
      "epoch": 1.6596284153005465,
      "grad_norm": 0.7120228912793429,
      "learning_rate": 0.00013184291204159534,
      "loss": 12.255,
      "step": 9491
    },
    {
      "epoch": 1.6598032786885246,
      "grad_norm": 0.7415858114349387,
      "learning_rate": 0.00013182948795718876,
      "loss": 12.4495,
      "step": 9492
    },
    {
      "epoch": 1.6599781420765027,
      "grad_norm": 0.660781532672647,
      "learning_rate": 0.0001318160632345033,
      "loss": 12.4052,
      "step": 9493
    },
    {
      "epoch": 1.6601530054644809,
      "grad_norm": 0.7097361573990478,
      "learning_rate": 0.00013180263787380804,
      "loss": 12.5229,
      "step": 9494
    },
    {
      "epoch": 1.660327868852459,
      "grad_norm": 0.8012062444346478,
      "learning_rate": 0.0001317892118753723,
      "loss": 12.3906,
      "step": 9495
    },
    {
      "epoch": 1.6605027322404373,
      "grad_norm": 0.7120466711994459,
      "learning_rate": 0.00013177578523946516,
      "loss": 12.2257,
      "step": 9496
    },
    {
      "epoch": 1.6606775956284152,
      "grad_norm": 0.7297403991767094,
      "learning_rate": 0.00013176235796635605,
      "loss": 12.1095,
      "step": 9497
    },
    {
      "epoch": 1.6608524590163936,
      "grad_norm": 0.6927320798971622,
      "learning_rate": 0.00013174893005631414,
      "loss": 12.1576,
      "step": 9498
    },
    {
      "epoch": 1.6610273224043715,
      "grad_norm": 0.7749749247202885,
      "learning_rate": 0.00013173550150960862,
      "loss": 12.455,
      "step": 9499
    },
    {
      "epoch": 1.6612021857923498,
      "grad_norm": 0.6699625534686929,
      "learning_rate": 0.00013172207232650893,
      "loss": 12.0352,
      "step": 9500
    },
    {
      "epoch": 1.661377049180328,
      "grad_norm": 0.7171847281814269,
      "learning_rate": 0.00013170864250728424,
      "loss": 12.2972,
      "step": 9501
    },
    {
      "epoch": 1.661551912568306,
      "grad_norm": 0.8899008831639932,
      "learning_rate": 0.00013169521205220395,
      "loss": 12.3243,
      "step": 9502
    },
    {
      "epoch": 1.6617267759562842,
      "grad_norm": 0.69402802789509,
      "learning_rate": 0.0001316817809615373,
      "loss": 12.4474,
      "step": 9503
    },
    {
      "epoch": 1.6619016393442623,
      "grad_norm": 0.8967176428657828,
      "learning_rate": 0.0001316683492355537,
      "loss": 12.2487,
      "step": 9504
    },
    {
      "epoch": 1.6620765027322404,
      "grad_norm": 0.7567884598272353,
      "learning_rate": 0.00013165491687452244,
      "loss": 12.1253,
      "step": 9505
    },
    {
      "epoch": 1.6622513661202185,
      "grad_norm": 0.9212071357242597,
      "learning_rate": 0.00013164148387871292,
      "loss": 12.2554,
      "step": 9506
    },
    {
      "epoch": 1.6624262295081969,
      "grad_norm": 0.6935138806854168,
      "learning_rate": 0.00013162805024839448,
      "loss": 12.2608,
      "step": 9507
    },
    {
      "epoch": 1.6626010928961747,
      "grad_norm": 0.8657013931309921,
      "learning_rate": 0.00013161461598383653,
      "loss": 12.4484,
      "step": 9508
    },
    {
      "epoch": 1.662775956284153,
      "grad_norm": 0.7333202625625195,
      "learning_rate": 0.00013160118108530847,
      "loss": 12.2838,
      "step": 9509
    },
    {
      "epoch": 1.662950819672131,
      "grad_norm": 0.8883151746286669,
      "learning_rate": 0.0001315877455530797,
      "loss": 12.6009,
      "step": 9510
    },
    {
      "epoch": 1.6631256830601093,
      "grad_norm": 0.7583342405139171,
      "learning_rate": 0.00013157430938741963,
      "loss": 12.1935,
      "step": 9511
    },
    {
      "epoch": 1.6633005464480874,
      "grad_norm": 0.7363672265126392,
      "learning_rate": 0.00013156087258859773,
      "loss": 12.4985,
      "step": 9512
    },
    {
      "epoch": 1.6634754098360656,
      "grad_norm": 0.8427215815465394,
      "learning_rate": 0.0001315474351568834,
      "loss": 12.4838,
      "step": 9513
    },
    {
      "epoch": 1.6636502732240437,
      "grad_norm": 0.6775203077726607,
      "learning_rate": 0.00013153399709254617,
      "loss": 11.9855,
      "step": 9514
    },
    {
      "epoch": 1.6638251366120218,
      "grad_norm": 1.012054957920294,
      "learning_rate": 0.00013152055839585545,
      "loss": 12.3771,
      "step": 9515
    },
    {
      "epoch": 1.6640000000000001,
      "grad_norm": 0.7075689704095872,
      "learning_rate": 0.00013150711906708077,
      "loss": 12.1535,
      "step": 9516
    },
    {
      "epoch": 1.664174863387978,
      "grad_norm": 0.7604973774083403,
      "learning_rate": 0.00013149367910649158,
      "loss": 12.3209,
      "step": 9517
    },
    {
      "epoch": 1.6643497267759564,
      "grad_norm": 0.8222180477288098,
      "learning_rate": 0.00013148023851435747,
      "loss": 12.2896,
      "step": 9518
    },
    {
      "epoch": 1.6645245901639343,
      "grad_norm": 0.712477317441159,
      "learning_rate": 0.0001314667972909479,
      "loss": 12.4813,
      "step": 9519
    },
    {
      "epoch": 1.6646994535519126,
      "grad_norm": 0.8423264471291022,
      "learning_rate": 0.0001314533554365324,
      "loss": 11.9857,
      "step": 9520
    },
    {
      "epoch": 1.6648743169398907,
      "grad_norm": 0.6781033016134895,
      "learning_rate": 0.00013143991295138058,
      "loss": 12.2481,
      "step": 9521
    },
    {
      "epoch": 1.6650491803278689,
      "grad_norm": 0.797194881572754,
      "learning_rate": 0.00013142646983576196,
      "loss": 12.2443,
      "step": 9522
    },
    {
      "epoch": 1.665224043715847,
      "grad_norm": 0.65780477580762,
      "learning_rate": 0.00013141302608994613,
      "loss": 12.3153,
      "step": 9523
    },
    {
      "epoch": 1.665398907103825,
      "grad_norm": 0.6795312844462444,
      "learning_rate": 0.0001313995817142027,
      "loss": 12.3299,
      "step": 9524
    },
    {
      "epoch": 1.6655737704918034,
      "grad_norm": 0.669444700535303,
      "learning_rate": 0.00013138613670880123,
      "loss": 12.4915,
      "step": 9525
    },
    {
      "epoch": 1.6657486338797813,
      "grad_norm": 0.7293795431445594,
      "learning_rate": 0.00013137269107401135,
      "loss": 12.5744,
      "step": 9526
    },
    {
      "epoch": 1.6659234972677597,
      "grad_norm": 0.6914076053189838,
      "learning_rate": 0.00013135924481010272,
      "loss": 12.4628,
      "step": 9527
    },
    {
      "epoch": 1.6660983606557376,
      "grad_norm": 0.6818506532166139,
      "learning_rate": 0.0001313457979173449,
      "loss": 12.3824,
      "step": 9528
    },
    {
      "epoch": 1.666273224043716,
      "grad_norm": 0.6948340706656517,
      "learning_rate": 0.0001313323503960076,
      "loss": 12.3952,
      "step": 9529
    },
    {
      "epoch": 1.666448087431694,
      "grad_norm": 0.7109442213549374,
      "learning_rate": 0.00013131890224636045,
      "loss": 12.3585,
      "step": 9530
    },
    {
      "epoch": 1.6666229508196722,
      "grad_norm": 0.6906155004369409,
      "learning_rate": 0.00013130545346867317,
      "loss": 12.449,
      "step": 9531
    },
    {
      "epoch": 1.6667978142076503,
      "grad_norm": 0.7593493873178359,
      "learning_rate": 0.00013129200406321545,
      "loss": 12.5351,
      "step": 9532
    },
    {
      "epoch": 1.6669726775956284,
      "grad_norm": 0.7870633151047808,
      "learning_rate": 0.00013127855403025692,
      "loss": 12.4863,
      "step": 9533
    },
    {
      "epoch": 1.6671475409836065,
      "grad_norm": 0.7222971784369858,
      "learning_rate": 0.0001312651033700674,
      "loss": 12.4581,
      "step": 9534
    },
    {
      "epoch": 1.6673224043715846,
      "grad_norm": 0.6443970692977585,
      "learning_rate": 0.00013125165208291653,
      "loss": 12.0091,
      "step": 9535
    },
    {
      "epoch": 1.667497267759563,
      "grad_norm": 0.7143575634786253,
      "learning_rate": 0.0001312382001690741,
      "loss": 12.4645,
      "step": 9536
    },
    {
      "epoch": 1.6676721311475409,
      "grad_norm": 0.6860827305734977,
      "learning_rate": 0.00013122474762880982,
      "loss": 12.6024,
      "step": 9537
    },
    {
      "epoch": 1.6678469945355192,
      "grad_norm": 0.762874412491159,
      "learning_rate": 0.0001312112944623935,
      "loss": 12.3396,
      "step": 9538
    },
    {
      "epoch": 1.668021857923497,
      "grad_norm": 0.7489623355667101,
      "learning_rate": 0.00013119784067009487,
      "loss": 12.0714,
      "step": 9539
    },
    {
      "epoch": 1.6681967213114755,
      "grad_norm": 0.7205682965223279,
      "learning_rate": 0.00013118438625218373,
      "loss": 12.3675,
      "step": 9540
    },
    {
      "epoch": 1.6683715846994536,
      "grad_norm": 0.7212598962931699,
      "learning_rate": 0.00013117093120892993,
      "loss": 12.2557,
      "step": 9541
    },
    {
      "epoch": 1.6685464480874317,
      "grad_norm": 0.6833205342261758,
      "learning_rate": 0.00013115747554060322,
      "loss": 12.4014,
      "step": 9542
    },
    {
      "epoch": 1.6687213114754098,
      "grad_norm": 0.6969826260580838,
      "learning_rate": 0.0001311440192474735,
      "loss": 12.3879,
      "step": 9543
    },
    {
      "epoch": 1.668896174863388,
      "grad_norm": 0.7696951539631165,
      "learning_rate": 0.00013113056232981054,
      "loss": 12.3668,
      "step": 9544
    },
    {
      "epoch": 1.6690710382513663,
      "grad_norm": 0.6848106377237686,
      "learning_rate": 0.00013111710478788425,
      "loss": 12.4155,
      "step": 9545
    },
    {
      "epoch": 1.6692459016393442,
      "grad_norm": 0.6631838941448781,
      "learning_rate": 0.00013110364662196445,
      "loss": 11.9436,
      "step": 9546
    },
    {
      "epoch": 1.6694207650273225,
      "grad_norm": 0.7087501214976449,
      "learning_rate": 0.00013109018783232103,
      "loss": 12.0031,
      "step": 9547
    },
    {
      "epoch": 1.6695956284153004,
      "grad_norm": 0.6630956918253926,
      "learning_rate": 0.00013107672841922388,
      "loss": 12.2876,
      "step": 9548
    },
    {
      "epoch": 1.6697704918032787,
      "grad_norm": 0.8042385316927183,
      "learning_rate": 0.0001310632683829429,
      "loss": 12.4667,
      "step": 9549
    },
    {
      "epoch": 1.6699453551912569,
      "grad_norm": 0.7333390768556453,
      "learning_rate": 0.00013104980772374807,
      "loss": 12.2738,
      "step": 9550
    },
    {
      "epoch": 1.670120218579235,
      "grad_norm": 0.678766689181608,
      "learning_rate": 0.00013103634644190921,
      "loss": 11.9106,
      "step": 9551
    },
    {
      "epoch": 1.670295081967213,
      "grad_norm": 0.6698929893647674,
      "learning_rate": 0.00013102288453769632,
      "loss": 12.3182,
      "step": 9552
    },
    {
      "epoch": 1.6704699453551912,
      "grad_norm": 0.8777882574213536,
      "learning_rate": 0.00013100942201137934,
      "loss": 11.9502,
      "step": 9553
    },
    {
      "epoch": 1.6706448087431696,
      "grad_norm": 0.7344941330330669,
      "learning_rate": 0.00013099595886322825,
      "loss": 12.349,
      "step": 9554
    },
    {
      "epoch": 1.6708196721311475,
      "grad_norm": 0.773702203831619,
      "learning_rate": 0.00013098249509351298,
      "loss": 12.3531,
      "step": 9555
    },
    {
      "epoch": 1.6709945355191258,
      "grad_norm": 0.7244681902310987,
      "learning_rate": 0.0001309690307025036,
      "loss": 12.1857,
      "step": 9556
    },
    {
      "epoch": 1.6711693989071037,
      "grad_norm": 0.7913118048551291,
      "learning_rate": 0.00013095556569047003,
      "loss": 12.5108,
      "step": 9557
    },
    {
      "epoch": 1.671344262295082,
      "grad_norm": 0.8377477250325067,
      "learning_rate": 0.00013094210005768232,
      "loss": 12.3052,
      "step": 9558
    },
    {
      "epoch": 1.6715191256830602,
      "grad_norm": 0.7473498255683261,
      "learning_rate": 0.00013092863380441053,
      "loss": 12.5166,
      "step": 9559
    },
    {
      "epoch": 1.6716939890710383,
      "grad_norm": 0.8075064821753225,
      "learning_rate": 0.00013091516693092465,
      "loss": 12.4918,
      "step": 9560
    },
    {
      "epoch": 1.6718688524590164,
      "grad_norm": 0.6987852112806919,
      "learning_rate": 0.00013090169943749476,
      "loss": 12.4651,
      "step": 9561
    },
    {
      "epoch": 1.6720437158469945,
      "grad_norm": 0.737676916743012,
      "learning_rate": 0.0001308882313243909,
      "loss": 12.4434,
      "step": 9562
    },
    {
      "epoch": 1.6722185792349726,
      "grad_norm": 0.7629496595654804,
      "learning_rate": 0.00013087476259188318,
      "loss": 12.2651,
      "step": 9563
    },
    {
      "epoch": 1.6723934426229508,
      "grad_norm": 0.739182895427066,
      "learning_rate": 0.00013086129324024165,
      "loss": 12.448,
      "step": 9564
    },
    {
      "epoch": 1.672568306010929,
      "grad_norm": 0.7136415525745038,
      "learning_rate": 0.00013084782326973647,
      "loss": 11.9771,
      "step": 9565
    },
    {
      "epoch": 1.672743169398907,
      "grad_norm": 0.6994902766467403,
      "learning_rate": 0.00013083435268063768,
      "loss": 12.3903,
      "step": 9566
    },
    {
      "epoch": 1.6729180327868853,
      "grad_norm": 0.6795955994078553,
      "learning_rate": 0.00013082088147321546,
      "loss": 12.4043,
      "step": 9567
    },
    {
      "epoch": 1.6730928961748632,
      "grad_norm": 0.7958989460133202,
      "learning_rate": 0.00013080740964773995,
      "loss": 12.4634,
      "step": 9568
    },
    {
      "epoch": 1.6732677595628416,
      "grad_norm": 0.7862719632267644,
      "learning_rate": 0.00013079393720448126,
      "loss": 12.6484,
      "step": 9569
    },
    {
      "epoch": 1.6734426229508197,
      "grad_norm": 0.6688502471755565,
      "learning_rate": 0.0001307804641437096,
      "loss": 12.3045,
      "step": 9570
    },
    {
      "epoch": 1.6736174863387978,
      "grad_norm": 0.7429039629190531,
      "learning_rate": 0.0001307669904656951,
      "loss": 12.4133,
      "step": 9571
    },
    {
      "epoch": 1.673792349726776,
      "grad_norm": 0.7497632991679174,
      "learning_rate": 0.00013075351617070798,
      "loss": 12.5327,
      "step": 9572
    },
    {
      "epoch": 1.673967213114754,
      "grad_norm": 0.6420815035337171,
      "learning_rate": 0.00013074004125901846,
      "loss": 11.9058,
      "step": 9573
    },
    {
      "epoch": 1.6741420765027324,
      "grad_norm": 0.6639143681099181,
      "learning_rate": 0.00013072656573089673,
      "loss": 12.3916,
      "step": 9574
    },
    {
      "epoch": 1.6743169398907103,
      "grad_norm": 0.6903711458331229,
      "learning_rate": 0.00013071308958661302,
      "loss": 12.4028,
      "step": 9575
    },
    {
      "epoch": 1.6744918032786886,
      "grad_norm": 0.6468221103372719,
      "learning_rate": 0.0001306996128264375,
      "loss": 12.2768,
      "step": 9576
    },
    {
      "epoch": 1.6746666666666665,
      "grad_norm": 0.6719822892950454,
      "learning_rate": 0.00013068613545064053,
      "loss": 12.2617,
      "step": 9577
    },
    {
      "epoch": 1.6748415300546449,
      "grad_norm": 0.6677540468750796,
      "learning_rate": 0.0001306726574594923,
      "loss": 12.4714,
      "step": 9578
    },
    {
      "epoch": 1.675016393442623,
      "grad_norm": 0.6779199643595583,
      "learning_rate": 0.00013065917885326313,
      "loss": 12.3983,
      "step": 9579
    },
    {
      "epoch": 1.675191256830601,
      "grad_norm": 0.7354158172671106,
      "learning_rate": 0.0001306456996322233,
      "loss": 12.5034,
      "step": 9580
    },
    {
      "epoch": 1.6753661202185792,
      "grad_norm": 0.7016357348405999,
      "learning_rate": 0.0001306322197966431,
      "loss": 12.4337,
      "step": 9581
    },
    {
      "epoch": 1.6755409836065573,
      "grad_norm": 0.7580414786947125,
      "learning_rate": 0.00013061873934679282,
      "loss": 12.2614,
      "step": 9582
    },
    {
      "epoch": 1.6757158469945357,
      "grad_norm": 0.6806643746278671,
      "learning_rate": 0.00013060525828294278,
      "loss": 12.3195,
      "step": 9583
    },
    {
      "epoch": 1.6758907103825136,
      "grad_norm": 0.7997384967381035,
      "learning_rate": 0.0001305917766053634,
      "loss": 12.2977,
      "step": 9584
    },
    {
      "epoch": 1.676065573770492,
      "grad_norm": 0.6849461287109233,
      "learning_rate": 0.0001305782943143249,
      "loss": 12.2264,
      "step": 9585
    },
    {
      "epoch": 1.6762404371584698,
      "grad_norm": 0.7120581713379511,
      "learning_rate": 0.00013056481141009778,
      "loss": 12.3061,
      "step": 9586
    },
    {
      "epoch": 1.6764153005464482,
      "grad_norm": 0.7038123020777951,
      "learning_rate": 0.00013055132789295234,
      "loss": 12.485,
      "step": 9587
    },
    {
      "epoch": 1.6765901639344263,
      "grad_norm": 0.679462868383612,
      "learning_rate": 0.00013053784376315888,
      "loss": 11.9815,
      "step": 9588
    },
    {
      "epoch": 1.6767650273224044,
      "grad_norm": 0.7406027687622943,
      "learning_rate": 0.00013052435902098801,
      "loss": 12.2659,
      "step": 9589
    },
    {
      "epoch": 1.6769398907103825,
      "grad_norm": 1.0348794879682646,
      "learning_rate": 0.00013051087366670994,
      "loss": 12.6685,
      "step": 9590
    },
    {
      "epoch": 1.6771147540983606,
      "grad_norm": 0.7121780292661144,
      "learning_rate": 0.00013049738770059521,
      "loss": 12.2018,
      "step": 9591
    },
    {
      "epoch": 1.6772896174863388,
      "grad_norm": 0.7067759015227137,
      "learning_rate": 0.0001304839011229142,
      "loss": 12.0639,
      "step": 9592
    },
    {
      "epoch": 1.6774644808743169,
      "grad_norm": 0.6759218963981786,
      "learning_rate": 0.0001304704139339374,
      "loss": 12.4181,
      "step": 9593
    },
    {
      "epoch": 1.6776393442622952,
      "grad_norm": 0.6795564119060518,
      "learning_rate": 0.0001304569261339352,
      "loss": 12.1462,
      "step": 9594
    },
    {
      "epoch": 1.6778142076502731,
      "grad_norm": 0.6743410971123867,
      "learning_rate": 0.00013044343772317815,
      "loss": 12.3508,
      "step": 9595
    },
    {
      "epoch": 1.6779890710382515,
      "grad_norm": 0.7317976630271456,
      "learning_rate": 0.0001304299487019367,
      "loss": 12.4337,
      "step": 9596
    },
    {
      "epoch": 1.6781639344262294,
      "grad_norm": 0.6843174699589774,
      "learning_rate": 0.0001304164590704813,
      "loss": 12.4514,
      "step": 9597
    },
    {
      "epoch": 1.6783387978142077,
      "grad_norm": 0.6992804154234166,
      "learning_rate": 0.00013040296882908255,
      "loss": 12.3966,
      "step": 9598
    },
    {
      "epoch": 1.6785136612021858,
      "grad_norm": 0.7717515678867172,
      "learning_rate": 0.00013038947797801095,
      "loss": 12.3234,
      "step": 9599
    },
    {
      "epoch": 1.678688524590164,
      "grad_norm": 0.6931916226959781,
      "learning_rate": 0.00013037598651753694,
      "loss": 12.3182,
      "step": 9600
    },
    {
      "epoch": 1.678863387978142,
      "grad_norm": 0.6841953190776614,
      "learning_rate": 0.0001303624944479312,
      "loss": 12.3663,
      "step": 9601
    },
    {
      "epoch": 1.6790382513661202,
      "grad_norm": 0.6767445199761944,
      "learning_rate": 0.00013034900176946417,
      "loss": 12.1336,
      "step": 9602
    },
    {
      "epoch": 1.6792131147540985,
      "grad_norm": 0.8131556689561624,
      "learning_rate": 0.0001303355084824065,
      "loss": 12.6326,
      "step": 9603
    },
    {
      "epoch": 1.6793879781420764,
      "grad_norm": 0.7066396471912272,
      "learning_rate": 0.0001303220145870287,
      "loss": 12.1402,
      "step": 9604
    },
    {
      "epoch": 1.6795628415300548,
      "grad_norm": 0.708373639691509,
      "learning_rate": 0.00013030852008360147,
      "loss": 12.4668,
      "step": 9605
    },
    {
      "epoch": 1.6797377049180326,
      "grad_norm": 0.6378390075210015,
      "learning_rate": 0.00013029502497239533,
      "loss": 12.3259,
      "step": 9606
    },
    {
      "epoch": 1.679912568306011,
      "grad_norm": 0.6874964477582733,
      "learning_rate": 0.0001302815292536809,
      "loss": 12.4808,
      "step": 9607
    },
    {
      "epoch": 1.680087431693989,
      "grad_norm": 0.7061544037959104,
      "learning_rate": 0.00013026803292772886,
      "loss": 12.2534,
      "step": 9608
    },
    {
      "epoch": 1.6802622950819672,
      "grad_norm": 0.6603006746041746,
      "learning_rate": 0.0001302545359948098,
      "loss": 12.3276,
      "step": 9609
    },
    {
      "epoch": 1.6804371584699453,
      "grad_norm": 0.7361350753857168,
      "learning_rate": 0.0001302410384551944,
      "loss": 12.1909,
      "step": 9610
    },
    {
      "epoch": 1.6806120218579235,
      "grad_norm": 0.7542978490900123,
      "learning_rate": 0.00013022754030915333,
      "loss": 12.3098,
      "step": 9611
    },
    {
      "epoch": 1.6807868852459016,
      "grad_norm": 0.7308395684467078,
      "learning_rate": 0.00013021404155695725,
      "loss": 12.4111,
      "step": 9612
    },
    {
      "epoch": 1.6809617486338797,
      "grad_norm": 0.7266005892071183,
      "learning_rate": 0.0001302005421988769,
      "loss": 12.3609,
      "step": 9613
    },
    {
      "epoch": 1.681136612021858,
      "grad_norm": 0.655776852902239,
      "learning_rate": 0.00013018704223518297,
      "loss": 12.3461,
      "step": 9614
    },
    {
      "epoch": 1.681311475409836,
      "grad_norm": 0.7809197412501825,
      "learning_rate": 0.00013017354166614613,
      "loss": 12.1989,
      "step": 9615
    },
    {
      "epoch": 1.6814863387978143,
      "grad_norm": 0.7105543468558398,
      "learning_rate": 0.00013016004049203712,
      "loss": 12.3587,
      "step": 9616
    },
    {
      "epoch": 1.6816612021857924,
      "grad_norm": 0.7466782045882543,
      "learning_rate": 0.0001301465387131267,
      "loss": 12.4143,
      "step": 9617
    },
    {
      "epoch": 1.6818360655737705,
      "grad_norm": 0.6723916982195086,
      "learning_rate": 0.00013013303632968562,
      "loss": 12.3394,
      "step": 9618
    },
    {
      "epoch": 1.6820109289617486,
      "grad_norm": 0.780454086398956,
      "learning_rate": 0.00013011953334198466,
      "loss": 12.4748,
      "step": 9619
    },
    {
      "epoch": 1.6821857923497268,
      "grad_norm": 0.7922179416800453,
      "learning_rate": 0.00013010602975029454,
      "loss": 12.2307,
      "step": 9620
    },
    {
      "epoch": 1.6823606557377049,
      "grad_norm": 0.7542964373412517,
      "learning_rate": 0.00013009252555488612,
      "loss": 12.3145,
      "step": 9621
    },
    {
      "epoch": 1.682535519125683,
      "grad_norm": 0.6962003551707342,
      "learning_rate": 0.00013007902075603015,
      "loss": 12.3927,
      "step": 9622
    },
    {
      "epoch": 1.6827103825136613,
      "grad_norm": 0.6539413874248747,
      "learning_rate": 0.0001300655153539975,
      "loss": 12.2681,
      "step": 9623
    },
    {
      "epoch": 1.6828852459016392,
      "grad_norm": 0.8128789323509636,
      "learning_rate": 0.0001300520093490589,
      "loss": 12.1172,
      "step": 9624
    },
    {
      "epoch": 1.6830601092896176,
      "grad_norm": 0.6950474952766711,
      "learning_rate": 0.0001300385027414853,
      "loss": 12.3366,
      "step": 9625
    },
    {
      "epoch": 1.6832349726775955,
      "grad_norm": 0.6372817973759548,
      "learning_rate": 0.0001300249955315474,
      "loss": 11.8073,
      "step": 9626
    },
    {
      "epoch": 1.6834098360655738,
      "grad_norm": 0.7689604834600302,
      "learning_rate": 0.00013001148771951622,
      "loss": 12.2074,
      "step": 9627
    },
    {
      "epoch": 1.683584699453552,
      "grad_norm": 0.675745308229099,
      "learning_rate": 0.00012999797930566252,
      "loss": 12.2948,
      "step": 9628
    },
    {
      "epoch": 1.68375956284153,
      "grad_norm": 0.7016424424902753,
      "learning_rate": 0.00012998447029025727,
      "loss": 12.5215,
      "step": 9629
    },
    {
      "epoch": 1.6839344262295082,
      "grad_norm": 0.6231695215300261,
      "learning_rate": 0.00012997096067357132,
      "loss": 12.1413,
      "step": 9630
    },
    {
      "epoch": 1.6841092896174863,
      "grad_norm": 0.705933682485569,
      "learning_rate": 0.00012995745045587555,
      "loss": 12.2013,
      "step": 9631
    },
    {
      "epoch": 1.6842841530054646,
      "grad_norm": 0.7044903510922654,
      "learning_rate": 0.00012994393963744095,
      "loss": 12.3287,
      "step": 9632
    },
    {
      "epoch": 1.6844590163934425,
      "grad_norm": 0.7102294133542536,
      "learning_rate": 0.0001299304282185384,
      "loss": 12.2156,
      "step": 9633
    },
    {
      "epoch": 1.6846338797814209,
      "grad_norm": 0.6513918615729162,
      "learning_rate": 0.0001299169161994389,
      "loss": 12.0747,
      "step": 9634
    },
    {
      "epoch": 1.6848087431693988,
      "grad_norm": 0.6609729562681965,
      "learning_rate": 0.00012990340358041334,
      "loss": 12.4229,
      "step": 9635
    },
    {
      "epoch": 1.6849836065573771,
      "grad_norm": 0.711376890402883,
      "learning_rate": 0.00012988989036173272,
      "loss": 12.4634,
      "step": 9636
    },
    {
      "epoch": 1.6851584699453552,
      "grad_norm": 0.6418583224765265,
      "learning_rate": 0.00012987637654366806,
      "loss": 12.4093,
      "step": 9637
    },
    {
      "epoch": 1.6853333333333333,
      "grad_norm": 0.8120066516327855,
      "learning_rate": 0.00012986286212649027,
      "loss": 12.4892,
      "step": 9638
    },
    {
      "epoch": 1.6855081967213115,
      "grad_norm": 0.7072440159994682,
      "learning_rate": 0.00012984934711047046,
      "loss": 12.3531,
      "step": 9639
    },
    {
      "epoch": 1.6856830601092896,
      "grad_norm": 0.8299410494954322,
      "learning_rate": 0.00012983583149587952,
      "loss": 12.4645,
      "step": 9640
    },
    {
      "epoch": 1.6858579234972677,
      "grad_norm": 0.7753860364523499,
      "learning_rate": 0.00012982231528298863,
      "loss": 12.131,
      "step": 9641
    },
    {
      "epoch": 1.6860327868852458,
      "grad_norm": 0.7047412316119067,
      "learning_rate": 0.0001298087984720687,
      "loss": 12.3158,
      "step": 9642
    },
    {
      "epoch": 1.6862076502732242,
      "grad_norm": 0.893527978504548,
      "learning_rate": 0.00012979528106339086,
      "loss": 12.1676,
      "step": 9643
    },
    {
      "epoch": 1.686382513661202,
      "grad_norm": 0.6985399788950999,
      "learning_rate": 0.00012978176305722614,
      "loss": 12.2536,
      "step": 9644
    },
    {
      "epoch": 1.6865573770491804,
      "grad_norm": 0.8524859255759947,
      "learning_rate": 0.0001297682444538456,
      "loss": 12.4583,
      "step": 9645
    },
    {
      "epoch": 1.6867322404371585,
      "grad_norm": 0.6892066221335281,
      "learning_rate": 0.0001297547252535204,
      "loss": 12.2962,
      "step": 9646
    },
    {
      "epoch": 1.6869071038251366,
      "grad_norm": 0.7469749060013221,
      "learning_rate": 0.00012974120545652156,
      "loss": 12.5647,
      "step": 9647
    },
    {
      "epoch": 1.6870819672131148,
      "grad_norm": 0.8062171792139117,
      "learning_rate": 0.00012972768506312027,
      "loss": 12.6493,
      "step": 9648
    },
    {
      "epoch": 1.6872568306010929,
      "grad_norm": 0.7148079787028666,
      "learning_rate": 0.00012971416407358763,
      "loss": 12.3155,
      "step": 9649
    },
    {
      "epoch": 1.687431693989071,
      "grad_norm": 0.7550587816450188,
      "learning_rate": 0.00012970064248819473,
      "loss": 12.5904,
      "step": 9650
    },
    {
      "epoch": 1.6876065573770491,
      "grad_norm": 0.6931513170659446,
      "learning_rate": 0.00012968712030721278,
      "loss": 12.441,
      "step": 9651
    },
    {
      "epoch": 1.6877814207650275,
      "grad_norm": 0.673233367036654,
      "learning_rate": 0.0001296735975309129,
      "loss": 12.2951,
      "step": 9652
    },
    {
      "epoch": 1.6879562841530054,
      "grad_norm": 0.6609862562731265,
      "learning_rate": 0.00012966007415956627,
      "loss": 12.2778,
      "step": 9653
    },
    {
      "epoch": 1.6881311475409837,
      "grad_norm": 0.7351015198107493,
      "learning_rate": 0.0001296465501934441,
      "loss": 12.3888,
      "step": 9654
    },
    {
      "epoch": 1.6883060109289616,
      "grad_norm": 0.6506841969066666,
      "learning_rate": 0.00012963302563281758,
      "loss": 12.0643,
      "step": 9655
    },
    {
      "epoch": 1.68848087431694,
      "grad_norm": 0.6579845110872168,
      "learning_rate": 0.0001296195004779579,
      "loss": 12.3013,
      "step": 9656
    },
    {
      "epoch": 1.688655737704918,
      "grad_norm": 0.731286158184017,
      "learning_rate": 0.00012960597472913628,
      "loss": 12.2679,
      "step": 9657
    },
    {
      "epoch": 1.6888306010928962,
      "grad_norm": 0.7341706253715291,
      "learning_rate": 0.000129592448386624,
      "loss": 12.1888,
      "step": 9658
    },
    {
      "epoch": 1.6890054644808743,
      "grad_norm": 0.6551442976135684,
      "learning_rate": 0.0001295789214506922,
      "loss": 12.2004,
      "step": 9659
    },
    {
      "epoch": 1.6891803278688524,
      "grad_norm": 0.6782474579172704,
      "learning_rate": 0.00012956539392161229,
      "loss": 12.3468,
      "step": 9660
    },
    {
      "epoch": 1.6893551912568308,
      "grad_norm": 0.7134740244699111,
      "learning_rate": 0.0001295518657996554,
      "loss": 12.3186,
      "step": 9661
    },
    {
      "epoch": 1.6895300546448087,
      "grad_norm": 0.7680441444267505,
      "learning_rate": 0.0001295383370850929,
      "loss": 12.3201,
      "step": 9662
    },
    {
      "epoch": 1.689704918032787,
      "grad_norm": 0.6641544937137207,
      "learning_rate": 0.000129524807778196,
      "loss": 12.2891,
      "step": 9663
    },
    {
      "epoch": 1.689879781420765,
      "grad_norm": 0.6804716966531326,
      "learning_rate": 0.0001295112778792361,
      "loss": 12.367,
      "step": 9664
    },
    {
      "epoch": 1.6900546448087432,
      "grad_norm": 0.8729045270455666,
      "learning_rate": 0.00012949774738848444,
      "loss": 12.6038,
      "step": 9665
    },
    {
      "epoch": 1.6902295081967214,
      "grad_norm": 0.7074651839250807,
      "learning_rate": 0.00012948421630621238,
      "loss": 12.4004,
      "step": 9666
    },
    {
      "epoch": 1.6904043715846995,
      "grad_norm": 0.7545251440579374,
      "learning_rate": 0.00012947068463269124,
      "loss": 12.2605,
      "step": 9667
    },
    {
      "epoch": 1.6905792349726776,
      "grad_norm": 0.7249460064889343,
      "learning_rate": 0.0001294571523681924,
      "loss": 12.48,
      "step": 9668
    },
    {
      "epoch": 1.6907540983606557,
      "grad_norm": 0.7007187505154067,
      "learning_rate": 0.00012944361951298722,
      "loss": 11.9765,
      "step": 9669
    },
    {
      "epoch": 1.6909289617486338,
      "grad_norm": 0.6467547977233066,
      "learning_rate": 0.00012943008606734706,
      "loss": 12.243,
      "step": 9670
    },
    {
      "epoch": 1.691103825136612,
      "grad_norm": 0.7169539322420995,
      "learning_rate": 0.00012941655203154332,
      "loss": 12.3728,
      "step": 9671
    },
    {
      "epoch": 1.6912786885245903,
      "grad_norm": 0.7140447491553671,
      "learning_rate": 0.00012940301740584737,
      "loss": 12.3862,
      "step": 9672
    },
    {
      "epoch": 1.6914535519125682,
      "grad_norm": 0.6902002400805769,
      "learning_rate": 0.00012938948219053066,
      "loss": 12.3929,
      "step": 9673
    },
    {
      "epoch": 1.6916284153005465,
      "grad_norm": 0.6919734437205731,
      "learning_rate": 0.00012937594638586457,
      "loss": 12.5384,
      "step": 9674
    },
    {
      "epoch": 1.6918032786885244,
      "grad_norm": 0.6722513160044972,
      "learning_rate": 0.00012936240999212057,
      "loss": 12.4308,
      "step": 9675
    },
    {
      "epoch": 1.6919781420765028,
      "grad_norm": 0.6863040819581945,
      "learning_rate": 0.0001293488730095701,
      "loss": 12.3026,
      "step": 9676
    },
    {
      "epoch": 1.6921530054644809,
      "grad_norm": 0.6996572430772049,
      "learning_rate": 0.00012933533543848461,
      "loss": 12.2432,
      "step": 9677
    },
    {
      "epoch": 1.692327868852459,
      "grad_norm": 0.7421272182847359,
      "learning_rate": 0.0001293217972791356,
      "loss": 12.2718,
      "step": 9678
    },
    {
      "epoch": 1.6925027322404371,
      "grad_norm": 0.6452690896339754,
      "learning_rate": 0.00012930825853179446,
      "loss": 12.3384,
      "step": 9679
    },
    {
      "epoch": 1.6926775956284152,
      "grad_norm": 0.6930082730693553,
      "learning_rate": 0.0001292947191967328,
      "loss": 12.0954,
      "step": 9680
    },
    {
      "epoch": 1.6928524590163936,
      "grad_norm": 0.7399034776088649,
      "learning_rate": 0.00012928117927422207,
      "loss": 12.4782,
      "step": 9681
    },
    {
      "epoch": 1.6930273224043715,
      "grad_norm": 0.6421879544715942,
      "learning_rate": 0.0001292676387645338,
      "loss": 12.2192,
      "step": 9682
    },
    {
      "epoch": 1.6932021857923498,
      "grad_norm": 0.7971971879927757,
      "learning_rate": 0.0001292540976679395,
      "loss": 12.3525,
      "step": 9683
    },
    {
      "epoch": 1.6933770491803277,
      "grad_norm": 0.6715549846128969,
      "learning_rate": 0.00012924055598471067,
      "loss": 12.418,
      "step": 9684
    },
    {
      "epoch": 1.693551912568306,
      "grad_norm": 0.7521746793734106,
      "learning_rate": 0.00012922701371511898,
      "loss": 12.399,
      "step": 9685
    },
    {
      "epoch": 1.6937267759562842,
      "grad_norm": 0.6928557901817847,
      "learning_rate": 0.0001292134708594359,
      "loss": 12.0107,
      "step": 9686
    },
    {
      "epoch": 1.6939016393442623,
      "grad_norm": 0.6695151463218842,
      "learning_rate": 0.00012919992741793307,
      "loss": 12.4108,
      "step": 9687
    },
    {
      "epoch": 1.6940765027322404,
      "grad_norm": 0.7034960008579233,
      "learning_rate": 0.00012918638339088197,
      "loss": 12.0481,
      "step": 9688
    },
    {
      "epoch": 1.6942513661202185,
      "grad_norm": 0.6854798792318278,
      "learning_rate": 0.0001291728387785543,
      "loss": 12.3803,
      "step": 9689
    },
    {
      "epoch": 1.6944262295081969,
      "grad_norm": 0.6800527443188212,
      "learning_rate": 0.00012915929358122164,
      "loss": 12.2573,
      "step": 9690
    },
    {
      "epoch": 1.6946010928961748,
      "grad_norm": 0.7508219534635109,
      "learning_rate": 0.00012914574779915563,
      "loss": 12.2734,
      "step": 9691
    },
    {
      "epoch": 1.6947759562841531,
      "grad_norm": 0.7578103147844442,
      "learning_rate": 0.00012913220143262786,
      "loss": 12.274,
      "step": 9692
    },
    {
      "epoch": 1.694950819672131,
      "grad_norm": 0.7085713357240376,
      "learning_rate": 0.00012911865448190997,
      "loss": 12.2176,
      "step": 9693
    },
    {
      "epoch": 1.6951256830601094,
      "grad_norm": 0.7307966935809623,
      "learning_rate": 0.00012910510694727373,
      "loss": 12.1131,
      "step": 9694
    },
    {
      "epoch": 1.6953005464480875,
      "grad_norm": 0.7652485285936725,
      "learning_rate": 0.00012909155882899072,
      "loss": 12.4033,
      "step": 9695
    },
    {
      "epoch": 1.6954754098360656,
      "grad_norm": 0.6891094946601014,
      "learning_rate": 0.0001290780101273326,
      "loss": 12.4839,
      "step": 9696
    },
    {
      "epoch": 1.6956502732240437,
      "grad_norm": 0.8368124161063707,
      "learning_rate": 0.0001290644608425711,
      "loss": 12.4094,
      "step": 9697
    },
    {
      "epoch": 1.6958251366120218,
      "grad_norm": 0.6515068467699117,
      "learning_rate": 0.00012905091097497795,
      "loss": 12.4218,
      "step": 9698
    },
    {
      "epoch": 1.696,
      "grad_norm": 0.9034524935339752,
      "learning_rate": 0.00012903736052482482,
      "loss": 12.4792,
      "step": 9699
    },
    {
      "epoch": 1.696174863387978,
      "grad_norm": 0.6433157869316022,
      "learning_rate": 0.00012902380949238342,
      "loss": 12.3991,
      "step": 9700
    },
    {
      "epoch": 1.6963497267759564,
      "grad_norm": 0.9480926640475871,
      "learning_rate": 0.00012901025787792557,
      "loss": 12.4257,
      "step": 9701
    },
    {
      "epoch": 1.6965245901639343,
      "grad_norm": 0.6759988020463764,
      "learning_rate": 0.00012899670568172295,
      "loss": 12.4493,
      "step": 9702
    },
    {
      "epoch": 1.6966994535519127,
      "grad_norm": 0.7676423043558478,
      "learning_rate": 0.00012898315290404736,
      "loss": 12.0871,
      "step": 9703
    },
    {
      "epoch": 1.6968743169398905,
      "grad_norm": 0.7537428747437206,
      "learning_rate": 0.00012896959954517055,
      "loss": 12.4595,
      "step": 9704
    },
    {
      "epoch": 1.697049180327869,
      "grad_norm": 0.6957202209893154,
      "learning_rate": 0.00012895604560536435,
      "loss": 12.2299,
      "step": 9705
    },
    {
      "epoch": 1.697224043715847,
      "grad_norm": 0.7377574773677574,
      "learning_rate": 0.0001289424910849005,
      "loss": 12.1966,
      "step": 9706
    },
    {
      "epoch": 1.6973989071038251,
      "grad_norm": 0.7339374163246207,
      "learning_rate": 0.00012892893598405087,
      "loss": 12.5001,
      "step": 9707
    },
    {
      "epoch": 1.6975737704918032,
      "grad_norm": 0.7514339683220083,
      "learning_rate": 0.00012891538030308723,
      "loss": 12.3117,
      "step": 9708
    },
    {
      "epoch": 1.6977486338797814,
      "grad_norm": 0.6500515790317647,
      "learning_rate": 0.00012890182404228144,
      "loss": 12.2413,
      "step": 9709
    },
    {
      "epoch": 1.6979234972677597,
      "grad_norm": 0.8214723238319593,
      "learning_rate": 0.00012888826720190534,
      "loss": 12.3838,
      "step": 9710
    },
    {
      "epoch": 1.6980983606557376,
      "grad_norm": 0.6884165056425076,
      "learning_rate": 0.00012887470978223078,
      "loss": 12.0881,
      "step": 9711
    },
    {
      "epoch": 1.698273224043716,
      "grad_norm": 0.7104603324476756,
      "learning_rate": 0.00012886115178352962,
      "loss": 12.2584,
      "step": 9712
    },
    {
      "epoch": 1.6984480874316938,
      "grad_norm": 0.6456139890844104,
      "learning_rate": 0.00012884759320607375,
      "loss": 12.4062,
      "step": 9713
    },
    {
      "epoch": 1.6986229508196722,
      "grad_norm": 0.6962774860996411,
      "learning_rate": 0.0001288340340501351,
      "loss": 12.5,
      "step": 9714
    },
    {
      "epoch": 1.6987978142076503,
      "grad_norm": 0.7290903003166549,
      "learning_rate": 0.0001288204743159855,
      "loss": 12.2823,
      "step": 9715
    },
    {
      "epoch": 1.6989726775956284,
      "grad_norm": 0.7106126642878542,
      "learning_rate": 0.0001288069140038969,
      "loss": 12.3355,
      "step": 9716
    },
    {
      "epoch": 1.6991475409836065,
      "grad_norm": 0.7271358428962509,
      "learning_rate": 0.00012879335311414124,
      "loss": 12.3676,
      "step": 9717
    },
    {
      "epoch": 1.6993224043715847,
      "grad_norm": 0.6567430161881557,
      "learning_rate": 0.00012877979164699043,
      "loss": 12.3028,
      "step": 9718
    },
    {
      "epoch": 1.699497267759563,
      "grad_norm": 0.743100947749993,
      "learning_rate": 0.00012876622960271644,
      "loss": 12.3797,
      "step": 9719
    },
    {
      "epoch": 1.699672131147541,
      "grad_norm": 0.688717154806989,
      "learning_rate": 0.0001287526669815912,
      "loss": 12.3242,
      "step": 9720
    },
    {
      "epoch": 1.6998469945355192,
      "grad_norm": 0.8229500465720174,
      "learning_rate": 0.00012873910378388676,
      "loss": 12.5196,
      "step": 9721
    },
    {
      "epoch": 1.7000218579234971,
      "grad_norm": 0.6936584772649681,
      "learning_rate": 0.00012872554000987498,
      "loss": 12.4136,
      "step": 9722
    },
    {
      "epoch": 1.7001967213114755,
      "grad_norm": 0.6922552810899081,
      "learning_rate": 0.000128711975659828,
      "loss": 12.3524,
      "step": 9723
    },
    {
      "epoch": 1.7003715846994536,
      "grad_norm": 0.8311198543982151,
      "learning_rate": 0.00012869841073401766,
      "loss": 12.3294,
      "step": 9724
    },
    {
      "epoch": 1.7005464480874317,
      "grad_norm": 0.6635042414551323,
      "learning_rate": 0.00012868484523271612,
      "loss": 12.3003,
      "step": 9725
    },
    {
      "epoch": 1.7007213114754098,
      "grad_norm": 0.8978936482852382,
      "learning_rate": 0.00012867127915619538,
      "loss": 12.5803,
      "step": 9726
    },
    {
      "epoch": 1.700896174863388,
      "grad_norm": 0.706033350859424,
      "learning_rate": 0.00012865771250472743,
      "loss": 12.5808,
      "step": 9727
    },
    {
      "epoch": 1.701071038251366,
      "grad_norm": 0.8949453053907257,
      "learning_rate": 0.00012864414527858438,
      "loss": 12.2194,
      "step": 9728
    },
    {
      "epoch": 1.7012459016393442,
      "grad_norm": 0.66227697077938,
      "learning_rate": 0.00012863057747803823,
      "loss": 12.2534,
      "step": 9729
    },
    {
      "epoch": 1.7014207650273225,
      "grad_norm": 0.8022630476334414,
      "learning_rate": 0.0001286170091033611,
      "loss": 12.4549,
      "step": 9730
    },
    {
      "epoch": 1.7015956284153004,
      "grad_norm": 0.7399953133494596,
      "learning_rate": 0.0001286034401548251,
      "loss": 12.3628,
      "step": 9731
    },
    {
      "epoch": 1.7017704918032788,
      "grad_norm": 0.7508597463785376,
      "learning_rate": 0.0001285898706327023,
      "loss": 12.4884,
      "step": 9732
    },
    {
      "epoch": 1.7019453551912567,
      "grad_norm": 0.6960149186391991,
      "learning_rate": 0.0001285763005372648,
      "loss": 12.1566,
      "step": 9733
    },
    {
      "epoch": 1.702120218579235,
      "grad_norm": 0.6520566450296618,
      "learning_rate": 0.0001285627298687847,
      "loss": 12.0716,
      "step": 9734
    },
    {
      "epoch": 1.7022950819672131,
      "grad_norm": 0.8235905573095029,
      "learning_rate": 0.00012854915862753422,
      "loss": 12.3264,
      "step": 9735
    },
    {
      "epoch": 1.7024699453551912,
      "grad_norm": 0.6916848403161673,
      "learning_rate": 0.0001285355868137854,
      "loss": 12.3716,
      "step": 9736
    },
    {
      "epoch": 1.7026448087431694,
      "grad_norm": 0.7332604964077671,
      "learning_rate": 0.0001285220144278105,
      "loss": 12.3189,
      "step": 9737
    },
    {
      "epoch": 1.7028196721311475,
      "grad_norm": 0.7099135500533605,
      "learning_rate": 0.00012850844146988162,
      "loss": 12.0958,
      "step": 9738
    },
    {
      "epoch": 1.7029945355191258,
      "grad_norm": 0.7050876564585085,
      "learning_rate": 0.00012849486794027094,
      "loss": 12.2452,
      "step": 9739
    },
    {
      "epoch": 1.7031693989071037,
      "grad_norm": 0.7377309154038781,
      "learning_rate": 0.0001284812938392507,
      "loss": 12.3013,
      "step": 9740
    },
    {
      "epoch": 1.703344262295082,
      "grad_norm": 0.7173025420044076,
      "learning_rate": 0.00012846771916709304,
      "loss": 12.2445,
      "step": 9741
    },
    {
      "epoch": 1.70351912568306,
      "grad_norm": 0.7479231353956298,
      "learning_rate": 0.0001284541439240702,
      "loss": 12.43,
      "step": 9742
    },
    {
      "epoch": 1.7036939890710383,
      "grad_norm": 0.6709364054918602,
      "learning_rate": 0.00012844056811045441,
      "loss": 12.2186,
      "step": 9743
    },
    {
      "epoch": 1.7038688524590164,
      "grad_norm": 0.7686419356122863,
      "learning_rate": 0.00012842699172651794,
      "loss": 12.0286,
      "step": 9744
    },
    {
      "epoch": 1.7040437158469945,
      "grad_norm": 0.7227944659098462,
      "learning_rate": 0.00012841341477253296,
      "loss": 12.4574,
      "step": 9745
    },
    {
      "epoch": 1.7042185792349727,
      "grad_norm": 0.75093038005454,
      "learning_rate": 0.0001283998372487718,
      "loss": 12.0884,
      "step": 9746
    },
    {
      "epoch": 1.7043934426229508,
      "grad_norm": 0.627902111514819,
      "learning_rate": 0.00012838625915550668,
      "loss": 12.2563,
      "step": 9747
    },
    {
      "epoch": 1.7045683060109291,
      "grad_norm": 0.7982048964471739,
      "learning_rate": 0.00012837268049300992,
      "loss": 12.0437,
      "step": 9748
    },
    {
      "epoch": 1.704743169398907,
      "grad_norm": 0.6584539467184519,
      "learning_rate": 0.0001283591012615538,
      "loss": 12.2171,
      "step": 9749
    },
    {
      "epoch": 1.7049180327868854,
      "grad_norm": 0.7806573955621897,
      "learning_rate": 0.00012834552146141065,
      "loss": 12.5702,
      "step": 9750
    },
    {
      "epoch": 1.7050928961748633,
      "grad_norm": 0.7168839549648078,
      "learning_rate": 0.00012833194109285274,
      "loss": 12.4128,
      "step": 9751
    },
    {
      "epoch": 1.7052677595628416,
      "grad_norm": 0.8068581541516233,
      "learning_rate": 0.00012831836015615242,
      "loss": 12.3424,
      "step": 9752
    },
    {
      "epoch": 1.7054426229508197,
      "grad_norm": 0.6819739924242668,
      "learning_rate": 0.00012830477865158202,
      "loss": 11.9873,
      "step": 9753
    },
    {
      "epoch": 1.7056174863387978,
      "grad_norm": 0.6956572419464316,
      "learning_rate": 0.00012829119657941393,
      "loss": 12.1315,
      "step": 9754
    },
    {
      "epoch": 1.705792349726776,
      "grad_norm": 0.7257241389999773,
      "learning_rate": 0.00012827761393992046,
      "loss": 12.4509,
      "step": 9755
    },
    {
      "epoch": 1.705967213114754,
      "grad_norm": 0.8992957674858177,
      "learning_rate": 0.00012826403073337402,
      "loss": 12.3315,
      "step": 9756
    },
    {
      "epoch": 1.7061420765027322,
      "grad_norm": 0.6668558502926066,
      "learning_rate": 0.00012825044696004696,
      "loss": 12.3329,
      "step": 9757
    },
    {
      "epoch": 1.7063169398907103,
      "grad_norm": 0.6749502890361808,
      "learning_rate": 0.00012823686262021172,
      "loss": 12.3556,
      "step": 9758
    },
    {
      "epoch": 1.7064918032786887,
      "grad_norm": 0.771553123853956,
      "learning_rate": 0.00012822327771414067,
      "loss": 12.4108,
      "step": 9759
    },
    {
      "epoch": 1.7066666666666666,
      "grad_norm": 0.639431872580387,
      "learning_rate": 0.00012820969224210625,
      "loss": 12.3879,
      "step": 9760
    },
    {
      "epoch": 1.706841530054645,
      "grad_norm": 0.7796421445248174,
      "learning_rate": 0.0001281961062043809,
      "loss": 12.0486,
      "step": 9761
    },
    {
      "epoch": 1.7070163934426228,
      "grad_norm": 0.7258376266120321,
      "learning_rate": 0.00012818251960123705,
      "loss": 12.4105,
      "step": 9762
    },
    {
      "epoch": 1.7071912568306011,
      "grad_norm": 0.6881783361665802,
      "learning_rate": 0.0001281689324329471,
      "loss": 12.2442,
      "step": 9763
    },
    {
      "epoch": 1.7073661202185793,
      "grad_norm": 0.9132120741557521,
      "learning_rate": 0.00012815534469978363,
      "loss": 12.3567,
      "step": 9764
    },
    {
      "epoch": 1.7075409836065574,
      "grad_norm": 0.7667439492141128,
      "learning_rate": 0.000128141756402019,
      "loss": 12.4597,
      "step": 9765
    },
    {
      "epoch": 1.7077158469945355,
      "grad_norm": 0.9063595639333836,
      "learning_rate": 0.00012812816753992578,
      "loss": 12.3774,
      "step": 9766
    },
    {
      "epoch": 1.7078907103825136,
      "grad_norm": 0.6541373656369537,
      "learning_rate": 0.00012811457811377644,
      "loss": 12.3888,
      "step": 9767
    },
    {
      "epoch": 1.708065573770492,
      "grad_norm": 0.9451259423845058,
      "learning_rate": 0.00012810098812384346,
      "loss": 12.2701,
      "step": 9768
    },
    {
      "epoch": 1.7082404371584698,
      "grad_norm": 0.7183113232561206,
      "learning_rate": 0.00012808739757039944,
      "loss": 12.3507,
      "step": 9769
    },
    {
      "epoch": 1.7084153005464482,
      "grad_norm": 0.8466009070107794,
      "learning_rate": 0.00012807380645371677,
      "loss": 12.4139,
      "step": 9770
    },
    {
      "epoch": 1.708590163934426,
      "grad_norm": 0.717914443822087,
      "learning_rate": 0.00012806021477406815,
      "loss": 12.4832,
      "step": 9771
    },
    {
      "epoch": 1.7087650273224044,
      "grad_norm": 0.7856705920530216,
      "learning_rate": 0.00012804662253172603,
      "loss": 12.4777,
      "step": 9772
    },
    {
      "epoch": 1.7089398907103825,
      "grad_norm": 0.7446651680559763,
      "learning_rate": 0.00012803302972696302,
      "loss": 12.3402,
      "step": 9773
    },
    {
      "epoch": 1.7091147540983607,
      "grad_norm": 0.6580507480669484,
      "learning_rate": 0.0001280194363600517,
      "loss": 12.3017,
      "step": 9774
    },
    {
      "epoch": 1.7092896174863388,
      "grad_norm": 0.8432214597885467,
      "learning_rate": 0.00012800584243126466,
      "loss": 12.5213,
      "step": 9775
    },
    {
      "epoch": 1.709464480874317,
      "grad_norm": 0.6203636085043089,
      "learning_rate": 0.0001279922479408745,
      "loss": 12.418,
      "step": 9776
    },
    {
      "epoch": 1.7096393442622952,
      "grad_norm": 0.705750276262796,
      "learning_rate": 0.0001279786528891538,
      "loss": 12.4087,
      "step": 9777
    },
    {
      "epoch": 1.7098142076502731,
      "grad_norm": 0.7737644786101289,
      "learning_rate": 0.00012796505727637522,
      "loss": 12.2738,
      "step": 9778
    },
    {
      "epoch": 1.7099890710382515,
      "grad_norm": 0.7394211971035516,
      "learning_rate": 0.00012795146110281135,
      "loss": 12.3676,
      "step": 9779
    },
    {
      "epoch": 1.7101639344262294,
      "grad_norm": 0.774067214406585,
      "learning_rate": 0.00012793786436873488,
      "loss": 12.0163,
      "step": 9780
    },
    {
      "epoch": 1.7103387978142077,
      "grad_norm": 0.675670571094516,
      "learning_rate": 0.00012792426707441845,
      "loss": 12.3589,
      "step": 9781
    },
    {
      "epoch": 1.7105136612021858,
      "grad_norm": 0.7833712141701508,
      "learning_rate": 0.0001279106692201347,
      "loss": 12.1193,
      "step": 9782
    },
    {
      "epoch": 1.710688524590164,
      "grad_norm": 0.7083634828269347,
      "learning_rate": 0.00012789707080615637,
      "loss": 11.8651,
      "step": 9783
    },
    {
      "epoch": 1.710863387978142,
      "grad_norm": 0.726015924656129,
      "learning_rate": 0.00012788347183275608,
      "loss": 11.9625,
      "step": 9784
    },
    {
      "epoch": 1.7110382513661202,
      "grad_norm": 0.7116940043979446,
      "learning_rate": 0.00012786987230020661,
      "loss": 12.0135,
      "step": 9785
    },
    {
      "epoch": 1.7112131147540983,
      "grad_norm": 0.6760342915688807,
      "learning_rate": 0.0001278562722087806,
      "loss": 12.4527,
      "step": 9786
    },
    {
      "epoch": 1.7113879781420764,
      "grad_norm": 0.769472008675468,
      "learning_rate": 0.00012784267155875083,
      "loss": 12.1916,
      "step": 9787
    },
    {
      "epoch": 1.7115628415300548,
      "grad_norm": 0.68213224411533,
      "learning_rate": 0.00012782907035038997,
      "loss": 12.0718,
      "step": 9788
    },
    {
      "epoch": 1.7117377049180327,
      "grad_norm": 0.7467879511907584,
      "learning_rate": 0.00012781546858397084,
      "loss": 12.2716,
      "step": 9789
    },
    {
      "epoch": 1.711912568306011,
      "grad_norm": 0.709416326865076,
      "learning_rate": 0.00012780186625976613,
      "loss": 12.1829,
      "step": 9790
    },
    {
      "epoch": 1.712087431693989,
      "grad_norm": 0.7330015403450401,
      "learning_rate": 0.00012778826337804866,
      "loss": 12.4242,
      "step": 9791
    },
    {
      "epoch": 1.7122622950819673,
      "grad_norm": 0.7829168957015548,
      "learning_rate": 0.00012777465993909118,
      "loss": 12.3947,
      "step": 9792
    },
    {
      "epoch": 1.7124371584699454,
      "grad_norm": 0.6472892154446629,
      "learning_rate": 0.00012776105594316647,
      "loss": 12.467,
      "step": 9793
    },
    {
      "epoch": 1.7126120218579235,
      "grad_norm": 0.8269764115496732,
      "learning_rate": 0.00012774745139054739,
      "loss": 12.2751,
      "step": 9794
    },
    {
      "epoch": 1.7127868852459016,
      "grad_norm": 0.7337867256032234,
      "learning_rate": 0.00012773384628150667,
      "loss": 12.0046,
      "step": 9795
    },
    {
      "epoch": 1.7129617486338797,
      "grad_norm": 0.8466892563583539,
      "learning_rate": 0.0001277202406163172,
      "loss": 12.273,
      "step": 9796
    },
    {
      "epoch": 1.713136612021858,
      "grad_norm": 0.7405853675237734,
      "learning_rate": 0.0001277066343952518,
      "loss": 12.1915,
      "step": 9797
    },
    {
      "epoch": 1.713311475409836,
      "grad_norm": 0.8152092801408383,
      "learning_rate": 0.0001276930276185833,
      "loss": 12.3853,
      "step": 9798
    },
    {
      "epoch": 1.7134863387978143,
      "grad_norm": 0.656893715898754,
      "learning_rate": 0.00012767942028658457,
      "loss": 12.0115,
      "step": 9799
    },
    {
      "epoch": 1.7136612021857922,
      "grad_norm": 0.7665813146408412,
      "learning_rate": 0.00012766581239952846,
      "loss": 12.4842,
      "step": 9800
    },
    {
      "epoch": 1.7138360655737706,
      "grad_norm": 0.6982168364488596,
      "learning_rate": 0.0001276522039576879,
      "loss": 12.3285,
      "step": 9801
    },
    {
      "epoch": 1.7140109289617487,
      "grad_norm": 0.715517725967766,
      "learning_rate": 0.00012763859496133572,
      "loss": 12.0165,
      "step": 9802
    },
    {
      "epoch": 1.7141857923497268,
      "grad_norm": 0.7053424023268603,
      "learning_rate": 0.00012762498541074485,
      "loss": 12.5053,
      "step": 9803
    },
    {
      "epoch": 1.714360655737705,
      "grad_norm": 0.7211823170864317,
      "learning_rate": 0.0001276113753061882,
      "loss": 12.3096,
      "step": 9804
    },
    {
      "epoch": 1.714535519125683,
      "grad_norm": 0.7045018127463064,
      "learning_rate": 0.0001275977646479387,
      "loss": 12.2389,
      "step": 9805
    },
    {
      "epoch": 1.7147103825136614,
      "grad_norm": 0.75636333535201,
      "learning_rate": 0.0001275841534362693,
      "loss": 12.3853,
      "step": 9806
    },
    {
      "epoch": 1.7148852459016393,
      "grad_norm": 0.699626181907553,
      "learning_rate": 0.00012757054167145292,
      "loss": 12.1992,
      "step": 9807
    },
    {
      "epoch": 1.7150601092896176,
      "grad_norm": 0.7089666659479225,
      "learning_rate": 0.00012755692935376254,
      "loss": 12.4421,
      "step": 9808
    },
    {
      "epoch": 1.7152349726775955,
      "grad_norm": 0.6703993184516688,
      "learning_rate": 0.00012754331648347108,
      "loss": 12.1655,
      "step": 9809
    },
    {
      "epoch": 1.7154098360655738,
      "grad_norm": 0.6650197083009006,
      "learning_rate": 0.00012752970306085162,
      "loss": 12.1575,
      "step": 9810
    },
    {
      "epoch": 1.715584699453552,
      "grad_norm": 0.7689688855195275,
      "learning_rate": 0.00012751608908617704,
      "loss": 12.2305,
      "step": 9811
    },
    {
      "epoch": 1.71575956284153,
      "grad_norm": 0.6292557779069824,
      "learning_rate": 0.00012750247455972036,
      "loss": 12.1915,
      "step": 9812
    },
    {
      "epoch": 1.7159344262295082,
      "grad_norm": 0.8378242912241837,
      "learning_rate": 0.00012748885948175466,
      "loss": 12.4517,
      "step": 9813
    },
    {
      "epoch": 1.7161092896174863,
      "grad_norm": 0.7110332244323302,
      "learning_rate": 0.00012747524385255292,
      "loss": 12.245,
      "step": 9814
    },
    {
      "epoch": 1.7162841530054644,
      "grad_norm": 0.8521097921464448,
      "learning_rate": 0.0001274616276723882,
      "loss": 12.384,
      "step": 9815
    },
    {
      "epoch": 1.7164590163934426,
      "grad_norm": 0.8255054839102092,
      "learning_rate": 0.0001274480109415335,
      "loss": 12.4593,
      "step": 9816
    },
    {
      "epoch": 1.716633879781421,
      "grad_norm": 0.7769756601487167,
      "learning_rate": 0.0001274343936602619,
      "loss": 12.3015,
      "step": 9817
    },
    {
      "epoch": 1.7168087431693988,
      "grad_norm": 0.9629147030264248,
      "learning_rate": 0.00012742077582884648,
      "loss": 12.5196,
      "step": 9818
    },
    {
      "epoch": 1.7169836065573771,
      "grad_norm": 0.7096116816349195,
      "learning_rate": 0.00012740715744756034,
      "loss": 12.0124,
      "step": 9819
    },
    {
      "epoch": 1.717158469945355,
      "grad_norm": 0.8550782530181887,
      "learning_rate": 0.00012739353851667651,
      "loss": 12.4173,
      "step": 9820
    },
    {
      "epoch": 1.7173333333333334,
      "grad_norm": 0.7302436325582148,
      "learning_rate": 0.0001273799190364681,
      "loss": 12.4219,
      "step": 9821
    },
    {
      "epoch": 1.7175081967213115,
      "grad_norm": 1.016982807728012,
      "learning_rate": 0.0001273662990072083,
      "loss": 12.462,
      "step": 9822
    },
    {
      "epoch": 1.7176830601092896,
      "grad_norm": 0.6815522513505766,
      "learning_rate": 0.00012735267842917014,
      "loss": 12.3718,
      "step": 9823
    },
    {
      "epoch": 1.7178579234972677,
      "grad_norm": 0.7944501401650209,
      "learning_rate": 0.00012733905730262683,
      "loss": 12.172,
      "step": 9824
    },
    {
      "epoch": 1.7180327868852459,
      "grad_norm": 0.7017917248674764,
      "learning_rate": 0.00012732543562785144,
      "loss": 12.2822,
      "step": 9825
    },
    {
      "epoch": 1.7182076502732242,
      "grad_norm": 0.7270785682776608,
      "learning_rate": 0.0001273118134051172,
      "loss": 12.4802,
      "step": 9826
    },
    {
      "epoch": 1.718382513661202,
      "grad_norm": 0.8337850108724352,
      "learning_rate": 0.00012729819063469724,
      "loss": 12.4669,
      "step": 9827
    },
    {
      "epoch": 1.7185573770491804,
      "grad_norm": 0.6338555808490939,
      "learning_rate": 0.0001272845673168647,
      "loss": 12.2804,
      "step": 9828
    },
    {
      "epoch": 1.7187322404371583,
      "grad_norm": 0.7150324498549778,
      "learning_rate": 0.00012727094345189286,
      "loss": 12.452,
      "step": 9829
    },
    {
      "epoch": 1.7189071038251367,
      "grad_norm": 0.7225012425954362,
      "learning_rate": 0.00012725731904005482,
      "loss": 12.3483,
      "step": 9830
    },
    {
      "epoch": 1.7190819672131148,
      "grad_norm": 0.7933414631224424,
      "learning_rate": 0.0001272436940816239,
      "loss": 12.4377,
      "step": 9831
    },
    {
      "epoch": 1.719256830601093,
      "grad_norm": 0.857793611038905,
      "learning_rate": 0.0001272300685768732,
      "loss": 12.2397,
      "step": 9832
    },
    {
      "epoch": 1.719431693989071,
      "grad_norm": 0.694230578239462,
      "learning_rate": 0.0001272164425260761,
      "loss": 12.1836,
      "step": 9833
    },
    {
      "epoch": 1.7196065573770491,
      "grad_norm": 0.818930919949025,
      "learning_rate": 0.00012720281592950572,
      "loss": 12.387,
      "step": 9834
    },
    {
      "epoch": 1.7197814207650273,
      "grad_norm": 0.75703920154244,
      "learning_rate": 0.00012718918878743533,
      "loss": 12.2703,
      "step": 9835
    },
    {
      "epoch": 1.7199562841530054,
      "grad_norm": 0.7531377329080469,
      "learning_rate": 0.00012717556110013826,
      "loss": 12.4649,
      "step": 9836
    },
    {
      "epoch": 1.7201311475409837,
      "grad_norm": 0.6726706496980154,
      "learning_rate": 0.00012716193286788774,
      "loss": 12.4348,
      "step": 9837
    },
    {
      "epoch": 1.7203060109289616,
      "grad_norm": 0.6848678330316315,
      "learning_rate": 0.00012714830409095706,
      "loss": 12.4398,
      "step": 9838
    },
    {
      "epoch": 1.72048087431694,
      "grad_norm": 0.7317470605275778,
      "learning_rate": 0.00012713467476961954,
      "loss": 12.1262,
      "step": 9839
    },
    {
      "epoch": 1.720655737704918,
      "grad_norm": 0.6631457125651167,
      "learning_rate": 0.00012712104490414844,
      "loss": 12.2635,
      "step": 9840
    },
    {
      "epoch": 1.7208306010928962,
      "grad_norm": 0.6803108417610331,
      "learning_rate": 0.00012710741449481716,
      "loss": 12.2391,
      "step": 9841
    },
    {
      "epoch": 1.7210054644808743,
      "grad_norm": 0.6428872692859815,
      "learning_rate": 0.00012709378354189895,
      "loss": 12.2328,
      "step": 9842
    },
    {
      "epoch": 1.7211803278688524,
      "grad_norm": 0.628836870441505,
      "learning_rate": 0.00012708015204566724,
      "loss": 12.2529,
      "step": 9843
    },
    {
      "epoch": 1.7213551912568306,
      "grad_norm": 0.6811208394142351,
      "learning_rate": 0.00012706652000639532,
      "loss": 12.3619,
      "step": 9844
    },
    {
      "epoch": 1.7215300546448087,
      "grad_norm": 0.7158559644209583,
      "learning_rate": 0.00012705288742435655,
      "loss": 12.1586,
      "step": 9845
    },
    {
      "epoch": 1.721704918032787,
      "grad_norm": 0.7345453393880647,
      "learning_rate": 0.00012703925429982432,
      "loss": 12.2354,
      "step": 9846
    },
    {
      "epoch": 1.721879781420765,
      "grad_norm": 0.6966606051351887,
      "learning_rate": 0.00012702562063307205,
      "loss": 12.5144,
      "step": 9847
    },
    {
      "epoch": 1.7220546448087433,
      "grad_norm": 0.6925054657215437,
      "learning_rate": 0.00012701198642437307,
      "loss": 12.0717,
      "step": 9848
    },
    {
      "epoch": 1.7222295081967212,
      "grad_norm": 0.7012171304999264,
      "learning_rate": 0.00012699835167400084,
      "loss": 12.1953,
      "step": 9849
    },
    {
      "epoch": 1.7224043715846995,
      "grad_norm": 0.6739343758719214,
      "learning_rate": 0.00012698471638222878,
      "loss": 12.2854,
      "step": 9850
    },
    {
      "epoch": 1.7225792349726776,
      "grad_norm": 0.6615902882459159,
      "learning_rate": 0.00012697108054933025,
      "loss": 12.3037,
      "step": 9851
    },
    {
      "epoch": 1.7227540983606557,
      "grad_norm": 0.7075953030818515,
      "learning_rate": 0.0001269574441755788,
      "loss": 12.2681,
      "step": 9852
    },
    {
      "epoch": 1.7229289617486339,
      "grad_norm": 0.6758430266935083,
      "learning_rate": 0.0001269438072612478,
      "loss": 12.1432,
      "step": 9853
    },
    {
      "epoch": 1.723103825136612,
      "grad_norm": 0.7800816773481275,
      "learning_rate": 0.00012693016980661075,
      "loss": 12.1929,
      "step": 9854
    },
    {
      "epoch": 1.7232786885245903,
      "grad_norm": 0.7064826101423742,
      "learning_rate": 0.00012691653181194109,
      "loss": 12.2385,
      "step": 9855
    },
    {
      "epoch": 1.7234535519125682,
      "grad_norm": 0.6955704134723748,
      "learning_rate": 0.00012690289327751234,
      "loss": 12.4295,
      "step": 9856
    },
    {
      "epoch": 1.7236284153005466,
      "grad_norm": 0.6282627610679835,
      "learning_rate": 0.000126889254203598,
      "loss": 12.2324,
      "step": 9857
    },
    {
      "epoch": 1.7238032786885245,
      "grad_norm": 0.7042128667139356,
      "learning_rate": 0.0001268756145904715,
      "loss": 12.3091,
      "step": 9858
    },
    {
      "epoch": 1.7239781420765028,
      "grad_norm": 0.6999114768804396,
      "learning_rate": 0.00012686197443840643,
      "loss": 12.2641,
      "step": 9859
    },
    {
      "epoch": 1.724153005464481,
      "grad_norm": 0.6495811919213721,
      "learning_rate": 0.0001268483337476763,
      "loss": 12.3053,
      "step": 9860
    },
    {
      "epoch": 1.724327868852459,
      "grad_norm": 0.6673608248483781,
      "learning_rate": 0.00012683469251855464,
      "loss": 12.0348,
      "step": 9861
    },
    {
      "epoch": 1.7245027322404372,
      "grad_norm": 0.6771589355384025,
      "learning_rate": 0.000126821050751315,
      "loss": 12.3087,
      "step": 9862
    },
    {
      "epoch": 1.7246775956284153,
      "grad_norm": 0.7414989015298556,
      "learning_rate": 0.00012680740844623096,
      "loss": 12.5424,
      "step": 9863
    },
    {
      "epoch": 1.7248524590163934,
      "grad_norm": 0.6187062960067333,
      "learning_rate": 0.00012679376560357602,
      "loss": 12.338,
      "step": 9864
    },
    {
      "epoch": 1.7250273224043715,
      "grad_norm": 0.6744809122252262,
      "learning_rate": 0.00012678012222362388,
      "loss": 12.3924,
      "step": 9865
    },
    {
      "epoch": 1.7252021857923499,
      "grad_norm": 0.648597648867973,
      "learning_rate": 0.000126766478306648,
      "loss": 12.4079,
      "step": 9866
    },
    {
      "epoch": 1.7253770491803277,
      "grad_norm": 0.7484730518778892,
      "learning_rate": 0.00012675283385292212,
      "loss": 12.5305,
      "step": 9867
    },
    {
      "epoch": 1.725551912568306,
      "grad_norm": 0.6444841757188374,
      "learning_rate": 0.0001267391888627197,
      "loss": 12.4515,
      "step": 9868
    },
    {
      "epoch": 1.725726775956284,
      "grad_norm": 0.6342217944472192,
      "learning_rate": 0.00012672554333631447,
      "loss": 12.2969,
      "step": 9869
    },
    {
      "epoch": 1.7259016393442623,
      "grad_norm": 0.7280411388140439,
      "learning_rate": 0.00012671189727398008,
      "loss": 12.3854,
      "step": 9870
    },
    {
      "epoch": 1.7260765027322404,
      "grad_norm": 0.6826633181899951,
      "learning_rate": 0.00012669825067599007,
      "loss": 12.3084,
      "step": 9871
    },
    {
      "epoch": 1.7262513661202186,
      "grad_norm": 0.7097771422032663,
      "learning_rate": 0.0001266846035426182,
      "loss": 12.4011,
      "step": 9872
    },
    {
      "epoch": 1.7264262295081967,
      "grad_norm": 0.7173492229253684,
      "learning_rate": 0.00012667095587413807,
      "loss": 12.533,
      "step": 9873
    },
    {
      "epoch": 1.7266010928961748,
      "grad_norm": 0.722062869210168,
      "learning_rate": 0.00012665730767082343,
      "loss": 12.3365,
      "step": 9874
    },
    {
      "epoch": 1.7267759562841531,
      "grad_norm": 0.6561811296964882,
      "learning_rate": 0.00012664365893294788,
      "loss": 12.4652,
      "step": 9875
    },
    {
      "epoch": 1.726950819672131,
      "grad_norm": 0.6388566974039374,
      "learning_rate": 0.00012663000966078516,
      "loss": 12.4154,
      "step": 9876
    },
    {
      "epoch": 1.7271256830601094,
      "grad_norm": 0.7125071727441201,
      "learning_rate": 0.000126616359854609,
      "loss": 12.4348,
      "step": 9877
    },
    {
      "epoch": 1.7273005464480873,
      "grad_norm": 0.7645950018225884,
      "learning_rate": 0.00012660270951469308,
      "loss": 12.3079,
      "step": 9878
    },
    {
      "epoch": 1.7274754098360656,
      "grad_norm": 0.7131657975954676,
      "learning_rate": 0.00012658905864131118,
      "loss": 12.4022,
      "step": 9879
    },
    {
      "epoch": 1.7276502732240437,
      "grad_norm": 0.7325521835647207,
      "learning_rate": 0.000126575407234737,
      "loss": 12.2722,
      "step": 9880
    },
    {
      "epoch": 1.7278251366120219,
      "grad_norm": 0.6874763115111685,
      "learning_rate": 0.0001265617552952443,
      "loss": 12.5159,
      "step": 9881
    },
    {
      "epoch": 1.728,
      "grad_norm": 0.7410809106850538,
      "learning_rate": 0.0001265481028231069,
      "loss": 12.37,
      "step": 9882
    },
    {
      "epoch": 1.728174863387978,
      "grad_norm": 0.6753608124739435,
      "learning_rate": 0.00012653444981859847,
      "loss": 12.297,
      "step": 9883
    },
    {
      "epoch": 1.7283497267759564,
      "grad_norm": 0.725427487589213,
      "learning_rate": 0.00012652079628199287,
      "loss": 12.1727,
      "step": 9884
    },
    {
      "epoch": 1.7285245901639343,
      "grad_norm": 0.7215933088239178,
      "learning_rate": 0.00012650714221356388,
      "loss": 12.1644,
      "step": 9885
    },
    {
      "epoch": 1.7286994535519127,
      "grad_norm": 0.6847975247769252,
      "learning_rate": 0.00012649348761358532,
      "loss": 12.3685,
      "step": 9886
    },
    {
      "epoch": 1.7288743169398906,
      "grad_norm": 0.7630644693497991,
      "learning_rate": 0.00012647983248233093,
      "loss": 12.2715,
      "step": 9887
    },
    {
      "epoch": 1.729049180327869,
      "grad_norm": 0.6465743855433582,
      "learning_rate": 0.00012646617682007463,
      "loss": 12.3988,
      "step": 9888
    },
    {
      "epoch": 1.729224043715847,
      "grad_norm": 0.7354707149975148,
      "learning_rate": 0.0001264525206270902,
      "loss": 12.3591,
      "step": 9889
    },
    {
      "epoch": 1.7293989071038252,
      "grad_norm": 0.6427414034578236,
      "learning_rate": 0.00012643886390365156,
      "loss": 11.9446,
      "step": 9890
    },
    {
      "epoch": 1.7295737704918033,
      "grad_norm": 0.6748487782975818,
      "learning_rate": 0.00012642520665003248,
      "loss": 12.305,
      "step": 9891
    },
    {
      "epoch": 1.7297486338797814,
      "grad_norm": 0.6923578330155408,
      "learning_rate": 0.00012641154886650687,
      "loss": 12.4128,
      "step": 9892
    },
    {
      "epoch": 1.7299234972677595,
      "grad_norm": 0.6766905136200115,
      "learning_rate": 0.00012639789055334864,
      "loss": 12.3775,
      "step": 9893
    },
    {
      "epoch": 1.7300983606557376,
      "grad_norm": 0.6757682365602093,
      "learning_rate": 0.00012638423171083165,
      "loss": 12.3212,
      "step": 9894
    },
    {
      "epoch": 1.730273224043716,
      "grad_norm": 0.70434520445989,
      "learning_rate": 0.00012637057233922977,
      "loss": 12.3506,
      "step": 9895
    },
    {
      "epoch": 1.7304480874316939,
      "grad_norm": 0.6751384540985659,
      "learning_rate": 0.000126356912438817,
      "loss": 12.3758,
      "step": 9896
    },
    {
      "epoch": 1.7306229508196722,
      "grad_norm": 0.6957025080818338,
      "learning_rate": 0.00012634325200986714,
      "loss": 12.5241,
      "step": 9897
    },
    {
      "epoch": 1.73079781420765,
      "grad_norm": 0.7283079401581378,
      "learning_rate": 0.00012632959105265422,
      "loss": 12.3951,
      "step": 9898
    },
    {
      "epoch": 1.7309726775956285,
      "grad_norm": 0.7041184707282699,
      "learning_rate": 0.0001263159295674522,
      "loss": 12.2021,
      "step": 9899
    },
    {
      "epoch": 1.7311475409836066,
      "grad_norm": 0.6830287410288004,
      "learning_rate": 0.00012630226755453494,
      "loss": 12.4845,
      "step": 9900
    },
    {
      "epoch": 1.7313224043715847,
      "grad_norm": 0.6979800658615173,
      "learning_rate": 0.0001262886050141765,
      "loss": 12.5467,
      "step": 9901
    },
    {
      "epoch": 1.7314972677595628,
      "grad_norm": 0.6945635163429194,
      "learning_rate": 0.00012627494194665076,
      "loss": 12.2395,
      "step": 9902
    },
    {
      "epoch": 1.731672131147541,
      "grad_norm": 0.6800481849202418,
      "learning_rate": 0.00012626127835223177,
      "loss": 12.182,
      "step": 9903
    },
    {
      "epoch": 1.7318469945355193,
      "grad_norm": 0.7148058831287123,
      "learning_rate": 0.00012624761423119356,
      "loss": 12.2653,
      "step": 9904
    },
    {
      "epoch": 1.7320218579234972,
      "grad_norm": 0.6455067185911147,
      "learning_rate": 0.00012623394958381006,
      "loss": 12.2246,
      "step": 9905
    },
    {
      "epoch": 1.7321967213114755,
      "grad_norm": 0.6740686100917117,
      "learning_rate": 0.00012622028441035534,
      "loss": 12.1201,
      "step": 9906
    },
    {
      "epoch": 1.7323715846994534,
      "grad_norm": 0.675968899521444,
      "learning_rate": 0.00012620661871110336,
      "loss": 12.2432,
      "step": 9907
    },
    {
      "epoch": 1.7325464480874317,
      "grad_norm": 0.7161398841795543,
      "learning_rate": 0.00012619295248632824,
      "loss": 12.4649,
      "step": 9908
    },
    {
      "epoch": 1.7327213114754099,
      "grad_norm": 0.7206580378132642,
      "learning_rate": 0.00012617928573630406,
      "loss": 12.05,
      "step": 9909
    },
    {
      "epoch": 1.732896174863388,
      "grad_norm": 0.6797347658113462,
      "learning_rate": 0.00012616561846130473,
      "loss": 12.3426,
      "step": 9910
    },
    {
      "epoch": 1.733071038251366,
      "grad_norm": 0.6837157555200168,
      "learning_rate": 0.00012615195066160447,
      "loss": 12.1605,
      "step": 9911
    },
    {
      "epoch": 1.7332459016393442,
      "grad_norm": 0.6395468962457537,
      "learning_rate": 0.00012613828233747727,
      "loss": 12.2833,
      "step": 9912
    },
    {
      "epoch": 1.7334207650273226,
      "grad_norm": 0.7752075775873539,
      "learning_rate": 0.0001261246134891973,
      "loss": 12.4677,
      "step": 9913
    },
    {
      "epoch": 1.7335956284153005,
      "grad_norm": 0.7064433758199271,
      "learning_rate": 0.00012611094411703853,
      "loss": 12.4584,
      "step": 9914
    },
    {
      "epoch": 1.7337704918032788,
      "grad_norm": 0.6311645697724872,
      "learning_rate": 0.00012609727422127525,
      "loss": 12.0961,
      "step": 9915
    },
    {
      "epoch": 1.7339453551912567,
      "grad_norm": 0.7034991762165996,
      "learning_rate": 0.00012608360380218144,
      "loss": 12.1158,
      "step": 9916
    },
    {
      "epoch": 1.734120218579235,
      "grad_norm": 0.68663791304173,
      "learning_rate": 0.0001260699328600313,
      "loss": 12.1181,
      "step": 9917
    },
    {
      "epoch": 1.7342950819672132,
      "grad_norm": 0.768630185171637,
      "learning_rate": 0.000126056261395099,
      "loss": 12.3898,
      "step": 9918
    },
    {
      "epoch": 1.7344699453551913,
      "grad_norm": 0.7039386188994597,
      "learning_rate": 0.0001260425894076586,
      "loss": 12.238,
      "step": 9919
    },
    {
      "epoch": 1.7346448087431694,
      "grad_norm": 0.6361569963353508,
      "learning_rate": 0.00012602891689798433,
      "loss": 12.473,
      "step": 9920
    },
    {
      "epoch": 1.7348196721311475,
      "grad_norm": 0.7814336958411379,
      "learning_rate": 0.00012601524386635036,
      "loss": 12.1284,
      "step": 9921
    },
    {
      "epoch": 1.7349945355191256,
      "grad_norm": 0.6822453550288025,
      "learning_rate": 0.0001260015703130309,
      "loss": 12.3106,
      "step": 9922
    },
    {
      "epoch": 1.7351693989071038,
      "grad_norm": 0.7108762208353381,
      "learning_rate": 0.00012598789623830008,
      "loss": 12.3804,
      "step": 9923
    },
    {
      "epoch": 1.735344262295082,
      "grad_norm": 0.6589426413556494,
      "learning_rate": 0.0001259742216424322,
      "loss": 12.1649,
      "step": 9924
    },
    {
      "epoch": 1.73551912568306,
      "grad_norm": 0.8036087697642129,
      "learning_rate": 0.00012596054652570134,
      "loss": 12.4818,
      "step": 9925
    },
    {
      "epoch": 1.7356939890710383,
      "grad_norm": 0.7778073447099932,
      "learning_rate": 0.00012594687088838185,
      "loss": 12.0997,
      "step": 9926
    },
    {
      "epoch": 1.7358688524590162,
      "grad_norm": 0.655914638975559,
      "learning_rate": 0.000125933194730748,
      "loss": 12.3058,
      "step": 9927
    },
    {
      "epoch": 1.7360437158469946,
      "grad_norm": 0.8353668887130825,
      "learning_rate": 0.00012591951805307388,
      "loss": 12.3009,
      "step": 9928
    },
    {
      "epoch": 1.7362185792349727,
      "grad_norm": 0.6841254759285212,
      "learning_rate": 0.0001259058408556339,
      "loss": 12.3823,
      "step": 9929
    },
    {
      "epoch": 1.7363934426229508,
      "grad_norm": 0.7863488322851788,
      "learning_rate": 0.00012589216313870223,
      "loss": 12.2601,
      "step": 9930
    },
    {
      "epoch": 1.736568306010929,
      "grad_norm": 0.7281615202662696,
      "learning_rate": 0.0001258784849025532,
      "loss": 12.4907,
      "step": 9931
    },
    {
      "epoch": 1.736743169398907,
      "grad_norm": 0.7288037656694715,
      "learning_rate": 0.0001258648061474611,
      "loss": 12.2394,
      "step": 9932
    },
    {
      "epoch": 1.7369180327868854,
      "grad_norm": 0.7005805453664802,
      "learning_rate": 0.00012585112687370022,
      "loss": 12.4935,
      "step": 9933
    },
    {
      "epoch": 1.7370928961748633,
      "grad_norm": 0.7306658311115047,
      "learning_rate": 0.00012583744708154486,
      "loss": 12.2553,
      "step": 9934
    },
    {
      "epoch": 1.7372677595628416,
      "grad_norm": 0.7788577312998346,
      "learning_rate": 0.00012582376677126937,
      "loss": 12.4194,
      "step": 9935
    },
    {
      "epoch": 1.7374426229508195,
      "grad_norm": 0.7183478611318928,
      "learning_rate": 0.00012581008594314804,
      "loss": 12.2046,
      "step": 9936
    },
    {
      "epoch": 1.7376174863387979,
      "grad_norm": 0.7487840533358506,
      "learning_rate": 0.00012579640459745527,
      "loss": 12.4701,
      "step": 9937
    },
    {
      "epoch": 1.737792349726776,
      "grad_norm": 0.7978108287620774,
      "learning_rate": 0.00012578272273446536,
      "loss": 12.4669,
      "step": 9938
    },
    {
      "epoch": 1.737967213114754,
      "grad_norm": 0.6547946835407651,
      "learning_rate": 0.0001257690403544527,
      "loss": 12.5364,
      "step": 9939
    },
    {
      "epoch": 1.7381420765027322,
      "grad_norm": 0.7530503283402877,
      "learning_rate": 0.0001257553574576917,
      "loss": 12.4315,
      "step": 9940
    },
    {
      "epoch": 1.7383169398907103,
      "grad_norm": 0.724361225890344,
      "learning_rate": 0.00012574167404445666,
      "loss": 12.4056,
      "step": 9941
    },
    {
      "epoch": 1.7384918032786887,
      "grad_norm": 0.7666846679792175,
      "learning_rate": 0.00012572799011502205,
      "loss": 12.3827,
      "step": 9942
    },
    {
      "epoch": 1.7386666666666666,
      "grad_norm": 0.7066209608657411,
      "learning_rate": 0.0001257143056696622,
      "loss": 12.4303,
      "step": 9943
    },
    {
      "epoch": 1.738841530054645,
      "grad_norm": 0.7870530215608724,
      "learning_rate": 0.00012570062070865157,
      "loss": 12.3993,
      "step": 9944
    },
    {
      "epoch": 1.7390163934426228,
      "grad_norm": 0.7201403236905035,
      "learning_rate": 0.0001256869352322646,
      "loss": 12.4482,
      "step": 9945
    },
    {
      "epoch": 1.7391912568306012,
      "grad_norm": 0.7726194827337902,
      "learning_rate": 0.00012567324924077572,
      "loss": 12.2405,
      "step": 9946
    },
    {
      "epoch": 1.7393661202185793,
      "grad_norm": 0.646233068776728,
      "learning_rate": 0.0001256595627344594,
      "loss": 11.9595,
      "step": 9947
    },
    {
      "epoch": 1.7395409836065574,
      "grad_norm": 0.7320274882193333,
      "learning_rate": 0.00012564587571359,
      "loss": 12.0389,
      "step": 9948
    },
    {
      "epoch": 1.7397158469945355,
      "grad_norm": 0.7587029764552937,
      "learning_rate": 0.00012563218817844207,
      "loss": 12.3055,
      "step": 9949
    },
    {
      "epoch": 1.7398907103825136,
      "grad_norm": 0.669374892237163,
      "learning_rate": 0.00012561850012929006,
      "loss": 12.3491,
      "step": 9950
    },
    {
      "epoch": 1.7400655737704918,
      "grad_norm": 0.6710324214237767,
      "learning_rate": 0.00012560481156640848,
      "loss": 11.9777,
      "step": 9951
    },
    {
      "epoch": 1.7402404371584699,
      "grad_norm": 0.7093708160598348,
      "learning_rate": 0.00012559112249007178,
      "loss": 12.2386,
      "step": 9952
    },
    {
      "epoch": 1.7404153005464482,
      "grad_norm": 0.7729542686256349,
      "learning_rate": 0.00012557743290055452,
      "loss": 12.3307,
      "step": 9953
    },
    {
      "epoch": 1.7405901639344261,
      "grad_norm": 0.6636580074539942,
      "learning_rate": 0.00012556374279813122,
      "loss": 12.179,
      "step": 9954
    },
    {
      "epoch": 1.7407650273224045,
      "grad_norm": 0.8287276870948247,
      "learning_rate": 0.00012555005218307634,
      "loss": 12.4422,
      "step": 9955
    },
    {
      "epoch": 1.7409398907103824,
      "grad_norm": 0.6251875347962199,
      "learning_rate": 0.00012553636105566452,
      "loss": 12.0643,
      "step": 9956
    },
    {
      "epoch": 1.7411147540983607,
      "grad_norm": 0.7419387655522421,
      "learning_rate": 0.00012552266941617018,
      "loss": 12.3693,
      "step": 9957
    },
    {
      "epoch": 1.7412896174863388,
      "grad_norm": 0.6505577489659413,
      "learning_rate": 0.000125508977264868,
      "loss": 12.4993,
      "step": 9958
    },
    {
      "epoch": 1.741464480874317,
      "grad_norm": 0.8240075173629623,
      "learning_rate": 0.0001254952846020325,
      "loss": 12.1451,
      "step": 9959
    },
    {
      "epoch": 1.741639344262295,
      "grad_norm": 0.6369548555653957,
      "learning_rate": 0.00012548159142793824,
      "loss": 12.3539,
      "step": 9960
    },
    {
      "epoch": 1.7418142076502732,
      "grad_norm": 0.7336576051120143,
      "learning_rate": 0.00012546789774285988,
      "loss": 12.313,
      "step": 9961
    },
    {
      "epoch": 1.7419890710382515,
      "grad_norm": 0.7182202951816496,
      "learning_rate": 0.0001254542035470719,
      "loss": 12.3706,
      "step": 9962
    },
    {
      "epoch": 1.7421639344262294,
      "grad_norm": 0.7289219523967989,
      "learning_rate": 0.00012544050884084909,
      "loss": 12.0547,
      "step": 9963
    },
    {
      "epoch": 1.7423387978142078,
      "grad_norm": 0.6878102129824737,
      "learning_rate": 0.00012542681362446586,
      "loss": 12.4507,
      "step": 9964
    },
    {
      "epoch": 1.7425136612021856,
      "grad_norm": 0.7092923746472868,
      "learning_rate": 0.00012541311789819704,
      "loss": 12.4381,
      "step": 9965
    },
    {
      "epoch": 1.742688524590164,
      "grad_norm": 0.7080767784480603,
      "learning_rate": 0.00012539942166231712,
      "loss": 12.0216,
      "step": 9966
    },
    {
      "epoch": 1.742863387978142,
      "grad_norm": 0.6642714629779226,
      "learning_rate": 0.0001253857249171008,
      "loss": 12.4803,
      "step": 9967
    },
    {
      "epoch": 1.7430382513661202,
      "grad_norm": 0.7033395725177936,
      "learning_rate": 0.00012537202766282278,
      "loss": 12.09,
      "step": 9968
    },
    {
      "epoch": 1.7432131147540983,
      "grad_norm": 0.739293919579558,
      "learning_rate": 0.00012535832989975766,
      "loss": 12.2236,
      "step": 9969
    },
    {
      "epoch": 1.7433879781420765,
      "grad_norm": 0.7350867825916231,
      "learning_rate": 0.00012534463162818023,
      "loss": 12.1513,
      "step": 9970
    },
    {
      "epoch": 1.7435628415300548,
      "grad_norm": 0.693956100360983,
      "learning_rate": 0.0001253309328483651,
      "loss": 12.3934,
      "step": 9971
    },
    {
      "epoch": 1.7437377049180327,
      "grad_norm": 0.7452463270074003,
      "learning_rate": 0.00012531723356058695,
      "loss": 12.1381,
      "step": 9972
    },
    {
      "epoch": 1.743912568306011,
      "grad_norm": 0.6538465842828878,
      "learning_rate": 0.00012530353376512057,
      "loss": 12.1965,
      "step": 9973
    },
    {
      "epoch": 1.744087431693989,
      "grad_norm": 0.7302512646712295,
      "learning_rate": 0.00012528983346224063,
      "loss": 12.1615,
      "step": 9974
    },
    {
      "epoch": 1.7442622950819673,
      "grad_norm": 0.662279323215206,
      "learning_rate": 0.00012527613265222187,
      "loss": 12.1561,
      "step": 9975
    },
    {
      "epoch": 1.7444371584699454,
      "grad_norm": 0.7147534646419335,
      "learning_rate": 0.00012526243133533904,
      "loss": 12.19,
      "step": 9976
    },
    {
      "epoch": 1.7446120218579235,
      "grad_norm": 0.704917043668193,
      "learning_rate": 0.00012524872951186693,
      "loss": 12.2295,
      "step": 9977
    },
    {
      "epoch": 1.7447868852459016,
      "grad_norm": 0.6397630219378847,
      "learning_rate": 0.00012523502718208028,
      "loss": 12.1095,
      "step": 9978
    },
    {
      "epoch": 1.7449617486338798,
      "grad_norm": 0.6376354466948919,
      "learning_rate": 0.0001252213243462538,
      "loss": 12.1349,
      "step": 9979
    },
    {
      "epoch": 1.7451366120218579,
      "grad_norm": 0.7142543549897259,
      "learning_rate": 0.00012520762100466233,
      "loss": 12.2501,
      "step": 9980
    },
    {
      "epoch": 1.745311475409836,
      "grad_norm": 0.6739398284036849,
      "learning_rate": 0.0001251939171575807,
      "loss": 12.3758,
      "step": 9981
    },
    {
      "epoch": 1.7454863387978143,
      "grad_norm": 0.6360651815200191,
      "learning_rate": 0.00012518021280528367,
      "loss": 12.1171,
      "step": 9982
    },
    {
      "epoch": 1.7456612021857922,
      "grad_norm": 0.6215295894869198,
      "learning_rate": 0.00012516650794804605,
      "loss": 12.3378,
      "step": 9983
    },
    {
      "epoch": 1.7458360655737706,
      "grad_norm": 0.6667640766848081,
      "learning_rate": 0.00012515280258614266,
      "loss": 12.4913,
      "step": 9984
    },
    {
      "epoch": 1.7460109289617485,
      "grad_norm": 0.6389960763767616,
      "learning_rate": 0.00012513909671984835,
      "loss": 12.4122,
      "step": 9985
    },
    {
      "epoch": 1.7461857923497268,
      "grad_norm": 0.6626577563417072,
      "learning_rate": 0.000125125390349438,
      "loss": 12.418,
      "step": 9986
    },
    {
      "epoch": 1.746360655737705,
      "grad_norm": 0.7554964998176463,
      "learning_rate": 0.00012511168347518637,
      "loss": 12.2012,
      "step": 9987
    },
    {
      "epoch": 1.746535519125683,
      "grad_norm": 0.6677395298615623,
      "learning_rate": 0.00012509797609736844,
      "loss": 12.1446,
      "step": 9988
    },
    {
      "epoch": 1.7467103825136612,
      "grad_norm": 0.6450279481891295,
      "learning_rate": 0.00012508426821625901,
      "loss": 12.4147,
      "step": 9989
    },
    {
      "epoch": 1.7468852459016393,
      "grad_norm": 0.6646995536079155,
      "learning_rate": 0.00012507055983213298,
      "loss": 12.3429,
      "step": 9990
    },
    {
      "epoch": 1.7470601092896176,
      "grad_norm": 0.8329973661147969,
      "learning_rate": 0.00012505685094526525,
      "loss": 12.2712,
      "step": 9991
    },
    {
      "epoch": 1.7472349726775955,
      "grad_norm": 0.6449333730721883,
      "learning_rate": 0.00012504314155593072,
      "loss": 12.336,
      "step": 9992
    },
    {
      "epoch": 1.7474098360655739,
      "grad_norm": 0.7240849953023193,
      "learning_rate": 0.0001250294316644043,
      "loss": 12.3062,
      "step": 9993
    },
    {
      "epoch": 1.7475846994535518,
      "grad_norm": 0.6831984571614947,
      "learning_rate": 0.00012501572127096092,
      "loss": 12.3821,
      "step": 9994
    },
    {
      "epoch": 1.7477595628415301,
      "grad_norm": 0.6464502982663861,
      "learning_rate": 0.00012500201037587557,
      "loss": 12.348,
      "step": 9995
    },
    {
      "epoch": 1.7479344262295082,
      "grad_norm": 0.6863472702164594,
      "learning_rate": 0.0001249882989794231,
      "loss": 12.2343,
      "step": 9996
    },
    {
      "epoch": 1.7481092896174864,
      "grad_norm": 0.7311590209823406,
      "learning_rate": 0.0001249745870818785,
      "loss": 12.317,
      "step": 9997
    },
    {
      "epoch": 1.7482841530054645,
      "grad_norm": 0.6686169814202514,
      "learning_rate": 0.00012496087468351677,
      "loss": 12.2152,
      "step": 9998
    },
    {
      "epoch": 1.7484590163934426,
      "grad_norm": 0.714462957567508,
      "learning_rate": 0.00012494716178461287,
      "loss": 12.5153,
      "step": 9999
    },
    {
      "epoch": 1.748633879781421,
      "grad_norm": 0.710972797478173,
      "learning_rate": 0.00012493344838544176,
      "loss": 12.494,
      "step": 10000
    },
    {
      "epoch": 1.7488087431693988,
      "grad_norm": 0.8683603103646957,
      "learning_rate": 0.00012491973448627847,
      "loss": 12.153,
      "step": 10001
    },
    {
      "epoch": 1.7489836065573772,
      "grad_norm": 0.7413783675147096,
      "learning_rate": 0.000124906020087398,
      "loss": 12.3524,
      "step": 10002
    },
    {
      "epoch": 1.749158469945355,
      "grad_norm": 0.6634122853610077,
      "learning_rate": 0.00012489230518907533,
      "loss": 12.4759,
      "step": 10003
    },
    {
      "epoch": 1.7493333333333334,
      "grad_norm": 0.7302572355294887,
      "learning_rate": 0.00012487858979158553,
      "loss": 12.4928,
      "step": 10004
    },
    {
      "epoch": 1.7495081967213115,
      "grad_norm": 0.6866385257376124,
      "learning_rate": 0.0001248648738952036,
      "loss": 12.3258,
      "step": 10005
    },
    {
      "epoch": 1.7496830601092896,
      "grad_norm": 0.689693299059991,
      "learning_rate": 0.00012485115750020462,
      "loss": 12.2204,
      "step": 10006
    },
    {
      "epoch": 1.7498579234972678,
      "grad_norm": 0.7939903146475624,
      "learning_rate": 0.00012483744060686361,
      "loss": 12.3467,
      "step": 10007
    },
    {
      "epoch": 1.7500327868852459,
      "grad_norm": 0.696450282463391,
      "learning_rate": 0.00012482372321545566,
      "loss": 12.3358,
      "step": 10008
    },
    {
      "epoch": 1.750207650273224,
      "grad_norm": 0.6798829478209076,
      "learning_rate": 0.0001248100053262559,
      "loss": 12.2621,
      "step": 10009
    },
    {
      "epoch": 1.7503825136612021,
      "grad_norm": 0.7776958544987623,
      "learning_rate": 0.00012479628693953926,
      "loss": 12.2424,
      "step": 10010
    },
    {
      "epoch": 1.7505573770491805,
      "grad_norm": 0.7068542499679783,
      "learning_rate": 0.000124782568055581,
      "loss": 12.3526,
      "step": 10011
    },
    {
      "epoch": 1.7507322404371584,
      "grad_norm": 0.7036698576986329,
      "learning_rate": 0.0001247688486746561,
      "loss": 12.4556,
      "step": 10012
    },
    {
      "epoch": 1.7509071038251367,
      "grad_norm": 0.7358496450865559,
      "learning_rate": 0.00012475512879703979,
      "loss": 12.3545,
      "step": 10013
    },
    {
      "epoch": 1.7510819672131146,
      "grad_norm": 0.6494074328524163,
      "learning_rate": 0.00012474140842300713,
      "loss": 12.3008,
      "step": 10014
    },
    {
      "epoch": 1.751256830601093,
      "grad_norm": 0.782166980616829,
      "learning_rate": 0.0001247276875528332,
      "loss": 12.4981,
      "step": 10015
    },
    {
      "epoch": 1.751431693989071,
      "grad_norm": 0.6669348391171227,
      "learning_rate": 0.0001247139661867933,
      "loss": 12.1543,
      "step": 10016
    },
    {
      "epoch": 1.7516065573770492,
      "grad_norm": 0.748643500610589,
      "learning_rate": 0.00012470024432516242,
      "loss": 12.4583,
      "step": 10017
    },
    {
      "epoch": 1.7517814207650273,
      "grad_norm": 0.6927551690881756,
      "learning_rate": 0.00012468652196821587,
      "loss": 12.4104,
      "step": 10018
    },
    {
      "epoch": 1.7519562841530054,
      "grad_norm": 0.7351862880822472,
      "learning_rate": 0.00012467279911622874,
      "loss": 12.4272,
      "step": 10019
    },
    {
      "epoch": 1.7521311475409838,
      "grad_norm": 0.6984716668847635,
      "learning_rate": 0.00012465907576947622,
      "loss": 12.3258,
      "step": 10020
    },
    {
      "epoch": 1.7523060109289617,
      "grad_norm": 0.7354083561695876,
      "learning_rate": 0.00012464535192823351,
      "loss": 12.3132,
      "step": 10021
    },
    {
      "epoch": 1.75248087431694,
      "grad_norm": 0.6443159968806182,
      "learning_rate": 0.00012463162759277584,
      "loss": 12.3633,
      "step": 10022
    },
    {
      "epoch": 1.752655737704918,
      "grad_norm": 0.7174791504039278,
      "learning_rate": 0.0001246179027633784,
      "loss": 12.2769,
      "step": 10023
    },
    {
      "epoch": 1.7528306010928962,
      "grad_norm": 0.6840727487864947,
      "learning_rate": 0.00012460417744031638,
      "loss": 12.293,
      "step": 10024
    },
    {
      "epoch": 1.7530054644808744,
      "grad_norm": 0.8056611431670104,
      "learning_rate": 0.00012459045162386512,
      "loss": 12.1292,
      "step": 10025
    },
    {
      "epoch": 1.7531803278688525,
      "grad_norm": 0.7414202576468258,
      "learning_rate": 0.00012457672531429977,
      "loss": 12.4545,
      "step": 10026
    },
    {
      "epoch": 1.7533551912568306,
      "grad_norm": 0.8075193834025941,
      "learning_rate": 0.00012456299851189563,
      "loss": 12.2562,
      "step": 10027
    },
    {
      "epoch": 1.7535300546448087,
      "grad_norm": 0.6760556501684833,
      "learning_rate": 0.00012454927121692796,
      "loss": 12.4214,
      "step": 10028
    },
    {
      "epoch": 1.7537049180327868,
      "grad_norm": 0.8221578930055091,
      "learning_rate": 0.000124535543429672,
      "loss": 12.4315,
      "step": 10029
    },
    {
      "epoch": 1.753879781420765,
      "grad_norm": 0.7257448045114566,
      "learning_rate": 0.0001245218151504031,
      "loss": 12.0707,
      "step": 10030
    },
    {
      "epoch": 1.7540546448087433,
      "grad_norm": 0.7393162542808638,
      "learning_rate": 0.00012450808637939646,
      "loss": 12.4851,
      "step": 10031
    },
    {
      "epoch": 1.7542295081967212,
      "grad_norm": 0.7332346779594123,
      "learning_rate": 0.00012449435711692748,
      "loss": 12.4848,
      "step": 10032
    },
    {
      "epoch": 1.7544043715846995,
      "grad_norm": 0.6880963842562843,
      "learning_rate": 0.00012448062736327141,
      "loss": 12.2669,
      "step": 10033
    },
    {
      "epoch": 1.7545792349726776,
      "grad_norm": 0.7922820021222102,
      "learning_rate": 0.0001244668971187036,
      "loss": 12.0151,
      "step": 10034
    },
    {
      "epoch": 1.7547540983606558,
      "grad_norm": 0.7669720392186065,
      "learning_rate": 0.0001244531663834994,
      "loss": 12.0861,
      "step": 10035
    },
    {
      "epoch": 1.7549289617486339,
      "grad_norm": 0.7657029556641853,
      "learning_rate": 0.0001244394351579341,
      "loss": 12.1978,
      "step": 10036
    },
    {
      "epoch": 1.755103825136612,
      "grad_norm": 0.6941190377473737,
      "learning_rate": 0.00012442570344228313,
      "loss": 12.1979,
      "step": 10037
    },
    {
      "epoch": 1.7552786885245901,
      "grad_norm": 0.7310028925357221,
      "learning_rate": 0.0001244119712368218,
      "loss": 12.3507,
      "step": 10038
    },
    {
      "epoch": 1.7554535519125682,
      "grad_norm": 0.8429258495042141,
      "learning_rate": 0.00012439823854182544,
      "loss": 12.1391,
      "step": 10039
    },
    {
      "epoch": 1.7556284153005466,
      "grad_norm": 0.668415088403265,
      "learning_rate": 0.00012438450535756955,
      "loss": 12.4285,
      "step": 10040
    },
    {
      "epoch": 1.7558032786885245,
      "grad_norm": 0.6749565848803235,
      "learning_rate": 0.0001243707716843294,
      "loss": 12.0356,
      "step": 10041
    },
    {
      "epoch": 1.7559781420765028,
      "grad_norm": 0.698302211819618,
      "learning_rate": 0.00012435703752238048,
      "loss": 12.126,
      "step": 10042
    },
    {
      "epoch": 1.7561530054644807,
      "grad_norm": 0.7440991767779657,
      "learning_rate": 0.0001243433028719982,
      "loss": 12.2001,
      "step": 10043
    },
    {
      "epoch": 1.756327868852459,
      "grad_norm": 0.6882139730429645,
      "learning_rate": 0.00012432956773345788,
      "loss": 12.3511,
      "step": 10044
    },
    {
      "epoch": 1.7565027322404372,
      "grad_norm": 0.6369902744721732,
      "learning_rate": 0.00012431583210703508,
      "loss": 12.4137,
      "step": 10045
    },
    {
      "epoch": 1.7566775956284153,
      "grad_norm": 0.7116779626577235,
      "learning_rate": 0.00012430209599300517,
      "loss": 12.1673,
      "step": 10046
    },
    {
      "epoch": 1.7568524590163934,
      "grad_norm": 0.653431490556103,
      "learning_rate": 0.00012428835939164363,
      "loss": 12.2324,
      "step": 10047
    },
    {
      "epoch": 1.7570273224043715,
      "grad_norm": 0.7177500193856188,
      "learning_rate": 0.0001242746223032259,
      "loss": 12.0312,
      "step": 10048
    },
    {
      "epoch": 1.7572021857923499,
      "grad_norm": 0.6426140929231761,
      "learning_rate": 0.00012426088472802747,
      "loss": 12.3012,
      "step": 10049
    },
    {
      "epoch": 1.7573770491803278,
      "grad_norm": 0.7275314559773507,
      "learning_rate": 0.0001242471466663238,
      "loss": 12.0033,
      "step": 10050
    },
    {
      "epoch": 1.7575519125683061,
      "grad_norm": 0.7267412339839958,
      "learning_rate": 0.00012423340811839036,
      "loss": 12.1717,
      "step": 10051
    },
    {
      "epoch": 1.757726775956284,
      "grad_norm": 0.7887184706002999,
      "learning_rate": 0.00012421966908450277,
      "loss": 12.2356,
      "step": 10052
    },
    {
      "epoch": 1.7579016393442624,
      "grad_norm": 0.7396702505257181,
      "learning_rate": 0.00012420592956493634,
      "loss": 12.3872,
      "step": 10053
    },
    {
      "epoch": 1.7580765027322405,
      "grad_norm": 0.7101049550133715,
      "learning_rate": 0.00012419218955996676,
      "loss": 12.3851,
      "step": 10054
    },
    {
      "epoch": 1.7582513661202186,
      "grad_norm": 0.7607714186445573,
      "learning_rate": 0.00012417844906986952,
      "loss": 12.4419,
      "step": 10055
    },
    {
      "epoch": 1.7584262295081967,
      "grad_norm": 0.6963294527406065,
      "learning_rate": 0.00012416470809492011,
      "loss": 12.3758,
      "step": 10056
    },
    {
      "epoch": 1.7586010928961748,
      "grad_norm": 0.6725527630913515,
      "learning_rate": 0.0001241509666353941,
      "loss": 12.1455,
      "step": 10057
    },
    {
      "epoch": 1.758775956284153,
      "grad_norm": 0.7235498222302938,
      "learning_rate": 0.00012413722469156708,
      "loss": 12.1803,
      "step": 10058
    },
    {
      "epoch": 1.758950819672131,
      "grad_norm": 0.668205695070279,
      "learning_rate": 0.0001241234822637146,
      "loss": 12.5043,
      "step": 10059
    },
    {
      "epoch": 1.7591256830601094,
      "grad_norm": 0.6940210716772859,
      "learning_rate": 0.00012410973935211218,
      "loss": 12.0714,
      "step": 10060
    },
    {
      "epoch": 1.7593005464480873,
      "grad_norm": 0.719468931923177,
      "learning_rate": 0.00012409599595703552,
      "loss": 12.2789,
      "step": 10061
    },
    {
      "epoch": 1.7594754098360657,
      "grad_norm": 0.6669694377973443,
      "learning_rate": 0.00012408225207876012,
      "loss": 12.2581,
      "step": 10062
    },
    {
      "epoch": 1.7596502732240438,
      "grad_norm": 0.741527132843904,
      "learning_rate": 0.0001240685077175616,
      "loss": 12.3267,
      "step": 10063
    },
    {
      "epoch": 1.759825136612022,
      "grad_norm": 0.7185213103867133,
      "learning_rate": 0.00012405476287371566,
      "loss": 12.4364,
      "step": 10064
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7061381386452605,
      "learning_rate": 0.00012404101754749782,
      "loss": 12.2479,
      "step": 10065
    },
    {
      "epoch": 1.7601748633879781,
      "grad_norm": 0.648271143955415,
      "learning_rate": 0.0001240272717391838,
      "loss": 12.4015,
      "step": 10066
    },
    {
      "epoch": 1.7603497267759562,
      "grad_norm": 0.7340375788402638,
      "learning_rate": 0.00012401352544904917,
      "loss": 12.5072,
      "step": 10067
    },
    {
      "epoch": 1.7605245901639344,
      "grad_norm": 0.6690075942648747,
      "learning_rate": 0.00012399977867736967,
      "loss": 12.4154,
      "step": 10068
    },
    {
      "epoch": 1.7606994535519127,
      "grad_norm": 0.6475327798919561,
      "learning_rate": 0.00012398603142442087,
      "loss": 12.2581,
      "step": 10069
    },
    {
      "epoch": 1.7608743169398906,
      "grad_norm": 0.725382140296607,
      "learning_rate": 0.00012397228369047852,
      "loss": 12.5537,
      "step": 10070
    },
    {
      "epoch": 1.761049180327869,
      "grad_norm": 0.716807811825303,
      "learning_rate": 0.00012395853547581828,
      "loss": 12.2625,
      "step": 10071
    },
    {
      "epoch": 1.7612240437158468,
      "grad_norm": 0.6247473426129058,
      "learning_rate": 0.00012394478678071582,
      "loss": 12.1932,
      "step": 10072
    },
    {
      "epoch": 1.7613989071038252,
      "grad_norm": 0.6567315360765105,
      "learning_rate": 0.0001239310376054469,
      "loss": 12.4321,
      "step": 10073
    },
    {
      "epoch": 1.7615737704918033,
      "grad_norm": 0.6855464833342239,
      "learning_rate": 0.00012391728795028718,
      "loss": 12.0971,
      "step": 10074
    },
    {
      "epoch": 1.7617486338797814,
      "grad_norm": 0.6974896075770064,
      "learning_rate": 0.0001239035378155124,
      "loss": 12.4746,
      "step": 10075
    },
    {
      "epoch": 1.7619234972677595,
      "grad_norm": 0.6610203355650214,
      "learning_rate": 0.00012388978720139824,
      "loss": 12.381,
      "step": 10076
    },
    {
      "epoch": 1.7620983606557377,
      "grad_norm": 0.7059156411929902,
      "learning_rate": 0.00012387603610822056,
      "loss": 12.4531,
      "step": 10077
    },
    {
      "epoch": 1.762273224043716,
      "grad_norm": 0.6957596670874698,
      "learning_rate": 0.00012386228453625505,
      "loss": 12.0692,
      "step": 10078
    },
    {
      "epoch": 1.762448087431694,
      "grad_norm": 0.6573690606169363,
      "learning_rate": 0.00012384853248577745,
      "loss": 12.304,
      "step": 10079
    },
    {
      "epoch": 1.7626229508196722,
      "grad_norm": 0.6633831086935555,
      "learning_rate": 0.00012383477995706353,
      "loss": 12.3544,
      "step": 10080
    },
    {
      "epoch": 1.7627978142076501,
      "grad_norm": 0.6944628526647797,
      "learning_rate": 0.00012382102695038914,
      "loss": 12.0695,
      "step": 10081
    },
    {
      "epoch": 1.7629726775956285,
      "grad_norm": 0.691509385523581,
      "learning_rate": 0.00012380727346602996,
      "loss": 12.1321,
      "step": 10082
    },
    {
      "epoch": 1.7631475409836066,
      "grad_norm": 0.6739436285681631,
      "learning_rate": 0.00012379351950426187,
      "loss": 12.4969,
      "step": 10083
    },
    {
      "epoch": 1.7633224043715847,
      "grad_norm": 0.9521857868175966,
      "learning_rate": 0.0001237797650653607,
      "loss": 11.8994,
      "step": 10084
    },
    {
      "epoch": 1.7634972677595628,
      "grad_norm": 0.7097473967958766,
      "learning_rate": 0.0001237660101496022,
      "loss": 12.1625,
      "step": 10085
    },
    {
      "epoch": 1.763672131147541,
      "grad_norm": 0.7170403439322293,
      "learning_rate": 0.0001237522547572622,
      "loss": 12.368,
      "step": 10086
    },
    {
      "epoch": 1.763846994535519,
      "grad_norm": 0.6749940705000759,
      "learning_rate": 0.00012373849888861656,
      "loss": 12.2085,
      "step": 10087
    },
    {
      "epoch": 1.7640218579234972,
      "grad_norm": 0.7189527512783928,
      "learning_rate": 0.00012372474254394118,
      "loss": 12.2491,
      "step": 10088
    },
    {
      "epoch": 1.7641967213114755,
      "grad_norm": 0.7387117037795587,
      "learning_rate": 0.00012371098572351184,
      "loss": 12.408,
      "step": 10089
    },
    {
      "epoch": 1.7643715846994534,
      "grad_norm": 0.7101595949997672,
      "learning_rate": 0.00012369722842760442,
      "loss": 12.3877,
      "step": 10090
    },
    {
      "epoch": 1.7645464480874318,
      "grad_norm": 0.7022008593377674,
      "learning_rate": 0.00012368347065649484,
      "loss": 12.1352,
      "step": 10091
    },
    {
      "epoch": 1.7647213114754097,
      "grad_norm": 0.6843781201705816,
      "learning_rate": 0.00012366971241045894,
      "loss": 12.5235,
      "step": 10092
    },
    {
      "epoch": 1.764896174863388,
      "grad_norm": 0.7055507422488394,
      "learning_rate": 0.0001236559536897726,
      "loss": 12.3619,
      "step": 10093
    },
    {
      "epoch": 1.7650710382513661,
      "grad_norm": 0.7742842963564194,
      "learning_rate": 0.0001236421944947118,
      "loss": 12.1336,
      "step": 10094
    },
    {
      "epoch": 1.7652459016393443,
      "grad_norm": 0.6484649116502766,
      "learning_rate": 0.00012362843482555238,
      "loss": 12.2592,
      "step": 10095
    },
    {
      "epoch": 1.7654207650273224,
      "grad_norm": 0.808614215289693,
      "learning_rate": 0.0001236146746825703,
      "loss": 12.279,
      "step": 10096
    },
    {
      "epoch": 1.7655956284153005,
      "grad_norm": 0.6754212658227726,
      "learning_rate": 0.0001236009140660415,
      "loss": 12.4259,
      "step": 10097
    },
    {
      "epoch": 1.7657704918032788,
      "grad_norm": 0.7704101264125504,
      "learning_rate": 0.00012358715297624192,
      "loss": 12.362,
      "step": 10098
    },
    {
      "epoch": 1.7659453551912567,
      "grad_norm": 0.7276191989480523,
      "learning_rate": 0.00012357339141344745,
      "loss": 12.1931,
      "step": 10099
    },
    {
      "epoch": 1.766120218579235,
      "grad_norm": 0.7823314341696856,
      "learning_rate": 0.00012355962937793418,
      "loss": 12.23,
      "step": 10100
    },
    {
      "epoch": 1.766295081967213,
      "grad_norm": 0.714751720895931,
      "learning_rate": 0.00012354586686997792,
      "loss": 12.2753,
      "step": 10101
    },
    {
      "epoch": 1.7664699453551913,
      "grad_norm": 0.6609287331921332,
      "learning_rate": 0.0001235321038898548,
      "loss": 12.2863,
      "step": 10102
    },
    {
      "epoch": 1.7666448087431694,
      "grad_norm": 0.7891699250216146,
      "learning_rate": 0.0001235183404378407,
      "loss": 12.2148,
      "step": 10103
    },
    {
      "epoch": 1.7668196721311475,
      "grad_norm": 0.7161165051210276,
      "learning_rate": 0.00012350457651421167,
      "loss": 12.4318,
      "step": 10104
    },
    {
      "epoch": 1.7669945355191257,
      "grad_norm": 0.6468441881513726,
      "learning_rate": 0.00012349081211924376,
      "loss": 12.3751,
      "step": 10105
    },
    {
      "epoch": 1.7671693989071038,
      "grad_norm": 0.7024587174418621,
      "learning_rate": 0.00012347704725321288,
      "loss": 12.3928,
      "step": 10106
    },
    {
      "epoch": 1.7673442622950821,
      "grad_norm": 0.6369151544709672,
      "learning_rate": 0.00012346328191639516,
      "loss": 12.4751,
      "step": 10107
    },
    {
      "epoch": 1.76751912568306,
      "grad_norm": 0.7244202144562149,
      "learning_rate": 0.00012344951610906654,
      "loss": 12.0898,
      "step": 10108
    },
    {
      "epoch": 1.7676939890710384,
      "grad_norm": 0.7219348386091048,
      "learning_rate": 0.00012343574983150318,
      "loss": 12.2456,
      "step": 10109
    },
    {
      "epoch": 1.7678688524590163,
      "grad_norm": 0.7160982511961327,
      "learning_rate": 0.00012342198308398108,
      "loss": 12.3136,
      "step": 10110
    },
    {
      "epoch": 1.7680437158469946,
      "grad_norm": 0.7189162010993372,
      "learning_rate": 0.00012340821586677623,
      "loss": 12.169,
      "step": 10111
    },
    {
      "epoch": 1.7682185792349727,
      "grad_norm": 0.7346455737165312,
      "learning_rate": 0.0001233944481801649,
      "loss": 11.8065,
      "step": 10112
    },
    {
      "epoch": 1.7683934426229508,
      "grad_norm": 0.7180389387923389,
      "learning_rate": 0.00012338068002442292,
      "loss": 11.9969,
      "step": 10113
    },
    {
      "epoch": 1.768568306010929,
      "grad_norm": 0.7089833218420452,
      "learning_rate": 0.00012336691139982662,
      "loss": 12.2153,
      "step": 10114
    },
    {
      "epoch": 1.768743169398907,
      "grad_norm": 0.6873665258603946,
      "learning_rate": 0.00012335314230665196,
      "loss": 12.3491,
      "step": 10115
    },
    {
      "epoch": 1.7689180327868852,
      "grad_norm": 0.6904367383890242,
      "learning_rate": 0.0001233393727451751,
      "loss": 12.2693,
      "step": 10116
    },
    {
      "epoch": 1.7690928961748633,
      "grad_norm": 0.7033725908035475,
      "learning_rate": 0.00012332560271567216,
      "loss": 11.9062,
      "step": 10117
    },
    {
      "epoch": 1.7692677595628417,
      "grad_norm": 0.6958474255205792,
      "learning_rate": 0.00012331183221841923,
      "loss": 12.4236,
      "step": 10118
    },
    {
      "epoch": 1.7694426229508196,
      "grad_norm": 0.7537697120600338,
      "learning_rate": 0.00012329806125369253,
      "loss": 12.2315,
      "step": 10119
    },
    {
      "epoch": 1.769617486338798,
      "grad_norm": 0.6539997124329141,
      "learning_rate": 0.00012328428982176813,
      "loss": 12.2943,
      "step": 10120
    },
    {
      "epoch": 1.7697923497267758,
      "grad_norm": 0.7001269817347595,
      "learning_rate": 0.00012327051792292227,
      "loss": 12.2182,
      "step": 10121
    },
    {
      "epoch": 1.7699672131147541,
      "grad_norm": 0.7338155682365639,
      "learning_rate": 0.00012325674555743106,
      "loss": 12.302,
      "step": 10122
    },
    {
      "epoch": 1.7701420765027323,
      "grad_norm": 0.7275532354553779,
      "learning_rate": 0.0001232429727255707,
      "loss": 12.1877,
      "step": 10123
    },
    {
      "epoch": 1.7703169398907104,
      "grad_norm": 0.8216181741157916,
      "learning_rate": 0.00012322919942761736,
      "loss": 12.3103,
      "step": 10124
    },
    {
      "epoch": 1.7704918032786885,
      "grad_norm": 0.7566851065159629,
      "learning_rate": 0.00012321542566384722,
      "loss": 12.5165,
      "step": 10125
    },
    {
      "epoch": 1.7706666666666666,
      "grad_norm": 0.7803660007896559,
      "learning_rate": 0.00012320165143453658,
      "loss": 12.1206,
      "step": 10126
    },
    {
      "epoch": 1.770841530054645,
      "grad_norm": 0.7328253775456597,
      "learning_rate": 0.00012318787673996153,
      "loss": 12.1988,
      "step": 10127
    },
    {
      "epoch": 1.7710163934426228,
      "grad_norm": 0.7007059515539635,
      "learning_rate": 0.0001231741015803984,
      "loss": 12.2225,
      "step": 10128
    },
    {
      "epoch": 1.7711912568306012,
      "grad_norm": 0.722387453234732,
      "learning_rate": 0.00012316032595612335,
      "loss": 12.1652,
      "step": 10129
    },
    {
      "epoch": 1.771366120218579,
      "grad_norm": 0.6948254709593936,
      "learning_rate": 0.00012314654986741265,
      "loss": 12.3215,
      "step": 10130
    },
    {
      "epoch": 1.7715409836065574,
      "grad_norm": 0.6935000576670025,
      "learning_rate": 0.00012313277331454256,
      "loss": 12.2649,
      "step": 10131
    },
    {
      "epoch": 1.7717158469945355,
      "grad_norm": 0.7098302028675375,
      "learning_rate": 0.00012311899629778935,
      "loss": 12.05,
      "step": 10132
    },
    {
      "epoch": 1.7718907103825137,
      "grad_norm": 0.6559312876647025,
      "learning_rate": 0.0001231052188174293,
      "loss": 11.9904,
      "step": 10133
    },
    {
      "epoch": 1.7720655737704918,
      "grad_norm": 0.6497622272161928,
      "learning_rate": 0.00012309144087373862,
      "loss": 12.3715,
      "step": 10134
    },
    {
      "epoch": 1.77224043715847,
      "grad_norm": 0.7022487886381181,
      "learning_rate": 0.00012307766246699367,
      "loss": 12.2486,
      "step": 10135
    },
    {
      "epoch": 1.7724153005464482,
      "grad_norm": 0.6476045682460227,
      "learning_rate": 0.00012306388359747074,
      "loss": 12.1351,
      "step": 10136
    },
    {
      "epoch": 1.7725901639344261,
      "grad_norm": 0.6491789422343002,
      "learning_rate": 0.00012305010426544614,
      "loss": 12.2879,
      "step": 10137
    },
    {
      "epoch": 1.7727650273224045,
      "grad_norm": 0.7784801887590417,
      "learning_rate": 0.00012303632447119615,
      "loss": 12.3465,
      "step": 10138
    },
    {
      "epoch": 1.7729398907103824,
      "grad_norm": 0.6797186596952196,
      "learning_rate": 0.00012302254421499718,
      "loss": 12.407,
      "step": 10139
    },
    {
      "epoch": 1.7731147540983607,
      "grad_norm": 0.6509747769132255,
      "learning_rate": 0.0001230087634971254,
      "loss": 12.1284,
      "step": 10140
    },
    {
      "epoch": 1.7732896174863388,
      "grad_norm": 0.6395875640902573,
      "learning_rate": 0.00012299498231785737,
      "loss": 12.3131,
      "step": 10141
    },
    {
      "epoch": 1.773464480874317,
      "grad_norm": 0.7191877031214092,
      "learning_rate": 0.00012298120067746928,
      "loss": 12.1819,
      "step": 10142
    },
    {
      "epoch": 1.773639344262295,
      "grad_norm": 0.6369903497473842,
      "learning_rate": 0.00012296741857623758,
      "loss": 11.8454,
      "step": 10143
    },
    {
      "epoch": 1.7738142076502732,
      "grad_norm": 0.6322263817090068,
      "learning_rate": 0.0001229536360144386,
      "loss": 12.2026,
      "step": 10144
    },
    {
      "epoch": 1.7739890710382513,
      "grad_norm": 0.623047664422909,
      "learning_rate": 0.00012293985299234876,
      "loss": 12.3571,
      "step": 10145
    },
    {
      "epoch": 1.7741639344262294,
      "grad_norm": 0.6606800820007682,
      "learning_rate": 0.00012292606951024447,
      "loss": 12.2533,
      "step": 10146
    },
    {
      "epoch": 1.7743387978142078,
      "grad_norm": 0.8491557268157746,
      "learning_rate": 0.00012291228556840202,
      "loss": 12.4044,
      "step": 10147
    },
    {
      "epoch": 1.7745136612021857,
      "grad_norm": 0.6446059415535625,
      "learning_rate": 0.00012289850116709795,
      "loss": 12.2052,
      "step": 10148
    },
    {
      "epoch": 1.774688524590164,
      "grad_norm": 0.6247056975853913,
      "learning_rate": 0.0001228847163066086,
      "loss": 12.1352,
      "step": 10149
    },
    {
      "epoch": 1.774863387978142,
      "grad_norm": 0.6532316291291381,
      "learning_rate": 0.0001228709309872104,
      "loss": 12.0842,
      "step": 10150
    },
    {
      "epoch": 1.7750382513661203,
      "grad_norm": 0.6818639060542652,
      "learning_rate": 0.00012285714520917986,
      "loss": 12.0824,
      "step": 10151
    },
    {
      "epoch": 1.7752131147540984,
      "grad_norm": 0.7244898327960612,
      "learning_rate": 0.00012284335897279333,
      "loss": 12.2208,
      "step": 10152
    },
    {
      "epoch": 1.7753879781420765,
      "grad_norm": 0.6460023758785526,
      "learning_rate": 0.00012282957227832735,
      "loss": 12.1723,
      "step": 10153
    },
    {
      "epoch": 1.7755628415300546,
      "grad_norm": 0.6477152311224083,
      "learning_rate": 0.00012281578512605832,
      "loss": 12.1435,
      "step": 10154
    },
    {
      "epoch": 1.7757377049180327,
      "grad_norm": 0.724212357282621,
      "learning_rate": 0.00012280199751626278,
      "loss": 12.5733,
      "step": 10155
    },
    {
      "epoch": 1.775912568306011,
      "grad_norm": 0.6766046677629733,
      "learning_rate": 0.00012278820944921714,
      "loss": 11.9651,
      "step": 10156
    },
    {
      "epoch": 1.776087431693989,
      "grad_norm": 0.756626262121075,
      "learning_rate": 0.00012277442092519796,
      "loss": 12.4018,
      "step": 10157
    },
    {
      "epoch": 1.7762622950819673,
      "grad_norm": 0.6727286577600449,
      "learning_rate": 0.0001227606319444817,
      "loss": 12.431,
      "step": 10158
    },
    {
      "epoch": 1.7764371584699452,
      "grad_norm": 0.7112348903805703,
      "learning_rate": 0.00012274684250734486,
      "loss": 12.4516,
      "step": 10159
    },
    {
      "epoch": 1.7766120218579236,
      "grad_norm": 0.7262269753529151,
      "learning_rate": 0.00012273305261406404,
      "loss": 12.2223,
      "step": 10160
    },
    {
      "epoch": 1.7767868852459017,
      "grad_norm": 0.6517077792853238,
      "learning_rate": 0.00012271926226491568,
      "loss": 12.3942,
      "step": 10161
    },
    {
      "epoch": 1.7769617486338798,
      "grad_norm": 0.7462912060150996,
      "learning_rate": 0.0001227054714601764,
      "loss": 12.4235,
      "step": 10162
    },
    {
      "epoch": 1.777136612021858,
      "grad_norm": 0.7534464589270172,
      "learning_rate": 0.00012269168020012265,
      "loss": 12.2747,
      "step": 10163
    },
    {
      "epoch": 1.777311475409836,
      "grad_norm": 0.6725475270785992,
      "learning_rate": 0.00012267788848503106,
      "loss": 12.2364,
      "step": 10164
    },
    {
      "epoch": 1.7774863387978144,
      "grad_norm": 0.7956325618441429,
      "learning_rate": 0.00012266409631517817,
      "loss": 12.3577,
      "step": 10165
    },
    {
      "epoch": 1.7776612021857923,
      "grad_norm": 0.6816748328754807,
      "learning_rate": 0.00012265030369084057,
      "loss": 12.3688,
      "step": 10166
    },
    {
      "epoch": 1.7778360655737706,
      "grad_norm": 0.7614405744519244,
      "learning_rate": 0.00012263651061229482,
      "loss": 12.2096,
      "step": 10167
    },
    {
      "epoch": 1.7780109289617485,
      "grad_norm": 0.7705927810099074,
      "learning_rate": 0.00012262271707981753,
      "loss": 12.4708,
      "step": 10168
    },
    {
      "epoch": 1.7781857923497268,
      "grad_norm": 0.7417245588808346,
      "learning_rate": 0.00012260892309368533,
      "loss": 12.3991,
      "step": 10169
    },
    {
      "epoch": 1.778360655737705,
      "grad_norm": 0.6280458723883752,
      "learning_rate": 0.00012259512865417477,
      "loss": 12.3714,
      "step": 10170
    },
    {
      "epoch": 1.778535519125683,
      "grad_norm": 0.7949792592061591,
      "learning_rate": 0.00012258133376156252,
      "loss": 12.0658,
      "step": 10171
    },
    {
      "epoch": 1.7787103825136612,
      "grad_norm": 0.6672369350486852,
      "learning_rate": 0.0001225675384161252,
      "loss": 12.4233,
      "step": 10172
    },
    {
      "epoch": 1.7788852459016393,
      "grad_norm": 0.8351633200779488,
      "learning_rate": 0.00012255374261813944,
      "loss": 12.4985,
      "step": 10173
    },
    {
      "epoch": 1.7790601092896174,
      "grad_norm": 0.6470201150004566,
      "learning_rate": 0.0001225399463678819,
      "loss": 12.2597,
      "step": 10174
    },
    {
      "epoch": 1.7792349726775956,
      "grad_norm": 0.7981548274250666,
      "learning_rate": 0.0001225261496656292,
      "loss": 12.3498,
      "step": 10175
    },
    {
      "epoch": 1.779409836065574,
      "grad_norm": 0.710191764497291,
      "learning_rate": 0.00012251235251165804,
      "loss": 12.3408,
      "step": 10176
    },
    {
      "epoch": 1.7795846994535518,
      "grad_norm": 0.8122300357210863,
      "learning_rate": 0.00012249855490624508,
      "loss": 12.2469,
      "step": 10177
    },
    {
      "epoch": 1.7797595628415301,
      "grad_norm": 0.726663976787163,
      "learning_rate": 0.00012248475684966705,
      "loss": 12.2146,
      "step": 10178
    },
    {
      "epoch": 1.779934426229508,
      "grad_norm": 0.7972344934435626,
      "learning_rate": 0.00012247095834220058,
      "loss": 12.4667,
      "step": 10179
    },
    {
      "epoch": 1.7801092896174864,
      "grad_norm": 0.8397016783655129,
      "learning_rate": 0.00012245715938412243,
      "loss": 12.3542,
      "step": 10180
    },
    {
      "epoch": 1.7802841530054645,
      "grad_norm": 0.6518096850253905,
      "learning_rate": 0.00012244335997570925,
      "loss": 12.1601,
      "step": 10181
    },
    {
      "epoch": 1.7804590163934426,
      "grad_norm": 0.9626257579486044,
      "learning_rate": 0.00012242956011723782,
      "loss": 12.3352,
      "step": 10182
    },
    {
      "epoch": 1.7806338797814207,
      "grad_norm": 0.6863701889421228,
      "learning_rate": 0.00012241575980898484,
      "loss": 12.168,
      "step": 10183
    },
    {
      "epoch": 1.7808087431693989,
      "grad_norm": 0.8751944459090073,
      "learning_rate": 0.00012240195905122703,
      "loss": 12.3832,
      "step": 10184
    },
    {
      "epoch": 1.7809836065573772,
      "grad_norm": 0.7488724211251084,
      "learning_rate": 0.00012238815784424117,
      "loss": 12.3395,
      "step": 10185
    },
    {
      "epoch": 1.781158469945355,
      "grad_norm": 0.7768420953735555,
      "learning_rate": 0.000122374356188304,
      "loss": 12.4894,
      "step": 10186
    },
    {
      "epoch": 1.7813333333333334,
      "grad_norm": 0.7630928816986834,
      "learning_rate": 0.00012236055408369233,
      "loss": 12.2174,
      "step": 10187
    },
    {
      "epoch": 1.7815081967213113,
      "grad_norm": 0.7530750261413385,
      "learning_rate": 0.00012234675153068283,
      "loss": 12.3476,
      "step": 10188
    },
    {
      "epoch": 1.7816830601092897,
      "grad_norm": 0.7836368023115985,
      "learning_rate": 0.00012233294852955238,
      "loss": 12.4156,
      "step": 10189
    },
    {
      "epoch": 1.7818579234972678,
      "grad_norm": 0.7035288290023877,
      "learning_rate": 0.00012231914508057772,
      "loss": 12.1547,
      "step": 10190
    },
    {
      "epoch": 1.782032786885246,
      "grad_norm": 0.8002349299528183,
      "learning_rate": 0.00012230534118403568,
      "loss": 12.382,
      "step": 10191
    },
    {
      "epoch": 1.782207650273224,
      "grad_norm": 0.6504901204616128,
      "learning_rate": 0.00012229153684020308,
      "loss": 12.457,
      "step": 10192
    },
    {
      "epoch": 1.7823825136612022,
      "grad_norm": 0.7527196410531372,
      "learning_rate": 0.00012227773204935672,
      "loss": 12.63,
      "step": 10193
    },
    {
      "epoch": 1.7825573770491805,
      "grad_norm": 0.7709084625860794,
      "learning_rate": 0.00012226392681177344,
      "loss": 12.065,
      "step": 10194
    },
    {
      "epoch": 1.7827322404371584,
      "grad_norm": 0.7369394218865435,
      "learning_rate": 0.00012225012112773001,
      "loss": 12.2104,
      "step": 10195
    },
    {
      "epoch": 1.7829071038251367,
      "grad_norm": 0.7018315475459501,
      "learning_rate": 0.0001222363149975034,
      "loss": 12.1519,
      "step": 10196
    },
    {
      "epoch": 1.7830819672131146,
      "grad_norm": 0.7050393429860239,
      "learning_rate": 0.00012222250842137036,
      "loss": 12.5221,
      "step": 10197
    },
    {
      "epoch": 1.783256830601093,
      "grad_norm": 0.6889284585823381,
      "learning_rate": 0.00012220870139960783,
      "loss": 12.3465,
      "step": 10198
    },
    {
      "epoch": 1.783431693989071,
      "grad_norm": 0.673514201724095,
      "learning_rate": 0.00012219489393249262,
      "loss": 12.4357,
      "step": 10199
    },
    {
      "epoch": 1.7836065573770492,
      "grad_norm": 0.7122552490747318,
      "learning_rate": 0.00012218108602030163,
      "loss": 12.3743,
      "step": 10200
    },
    {
      "epoch": 1.7837814207650273,
      "grad_norm": 0.6942952721580313,
      "learning_rate": 0.0001221672776633118,
      "loss": 12.3725,
      "step": 10201
    },
    {
      "epoch": 1.7839562841530054,
      "grad_norm": 0.681961395511829,
      "learning_rate": 0.00012215346886179995,
      "loss": 12.1109,
      "step": 10202
    },
    {
      "epoch": 1.7841311475409836,
      "grad_norm": 0.7242454251800965,
      "learning_rate": 0.00012213965961604308,
      "loss": 12.1854,
      "step": 10203
    },
    {
      "epoch": 1.7843060109289617,
      "grad_norm": 0.7648149977489737,
      "learning_rate": 0.000122125849926318,
      "loss": 12.333,
      "step": 10204
    },
    {
      "epoch": 1.78448087431694,
      "grad_norm": 0.6387513427061517,
      "learning_rate": 0.00012211203979290173,
      "loss": 12.1381,
      "step": 10205
    },
    {
      "epoch": 1.784655737704918,
      "grad_norm": 0.6705127428506077,
      "learning_rate": 0.00012209822921607115,
      "loss": 12.2805,
      "step": 10206
    },
    {
      "epoch": 1.7848306010928963,
      "grad_norm": 0.6812188117727772,
      "learning_rate": 0.00012208441819610324,
      "loss": 12.5182,
      "step": 10207
    },
    {
      "epoch": 1.7850054644808742,
      "grad_norm": 0.8018622332681082,
      "learning_rate": 0.00012207060673327492,
      "loss": 12.3673,
      "step": 10208
    },
    {
      "epoch": 1.7851803278688525,
      "grad_norm": 0.6580287252370319,
      "learning_rate": 0.00012205679482786317,
      "loss": 12.1853,
      "step": 10209
    },
    {
      "epoch": 1.7853551912568306,
      "grad_norm": 0.7545164281562918,
      "learning_rate": 0.00012204298248014499,
      "loss": 12.335,
      "step": 10210
    },
    {
      "epoch": 1.7855300546448087,
      "grad_norm": 0.6602542610220565,
      "learning_rate": 0.00012202916969039728,
      "loss": 12.3939,
      "step": 10211
    },
    {
      "epoch": 1.7857049180327869,
      "grad_norm": 0.6995742505732353,
      "learning_rate": 0.00012201535645889712,
      "loss": 12.0542,
      "step": 10212
    },
    {
      "epoch": 1.785879781420765,
      "grad_norm": 0.6489213574243243,
      "learning_rate": 0.00012200154278592146,
      "loss": 12.2713,
      "step": 10213
    },
    {
      "epoch": 1.7860546448087433,
      "grad_norm": 0.7004037066787561,
      "learning_rate": 0.0001219877286717473,
      "loss": 12.0571,
      "step": 10214
    },
    {
      "epoch": 1.7862295081967212,
      "grad_norm": 0.6709550680803584,
      "learning_rate": 0.00012197391411665168,
      "loss": 12.4643,
      "step": 10215
    },
    {
      "epoch": 1.7864043715846996,
      "grad_norm": 0.6464182496195225,
      "learning_rate": 0.00012196009912091162,
      "loss": 12.2329,
      "step": 10216
    },
    {
      "epoch": 1.7865792349726775,
      "grad_norm": 0.7215888524903372,
      "learning_rate": 0.00012194628368480412,
      "loss": 12.1418,
      "step": 10217
    },
    {
      "epoch": 1.7867540983606558,
      "grad_norm": 0.6475150437357547,
      "learning_rate": 0.00012193246780860628,
      "loss": 12.2142,
      "step": 10218
    },
    {
      "epoch": 1.786928961748634,
      "grad_norm": 0.6863177487290107,
      "learning_rate": 0.0001219186514925951,
      "loss": 12.246,
      "step": 10219
    },
    {
      "epoch": 1.787103825136612,
      "grad_norm": 0.6609117898933433,
      "learning_rate": 0.00012190483473704766,
      "loss": 12.2397,
      "step": 10220
    },
    {
      "epoch": 1.7872786885245902,
      "grad_norm": 0.7944468333102841,
      "learning_rate": 0.00012189101754224102,
      "loss": 12.0642,
      "step": 10221
    },
    {
      "epoch": 1.7874535519125683,
      "grad_norm": 0.7027976832797528,
      "learning_rate": 0.00012187719990845227,
      "loss": 12.2899,
      "step": 10222
    },
    {
      "epoch": 1.7876284153005466,
      "grad_norm": 0.7353425797277033,
      "learning_rate": 0.00012186338183595849,
      "loss": 12.325,
      "step": 10223
    },
    {
      "epoch": 1.7878032786885245,
      "grad_norm": 0.6853939377446976,
      "learning_rate": 0.00012184956332503678,
      "loss": 12.4061,
      "step": 10224
    },
    {
      "epoch": 1.7879781420765029,
      "grad_norm": 0.689698495628008,
      "learning_rate": 0.00012183574437596423,
      "loss": 12.3007,
      "step": 10225
    },
    {
      "epoch": 1.7881530054644807,
      "grad_norm": 0.6250660807055155,
      "learning_rate": 0.00012182192498901797,
      "loss": 12.1105,
      "step": 10226
    },
    {
      "epoch": 1.788327868852459,
      "grad_norm": 0.6649092061921982,
      "learning_rate": 0.00012180810516447512,
      "loss": 12.458,
      "step": 10227
    },
    {
      "epoch": 1.7885027322404372,
      "grad_norm": 0.6805803992177746,
      "learning_rate": 0.00012179428490261278,
      "loss": 12.5307,
      "step": 10228
    },
    {
      "epoch": 1.7886775956284153,
      "grad_norm": 0.6573333292658691,
      "learning_rate": 0.00012178046420370812,
      "loss": 12.1098,
      "step": 10229
    },
    {
      "epoch": 1.7888524590163934,
      "grad_norm": 0.6715390626300707,
      "learning_rate": 0.00012176664306803829,
      "loss": 12.2461,
      "step": 10230
    },
    {
      "epoch": 1.7890273224043716,
      "grad_norm": 0.6563208177267794,
      "learning_rate": 0.0001217528214958804,
      "loss": 12.3099,
      "step": 10231
    },
    {
      "epoch": 1.7892021857923497,
      "grad_norm": 0.6905660686114209,
      "learning_rate": 0.00012173899948751168,
      "loss": 12.2926,
      "step": 10232
    },
    {
      "epoch": 1.7893770491803278,
      "grad_norm": 0.7609991443714046,
      "learning_rate": 0.00012172517704320927,
      "loss": 12.4837,
      "step": 10233
    },
    {
      "epoch": 1.7895519125683061,
      "grad_norm": 0.6954544798075702,
      "learning_rate": 0.00012171135416325035,
      "loss": 12.1362,
      "step": 10234
    },
    {
      "epoch": 1.789726775956284,
      "grad_norm": 0.7251665035995942,
      "learning_rate": 0.00012169753084791213,
      "loss": 12.4049,
      "step": 10235
    },
    {
      "epoch": 1.7899016393442624,
      "grad_norm": 0.6405676711395185,
      "learning_rate": 0.00012168370709747177,
      "loss": 12.2539,
      "step": 10236
    },
    {
      "epoch": 1.7900765027322403,
      "grad_norm": 0.6528174513358344,
      "learning_rate": 0.00012166988291220654,
      "loss": 11.9791,
      "step": 10237
    },
    {
      "epoch": 1.7902513661202186,
      "grad_norm": 0.7503088147131741,
      "learning_rate": 0.00012165605829239357,
      "loss": 12.2466,
      "step": 10238
    },
    {
      "epoch": 1.7904262295081967,
      "grad_norm": 0.8202019114940517,
      "learning_rate": 0.00012164223323831016,
      "loss": 12.2408,
      "step": 10239
    },
    {
      "epoch": 1.7906010928961749,
      "grad_norm": 0.6966744361101842,
      "learning_rate": 0.00012162840775023354,
      "loss": 12.349,
      "step": 10240
    },
    {
      "epoch": 1.790775956284153,
      "grad_norm": 0.6270603342479507,
      "learning_rate": 0.00012161458182844094,
      "loss": 12.1573,
      "step": 10241
    },
    {
      "epoch": 1.790950819672131,
      "grad_norm": 0.6976466312868066,
      "learning_rate": 0.00012160075547320961,
      "loss": 12.3352,
      "step": 10242
    },
    {
      "epoch": 1.7911256830601094,
      "grad_norm": 0.6936027762080507,
      "learning_rate": 0.00012158692868481678,
      "loss": 12.2976,
      "step": 10243
    },
    {
      "epoch": 1.7913005464480873,
      "grad_norm": 0.7495655980737852,
      "learning_rate": 0.00012157310146353977,
      "loss": 12.412,
      "step": 10244
    },
    {
      "epoch": 1.7914754098360657,
      "grad_norm": 0.8999597050928656,
      "learning_rate": 0.00012155927380965582,
      "loss": 12.3647,
      "step": 10245
    },
    {
      "epoch": 1.7916502732240436,
      "grad_norm": 0.6767014780698555,
      "learning_rate": 0.00012154544572344226,
      "loss": 12.2936,
      "step": 10246
    },
    {
      "epoch": 1.791825136612022,
      "grad_norm": 0.681690966524557,
      "learning_rate": 0.00012153161720517636,
      "loss": 12.415,
      "step": 10247
    },
    {
      "epoch": 1.792,
      "grad_norm": 0.7354389297838164,
      "learning_rate": 0.0001215177882551354,
      "loss": 12.2553,
      "step": 10248
    },
    {
      "epoch": 1.7921748633879782,
      "grad_norm": 0.6524238414056033,
      "learning_rate": 0.00012150395887359675,
      "loss": 12.019,
      "step": 10249
    },
    {
      "epoch": 1.7923497267759563,
      "grad_norm": 0.8888449180444176,
      "learning_rate": 0.00012149012906083765,
      "loss": 12.3496,
      "step": 10250
    },
    {
      "epoch": 1.7925245901639344,
      "grad_norm": 0.6619520556519519,
      "learning_rate": 0.00012147629881713553,
      "loss": 12.2975,
      "step": 10251
    },
    {
      "epoch": 1.7926994535519125,
      "grad_norm": 0.6864589688948882,
      "learning_rate": 0.00012146246814276763,
      "loss": 12.0139,
      "step": 10252
    },
    {
      "epoch": 1.7928743169398906,
      "grad_norm": 0.8027758442239861,
      "learning_rate": 0.00012144863703801139,
      "loss": 12.2305,
      "step": 10253
    },
    {
      "epoch": 1.793049180327869,
      "grad_norm": 0.8192250536466478,
      "learning_rate": 0.0001214348055031441,
      "loss": 11.9993,
      "step": 10254
    },
    {
      "epoch": 1.7932240437158469,
      "grad_norm": 0.8355948364121973,
      "learning_rate": 0.00012142097353844315,
      "loss": 12.5069,
      "step": 10255
    },
    {
      "epoch": 1.7933989071038252,
      "grad_norm": 0.6960927738690609,
      "learning_rate": 0.00012140714114418592,
      "loss": 12.4657,
      "step": 10256
    },
    {
      "epoch": 1.7935737704918033,
      "grad_norm": 0.8444097246427461,
      "learning_rate": 0.00012139330832064974,
      "loss": 12.5098,
      "step": 10257
    },
    {
      "epoch": 1.7937486338797815,
      "grad_norm": 0.6858882917943764,
      "learning_rate": 0.00012137947506811209,
      "loss": 12.3373,
      "step": 10258
    },
    {
      "epoch": 1.7939234972677596,
      "grad_norm": 0.841731113985785,
      "learning_rate": 0.00012136564138685029,
      "loss": 12.2107,
      "step": 10259
    },
    {
      "epoch": 1.7940983606557377,
      "grad_norm": 0.6504529575784019,
      "learning_rate": 0.00012135180727714179,
      "loss": 12.3385,
      "step": 10260
    },
    {
      "epoch": 1.7942732240437158,
      "grad_norm": 0.6957899874952456,
      "learning_rate": 0.00012133797273926401,
      "loss": 12.3115,
      "step": 10261
    },
    {
      "epoch": 1.794448087431694,
      "grad_norm": 0.8672669890760916,
      "learning_rate": 0.00012132413777349433,
      "loss": 12.2557,
      "step": 10262
    },
    {
      "epoch": 1.7946229508196723,
      "grad_norm": 0.66055915852558,
      "learning_rate": 0.00012131030238011025,
      "loss": 12.3192,
      "step": 10263
    },
    {
      "epoch": 1.7947978142076502,
      "grad_norm": 0.7736776533611411,
      "learning_rate": 0.00012129646655938916,
      "loss": 12.3982,
      "step": 10264
    },
    {
      "epoch": 1.7949726775956285,
      "grad_norm": 0.6252568025297726,
      "learning_rate": 0.00012128263031160851,
      "loss": 12.367,
      "step": 10265
    },
    {
      "epoch": 1.7951475409836064,
      "grad_norm": 0.7278247108796613,
      "learning_rate": 0.0001212687936370458,
      "loss": 12.3438,
      "step": 10266
    },
    {
      "epoch": 1.7953224043715847,
      "grad_norm": 0.6888014560514235,
      "learning_rate": 0.00012125495653597846,
      "loss": 11.9398,
      "step": 10267
    },
    {
      "epoch": 1.7954972677595629,
      "grad_norm": 0.6832005890810957,
      "learning_rate": 0.00012124111900868398,
      "loss": 12.4122,
      "step": 10268
    },
    {
      "epoch": 1.795672131147541,
      "grad_norm": 0.7148520050234691,
      "learning_rate": 0.00012122728105543985,
      "loss": 12.4104,
      "step": 10269
    },
    {
      "epoch": 1.795846994535519,
      "grad_norm": 0.7238831887237575,
      "learning_rate": 0.00012121344267652356,
      "loss": 12.3465,
      "step": 10270
    },
    {
      "epoch": 1.7960218579234972,
      "grad_norm": 0.6654212840674028,
      "learning_rate": 0.00012119960387221261,
      "loss": 12.4627,
      "step": 10271
    },
    {
      "epoch": 1.7961967213114756,
      "grad_norm": 0.6339043885242167,
      "learning_rate": 0.0001211857646427845,
      "loss": 12.3473,
      "step": 10272
    },
    {
      "epoch": 1.7963715846994535,
      "grad_norm": 0.6265290785035575,
      "learning_rate": 0.00012117192498851677,
      "loss": 12.0218,
      "step": 10273
    },
    {
      "epoch": 1.7965464480874318,
      "grad_norm": 0.6803342838318234,
      "learning_rate": 0.00012115808490968694,
      "loss": 12.208,
      "step": 10274
    },
    {
      "epoch": 1.7967213114754097,
      "grad_norm": 0.6559304189856264,
      "learning_rate": 0.00012114424440657255,
      "loss": 12.0907,
      "step": 10275
    },
    {
      "epoch": 1.796896174863388,
      "grad_norm": 0.6691877260797059,
      "learning_rate": 0.00012113040347945113,
      "loss": 12.4312,
      "step": 10276
    },
    {
      "epoch": 1.7970710382513662,
      "grad_norm": 0.6887058837412173,
      "learning_rate": 0.00012111656212860023,
      "loss": 12.168,
      "step": 10277
    },
    {
      "epoch": 1.7972459016393443,
      "grad_norm": 0.6456903398042135,
      "learning_rate": 0.00012110272035429743,
      "loss": 12.078,
      "step": 10278
    },
    {
      "epoch": 1.7974207650273224,
      "grad_norm": 0.7505778376083985,
      "learning_rate": 0.00012108887815682031,
      "loss": 12.4737,
      "step": 10279
    },
    {
      "epoch": 1.7975956284153005,
      "grad_norm": 0.7047000411381014,
      "learning_rate": 0.00012107503553644641,
      "loss": 12.1798,
      "step": 10280
    },
    {
      "epoch": 1.7977704918032786,
      "grad_norm": 0.68821322664469,
      "learning_rate": 0.00012106119249345336,
      "loss": 12.3733,
      "step": 10281
    },
    {
      "epoch": 1.7979453551912568,
      "grad_norm": 0.7267539469086162,
      "learning_rate": 0.00012104734902811871,
      "loss": 12.0107,
      "step": 10282
    },
    {
      "epoch": 1.798120218579235,
      "grad_norm": 0.6270513891115582,
      "learning_rate": 0.00012103350514072011,
      "loss": 12.2627,
      "step": 10283
    },
    {
      "epoch": 1.798295081967213,
      "grad_norm": 0.6932317038447717,
      "learning_rate": 0.00012101966083153514,
      "loss": 12.287,
      "step": 10284
    },
    {
      "epoch": 1.7984699453551913,
      "grad_norm": 0.7682802571019307,
      "learning_rate": 0.00012100581610084143,
      "loss": 12.2175,
      "step": 10285
    },
    {
      "epoch": 1.7986448087431695,
      "grad_norm": 0.7607852237987005,
      "learning_rate": 0.00012099197094891659,
      "loss": 12.222,
      "step": 10286
    },
    {
      "epoch": 1.7988196721311476,
      "grad_norm": 0.6748025163465688,
      "learning_rate": 0.00012097812537603829,
      "loss": 12.4669,
      "step": 10287
    },
    {
      "epoch": 1.7989945355191257,
      "grad_norm": 0.6834378902049861,
      "learning_rate": 0.0001209642793824842,
      "loss": 12.1874,
      "step": 10288
    },
    {
      "epoch": 1.7991693989071038,
      "grad_norm": 0.7087195331321526,
      "learning_rate": 0.0001209504329685319,
      "loss": 12.1675,
      "step": 10289
    },
    {
      "epoch": 1.799344262295082,
      "grad_norm": 0.7518461468578629,
      "learning_rate": 0.00012093658613445913,
      "loss": 12.2376,
      "step": 10290
    },
    {
      "epoch": 1.79951912568306,
      "grad_norm": 0.7519431300307702,
      "learning_rate": 0.00012092273888054348,
      "loss": 12.5598,
      "step": 10291
    },
    {
      "epoch": 1.7996939890710384,
      "grad_norm": 0.7007613918425116,
      "learning_rate": 0.00012090889120706273,
      "loss": 12.2156,
      "step": 10292
    },
    {
      "epoch": 1.7998688524590163,
      "grad_norm": 0.7099894792226743,
      "learning_rate": 0.00012089504311429444,
      "loss": 12.2957,
      "step": 10293
    },
    {
      "epoch": 1.8000437158469946,
      "grad_norm": 0.6791907763744367,
      "learning_rate": 0.00012088119460251643,
      "loss": 12.3572,
      "step": 10294
    },
    {
      "epoch": 1.8002185792349725,
      "grad_norm": 0.679145735704672,
      "learning_rate": 0.00012086734567200635,
      "loss": 12.3421,
      "step": 10295
    },
    {
      "epoch": 1.8003934426229509,
      "grad_norm": 0.7012157634011296,
      "learning_rate": 0.0001208534963230419,
      "loss": 12.445,
      "step": 10296
    },
    {
      "epoch": 1.800568306010929,
      "grad_norm": 0.679990526191735,
      "learning_rate": 0.00012083964655590086,
      "loss": 12.0649,
      "step": 10297
    },
    {
      "epoch": 1.800743169398907,
      "grad_norm": 0.6867684103469326,
      "learning_rate": 0.00012082579637086087,
      "loss": 12.3462,
      "step": 10298
    },
    {
      "epoch": 1.8009180327868852,
      "grad_norm": 0.6527311573722426,
      "learning_rate": 0.00012081194576819974,
      "loss": 12.3492,
      "step": 10299
    },
    {
      "epoch": 1.8010928961748633,
      "grad_norm": 0.8526847178850415,
      "learning_rate": 0.00012079809474819519,
      "loss": 12.4553,
      "step": 10300
    },
    {
      "epoch": 1.8012677595628417,
      "grad_norm": 0.7483607813155417,
      "learning_rate": 0.000120784243311125,
      "loss": 12.3181,
      "step": 10301
    },
    {
      "epoch": 1.8014426229508196,
      "grad_norm": 0.9853733214985698,
      "learning_rate": 0.00012077039145726692,
      "loss": 12.181,
      "step": 10302
    },
    {
      "epoch": 1.801617486338798,
      "grad_norm": 0.728031595821851,
      "learning_rate": 0.00012075653918689868,
      "loss": 12.4576,
      "step": 10303
    },
    {
      "epoch": 1.8017923497267758,
      "grad_norm": 0.7799533415063791,
      "learning_rate": 0.00012074268650029811,
      "loss": 12.1143,
      "step": 10304
    },
    {
      "epoch": 1.8019672131147542,
      "grad_norm": 0.7395950780142942,
      "learning_rate": 0.00012072883339774299,
      "loss": 12.3154,
      "step": 10305
    },
    {
      "epoch": 1.8021420765027323,
      "grad_norm": 0.7380583607699489,
      "learning_rate": 0.00012071497987951115,
      "loss": 12.2368,
      "step": 10306
    },
    {
      "epoch": 1.8023169398907104,
      "grad_norm": 0.8542395174822731,
      "learning_rate": 0.00012070112594588032,
      "loss": 12.4487,
      "step": 10307
    },
    {
      "epoch": 1.8024918032786885,
      "grad_norm": 0.6705326565868831,
      "learning_rate": 0.00012068727159712838,
      "loss": 12.158,
      "step": 10308
    },
    {
      "epoch": 1.8026666666666666,
      "grad_norm": 0.7485880602109327,
      "learning_rate": 0.0001206734168335331,
      "loss": 12.3334,
      "step": 10309
    },
    {
      "epoch": 1.8028415300546448,
      "grad_norm": 0.7029309204375063,
      "learning_rate": 0.00012065956165537236,
      "loss": 12.2653,
      "step": 10310
    },
    {
      "epoch": 1.8030163934426229,
      "grad_norm": 0.7937136456989194,
      "learning_rate": 0.00012064570606292396,
      "loss": 12.2884,
      "step": 10311
    },
    {
      "epoch": 1.8031912568306012,
      "grad_norm": 0.760914483130921,
      "learning_rate": 0.00012063185005646576,
      "loss": 11.9125,
      "step": 10312
    },
    {
      "epoch": 1.8033661202185791,
      "grad_norm": 0.6974710056051386,
      "learning_rate": 0.00012061799363627564,
      "loss": 12.1701,
      "step": 10313
    },
    {
      "epoch": 1.8035409836065575,
      "grad_norm": 0.7733496483673491,
      "learning_rate": 0.00012060413680263142,
      "loss": 12.3982,
      "step": 10314
    },
    {
      "epoch": 1.8037158469945354,
      "grad_norm": 0.6907787009690699,
      "learning_rate": 0.00012059027955581099,
      "loss": 12.2125,
      "step": 10315
    },
    {
      "epoch": 1.8038907103825137,
      "grad_norm": 0.7766152732986601,
      "learning_rate": 0.00012057642189609226,
      "loss": 12.2628,
      "step": 10316
    },
    {
      "epoch": 1.8040655737704918,
      "grad_norm": 0.6706168590348005,
      "learning_rate": 0.00012056256382375308,
      "loss": 12.2464,
      "step": 10317
    },
    {
      "epoch": 1.80424043715847,
      "grad_norm": 0.6848268398766816,
      "learning_rate": 0.00012054870533907134,
      "loss": 12.1105,
      "step": 10318
    },
    {
      "epoch": 1.804415300546448,
      "grad_norm": 0.6533131179933588,
      "learning_rate": 0.000120534846442325,
      "loss": 12.3754,
      "step": 10319
    },
    {
      "epoch": 1.8045901639344262,
      "grad_norm": 0.7496482502416443,
      "learning_rate": 0.00012052098713379192,
      "loss": 12.5389,
      "step": 10320
    },
    {
      "epoch": 1.8047650273224045,
      "grad_norm": 0.6844789126682581,
      "learning_rate": 0.00012050712741375003,
      "loss": 12.1253,
      "step": 10321
    },
    {
      "epoch": 1.8049398907103824,
      "grad_norm": 0.6492091181279697,
      "learning_rate": 0.00012049326728247729,
      "loss": 12.195,
      "step": 10322
    },
    {
      "epoch": 1.8051147540983608,
      "grad_norm": 0.6442259095942419,
      "learning_rate": 0.00012047940674025162,
      "loss": 12.3358,
      "step": 10323
    },
    {
      "epoch": 1.8052896174863386,
      "grad_norm": 0.6885932242088919,
      "learning_rate": 0.00012046554578735096,
      "loss": 12.2335,
      "step": 10324
    },
    {
      "epoch": 1.805464480874317,
      "grad_norm": 0.716562589468281,
      "learning_rate": 0.00012045168442405326,
      "loss": 12.2215,
      "step": 10325
    },
    {
      "epoch": 1.805639344262295,
      "grad_norm": 0.7100603232486686,
      "learning_rate": 0.0001204378226506365,
      "loss": 12.2862,
      "step": 10326
    },
    {
      "epoch": 1.8058142076502732,
      "grad_norm": 0.8026678697939291,
      "learning_rate": 0.00012042396046737865,
      "loss": 12.3798,
      "step": 10327
    },
    {
      "epoch": 1.8059890710382513,
      "grad_norm": 0.6479430901542572,
      "learning_rate": 0.00012041009787455766,
      "loss": 12.0121,
      "step": 10328
    },
    {
      "epoch": 1.8061639344262295,
      "grad_norm": 0.6678386054554637,
      "learning_rate": 0.00012039623487245156,
      "loss": 12.4679,
      "step": 10329
    },
    {
      "epoch": 1.8063387978142078,
      "grad_norm": 0.7529663664481676,
      "learning_rate": 0.0001203823714613383,
      "loss": 12.2541,
      "step": 10330
    },
    {
      "epoch": 1.8065136612021857,
      "grad_norm": 0.6347465182205172,
      "learning_rate": 0.00012036850764149596,
      "loss": 12.2154,
      "step": 10331
    },
    {
      "epoch": 1.806688524590164,
      "grad_norm": 0.7479691448266044,
      "learning_rate": 0.00012035464341320246,
      "loss": 12.0721,
      "step": 10332
    },
    {
      "epoch": 1.806863387978142,
      "grad_norm": 0.7531424666280535,
      "learning_rate": 0.0001203407787767359,
      "loss": 12.2852,
      "step": 10333
    },
    {
      "epoch": 1.8070382513661203,
      "grad_norm": 0.5899383927148004,
      "learning_rate": 0.00012032691373237424,
      "loss": 12.2748,
      "step": 10334
    },
    {
      "epoch": 1.8072131147540984,
      "grad_norm": 0.7629172984363051,
      "learning_rate": 0.00012031304828039554,
      "loss": 12.2135,
      "step": 10335
    },
    {
      "epoch": 1.8073879781420765,
      "grad_norm": 0.6154048327311151,
      "learning_rate": 0.00012029918242107787,
      "loss": 12.2816,
      "step": 10336
    },
    {
      "epoch": 1.8075628415300546,
      "grad_norm": 0.6879483356014506,
      "learning_rate": 0.00012028531615469927,
      "loss": 12.2854,
      "step": 10337
    },
    {
      "epoch": 1.8077377049180328,
      "grad_norm": 0.7667851439566601,
      "learning_rate": 0.00012027144948153779,
      "loss": 12.5241,
      "step": 10338
    },
    {
      "epoch": 1.8079125683060109,
      "grad_norm": 0.7023571913238392,
      "learning_rate": 0.0001202575824018715,
      "loss": 12.2487,
      "step": 10339
    },
    {
      "epoch": 1.808087431693989,
      "grad_norm": 0.6404253698767475,
      "learning_rate": 0.0001202437149159785,
      "loss": 12.1306,
      "step": 10340
    },
    {
      "epoch": 1.8082622950819673,
      "grad_norm": 0.6488031830316692,
      "learning_rate": 0.00012022984702413684,
      "loss": 12.2936,
      "step": 10341
    },
    {
      "epoch": 1.8084371584699452,
      "grad_norm": 0.6660093745397535,
      "learning_rate": 0.00012021597872662466,
      "loss": 12.186,
      "step": 10342
    },
    {
      "epoch": 1.8086120218579236,
      "grad_norm": 0.7471908002263993,
      "learning_rate": 0.00012020211002372002,
      "loss": 12.2152,
      "step": 10343
    },
    {
      "epoch": 1.8087868852459015,
      "grad_norm": 0.7129574294197101,
      "learning_rate": 0.00012018824091570103,
      "loss": 12.1469,
      "step": 10344
    },
    {
      "epoch": 1.8089617486338798,
      "grad_norm": 0.8968234047339543,
      "learning_rate": 0.00012017437140284586,
      "loss": 12.1916,
      "step": 10345
    },
    {
      "epoch": 1.809136612021858,
      "grad_norm": 0.7516199943164519,
      "learning_rate": 0.00012016050148543254,
      "loss": 12.2171,
      "step": 10346
    },
    {
      "epoch": 1.809311475409836,
      "grad_norm": 0.8326923117346515,
      "learning_rate": 0.00012014663116373934,
      "loss": 12.3943,
      "step": 10347
    },
    {
      "epoch": 1.8094863387978142,
      "grad_norm": 0.6610711534754395,
      "learning_rate": 0.00012013276043804426,
      "loss": 12.3534,
      "step": 10348
    },
    {
      "epoch": 1.8096612021857923,
      "grad_norm": 0.9280303504497381,
      "learning_rate": 0.00012011888930862558,
      "loss": 12.2941,
      "step": 10349
    },
    {
      "epoch": 1.8098360655737706,
      "grad_norm": 0.7094446792597882,
      "learning_rate": 0.00012010501777576135,
      "loss": 12.2679,
      "step": 10350
    },
    {
      "epoch": 1.8100109289617485,
      "grad_norm": 0.7191347243024175,
      "learning_rate": 0.0001200911458397298,
      "loss": 12.3405,
      "step": 10351
    },
    {
      "epoch": 1.8101857923497269,
      "grad_norm": 0.7565381358454611,
      "learning_rate": 0.00012007727350080909,
      "loss": 12.5051,
      "step": 10352
    },
    {
      "epoch": 1.8103606557377048,
      "grad_norm": 0.6595319006640622,
      "learning_rate": 0.00012006340075927736,
      "loss": 12.2906,
      "step": 10353
    },
    {
      "epoch": 1.8105355191256831,
      "grad_norm": 0.9387627064780933,
      "learning_rate": 0.00012004952761541291,
      "loss": 12.3272,
      "step": 10354
    },
    {
      "epoch": 1.8107103825136612,
      "grad_norm": 0.6609002102577387,
      "learning_rate": 0.00012003565406949384,
      "loss": 12.351,
      "step": 10355
    },
    {
      "epoch": 1.8108852459016394,
      "grad_norm": 0.8091041681399647,
      "learning_rate": 0.00012002178012179839,
      "loss": 12.1209,
      "step": 10356
    },
    {
      "epoch": 1.8110601092896175,
      "grad_norm": 0.7028304281684368,
      "learning_rate": 0.00012000790577260477,
      "loss": 12.4361,
      "step": 10357
    },
    {
      "epoch": 1.8112349726775956,
      "grad_norm": 0.6785217024528947,
      "learning_rate": 0.0001199940310221912,
      "loss": 12.2868,
      "step": 10358
    },
    {
      "epoch": 1.811409836065574,
      "grad_norm": 0.7601114484727142,
      "learning_rate": 0.0001199801558708359,
      "loss": 12.1743,
      "step": 10359
    },
    {
      "epoch": 1.8115846994535518,
      "grad_norm": 0.7182099397628492,
      "learning_rate": 0.00011996628031881719,
      "loss": 12.2781,
      "step": 10360
    },
    {
      "epoch": 1.8117595628415302,
      "grad_norm": 0.740463039738209,
      "learning_rate": 0.00011995240436641319,
      "loss": 12.2786,
      "step": 10361
    },
    {
      "epoch": 1.811934426229508,
      "grad_norm": 0.8121333614134119,
      "learning_rate": 0.00011993852801390226,
      "loss": 12.3854,
      "step": 10362
    },
    {
      "epoch": 1.8121092896174864,
      "grad_norm": 0.6854411543253214,
      "learning_rate": 0.00011992465126156261,
      "loss": 12.2877,
      "step": 10363
    },
    {
      "epoch": 1.8122841530054645,
      "grad_norm": 0.7192243358244496,
      "learning_rate": 0.00011991077410967253,
      "loss": 12.1117,
      "step": 10364
    },
    {
      "epoch": 1.8124590163934426,
      "grad_norm": 0.6439727683390284,
      "learning_rate": 0.00011989689655851029,
      "loss": 12.0907,
      "step": 10365
    },
    {
      "epoch": 1.8126338797814208,
      "grad_norm": 0.6648697571869623,
      "learning_rate": 0.00011988301860835419,
      "loss": 12.0422,
      "step": 10366
    },
    {
      "epoch": 1.8128087431693989,
      "grad_norm": 0.6981636378359546,
      "learning_rate": 0.00011986914025948251,
      "loss": 12.2701,
      "step": 10367
    },
    {
      "epoch": 1.812983606557377,
      "grad_norm": 0.7176893485707879,
      "learning_rate": 0.00011985526151217357,
      "loss": 12.1738,
      "step": 10368
    },
    {
      "epoch": 1.8131584699453551,
      "grad_norm": 0.8606963409982413,
      "learning_rate": 0.00011984138236670566,
      "loss": 12.3813,
      "step": 10369
    },
    {
      "epoch": 1.8133333333333335,
      "grad_norm": 0.6923426021059464,
      "learning_rate": 0.00011982750282335713,
      "loss": 12.1355,
      "step": 10370
    },
    {
      "epoch": 1.8135081967213114,
      "grad_norm": 0.7818070578983979,
      "learning_rate": 0.00011981362288240627,
      "loss": 12.3819,
      "step": 10371
    },
    {
      "epoch": 1.8136830601092897,
      "grad_norm": 0.7487863316452487,
      "learning_rate": 0.00011979974254413144,
      "loss": 12.2252,
      "step": 10372
    },
    {
      "epoch": 1.8138579234972676,
      "grad_norm": 0.6888093344723999,
      "learning_rate": 0.00011978586180881099,
      "loss": 12.3469,
      "step": 10373
    },
    {
      "epoch": 1.814032786885246,
      "grad_norm": 0.7054219772763806,
      "learning_rate": 0.00011977198067672329,
      "loss": 12.2757,
      "step": 10374
    },
    {
      "epoch": 1.814207650273224,
      "grad_norm": 0.684567307567541,
      "learning_rate": 0.00011975809914814662,
      "loss": 12.518,
      "step": 10375
    },
    {
      "epoch": 1.8143825136612022,
      "grad_norm": 0.6946549546469781,
      "learning_rate": 0.0001197442172233594,
      "loss": 12.1498,
      "step": 10376
    },
    {
      "epoch": 1.8145573770491803,
      "grad_norm": 0.6724772735952873,
      "learning_rate": 0.00011973033490264001,
      "loss": 12.1305,
      "step": 10377
    },
    {
      "epoch": 1.8147322404371584,
      "grad_norm": 0.7465545303131381,
      "learning_rate": 0.00011971645218626684,
      "loss": 11.836,
      "step": 10378
    },
    {
      "epoch": 1.8149071038251368,
      "grad_norm": 0.721532751630581,
      "learning_rate": 0.00011970256907451826,
      "loss": 12.5098,
      "step": 10379
    },
    {
      "epoch": 1.8150819672131147,
      "grad_norm": 0.652851975129561,
      "learning_rate": 0.00011968868556767266,
      "loss": 12.2396,
      "step": 10380
    },
    {
      "epoch": 1.815256830601093,
      "grad_norm": 0.695051640647797,
      "learning_rate": 0.00011967480166600849,
      "loss": 12.3353,
      "step": 10381
    },
    {
      "epoch": 1.815431693989071,
      "grad_norm": 0.6767321156224745,
      "learning_rate": 0.00011966091736980409,
      "loss": 12.3557,
      "step": 10382
    },
    {
      "epoch": 1.8156065573770492,
      "grad_norm": 0.6790329673317218,
      "learning_rate": 0.00011964703267933798,
      "loss": 12.3369,
      "step": 10383
    },
    {
      "epoch": 1.8157814207650274,
      "grad_norm": 0.6618326565328603,
      "learning_rate": 0.00011963314759488851,
      "loss": 12.0535,
      "step": 10384
    },
    {
      "epoch": 1.8159562841530055,
      "grad_norm": 0.8116802098798352,
      "learning_rate": 0.00011961926211673416,
      "loss": 11.9729,
      "step": 10385
    },
    {
      "epoch": 1.8161311475409836,
      "grad_norm": 0.6474203380759406,
      "learning_rate": 0.00011960537624515338,
      "loss": 12.0673,
      "step": 10386
    },
    {
      "epoch": 1.8163060109289617,
      "grad_norm": 0.7293279002766092,
      "learning_rate": 0.00011959148998042458,
      "loss": 12.3861,
      "step": 10387
    },
    {
      "epoch": 1.81648087431694,
      "grad_norm": 0.7132486727036433,
      "learning_rate": 0.00011957760332282628,
      "loss": 12.368,
      "step": 10388
    },
    {
      "epoch": 1.816655737704918,
      "grad_norm": 0.645663921255947,
      "learning_rate": 0.00011956371627263687,
      "loss": 12.2536,
      "step": 10389
    },
    {
      "epoch": 1.8168306010928963,
      "grad_norm": 0.7239658649330558,
      "learning_rate": 0.00011954982883013495,
      "loss": 12.0877,
      "step": 10390
    },
    {
      "epoch": 1.8170054644808742,
      "grad_norm": 0.6462982433600298,
      "learning_rate": 0.0001195359409955989,
      "loss": 12.1993,
      "step": 10391
    },
    {
      "epoch": 1.8171803278688525,
      "grad_norm": 0.7067025276274909,
      "learning_rate": 0.00011952205276930723,
      "loss": 12.5253,
      "step": 10392
    },
    {
      "epoch": 1.8173551912568306,
      "grad_norm": 0.7409057846887571,
      "learning_rate": 0.00011950816415153852,
      "loss": 12.2557,
      "step": 10393
    },
    {
      "epoch": 1.8175300546448088,
      "grad_norm": 0.7466516700919559,
      "learning_rate": 0.00011949427514257117,
      "loss": 12.2012,
      "step": 10394
    },
    {
      "epoch": 1.8177049180327869,
      "grad_norm": 0.6267415130706756,
      "learning_rate": 0.00011948038574268377,
      "loss": 12.3811,
      "step": 10395
    },
    {
      "epoch": 1.817879781420765,
      "grad_norm": 0.6567736258585618,
      "learning_rate": 0.0001194664959521548,
      "loss": 12.1572,
      "step": 10396
    },
    {
      "epoch": 1.8180546448087431,
      "grad_norm": 0.7526601956860078,
      "learning_rate": 0.00011945260577126284,
      "loss": 12.4087,
      "step": 10397
    },
    {
      "epoch": 1.8182295081967212,
      "grad_norm": 0.6625349888475613,
      "learning_rate": 0.00011943871520028642,
      "loss": 12.1389,
      "step": 10398
    },
    {
      "epoch": 1.8184043715846996,
      "grad_norm": 0.7289154834205158,
      "learning_rate": 0.00011942482423950406,
      "loss": 12.3432,
      "step": 10399
    },
    {
      "epoch": 1.8185792349726775,
      "grad_norm": 0.7219447760252579,
      "learning_rate": 0.00011941093288919432,
      "loss": 12.391,
      "step": 10400
    },
    {
      "epoch": 1.8187540983606558,
      "grad_norm": 0.6600883121017213,
      "learning_rate": 0.00011939704114963576,
      "loss": 12.5486,
      "step": 10401
    },
    {
      "epoch": 1.8189289617486337,
      "grad_norm": 0.680673607584257,
      "learning_rate": 0.00011938314902110701,
      "loss": 12.0249,
      "step": 10402
    },
    {
      "epoch": 1.819103825136612,
      "grad_norm": 0.7354847488435807,
      "learning_rate": 0.0001193692565038866,
      "loss": 12.4182,
      "step": 10403
    },
    {
      "epoch": 1.8192786885245902,
      "grad_norm": 0.7514676600290663,
      "learning_rate": 0.00011935536359825312,
      "loss": 12.3373,
      "step": 10404
    },
    {
      "epoch": 1.8194535519125683,
      "grad_norm": 0.7992555743965304,
      "learning_rate": 0.00011934147030448517,
      "loss": 12.4137,
      "step": 10405
    },
    {
      "epoch": 1.8196284153005464,
      "grad_norm": 0.787723706574635,
      "learning_rate": 0.00011932757662286133,
      "loss": 12.2839,
      "step": 10406
    },
    {
      "epoch": 1.8198032786885245,
      "grad_norm": 0.6888669192511104,
      "learning_rate": 0.00011931368255366027,
      "loss": 12.3848,
      "step": 10407
    },
    {
      "epoch": 1.8199781420765029,
      "grad_norm": 0.731726499570599,
      "learning_rate": 0.00011929978809716055,
      "loss": 12.3018,
      "step": 10408
    },
    {
      "epoch": 1.8201530054644808,
      "grad_norm": 0.728722704440417,
      "learning_rate": 0.00011928589325364083,
      "loss": 12.3121,
      "step": 10409
    },
    {
      "epoch": 1.8203278688524591,
      "grad_norm": 0.648098887088745,
      "learning_rate": 0.00011927199802337976,
      "loss": 12.3435,
      "step": 10410
    },
    {
      "epoch": 1.820502732240437,
      "grad_norm": 1.0006188309176047,
      "learning_rate": 0.00011925810240665591,
      "loss": 12.5366,
      "step": 10411
    },
    {
      "epoch": 1.8206775956284154,
      "grad_norm": 0.7037154665103972,
      "learning_rate": 0.000119244206403748,
      "loss": 12.3367,
      "step": 10412
    },
    {
      "epoch": 1.8208524590163935,
      "grad_norm": 1.0176563286509752,
      "learning_rate": 0.00011923031001493467,
      "loss": 12.2684,
      "step": 10413
    },
    {
      "epoch": 1.8210273224043716,
      "grad_norm": 0.7554725538431981,
      "learning_rate": 0.00011921641324049458,
      "loss": 12.1654,
      "step": 10414
    },
    {
      "epoch": 1.8212021857923497,
      "grad_norm": 0.9339788242931154,
      "learning_rate": 0.00011920251608070638,
      "loss": 12.4455,
      "step": 10415
    },
    {
      "epoch": 1.8213770491803278,
      "grad_norm": 0.793535837330081,
      "learning_rate": 0.0001191886185358488,
      "loss": 12.4076,
      "step": 10416
    },
    {
      "epoch": 1.8215519125683062,
      "grad_norm": 0.7487958678681811,
      "learning_rate": 0.00011917472060620049,
      "loss": 12.2977,
      "step": 10417
    },
    {
      "epoch": 1.821726775956284,
      "grad_norm": 0.9302862110029029,
      "learning_rate": 0.00011916082229204015,
      "loss": 12.3518,
      "step": 10418
    },
    {
      "epoch": 1.8219016393442624,
      "grad_norm": 0.662540773533255,
      "learning_rate": 0.00011914692359364652,
      "loss": 12.3917,
      "step": 10419
    },
    {
      "epoch": 1.8220765027322403,
      "grad_norm": 0.8498325464361741,
      "learning_rate": 0.00011913302451129826,
      "loss": 12.579,
      "step": 10420
    },
    {
      "epoch": 1.8222513661202187,
      "grad_norm": 0.7949953144359796,
      "learning_rate": 0.00011911912504527411,
      "loss": 12.3287,
      "step": 10421
    },
    {
      "epoch": 1.8224262295081968,
      "grad_norm": 0.7111310403231312,
      "learning_rate": 0.00011910522519585283,
      "loss": 12.3867,
      "step": 10422
    },
    {
      "epoch": 1.822601092896175,
      "grad_norm": 0.8988485439917544,
      "learning_rate": 0.00011909132496331308,
      "loss": 12.0627,
      "step": 10423
    },
    {
      "epoch": 1.822775956284153,
      "grad_norm": 0.726631050207872,
      "learning_rate": 0.00011907742434793365,
      "loss": 12.487,
      "step": 10424
    },
    {
      "epoch": 1.8229508196721311,
      "grad_norm": 0.7827759095855651,
      "learning_rate": 0.00011906352334999331,
      "loss": 12.4216,
      "step": 10425
    },
    {
      "epoch": 1.8231256830601092,
      "grad_norm": 0.8546061177439753,
      "learning_rate": 0.00011904962196977079,
      "loss": 12.368,
      "step": 10426
    },
    {
      "epoch": 1.8233005464480874,
      "grad_norm": 0.6605400308824634,
      "learning_rate": 0.00011903572020754488,
      "loss": 12.2373,
      "step": 10427
    },
    {
      "epoch": 1.8234754098360657,
      "grad_norm": 0.8851197645290161,
      "learning_rate": 0.00011902181806359427,
      "loss": 12.5874,
      "step": 10428
    },
    {
      "epoch": 1.8236502732240436,
      "grad_norm": 0.8155569776394306,
      "learning_rate": 0.00011900791553819787,
      "loss": 12.5128,
      "step": 10429
    },
    {
      "epoch": 1.823825136612022,
      "grad_norm": 0.7174499806276234,
      "learning_rate": 0.00011899401263163434,
      "loss": 12.4168,
      "step": 10430
    },
    {
      "epoch": 1.8239999999999998,
      "grad_norm": 0.8310379332854575,
      "learning_rate": 0.0001189801093441826,
      "loss": 12.3621,
      "step": 10431
    },
    {
      "epoch": 1.8241748633879782,
      "grad_norm": 0.6585678728088499,
      "learning_rate": 0.00011896620567612133,
      "loss": 12.2921,
      "step": 10432
    },
    {
      "epoch": 1.8243497267759563,
      "grad_norm": 0.7483897255758394,
      "learning_rate": 0.00011895230162772943,
      "loss": 12.4474,
      "step": 10433
    },
    {
      "epoch": 1.8245245901639344,
      "grad_norm": 0.7097013305295237,
      "learning_rate": 0.00011893839719928573,
      "loss": 12.3064,
      "step": 10434
    },
    {
      "epoch": 1.8246994535519125,
      "grad_norm": 0.686203064941276,
      "learning_rate": 0.00011892449239106896,
      "loss": 12.0674,
      "step": 10435
    },
    {
      "epoch": 1.8248743169398907,
      "grad_norm": 0.7962859351269781,
      "learning_rate": 0.00011891058720335805,
      "loss": 12.2807,
      "step": 10436
    },
    {
      "epoch": 1.825049180327869,
      "grad_norm": 0.6941791742177725,
      "learning_rate": 0.00011889668163643177,
      "loss": 12.2745,
      "step": 10437
    },
    {
      "epoch": 1.825224043715847,
      "grad_norm": 0.74264000182552,
      "learning_rate": 0.00011888277569056905,
      "loss": 12.1883,
      "step": 10438
    },
    {
      "epoch": 1.8253989071038252,
      "grad_norm": 0.6614358982429069,
      "learning_rate": 0.00011886886936604866,
      "loss": 12.243,
      "step": 10439
    },
    {
      "epoch": 1.8255737704918031,
      "grad_norm": 0.706328028357419,
      "learning_rate": 0.00011885496266314951,
      "loss": 12.2369,
      "step": 10440
    },
    {
      "epoch": 1.8257486338797815,
      "grad_norm": 0.6983319970042818,
      "learning_rate": 0.00011884105558215046,
      "loss": 12.2625,
      "step": 10441
    },
    {
      "epoch": 1.8259234972677596,
      "grad_norm": 0.7183070515579608,
      "learning_rate": 0.00011882714812333041,
      "loss": 12.3995,
      "step": 10442
    },
    {
      "epoch": 1.8260983606557377,
      "grad_norm": 0.6908995422081966,
      "learning_rate": 0.00011881324028696824,
      "loss": 12.1459,
      "step": 10443
    },
    {
      "epoch": 1.8262732240437158,
      "grad_norm": 0.7177082236684474,
      "learning_rate": 0.0001187993320733428,
      "loss": 12.5073,
      "step": 10444
    },
    {
      "epoch": 1.826448087431694,
      "grad_norm": 0.6613542259237717,
      "learning_rate": 0.00011878542348273307,
      "loss": 12.22,
      "step": 10445
    },
    {
      "epoch": 1.8266229508196723,
      "grad_norm": 0.6961825671662438,
      "learning_rate": 0.00011877151451541792,
      "loss": 11.9548,
      "step": 10446
    },
    {
      "epoch": 1.8267978142076502,
      "grad_norm": 0.7211093270596636,
      "learning_rate": 0.00011875760517167624,
      "loss": 12.3805,
      "step": 10447
    },
    {
      "epoch": 1.8269726775956285,
      "grad_norm": 0.8385835166607105,
      "learning_rate": 0.000118743695451787,
      "loss": 12.3726,
      "step": 10448
    },
    {
      "epoch": 1.8271475409836064,
      "grad_norm": 0.6432517411062131,
      "learning_rate": 0.00011872978535602912,
      "loss": 12.0874,
      "step": 10449
    },
    {
      "epoch": 1.8273224043715848,
      "grad_norm": 0.7823389221044652,
      "learning_rate": 0.00011871587488468154,
      "loss": 12.0555,
      "step": 10450
    },
    {
      "epoch": 1.827497267759563,
      "grad_norm": 0.6946340637505702,
      "learning_rate": 0.00011870196403802318,
      "loss": 12.2711,
      "step": 10451
    },
    {
      "epoch": 1.827672131147541,
      "grad_norm": 0.7508559186110607,
      "learning_rate": 0.00011868805281633304,
      "loss": 12.1829,
      "step": 10452
    },
    {
      "epoch": 1.8278469945355191,
      "grad_norm": 0.6353982955053664,
      "learning_rate": 0.00011867414121989004,
      "loss": 12.1813,
      "step": 10453
    },
    {
      "epoch": 1.8280218579234973,
      "grad_norm": 0.7359379930372744,
      "learning_rate": 0.0001186602292489732,
      "loss": 12.3829,
      "step": 10454
    },
    {
      "epoch": 1.8281967213114754,
      "grad_norm": 0.7500666416781139,
      "learning_rate": 0.00011864631690386146,
      "loss": 12.2202,
      "step": 10455
    },
    {
      "epoch": 1.8283715846994535,
      "grad_norm": 0.7687124429957694,
      "learning_rate": 0.00011863240418483381,
      "loss": 12.0925,
      "step": 10456
    },
    {
      "epoch": 1.8285464480874318,
      "grad_norm": 0.6787309534484582,
      "learning_rate": 0.00011861849109216923,
      "loss": 12.2652,
      "step": 10457
    },
    {
      "epoch": 1.8287213114754097,
      "grad_norm": 0.6745869476967906,
      "learning_rate": 0.00011860457762614676,
      "loss": 12.3337,
      "step": 10458
    },
    {
      "epoch": 1.828896174863388,
      "grad_norm": 0.6593578413289605,
      "learning_rate": 0.00011859066378704538,
      "loss": 12.4229,
      "step": 10459
    },
    {
      "epoch": 1.829071038251366,
      "grad_norm": 0.6515238579866149,
      "learning_rate": 0.00011857674957514411,
      "loss": 12.2562,
      "step": 10460
    },
    {
      "epoch": 1.8292459016393443,
      "grad_norm": 0.6835440048848802,
      "learning_rate": 0.00011856283499072196,
      "loss": 12.2386,
      "step": 10461
    },
    {
      "epoch": 1.8294207650273224,
      "grad_norm": 0.7920154699400017,
      "learning_rate": 0.00011854892003405798,
      "loss": 12.2161,
      "step": 10462
    },
    {
      "epoch": 1.8295956284153005,
      "grad_norm": 0.7234338502881577,
      "learning_rate": 0.00011853500470543122,
      "loss": 12.4336,
      "step": 10463
    },
    {
      "epoch": 1.8297704918032787,
      "grad_norm": 0.7513502232468015,
      "learning_rate": 0.00011852108900512067,
      "loss": 12.2478,
      "step": 10464
    },
    {
      "epoch": 1.8299453551912568,
      "grad_norm": 0.7497802454714251,
      "learning_rate": 0.00011850717293340545,
      "loss": 12.3526,
      "step": 10465
    },
    {
      "epoch": 1.8301202185792351,
      "grad_norm": 0.7104009081289469,
      "learning_rate": 0.00011849325649056458,
      "loss": 11.9822,
      "step": 10466
    },
    {
      "epoch": 1.830295081967213,
      "grad_norm": 0.6841659201645351,
      "learning_rate": 0.00011847933967687712,
      "loss": 11.9511,
      "step": 10467
    },
    {
      "epoch": 1.8304699453551914,
      "grad_norm": 0.6766436480346371,
      "learning_rate": 0.00011846542249262218,
      "loss": 12.2643,
      "step": 10468
    },
    {
      "epoch": 1.8306448087431693,
      "grad_norm": 0.6763277426231803,
      "learning_rate": 0.00011845150493807881,
      "loss": 12.2852,
      "step": 10469
    },
    {
      "epoch": 1.8308196721311476,
      "grad_norm": 0.7098491397919613,
      "learning_rate": 0.00011843758701352614,
      "loss": 12.2688,
      "step": 10470
    },
    {
      "epoch": 1.8309945355191257,
      "grad_norm": 0.6215651172285561,
      "learning_rate": 0.0001184236687192432,
      "loss": 12.1837,
      "step": 10471
    },
    {
      "epoch": 1.8311693989071038,
      "grad_norm": 0.8065305819836472,
      "learning_rate": 0.00011840975005550914,
      "loss": 12.2517,
      "step": 10472
    },
    {
      "epoch": 1.831344262295082,
      "grad_norm": 0.6780796723478403,
      "learning_rate": 0.00011839583102260307,
      "loss": 12.3398,
      "step": 10473
    },
    {
      "epoch": 1.83151912568306,
      "grad_norm": 0.74297508688975,
      "learning_rate": 0.00011838191162080412,
      "loss": 12.1843,
      "step": 10474
    },
    {
      "epoch": 1.8316939890710382,
      "grad_norm": 0.7323194548715827,
      "learning_rate": 0.00011836799185039142,
      "loss": 12.4365,
      "step": 10475
    },
    {
      "epoch": 1.8318688524590163,
      "grad_norm": 0.7088548561229682,
      "learning_rate": 0.00011835407171164403,
      "loss": 12.204,
      "step": 10476
    },
    {
      "epoch": 1.8320437158469947,
      "grad_norm": 0.7013942747817573,
      "learning_rate": 0.0001183401512048412,
      "loss": 12.1763,
      "step": 10477
    },
    {
      "epoch": 1.8322185792349726,
      "grad_norm": 0.6326104570576221,
      "learning_rate": 0.000118326230330262,
      "loss": 12.2832,
      "step": 10478
    },
    {
      "epoch": 1.832393442622951,
      "grad_norm": 0.6540954768923558,
      "learning_rate": 0.00011831230908818563,
      "loss": 12.212,
      "step": 10479
    },
    {
      "epoch": 1.832568306010929,
      "grad_norm": 0.6321554791698788,
      "learning_rate": 0.00011829838747889122,
      "loss": 12.0021,
      "step": 10480
    },
    {
      "epoch": 1.8327431693989071,
      "grad_norm": 0.6773565807116645,
      "learning_rate": 0.00011828446550265795,
      "loss": 12.3223,
      "step": 10481
    },
    {
      "epoch": 1.8329180327868853,
      "grad_norm": 0.6174289383987147,
      "learning_rate": 0.00011827054315976504,
      "loss": 12.215,
      "step": 10482
    },
    {
      "epoch": 1.8330928961748634,
      "grad_norm": 0.785797822206185,
      "learning_rate": 0.00011825662045049159,
      "loss": 12.3301,
      "step": 10483
    },
    {
      "epoch": 1.8332677595628415,
      "grad_norm": 0.6227444217372083,
      "learning_rate": 0.0001182426973751169,
      "loss": 12.3488,
      "step": 10484
    },
    {
      "epoch": 1.8334426229508196,
      "grad_norm": 0.8062764058637277,
      "learning_rate": 0.00011822877393392006,
      "loss": 12.3625,
      "step": 10485
    },
    {
      "epoch": 1.833617486338798,
      "grad_norm": 0.7243419526778725,
      "learning_rate": 0.00011821485012718039,
      "loss": 12.4043,
      "step": 10486
    },
    {
      "epoch": 1.8337923497267758,
      "grad_norm": 0.689524865766204,
      "learning_rate": 0.00011820092595517702,
      "loss": 12.3691,
      "step": 10487
    },
    {
      "epoch": 1.8339672131147542,
      "grad_norm": 0.5906744433513197,
      "learning_rate": 0.00011818700141818921,
      "loss": 12.242,
      "step": 10488
    },
    {
      "epoch": 1.834142076502732,
      "grad_norm": 0.705694373513565,
      "learning_rate": 0.00011817307651649616,
      "loss": 12.3176,
      "step": 10489
    },
    {
      "epoch": 1.8343169398907104,
      "grad_norm": 0.6665704680498218,
      "learning_rate": 0.00011815915125037714,
      "loss": 12.3683,
      "step": 10490
    },
    {
      "epoch": 1.8344918032786885,
      "grad_norm": 0.6351623093820067,
      "learning_rate": 0.0001181452256201114,
      "loss": 12.2784,
      "step": 10491
    },
    {
      "epoch": 1.8346666666666667,
      "grad_norm": 0.6551897571061909,
      "learning_rate": 0.0001181312996259781,
      "loss": 12.1736,
      "step": 10492
    },
    {
      "epoch": 1.8348415300546448,
      "grad_norm": 0.6982572973626199,
      "learning_rate": 0.00011811737326825667,
      "loss": 12.1946,
      "step": 10493
    },
    {
      "epoch": 1.835016393442623,
      "grad_norm": 0.6792938927679252,
      "learning_rate": 0.00011810344654722623,
      "loss": 12.0501,
      "step": 10494
    },
    {
      "epoch": 1.8351912568306012,
      "grad_norm": 0.6594915629481004,
      "learning_rate": 0.00011808951946316611,
      "loss": 11.9794,
      "step": 10495
    },
    {
      "epoch": 1.8353661202185791,
      "grad_norm": 0.656988591152244,
      "learning_rate": 0.00011807559201635556,
      "loss": 12.3066,
      "step": 10496
    },
    {
      "epoch": 1.8355409836065575,
      "grad_norm": 0.6608125852766337,
      "learning_rate": 0.00011806166420707392,
      "loss": 12.3838,
      "step": 10497
    },
    {
      "epoch": 1.8357158469945354,
      "grad_norm": 0.6339303908755561,
      "learning_rate": 0.00011804773603560045,
      "loss": 12.3354,
      "step": 10498
    },
    {
      "epoch": 1.8358907103825137,
      "grad_norm": 0.6822809362618157,
      "learning_rate": 0.00011803380750221444,
      "loss": 12.299,
      "step": 10499
    },
    {
      "epoch": 1.8360655737704918,
      "grad_norm": 0.6449239425907762,
      "learning_rate": 0.00011801987860719523,
      "loss": 12.3288,
      "step": 10500
    },
    {
      "epoch": 1.83624043715847,
      "grad_norm": 0.7000075017449864,
      "learning_rate": 0.00011800594935082214,
      "loss": 12.1786,
      "step": 10501
    },
    {
      "epoch": 1.836415300546448,
      "grad_norm": 0.6441206999494452,
      "learning_rate": 0.00011799201973337446,
      "loss": 12.4021,
      "step": 10502
    },
    {
      "epoch": 1.8365901639344262,
      "grad_norm": 0.7915549167751564,
      "learning_rate": 0.00011797808975513156,
      "loss": 12.3012,
      "step": 10503
    },
    {
      "epoch": 1.8367650273224043,
      "grad_norm": 0.6673940608361867,
      "learning_rate": 0.00011796415941637276,
      "loss": 12.1088,
      "step": 10504
    },
    {
      "epoch": 1.8369398907103824,
      "grad_norm": 0.7081586009579994,
      "learning_rate": 0.00011795022871737737,
      "loss": 12.361,
      "step": 10505
    },
    {
      "epoch": 1.8371147540983608,
      "grad_norm": 0.7673718896297251,
      "learning_rate": 0.00011793629765842482,
      "loss": 12.5016,
      "step": 10506
    },
    {
      "epoch": 1.8372896174863387,
      "grad_norm": 0.6958390347131358,
      "learning_rate": 0.00011792236623979442,
      "loss": 12.2402,
      "step": 10507
    },
    {
      "epoch": 1.837464480874317,
      "grad_norm": 0.6917053473896497,
      "learning_rate": 0.00011790843446176554,
      "loss": 12.3103,
      "step": 10508
    },
    {
      "epoch": 1.837639344262295,
      "grad_norm": 0.6971660451171443,
      "learning_rate": 0.00011789450232461756,
      "loss": 12.2928,
      "step": 10509
    },
    {
      "epoch": 1.8378142076502733,
      "grad_norm": 0.610202794094046,
      "learning_rate": 0.00011788056982862987,
      "loss": 12.0944,
      "step": 10510
    },
    {
      "epoch": 1.8379890710382514,
      "grad_norm": 0.6570428028244779,
      "learning_rate": 0.00011786663697408186,
      "loss": 12.3385,
      "step": 10511
    },
    {
      "epoch": 1.8381639344262295,
      "grad_norm": 0.626412901588855,
      "learning_rate": 0.00011785270376125292,
      "loss": 12.2348,
      "step": 10512
    },
    {
      "epoch": 1.8383387978142076,
      "grad_norm": 0.723730951703196,
      "learning_rate": 0.00011783877019042247,
      "loss": 12.1436,
      "step": 10513
    },
    {
      "epoch": 1.8385136612021857,
      "grad_norm": 0.6089721796548709,
      "learning_rate": 0.00011782483626186987,
      "loss": 12.418,
      "step": 10514
    },
    {
      "epoch": 1.838688524590164,
      "grad_norm": 0.6642037527668233,
      "learning_rate": 0.00011781090197587459,
      "loss": 12.0562,
      "step": 10515
    },
    {
      "epoch": 1.838863387978142,
      "grad_norm": 0.6551104537841942,
      "learning_rate": 0.00011779696733271604,
      "loss": 12.1841,
      "step": 10516
    },
    {
      "epoch": 1.8390382513661203,
      "grad_norm": 0.7191142735208057,
      "learning_rate": 0.00011778303233267365,
      "loss": 12.3076,
      "step": 10517
    },
    {
      "epoch": 1.8392131147540982,
      "grad_norm": 0.7837421025712811,
      "learning_rate": 0.00011776909697602689,
      "loss": 12.3215,
      "step": 10518
    },
    {
      "epoch": 1.8393879781420766,
      "grad_norm": 0.666123791475688,
      "learning_rate": 0.00011775516126305513,
      "loss": 12.1901,
      "step": 10519
    },
    {
      "epoch": 1.8395628415300547,
      "grad_norm": 0.7519662464223223,
      "learning_rate": 0.00011774122519403788,
      "loss": 12.2399,
      "step": 10520
    },
    {
      "epoch": 1.8397377049180328,
      "grad_norm": 0.7572941597282482,
      "learning_rate": 0.00011772728876925461,
      "loss": 12.4991,
      "step": 10521
    },
    {
      "epoch": 1.839912568306011,
      "grad_norm": 0.6997623730574818,
      "learning_rate": 0.00011771335198898477,
      "loss": 12.4525,
      "step": 10522
    },
    {
      "epoch": 1.840087431693989,
      "grad_norm": 0.836871600360832,
      "learning_rate": 0.00011769941485350784,
      "loss": 12.1623,
      "step": 10523
    },
    {
      "epoch": 1.8402622950819674,
      "grad_norm": 0.6384545310528477,
      "learning_rate": 0.00011768547736310327,
      "loss": 12.4572,
      "step": 10524
    },
    {
      "epoch": 1.8404371584699453,
      "grad_norm": 0.7475356157207016,
      "learning_rate": 0.00011767153951805062,
      "loss": 12.1693,
      "step": 10525
    },
    {
      "epoch": 1.8406120218579236,
      "grad_norm": 0.7302353759863321,
      "learning_rate": 0.00011765760131862929,
      "loss": 12.176,
      "step": 10526
    },
    {
      "epoch": 1.8407868852459015,
      "grad_norm": 0.735542225952714,
      "learning_rate": 0.0001176436627651189,
      "loss": 12.1696,
      "step": 10527
    },
    {
      "epoch": 1.8409617486338798,
      "grad_norm": 0.678974510326267,
      "learning_rate": 0.00011762972385779882,
      "loss": 12.4639,
      "step": 10528
    },
    {
      "epoch": 1.841136612021858,
      "grad_norm": 0.6873528571210649,
      "learning_rate": 0.00011761578459694868,
      "loss": 12.2516,
      "step": 10529
    },
    {
      "epoch": 1.841311475409836,
      "grad_norm": 0.7315229639612709,
      "learning_rate": 0.000117601844982848,
      "loss": 12.2184,
      "step": 10530
    },
    {
      "epoch": 1.8414863387978142,
      "grad_norm": 0.6822302515178454,
      "learning_rate": 0.00011758790501577625,
      "loss": 12.1517,
      "step": 10531
    },
    {
      "epoch": 1.8416612021857923,
      "grad_norm": 0.6868357019809754,
      "learning_rate": 0.00011757396469601305,
      "loss": 12.3307,
      "step": 10532
    },
    {
      "epoch": 1.8418360655737704,
      "grad_norm": 0.6616645285799122,
      "learning_rate": 0.00011756002402383783,
      "loss": 12.2365,
      "step": 10533
    },
    {
      "epoch": 1.8420109289617486,
      "grad_norm": 0.6795387991312197,
      "learning_rate": 0.00011754608299953027,
      "loss": 12.3823,
      "step": 10534
    },
    {
      "epoch": 1.842185792349727,
      "grad_norm": 0.7795991313352337,
      "learning_rate": 0.00011753214162336985,
      "loss": 12.1586,
      "step": 10535
    },
    {
      "epoch": 1.8423606557377048,
      "grad_norm": 0.66604716073428,
      "learning_rate": 0.00011751819989563617,
      "loss": 12.1707,
      "step": 10536
    },
    {
      "epoch": 1.8425355191256831,
      "grad_norm": 0.7374213680806052,
      "learning_rate": 0.00011750425781660879,
      "loss": 12.3237,
      "step": 10537
    },
    {
      "epoch": 1.842710382513661,
      "grad_norm": 0.7007644858186254,
      "learning_rate": 0.00011749031538656725,
      "loss": 12.1375,
      "step": 10538
    },
    {
      "epoch": 1.8428852459016394,
      "grad_norm": 0.695607632495191,
      "learning_rate": 0.00011747637260579126,
      "loss": 12.3173,
      "step": 10539
    },
    {
      "epoch": 1.8430601092896175,
      "grad_norm": 0.840330738833329,
      "learning_rate": 0.00011746242947456028,
      "loss": 12.2132,
      "step": 10540
    },
    {
      "epoch": 1.8432349726775956,
      "grad_norm": 0.6279883657990959,
      "learning_rate": 0.00011744848599315401,
      "loss": 12.3524,
      "step": 10541
    },
    {
      "epoch": 1.8434098360655737,
      "grad_norm": 0.7582874600372655,
      "learning_rate": 0.00011743454216185201,
      "loss": 12.2315,
      "step": 10542
    },
    {
      "epoch": 1.8435846994535519,
      "grad_norm": 0.6994606677543395,
      "learning_rate": 0.0001174205979809339,
      "loss": 12.2717,
      "step": 10543
    },
    {
      "epoch": 1.8437595628415302,
      "grad_norm": 0.7319801401591198,
      "learning_rate": 0.00011740665345067932,
      "loss": 12.3418,
      "step": 10544
    },
    {
      "epoch": 1.843934426229508,
      "grad_norm": 0.7603310040972344,
      "learning_rate": 0.00011739270857136788,
      "loss": 12.3753,
      "step": 10545
    },
    {
      "epoch": 1.8441092896174864,
      "grad_norm": 0.6726899747920454,
      "learning_rate": 0.00011737876334327924,
      "loss": 11.9828,
      "step": 10546
    },
    {
      "epoch": 1.8442841530054643,
      "grad_norm": 0.7762979339872361,
      "learning_rate": 0.00011736481776669306,
      "loss": 12.1784,
      "step": 10547
    },
    {
      "epoch": 1.8444590163934427,
      "grad_norm": 0.7289921214600743,
      "learning_rate": 0.00011735087184188893,
      "loss": 12.4756,
      "step": 10548
    },
    {
      "epoch": 1.8446338797814208,
      "grad_norm": 0.8420378185833752,
      "learning_rate": 0.00011733692556914654,
      "loss": 12.5196,
      "step": 10549
    },
    {
      "epoch": 1.844808743169399,
      "grad_norm": 0.7297642951831281,
      "learning_rate": 0.00011732297894874559,
      "loss": 12.1994,
      "step": 10550
    },
    {
      "epoch": 1.844983606557377,
      "grad_norm": 0.6940989902639505,
      "learning_rate": 0.00011730903198096573,
      "loss": 12.4764,
      "step": 10551
    },
    {
      "epoch": 1.8451584699453552,
      "grad_norm": 0.7169375740955269,
      "learning_rate": 0.00011729508466608661,
      "loss": 12.4588,
      "step": 10552
    },
    {
      "epoch": 1.8453333333333335,
      "grad_norm": 0.7876759496688356,
      "learning_rate": 0.00011728113700438794,
      "loss": 12.4328,
      "step": 10553
    },
    {
      "epoch": 1.8455081967213114,
      "grad_norm": 0.7598536868757464,
      "learning_rate": 0.00011726718899614942,
      "loss": 12.2456,
      "step": 10554
    },
    {
      "epoch": 1.8456830601092897,
      "grad_norm": 0.7500562421800662,
      "learning_rate": 0.00011725324064165075,
      "loss": 12.488,
      "step": 10555
    },
    {
      "epoch": 1.8458579234972676,
      "grad_norm": 0.6435580827449187,
      "learning_rate": 0.00011723929194117163,
      "loss": 12.1386,
      "step": 10556
    },
    {
      "epoch": 1.846032786885246,
      "grad_norm": 0.829446923088484,
      "learning_rate": 0.00011722534289499177,
      "loss": 12.2521,
      "step": 10557
    },
    {
      "epoch": 1.846207650273224,
      "grad_norm": 0.7062896309024408,
      "learning_rate": 0.00011721139350339088,
      "loss": 12.1692,
      "step": 10558
    },
    {
      "epoch": 1.8463825136612022,
      "grad_norm": 0.6666835662349475,
      "learning_rate": 0.00011719744376664872,
      "loss": 12.2632,
      "step": 10559
    },
    {
      "epoch": 1.8465573770491803,
      "grad_norm": 0.6994578368668833,
      "learning_rate": 0.000117183493685045,
      "loss": 12.2371,
      "step": 10560
    },
    {
      "epoch": 1.8467322404371584,
      "grad_norm": 0.7352653163861769,
      "learning_rate": 0.00011716954325885949,
      "loss": 12.472,
      "step": 10561
    },
    {
      "epoch": 1.8469071038251366,
      "grad_norm": 0.7426540286283264,
      "learning_rate": 0.00011715559248837191,
      "loss": 11.9056,
      "step": 10562
    },
    {
      "epoch": 1.8470819672131147,
      "grad_norm": 0.6407494875914878,
      "learning_rate": 0.00011714164137386202,
      "loss": 12.2992,
      "step": 10563
    },
    {
      "epoch": 1.847256830601093,
      "grad_norm": 0.8682277753891856,
      "learning_rate": 0.0001171276899156096,
      "loss": 11.9955,
      "step": 10564
    },
    {
      "epoch": 1.847431693989071,
      "grad_norm": 0.6969468805013829,
      "learning_rate": 0.00011711373811389439,
      "loss": 12.1113,
      "step": 10565
    },
    {
      "epoch": 1.8476065573770493,
      "grad_norm": 0.6991356266700313,
      "learning_rate": 0.00011709978596899622,
      "loss": 12.1993,
      "step": 10566
    },
    {
      "epoch": 1.8477814207650272,
      "grad_norm": 0.7109675130350683,
      "learning_rate": 0.0001170858334811948,
      "loss": 12.4311,
      "step": 10567
    },
    {
      "epoch": 1.8479562841530055,
      "grad_norm": 0.7273765244289374,
      "learning_rate": 0.00011707188065076998,
      "loss": 12.4479,
      "step": 10568
    },
    {
      "epoch": 1.8481311475409836,
      "grad_norm": 0.6846895019197751,
      "learning_rate": 0.00011705792747800153,
      "loss": 12.2242,
      "step": 10569
    },
    {
      "epoch": 1.8483060109289617,
      "grad_norm": 0.6782198374621382,
      "learning_rate": 0.00011704397396316924,
      "loss": 12.2762,
      "step": 10570
    },
    {
      "epoch": 1.8484808743169399,
      "grad_norm": 0.6395606035463163,
      "learning_rate": 0.000117030020106553,
      "loss": 12.3125,
      "step": 10571
    },
    {
      "epoch": 1.848655737704918,
      "grad_norm": 0.6749389234381163,
      "learning_rate": 0.0001170160659084325,
      "loss": 12.1613,
      "step": 10572
    },
    {
      "epoch": 1.8488306010928963,
      "grad_norm": 0.6685303941399985,
      "learning_rate": 0.00011700211136908767,
      "loss": 11.8911,
      "step": 10573
    },
    {
      "epoch": 1.8490054644808742,
      "grad_norm": 0.6658859657640276,
      "learning_rate": 0.00011698815648879828,
      "loss": 11.8584,
      "step": 10574
    },
    {
      "epoch": 1.8491803278688526,
      "grad_norm": 0.6695705599550132,
      "learning_rate": 0.00011697420126784422,
      "loss": 12.2059,
      "step": 10575
    },
    {
      "epoch": 1.8493551912568305,
      "grad_norm": 0.7304989147151635,
      "learning_rate": 0.00011696024570650528,
      "loss": 12.3233,
      "step": 10576
    },
    {
      "epoch": 1.8495300546448088,
      "grad_norm": 0.6097206820640632,
      "learning_rate": 0.00011694628980506135,
      "loss": 12.1125,
      "step": 10577
    },
    {
      "epoch": 1.849704918032787,
      "grad_norm": 0.6955212627781535,
      "learning_rate": 0.0001169323335637923,
      "loss": 12.3298,
      "step": 10578
    },
    {
      "epoch": 1.849879781420765,
      "grad_norm": 0.6686460892729875,
      "learning_rate": 0.00011691837698297792,
      "loss": 12.2861,
      "step": 10579
    },
    {
      "epoch": 1.8500546448087432,
      "grad_norm": 0.9708150130412379,
      "learning_rate": 0.00011690442006289819,
      "loss": 12.5437,
      "step": 10580
    },
    {
      "epoch": 1.8502295081967213,
      "grad_norm": 0.6595538528894553,
      "learning_rate": 0.0001168904628038329,
      "loss": 12.4528,
      "step": 10581
    },
    {
      "epoch": 1.8504043715846996,
      "grad_norm": 0.7612244987304712,
      "learning_rate": 0.00011687650520606199,
      "loss": 12.248,
      "step": 10582
    },
    {
      "epoch": 1.8505792349726775,
      "grad_norm": 0.6485759129598948,
      "learning_rate": 0.00011686254726986532,
      "loss": 11.853,
      "step": 10583
    },
    {
      "epoch": 1.8507540983606559,
      "grad_norm": 0.6752374358432064,
      "learning_rate": 0.0001168485889955228,
      "loss": 12.3683,
      "step": 10584
    },
    {
      "epoch": 1.8509289617486337,
      "grad_norm": 0.72403382975009,
      "learning_rate": 0.00011683463038331434,
      "loss": 12.1367,
      "step": 10585
    },
    {
      "epoch": 1.851103825136612,
      "grad_norm": 0.6696031533375164,
      "learning_rate": 0.00011682067143351985,
      "loss": 12.1327,
      "step": 10586
    },
    {
      "epoch": 1.8512786885245902,
      "grad_norm": 0.809416990668271,
      "learning_rate": 0.00011680671214641927,
      "loss": 12.3079,
      "step": 10587
    },
    {
      "epoch": 1.8514535519125683,
      "grad_norm": 0.7614065320946105,
      "learning_rate": 0.00011679275252229248,
      "loss": 12.2542,
      "step": 10588
    },
    {
      "epoch": 1.8516284153005464,
      "grad_norm": 0.6874717114555995,
      "learning_rate": 0.00011677879256141949,
      "loss": 12.1802,
      "step": 10589
    },
    {
      "epoch": 1.8518032786885246,
      "grad_norm": 0.7900274751393097,
      "learning_rate": 0.00011676483226408014,
      "loss": 12.4135,
      "step": 10590
    },
    {
      "epoch": 1.8519781420765027,
      "grad_norm": 0.819900699486473,
      "learning_rate": 0.00011675087163055445,
      "loss": 12.3115,
      "step": 10591
    },
    {
      "epoch": 1.8521530054644808,
      "grad_norm": 0.7881903536596455,
      "learning_rate": 0.00011673691066112236,
      "loss": 12.2633,
      "step": 10592
    },
    {
      "epoch": 1.8523278688524591,
      "grad_norm": 0.691823638266554,
      "learning_rate": 0.0001167229493560638,
      "loss": 12.3318,
      "step": 10593
    },
    {
      "epoch": 1.852502732240437,
      "grad_norm": 0.7616087184335557,
      "learning_rate": 0.00011670898771565877,
      "loss": 12.2688,
      "step": 10594
    },
    {
      "epoch": 1.8526775956284154,
      "grad_norm": 0.630709684786522,
      "learning_rate": 0.00011669502574018722,
      "loss": 12.194,
      "step": 10595
    },
    {
      "epoch": 1.8528524590163933,
      "grad_norm": 0.8047587948547291,
      "learning_rate": 0.00011668106342992917,
      "loss": 12.1406,
      "step": 10596
    },
    {
      "epoch": 1.8530273224043716,
      "grad_norm": 0.6462163338374004,
      "learning_rate": 0.00011666710078516456,
      "loss": 12.1505,
      "step": 10597
    },
    {
      "epoch": 1.8532021857923497,
      "grad_norm": 0.6848979844449046,
      "learning_rate": 0.00011665313780617342,
      "loss": 12.3848,
      "step": 10598
    },
    {
      "epoch": 1.8533770491803279,
      "grad_norm": 0.6870422334650771,
      "learning_rate": 0.00011663917449323573,
      "loss": 12.4791,
      "step": 10599
    },
    {
      "epoch": 1.853551912568306,
      "grad_norm": 0.6284224614893039,
      "learning_rate": 0.0001166252108466315,
      "loss": 12.2032,
      "step": 10600
    },
    {
      "epoch": 1.853726775956284,
      "grad_norm": 0.7790426919467173,
      "learning_rate": 0.00011661124686664075,
      "loss": 12.0947,
      "step": 10601
    },
    {
      "epoch": 1.8539016393442624,
      "grad_norm": 0.6600786052526307,
      "learning_rate": 0.00011659728255354349,
      "loss": 12.3053,
      "step": 10602
    },
    {
      "epoch": 1.8540765027322403,
      "grad_norm": 0.657912371057612,
      "learning_rate": 0.00011658331790761976,
      "loss": 12.3002,
      "step": 10603
    },
    {
      "epoch": 1.8542513661202187,
      "grad_norm": 0.6996114151671443,
      "learning_rate": 0.00011656935292914958,
      "loss": 12.2544,
      "step": 10604
    },
    {
      "epoch": 1.8544262295081966,
      "grad_norm": 0.7237326803505363,
      "learning_rate": 0.000116555387618413,
      "loss": 12.1958,
      "step": 10605
    },
    {
      "epoch": 1.854601092896175,
      "grad_norm": 0.719275957433355,
      "learning_rate": 0.00011654142197569006,
      "loss": 12.0843,
      "step": 10606
    },
    {
      "epoch": 1.854775956284153,
      "grad_norm": 0.7312141219564613,
      "learning_rate": 0.00011652745600126087,
      "loss": 12.1728,
      "step": 10607
    },
    {
      "epoch": 1.8549508196721312,
      "grad_norm": 0.818012329958823,
      "learning_rate": 0.00011651348969540539,
      "loss": 12.386,
      "step": 10608
    },
    {
      "epoch": 1.8551256830601093,
      "grad_norm": 0.7391116112548994,
      "learning_rate": 0.00011649952305840374,
      "loss": 12.0379,
      "step": 10609
    },
    {
      "epoch": 1.8553005464480874,
      "grad_norm": 0.7070458486872988,
      "learning_rate": 0.00011648555609053601,
      "loss": 12.0753,
      "step": 10610
    },
    {
      "epoch": 1.8554754098360657,
      "grad_norm": 0.7116541350193669,
      "learning_rate": 0.00011647158879208225,
      "loss": 12.0779,
      "step": 10611
    },
    {
      "epoch": 1.8556502732240436,
      "grad_norm": 0.6162590314247113,
      "learning_rate": 0.00011645762116332257,
      "loss": 11.9481,
      "step": 10612
    },
    {
      "epoch": 1.855825136612022,
      "grad_norm": 0.7214178895025539,
      "learning_rate": 0.00011644365320453705,
      "loss": 12.0722,
      "step": 10613
    },
    {
      "epoch": 1.8559999999999999,
      "grad_norm": 0.6338981366418125,
      "learning_rate": 0.00011642968491600581,
      "loss": 12.1169,
      "step": 10614
    },
    {
      "epoch": 1.8561748633879782,
      "grad_norm": 0.7433931941089665,
      "learning_rate": 0.00011641571629800892,
      "loss": 12.4538,
      "step": 10615
    },
    {
      "epoch": 1.8563497267759563,
      "grad_norm": 0.6635168570352963,
      "learning_rate": 0.00011640174735082654,
      "loss": 12.4901,
      "step": 10616
    },
    {
      "epoch": 1.8565245901639345,
      "grad_norm": 0.7458434111248914,
      "learning_rate": 0.00011638777807473874,
      "loss": 12.4204,
      "step": 10617
    },
    {
      "epoch": 1.8566994535519126,
      "grad_norm": 0.7033021041259995,
      "learning_rate": 0.00011637380847002569,
      "loss": 12.2273,
      "step": 10618
    },
    {
      "epoch": 1.8568743169398907,
      "grad_norm": 0.7018324365817893,
      "learning_rate": 0.0001163598385369675,
      "loss": 12.326,
      "step": 10619
    },
    {
      "epoch": 1.8570491803278688,
      "grad_norm": 0.6565788813056594,
      "learning_rate": 0.0001163458682758443,
      "loss": 12.1398,
      "step": 10620
    },
    {
      "epoch": 1.857224043715847,
      "grad_norm": 0.6837577918888097,
      "learning_rate": 0.00011633189768693628,
      "loss": 12.3842,
      "step": 10621
    },
    {
      "epoch": 1.8573989071038253,
      "grad_norm": 0.6936110374046345,
      "learning_rate": 0.00011631792677052353,
      "loss": 12.4199,
      "step": 10622
    },
    {
      "epoch": 1.8575737704918032,
      "grad_norm": 0.76099220622734,
      "learning_rate": 0.0001163039555268863,
      "loss": 12.4339,
      "step": 10623
    },
    {
      "epoch": 1.8577486338797815,
      "grad_norm": 0.7046752653007271,
      "learning_rate": 0.00011628998395630464,
      "loss": 12.2491,
      "step": 10624
    },
    {
      "epoch": 1.8579234972677594,
      "grad_norm": 0.6563397401784965,
      "learning_rate": 0.00011627601205905883,
      "loss": 12.382,
      "step": 10625
    },
    {
      "epoch": 1.8580983606557377,
      "grad_norm": 0.7240152228554309,
      "learning_rate": 0.000116262039835429,
      "loss": 11.7304,
      "step": 10626
    },
    {
      "epoch": 1.8582732240437159,
      "grad_norm": 0.6429472144841546,
      "learning_rate": 0.0001162480672856953,
      "loss": 11.8401,
      "step": 10627
    },
    {
      "epoch": 1.858448087431694,
      "grad_norm": 0.6014617317613083,
      "learning_rate": 0.00011623409441013802,
      "loss": 12.368,
      "step": 10628
    },
    {
      "epoch": 1.858622950819672,
      "grad_norm": 0.6507304293116094,
      "learning_rate": 0.00011622012120903724,
      "loss": 12.178,
      "step": 10629
    },
    {
      "epoch": 1.8587978142076502,
      "grad_norm": 0.7168167808309047,
      "learning_rate": 0.00011620614768267328,
      "loss": 12.4559,
      "step": 10630
    },
    {
      "epoch": 1.8589726775956286,
      "grad_norm": 0.7352144281562023,
      "learning_rate": 0.0001161921738313263,
      "loss": 12.4089,
      "step": 10631
    },
    {
      "epoch": 1.8591475409836065,
      "grad_norm": 0.7891731930555754,
      "learning_rate": 0.0001161781996552765,
      "loss": 12.4274,
      "step": 10632
    },
    {
      "epoch": 1.8593224043715848,
      "grad_norm": 0.610438825193436,
      "learning_rate": 0.00011616422515480412,
      "loss": 12.1851,
      "step": 10633
    },
    {
      "epoch": 1.8594972677595627,
      "grad_norm": 0.642781233502894,
      "learning_rate": 0.00011615025033018936,
      "loss": 12.1422,
      "step": 10634
    },
    {
      "epoch": 1.859672131147541,
      "grad_norm": 0.7502307325613955,
      "learning_rate": 0.00011613627518171255,
      "loss": 12.4465,
      "step": 10635
    },
    {
      "epoch": 1.8598469945355192,
      "grad_norm": 0.6540961970906896,
      "learning_rate": 0.00011612229970965385,
      "loss": 12.2702,
      "step": 10636
    },
    {
      "epoch": 1.8600218579234973,
      "grad_norm": 0.6373243026684021,
      "learning_rate": 0.00011610832391429356,
      "loss": 12.2604,
      "step": 10637
    },
    {
      "epoch": 1.8601967213114754,
      "grad_norm": 0.7143882861180567,
      "learning_rate": 0.0001160943477959119,
      "loss": 12.1841,
      "step": 10638
    },
    {
      "epoch": 1.8603715846994535,
      "grad_norm": 0.6478700119275741,
      "learning_rate": 0.00011608037135478915,
      "loss": 12.3689,
      "step": 10639
    },
    {
      "epoch": 1.8605464480874319,
      "grad_norm": 0.7077462385937335,
      "learning_rate": 0.00011606639459120561,
      "loss": 12.0506,
      "step": 10640
    },
    {
      "epoch": 1.8607213114754098,
      "grad_norm": 0.6532344090000071,
      "learning_rate": 0.0001160524175054415,
      "loss": 12.3428,
      "step": 10641
    },
    {
      "epoch": 1.860896174863388,
      "grad_norm": 0.6878068587136328,
      "learning_rate": 0.00011603844009777714,
      "loss": 12.3668,
      "step": 10642
    },
    {
      "epoch": 1.861071038251366,
      "grad_norm": 0.74701880533508,
      "learning_rate": 0.00011602446236849278,
      "loss": 12.5609,
      "step": 10643
    },
    {
      "epoch": 1.8612459016393443,
      "grad_norm": 0.665642137567142,
      "learning_rate": 0.00011601048431786878,
      "loss": 12.0883,
      "step": 10644
    },
    {
      "epoch": 1.8614207650273225,
      "grad_norm": 0.6858000335449493,
      "learning_rate": 0.00011599650594618541,
      "loss": 12.106,
      "step": 10645
    },
    {
      "epoch": 1.8615956284153006,
      "grad_norm": 0.6478286042871902,
      "learning_rate": 0.00011598252725372298,
      "loss": 12.2261,
      "step": 10646
    },
    {
      "epoch": 1.8617704918032787,
      "grad_norm": 0.6666749887609265,
      "learning_rate": 0.0001159685482407618,
      "loss": 12.3204,
      "step": 10647
    },
    {
      "epoch": 1.8619453551912568,
      "grad_norm": 0.7009631941003355,
      "learning_rate": 0.00011595456890758218,
      "loss": 12.0529,
      "step": 10648
    },
    {
      "epoch": 1.862120218579235,
      "grad_norm": 0.7409195936940395,
      "learning_rate": 0.0001159405892544645,
      "loss": 12.414,
      "step": 10649
    },
    {
      "epoch": 1.862295081967213,
      "grad_norm": 0.73951345286935,
      "learning_rate": 0.00011592660928168904,
      "loss": 11.8542,
      "step": 10650
    },
    {
      "epoch": 1.8624699453551914,
      "grad_norm": 0.6297877075071785,
      "learning_rate": 0.0001159126289895362,
      "loss": 12.2274,
      "step": 10651
    },
    {
      "epoch": 1.8626448087431693,
      "grad_norm": 0.7016217446799725,
      "learning_rate": 0.00011589864837828624,
      "loss": 12.4559,
      "step": 10652
    },
    {
      "epoch": 1.8628196721311476,
      "grad_norm": 0.6729666526437273,
      "learning_rate": 0.00011588466744821958,
      "loss": 12.3808,
      "step": 10653
    },
    {
      "epoch": 1.8629945355191255,
      "grad_norm": 0.7372301287603132,
      "learning_rate": 0.00011587068619961657,
      "loss": 11.7656,
      "step": 10654
    },
    {
      "epoch": 1.8631693989071039,
      "grad_norm": 0.7656205543922069,
      "learning_rate": 0.0001158567046327576,
      "loss": 12.3367,
      "step": 10655
    },
    {
      "epoch": 1.863344262295082,
      "grad_norm": 0.7339811092138193,
      "learning_rate": 0.00011584272274792296,
      "loss": 12.4352,
      "step": 10656
    },
    {
      "epoch": 1.86351912568306,
      "grad_norm": 0.6773764245852378,
      "learning_rate": 0.00011582874054539312,
      "loss": 12.1465,
      "step": 10657
    },
    {
      "epoch": 1.8636939890710382,
      "grad_norm": 0.6874143914579243,
      "learning_rate": 0.00011581475802544841,
      "loss": 12.159,
      "step": 10658
    },
    {
      "epoch": 1.8638688524590163,
      "grad_norm": 0.6210477605770612,
      "learning_rate": 0.00011580077518836927,
      "loss": 12.059,
      "step": 10659
    },
    {
      "epoch": 1.8640437158469947,
      "grad_norm": 0.7454711246318431,
      "learning_rate": 0.00011578679203443605,
      "loss": 12.4057,
      "step": 10660
    },
    {
      "epoch": 1.8642185792349726,
      "grad_norm": 0.7606090317387648,
      "learning_rate": 0.0001157728085639292,
      "loss": 12.4794,
      "step": 10661
    },
    {
      "epoch": 1.864393442622951,
      "grad_norm": 0.7197300677712464,
      "learning_rate": 0.00011575882477712912,
      "loss": 12.3069,
      "step": 10662
    },
    {
      "epoch": 1.8645683060109288,
      "grad_norm": 0.7897700729864234,
      "learning_rate": 0.00011574484067431617,
      "loss": 12.4059,
      "step": 10663
    },
    {
      "epoch": 1.8647431693989072,
      "grad_norm": 0.6630557168003879,
      "learning_rate": 0.00011573085625577087,
      "loss": 12.186,
      "step": 10664
    },
    {
      "epoch": 1.8649180327868853,
      "grad_norm": 0.6898969571250771,
      "learning_rate": 0.00011571687152177357,
      "loss": 12.2326,
      "step": 10665
    },
    {
      "epoch": 1.8650928961748634,
      "grad_norm": 0.669492230885958,
      "learning_rate": 0.00011570288647260474,
      "loss": 12.36,
      "step": 10666
    },
    {
      "epoch": 1.8652677595628415,
      "grad_norm": 0.6774895876512416,
      "learning_rate": 0.00011568890110854485,
      "loss": 12.2795,
      "step": 10667
    },
    {
      "epoch": 1.8654426229508196,
      "grad_norm": 0.7647590449247169,
      "learning_rate": 0.00011567491542987427,
      "loss": 12.292,
      "step": 10668
    },
    {
      "epoch": 1.8656174863387978,
      "grad_norm": 0.6517498201224925,
      "learning_rate": 0.00011566092943687356,
      "loss": 12.3148,
      "step": 10669
    },
    {
      "epoch": 1.8657923497267759,
      "grad_norm": 0.7273475678940274,
      "learning_rate": 0.0001156469431298231,
      "loss": 12.4289,
      "step": 10670
    },
    {
      "epoch": 1.8659672131147542,
      "grad_norm": 0.6569490987408645,
      "learning_rate": 0.00011563295650900344,
      "loss": 12.1034,
      "step": 10671
    },
    {
      "epoch": 1.8661420765027321,
      "grad_norm": 0.8568255592562449,
      "learning_rate": 0.00011561896957469491,
      "loss": 11.9705,
      "step": 10672
    },
    {
      "epoch": 1.8663169398907105,
      "grad_norm": 0.6870844511958872,
      "learning_rate": 0.00011560498232717815,
      "loss": 12.4075,
      "step": 10673
    },
    {
      "epoch": 1.8664918032786886,
      "grad_norm": 0.8186337696514165,
      "learning_rate": 0.00011559099476673357,
      "loss": 12.3748,
      "step": 10674
    },
    {
      "epoch": 1.8666666666666667,
      "grad_norm": 0.6672292448923971,
      "learning_rate": 0.00011557700689364166,
      "loss": 12.1533,
      "step": 10675
    },
    {
      "epoch": 1.8668415300546448,
      "grad_norm": 0.7353203852047145,
      "learning_rate": 0.00011556301870818296,
      "loss": 12.3583,
      "step": 10676
    },
    {
      "epoch": 1.867016393442623,
      "grad_norm": 0.7651199817817853,
      "learning_rate": 0.0001155490302106379,
      "loss": 11.9264,
      "step": 10677
    },
    {
      "epoch": 1.867191256830601,
      "grad_norm": 0.7771651311295122,
      "learning_rate": 0.0001155350414012871,
      "loss": 12.3668,
      "step": 10678
    },
    {
      "epoch": 1.8673661202185792,
      "grad_norm": 0.704792769091153,
      "learning_rate": 0.00011552105228041099,
      "loss": 12.1992,
      "step": 10679
    },
    {
      "epoch": 1.8675409836065575,
      "grad_norm": 0.6410350540832389,
      "learning_rate": 0.00011550706284829013,
      "loss": 12.4146,
      "step": 10680
    },
    {
      "epoch": 1.8677158469945354,
      "grad_norm": 0.7920581132089323,
      "learning_rate": 0.00011549307310520505,
      "loss": 12.109,
      "step": 10681
    },
    {
      "epoch": 1.8678907103825138,
      "grad_norm": 0.6696442142654635,
      "learning_rate": 0.00011547908305143629,
      "loss": 12.2908,
      "step": 10682
    },
    {
      "epoch": 1.8680655737704916,
      "grad_norm": 0.8462788338991629,
      "learning_rate": 0.00011546509268726438,
      "loss": 12.442,
      "step": 10683
    },
    {
      "epoch": 1.86824043715847,
      "grad_norm": 0.7576080121308814,
      "learning_rate": 0.00011545110201296985,
      "loss": 12.1786,
      "step": 10684
    },
    {
      "epoch": 1.868415300546448,
      "grad_norm": 0.9572206909401481,
      "learning_rate": 0.00011543711102883334,
      "loss": 12.122,
      "step": 10685
    },
    {
      "epoch": 1.8685901639344262,
      "grad_norm": 0.7087734608047881,
      "learning_rate": 0.00011542311973513534,
      "loss": 12.2778,
      "step": 10686
    },
    {
      "epoch": 1.8687650273224043,
      "grad_norm": 0.7246256534336206,
      "learning_rate": 0.00011540912813215641,
      "loss": 12.1698,
      "step": 10687
    },
    {
      "epoch": 1.8689398907103825,
      "grad_norm": 0.8718718138518106,
      "learning_rate": 0.00011539513622017715,
      "loss": 12.3752,
      "step": 10688
    },
    {
      "epoch": 1.8691147540983608,
      "grad_norm": 0.671726301714031,
      "learning_rate": 0.00011538114399947815,
      "loss": 12.3698,
      "step": 10689
    },
    {
      "epoch": 1.8692896174863387,
      "grad_norm": 0.7451307839192589,
      "learning_rate": 0.00011536715147033998,
      "loss": 12.0984,
      "step": 10690
    },
    {
      "epoch": 1.869464480874317,
      "grad_norm": 0.7436032877205784,
      "learning_rate": 0.00011535315863304326,
      "loss": 12.4015,
      "step": 10691
    },
    {
      "epoch": 1.869639344262295,
      "grad_norm": 0.7683499590296042,
      "learning_rate": 0.00011533916548786857,
      "loss": 12.2422,
      "step": 10692
    },
    {
      "epoch": 1.8698142076502733,
      "grad_norm": 0.6657941993998474,
      "learning_rate": 0.00011532517203509648,
      "loss": 12.0953,
      "step": 10693
    },
    {
      "epoch": 1.8699890710382514,
      "grad_norm": 0.6648913262444677,
      "learning_rate": 0.00011531117827500765,
      "loss": 12.2436,
      "step": 10694
    },
    {
      "epoch": 1.8701639344262295,
      "grad_norm": 0.7631596141039421,
      "learning_rate": 0.00011529718420788269,
      "loss": 12.4977,
      "step": 10695
    },
    {
      "epoch": 1.8703387978142076,
      "grad_norm": 0.742781505410907,
      "learning_rate": 0.00011528318983400221,
      "loss": 12.2013,
      "step": 10696
    },
    {
      "epoch": 1.8705136612021858,
      "grad_norm": 0.7785895945274525,
      "learning_rate": 0.00011526919515364687,
      "loss": 12.2705,
      "step": 10697
    },
    {
      "epoch": 1.8706885245901639,
      "grad_norm": 0.6217338473981474,
      "learning_rate": 0.00011525520016709727,
      "loss": 12.2865,
      "step": 10698
    },
    {
      "epoch": 1.870863387978142,
      "grad_norm": 0.7645899029779888,
      "learning_rate": 0.00011524120487463405,
      "loss": 12.282,
      "step": 10699
    },
    {
      "epoch": 1.8710382513661203,
      "grad_norm": 0.6469623395044936,
      "learning_rate": 0.0001152272092765379,
      "loss": 12.1574,
      "step": 10700
    },
    {
      "epoch": 1.8712131147540982,
      "grad_norm": 0.7634625296544232,
      "learning_rate": 0.00011521321337308946,
      "loss": 12.3572,
      "step": 10701
    },
    {
      "epoch": 1.8713879781420766,
      "grad_norm": 0.7144472468560703,
      "learning_rate": 0.00011519921716456936,
      "loss": 12.4743,
      "step": 10702
    },
    {
      "epoch": 1.8715628415300547,
      "grad_norm": 0.8172201083556375,
      "learning_rate": 0.00011518522065125832,
      "loss": 12.2847,
      "step": 10703
    },
    {
      "epoch": 1.8717377049180328,
      "grad_norm": 0.7574779486545595,
      "learning_rate": 0.00011517122383343692,
      "loss": 12.0,
      "step": 10704
    },
    {
      "epoch": 1.871912568306011,
      "grad_norm": 0.7921485573389341,
      "learning_rate": 0.00011515722671138593,
      "loss": 12.2088,
      "step": 10705
    },
    {
      "epoch": 1.872087431693989,
      "grad_norm": 0.9160516653458776,
      "learning_rate": 0.00011514322928538601,
      "loss": 12.1799,
      "step": 10706
    },
    {
      "epoch": 1.8722622950819672,
      "grad_norm": 0.6563118645727016,
      "learning_rate": 0.00011512923155571784,
      "loss": 11.9976,
      "step": 10707
    },
    {
      "epoch": 1.8724371584699453,
      "grad_norm": 0.8180721850408869,
      "learning_rate": 0.00011511523352266213,
      "loss": 12.1018,
      "step": 10708
    },
    {
      "epoch": 1.8726120218579236,
      "grad_norm": 0.6873675055738453,
      "learning_rate": 0.0001151012351864996,
      "loss": 12.1142,
      "step": 10709
    },
    {
      "epoch": 1.8727868852459015,
      "grad_norm": 0.7130705590904474,
      "learning_rate": 0.00011508723654751093,
      "loss": 12.4222,
      "step": 10710
    },
    {
      "epoch": 1.8729617486338799,
      "grad_norm": 0.7545893499401195,
      "learning_rate": 0.00011507323760597682,
      "loss": 12.3311,
      "step": 10711
    },
    {
      "epoch": 1.8731366120218578,
      "grad_norm": 0.7018756419985939,
      "learning_rate": 0.00011505923836217804,
      "loss": 12.2295,
      "step": 10712
    },
    {
      "epoch": 1.8733114754098361,
      "grad_norm": 0.829928696307286,
      "learning_rate": 0.00011504523881639526,
      "loss": 12.1426,
      "step": 10713
    },
    {
      "epoch": 1.8734863387978142,
      "grad_norm": 0.6939107424776823,
      "learning_rate": 0.00011503123896890927,
      "loss": 12.2465,
      "step": 10714
    },
    {
      "epoch": 1.8736612021857924,
      "grad_norm": 0.7944312507267905,
      "learning_rate": 0.00011501723882000079,
      "loss": 12.2664,
      "step": 10715
    },
    {
      "epoch": 1.8738360655737705,
      "grad_norm": 0.7284145422835815,
      "learning_rate": 0.00011500323836995052,
      "loss": 12.2439,
      "step": 10716
    },
    {
      "epoch": 1.8740109289617486,
      "grad_norm": 0.71600094096622,
      "learning_rate": 0.00011498923761903932,
      "loss": 12.2704,
      "step": 10717
    },
    {
      "epoch": 1.874185792349727,
      "grad_norm": 0.7070707279401791,
      "learning_rate": 0.00011497523656754783,
      "loss": 12.315,
      "step": 10718
    },
    {
      "epoch": 1.8743606557377048,
      "grad_norm": 0.7453855765818223,
      "learning_rate": 0.00011496123521575692,
      "loss": 12.1204,
      "step": 10719
    },
    {
      "epoch": 1.8745355191256832,
      "grad_norm": 0.6988226081901474,
      "learning_rate": 0.00011494723356394726,
      "loss": 12.3901,
      "step": 10720
    },
    {
      "epoch": 1.874710382513661,
      "grad_norm": 0.7156404651153867,
      "learning_rate": 0.0001149332316123997,
      "loss": 12.3662,
      "step": 10721
    },
    {
      "epoch": 1.8748852459016394,
      "grad_norm": 0.6720111375370462,
      "learning_rate": 0.00011491922936139499,
      "loss": 12.3254,
      "step": 10722
    },
    {
      "epoch": 1.8750601092896175,
      "grad_norm": 0.7210615956719395,
      "learning_rate": 0.00011490522681121389,
      "loss": 11.9823,
      "step": 10723
    },
    {
      "epoch": 1.8752349726775956,
      "grad_norm": 0.7010795706861056,
      "learning_rate": 0.00011489122396213727,
      "loss": 12.3396,
      "step": 10724
    },
    {
      "epoch": 1.8754098360655738,
      "grad_norm": 0.6927601462891205,
      "learning_rate": 0.00011487722081444585,
      "loss": 12.4576,
      "step": 10725
    },
    {
      "epoch": 1.8755846994535519,
      "grad_norm": 0.6463732489897895,
      "learning_rate": 0.00011486321736842052,
      "loss": 12.2221,
      "step": 10726
    },
    {
      "epoch": 1.87575956284153,
      "grad_norm": 0.7018967127313547,
      "learning_rate": 0.00011484921362434202,
      "loss": 12.01,
      "step": 10727
    },
    {
      "epoch": 1.8759344262295081,
      "grad_norm": 0.6605052516720888,
      "learning_rate": 0.00011483520958249121,
      "loss": 11.8673,
      "step": 10728
    },
    {
      "epoch": 1.8761092896174865,
      "grad_norm": 0.6489052575983075,
      "learning_rate": 0.00011482120524314888,
      "loss": 12.4086,
      "step": 10729
    },
    {
      "epoch": 1.8762841530054644,
      "grad_norm": 0.6113591480722693,
      "learning_rate": 0.00011480720060659587,
      "loss": 12.2234,
      "step": 10730
    },
    {
      "epoch": 1.8764590163934427,
      "grad_norm": 0.8315559771676017,
      "learning_rate": 0.00011479319567311304,
      "loss": 12.1767,
      "step": 10731
    },
    {
      "epoch": 1.8766338797814206,
      "grad_norm": 0.6766219058399662,
      "learning_rate": 0.00011477919044298118,
      "loss": 12.3971,
      "step": 10732
    },
    {
      "epoch": 1.876808743169399,
      "grad_norm": 0.7044475074112283,
      "learning_rate": 0.00011476518491648123,
      "loss": 12.444,
      "step": 10733
    },
    {
      "epoch": 1.876983606557377,
      "grad_norm": 0.6550848617648155,
      "learning_rate": 0.00011475117909389397,
      "loss": 12.2398,
      "step": 10734
    },
    {
      "epoch": 1.8771584699453552,
      "grad_norm": 0.8118767594108653,
      "learning_rate": 0.00011473717297550027,
      "loss": 12.1011,
      "step": 10735
    },
    {
      "epoch": 1.8773333333333333,
      "grad_norm": 0.6411058239933479,
      "learning_rate": 0.000114723166561581,
      "loss": 12.1917,
      "step": 10736
    },
    {
      "epoch": 1.8775081967213114,
      "grad_norm": 0.7971794891589659,
      "learning_rate": 0.00011470915985241702,
      "loss": 12.2994,
      "step": 10737
    },
    {
      "epoch": 1.8776830601092898,
      "grad_norm": 0.6608736294011253,
      "learning_rate": 0.00011469515284828924,
      "loss": 12.1747,
      "step": 10738
    },
    {
      "epoch": 1.8778579234972677,
      "grad_norm": 0.6774080839967894,
      "learning_rate": 0.00011468114554947852,
      "loss": 12.3304,
      "step": 10739
    },
    {
      "epoch": 1.878032786885246,
      "grad_norm": 0.6964940198472557,
      "learning_rate": 0.00011466713795626576,
      "loss": 12.5407,
      "step": 10740
    },
    {
      "epoch": 1.878207650273224,
      "grad_norm": 0.6834646024918175,
      "learning_rate": 0.00011465313006893184,
      "loss": 12.1017,
      "step": 10741
    },
    {
      "epoch": 1.8783825136612022,
      "grad_norm": 0.6501479825643359,
      "learning_rate": 0.0001146391218877577,
      "loss": 12.0734,
      "step": 10742
    },
    {
      "epoch": 1.8785573770491804,
      "grad_norm": 0.6606779209672691,
      "learning_rate": 0.00011462511341302418,
      "loss": 12.3975,
      "step": 10743
    },
    {
      "epoch": 1.8787322404371585,
      "grad_norm": 0.6926577227040837,
      "learning_rate": 0.00011461110464501225,
      "loss": 12.2353,
      "step": 10744
    },
    {
      "epoch": 1.8789071038251366,
      "grad_norm": 0.7174232799283116,
      "learning_rate": 0.0001145970955840028,
      "loss": 12.3926,
      "step": 10745
    },
    {
      "epoch": 1.8790819672131147,
      "grad_norm": 0.7102917217574207,
      "learning_rate": 0.00011458308623027678,
      "loss": 12.3785,
      "step": 10746
    },
    {
      "epoch": 1.879256830601093,
      "grad_norm": 0.5959022106604016,
      "learning_rate": 0.00011456907658411509,
      "loss": 12.2157,
      "step": 10747
    },
    {
      "epoch": 1.879431693989071,
      "grad_norm": 0.6572837019218448,
      "learning_rate": 0.00011455506664579869,
      "loss": 12.1,
      "step": 10748
    },
    {
      "epoch": 1.8796065573770493,
      "grad_norm": 0.6699058004043843,
      "learning_rate": 0.0001145410564156085,
      "loss": 12.1555,
      "step": 10749
    },
    {
      "epoch": 1.8797814207650272,
      "grad_norm": 0.6853622570308944,
      "learning_rate": 0.0001145270458938255,
      "loss": 12.2203,
      "step": 10750
    },
    {
      "epoch": 1.8799562841530055,
      "grad_norm": 0.7341247638424735,
      "learning_rate": 0.00011451303508073066,
      "loss": 12.3806,
      "step": 10751
    },
    {
      "epoch": 1.8801311475409836,
      "grad_norm": 0.6555739167804499,
      "learning_rate": 0.00011449902397660483,
      "loss": 12.3045,
      "step": 10752
    },
    {
      "epoch": 1.8803060109289618,
      "grad_norm": 0.7847244492604145,
      "learning_rate": 0.0001144850125817291,
      "loss": 12.3309,
      "step": 10753
    },
    {
      "epoch": 1.8804808743169399,
      "grad_norm": 0.7119460482001424,
      "learning_rate": 0.00011447100089638439,
      "loss": 12.0751,
      "step": 10754
    },
    {
      "epoch": 1.880655737704918,
      "grad_norm": 0.6862922615093963,
      "learning_rate": 0.00011445698892085166,
      "loss": 12.2328,
      "step": 10755
    },
    {
      "epoch": 1.8808306010928961,
      "grad_norm": 0.6011616924043272,
      "learning_rate": 0.0001144429766554119,
      "loss": 12.2803,
      "step": 10756
    },
    {
      "epoch": 1.8810054644808742,
      "grad_norm": 0.6976595889806708,
      "learning_rate": 0.00011442896410034616,
      "loss": 12.2604,
      "step": 10757
    },
    {
      "epoch": 1.8811803278688526,
      "grad_norm": 0.7816158831613459,
      "learning_rate": 0.00011441495125593538,
      "loss": 12.243,
      "step": 10758
    },
    {
      "epoch": 1.8813551912568305,
      "grad_norm": 0.6438313541293589,
      "learning_rate": 0.00011440093812246054,
      "loss": 12.3978,
      "step": 10759
    },
    {
      "epoch": 1.8815300546448088,
      "grad_norm": 0.759314474662377,
      "learning_rate": 0.00011438692470020269,
      "loss": 12.3658,
      "step": 10760
    },
    {
      "epoch": 1.8817049180327867,
      "grad_norm": 0.7051740493812506,
      "learning_rate": 0.0001143729109894428,
      "loss": 12.3056,
      "step": 10761
    },
    {
      "epoch": 1.881879781420765,
      "grad_norm": 0.6431097598493004,
      "learning_rate": 0.00011435889699046193,
      "loss": 12.334,
      "step": 10762
    },
    {
      "epoch": 1.8820546448087432,
      "grad_norm": 0.6310330956255438,
      "learning_rate": 0.00011434488270354111,
      "loss": 11.9241,
      "step": 10763
    },
    {
      "epoch": 1.8822295081967213,
      "grad_norm": 0.6366924859105733,
      "learning_rate": 0.00011433086812896132,
      "loss": 12.3912,
      "step": 10764
    },
    {
      "epoch": 1.8824043715846994,
      "grad_norm": 0.6871899250886849,
      "learning_rate": 0.00011431685326700365,
      "loss": 12.1031,
      "step": 10765
    },
    {
      "epoch": 1.8825792349726775,
      "grad_norm": 0.6636571325365545,
      "learning_rate": 0.00011430283811794908,
      "loss": 12.3006,
      "step": 10766
    },
    {
      "epoch": 1.8827540983606559,
      "grad_norm": 0.6158632760460322,
      "learning_rate": 0.00011428882268207872,
      "loss": 12.2465,
      "step": 10767
    },
    {
      "epoch": 1.8829289617486338,
      "grad_norm": 0.6024970084315922,
      "learning_rate": 0.00011427480695967356,
      "loss": 12.0763,
      "step": 10768
    },
    {
      "epoch": 1.8831038251366121,
      "grad_norm": 0.6232500973204272,
      "learning_rate": 0.00011426079095101473,
      "loss": 12.259,
      "step": 10769
    },
    {
      "epoch": 1.88327868852459,
      "grad_norm": 0.6436341583388282,
      "learning_rate": 0.00011424677465638325,
      "loss": 12.2024,
      "step": 10770
    },
    {
      "epoch": 1.8834535519125684,
      "grad_norm": 0.6180914305788219,
      "learning_rate": 0.00011423275807606015,
      "loss": 12.0945,
      "step": 10771
    },
    {
      "epoch": 1.8836284153005465,
      "grad_norm": 0.6845796498183676,
      "learning_rate": 0.0001142187412103266,
      "loss": 12.3308,
      "step": 10772
    },
    {
      "epoch": 1.8838032786885246,
      "grad_norm": 0.7339092270275489,
      "learning_rate": 0.00011420472405946358,
      "loss": 12.3793,
      "step": 10773
    },
    {
      "epoch": 1.8839781420765027,
      "grad_norm": 0.6521256599911763,
      "learning_rate": 0.00011419070662375228,
      "loss": 12.284,
      "step": 10774
    },
    {
      "epoch": 1.8841530054644808,
      "grad_norm": 0.6332764770425527,
      "learning_rate": 0.00011417668890347372,
      "loss": 12.1901,
      "step": 10775
    },
    {
      "epoch": 1.8843278688524592,
      "grad_norm": 0.6539200014995443,
      "learning_rate": 0.00011416267089890901,
      "loss": 12.3563,
      "step": 10776
    },
    {
      "epoch": 1.884502732240437,
      "grad_norm": 0.6013200598051623,
      "learning_rate": 0.0001141486526103393,
      "loss": 12.1911,
      "step": 10777
    },
    {
      "epoch": 1.8846775956284154,
      "grad_norm": 0.7600704539615878,
      "learning_rate": 0.00011413463403804563,
      "loss": 12.157,
      "step": 10778
    },
    {
      "epoch": 1.8848524590163933,
      "grad_norm": 0.6792564638485555,
      "learning_rate": 0.00011412061518230914,
      "loss": 12.0927,
      "step": 10779
    },
    {
      "epoch": 1.8850273224043717,
      "grad_norm": 0.6807601385257588,
      "learning_rate": 0.00011410659604341097,
      "loss": 12.1557,
      "step": 10780
    },
    {
      "epoch": 1.8852021857923498,
      "grad_norm": 0.6603429763643458,
      "learning_rate": 0.00011409257662163227,
      "loss": 12.2232,
      "step": 10781
    },
    {
      "epoch": 1.885377049180328,
      "grad_norm": 0.6487583522258137,
      "learning_rate": 0.00011407855691725411,
      "loss": 12.4943,
      "step": 10782
    },
    {
      "epoch": 1.885551912568306,
      "grad_norm": 0.6454669090776016,
      "learning_rate": 0.00011406453693055766,
      "loss": 12.0912,
      "step": 10783
    },
    {
      "epoch": 1.8857267759562841,
      "grad_norm": 0.6048012734508887,
      "learning_rate": 0.00011405051666182405,
      "loss": 12.4102,
      "step": 10784
    },
    {
      "epoch": 1.8859016393442622,
      "grad_norm": 0.6932491012081291,
      "learning_rate": 0.00011403649611133444,
      "loss": 12.143,
      "step": 10785
    },
    {
      "epoch": 1.8860765027322404,
      "grad_norm": 0.7063926086261723,
      "learning_rate": 0.00011402247527937,
      "loss": 12.2468,
      "step": 10786
    },
    {
      "epoch": 1.8862513661202187,
      "grad_norm": 0.6756395565202896,
      "learning_rate": 0.00011400845416621186,
      "loss": 12.2506,
      "step": 10787
    },
    {
      "epoch": 1.8864262295081966,
      "grad_norm": 0.6634741747947239,
      "learning_rate": 0.00011399443277214121,
      "loss": 12.123,
      "step": 10788
    },
    {
      "epoch": 1.886601092896175,
      "grad_norm": 0.6193058092651096,
      "learning_rate": 0.00011398041109743922,
      "loss": 12.1966,
      "step": 10789
    },
    {
      "epoch": 1.8867759562841528,
      "grad_norm": 0.7299814489155592,
      "learning_rate": 0.00011396638914238707,
      "loss": 12.094,
      "step": 10790
    },
    {
      "epoch": 1.8869508196721312,
      "grad_norm": 0.6448493957184277,
      "learning_rate": 0.00011395236690726592,
      "loss": 12.4466,
      "step": 10791
    },
    {
      "epoch": 1.8871256830601093,
      "grad_norm": 0.656541925409768,
      "learning_rate": 0.00011393834439235698,
      "loss": 12.2826,
      "step": 10792
    },
    {
      "epoch": 1.8873005464480874,
      "grad_norm": 0.7333213425656419,
      "learning_rate": 0.00011392432159794142,
      "loss": 12.1189,
      "step": 10793
    },
    {
      "epoch": 1.8874754098360655,
      "grad_norm": 0.7968532544530981,
      "learning_rate": 0.00011391029852430048,
      "loss": 12.1521,
      "step": 10794
    },
    {
      "epoch": 1.8876502732240437,
      "grad_norm": 0.5993618245752648,
      "learning_rate": 0.00011389627517171532,
      "loss": 12.3325,
      "step": 10795
    },
    {
      "epoch": 1.887825136612022,
      "grad_norm": 0.6650573009522296,
      "learning_rate": 0.00011388225154046721,
      "loss": 11.9939,
      "step": 10796
    },
    {
      "epoch": 1.888,
      "grad_norm": 0.7616763988190722,
      "learning_rate": 0.00011386822763083733,
      "loss": 12.4264,
      "step": 10797
    },
    {
      "epoch": 1.8881748633879782,
      "grad_norm": 0.6202508997303093,
      "learning_rate": 0.00011385420344310688,
      "loss": 12.3248,
      "step": 10798
    },
    {
      "epoch": 1.8883497267759561,
      "grad_norm": 0.7245769397572099,
      "learning_rate": 0.00011384017897755715,
      "loss": 12.4033,
      "step": 10799
    },
    {
      "epoch": 1.8885245901639345,
      "grad_norm": 0.6444209464160481,
      "learning_rate": 0.00011382615423446929,
      "loss": 12.3418,
      "step": 10800
    },
    {
      "epoch": 1.8886994535519126,
      "grad_norm": 0.7908378990371715,
      "learning_rate": 0.0001138121292141246,
      "loss": 12.2273,
      "step": 10801
    },
    {
      "epoch": 1.8888743169398907,
      "grad_norm": 0.6547714989361045,
      "learning_rate": 0.00011379810391680432,
      "loss": 12.3552,
      "step": 10802
    },
    {
      "epoch": 1.8890491803278688,
      "grad_norm": 0.7342397100296824,
      "learning_rate": 0.0001137840783427897,
      "loss": 12.2792,
      "step": 10803
    },
    {
      "epoch": 1.889224043715847,
      "grad_norm": 0.7158197552201259,
      "learning_rate": 0.00011377005249236197,
      "loss": 12.2863,
      "step": 10804
    },
    {
      "epoch": 1.8893989071038253,
      "grad_norm": 0.6575597513663182,
      "learning_rate": 0.0001137560263658024,
      "loss": 12.2533,
      "step": 10805
    },
    {
      "epoch": 1.8895737704918032,
      "grad_norm": 0.6634288837852,
      "learning_rate": 0.00011374199996339232,
      "loss": 12.3053,
      "step": 10806
    },
    {
      "epoch": 1.8897486338797815,
      "grad_norm": 0.7077157252810293,
      "learning_rate": 0.00011372797328541287,
      "loss": 12.3502,
      "step": 10807
    },
    {
      "epoch": 1.8899234972677594,
      "grad_norm": 0.6705145295904036,
      "learning_rate": 0.00011371394633214547,
      "loss": 12.1668,
      "step": 10808
    },
    {
      "epoch": 1.8900983606557378,
      "grad_norm": 0.7027229280832705,
      "learning_rate": 0.00011369991910387128,
      "loss": 12.3846,
      "step": 10809
    },
    {
      "epoch": 1.890273224043716,
      "grad_norm": 0.6508895586735183,
      "learning_rate": 0.00011368589160087169,
      "loss": 12.3228,
      "step": 10810
    },
    {
      "epoch": 1.890448087431694,
      "grad_norm": 0.9014906163855991,
      "learning_rate": 0.00011367186382342794,
      "loss": 12.2966,
      "step": 10811
    },
    {
      "epoch": 1.8906229508196721,
      "grad_norm": 0.734658040982748,
      "learning_rate": 0.00011365783577182132,
      "loss": 12.3526,
      "step": 10812
    },
    {
      "epoch": 1.8907978142076503,
      "grad_norm": 0.6782195356004381,
      "learning_rate": 0.00011364380744633321,
      "loss": 12.0987,
      "step": 10813
    },
    {
      "epoch": 1.8909726775956284,
      "grad_norm": 0.6472295109445322,
      "learning_rate": 0.00011362977884724481,
      "loss": 12.0789,
      "step": 10814
    },
    {
      "epoch": 1.8911475409836065,
      "grad_norm": 0.6440549624990242,
      "learning_rate": 0.00011361574997483755,
      "loss": 11.9603,
      "step": 10815
    },
    {
      "epoch": 1.8913224043715848,
      "grad_norm": 0.6466976199650911,
      "learning_rate": 0.00011360172082939265,
      "loss": 12.1713,
      "step": 10816
    },
    {
      "epoch": 1.8914972677595627,
      "grad_norm": 0.6401082854121671,
      "learning_rate": 0.00011358769141119152,
      "loss": 12.448,
      "step": 10817
    },
    {
      "epoch": 1.891672131147541,
      "grad_norm": 0.6759336244348557,
      "learning_rate": 0.00011357366172051545,
      "loss": 12.316,
      "step": 10818
    },
    {
      "epoch": 1.891846994535519,
      "grad_norm": 0.6524925491813907,
      "learning_rate": 0.00011355963175764575,
      "loss": 12.3364,
      "step": 10819
    },
    {
      "epoch": 1.8920218579234973,
      "grad_norm": 0.6584970953421615,
      "learning_rate": 0.00011354560152286384,
      "loss": 12.1572,
      "step": 10820
    },
    {
      "epoch": 1.8921967213114754,
      "grad_norm": 0.651482724897238,
      "learning_rate": 0.000113531571016451,
      "loss": 11.9814,
      "step": 10821
    },
    {
      "epoch": 1.8923715846994535,
      "grad_norm": 0.6855204272259102,
      "learning_rate": 0.00011351754023868865,
      "loss": 12.3575,
      "step": 10822
    },
    {
      "epoch": 1.8925464480874317,
      "grad_norm": 0.6437873483540589,
      "learning_rate": 0.0001135035091898581,
      "loss": 12.3751,
      "step": 10823
    },
    {
      "epoch": 1.8927213114754098,
      "grad_norm": 0.7085142383264157,
      "learning_rate": 0.00011348947787024072,
      "loss": 11.7772,
      "step": 10824
    },
    {
      "epoch": 1.8928961748633881,
      "grad_norm": 0.627720016293472,
      "learning_rate": 0.0001134754462801179,
      "loss": 12.2271,
      "step": 10825
    },
    {
      "epoch": 1.893071038251366,
      "grad_norm": 0.649427073324588,
      "learning_rate": 0.000113461414419771,
      "loss": 12.0496,
      "step": 10826
    },
    {
      "epoch": 1.8932459016393444,
      "grad_norm": 0.726036805440053,
      "learning_rate": 0.0001134473822894814,
      "loss": 12.1211,
      "step": 10827
    },
    {
      "epoch": 1.8934207650273223,
      "grad_norm": 0.7480861518527475,
      "learning_rate": 0.0001134333498895305,
      "loss": 12.4897,
      "step": 10828
    },
    {
      "epoch": 1.8935956284153006,
      "grad_norm": 0.6663913775426953,
      "learning_rate": 0.00011341931722019971,
      "loss": 12.2213,
      "step": 10829
    },
    {
      "epoch": 1.8937704918032787,
      "grad_norm": 0.737965149746535,
      "learning_rate": 0.0001134052842817704,
      "loss": 12.2645,
      "step": 10830
    },
    {
      "epoch": 1.8939453551912568,
      "grad_norm": 0.7368584321861945,
      "learning_rate": 0.00011339125107452395,
      "loss": 12.2543,
      "step": 10831
    },
    {
      "epoch": 1.894120218579235,
      "grad_norm": 0.6549439083394075,
      "learning_rate": 0.00011337721759874185,
      "loss": 12.3057,
      "step": 10832
    },
    {
      "epoch": 1.894295081967213,
      "grad_norm": 0.6432355612187786,
      "learning_rate": 0.00011336318385470542,
      "loss": 12.4273,
      "step": 10833
    },
    {
      "epoch": 1.8944699453551914,
      "grad_norm": 0.6250209048088127,
      "learning_rate": 0.00011334914984269613,
      "loss": 12.3141,
      "step": 10834
    },
    {
      "epoch": 1.8946448087431693,
      "grad_norm": 0.6708172987644676,
      "learning_rate": 0.00011333511556299541,
      "loss": 12.2286,
      "step": 10835
    },
    {
      "epoch": 1.8948196721311477,
      "grad_norm": 0.6830032966844601,
      "learning_rate": 0.00011332108101588468,
      "loss": 12.3779,
      "step": 10836
    },
    {
      "epoch": 1.8949945355191256,
      "grad_norm": 0.6487055442264922,
      "learning_rate": 0.00011330704620164538,
      "loss": 12.1602,
      "step": 10837
    },
    {
      "epoch": 1.895169398907104,
      "grad_norm": 0.6603932853051842,
      "learning_rate": 0.00011329301112055893,
      "loss": 12.3197,
      "step": 10838
    },
    {
      "epoch": 1.895344262295082,
      "grad_norm": 0.7098438978277921,
      "learning_rate": 0.0001132789757729068,
      "loss": 12.2021,
      "step": 10839
    },
    {
      "epoch": 1.8955191256830601,
      "grad_norm": 0.636311652743196,
      "learning_rate": 0.00011326494015897043,
      "loss": 12.4958,
      "step": 10840
    },
    {
      "epoch": 1.8956939890710383,
      "grad_norm": 0.6334549147153029,
      "learning_rate": 0.0001132509042790313,
      "loss": 12.0683,
      "step": 10841
    },
    {
      "epoch": 1.8958688524590164,
      "grad_norm": 0.6604571070923643,
      "learning_rate": 0.00011323686813337083,
      "loss": 12.2398,
      "step": 10842
    },
    {
      "epoch": 1.8960437158469945,
      "grad_norm": 0.6905616066808363,
      "learning_rate": 0.0001132228317222705,
      "loss": 12.4101,
      "step": 10843
    },
    {
      "epoch": 1.8962185792349726,
      "grad_norm": 0.7241229368865222,
      "learning_rate": 0.00011320879504601183,
      "loss": 12.3265,
      "step": 10844
    },
    {
      "epoch": 1.896393442622951,
      "grad_norm": 0.6581773521207416,
      "learning_rate": 0.00011319475810487623,
      "loss": 12.1922,
      "step": 10845
    },
    {
      "epoch": 1.8965683060109289,
      "grad_norm": 0.7197301220881384,
      "learning_rate": 0.00011318072089914524,
      "loss": 12.2794,
      "step": 10846
    },
    {
      "epoch": 1.8967431693989072,
      "grad_norm": 0.7151532407920479,
      "learning_rate": 0.00011316668342910035,
      "loss": 11.8077,
      "step": 10847
    },
    {
      "epoch": 1.896918032786885,
      "grad_norm": 0.6590398532328811,
      "learning_rate": 0.00011315264569502298,
      "loss": 12.2194,
      "step": 10848
    },
    {
      "epoch": 1.8970928961748634,
      "grad_norm": 0.6088363146473238,
      "learning_rate": 0.0001131386076971947,
      "loss": 12.1876,
      "step": 10849
    },
    {
      "epoch": 1.8972677595628415,
      "grad_norm": 0.7354488409070287,
      "learning_rate": 0.00011312456943589698,
      "loss": 12.3146,
      "step": 10850
    },
    {
      "epoch": 1.8974426229508197,
      "grad_norm": 0.6066620480783336,
      "learning_rate": 0.00011311053091141137,
      "loss": 12.1515,
      "step": 10851
    },
    {
      "epoch": 1.8976174863387978,
      "grad_norm": 0.6488881936112345,
      "learning_rate": 0.00011309649212401934,
      "loss": 12.176,
      "step": 10852
    },
    {
      "epoch": 1.897792349726776,
      "grad_norm": 0.7370565739685754,
      "learning_rate": 0.00011308245307400244,
      "loss": 12.2988,
      "step": 10853
    },
    {
      "epoch": 1.8979672131147542,
      "grad_norm": 0.672157941679539,
      "learning_rate": 0.00011306841376164221,
      "loss": 12.2295,
      "step": 10854
    },
    {
      "epoch": 1.8981420765027321,
      "grad_norm": 0.688251103847374,
      "learning_rate": 0.00011305437418722011,
      "loss": 12.156,
      "step": 10855
    },
    {
      "epoch": 1.8983169398907105,
      "grad_norm": 0.7021487264103726,
      "learning_rate": 0.0001130403343510178,
      "loss": 12.2455,
      "step": 10856
    },
    {
      "epoch": 1.8984918032786884,
      "grad_norm": 0.6914865721013999,
      "learning_rate": 0.00011302629425331666,
      "loss": 12.4732,
      "step": 10857
    },
    {
      "epoch": 1.8986666666666667,
      "grad_norm": 0.8092428451163401,
      "learning_rate": 0.00011301225389439838,
      "loss": 12.3039,
      "step": 10858
    },
    {
      "epoch": 1.8988415300546448,
      "grad_norm": 0.696343563742947,
      "learning_rate": 0.00011299821327454445,
      "loss": 12.1567,
      "step": 10859
    },
    {
      "epoch": 1.899016393442623,
      "grad_norm": 0.7710214837141676,
      "learning_rate": 0.0001129841723940364,
      "loss": 12.1111,
      "step": 10860
    },
    {
      "epoch": 1.899191256830601,
      "grad_norm": 0.7274391139405161,
      "learning_rate": 0.00011297013125315587,
      "loss": 12.3114,
      "step": 10861
    },
    {
      "epoch": 1.8993661202185792,
      "grad_norm": 0.679224457166071,
      "learning_rate": 0.00011295608985218435,
      "loss": 12.1967,
      "step": 10862
    },
    {
      "epoch": 1.8995409836065575,
      "grad_norm": 0.7183655893058251,
      "learning_rate": 0.00011294204819140348,
      "loss": 12.5305,
      "step": 10863
    },
    {
      "epoch": 1.8997158469945354,
      "grad_norm": 0.6643522524314204,
      "learning_rate": 0.00011292800627109477,
      "loss": 12.35,
      "step": 10864
    },
    {
      "epoch": 1.8998907103825138,
      "grad_norm": 0.6550881139443193,
      "learning_rate": 0.00011291396409153989,
      "loss": 12.2801,
      "step": 10865
    },
    {
      "epoch": 1.9000655737704917,
      "grad_norm": 0.6066165996803916,
      "learning_rate": 0.00011289992165302035,
      "loss": 12.1691,
      "step": 10866
    },
    {
      "epoch": 1.90024043715847,
      "grad_norm": 0.646881198023037,
      "learning_rate": 0.00011288587895581774,
      "loss": 12.4407,
      "step": 10867
    },
    {
      "epoch": 1.9004153005464481,
      "grad_norm": 0.6588583220725575,
      "learning_rate": 0.00011287183600021375,
      "loss": 12.3412,
      "step": 10868
    },
    {
      "epoch": 1.9005901639344263,
      "grad_norm": 0.6647271417148042,
      "learning_rate": 0.00011285779278648989,
      "loss": 12.2423,
      "step": 10869
    },
    {
      "epoch": 1.9007650273224044,
      "grad_norm": 0.6617396563135864,
      "learning_rate": 0.00011284374931492783,
      "loss": 12.2936,
      "step": 10870
    },
    {
      "epoch": 1.9009398907103825,
      "grad_norm": 0.6821574417196051,
      "learning_rate": 0.00011282970558580914,
      "loss": 12.1541,
      "step": 10871
    },
    {
      "epoch": 1.9011147540983606,
      "grad_norm": 0.670304724731871,
      "learning_rate": 0.00011281566159941548,
      "loss": 12.4189,
      "step": 10872
    },
    {
      "epoch": 1.9012896174863387,
      "grad_norm": 0.711129647648126,
      "learning_rate": 0.00011280161735602846,
      "loss": 12.3163,
      "step": 10873
    },
    {
      "epoch": 1.901464480874317,
      "grad_norm": 0.6503726193011183,
      "learning_rate": 0.00011278757285592967,
      "loss": 12.3762,
      "step": 10874
    },
    {
      "epoch": 1.901639344262295,
      "grad_norm": 0.6978251452952507,
      "learning_rate": 0.00011277352809940081,
      "loss": 12.2743,
      "step": 10875
    },
    {
      "epoch": 1.9018142076502733,
      "grad_norm": 0.6996971502385182,
      "learning_rate": 0.00011275948308672347,
      "loss": 12.2491,
      "step": 10876
    },
    {
      "epoch": 1.9019890710382512,
      "grad_norm": 0.6748680880290795,
      "learning_rate": 0.00011274543781817934,
      "loss": 12.2586,
      "step": 10877
    },
    {
      "epoch": 1.9021639344262296,
      "grad_norm": 0.6480792203379755,
      "learning_rate": 0.00011273139229405003,
      "loss": 12.295,
      "step": 10878
    },
    {
      "epoch": 1.9023387978142077,
      "grad_norm": 0.6719957074192345,
      "learning_rate": 0.00011271734651461722,
      "loss": 12.2157,
      "step": 10879
    },
    {
      "epoch": 1.9025136612021858,
      "grad_norm": 0.5893092322427503,
      "learning_rate": 0.00011270330048016258,
      "loss": 12.2473,
      "step": 10880
    },
    {
      "epoch": 1.902688524590164,
      "grad_norm": 0.6733335583337411,
      "learning_rate": 0.00011268925419096774,
      "loss": 12.2679,
      "step": 10881
    },
    {
      "epoch": 1.902863387978142,
      "grad_norm": 0.6293223447455172,
      "learning_rate": 0.00011267520764731441,
      "loss": 12.3081,
      "step": 10882
    },
    {
      "epoch": 1.9030382513661204,
      "grad_norm": 0.7360878058747793,
      "learning_rate": 0.00011266116084948424,
      "loss": 12.1144,
      "step": 10883
    },
    {
      "epoch": 1.9032131147540983,
      "grad_norm": 0.6777669139122676,
      "learning_rate": 0.00011264711379775892,
      "loss": 12.0512,
      "step": 10884
    },
    {
      "epoch": 1.9033879781420766,
      "grad_norm": 0.6886261236167526,
      "learning_rate": 0.00011263306649242014,
      "loss": 12.3056,
      "step": 10885
    },
    {
      "epoch": 1.9035628415300545,
      "grad_norm": 0.6553630903555844,
      "learning_rate": 0.00011261901893374957,
      "loss": 12.3638,
      "step": 10886
    },
    {
      "epoch": 1.9037377049180328,
      "grad_norm": 0.6881632784535026,
      "learning_rate": 0.00011260497112202895,
      "loss": 12.3108,
      "step": 10887
    },
    {
      "epoch": 1.903912568306011,
      "grad_norm": 0.6989779407627484,
      "learning_rate": 0.00011259092305753993,
      "loss": 12.2303,
      "step": 10888
    },
    {
      "epoch": 1.904087431693989,
      "grad_norm": 0.6175098221654517,
      "learning_rate": 0.00011257687474056426,
      "loss": 12.2574,
      "step": 10889
    },
    {
      "epoch": 1.9042622950819672,
      "grad_norm": 0.6535488232926953,
      "learning_rate": 0.00011256282617138365,
      "loss": 12.0806,
      "step": 10890
    },
    {
      "epoch": 1.9044371584699453,
      "grad_norm": 0.7010592484759763,
      "learning_rate": 0.00011254877735027977,
      "loss": 12.2271,
      "step": 10891
    },
    {
      "epoch": 1.9046120218579234,
      "grad_norm": 0.6867922375135358,
      "learning_rate": 0.00011253472827753442,
      "loss": 12.1596,
      "step": 10892
    },
    {
      "epoch": 1.9047868852459016,
      "grad_norm": 0.6148335879591107,
      "learning_rate": 0.00011252067895342923,
      "loss": 12.219,
      "step": 10893
    },
    {
      "epoch": 1.90496174863388,
      "grad_norm": 0.7416587718582831,
      "learning_rate": 0.00011250662937824601,
      "loss": 11.9547,
      "step": 10894
    },
    {
      "epoch": 1.9051366120218578,
      "grad_norm": 0.6736042815744228,
      "learning_rate": 0.00011249257955226648,
      "loss": 12.2537,
      "step": 10895
    },
    {
      "epoch": 1.9053114754098361,
      "grad_norm": 0.7988725099290506,
      "learning_rate": 0.00011247852947577235,
      "loss": 12.2721,
      "step": 10896
    },
    {
      "epoch": 1.9054863387978143,
      "grad_norm": 0.6050591175953184,
      "learning_rate": 0.00011246447914904543,
      "loss": 12.2399,
      "step": 10897
    },
    {
      "epoch": 1.9056612021857924,
      "grad_norm": 0.8320468820573955,
      "learning_rate": 0.00011245042857236738,
      "loss": 12.3506,
      "step": 10898
    },
    {
      "epoch": 1.9058360655737705,
      "grad_norm": 0.6978773646853764,
      "learning_rate": 0.00011243637774602003,
      "loss": 12.1961,
      "step": 10899
    },
    {
      "epoch": 1.9060109289617486,
      "grad_norm": 0.6739104046378399,
      "learning_rate": 0.00011242232667028512,
      "loss": 11.836,
      "step": 10900
    },
    {
      "epoch": 1.9061857923497267,
      "grad_norm": 0.7561661884842187,
      "learning_rate": 0.00011240827534544443,
      "loss": 12.2103,
      "step": 10901
    },
    {
      "epoch": 1.9063606557377049,
      "grad_norm": 0.7151727369722085,
      "learning_rate": 0.00011239422377177973,
      "loss": 12.0573,
      "step": 10902
    },
    {
      "epoch": 1.9065355191256832,
      "grad_norm": 0.7363177755178257,
      "learning_rate": 0.00011238017194957274,
      "loss": 11.9385,
      "step": 10903
    },
    {
      "epoch": 1.906710382513661,
      "grad_norm": 0.6316894810153365,
      "learning_rate": 0.00011236611987910534,
      "loss": 12.1686,
      "step": 10904
    },
    {
      "epoch": 1.9068852459016394,
      "grad_norm": 0.8184662378930561,
      "learning_rate": 0.00011235206756065924,
      "loss": 12.4356,
      "step": 10905
    },
    {
      "epoch": 1.9070601092896173,
      "grad_norm": 0.6918273084876335,
      "learning_rate": 0.00011233801499451628,
      "loss": 12.1017,
      "step": 10906
    },
    {
      "epoch": 1.9072349726775957,
      "grad_norm": 0.6859495734317084,
      "learning_rate": 0.00011232396218095821,
      "loss": 12.3534,
      "step": 10907
    },
    {
      "epoch": 1.9074098360655738,
      "grad_norm": 0.7706236839369514,
      "learning_rate": 0.00011230990912026686,
      "loss": 12.2092,
      "step": 10908
    },
    {
      "epoch": 1.907584699453552,
      "grad_norm": 0.6372402146437077,
      "learning_rate": 0.00011229585581272408,
      "loss": 12.2298,
      "step": 10909
    },
    {
      "epoch": 1.90775956284153,
      "grad_norm": 0.6586114540212912,
      "learning_rate": 0.00011228180225861158,
      "loss": 12.3101,
      "step": 10910
    },
    {
      "epoch": 1.9079344262295082,
      "grad_norm": 0.6758999195988832,
      "learning_rate": 0.00011226774845821129,
      "loss": 12.2511,
      "step": 10911
    },
    {
      "epoch": 1.9081092896174865,
      "grad_norm": 0.712485547980539,
      "learning_rate": 0.00011225369441180491,
      "loss": 12.387,
      "step": 10912
    },
    {
      "epoch": 1.9082841530054644,
      "grad_norm": 0.6624508928547966,
      "learning_rate": 0.00011223964011967441,
      "loss": 12.3636,
      "step": 10913
    },
    {
      "epoch": 1.9084590163934427,
      "grad_norm": 0.6771508219767437,
      "learning_rate": 0.0001122255855821015,
      "loss": 12.3595,
      "step": 10914
    },
    {
      "epoch": 1.9086338797814206,
      "grad_norm": 0.6276253615311886,
      "learning_rate": 0.00011221153079936806,
      "loss": 12.2028,
      "step": 10915
    },
    {
      "epoch": 1.908808743169399,
      "grad_norm": 0.6689101557684509,
      "learning_rate": 0.00011219747577175595,
      "loss": 12.2858,
      "step": 10916
    },
    {
      "epoch": 1.908983606557377,
      "grad_norm": 0.7022224684173858,
      "learning_rate": 0.00011218342049954696,
      "loss": 12.4067,
      "step": 10917
    },
    {
      "epoch": 1.9091584699453552,
      "grad_norm": 0.6440574475356022,
      "learning_rate": 0.00011216936498302304,
      "loss": 12.2812,
      "step": 10918
    },
    {
      "epoch": 1.9093333333333333,
      "grad_norm": 0.6890501805082712,
      "learning_rate": 0.00011215530922246595,
      "loss": 11.9979,
      "step": 10919
    },
    {
      "epoch": 1.9095081967213114,
      "grad_norm": 0.6300887149192167,
      "learning_rate": 0.0001121412532181576,
      "loss": 12.0961,
      "step": 10920
    },
    {
      "epoch": 1.9096830601092896,
      "grad_norm": 0.6750099997392394,
      "learning_rate": 0.00011212719697037983,
      "loss": 12.3045,
      "step": 10921
    },
    {
      "epoch": 1.9098579234972677,
      "grad_norm": 0.7339789096263423,
      "learning_rate": 0.00011211314047941455,
      "loss": 12.2167,
      "step": 10922
    },
    {
      "epoch": 1.910032786885246,
      "grad_norm": 0.7669682692519533,
      "learning_rate": 0.00011209908374554359,
      "loss": 11.9669,
      "step": 10923
    },
    {
      "epoch": 1.910207650273224,
      "grad_norm": 0.6884983543499428,
      "learning_rate": 0.00011208502676904886,
      "loss": 12.0817,
      "step": 10924
    },
    {
      "epoch": 1.9103825136612023,
      "grad_norm": 0.6985766716884025,
      "learning_rate": 0.00011207096955021224,
      "loss": 12.1298,
      "step": 10925
    },
    {
      "epoch": 1.9105573770491804,
      "grad_norm": 0.7007244087591634,
      "learning_rate": 0.00011205691208931561,
      "loss": 12.3844,
      "step": 10926
    },
    {
      "epoch": 1.9107322404371585,
      "grad_norm": 0.708600933341796,
      "learning_rate": 0.00011204285438664089,
      "loss": 12.2999,
      "step": 10927
    },
    {
      "epoch": 1.9109071038251366,
      "grad_norm": 0.7637411535196259,
      "learning_rate": 0.00011202879644246995,
      "loss": 12.1868,
      "step": 10928
    },
    {
      "epoch": 1.9110819672131147,
      "grad_norm": 0.7032225193125321,
      "learning_rate": 0.00011201473825708471,
      "loss": 12.3599,
      "step": 10929
    },
    {
      "epoch": 1.9112568306010929,
      "grad_norm": 0.7074756780241429,
      "learning_rate": 0.00011200067983076708,
      "loss": 12.2622,
      "step": 10930
    },
    {
      "epoch": 1.911431693989071,
      "grad_norm": 0.697818370892589,
      "learning_rate": 0.00011198662116379897,
      "loss": 12.2013,
      "step": 10931
    },
    {
      "epoch": 1.9116065573770493,
      "grad_norm": 0.7122891069066545,
      "learning_rate": 0.00011197256225646229,
      "loss": 12.1337,
      "step": 10932
    },
    {
      "epoch": 1.9117814207650272,
      "grad_norm": 0.6567146044556199,
      "learning_rate": 0.000111958503109039,
      "loss": 12.4334,
      "step": 10933
    },
    {
      "epoch": 1.9119562841530056,
      "grad_norm": 0.6925700257613836,
      "learning_rate": 0.00011194444372181099,
      "loss": 12.152,
      "step": 10934
    },
    {
      "epoch": 1.9121311475409835,
      "grad_norm": 0.672774758490054,
      "learning_rate": 0.00011193038409506021,
      "loss": 12.1797,
      "step": 10935
    },
    {
      "epoch": 1.9123060109289618,
      "grad_norm": 0.676185192867284,
      "learning_rate": 0.00011191632422906859,
      "loss": 12.1932,
      "step": 10936
    },
    {
      "epoch": 1.91248087431694,
      "grad_norm": 0.7082706470104738,
      "learning_rate": 0.00011190226412411808,
      "loss": 12.338,
      "step": 10937
    },
    {
      "epoch": 1.912655737704918,
      "grad_norm": 0.6979462474785121,
      "learning_rate": 0.00011188820378049065,
      "loss": 12.3768,
      "step": 10938
    },
    {
      "epoch": 1.9128306010928962,
      "grad_norm": 0.7804413124695068,
      "learning_rate": 0.00011187414319846823,
      "loss": 12.0384,
      "step": 10939
    },
    {
      "epoch": 1.9130054644808743,
      "grad_norm": 0.7001823807138132,
      "learning_rate": 0.00011186008237833277,
      "loss": 12.3419,
      "step": 10940
    },
    {
      "epoch": 1.9131803278688526,
      "grad_norm": 0.6401458398057861,
      "learning_rate": 0.00011184602132036626,
      "loss": 12.0903,
      "step": 10941
    },
    {
      "epoch": 1.9133551912568305,
      "grad_norm": 0.6584731660590457,
      "learning_rate": 0.00011183196002485064,
      "loss": 12.3054,
      "step": 10942
    },
    {
      "epoch": 1.9135300546448089,
      "grad_norm": 0.6716356049092078,
      "learning_rate": 0.0001118178984920679,
      "loss": 12.2373,
      "step": 10943
    },
    {
      "epoch": 1.9137049180327868,
      "grad_norm": 0.686132098895601,
      "learning_rate": 0.00011180383672229997,
      "loss": 12.1238,
      "step": 10944
    },
    {
      "epoch": 1.913879781420765,
      "grad_norm": 0.7344354944565885,
      "learning_rate": 0.00011178977471582892,
      "loss": 12.3188,
      "step": 10945
    },
    {
      "epoch": 1.9140546448087432,
      "grad_norm": 0.6774295891469537,
      "learning_rate": 0.00011177571247293665,
      "loss": 12.3409,
      "step": 10946
    },
    {
      "epoch": 1.9142295081967213,
      "grad_norm": 0.6967457120655007,
      "learning_rate": 0.00011176164999390522,
      "loss": 12.1834,
      "step": 10947
    },
    {
      "epoch": 1.9144043715846994,
      "grad_norm": 0.752047057114326,
      "learning_rate": 0.00011174758727901657,
      "loss": 12.2783,
      "step": 10948
    },
    {
      "epoch": 1.9145792349726776,
      "grad_norm": 0.6772557657154468,
      "learning_rate": 0.00011173352432855273,
      "loss": 12.4089,
      "step": 10949
    },
    {
      "epoch": 1.9147540983606557,
      "grad_norm": 0.7100053506100534,
      "learning_rate": 0.00011171946114279575,
      "loss": 12.3527,
      "step": 10950
    },
    {
      "epoch": 1.9149289617486338,
      "grad_norm": 0.7426101165057826,
      "learning_rate": 0.0001117053977220275,
      "loss": 12.4475,
      "step": 10951
    },
    {
      "epoch": 1.9151038251366121,
      "grad_norm": 0.6899933361208874,
      "learning_rate": 0.00011169133406653016,
      "loss": 11.9782,
      "step": 10952
    },
    {
      "epoch": 1.91527868852459,
      "grad_norm": 0.6576284859694288,
      "learning_rate": 0.00011167727017658562,
      "loss": 12.319,
      "step": 10953
    },
    {
      "epoch": 1.9154535519125684,
      "grad_norm": 0.7113145219176392,
      "learning_rate": 0.00011166320605247602,
      "loss": 12.1731,
      "step": 10954
    },
    {
      "epoch": 1.9156284153005463,
      "grad_norm": 0.6474165360876366,
      "learning_rate": 0.00011164914169448328,
      "loss": 12.2823,
      "step": 10955
    },
    {
      "epoch": 1.9158032786885246,
      "grad_norm": 0.7053385519505976,
      "learning_rate": 0.0001116350771028895,
      "loss": 12.163,
      "step": 10956
    },
    {
      "epoch": 1.9159781420765027,
      "grad_norm": 0.7018440724868943,
      "learning_rate": 0.0001116210122779767,
      "loss": 12.2376,
      "step": 10957
    },
    {
      "epoch": 1.9161530054644809,
      "grad_norm": 0.7525084416290988,
      "learning_rate": 0.00011160694722002692,
      "loss": 12.2502,
      "step": 10958
    },
    {
      "epoch": 1.916327868852459,
      "grad_norm": 0.7184650093979285,
      "learning_rate": 0.00011159288192932223,
      "loss": 12.4588,
      "step": 10959
    },
    {
      "epoch": 1.916502732240437,
      "grad_norm": 0.6680517578629179,
      "learning_rate": 0.0001115788164061446,
      "loss": 12.5748,
      "step": 10960
    },
    {
      "epoch": 1.9166775956284154,
      "grad_norm": 0.7233069425805277,
      "learning_rate": 0.0001115647506507762,
      "loss": 12.464,
      "step": 10961
    },
    {
      "epoch": 1.9168524590163933,
      "grad_norm": 0.6937423964755474,
      "learning_rate": 0.00011155068466349905,
      "loss": 12.2834,
      "step": 10962
    },
    {
      "epoch": 1.9170273224043717,
      "grad_norm": 0.6765487231674228,
      "learning_rate": 0.00011153661844459517,
      "loss": 12.3525,
      "step": 10963
    },
    {
      "epoch": 1.9172021857923496,
      "grad_norm": 0.7054280081256779,
      "learning_rate": 0.00011152255199434671,
      "loss": 12.3378,
      "step": 10964
    },
    {
      "epoch": 1.917377049180328,
      "grad_norm": 0.6408452350361329,
      "learning_rate": 0.00011150848531303567,
      "loss": 12.3986,
      "step": 10965
    },
    {
      "epoch": 1.917551912568306,
      "grad_norm": 0.6618381682131772,
      "learning_rate": 0.0001114944184009442,
      "loss": 12.1061,
      "step": 10966
    },
    {
      "epoch": 1.9177267759562842,
      "grad_norm": 0.6468886708174111,
      "learning_rate": 0.0001114803512583543,
      "loss": 12.2874,
      "step": 10967
    },
    {
      "epoch": 1.9179016393442623,
      "grad_norm": 0.7818758332853372,
      "learning_rate": 0.00011146628388554815,
      "loss": 12.1635,
      "step": 10968
    },
    {
      "epoch": 1.9180765027322404,
      "grad_norm": 0.6065092813498844,
      "learning_rate": 0.00011145221628280777,
      "loss": 11.8878,
      "step": 10969
    },
    {
      "epoch": 1.9182513661202187,
      "grad_norm": 0.6651913437514316,
      "learning_rate": 0.00011143814845041529,
      "loss": 12.2938,
      "step": 10970
    },
    {
      "epoch": 1.9184262295081966,
      "grad_norm": 0.7519967392956615,
      "learning_rate": 0.00011142408038865284,
      "loss": 12.3907,
      "step": 10971
    },
    {
      "epoch": 1.918601092896175,
      "grad_norm": 0.6881734609064896,
      "learning_rate": 0.00011141001209780249,
      "loss": 11.8845,
      "step": 10972
    },
    {
      "epoch": 1.9187759562841529,
      "grad_norm": 0.6796687914232183,
      "learning_rate": 0.00011139594357814635,
      "loss": 12.3122,
      "step": 10973
    },
    {
      "epoch": 1.9189508196721312,
      "grad_norm": 0.6661704270609288,
      "learning_rate": 0.00011138187482996658,
      "loss": 12.1679,
      "step": 10974
    },
    {
      "epoch": 1.9191256830601093,
      "grad_norm": 0.678108775340567,
      "learning_rate": 0.00011136780585354525,
      "loss": 12.1383,
      "step": 10975
    },
    {
      "epoch": 1.9193005464480875,
      "grad_norm": 0.6927366806289179,
      "learning_rate": 0.00011135373664916451,
      "loss": 12.4656,
      "step": 10976
    },
    {
      "epoch": 1.9194754098360656,
      "grad_norm": 0.8995722899752441,
      "learning_rate": 0.00011133966721710649,
      "loss": 12.1045,
      "step": 10977
    },
    {
      "epoch": 1.9196502732240437,
      "grad_norm": 0.7333009541019903,
      "learning_rate": 0.00011132559755765335,
      "loss": 12.1508,
      "step": 10978
    },
    {
      "epoch": 1.9198251366120218,
      "grad_norm": 0.8552653736427931,
      "learning_rate": 0.00011131152767108717,
      "loss": 12.2431,
      "step": 10979
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7216401275321261,
      "learning_rate": 0.00011129745755769014,
      "loss": 11.8586,
      "step": 10980
    },
    {
      "epoch": 1.9201748633879783,
      "grad_norm": 0.7008555799532895,
      "learning_rate": 0.00011128338721774441,
      "loss": 12.2575,
      "step": 10981
    },
    {
      "epoch": 1.9203497267759562,
      "grad_norm": 0.7656830697457011,
      "learning_rate": 0.00011126931665153212,
      "loss": 12.0501,
      "step": 10982
    },
    {
      "epoch": 1.9205245901639345,
      "grad_norm": 0.7126790467916868,
      "learning_rate": 0.00011125524585933542,
      "loss": 12.3696,
      "step": 10983
    },
    {
      "epoch": 1.9206994535519124,
      "grad_norm": 0.81221538870503,
      "learning_rate": 0.0001112411748414365,
      "loss": 12.6383,
      "step": 10984
    },
    {
      "epoch": 1.9208743169398907,
      "grad_norm": 0.7165059310033284,
      "learning_rate": 0.00011122710359811748,
      "loss": 12.1954,
      "step": 10985
    },
    {
      "epoch": 1.9210491803278689,
      "grad_norm": 0.6943608997122616,
      "learning_rate": 0.00011121303212966059,
      "loss": 12.1418,
      "step": 10986
    },
    {
      "epoch": 1.921224043715847,
      "grad_norm": 0.744806918717403,
      "learning_rate": 0.00011119896043634797,
      "loss": 12.2718,
      "step": 10987
    },
    {
      "epoch": 1.921398907103825,
      "grad_norm": 0.653177278254652,
      "learning_rate": 0.00011118488851846178,
      "loss": 12.1978,
      "step": 10988
    },
    {
      "epoch": 1.9215737704918032,
      "grad_norm": 0.6456790867722251,
      "learning_rate": 0.00011117081637628426,
      "loss": 12.3161,
      "step": 10989
    },
    {
      "epoch": 1.9217486338797816,
      "grad_norm": 0.7014499882505626,
      "learning_rate": 0.00011115674401009757,
      "loss": 12.286,
      "step": 10990
    },
    {
      "epoch": 1.9219234972677595,
      "grad_norm": 0.751103164914308,
      "learning_rate": 0.00011114267142018392,
      "loss": 12.0784,
      "step": 10991
    },
    {
      "epoch": 1.9220983606557378,
      "grad_norm": 0.6589109076192879,
      "learning_rate": 0.00011112859860682547,
      "loss": 12.2676,
      "step": 10992
    },
    {
      "epoch": 1.9222732240437157,
      "grad_norm": 0.6807344845578593,
      "learning_rate": 0.00011111452557030449,
      "loss": 12.1338,
      "step": 10993
    },
    {
      "epoch": 1.922448087431694,
      "grad_norm": 0.6269509757479722,
      "learning_rate": 0.00011110045231090309,
      "loss": 12.1703,
      "step": 10994
    },
    {
      "epoch": 1.9226229508196722,
      "grad_norm": 0.6632926417238296,
      "learning_rate": 0.00011108637882890356,
      "loss": 12.1818,
      "step": 10995
    },
    {
      "epoch": 1.9227978142076503,
      "grad_norm": 0.6108398182894433,
      "learning_rate": 0.00011107230512458811,
      "loss": 12.1513,
      "step": 10996
    },
    {
      "epoch": 1.9229726775956284,
      "grad_norm": 0.7141646286299497,
      "learning_rate": 0.00011105823119823893,
      "loss": 12.4884,
      "step": 10997
    },
    {
      "epoch": 1.9231475409836065,
      "grad_norm": 0.636422923024968,
      "learning_rate": 0.00011104415705013829,
      "loss": 12.2526,
      "step": 10998
    },
    {
      "epoch": 1.9233224043715849,
      "grad_norm": 0.7210893007201145,
      "learning_rate": 0.00011103008268056834,
      "loss": 12.0684,
      "step": 10999
    },
    {
      "epoch": 1.9234972677595628,
      "grad_norm": 0.6569406748803778,
      "learning_rate": 0.00011101600808981141,
      "loss": 12.1767,
      "step": 11000
    },
    {
      "epoch": 1.923672131147541,
      "grad_norm": 0.6462836757689455,
      "learning_rate": 0.00011100193327814964,
      "loss": 12.2354,
      "step": 11001
    },
    {
      "epoch": 1.923846994535519,
      "grad_norm": 0.6595667064184937,
      "learning_rate": 0.0001109878582458654,
      "loss": 12.3075,
      "step": 11002
    },
    {
      "epoch": 1.9240218579234973,
      "grad_norm": 0.7335438775600093,
      "learning_rate": 0.00011097378299324084,
      "loss": 12.0935,
      "step": 11003
    },
    {
      "epoch": 1.9241967213114755,
      "grad_norm": 0.6745726708099875,
      "learning_rate": 0.00011095970752055819,
      "loss": 12.2245,
      "step": 11004
    },
    {
      "epoch": 1.9243715846994536,
      "grad_norm": 0.6161456786237123,
      "learning_rate": 0.00011094563182809982,
      "loss": 12.0713,
      "step": 11005
    },
    {
      "epoch": 1.9245464480874317,
      "grad_norm": 0.7184335210190956,
      "learning_rate": 0.00011093155591614787,
      "loss": 12.3023,
      "step": 11006
    },
    {
      "epoch": 1.9247213114754098,
      "grad_norm": 0.7170080777325899,
      "learning_rate": 0.0001109174797849847,
      "loss": 12.1147,
      "step": 11007
    },
    {
      "epoch": 1.924896174863388,
      "grad_norm": 0.8169397925609353,
      "learning_rate": 0.0001109034034348925,
      "loss": 12.0821,
      "step": 11008
    },
    {
      "epoch": 1.925071038251366,
      "grad_norm": 0.8255642707348549,
      "learning_rate": 0.00011088932686615363,
      "loss": 12.1116,
      "step": 11009
    },
    {
      "epoch": 1.9252459016393444,
      "grad_norm": 0.7951070042541463,
      "learning_rate": 0.00011087525007905031,
      "loss": 12.1892,
      "step": 11010
    },
    {
      "epoch": 1.9254207650273223,
      "grad_norm": 0.7611743364648156,
      "learning_rate": 0.0001108611730738648,
      "loss": 12.0285,
      "step": 11011
    },
    {
      "epoch": 1.9255956284153006,
      "grad_norm": 0.6693653540191639,
      "learning_rate": 0.00011084709585087947,
      "loss": 12.2995,
      "step": 11012
    },
    {
      "epoch": 1.9257704918032785,
      "grad_norm": 0.7812467038695071,
      "learning_rate": 0.0001108330184103765,
      "loss": 12.056,
      "step": 11013
    },
    {
      "epoch": 1.9259453551912569,
      "grad_norm": 0.6558112948406994,
      "learning_rate": 0.0001108189407526383,
      "loss": 12.2729,
      "step": 11014
    },
    {
      "epoch": 1.926120218579235,
      "grad_norm": 0.7040191538822996,
      "learning_rate": 0.00011080486287794712,
      "loss": 12.3201,
      "step": 11015
    },
    {
      "epoch": 1.926295081967213,
      "grad_norm": 0.713362382077782,
      "learning_rate": 0.00011079078478658525,
      "loss": 12.246,
      "step": 11016
    },
    {
      "epoch": 1.9264699453551912,
      "grad_norm": 0.6847573592160284,
      "learning_rate": 0.000110776706478835,
      "loss": 11.98,
      "step": 11017
    },
    {
      "epoch": 1.9266448087431693,
      "grad_norm": 0.7182529264179015,
      "learning_rate": 0.00011076262795497871,
      "loss": 12.2773,
      "step": 11018
    },
    {
      "epoch": 1.9268196721311477,
      "grad_norm": 0.6764396999613442,
      "learning_rate": 0.00011074854921529869,
      "loss": 12.3921,
      "step": 11019
    },
    {
      "epoch": 1.9269945355191256,
      "grad_norm": 0.6919141378247681,
      "learning_rate": 0.00011073447026007724,
      "loss": 12.2825,
      "step": 11020
    },
    {
      "epoch": 1.927169398907104,
      "grad_norm": 0.676526462385471,
      "learning_rate": 0.00011072039108959673,
      "loss": 12.2189,
      "step": 11021
    },
    {
      "epoch": 1.9273442622950818,
      "grad_norm": 0.7154285172415649,
      "learning_rate": 0.00011070631170413944,
      "loss": 12.3554,
      "step": 11022
    },
    {
      "epoch": 1.9275191256830602,
      "grad_norm": 0.6674789672239002,
      "learning_rate": 0.00011069223210398774,
      "loss": 12.3403,
      "step": 11023
    },
    {
      "epoch": 1.9276939890710383,
      "grad_norm": 0.6517923273448106,
      "learning_rate": 0.00011067815228942394,
      "loss": 12.2638,
      "step": 11024
    },
    {
      "epoch": 1.9278688524590164,
      "grad_norm": 0.714416364397663,
      "learning_rate": 0.00011066407226073041,
      "loss": 12.3249,
      "step": 11025
    },
    {
      "epoch": 1.9280437158469945,
      "grad_norm": 0.6725235947862057,
      "learning_rate": 0.00011064999201818948,
      "loss": 12.1212,
      "step": 11026
    },
    {
      "epoch": 1.9282185792349726,
      "grad_norm": 0.6695461067069584,
      "learning_rate": 0.00011063591156208351,
      "loss": 12.0334,
      "step": 11027
    },
    {
      "epoch": 1.928393442622951,
      "grad_norm": 0.7109745627326449,
      "learning_rate": 0.00011062183089269487,
      "loss": 12.2197,
      "step": 11028
    },
    {
      "epoch": 1.9285683060109289,
      "grad_norm": 0.6487239617406464,
      "learning_rate": 0.0001106077500103059,
      "loss": 12.2668,
      "step": 11029
    },
    {
      "epoch": 1.9287431693989072,
      "grad_norm": 0.6993795996845925,
      "learning_rate": 0.00011059366891519898,
      "loss": 12.2556,
      "step": 11030
    },
    {
      "epoch": 1.9289180327868851,
      "grad_norm": 0.9605737666815597,
      "learning_rate": 0.00011057958760765645,
      "loss": 12.2089,
      "step": 11031
    },
    {
      "epoch": 1.9290928961748635,
      "grad_norm": 0.7655765348213882,
      "learning_rate": 0.00011056550608796072,
      "loss": 12.299,
      "step": 11032
    },
    {
      "epoch": 1.9292677595628416,
      "grad_norm": 0.6677114790121761,
      "learning_rate": 0.00011055142435639416,
      "loss": 12.2147,
      "step": 11033
    },
    {
      "epoch": 1.9294426229508197,
      "grad_norm": 0.6867767007913587,
      "learning_rate": 0.00011053734241323914,
      "loss": 11.9676,
      "step": 11034
    },
    {
      "epoch": 1.9296174863387978,
      "grad_norm": 0.7421990565095635,
      "learning_rate": 0.00011052326025877804,
      "loss": 12.2499,
      "step": 11035
    },
    {
      "epoch": 1.929792349726776,
      "grad_norm": 0.7014005351096636,
      "learning_rate": 0.00011050917789329327,
      "loss": 12.1946,
      "step": 11036
    },
    {
      "epoch": 1.929967213114754,
      "grad_norm": 0.6766476302635738,
      "learning_rate": 0.0001104950953170672,
      "loss": 11.8731,
      "step": 11037
    },
    {
      "epoch": 1.9301420765027322,
      "grad_norm": 0.713721049419984,
      "learning_rate": 0.00011048101253038228,
      "loss": 12.1746,
      "step": 11038
    },
    {
      "epoch": 1.9303169398907105,
      "grad_norm": 0.7397532814594053,
      "learning_rate": 0.00011046692953352087,
      "loss": 12.2297,
      "step": 11039
    },
    {
      "epoch": 1.9304918032786884,
      "grad_norm": 0.7109024110311387,
      "learning_rate": 0.00011045284632676536,
      "loss": 12.4084,
      "step": 11040
    },
    {
      "epoch": 1.9306666666666668,
      "grad_norm": 0.6477801983367123,
      "learning_rate": 0.00011043876291039821,
      "loss": 12.354,
      "step": 11041
    },
    {
      "epoch": 1.9308415300546447,
      "grad_norm": 0.7040642047474225,
      "learning_rate": 0.00011042467928470177,
      "loss": 12.2928,
      "step": 11042
    },
    {
      "epoch": 1.931016393442623,
      "grad_norm": 0.7392775176724999,
      "learning_rate": 0.00011041059544995855,
      "loss": 12.3893,
      "step": 11043
    },
    {
      "epoch": 1.931191256830601,
      "grad_norm": 0.6863310482759213,
      "learning_rate": 0.00011039651140645091,
      "loss": 12.303,
      "step": 11044
    },
    {
      "epoch": 1.9313661202185792,
      "grad_norm": 0.7796181773302151,
      "learning_rate": 0.0001103824271544613,
      "loss": 12.2449,
      "step": 11045
    },
    {
      "epoch": 1.9315409836065573,
      "grad_norm": 0.708399772158717,
      "learning_rate": 0.00011036834269427214,
      "loss": 12.2864,
      "step": 11046
    },
    {
      "epoch": 1.9317158469945355,
      "grad_norm": 0.6134468321608076,
      "learning_rate": 0.00011035425802616586,
      "loss": 12.3538,
      "step": 11047
    },
    {
      "epoch": 1.9318907103825138,
      "grad_norm": 0.795576123855212,
      "learning_rate": 0.00011034017315042495,
      "loss": 12.2641,
      "step": 11048
    },
    {
      "epoch": 1.9320655737704917,
      "grad_norm": 0.6741366618680986,
      "learning_rate": 0.00011032608806733178,
      "loss": 12.1211,
      "step": 11049
    },
    {
      "epoch": 1.93224043715847,
      "grad_norm": 0.926347202238897,
      "learning_rate": 0.00011031200277716889,
      "loss": 12.5891,
      "step": 11050
    },
    {
      "epoch": 1.932415300546448,
      "grad_norm": 0.6508069953421324,
      "learning_rate": 0.00011029791728021865,
      "loss": 12.2941,
      "step": 11051
    },
    {
      "epoch": 1.9325901639344263,
      "grad_norm": 0.7513660609813528,
      "learning_rate": 0.00011028383157676355,
      "loss": 12.0904,
      "step": 11052
    },
    {
      "epoch": 1.9327650273224044,
      "grad_norm": 0.6869648857055504,
      "learning_rate": 0.00011026974566708606,
      "loss": 12.1595,
      "step": 11053
    },
    {
      "epoch": 1.9329398907103825,
      "grad_norm": 0.6916542511929946,
      "learning_rate": 0.0001102556595514686,
      "loss": 12.1933,
      "step": 11054
    },
    {
      "epoch": 1.9331147540983606,
      "grad_norm": 0.8363892764672229,
      "learning_rate": 0.00011024157323019373,
      "loss": 12.4437,
      "step": 11055
    },
    {
      "epoch": 1.9332896174863388,
      "grad_norm": 0.7085483798929427,
      "learning_rate": 0.00011022748670354384,
      "loss": 12.3527,
      "step": 11056
    },
    {
      "epoch": 1.933464480874317,
      "grad_norm": 0.7132337204150467,
      "learning_rate": 0.00011021339997180147,
      "loss": 12.3895,
      "step": 11057
    },
    {
      "epoch": 1.933639344262295,
      "grad_norm": 0.7096592786886853,
      "learning_rate": 0.00011019931303524904,
      "loss": 12.226,
      "step": 11058
    },
    {
      "epoch": 1.9338142076502733,
      "grad_norm": 0.6379775495047113,
      "learning_rate": 0.00011018522589416907,
      "loss": 12.4108,
      "step": 11059
    },
    {
      "epoch": 1.9339890710382512,
      "grad_norm": 0.7512106331598007,
      "learning_rate": 0.00011017113854884407,
      "loss": 12.2605,
      "step": 11060
    },
    {
      "epoch": 1.9341639344262296,
      "grad_norm": 0.6598372476929392,
      "learning_rate": 0.00011015705099955646,
      "loss": 12.1162,
      "step": 11061
    },
    {
      "epoch": 1.9343387978142077,
      "grad_norm": 0.7571854093004992,
      "learning_rate": 0.00011014296324658886,
      "loss": 12.2881,
      "step": 11062
    },
    {
      "epoch": 1.9345136612021858,
      "grad_norm": 0.6160176552669577,
      "learning_rate": 0.00011012887529022367,
      "loss": 12.0605,
      "step": 11063
    },
    {
      "epoch": 1.934688524590164,
      "grad_norm": 0.6820118078490799,
      "learning_rate": 0.00011011478713074343,
      "loss": 12.282,
      "step": 11064
    },
    {
      "epoch": 1.934863387978142,
      "grad_norm": 0.7476055099504951,
      "learning_rate": 0.00011010069876843066,
      "loss": 12.3508,
      "step": 11065
    },
    {
      "epoch": 1.9350382513661202,
      "grad_norm": 0.6525165103188423,
      "learning_rate": 0.00011008661020356785,
      "loss": 12.0323,
      "step": 11066
    },
    {
      "epoch": 1.9352131147540983,
      "grad_norm": 0.8036365179888544,
      "learning_rate": 0.00011007252143643754,
      "loss": 12.2271,
      "step": 11067
    },
    {
      "epoch": 1.9353879781420766,
      "grad_norm": 0.6535894400508389,
      "learning_rate": 0.00011005843246732226,
      "loss": 12.1939,
      "step": 11068
    },
    {
      "epoch": 1.9355628415300545,
      "grad_norm": 0.6981212787886519,
      "learning_rate": 0.00011004434329650452,
      "loss": 12.1504,
      "step": 11069
    },
    {
      "epoch": 1.9357377049180329,
      "grad_norm": 0.6820982844197849,
      "learning_rate": 0.00011003025392426687,
      "loss": 12.2431,
      "step": 11070
    },
    {
      "epoch": 1.9359125683060108,
      "grad_norm": 0.6368239149047031,
      "learning_rate": 0.00011001616435089182,
      "loss": 11.8366,
      "step": 11071
    },
    {
      "epoch": 1.9360874316939891,
      "grad_norm": 0.6892758349525264,
      "learning_rate": 0.00011000207457666191,
      "loss": 12.2847,
      "step": 11072
    },
    {
      "epoch": 1.9362622950819672,
      "grad_norm": 0.8416141581363508,
      "learning_rate": 0.00010998798460185971,
      "loss": 12.2337,
      "step": 11073
    },
    {
      "epoch": 1.9364371584699454,
      "grad_norm": 0.7263455782426519,
      "learning_rate": 0.00010997389442676773,
      "loss": 12.4374,
      "step": 11074
    },
    {
      "epoch": 1.9366120218579235,
      "grad_norm": 0.6868643174799416,
      "learning_rate": 0.00010995980405166857,
      "loss": 12.1369,
      "step": 11075
    },
    {
      "epoch": 1.9367868852459016,
      "grad_norm": 0.6868354027068615,
      "learning_rate": 0.00010994571347684475,
      "loss": 12.2909,
      "step": 11076
    },
    {
      "epoch": 1.93696174863388,
      "grad_norm": 0.7621521829186043,
      "learning_rate": 0.00010993162270257882,
      "loss": 11.9521,
      "step": 11077
    },
    {
      "epoch": 1.9371366120218578,
      "grad_norm": 0.6453980621893535,
      "learning_rate": 0.00010991753172915338,
      "loss": 12.0638,
      "step": 11078
    },
    {
      "epoch": 1.9373114754098362,
      "grad_norm": 0.8616234996478886,
      "learning_rate": 0.00010990344055685098,
      "loss": 12.3328,
      "step": 11079
    },
    {
      "epoch": 1.937486338797814,
      "grad_norm": 0.6464254848662898,
      "learning_rate": 0.00010988934918595418,
      "loss": 12.2182,
      "step": 11080
    },
    {
      "epoch": 1.9376612021857924,
      "grad_norm": 0.794289396970802,
      "learning_rate": 0.00010987525761674558,
      "loss": 12.1818,
      "step": 11081
    },
    {
      "epoch": 1.9378360655737705,
      "grad_norm": 0.7307373140129121,
      "learning_rate": 0.00010986116584950774,
      "loss": 12.2783,
      "step": 11082
    },
    {
      "epoch": 1.9380109289617486,
      "grad_norm": 0.7103532884507617,
      "learning_rate": 0.00010984707388452322,
      "loss": 11.9301,
      "step": 11083
    },
    {
      "epoch": 1.9381857923497268,
      "grad_norm": 0.6966457503678173,
      "learning_rate": 0.00010983298172207463,
      "loss": 12.3252,
      "step": 11084
    },
    {
      "epoch": 1.9383606557377049,
      "grad_norm": 0.6830654005160322,
      "learning_rate": 0.0001098188893624446,
      "loss": 12.1734,
      "step": 11085
    },
    {
      "epoch": 1.9385355191256832,
      "grad_norm": 0.7126285150886901,
      "learning_rate": 0.00010980479680591569,
      "loss": 12.2074,
      "step": 11086
    },
    {
      "epoch": 1.9387103825136611,
      "grad_norm": 0.6517033041739054,
      "learning_rate": 0.0001097907040527705,
      "loss": 12.2713,
      "step": 11087
    },
    {
      "epoch": 1.9388852459016395,
      "grad_norm": 0.8196803004427793,
      "learning_rate": 0.0001097766111032916,
      "loss": 12.4283,
      "step": 11088
    },
    {
      "epoch": 1.9390601092896174,
      "grad_norm": 0.7853927611061063,
      "learning_rate": 0.00010976251795776167,
      "loss": 12.1277,
      "step": 11089
    },
    {
      "epoch": 1.9392349726775957,
      "grad_norm": 0.6824588593422259,
      "learning_rate": 0.00010974842461646324,
      "loss": 12.3552,
      "step": 11090
    },
    {
      "epoch": 1.9394098360655738,
      "grad_norm": 0.7854356163812259,
      "learning_rate": 0.00010973433107967902,
      "loss": 12.3364,
      "step": 11091
    },
    {
      "epoch": 1.939584699453552,
      "grad_norm": 0.6693647403675527,
      "learning_rate": 0.0001097202373476915,
      "loss": 11.9897,
      "step": 11092
    },
    {
      "epoch": 1.93975956284153,
      "grad_norm": 0.7416416822596555,
      "learning_rate": 0.0001097061434207834,
      "loss": 12.3228,
      "step": 11093
    },
    {
      "epoch": 1.9399344262295082,
      "grad_norm": 0.734196800815501,
      "learning_rate": 0.00010969204929923736,
      "loss": 12.3468,
      "step": 11094
    },
    {
      "epoch": 1.9401092896174863,
      "grad_norm": 0.6638671047857551,
      "learning_rate": 0.00010967795498333593,
      "loss": 12.3251,
      "step": 11095
    },
    {
      "epoch": 1.9402841530054644,
      "grad_norm": 0.8432304848388915,
      "learning_rate": 0.00010966386047336183,
      "loss": 12.246,
      "step": 11096
    },
    {
      "epoch": 1.9404590163934428,
      "grad_norm": 0.7045305736169717,
      "learning_rate": 0.0001096497657695976,
      "loss": 12.28,
      "step": 11097
    },
    {
      "epoch": 1.9406338797814207,
      "grad_norm": 0.7209626451202852,
      "learning_rate": 0.000109635670872326,
      "loss": 12.1523,
      "step": 11098
    },
    {
      "epoch": 1.940808743169399,
      "grad_norm": 0.7748571680098777,
      "learning_rate": 0.00010962157578182959,
      "loss": 12.3861,
      "step": 11099
    },
    {
      "epoch": 1.940983606557377,
      "grad_norm": 0.6216061463571076,
      "learning_rate": 0.00010960748049839103,
      "loss": 12.2711,
      "step": 11100
    },
    {
      "epoch": 1.9411584699453552,
      "grad_norm": 0.7542636280213612,
      "learning_rate": 0.000109593385022293,
      "loss": 12.307,
      "step": 11101
    },
    {
      "epoch": 1.9413333333333334,
      "grad_norm": 0.7291754522801135,
      "learning_rate": 0.00010957928935381812,
      "loss": 12.0914,
      "step": 11102
    },
    {
      "epoch": 1.9415081967213115,
      "grad_norm": 0.7884348119579654,
      "learning_rate": 0.00010956519349324909,
      "loss": 12.3134,
      "step": 11103
    },
    {
      "epoch": 1.9416830601092896,
      "grad_norm": 0.7555161904408905,
      "learning_rate": 0.00010955109744086856,
      "loss": 12.337,
      "step": 11104
    },
    {
      "epoch": 1.9418579234972677,
      "grad_norm": 0.7810601383703109,
      "learning_rate": 0.00010953700119695922,
      "loss": 12.2879,
      "step": 11105
    },
    {
      "epoch": 1.942032786885246,
      "grad_norm": 0.7973958974976202,
      "learning_rate": 0.00010952290476180371,
      "loss": 12.2657,
      "step": 11106
    },
    {
      "epoch": 1.942207650273224,
      "grad_norm": 0.6570037765564561,
      "learning_rate": 0.00010950880813568471,
      "loss": 12.4453,
      "step": 11107
    },
    {
      "epoch": 1.9423825136612023,
      "grad_norm": 0.7085309373045636,
      "learning_rate": 0.00010949471131888493,
      "loss": 12.0353,
      "step": 11108
    },
    {
      "epoch": 1.9425573770491802,
      "grad_norm": 0.6930316024549915,
      "learning_rate": 0.00010948061431168701,
      "loss": 12.1337,
      "step": 11109
    },
    {
      "epoch": 1.9427322404371585,
      "grad_norm": 0.6800168971637627,
      "learning_rate": 0.0001094665171143737,
      "loss": 12.1973,
      "step": 11110
    },
    {
      "epoch": 1.9429071038251366,
      "grad_norm": 0.7588920786892859,
      "learning_rate": 0.00010945241972722765,
      "loss": 12.2979,
      "step": 11111
    },
    {
      "epoch": 1.9430819672131148,
      "grad_norm": 0.6879036715504842,
      "learning_rate": 0.00010943832215053153,
      "loss": 11.7318,
      "step": 11112
    },
    {
      "epoch": 1.9432568306010929,
      "grad_norm": 0.7556761080009112,
      "learning_rate": 0.0001094242243845681,
      "loss": 12.3735,
      "step": 11113
    },
    {
      "epoch": 1.943431693989071,
      "grad_norm": 0.7630698784109929,
      "learning_rate": 0.00010941012642962003,
      "loss": 12.1979,
      "step": 11114
    },
    {
      "epoch": 1.9436065573770491,
      "grad_norm": 1.0525635243720106,
      "learning_rate": 0.00010939602828597002,
      "loss": 12.2473,
      "step": 11115
    },
    {
      "epoch": 1.9437814207650272,
      "grad_norm": 0.7282871585680677,
      "learning_rate": 0.0001093819299539008,
      "loss": 12.0296,
      "step": 11116
    },
    {
      "epoch": 1.9439562841530056,
      "grad_norm": 0.6745542677883436,
      "learning_rate": 0.00010936783143369507,
      "loss": 11.9387,
      "step": 11117
    },
    {
      "epoch": 1.9441311475409835,
      "grad_norm": 0.7278692625955026,
      "learning_rate": 0.00010935373272563556,
      "loss": 12.1989,
      "step": 11118
    },
    {
      "epoch": 1.9443060109289618,
      "grad_norm": 0.6793811971389324,
      "learning_rate": 0.000109339633830005,
      "loss": 12.368,
      "step": 11119
    },
    {
      "epoch": 1.94448087431694,
      "grad_norm": 0.6369063788432606,
      "learning_rate": 0.00010932553474708612,
      "loss": 12.1706,
      "step": 11120
    },
    {
      "epoch": 1.944655737704918,
      "grad_norm": 0.7170151586640904,
      "learning_rate": 0.00010931143547716161,
      "loss": 12.1874,
      "step": 11121
    },
    {
      "epoch": 1.9448306010928962,
      "grad_norm": 0.7219035099226354,
      "learning_rate": 0.00010929733602051422,
      "loss": 12.3443,
      "step": 11122
    },
    {
      "epoch": 1.9450054644808743,
      "grad_norm": 0.7021655193643171,
      "learning_rate": 0.0001092832363774267,
      "loss": 12.4843,
      "step": 11123
    },
    {
      "epoch": 1.9451803278688524,
      "grad_norm": 0.6659651352301682,
      "learning_rate": 0.0001092691365481818,
      "loss": 12.0631,
      "step": 11124
    },
    {
      "epoch": 1.9453551912568305,
      "grad_norm": 0.6800656462396991,
      "learning_rate": 0.00010925503653306226,
      "loss": 12.3003,
      "step": 11125
    },
    {
      "epoch": 1.9455300546448089,
      "grad_norm": 0.7207725835134732,
      "learning_rate": 0.0001092409363323508,
      "loss": 12.1788,
      "step": 11126
    },
    {
      "epoch": 1.9457049180327868,
      "grad_norm": 0.6446154405711129,
      "learning_rate": 0.00010922683594633021,
      "loss": 12.2125,
      "step": 11127
    },
    {
      "epoch": 1.9458797814207651,
      "grad_norm": 0.681435736847313,
      "learning_rate": 0.00010921273537528322,
      "loss": 12.2805,
      "step": 11128
    },
    {
      "epoch": 1.946054644808743,
      "grad_norm": 0.6551345819997205,
      "learning_rate": 0.0001091986346194926,
      "loss": 12.3319,
      "step": 11129
    },
    {
      "epoch": 1.9462295081967214,
      "grad_norm": 0.6742581878630168,
      "learning_rate": 0.00010918453367924113,
      "loss": 12.2559,
      "step": 11130
    },
    {
      "epoch": 1.9464043715846995,
      "grad_norm": 0.6979293297099485,
      "learning_rate": 0.00010917043255481152,
      "loss": 12.3311,
      "step": 11131
    },
    {
      "epoch": 1.9465792349726776,
      "grad_norm": 0.7051907963866995,
      "learning_rate": 0.0001091563312464866,
      "loss": 12.3192,
      "step": 11132
    },
    {
      "epoch": 1.9467540983606557,
      "grad_norm": 0.7003085038094243,
      "learning_rate": 0.00010914222975454913,
      "loss": 12.5381,
      "step": 11133
    },
    {
      "epoch": 1.9469289617486338,
      "grad_norm": 0.6999279157538388,
      "learning_rate": 0.00010912812807928188,
      "loss": 12.2051,
      "step": 11134
    },
    {
      "epoch": 1.9471038251366122,
      "grad_norm": 0.6724293160095248,
      "learning_rate": 0.00010911402622096767,
      "loss": 12.3567,
      "step": 11135
    },
    {
      "epoch": 1.94727868852459,
      "grad_norm": 0.63286885581216,
      "learning_rate": 0.00010909992417988919,
      "loss": 11.8798,
      "step": 11136
    },
    {
      "epoch": 1.9474535519125684,
      "grad_norm": 0.6989146150520226,
      "learning_rate": 0.00010908582195632935,
      "loss": 12.2856,
      "step": 11137
    },
    {
      "epoch": 1.9476284153005463,
      "grad_norm": 0.6592228624962221,
      "learning_rate": 0.0001090717195505708,
      "loss": 12.0439,
      "step": 11138
    },
    {
      "epoch": 1.9478032786885247,
      "grad_norm": 0.6673788892741546,
      "learning_rate": 0.00010905761696289651,
      "loss": 12.2537,
      "step": 11139
    },
    {
      "epoch": 1.9479781420765028,
      "grad_norm": 0.6775592484080019,
      "learning_rate": 0.00010904351419358911,
      "loss": 12.0901,
      "step": 11140
    },
    {
      "epoch": 1.948153005464481,
      "grad_norm": 0.7457384898618675,
      "learning_rate": 0.00010902941124293153,
      "loss": 12.2226,
      "step": 11141
    },
    {
      "epoch": 1.948327868852459,
      "grad_norm": 0.7163018012216865,
      "learning_rate": 0.00010901530811120655,
      "loss": 12.2721,
      "step": 11142
    },
    {
      "epoch": 1.9485027322404371,
      "grad_norm": 0.6562880852480956,
      "learning_rate": 0.0001090012047986969,
      "loss": 12.065,
      "step": 11143
    },
    {
      "epoch": 1.9486775956284152,
      "grad_norm": 0.7236846629381438,
      "learning_rate": 0.00010898710130568553,
      "loss": 12.1217,
      "step": 11144
    },
    {
      "epoch": 1.9488524590163934,
      "grad_norm": 0.6675757481035912,
      "learning_rate": 0.00010897299763245512,
      "loss": 12.1612,
      "step": 11145
    },
    {
      "epoch": 1.9490273224043717,
      "grad_norm": 0.6410848330459872,
      "learning_rate": 0.00010895889377928859,
      "loss": 12.2816,
      "step": 11146
    },
    {
      "epoch": 1.9492021857923496,
      "grad_norm": 0.6401679226520106,
      "learning_rate": 0.00010894478974646873,
      "loss": 12.1585,
      "step": 11147
    },
    {
      "epoch": 1.949377049180328,
      "grad_norm": 0.6616158912311577,
      "learning_rate": 0.00010893068553427837,
      "loss": 12.2464,
      "step": 11148
    },
    {
      "epoch": 1.9495519125683058,
      "grad_norm": 0.7169424809502094,
      "learning_rate": 0.00010891658114300034,
      "loss": 12.1306,
      "step": 11149
    },
    {
      "epoch": 1.9497267759562842,
      "grad_norm": 0.7294795100019208,
      "learning_rate": 0.00010890247657291745,
      "loss": 12.0548,
      "step": 11150
    },
    {
      "epoch": 1.9499016393442623,
      "grad_norm": 0.6866772521516503,
      "learning_rate": 0.00010888837182431262,
      "loss": 12.2247,
      "step": 11151
    },
    {
      "epoch": 1.9500765027322404,
      "grad_norm": 0.6368389990919946,
      "learning_rate": 0.0001088742668974686,
      "loss": 11.9063,
      "step": 11152
    },
    {
      "epoch": 1.9502513661202185,
      "grad_norm": 0.6432880666621971,
      "learning_rate": 0.00010886016179266833,
      "loss": 12.0866,
      "step": 11153
    },
    {
      "epoch": 1.9504262295081967,
      "grad_norm": 0.7338577433704114,
      "learning_rate": 0.00010884605651019459,
      "loss": 11.9866,
      "step": 11154
    },
    {
      "epoch": 1.950601092896175,
      "grad_norm": 0.6803817388159531,
      "learning_rate": 0.00010883195105033026,
      "loss": 12.4101,
      "step": 11155
    },
    {
      "epoch": 1.950775956284153,
      "grad_norm": 0.6858171635920073,
      "learning_rate": 0.00010881784541335817,
      "loss": 12.241,
      "step": 11156
    },
    {
      "epoch": 1.9509508196721312,
      "grad_norm": 0.7418674039044869,
      "learning_rate": 0.00010880373959956122,
      "loss": 12.3223,
      "step": 11157
    },
    {
      "epoch": 1.9511256830601091,
      "grad_norm": 0.6777288865598536,
      "learning_rate": 0.00010878963360922228,
      "loss": 12.0515,
      "step": 11158
    },
    {
      "epoch": 1.9513005464480875,
      "grad_norm": 0.7694146868819445,
      "learning_rate": 0.00010877552744262418,
      "loss": 12.1761,
      "step": 11159
    },
    {
      "epoch": 1.9514754098360656,
      "grad_norm": 0.7246575920567481,
      "learning_rate": 0.0001087614211000498,
      "loss": 12.1864,
      "step": 11160
    },
    {
      "epoch": 1.9516502732240437,
      "grad_norm": 0.7557425303400729,
      "learning_rate": 0.00010874731458178205,
      "loss": 12.1931,
      "step": 11161
    },
    {
      "epoch": 1.9518251366120218,
      "grad_norm": 0.7030194695912694,
      "learning_rate": 0.00010873320788810378,
      "loss": 12.0929,
      "step": 11162
    },
    {
      "epoch": 1.952,
      "grad_norm": 0.6453646322599011,
      "learning_rate": 0.00010871910101929785,
      "loss": 12.0883,
      "step": 11163
    },
    {
      "epoch": 1.9521748633879783,
      "grad_norm": 0.7760694742296966,
      "learning_rate": 0.0001087049939756472,
      "loss": 12.3578,
      "step": 11164
    },
    {
      "epoch": 1.9523497267759562,
      "grad_norm": 0.6087245669303658,
      "learning_rate": 0.00010869088675743468,
      "loss": 12.0322,
      "step": 11165
    },
    {
      "epoch": 1.9525245901639345,
      "grad_norm": 0.7215400093035942,
      "learning_rate": 0.0001086767793649432,
      "loss": 12.0445,
      "step": 11166
    },
    {
      "epoch": 1.9526994535519124,
      "grad_norm": 0.6554259432404438,
      "learning_rate": 0.00010866267179845566,
      "loss": 12.2757,
      "step": 11167
    },
    {
      "epoch": 1.9528743169398908,
      "grad_norm": 0.6592734954234069,
      "learning_rate": 0.00010864856405825495,
      "loss": 11.8976,
      "step": 11168
    },
    {
      "epoch": 1.953049180327869,
      "grad_norm": 0.7958216241746008,
      "learning_rate": 0.00010863445614462395,
      "loss": 12.3149,
      "step": 11169
    },
    {
      "epoch": 1.953224043715847,
      "grad_norm": 0.6431159293965221,
      "learning_rate": 0.00010862034805784562,
      "loss": 12.1698,
      "step": 11170
    },
    {
      "epoch": 1.9533989071038251,
      "grad_norm": 0.7673573286531506,
      "learning_rate": 0.00010860623979820283,
      "loss": 12.2289,
      "step": 11171
    },
    {
      "epoch": 1.9535737704918033,
      "grad_norm": 0.7451238444033601,
      "learning_rate": 0.00010859213136597853,
      "loss": 12.3939,
      "step": 11172
    },
    {
      "epoch": 1.9537486338797814,
      "grad_norm": 0.6169340834998973,
      "learning_rate": 0.0001085780227614556,
      "loss": 12.3093,
      "step": 11173
    },
    {
      "epoch": 1.9539234972677595,
      "grad_norm": 0.6579034979414184,
      "learning_rate": 0.00010856391398491699,
      "loss": 12.2702,
      "step": 11174
    },
    {
      "epoch": 1.9540983606557378,
      "grad_norm": 0.7220167186043827,
      "learning_rate": 0.00010854980503664558,
      "loss": 12.1166,
      "step": 11175
    },
    {
      "epoch": 1.9542732240437157,
      "grad_norm": 0.6927596435007597,
      "learning_rate": 0.00010853569591692434,
      "loss": 12.1352,
      "step": 11176
    },
    {
      "epoch": 1.954448087431694,
      "grad_norm": 0.7449022764642076,
      "learning_rate": 0.0001085215866260362,
      "loss": 12.0982,
      "step": 11177
    },
    {
      "epoch": 1.954622950819672,
      "grad_norm": 0.6565972789502156,
      "learning_rate": 0.0001085074771642641,
      "loss": 12.3406,
      "step": 11178
    },
    {
      "epoch": 1.9547978142076503,
      "grad_norm": 0.7246592916990683,
      "learning_rate": 0.00010849336753189093,
      "loss": 12.2627,
      "step": 11179
    },
    {
      "epoch": 1.9549726775956284,
      "grad_norm": 0.6137929170319504,
      "learning_rate": 0.00010847925772919967,
      "loss": 12.1693,
      "step": 11180
    },
    {
      "epoch": 1.9551475409836065,
      "grad_norm": 0.6829385592992334,
      "learning_rate": 0.00010846514775647325,
      "loss": 12.4132,
      "step": 11181
    },
    {
      "epoch": 1.9553224043715847,
      "grad_norm": 0.6486252495868596,
      "learning_rate": 0.00010845103761399462,
      "loss": 12.3253,
      "step": 11182
    },
    {
      "epoch": 1.9554972677595628,
      "grad_norm": 0.6326750567043576,
      "learning_rate": 0.00010843692730204677,
      "loss": 12.0556,
      "step": 11183
    },
    {
      "epoch": 1.9556721311475411,
      "grad_norm": 0.665019677524423,
      "learning_rate": 0.0001084228168209126,
      "loss": 12.2898,
      "step": 11184
    },
    {
      "epoch": 1.955846994535519,
      "grad_norm": 0.6737517000029963,
      "learning_rate": 0.00010840870617087514,
      "loss": 12.1361,
      "step": 11185
    },
    {
      "epoch": 1.9560218579234974,
      "grad_norm": 0.6432931274403751,
      "learning_rate": 0.00010839459535221721,
      "loss": 12.2543,
      "step": 11186
    },
    {
      "epoch": 1.9561967213114753,
      "grad_norm": 0.744922078228936,
      "learning_rate": 0.00010838048436522195,
      "loss": 12.0679,
      "step": 11187
    },
    {
      "epoch": 1.9563715846994536,
      "grad_norm": 0.7582714170758118,
      "learning_rate": 0.0001083663732101722,
      "loss": 12.3123,
      "step": 11188
    },
    {
      "epoch": 1.9565464480874317,
      "grad_norm": 0.7935369854582822,
      "learning_rate": 0.00010835226188735099,
      "loss": 12.4519,
      "step": 11189
    },
    {
      "epoch": 1.9567213114754098,
      "grad_norm": 0.7511520524016266,
      "learning_rate": 0.00010833815039704132,
      "loss": 12.1085,
      "step": 11190
    },
    {
      "epoch": 1.956896174863388,
      "grad_norm": 0.6618641061313258,
      "learning_rate": 0.00010832403873952606,
      "loss": 12.0879,
      "step": 11191
    },
    {
      "epoch": 1.957071038251366,
      "grad_norm": 0.7422421308759843,
      "learning_rate": 0.00010830992691508834,
      "loss": 12.2766,
      "step": 11192
    },
    {
      "epoch": 1.9572459016393444,
      "grad_norm": 0.6921240400032549,
      "learning_rate": 0.00010829581492401102,
      "loss": 11.9992,
      "step": 11193
    },
    {
      "epoch": 1.9574207650273223,
      "grad_norm": 0.7023375953554212,
      "learning_rate": 0.00010828170276657717,
      "loss": 12.2427,
      "step": 11194
    },
    {
      "epoch": 1.9575956284153007,
      "grad_norm": 0.7665205190292017,
      "learning_rate": 0.00010826759044306973,
      "loss": 12.1343,
      "step": 11195
    },
    {
      "epoch": 1.9577704918032786,
      "grad_norm": 0.6687168929025263,
      "learning_rate": 0.00010825347795377174,
      "loss": 12.2734,
      "step": 11196
    },
    {
      "epoch": 1.957945355191257,
      "grad_norm": 0.8415464158901039,
      "learning_rate": 0.00010823936529896614,
      "loss": 12.2542,
      "step": 11197
    },
    {
      "epoch": 1.958120218579235,
      "grad_norm": 0.6244033823707701,
      "learning_rate": 0.00010822525247893598,
      "loss": 12.3716,
      "step": 11198
    },
    {
      "epoch": 1.9582950819672131,
      "grad_norm": 0.8443548573514842,
      "learning_rate": 0.00010821113949396428,
      "loss": 12.2491,
      "step": 11199
    },
    {
      "epoch": 1.9584699453551913,
      "grad_norm": 0.6955328796849567,
      "learning_rate": 0.00010819702634433398,
      "loss": 12.28,
      "step": 11200
    },
    {
      "epoch": 1.9586448087431694,
      "grad_norm": 0.6048495290747885,
      "learning_rate": 0.00010818291303032817,
      "loss": 12.2406,
      "step": 11201
    },
    {
      "epoch": 1.9588196721311475,
      "grad_norm": 0.6872612666222921,
      "learning_rate": 0.00010816879955222983,
      "loss": 12.1979,
      "step": 11202
    },
    {
      "epoch": 1.9589945355191256,
      "grad_norm": 0.6333195359246103,
      "learning_rate": 0.00010815468591032196,
      "loss": 12.2326,
      "step": 11203
    },
    {
      "epoch": 1.959169398907104,
      "grad_norm": 0.7040247081227522,
      "learning_rate": 0.0001081405721048876,
      "loss": 12.3901,
      "step": 11204
    },
    {
      "epoch": 1.9593442622950819,
      "grad_norm": 0.644172606025199,
      "learning_rate": 0.00010812645813620976,
      "loss": 12.3102,
      "step": 11205
    },
    {
      "epoch": 1.9595191256830602,
      "grad_norm": 0.693379725691512,
      "learning_rate": 0.0001081123440045715,
      "loss": 12.3031,
      "step": 11206
    },
    {
      "epoch": 1.959693989071038,
      "grad_norm": 0.7931623922327836,
      "learning_rate": 0.00010809822971025584,
      "loss": 12.2514,
      "step": 11207
    },
    {
      "epoch": 1.9598688524590164,
      "grad_norm": 0.7485716216413252,
      "learning_rate": 0.0001080841152535458,
      "loss": 12.1814,
      "step": 11208
    },
    {
      "epoch": 1.9600437158469945,
      "grad_norm": 0.6785215954297462,
      "learning_rate": 0.00010807000063472444,
      "loss": 12.4463,
      "step": 11209
    },
    {
      "epoch": 1.9602185792349727,
      "grad_norm": 0.6773701236762878,
      "learning_rate": 0.00010805588585407478,
      "loss": 11.9492,
      "step": 11210
    },
    {
      "epoch": 1.9603934426229508,
      "grad_norm": 0.6580848530870373,
      "learning_rate": 0.00010804177091187988,
      "loss": 12.3566,
      "step": 11211
    },
    {
      "epoch": 1.960568306010929,
      "grad_norm": 0.6893117014805192,
      "learning_rate": 0.00010802765580842277,
      "loss": 12.3453,
      "step": 11212
    },
    {
      "epoch": 1.9607431693989072,
      "grad_norm": 0.6264387545262672,
      "learning_rate": 0.00010801354054398651,
      "loss": 12.3165,
      "step": 11213
    },
    {
      "epoch": 1.9609180327868851,
      "grad_norm": 0.6793992020870926,
      "learning_rate": 0.00010799942511885418,
      "loss": 12.3064,
      "step": 11214
    },
    {
      "epoch": 1.9610928961748635,
      "grad_norm": 0.6830450131031597,
      "learning_rate": 0.00010798530953330879,
      "loss": 12.1691,
      "step": 11215
    },
    {
      "epoch": 1.9612677595628414,
      "grad_norm": 0.7123290748805651,
      "learning_rate": 0.00010797119378763346,
      "loss": 11.9986,
      "step": 11216
    },
    {
      "epoch": 1.9614426229508197,
      "grad_norm": 0.7147738487992109,
      "learning_rate": 0.00010795707788211118,
      "loss": 12.1583,
      "step": 11217
    },
    {
      "epoch": 1.9616174863387978,
      "grad_norm": 0.6812693231828679,
      "learning_rate": 0.00010794296181702508,
      "loss": 12.2705,
      "step": 11218
    },
    {
      "epoch": 1.961792349726776,
      "grad_norm": 0.634383826522439,
      "learning_rate": 0.00010792884559265819,
      "loss": 11.9686,
      "step": 11219
    },
    {
      "epoch": 1.961967213114754,
      "grad_norm": 0.6551360507890838,
      "learning_rate": 0.00010791472920929361,
      "loss": 12.2532,
      "step": 11220
    },
    {
      "epoch": 1.9621420765027322,
      "grad_norm": 0.685393736723234,
      "learning_rate": 0.00010790061266721441,
      "loss": 12.2058,
      "step": 11221
    },
    {
      "epoch": 1.9623169398907105,
      "grad_norm": 0.7305375671234046,
      "learning_rate": 0.00010788649596670366,
      "loss": 12.4289,
      "step": 11222
    },
    {
      "epoch": 1.9624918032786884,
      "grad_norm": 0.6573649808166194,
      "learning_rate": 0.00010787237910804446,
      "loss": 12.0856,
      "step": 11223
    },
    {
      "epoch": 1.9626666666666668,
      "grad_norm": 0.7509456678829651,
      "learning_rate": 0.0001078582620915199,
      "loss": 12.1097,
      "step": 11224
    },
    {
      "epoch": 1.9628415300546447,
      "grad_norm": 0.6955244531258865,
      "learning_rate": 0.00010784414491741304,
      "loss": 12.2062,
      "step": 11225
    },
    {
      "epoch": 1.963016393442623,
      "grad_norm": 0.659095012894107,
      "learning_rate": 0.00010783002758600702,
      "loss": 12.1093,
      "step": 11226
    },
    {
      "epoch": 1.9631912568306011,
      "grad_norm": 0.7717590619886087,
      "learning_rate": 0.00010781591009758484,
      "loss": 12.0298,
      "step": 11227
    },
    {
      "epoch": 1.9633661202185793,
      "grad_norm": 0.6992489322271175,
      "learning_rate": 0.00010780179245242972,
      "loss": 12.0597,
      "step": 11228
    },
    {
      "epoch": 1.9635409836065574,
      "grad_norm": 0.7596709210802898,
      "learning_rate": 0.00010778767465082468,
      "loss": 11.9545,
      "step": 11229
    },
    {
      "epoch": 1.9637158469945355,
      "grad_norm": 0.7432869226900648,
      "learning_rate": 0.00010777355669305287,
      "loss": 11.9834,
      "step": 11230
    },
    {
      "epoch": 1.9638907103825136,
      "grad_norm": 0.799992064648911,
      "learning_rate": 0.00010775943857939739,
      "loss": 12.3209,
      "step": 11231
    },
    {
      "epoch": 1.9640655737704917,
      "grad_norm": 0.684951399591446,
      "learning_rate": 0.0001077453203101413,
      "loss": 11.9334,
      "step": 11232
    },
    {
      "epoch": 1.96424043715847,
      "grad_norm": 0.6665111568463726,
      "learning_rate": 0.0001077312018855678,
      "loss": 12.4081,
      "step": 11233
    },
    {
      "epoch": 1.964415300546448,
      "grad_norm": 0.7708870046612503,
      "learning_rate": 0.00010771708330595993,
      "loss": 12.3554,
      "step": 11234
    },
    {
      "epoch": 1.9645901639344263,
      "grad_norm": 0.6676549428936472,
      "learning_rate": 0.00010770296457160088,
      "loss": 12.0839,
      "step": 11235
    },
    {
      "epoch": 1.9647650273224042,
      "grad_norm": 0.7533960917214106,
      "learning_rate": 0.00010768884568277369,
      "loss": 12.2709,
      "step": 11236
    },
    {
      "epoch": 1.9649398907103826,
      "grad_norm": 0.7855433978772572,
      "learning_rate": 0.00010767472663976157,
      "loss": 11.9247,
      "step": 11237
    },
    {
      "epoch": 1.9651147540983607,
      "grad_norm": 0.6425554108379177,
      "learning_rate": 0.00010766060744284763,
      "loss": 12.1329,
      "step": 11238
    },
    {
      "epoch": 1.9652896174863388,
      "grad_norm": 0.7142414577996637,
      "learning_rate": 0.00010764648809231494,
      "loss": 12.3671,
      "step": 11239
    },
    {
      "epoch": 1.965464480874317,
      "grad_norm": 0.7959202332861027,
      "learning_rate": 0.00010763236858844674,
      "loss": 12.2471,
      "step": 11240
    },
    {
      "epoch": 1.965639344262295,
      "grad_norm": 0.6866985353155475,
      "learning_rate": 0.00010761824893152608,
      "loss": 12.2018,
      "step": 11241
    },
    {
      "epoch": 1.9658142076502734,
      "grad_norm": 0.6798757208579818,
      "learning_rate": 0.00010760412912183617,
      "loss": 12.098,
      "step": 11242
    },
    {
      "epoch": 1.9659890710382513,
      "grad_norm": 0.6919816983439442,
      "learning_rate": 0.00010759000915966011,
      "loss": 12.3415,
      "step": 11243
    },
    {
      "epoch": 1.9661639344262296,
      "grad_norm": 0.6436629472535738,
      "learning_rate": 0.00010757588904528106,
      "loss": 12.1937,
      "step": 11244
    },
    {
      "epoch": 1.9663387978142075,
      "grad_norm": 0.6641508036541045,
      "learning_rate": 0.00010756176877898216,
      "loss": 12.2029,
      "step": 11245
    },
    {
      "epoch": 1.9665136612021858,
      "grad_norm": 0.6219500874777595,
      "learning_rate": 0.00010754764836104657,
      "loss": 12.1018,
      "step": 11246
    },
    {
      "epoch": 1.966688524590164,
      "grad_norm": 0.6516755977660155,
      "learning_rate": 0.0001075335277917575,
      "loss": 12.1687,
      "step": 11247
    },
    {
      "epoch": 1.966863387978142,
      "grad_norm": 0.6634452866975449,
      "learning_rate": 0.000107519407071398,
      "loss": 12.166,
      "step": 11248
    },
    {
      "epoch": 1.9670382513661202,
      "grad_norm": 0.6526368964508371,
      "learning_rate": 0.00010750528620025137,
      "loss": 12.1388,
      "step": 11249
    },
    {
      "epoch": 1.9672131147540983,
      "grad_norm": 0.7096258224108399,
      "learning_rate": 0.00010749116517860067,
      "loss": 12.407,
      "step": 11250
    },
    {
      "epoch": 1.9673879781420767,
      "grad_norm": 0.6412348772439148,
      "learning_rate": 0.00010747704400672913,
      "loss": 12.3259,
      "step": 11251
    },
    {
      "epoch": 1.9675628415300546,
      "grad_norm": 0.8426844611121677,
      "learning_rate": 0.00010746292268491989,
      "loss": 12.5389,
      "step": 11252
    },
    {
      "epoch": 1.967737704918033,
      "grad_norm": 0.6120449224879546,
      "learning_rate": 0.00010744880121345613,
      "loss": 12.1881,
      "step": 11253
    },
    {
      "epoch": 1.9679125683060108,
      "grad_norm": 0.8024065878628199,
      "learning_rate": 0.00010743467959262103,
      "loss": 12.3155,
      "step": 11254
    },
    {
      "epoch": 1.9680874316939891,
      "grad_norm": 0.6434403979537701,
      "learning_rate": 0.0001074205578226978,
      "loss": 12.3187,
      "step": 11255
    },
    {
      "epoch": 1.9682622950819673,
      "grad_norm": 0.7862027501950828,
      "learning_rate": 0.00010740643590396957,
      "loss": 12.0723,
      "step": 11256
    },
    {
      "epoch": 1.9684371584699454,
      "grad_norm": 0.6860273754052608,
      "learning_rate": 0.00010739231383671957,
      "loss": 12.1068,
      "step": 11257
    },
    {
      "epoch": 1.9686120218579235,
      "grad_norm": 0.7127067902363029,
      "learning_rate": 0.000107378191621231,
      "loss": 12.3313,
      "step": 11258
    },
    {
      "epoch": 1.9687868852459016,
      "grad_norm": 0.7106270030468937,
      "learning_rate": 0.00010736406925778702,
      "loss": 12.12,
      "step": 11259
    },
    {
      "epoch": 1.9689617486338797,
      "grad_norm": 0.703175204189451,
      "learning_rate": 0.00010734994674667082,
      "loss": 12.1177,
      "step": 11260
    },
    {
      "epoch": 1.9691366120218579,
      "grad_norm": 0.6549197844133872,
      "learning_rate": 0.00010733582408816564,
      "loss": 11.9939,
      "step": 11261
    },
    {
      "epoch": 1.9693114754098362,
      "grad_norm": 0.7260774442090973,
      "learning_rate": 0.00010732170128255468,
      "loss": 12.2033,
      "step": 11262
    },
    {
      "epoch": 1.969486338797814,
      "grad_norm": 0.6373714308128344,
      "learning_rate": 0.0001073075783301211,
      "loss": 12.0112,
      "step": 11263
    },
    {
      "epoch": 1.9696612021857924,
      "grad_norm": 0.6932148758778213,
      "learning_rate": 0.00010729345523114815,
      "loss": 12.2911,
      "step": 11264
    },
    {
      "epoch": 1.9698360655737703,
      "grad_norm": 0.6761932275799586,
      "learning_rate": 0.00010727933198591902,
      "loss": 12.0919,
      "step": 11265
    },
    {
      "epoch": 1.9700109289617487,
      "grad_norm": 0.72624912469226,
      "learning_rate": 0.00010726520859471696,
      "loss": 12.2637,
      "step": 11266
    },
    {
      "epoch": 1.9701857923497268,
      "grad_norm": 0.6895929369957075,
      "learning_rate": 0.00010725108505782513,
      "loss": 12.126,
      "step": 11267
    },
    {
      "epoch": 1.970360655737705,
      "grad_norm": 0.6906543641966973,
      "learning_rate": 0.0001072369613755268,
      "loss": 12.1107,
      "step": 11268
    },
    {
      "epoch": 1.970535519125683,
      "grad_norm": 0.760326594529459,
      "learning_rate": 0.00010722283754810515,
      "loss": 12.086,
      "step": 11269
    },
    {
      "epoch": 1.9707103825136612,
      "grad_norm": 0.6388667236811773,
      "learning_rate": 0.00010720871357584345,
      "loss": 12.4059,
      "step": 11270
    },
    {
      "epoch": 1.9708852459016395,
      "grad_norm": 0.7756363074633353,
      "learning_rate": 0.00010719458945902492,
      "loss": 12.221,
      "step": 11271
    },
    {
      "epoch": 1.9710601092896174,
      "grad_norm": 0.6994618160321604,
      "learning_rate": 0.00010718046519793276,
      "loss": 12.1374,
      "step": 11272
    },
    {
      "epoch": 1.9712349726775957,
      "grad_norm": 0.6595537191916014,
      "learning_rate": 0.00010716634079285024,
      "loss": 12.1771,
      "step": 11273
    },
    {
      "epoch": 1.9714098360655736,
      "grad_norm": 0.68603637739114,
      "learning_rate": 0.00010715221624406057,
      "loss": 11.9014,
      "step": 11274
    },
    {
      "epoch": 1.971584699453552,
      "grad_norm": 0.6634022659994052,
      "learning_rate": 0.000107138091551847,
      "loss": 12.4631,
      "step": 11275
    },
    {
      "epoch": 1.97175956284153,
      "grad_norm": 0.6562766577299519,
      "learning_rate": 0.00010712396671649278,
      "loss": 12.3544,
      "step": 11276
    },
    {
      "epoch": 1.9719344262295082,
      "grad_norm": 0.6151271118870463,
      "learning_rate": 0.00010710984173828118,
      "loss": 12.213,
      "step": 11277
    },
    {
      "epoch": 1.9721092896174863,
      "grad_norm": 0.6641862195551564,
      "learning_rate": 0.0001070957166174954,
      "loss": 12.4176,
      "step": 11278
    },
    {
      "epoch": 1.9722841530054644,
      "grad_norm": 0.6256770102595421,
      "learning_rate": 0.00010708159135441873,
      "loss": 12.5216,
      "step": 11279
    },
    {
      "epoch": 1.9724590163934428,
      "grad_norm": 0.7362410208117016,
      "learning_rate": 0.00010706746594933438,
      "loss": 12.2624,
      "step": 11280
    },
    {
      "epoch": 1.9726338797814207,
      "grad_norm": 0.6793459353506182,
      "learning_rate": 0.00010705334040252569,
      "loss": 12.2968,
      "step": 11281
    },
    {
      "epoch": 1.972808743169399,
      "grad_norm": 0.6356694601106032,
      "learning_rate": 0.00010703921471427581,
      "loss": 12.2151,
      "step": 11282
    },
    {
      "epoch": 1.972983606557377,
      "grad_norm": 0.6733517661047346,
      "learning_rate": 0.0001070250888848681,
      "loss": 12.3096,
      "step": 11283
    },
    {
      "epoch": 1.9731584699453553,
      "grad_norm": 0.6357545985287666,
      "learning_rate": 0.00010701096291458576,
      "loss": 12.0459,
      "step": 11284
    },
    {
      "epoch": 1.9733333333333334,
      "grad_norm": 0.7065070811474686,
      "learning_rate": 0.00010699683680371211,
      "loss": 12.3486,
      "step": 11285
    },
    {
      "epoch": 1.9735081967213115,
      "grad_norm": 0.6784790381245599,
      "learning_rate": 0.00010698271055253042,
      "loss": 12.2509,
      "step": 11286
    },
    {
      "epoch": 1.9736830601092896,
      "grad_norm": 0.678882859425327,
      "learning_rate": 0.0001069685841613239,
      "loss": 11.8786,
      "step": 11287
    },
    {
      "epoch": 1.9738579234972677,
      "grad_norm": 0.8297669726617367,
      "learning_rate": 0.00010695445763037589,
      "loss": 12.2856,
      "step": 11288
    },
    {
      "epoch": 1.9740327868852459,
      "grad_norm": 0.6211960660906767,
      "learning_rate": 0.00010694033095996962,
      "loss": 12.2279,
      "step": 11289
    },
    {
      "epoch": 1.974207650273224,
      "grad_norm": 0.884773380017266,
      "learning_rate": 0.00010692620415038845,
      "loss": 12.4636,
      "step": 11290
    },
    {
      "epoch": 1.9743825136612023,
      "grad_norm": 0.7588243730647618,
      "learning_rate": 0.0001069120772019156,
      "loss": 12.1316,
      "step": 11291
    },
    {
      "epoch": 1.9745573770491802,
      "grad_norm": 0.8014515762260723,
      "learning_rate": 0.00010689795011483439,
      "loss": 12.326,
      "step": 11292
    },
    {
      "epoch": 1.9747322404371586,
      "grad_norm": 0.8436048765365113,
      "learning_rate": 0.0001068838228894281,
      "loss": 12.2222,
      "step": 11293
    },
    {
      "epoch": 1.9749071038251365,
      "grad_norm": 0.6797879525180385,
      "learning_rate": 0.00010686969552597999,
      "loss": 12.0483,
      "step": 11294
    },
    {
      "epoch": 1.9750819672131148,
      "grad_norm": 1.0275745458285457,
      "learning_rate": 0.00010685556802477345,
      "loss": 12.4419,
      "step": 11295
    },
    {
      "epoch": 1.975256830601093,
      "grad_norm": 0.6431324627448253,
      "learning_rate": 0.00010684144038609168,
      "loss": 12.1403,
      "step": 11296
    },
    {
      "epoch": 1.975431693989071,
      "grad_norm": 0.827617089586682,
      "learning_rate": 0.00010682731261021805,
      "loss": 12.2987,
      "step": 11297
    },
    {
      "epoch": 1.9756065573770492,
      "grad_norm": 0.7482972862862131,
      "learning_rate": 0.00010681318469743582,
      "loss": 12.176,
      "step": 11298
    },
    {
      "epoch": 1.9757814207650273,
      "grad_norm": 0.754166201067388,
      "learning_rate": 0.00010679905664802831,
      "loss": 12.117,
      "step": 11299
    },
    {
      "epoch": 1.9759562841530056,
      "grad_norm": 0.9049809385320694,
      "learning_rate": 0.00010678492846227887,
      "loss": 12.1898,
      "step": 11300
    },
    {
      "epoch": 1.9761311475409835,
      "grad_norm": 0.7146153685457443,
      "learning_rate": 0.00010677080014047076,
      "loss": 11.9573,
      "step": 11301
    },
    {
      "epoch": 1.9763060109289619,
      "grad_norm": 0.7234076153655692,
      "learning_rate": 0.00010675667168288732,
      "loss": 12.1973,
      "step": 11302
    },
    {
      "epoch": 1.9764808743169398,
      "grad_norm": 0.7669890641917009,
      "learning_rate": 0.00010674254308981187,
      "loss": 12.1399,
      "step": 11303
    },
    {
      "epoch": 1.976655737704918,
      "grad_norm": 0.6755557220357492,
      "learning_rate": 0.00010672841436152771,
      "loss": 12.1362,
      "step": 11304
    },
    {
      "epoch": 1.9768306010928962,
      "grad_norm": 0.8358330808457862,
      "learning_rate": 0.0001067142854983182,
      "loss": 12.3929,
      "step": 11305
    },
    {
      "epoch": 1.9770054644808743,
      "grad_norm": 0.6168994909535782,
      "learning_rate": 0.00010670015650046667,
      "loss": 12.0052,
      "step": 11306
    },
    {
      "epoch": 1.9771803278688524,
      "grad_norm": 0.870368462197377,
      "learning_rate": 0.00010668602736825641,
      "loss": 12.1299,
      "step": 11307
    },
    {
      "epoch": 1.9773551912568306,
      "grad_norm": 0.6687709527079923,
      "learning_rate": 0.00010667189810197079,
      "loss": 12.2568,
      "step": 11308
    },
    {
      "epoch": 1.9775300546448087,
      "grad_norm": 0.7109029458459535,
      "learning_rate": 0.00010665776870189308,
      "loss": 12.2287,
      "step": 11309
    },
    {
      "epoch": 1.9777049180327868,
      "grad_norm": 0.8831072693706977,
      "learning_rate": 0.0001066436391683067,
      "loss": 12.2521,
      "step": 11310
    },
    {
      "epoch": 1.9778797814207651,
      "grad_norm": 0.6441307618376514,
      "learning_rate": 0.00010662950950149495,
      "loss": 12.16,
      "step": 11311
    },
    {
      "epoch": 1.978054644808743,
      "grad_norm": 0.7381639270764812,
      "learning_rate": 0.00010661537970174118,
      "loss": 12.0931,
      "step": 11312
    },
    {
      "epoch": 1.9782295081967214,
      "grad_norm": 0.6876294301038841,
      "learning_rate": 0.00010660124976932875,
      "loss": 12.2216,
      "step": 11313
    },
    {
      "epoch": 1.9784043715846995,
      "grad_norm": 0.6470685010232936,
      "learning_rate": 0.00010658711970454099,
      "loss": 12.0189,
      "step": 11314
    },
    {
      "epoch": 1.9785792349726776,
      "grad_norm": 0.8662386022237264,
      "learning_rate": 0.00010657298950766126,
      "loss": 12.1428,
      "step": 11315
    },
    {
      "epoch": 1.9787540983606557,
      "grad_norm": 0.7210835779417488,
      "learning_rate": 0.00010655885917897286,
      "loss": 12.416,
      "step": 11316
    },
    {
      "epoch": 1.9789289617486339,
      "grad_norm": 0.8292798173898893,
      "learning_rate": 0.00010654472871875922,
      "loss": 12.2624,
      "step": 11317
    },
    {
      "epoch": 1.979103825136612,
      "grad_norm": 0.6838360921080406,
      "learning_rate": 0.0001065305981273037,
      "loss": 12.1084,
      "step": 11318
    },
    {
      "epoch": 1.97927868852459,
      "grad_norm": 0.6535723874580981,
      "learning_rate": 0.00010651646740488961,
      "loss": 12.2361,
      "step": 11319
    },
    {
      "epoch": 1.9794535519125684,
      "grad_norm": 0.785830991352303,
      "learning_rate": 0.00010650233655180035,
      "loss": 12.3491,
      "step": 11320
    },
    {
      "epoch": 1.9796284153005463,
      "grad_norm": 0.5812711985919568,
      "learning_rate": 0.00010648820556831929,
      "loss": 12.1787,
      "step": 11321
    },
    {
      "epoch": 1.9798032786885247,
      "grad_norm": 0.7089992756825152,
      "learning_rate": 0.00010647407445472979,
      "loss": 12.1567,
      "step": 11322
    },
    {
      "epoch": 1.9799781420765026,
      "grad_norm": 0.6487619190965876,
      "learning_rate": 0.00010645994321131518,
      "loss": 12.3124,
      "step": 11323
    },
    {
      "epoch": 1.980153005464481,
      "grad_norm": 0.6897813204492246,
      "learning_rate": 0.00010644581183835891,
      "loss": 12.5547,
      "step": 11324
    },
    {
      "epoch": 1.980327868852459,
      "grad_norm": 0.6119592849863855,
      "learning_rate": 0.0001064316803361443,
      "loss": 12.2217,
      "step": 11325
    },
    {
      "epoch": 1.9805027322404372,
      "grad_norm": 0.6047924314195524,
      "learning_rate": 0.00010641754870495474,
      "loss": 11.9545,
      "step": 11326
    },
    {
      "epoch": 1.9806775956284153,
      "grad_norm": 0.6241208744336816,
      "learning_rate": 0.00010640341694507367,
      "loss": 12.1631,
      "step": 11327
    },
    {
      "epoch": 1.9808524590163934,
      "grad_norm": 0.6427417004622001,
      "learning_rate": 0.00010638928505678438,
      "loss": 12.3663,
      "step": 11328
    },
    {
      "epoch": 1.9810273224043717,
      "grad_norm": 0.6453784793876367,
      "learning_rate": 0.00010637515304037036,
      "loss": 12.2469,
      "step": 11329
    },
    {
      "epoch": 1.9812021857923496,
      "grad_norm": 0.6748600760096624,
      "learning_rate": 0.00010636102089611491,
      "loss": 12.4081,
      "step": 11330
    },
    {
      "epoch": 1.981377049180328,
      "grad_norm": 0.6723848420120706,
      "learning_rate": 0.00010634688862430146,
      "loss": 12.1202,
      "step": 11331
    },
    {
      "epoch": 1.9815519125683059,
      "grad_norm": 0.6406478256328245,
      "learning_rate": 0.00010633275622521339,
      "loss": 12.2908,
      "step": 11332
    },
    {
      "epoch": 1.9817267759562842,
      "grad_norm": 0.7352820698416223,
      "learning_rate": 0.00010631862369913416,
      "loss": 12.5164,
      "step": 11333
    },
    {
      "epoch": 1.9819016393442623,
      "grad_norm": 0.672678884633417,
      "learning_rate": 0.00010630449104634712,
      "loss": 12.4658,
      "step": 11334
    },
    {
      "epoch": 1.9820765027322405,
      "grad_norm": 0.6122427050776609,
      "learning_rate": 0.00010629035826713563,
      "loss": 12.3888,
      "step": 11335
    },
    {
      "epoch": 1.9822513661202186,
      "grad_norm": 0.6518143551556084,
      "learning_rate": 0.00010627622536178319,
      "loss": 12.1817,
      "step": 11336
    },
    {
      "epoch": 1.9824262295081967,
      "grad_norm": 0.6543809116521477,
      "learning_rate": 0.00010626209233057311,
      "loss": 12.4159,
      "step": 11337
    },
    {
      "epoch": 1.9826010928961748,
      "grad_norm": 0.6288704749717338,
      "learning_rate": 0.00010624795917378891,
      "loss": 11.8711,
      "step": 11338
    },
    {
      "epoch": 1.982775956284153,
      "grad_norm": 0.7298234906638249,
      "learning_rate": 0.00010623382589171392,
      "loss": 12.1501,
      "step": 11339
    },
    {
      "epoch": 1.9829508196721313,
      "grad_norm": 0.6332241851227948,
      "learning_rate": 0.00010621969248463157,
      "loss": 12.3439,
      "step": 11340
    },
    {
      "epoch": 1.9831256830601092,
      "grad_norm": 0.6817321097225828,
      "learning_rate": 0.00010620555895282531,
      "loss": 12.3459,
      "step": 11341
    },
    {
      "epoch": 1.9833005464480875,
      "grad_norm": 0.666154242401484,
      "learning_rate": 0.0001061914252965785,
      "loss": 11.9928,
      "step": 11342
    },
    {
      "epoch": 1.9834754098360656,
      "grad_norm": 0.8562049840544128,
      "learning_rate": 0.00010617729151617465,
      "loss": 12.1502,
      "step": 11343
    },
    {
      "epoch": 1.9836502732240437,
      "grad_norm": 0.6905899550934105,
      "learning_rate": 0.0001061631576118971,
      "loss": 12.3438,
      "step": 11344
    },
    {
      "epoch": 1.9838251366120219,
      "grad_norm": 0.7067575911358251,
      "learning_rate": 0.00010614902358402936,
      "loss": 12.2225,
      "step": 11345
    },
    {
      "epoch": 1.984,
      "grad_norm": 0.7509714562089332,
      "learning_rate": 0.00010613488943285478,
      "loss": 11.9411,
      "step": 11346
    },
    {
      "epoch": 1.984174863387978,
      "grad_norm": 0.7439995156965719,
      "learning_rate": 0.00010612075515865685,
      "loss": 12.2935,
      "step": 11347
    },
    {
      "epoch": 1.9843497267759562,
      "grad_norm": 0.7490467594999668,
      "learning_rate": 0.00010610662076171898,
      "loss": 12.4185,
      "step": 11348
    },
    {
      "epoch": 1.9845245901639346,
      "grad_norm": 0.691969622318414,
      "learning_rate": 0.00010609248624232459,
      "loss": 12.1877,
      "step": 11349
    },
    {
      "epoch": 1.9846994535519125,
      "grad_norm": 0.6982499046702286,
      "learning_rate": 0.00010607835160075719,
      "loss": 12.2989,
      "step": 11350
    },
    {
      "epoch": 1.9848743169398908,
      "grad_norm": 0.6346159304090131,
      "learning_rate": 0.00010606421683730014,
      "loss": 11.9258,
      "step": 11351
    },
    {
      "epoch": 1.9850491803278687,
      "grad_norm": 0.6578203020853065,
      "learning_rate": 0.00010605008195223694,
      "loss": 12.2677,
      "step": 11352
    },
    {
      "epoch": 1.985224043715847,
      "grad_norm": 0.6569210874762308,
      "learning_rate": 0.00010603594694585102,
      "loss": 12.2188,
      "step": 11353
    },
    {
      "epoch": 1.9853989071038252,
      "grad_norm": 0.6952385126196351,
      "learning_rate": 0.0001060218118184258,
      "loss": 12.3244,
      "step": 11354
    },
    {
      "epoch": 1.9855737704918033,
      "grad_norm": 0.7705391657944366,
      "learning_rate": 0.00010600767657024482,
      "loss": 11.9359,
      "step": 11355
    },
    {
      "epoch": 1.9857486338797814,
      "grad_norm": 0.6211920183762676,
      "learning_rate": 0.00010599354120159144,
      "loss": 12.13,
      "step": 11356
    },
    {
      "epoch": 1.9859234972677595,
      "grad_norm": 0.5835415949381086,
      "learning_rate": 0.00010597940571274916,
      "loss": 11.8463,
      "step": 11357
    },
    {
      "epoch": 1.9860983606557379,
      "grad_norm": 0.7378123058477039,
      "learning_rate": 0.00010596527010400145,
      "loss": 12.2571,
      "step": 11358
    },
    {
      "epoch": 1.9862732240437158,
      "grad_norm": 0.7299268269873478,
      "learning_rate": 0.00010595113437563176,
      "loss": 12.4778,
      "step": 11359
    },
    {
      "epoch": 1.986448087431694,
      "grad_norm": 0.7444975723320824,
      "learning_rate": 0.00010593699852792354,
      "loss": 12.0464,
      "step": 11360
    },
    {
      "epoch": 1.986622950819672,
      "grad_norm": 0.6421375323907834,
      "learning_rate": 0.00010592286256116027,
      "loss": 12.0473,
      "step": 11361
    },
    {
      "epoch": 1.9867978142076503,
      "grad_norm": 0.801065666147179,
      "learning_rate": 0.00010590872647562541,
      "loss": 12.1464,
      "step": 11362
    },
    {
      "epoch": 1.9869726775956285,
      "grad_norm": 0.7011342652846599,
      "learning_rate": 0.00010589459027160248,
      "loss": 12.198,
      "step": 11363
    },
    {
      "epoch": 1.9871475409836066,
      "grad_norm": 0.7091816642280596,
      "learning_rate": 0.00010588045394937489,
      "loss": 12.3156,
      "step": 11364
    },
    {
      "epoch": 1.9873224043715847,
      "grad_norm": 0.7372897211471042,
      "learning_rate": 0.00010586631750922614,
      "loss": 12.3736,
      "step": 11365
    },
    {
      "epoch": 1.9874972677595628,
      "grad_norm": 0.6986870339297098,
      "learning_rate": 0.00010585218095143972,
      "loss": 12.0787,
      "step": 11366
    },
    {
      "epoch": 1.987672131147541,
      "grad_norm": 0.9287045759416381,
      "learning_rate": 0.00010583804427629911,
      "loss": 12.5629,
      "step": 11367
    },
    {
      "epoch": 1.987846994535519,
      "grad_norm": 0.6775636307635984,
      "learning_rate": 0.00010582390748408777,
      "loss": 12.4149,
      "step": 11368
    },
    {
      "epoch": 1.9880218579234974,
      "grad_norm": 0.7239020842308317,
      "learning_rate": 0.0001058097705750892,
      "loss": 12.3783,
      "step": 11369
    },
    {
      "epoch": 1.9881967213114753,
      "grad_norm": 0.6401258490845884,
      "learning_rate": 0.00010579563354958692,
      "loss": 12.1127,
      "step": 11370
    },
    {
      "epoch": 1.9883715846994536,
      "grad_norm": 0.7105477496578566,
      "learning_rate": 0.00010578149640786437,
      "loss": 12.2924,
      "step": 11371
    },
    {
      "epoch": 1.9885464480874315,
      "grad_norm": 0.6602092266288349,
      "learning_rate": 0.00010576735915020508,
      "loss": 12.2976,
      "step": 11372
    },
    {
      "epoch": 1.9887213114754099,
      "grad_norm": 0.6812440016013469,
      "learning_rate": 0.00010575322177689251,
      "loss": 12.1762,
      "step": 11373
    },
    {
      "epoch": 1.988896174863388,
      "grad_norm": 0.7066614151475651,
      "learning_rate": 0.0001057390842882102,
      "loss": 12.3753,
      "step": 11374
    },
    {
      "epoch": 1.989071038251366,
      "grad_norm": 0.7122068699024164,
      "learning_rate": 0.00010572494668444163,
      "loss": 12.0903,
      "step": 11375
    },
    {
      "epoch": 1.9892459016393442,
      "grad_norm": 0.6569747076378881,
      "learning_rate": 0.00010571080896587027,
      "loss": 12.3981,
      "step": 11376
    },
    {
      "epoch": 1.9894207650273223,
      "grad_norm": 0.7924327797190751,
      "learning_rate": 0.00010569667113277971,
      "loss": 12.4661,
      "step": 11377
    },
    {
      "epoch": 1.9895956284153007,
      "grad_norm": 0.6252037264283313,
      "learning_rate": 0.00010568253318545333,
      "loss": 12.1169,
      "step": 11378
    },
    {
      "epoch": 1.9897704918032786,
      "grad_norm": 0.7296112626912047,
      "learning_rate": 0.00010566839512417479,
      "loss": 12.1669,
      "step": 11379
    },
    {
      "epoch": 1.989945355191257,
      "grad_norm": 0.7586986326806932,
      "learning_rate": 0.00010565425694922745,
      "loss": 12.1908,
      "step": 11380
    },
    {
      "epoch": 1.9901202185792348,
      "grad_norm": 0.6307943461084272,
      "learning_rate": 0.00010564011866089496,
      "loss": 11.8816,
      "step": 11381
    },
    {
      "epoch": 1.9902950819672132,
      "grad_norm": 0.6533372436186261,
      "learning_rate": 0.00010562598025946073,
      "loss": 12.3772,
      "step": 11382
    },
    {
      "epoch": 1.9904699453551913,
      "grad_norm": 0.6991346977897506,
      "learning_rate": 0.00010561184174520833,
      "loss": 12.2736,
      "step": 11383
    },
    {
      "epoch": 1.9906448087431694,
      "grad_norm": 0.685946841130245,
      "learning_rate": 0.0001055977031184213,
      "loss": 12.2396,
      "step": 11384
    },
    {
      "epoch": 1.9908196721311475,
      "grad_norm": 0.6325093099217145,
      "learning_rate": 0.00010558356437938308,
      "loss": 12.0823,
      "step": 11385
    },
    {
      "epoch": 1.9909945355191256,
      "grad_norm": 0.6952033220640795,
      "learning_rate": 0.00010556942552837732,
      "loss": 12.0589,
      "step": 11386
    },
    {
      "epoch": 1.991169398907104,
      "grad_norm": 0.6547698896620844,
      "learning_rate": 0.00010555528656568745,
      "loss": 12.2961,
      "step": 11387
    },
    {
      "epoch": 1.9913442622950819,
      "grad_norm": 0.6608361515092187,
      "learning_rate": 0.000105541147491597,
      "loss": 12.0757,
      "step": 11388
    },
    {
      "epoch": 1.9915191256830602,
      "grad_norm": 0.7144451901777532,
      "learning_rate": 0.00010552700830638955,
      "loss": 12.4754,
      "step": 11389
    },
    {
      "epoch": 1.9916939890710381,
      "grad_norm": 0.7630771134672876,
      "learning_rate": 0.00010551286901034863,
      "loss": 12.2077,
      "step": 11390
    },
    {
      "epoch": 1.9918688524590165,
      "grad_norm": 0.7275492628185525,
      "learning_rate": 0.00010549872960375773,
      "loss": 12.3134,
      "step": 11391
    },
    {
      "epoch": 1.9920437158469946,
      "grad_norm": 0.6521927679008185,
      "learning_rate": 0.00010548459008690041,
      "loss": 12.1799,
      "step": 11392
    },
    {
      "epoch": 1.9922185792349727,
      "grad_norm": 0.716448975636987,
      "learning_rate": 0.00010547045046006028,
      "loss": 12.2741,
      "step": 11393
    },
    {
      "epoch": 1.9923934426229508,
      "grad_norm": 0.6440995038587529,
      "learning_rate": 0.00010545631072352078,
      "loss": 12.0373,
      "step": 11394
    },
    {
      "epoch": 1.992568306010929,
      "grad_norm": 0.6170809310186249,
      "learning_rate": 0.0001054421708775655,
      "loss": 12.2032,
      "step": 11395
    },
    {
      "epoch": 1.992743169398907,
      "grad_norm": 0.6665963026042717,
      "learning_rate": 0.00010542803092247798,
      "loss": 12.0465,
      "step": 11396
    },
    {
      "epoch": 1.9929180327868852,
      "grad_norm": 0.6439466273437296,
      "learning_rate": 0.00010541389085854176,
      "loss": 12.094,
      "step": 11397
    },
    {
      "epoch": 1.9930928961748635,
      "grad_norm": 0.6473881232267258,
      "learning_rate": 0.00010539975068604043,
      "loss": 11.9911,
      "step": 11398
    },
    {
      "epoch": 1.9932677595628414,
      "grad_norm": 0.6955976384076634,
      "learning_rate": 0.00010538561040525752,
      "loss": 12.3017,
      "step": 11399
    },
    {
      "epoch": 1.9934426229508198,
      "grad_norm": 0.6449806188071737,
      "learning_rate": 0.00010537147001647656,
      "loss": 12.2267,
      "step": 11400
    },
    {
      "epoch": 1.9936174863387977,
      "grad_norm": 0.7002995676831624,
      "learning_rate": 0.00010535732951998117,
      "loss": 12.013,
      "step": 11401
    },
    {
      "epoch": 1.993792349726776,
      "grad_norm": 0.7345086675547635,
      "learning_rate": 0.00010534318891605483,
      "loss": 12.3074,
      "step": 11402
    },
    {
      "epoch": 1.9939672131147541,
      "grad_norm": 0.6142100623930219,
      "learning_rate": 0.00010532904820498118,
      "loss": 11.9606,
      "step": 11403
    },
    {
      "epoch": 1.9941420765027322,
      "grad_norm": 0.6456692471574496,
      "learning_rate": 0.00010531490738704375,
      "loss": 12.2187,
      "step": 11404
    },
    {
      "epoch": 1.9943169398907103,
      "grad_norm": 0.6551591764155362,
      "learning_rate": 0.0001053007664625261,
      "loss": 12.3058,
      "step": 11405
    },
    {
      "epoch": 1.9944918032786885,
      "grad_norm": 0.7261241903104162,
      "learning_rate": 0.0001052866254317118,
      "loss": 12.1313,
      "step": 11406
    },
    {
      "epoch": 1.9946666666666668,
      "grad_norm": 0.6979942002945572,
      "learning_rate": 0.00010527248429488443,
      "loss": 12.1992,
      "step": 11407
    },
    {
      "epoch": 1.9948415300546447,
      "grad_norm": 0.6578779103907272,
      "learning_rate": 0.00010525834305232755,
      "loss": 12.247,
      "step": 11408
    },
    {
      "epoch": 1.995016393442623,
      "grad_norm": 0.6593294143874759,
      "learning_rate": 0.00010524420170432479,
      "loss": 12.1715,
      "step": 11409
    },
    {
      "epoch": 1.995191256830601,
      "grad_norm": 0.7369042000931258,
      "learning_rate": 0.00010523006025115965,
      "loss": 12.1643,
      "step": 11410
    },
    {
      "epoch": 1.9953661202185793,
      "grad_norm": 0.6940551460082206,
      "learning_rate": 0.00010521591869311578,
      "loss": 12.1225,
      "step": 11411
    },
    {
      "epoch": 1.9955409836065574,
      "grad_norm": 0.6663969587662347,
      "learning_rate": 0.00010520177703047666,
      "loss": 12.3347,
      "step": 11412
    },
    {
      "epoch": 1.9957158469945355,
      "grad_norm": 0.7184844689787407,
      "learning_rate": 0.00010518763526352597,
      "loss": 12.1329,
      "step": 11413
    },
    {
      "epoch": 1.9958907103825136,
      "grad_norm": 0.6332997712727627,
      "learning_rate": 0.00010517349339254727,
      "loss": 12.1148,
      "step": 11414
    },
    {
      "epoch": 1.9960655737704918,
      "grad_norm": 0.6256930213186597,
      "learning_rate": 0.00010515935141782414,
      "loss": 12.3376,
      "step": 11415
    },
    {
      "epoch": 1.99624043715847,
      "grad_norm": 0.6357582349735288,
      "learning_rate": 0.00010514520933964018,
      "loss": 12.1351,
      "step": 11416
    },
    {
      "epoch": 1.996415300546448,
      "grad_norm": 0.6103717620059921,
      "learning_rate": 0.00010513106715827896,
      "loss": 12.3588,
      "step": 11417
    },
    {
      "epoch": 1.9965901639344263,
      "grad_norm": 0.6655336324159813,
      "learning_rate": 0.00010511692487402414,
      "loss": 12.1611,
      "step": 11418
    },
    {
      "epoch": 1.9967650273224042,
      "grad_norm": 0.6325973539148457,
      "learning_rate": 0.00010510278248715919,
      "loss": 12.1827,
      "step": 11419
    },
    {
      "epoch": 1.9969398907103826,
      "grad_norm": 0.7150002622558357,
      "learning_rate": 0.00010508863999796782,
      "loss": 12.0905,
      "step": 11420
    },
    {
      "epoch": 1.9971147540983607,
      "grad_norm": 0.6338628761232917,
      "learning_rate": 0.00010507449740673359,
      "loss": 12.2508,
      "step": 11421
    },
    {
      "epoch": 1.9972896174863388,
      "grad_norm": 0.7329469879557072,
      "learning_rate": 0.00010506035471374011,
      "loss": 12.2805,
      "step": 11422
    },
    {
      "epoch": 1.997464480874317,
      "grad_norm": 0.6539070172905247,
      "learning_rate": 0.000105046211919271,
      "loss": 12.2524,
      "step": 11423
    },
    {
      "epoch": 1.997639344262295,
      "grad_norm": 0.694171964924821,
      "learning_rate": 0.0001050320690236098,
      "loss": 12.258,
      "step": 11424
    },
    {
      "epoch": 1.9978142076502732,
      "grad_norm": 0.6422624216336815,
      "learning_rate": 0.00010501792602704022,
      "loss": 12.2353,
      "step": 11425
    },
    {
      "epoch": 1.9979890710382513,
      "grad_norm": 0.7420845287593487,
      "learning_rate": 0.00010500378292984576,
      "loss": 12.0977,
      "step": 11426
    },
    {
      "epoch": 1.9981639344262296,
      "grad_norm": 0.7383200012434621,
      "learning_rate": 0.00010498963973231013,
      "loss": 12.0243,
      "step": 11427
    },
    {
      "epoch": 1.9983387978142075,
      "grad_norm": 0.7526420659944598,
      "learning_rate": 0.00010497549643471687,
      "loss": 12.2087,
      "step": 11428
    },
    {
      "epoch": 1.9985136612021859,
      "grad_norm": 0.6411186916173992,
      "learning_rate": 0.00010496135303734967,
      "loss": 12.116,
      "step": 11429
    },
    {
      "epoch": 1.9986885245901638,
      "grad_norm": 0.7183197769911387,
      "learning_rate": 0.00010494720954049209,
      "loss": 12.4232,
      "step": 11430
    },
    {
      "epoch": 1.9988633879781421,
      "grad_norm": 0.585555972089451,
      "learning_rate": 0.00010493306594442773,
      "loss": 12.2454,
      "step": 11431
    },
    {
      "epoch": 1.9990382513661202,
      "grad_norm": 0.6894847600373266,
      "learning_rate": 0.00010491892224944031,
      "loss": 12.393,
      "step": 11432
    },
    {
      "epoch": 1.9992131147540984,
      "grad_norm": 0.6699950474022041,
      "learning_rate": 0.00010490477845581337,
      "loss": 11.9398,
      "step": 11433
    },
    {
      "epoch": 1.9993879781420765,
      "grad_norm": 0.6928153503292406,
      "learning_rate": 0.00010489063456383057,
      "loss": 12.2135,
      "step": 11434
    },
    {
      "epoch": 1.9995628415300546,
      "grad_norm": 0.6720104619336751,
      "learning_rate": 0.00010487649057377553,
      "loss": 12.395,
      "step": 11435
    },
    {
      "epoch": 1.999737704918033,
      "grad_norm": 0.6922773987524429,
      "learning_rate": 0.00010486234648593188,
      "loss": 12.3278,
      "step": 11436
    },
    {
      "epoch": 1.9999125683060108,
      "grad_norm": 0.7116501074357043,
      "learning_rate": 0.00010484820230058325,
      "loss": 12.3744,
      "step": 11437
    },
    {
      "epoch": 2.000087431693989,
      "grad_norm": 0.7085174658808369,
      "learning_rate": 0.00010483405801801327,
      "loss": 11.4048,
      "step": 11438
    },
    {
      "epoch": 2.000262295081967,
      "grad_norm": 0.7891795734042671,
      "learning_rate": 0.0001048199136385056,
      "loss": 11.9798,
      "step": 11439
    },
    {
      "epoch": 2.0004371584699454,
      "grad_norm": 0.6218003248481266,
      "learning_rate": 0.00010480576916234384,
      "loss": 12.2728,
      "step": 11440
    },
    {
      "epoch": 2.0006120218579233,
      "grad_norm": 0.7326963465070171,
      "learning_rate": 0.00010479162458981168,
      "loss": 12.1194,
      "step": 11441
    },
    {
      "epoch": 2.0007868852459016,
      "grad_norm": 0.6988167048902554,
      "learning_rate": 0.00010477747992119273,
      "loss": 11.6144,
      "step": 11442
    },
    {
      "epoch": 2.00096174863388,
      "grad_norm": 0.6906441899001037,
      "learning_rate": 0.00010476333515677064,
      "loss": 12.084,
      "step": 11443
    },
    {
      "epoch": 2.001136612021858,
      "grad_norm": 0.7892543738162127,
      "learning_rate": 0.00010474919029682903,
      "loss": 12.0149,
      "step": 11444
    },
    {
      "epoch": 2.0013114754098362,
      "grad_norm": 0.7078739074534438,
      "learning_rate": 0.0001047350453416516,
      "loss": 11.9808,
      "step": 11445
    },
    {
      "epoch": 2.001486338797814,
      "grad_norm": 0.6793109606575872,
      "learning_rate": 0.00010472090029152196,
      "loss": 11.9663,
      "step": 11446
    },
    {
      "epoch": 2.0016612021857925,
      "grad_norm": 0.6262514919088523,
      "learning_rate": 0.00010470675514672377,
      "loss": 11.7213,
      "step": 11447
    },
    {
      "epoch": 2.0018360655737704,
      "grad_norm": 0.6861405837759686,
      "learning_rate": 0.00010469260990754068,
      "loss": 12.1197,
      "step": 11448
    },
    {
      "epoch": 2.0020109289617487,
      "grad_norm": 0.6517349904048979,
      "learning_rate": 0.00010467846457425636,
      "loss": 11.8756,
      "step": 11449
    },
    {
      "epoch": 2.0021857923497266,
      "grad_norm": 0.6534647408825905,
      "learning_rate": 0.00010466431914715449,
      "loss": 11.9253,
      "step": 11450
    },
    {
      "epoch": 2.002360655737705,
      "grad_norm": 0.6151433554548947,
      "learning_rate": 0.00010465017362651868,
      "loss": 12.0141,
      "step": 11451
    },
    {
      "epoch": 2.002535519125683,
      "grad_norm": 0.6736736770620559,
      "learning_rate": 0.0001046360280126326,
      "loss": 12.1378,
      "step": 11452
    },
    {
      "epoch": 2.002710382513661,
      "grad_norm": 0.6523444802312764,
      "learning_rate": 0.00010462188230577992,
      "loss": 11.7528,
      "step": 11453
    },
    {
      "epoch": 2.0028852459016395,
      "grad_norm": 0.7473582434976749,
      "learning_rate": 0.00010460773650624431,
      "loss": 11.8546,
      "step": 11454
    },
    {
      "epoch": 2.0030601092896174,
      "grad_norm": 0.7074134412246925,
      "learning_rate": 0.00010459359061430947,
      "loss": 11.7374,
      "step": 11455
    },
    {
      "epoch": 2.0032349726775958,
      "grad_norm": 0.6343738753353977,
      "learning_rate": 0.00010457944463025899,
      "loss": 12.1657,
      "step": 11456
    },
    {
      "epoch": 2.0034098360655737,
      "grad_norm": 0.7240869728721676,
      "learning_rate": 0.0001045652985543766,
      "loss": 12.1728,
      "step": 11457
    },
    {
      "epoch": 2.003584699453552,
      "grad_norm": 0.6305105742300793,
      "learning_rate": 0.00010455115238694597,
      "loss": 11.9151,
      "step": 11458
    },
    {
      "epoch": 2.00375956284153,
      "grad_norm": 0.7027266840226898,
      "learning_rate": 0.00010453700612825077,
      "loss": 12.021,
      "step": 11459
    },
    {
      "epoch": 2.0039344262295082,
      "grad_norm": 0.6651546665854513,
      "learning_rate": 0.00010452285977857463,
      "loss": 12.0224,
      "step": 11460
    },
    {
      "epoch": 2.004109289617486,
      "grad_norm": 0.6871770527791581,
      "learning_rate": 0.00010450871333820129,
      "loss": 12.0438,
      "step": 11461
    },
    {
      "epoch": 2.0042841530054645,
      "grad_norm": 0.7284903116355121,
      "learning_rate": 0.00010449456680741438,
      "loss": 12.019,
      "step": 11462
    },
    {
      "epoch": 2.004459016393443,
      "grad_norm": 0.7000986402738196,
      "learning_rate": 0.00010448042018649762,
      "loss": 11.9796,
      "step": 11463
    },
    {
      "epoch": 2.0046338797814207,
      "grad_norm": 0.6132713474819632,
      "learning_rate": 0.0001044662734757347,
      "loss": 11.9284,
      "step": 11464
    },
    {
      "epoch": 2.004808743169399,
      "grad_norm": 0.6201038865200729,
      "learning_rate": 0.00010445212667540924,
      "loss": 11.9192,
      "step": 11465
    },
    {
      "epoch": 2.004983606557377,
      "grad_norm": 0.6034094971921897,
      "learning_rate": 0.00010443797978580502,
      "loss": 11.8495,
      "step": 11466
    },
    {
      "epoch": 2.0051584699453553,
      "grad_norm": 0.7550799561148454,
      "learning_rate": 0.00010442383280720564,
      "loss": 11.9745,
      "step": 11467
    },
    {
      "epoch": 2.005333333333333,
      "grad_norm": 0.6917786593458577,
      "learning_rate": 0.00010440968573989486,
      "loss": 12.0503,
      "step": 11468
    },
    {
      "epoch": 2.0055081967213115,
      "grad_norm": 0.6814088851387192,
      "learning_rate": 0.0001043955385841563,
      "loss": 11.8587,
      "step": 11469
    },
    {
      "epoch": 2.0056830601092894,
      "grad_norm": 0.6352499644200021,
      "learning_rate": 0.00010438139134027372,
      "loss": 11.5082,
      "step": 11470
    },
    {
      "epoch": 2.0058579234972678,
      "grad_norm": 0.6450557228115583,
      "learning_rate": 0.00010436724400853081,
      "loss": 11.6555,
      "step": 11471
    },
    {
      "epoch": 2.006032786885246,
      "grad_norm": 0.7712964608777182,
      "learning_rate": 0.00010435309658921122,
      "loss": 12.0241,
      "step": 11472
    },
    {
      "epoch": 2.006207650273224,
      "grad_norm": 0.6493242473944092,
      "learning_rate": 0.00010433894908259869,
      "loss": 12.0021,
      "step": 11473
    },
    {
      "epoch": 2.0063825136612023,
      "grad_norm": 0.6771317328161395,
      "learning_rate": 0.0001043248014889769,
      "loss": 11.9695,
      "step": 11474
    },
    {
      "epoch": 2.0065573770491802,
      "grad_norm": 0.6302081014711041,
      "learning_rate": 0.00010431065380862959,
      "loss": 11.9335,
      "step": 11475
    },
    {
      "epoch": 2.0067322404371586,
      "grad_norm": 0.6439682933302393,
      "learning_rate": 0.00010429650604184038,
      "loss": 12.0183,
      "step": 11476
    },
    {
      "epoch": 2.0069071038251365,
      "grad_norm": 0.617847871488541,
      "learning_rate": 0.0001042823581888931,
      "loss": 11.9474,
      "step": 11477
    },
    {
      "epoch": 2.007081967213115,
      "grad_norm": 0.6622037415165232,
      "learning_rate": 0.00010426821025007134,
      "loss": 11.7775,
      "step": 11478
    },
    {
      "epoch": 2.0072568306010927,
      "grad_norm": 0.6688628529154064,
      "learning_rate": 0.00010425406222565885,
      "loss": 11.9033,
      "step": 11479
    },
    {
      "epoch": 2.007431693989071,
      "grad_norm": 0.6957376699198959,
      "learning_rate": 0.00010423991411593939,
      "loss": 11.8305,
      "step": 11480
    },
    {
      "epoch": 2.007606557377049,
      "grad_norm": 0.6588457985455033,
      "learning_rate": 0.0001042257659211966,
      "loss": 12.007,
      "step": 11481
    },
    {
      "epoch": 2.0077814207650273,
      "grad_norm": 0.6352285481238666,
      "learning_rate": 0.00010421161764171426,
      "loss": 11.757,
      "step": 11482
    },
    {
      "epoch": 2.0079562841530056,
      "grad_norm": 0.6117864334188431,
      "learning_rate": 0.00010419746927777605,
      "loss": 11.8647,
      "step": 11483
    },
    {
      "epoch": 2.0081311475409835,
      "grad_norm": 0.6863259382304231,
      "learning_rate": 0.00010418332082966567,
      "loss": 11.8934,
      "step": 11484
    },
    {
      "epoch": 2.008306010928962,
      "grad_norm": 0.6805718641058119,
      "learning_rate": 0.00010416917229766687,
      "loss": 11.9819,
      "step": 11485
    },
    {
      "epoch": 2.00848087431694,
      "grad_norm": 0.6643588512749375,
      "learning_rate": 0.00010415502368206339,
      "loss": 11.9703,
      "step": 11486
    },
    {
      "epoch": 2.008655737704918,
      "grad_norm": 0.6360681717421941,
      "learning_rate": 0.0001041408749831389,
      "loss": 12.116,
      "step": 11487
    },
    {
      "epoch": 2.008830601092896,
      "grad_norm": 0.667388509919463,
      "learning_rate": 0.00010412672620117714,
      "loss": 11.8802,
      "step": 11488
    },
    {
      "epoch": 2.0090054644808744,
      "grad_norm": 0.6794670069239708,
      "learning_rate": 0.00010411257733646189,
      "loss": 12.0328,
      "step": 11489
    },
    {
      "epoch": 2.0091803278688523,
      "grad_norm": 0.634108719079902,
      "learning_rate": 0.00010409842838927682,
      "loss": 12.057,
      "step": 11490
    },
    {
      "epoch": 2.0093551912568306,
      "grad_norm": 0.7895009769785681,
      "learning_rate": 0.00010408427935990568,
      "loss": 12.0503,
      "step": 11491
    },
    {
      "epoch": 2.009530054644809,
      "grad_norm": 0.6900011388101996,
      "learning_rate": 0.00010407013024863219,
      "loss": 12.0061,
      "step": 11492
    },
    {
      "epoch": 2.009704918032787,
      "grad_norm": 0.7172293427044801,
      "learning_rate": 0.00010405598105574008,
      "loss": 11.9855,
      "step": 11493
    },
    {
      "epoch": 2.009879781420765,
      "grad_norm": 0.6485489584896484,
      "learning_rate": 0.00010404183178151312,
      "loss": 12.0813,
      "step": 11494
    },
    {
      "epoch": 2.010054644808743,
      "grad_norm": 0.6322263316520423,
      "learning_rate": 0.000104027682426235,
      "loss": 12.1194,
      "step": 11495
    },
    {
      "epoch": 2.0102295081967214,
      "grad_norm": 0.6272580882409236,
      "learning_rate": 0.0001040135329901895,
      "loss": 12.0845,
      "step": 11496
    },
    {
      "epoch": 2.0104043715846993,
      "grad_norm": 0.7101898856671656,
      "learning_rate": 0.00010399938347366034,
      "loss": 11.9014,
      "step": 11497
    },
    {
      "epoch": 2.0105792349726777,
      "grad_norm": 0.6589603323331092,
      "learning_rate": 0.00010398523387693124,
      "loss": 11.8448,
      "step": 11498
    },
    {
      "epoch": 2.0107540983606556,
      "grad_norm": 0.6316992401128498,
      "learning_rate": 0.00010397108420028598,
      "loss": 11.9141,
      "step": 11499
    },
    {
      "epoch": 2.010928961748634,
      "grad_norm": 0.6572128782816388,
      "learning_rate": 0.00010395693444400829,
      "loss": 11.892,
      "step": 11500
    },
    {
      "epoch": 2.0111038251366122,
      "grad_norm": 0.6318729827232925,
      "learning_rate": 0.00010394278460838192,
      "loss": 11.8647,
      "step": 11501
    },
    {
      "epoch": 2.01127868852459,
      "grad_norm": 0.6454115653882396,
      "learning_rate": 0.0001039286346936906,
      "loss": 11.8448,
      "step": 11502
    },
    {
      "epoch": 2.0114535519125685,
      "grad_norm": 0.7030997692039274,
      "learning_rate": 0.0001039144847002181,
      "loss": 11.9971,
      "step": 11503
    },
    {
      "epoch": 2.0116284153005464,
      "grad_norm": 0.6542048141434951,
      "learning_rate": 0.00010390033462824817,
      "loss": 11.9427,
      "step": 11504
    },
    {
      "epoch": 2.0118032786885247,
      "grad_norm": 0.6661234102541929,
      "learning_rate": 0.00010388618447806455,
      "loss": 11.8679,
      "step": 11505
    },
    {
      "epoch": 2.0119781420765026,
      "grad_norm": 0.6735778566384226,
      "learning_rate": 0.00010387203424995098,
      "loss": 11.8357,
      "step": 11506
    },
    {
      "epoch": 2.012153005464481,
      "grad_norm": 0.6310628175789863,
      "learning_rate": 0.00010385788394419127,
      "loss": 11.6466,
      "step": 11507
    },
    {
      "epoch": 2.012327868852459,
      "grad_norm": 0.7707759844661036,
      "learning_rate": 0.00010384373356106911,
      "loss": 11.8885,
      "step": 11508
    },
    {
      "epoch": 2.012502732240437,
      "grad_norm": 0.6379777769358127,
      "learning_rate": 0.0001038295831008683,
      "loss": 11.6574,
      "step": 11509
    },
    {
      "epoch": 2.012677595628415,
      "grad_norm": 0.681057775386774,
      "learning_rate": 0.00010381543256387259,
      "loss": 12.052,
      "step": 11510
    },
    {
      "epoch": 2.0128524590163934,
      "grad_norm": 0.6927564050465735,
      "learning_rate": 0.00010380128195036576,
      "loss": 12.0338,
      "step": 11511
    },
    {
      "epoch": 2.0130273224043718,
      "grad_norm": 0.7196058528922978,
      "learning_rate": 0.00010378713126063153,
      "loss": 12.0326,
      "step": 11512
    },
    {
      "epoch": 2.0132021857923497,
      "grad_norm": 0.7969377277771561,
      "learning_rate": 0.0001037729804949537,
      "loss": 12.1465,
      "step": 11513
    },
    {
      "epoch": 2.013377049180328,
      "grad_norm": 0.619921032202042,
      "learning_rate": 0.00010375882965361605,
      "loss": 12.0008,
      "step": 11514
    },
    {
      "epoch": 2.013551912568306,
      "grad_norm": 0.6668984136148671,
      "learning_rate": 0.00010374467873690228,
      "loss": 11.8148,
      "step": 11515
    },
    {
      "epoch": 2.0137267759562842,
      "grad_norm": 0.7534176076371693,
      "learning_rate": 0.00010373052774509624,
      "loss": 11.8719,
      "step": 11516
    },
    {
      "epoch": 2.013901639344262,
      "grad_norm": 0.7389165200632354,
      "learning_rate": 0.00010371637667848161,
      "loss": 11.9613,
      "step": 11517
    },
    {
      "epoch": 2.0140765027322405,
      "grad_norm": 0.7025869433094909,
      "learning_rate": 0.00010370222553734228,
      "loss": 12.0485,
      "step": 11518
    },
    {
      "epoch": 2.0142513661202184,
      "grad_norm": 0.7786148873339083,
      "learning_rate": 0.00010368807432196194,
      "loss": 11.8372,
      "step": 11519
    },
    {
      "epoch": 2.0144262295081967,
      "grad_norm": 0.6465107879940448,
      "learning_rate": 0.00010367392303262438,
      "loss": 11.7431,
      "step": 11520
    },
    {
      "epoch": 2.014601092896175,
      "grad_norm": 0.8535322528972139,
      "learning_rate": 0.0001036597716696134,
      "loss": 11.6779,
      "step": 11521
    },
    {
      "epoch": 2.014775956284153,
      "grad_norm": 0.6143760602070802,
      "learning_rate": 0.00010364562023321272,
      "loss": 11.8118,
      "step": 11522
    },
    {
      "epoch": 2.0149508196721313,
      "grad_norm": 0.7462589067006141,
      "learning_rate": 0.00010363146872370622,
      "loss": 12.106,
      "step": 11523
    },
    {
      "epoch": 2.015125683060109,
      "grad_norm": 0.6270354203791894,
      "learning_rate": 0.00010361731714137757,
      "loss": 11.8055,
      "step": 11524
    },
    {
      "epoch": 2.0153005464480875,
      "grad_norm": 0.7566897643514511,
      "learning_rate": 0.00010360316548651063,
      "loss": 12.0479,
      "step": 11525
    },
    {
      "epoch": 2.0154754098360654,
      "grad_norm": 0.6658738635806455,
      "learning_rate": 0.00010358901375938916,
      "loss": 12.1381,
      "step": 11526
    },
    {
      "epoch": 2.0156502732240438,
      "grad_norm": 0.6715350054222256,
      "learning_rate": 0.00010357486196029692,
      "loss": 11.7741,
      "step": 11527
    },
    {
      "epoch": 2.0158251366120217,
      "grad_norm": 0.6682341848768882,
      "learning_rate": 0.00010356071008951775,
      "loss": 11.9461,
      "step": 11528
    },
    {
      "epoch": 2.016,
      "grad_norm": 0.6435872983686965,
      "learning_rate": 0.00010354655814733538,
      "loss": 12.081,
      "step": 11529
    },
    {
      "epoch": 2.0161748633879784,
      "grad_norm": 0.6865452348495916,
      "learning_rate": 0.00010353240613403368,
      "loss": 12.0441,
      "step": 11530
    },
    {
      "epoch": 2.0163497267759563,
      "grad_norm": 0.5887338993516024,
      "learning_rate": 0.00010351825404989635,
      "loss": 11.9067,
      "step": 11531
    },
    {
      "epoch": 2.0165245901639346,
      "grad_norm": 0.6130583538017462,
      "learning_rate": 0.00010350410189520723,
      "loss": 12.1212,
      "step": 11532
    },
    {
      "epoch": 2.0166994535519125,
      "grad_norm": 0.6664357679120047,
      "learning_rate": 0.00010348994967025012,
      "loss": 12.1026,
      "step": 11533
    },
    {
      "epoch": 2.016874316939891,
      "grad_norm": 0.6095211526505853,
      "learning_rate": 0.00010347579737530879,
      "loss": 11.9176,
      "step": 11534
    },
    {
      "epoch": 2.0170491803278687,
      "grad_norm": 0.7194609497268513,
      "learning_rate": 0.00010346164501066706,
      "loss": 12.0229,
      "step": 11535
    },
    {
      "epoch": 2.017224043715847,
      "grad_norm": 0.6330044274748657,
      "learning_rate": 0.00010344749257660868,
      "loss": 11.8056,
      "step": 11536
    },
    {
      "epoch": 2.017398907103825,
      "grad_norm": 0.6749755773323765,
      "learning_rate": 0.00010343334007341754,
      "loss": 12.0125,
      "step": 11537
    },
    {
      "epoch": 2.0175737704918033,
      "grad_norm": 0.6458571015218282,
      "learning_rate": 0.00010341918750137737,
      "loss": 12.1225,
      "step": 11538
    },
    {
      "epoch": 2.017748633879781,
      "grad_norm": 0.5998842435876202,
      "learning_rate": 0.000103405034860772,
      "loss": 11.8914,
      "step": 11539
    },
    {
      "epoch": 2.0179234972677595,
      "grad_norm": 0.670394874690994,
      "learning_rate": 0.00010339088215188524,
      "loss": 12.0057,
      "step": 11540
    },
    {
      "epoch": 2.018098360655738,
      "grad_norm": 0.6189986822617894,
      "learning_rate": 0.00010337672937500085,
      "loss": 11.8282,
      "step": 11541
    },
    {
      "epoch": 2.018273224043716,
      "grad_norm": 0.6341040356145229,
      "learning_rate": 0.00010336257653040269,
      "loss": 11.8017,
      "step": 11542
    },
    {
      "epoch": 2.018448087431694,
      "grad_norm": 0.6500019639317532,
      "learning_rate": 0.00010334842361837454,
      "loss": 12.0625,
      "step": 11543
    },
    {
      "epoch": 2.018622950819672,
      "grad_norm": 0.7034969906918085,
      "learning_rate": 0.0001033342706392002,
      "loss": 11.7456,
      "step": 11544
    },
    {
      "epoch": 2.0187978142076504,
      "grad_norm": 0.7174512539270851,
      "learning_rate": 0.00010332011759316351,
      "loss": 11.9122,
      "step": 11545
    },
    {
      "epoch": 2.0189726775956283,
      "grad_norm": 0.6452965595337758,
      "learning_rate": 0.00010330596448054826,
      "loss": 12.0864,
      "step": 11546
    },
    {
      "epoch": 2.0191475409836066,
      "grad_norm": 0.6226094711693787,
      "learning_rate": 0.00010329181130163828,
      "loss": 11.7728,
      "step": 11547
    },
    {
      "epoch": 2.0193224043715845,
      "grad_norm": 0.6223165703252119,
      "learning_rate": 0.00010327765805671739,
      "loss": 11.8751,
      "step": 11548
    },
    {
      "epoch": 2.019497267759563,
      "grad_norm": 0.6124090823540658,
      "learning_rate": 0.00010326350474606937,
      "loss": 11.897,
      "step": 11549
    },
    {
      "epoch": 2.019672131147541,
      "grad_norm": 0.6514637276215559,
      "learning_rate": 0.00010324935136997806,
      "loss": 11.9593,
      "step": 11550
    },
    {
      "epoch": 2.019846994535519,
      "grad_norm": 0.6327131606494091,
      "learning_rate": 0.00010323519792872729,
      "loss": 12.0745,
      "step": 11551
    },
    {
      "epoch": 2.0200218579234974,
      "grad_norm": 0.6487066093546905,
      "learning_rate": 0.00010322104442260086,
      "loss": 12.0038,
      "step": 11552
    },
    {
      "epoch": 2.0201967213114753,
      "grad_norm": 0.675596626497308,
      "learning_rate": 0.00010320689085188259,
      "loss": 12.1064,
      "step": 11553
    },
    {
      "epoch": 2.0203715846994537,
      "grad_norm": 0.6716797113870834,
      "learning_rate": 0.00010319273721685634,
      "loss": 11.8412,
      "step": 11554
    },
    {
      "epoch": 2.0205464480874316,
      "grad_norm": 0.6849738020095435,
      "learning_rate": 0.0001031785835178059,
      "loss": 12.0589,
      "step": 11555
    },
    {
      "epoch": 2.02072131147541,
      "grad_norm": 0.6937131593100887,
      "learning_rate": 0.00010316442975501508,
      "loss": 11.9985,
      "step": 11556
    },
    {
      "epoch": 2.020896174863388,
      "grad_norm": 0.6206553861356474,
      "learning_rate": 0.00010315027592876776,
      "loss": 12.1163,
      "step": 11557
    },
    {
      "epoch": 2.021071038251366,
      "grad_norm": 0.6518243299827509,
      "learning_rate": 0.00010313612203934769,
      "loss": 11.9828,
      "step": 11558
    },
    {
      "epoch": 2.0212459016393445,
      "grad_norm": 0.6622874109019751,
      "learning_rate": 0.00010312196808703876,
      "loss": 11.7576,
      "step": 11559
    },
    {
      "epoch": 2.0214207650273224,
      "grad_norm": 0.6782087951768867,
      "learning_rate": 0.0001031078140721248,
      "loss": 11.7346,
      "step": 11560
    },
    {
      "epoch": 2.0215956284153007,
      "grad_norm": 0.6645635590298039,
      "learning_rate": 0.0001030936599948896,
      "loss": 12.1384,
      "step": 11561
    },
    {
      "epoch": 2.0217704918032786,
      "grad_norm": 0.6559634377872223,
      "learning_rate": 0.00010307950585561706,
      "loss": 12.0244,
      "step": 11562
    },
    {
      "epoch": 2.021945355191257,
      "grad_norm": 0.6549113752791027,
      "learning_rate": 0.00010306535165459092,
      "loss": 12.119,
      "step": 11563
    },
    {
      "epoch": 2.022120218579235,
      "grad_norm": 0.6873814049555252,
      "learning_rate": 0.00010305119739209513,
      "loss": 11.989,
      "step": 11564
    },
    {
      "epoch": 2.022295081967213,
      "grad_norm": 0.684089628532615,
      "learning_rate": 0.0001030370430684134,
      "loss": 11.9232,
      "step": 11565
    },
    {
      "epoch": 2.022469945355191,
      "grad_norm": 0.6381009460973398,
      "learning_rate": 0.00010302288868382968,
      "loss": 11.6511,
      "step": 11566
    },
    {
      "epoch": 2.0226448087431694,
      "grad_norm": 0.6369396202683658,
      "learning_rate": 0.00010300873423862773,
      "loss": 11.9378,
      "step": 11567
    },
    {
      "epoch": 2.0228196721311473,
      "grad_norm": 0.6952867871905837,
      "learning_rate": 0.00010299457973309142,
      "loss": 12.235,
      "step": 11568
    },
    {
      "epoch": 2.0229945355191257,
      "grad_norm": 0.6809745276469829,
      "learning_rate": 0.00010298042516750462,
      "loss": 12.0817,
      "step": 11569
    },
    {
      "epoch": 2.023169398907104,
      "grad_norm": 0.6255769150422587,
      "learning_rate": 0.00010296627054215112,
      "loss": 12.0083,
      "step": 11570
    },
    {
      "epoch": 2.023344262295082,
      "grad_norm": 0.6606592456034257,
      "learning_rate": 0.00010295211585731482,
      "loss": 11.928,
      "step": 11571
    },
    {
      "epoch": 2.0235191256830602,
      "grad_norm": 0.7009356883136416,
      "learning_rate": 0.00010293796111327947,
      "loss": 11.8752,
      "step": 11572
    },
    {
      "epoch": 2.023693989071038,
      "grad_norm": 0.6155526875877783,
      "learning_rate": 0.00010292380631032903,
      "loss": 11.926,
      "step": 11573
    },
    {
      "epoch": 2.0238688524590165,
      "grad_norm": 0.6632876251686355,
      "learning_rate": 0.0001029096514487473,
      "loss": 11.9299,
      "step": 11574
    },
    {
      "epoch": 2.0240437158469944,
      "grad_norm": 0.6809195180845375,
      "learning_rate": 0.00010289549652881809,
      "loss": 11.8413,
      "step": 11575
    },
    {
      "epoch": 2.0242185792349727,
      "grad_norm": 0.6866731435270795,
      "learning_rate": 0.00010288134155082532,
      "loss": 12.0524,
      "step": 11576
    },
    {
      "epoch": 2.0243934426229506,
      "grad_norm": 0.7136895744016513,
      "learning_rate": 0.00010286718651505275,
      "loss": 11.8378,
      "step": 11577
    },
    {
      "epoch": 2.024568306010929,
      "grad_norm": 0.61056657239057,
      "learning_rate": 0.00010285303142178435,
      "loss": 11.8511,
      "step": 11578
    },
    {
      "epoch": 2.0247431693989073,
      "grad_norm": 0.7143377133916817,
      "learning_rate": 0.00010283887627130388,
      "loss": 11.9747,
      "step": 11579
    },
    {
      "epoch": 2.024918032786885,
      "grad_norm": 0.6922440475863975,
      "learning_rate": 0.00010282472106389523,
      "loss": 11.8757,
      "step": 11580
    },
    {
      "epoch": 2.0250928961748635,
      "grad_norm": 0.6763479711790348,
      "learning_rate": 0.00010281056579984225,
      "loss": 11.7755,
      "step": 11581
    },
    {
      "epoch": 2.0252677595628414,
      "grad_norm": 0.7465482073876204,
      "learning_rate": 0.0001027964104794288,
      "loss": 12.0848,
      "step": 11582
    },
    {
      "epoch": 2.02544262295082,
      "grad_norm": 0.6444160190785481,
      "learning_rate": 0.00010278225510293873,
      "loss": 11.9285,
      "step": 11583
    },
    {
      "epoch": 2.0256174863387977,
      "grad_norm": 0.7677590156426708,
      "learning_rate": 0.00010276809967065589,
      "loss": 11.9618,
      "step": 11584
    },
    {
      "epoch": 2.025792349726776,
      "grad_norm": 0.6517719121074607,
      "learning_rate": 0.00010275394418286417,
      "loss": 11.9453,
      "step": 11585
    },
    {
      "epoch": 2.025967213114754,
      "grad_norm": 0.7590586509559876,
      "learning_rate": 0.00010273978863984742,
      "loss": 11.9765,
      "step": 11586
    },
    {
      "epoch": 2.0261420765027323,
      "grad_norm": 0.6747549859743238,
      "learning_rate": 0.00010272563304188947,
      "loss": 11.841,
      "step": 11587
    },
    {
      "epoch": 2.0263169398907106,
      "grad_norm": 0.6627455246551369,
      "learning_rate": 0.00010271147738927423,
      "loss": 11.8938,
      "step": 11588
    },
    {
      "epoch": 2.0264918032786885,
      "grad_norm": 0.6663423475407295,
      "learning_rate": 0.00010269732168228551,
      "loss": 11.7596,
      "step": 11589
    },
    {
      "epoch": 2.026666666666667,
      "grad_norm": 0.7185355058100052,
      "learning_rate": 0.00010268316592120724,
      "loss": 12.0082,
      "step": 11590
    },
    {
      "epoch": 2.0268415300546447,
      "grad_norm": 0.6053583467847634,
      "learning_rate": 0.00010266901010632324,
      "loss": 12.0281,
      "step": 11591
    },
    {
      "epoch": 2.027016393442623,
      "grad_norm": 0.6526599296201416,
      "learning_rate": 0.0001026548542379174,
      "loss": 11.9724,
      "step": 11592
    },
    {
      "epoch": 2.027191256830601,
      "grad_norm": 0.685528674129262,
      "learning_rate": 0.00010264069831627359,
      "loss": 11.9049,
      "step": 11593
    },
    {
      "epoch": 2.0273661202185793,
      "grad_norm": 0.6386093153693693,
      "learning_rate": 0.00010262654234167565,
      "loss": 11.8727,
      "step": 11594
    },
    {
      "epoch": 2.027540983606557,
      "grad_norm": 0.667465140251981,
      "learning_rate": 0.00010261238631440748,
      "loss": 11.9513,
      "step": 11595
    },
    {
      "epoch": 2.0277158469945356,
      "grad_norm": 0.5866240837606936,
      "learning_rate": 0.00010259823023475294,
      "loss": 12.1546,
      "step": 11596
    },
    {
      "epoch": 2.0278907103825135,
      "grad_norm": 0.6175639445682707,
      "learning_rate": 0.00010258407410299593,
      "loss": 11.9395,
      "step": 11597
    },
    {
      "epoch": 2.028065573770492,
      "grad_norm": 0.6797288485783662,
      "learning_rate": 0.00010256991791942028,
      "loss": 11.9108,
      "step": 11598
    },
    {
      "epoch": 2.02824043715847,
      "grad_norm": 0.6443542700296144,
      "learning_rate": 0.00010255576168430989,
      "loss": 11.8255,
      "step": 11599
    },
    {
      "epoch": 2.028415300546448,
      "grad_norm": 0.7068634983625908,
      "learning_rate": 0.00010254160539794862,
      "loss": 11.993,
      "step": 11600
    },
    {
      "epoch": 2.0285901639344264,
      "grad_norm": 0.6463491391052291,
      "learning_rate": 0.00010252744906062037,
      "loss": 11.9887,
      "step": 11601
    },
    {
      "epoch": 2.0287650273224043,
      "grad_norm": 0.7397211004858135,
      "learning_rate": 0.00010251329267260901,
      "loss": 12.1187,
      "step": 11602
    },
    {
      "epoch": 2.0289398907103826,
      "grad_norm": 0.6405519283391906,
      "learning_rate": 0.00010249913623419842,
      "loss": 12.0549,
      "step": 11603
    },
    {
      "epoch": 2.0291147540983605,
      "grad_norm": 0.6741086091581272,
      "learning_rate": 0.00010248497974567244,
      "loss": 11.8072,
      "step": 11604
    },
    {
      "epoch": 2.029289617486339,
      "grad_norm": 0.5905221186858797,
      "learning_rate": 0.00010247082320731503,
      "loss": 12.0095,
      "step": 11605
    },
    {
      "epoch": 2.0294644808743167,
      "grad_norm": 0.7040994374521465,
      "learning_rate": 0.00010245666661940998,
      "loss": 11.8439,
      "step": 11606
    },
    {
      "epoch": 2.029639344262295,
      "grad_norm": 0.6495603896270894,
      "learning_rate": 0.00010244250998224126,
      "loss": 11.8065,
      "step": 11607
    },
    {
      "epoch": 2.0298142076502734,
      "grad_norm": 0.6331278266886643,
      "learning_rate": 0.0001024283532960927,
      "loss": 12.0816,
      "step": 11608
    },
    {
      "epoch": 2.0299890710382513,
      "grad_norm": 0.6352452976667267,
      "learning_rate": 0.0001024141965612482,
      "loss": 11.9214,
      "step": 11609
    },
    {
      "epoch": 2.0301639344262297,
      "grad_norm": 0.6842153086544611,
      "learning_rate": 0.00010240003977799166,
      "loss": 12.1045,
      "step": 11610
    },
    {
      "epoch": 2.0303387978142076,
      "grad_norm": 0.6293507740787458,
      "learning_rate": 0.00010238588294660693,
      "loss": 11.9419,
      "step": 11611
    },
    {
      "epoch": 2.030513661202186,
      "grad_norm": 0.7015815958644058,
      "learning_rate": 0.00010237172606737795,
      "loss": 11.9885,
      "step": 11612
    },
    {
      "epoch": 2.030688524590164,
      "grad_norm": 0.6674632453559942,
      "learning_rate": 0.00010235756914058856,
      "loss": 12.0117,
      "step": 11613
    },
    {
      "epoch": 2.030863387978142,
      "grad_norm": 0.6357435615843383,
      "learning_rate": 0.00010234341216652268,
      "loss": 12.0776,
      "step": 11614
    },
    {
      "epoch": 2.03103825136612,
      "grad_norm": 0.8196215820923937,
      "learning_rate": 0.0001023292551454642,
      "loss": 11.8097,
      "step": 11615
    },
    {
      "epoch": 2.0312131147540984,
      "grad_norm": 0.6558096312841755,
      "learning_rate": 0.00010231509807769696,
      "loss": 12.0034,
      "step": 11616
    },
    {
      "epoch": 2.0313879781420763,
      "grad_norm": 0.769799645213425,
      "learning_rate": 0.00010230094096350496,
      "loss": 11.982,
      "step": 11617
    },
    {
      "epoch": 2.0315628415300546,
      "grad_norm": 0.6368818019372005,
      "learning_rate": 0.00010228678380317196,
      "loss": 12.1722,
      "step": 11618
    },
    {
      "epoch": 2.031737704918033,
      "grad_norm": 0.7160642632014436,
      "learning_rate": 0.00010227262659698198,
      "loss": 12.0924,
      "step": 11619
    },
    {
      "epoch": 2.031912568306011,
      "grad_norm": 0.611853043484903,
      "learning_rate": 0.00010225846934521881,
      "loss": 12.032,
      "step": 11620
    },
    {
      "epoch": 2.032087431693989,
      "grad_norm": 0.665841050823645,
      "learning_rate": 0.00010224431204816645,
      "loss": 12.0827,
      "step": 11621
    },
    {
      "epoch": 2.032262295081967,
      "grad_norm": 0.6269075511708081,
      "learning_rate": 0.00010223015470610871,
      "loss": 11.9374,
      "step": 11622
    },
    {
      "epoch": 2.0324371584699454,
      "grad_norm": 0.6918100723461905,
      "learning_rate": 0.00010221599731932952,
      "loss": 12.0167,
      "step": 11623
    },
    {
      "epoch": 2.0326120218579233,
      "grad_norm": 0.6072764152141982,
      "learning_rate": 0.00010220183988811279,
      "loss": 11.9285,
      "step": 11624
    },
    {
      "epoch": 2.0327868852459017,
      "grad_norm": 0.6220785081563964,
      "learning_rate": 0.00010218768241274237,
      "loss": 12.0556,
      "step": 11625
    },
    {
      "epoch": 2.0329617486338796,
      "grad_norm": 0.6319730231168589,
      "learning_rate": 0.00010217352489350225,
      "loss": 11.9691,
      "step": 11626
    },
    {
      "epoch": 2.033136612021858,
      "grad_norm": 0.6604941448238616,
      "learning_rate": 0.00010215936733067627,
      "loss": 11.8124,
      "step": 11627
    },
    {
      "epoch": 2.0333114754098363,
      "grad_norm": 0.6564927465998239,
      "learning_rate": 0.00010214520972454835,
      "loss": 11.9979,
      "step": 11628
    },
    {
      "epoch": 2.033486338797814,
      "grad_norm": 0.6726031474836783,
      "learning_rate": 0.00010213105207540237,
      "loss": 11.9564,
      "step": 11629
    },
    {
      "epoch": 2.0336612021857925,
      "grad_norm": 0.7370816195130274,
      "learning_rate": 0.00010211689438352225,
      "loss": 12.2587,
      "step": 11630
    },
    {
      "epoch": 2.0338360655737704,
      "grad_norm": 0.6752810379874197,
      "learning_rate": 0.00010210273664919191,
      "loss": 11.7597,
      "step": 11631
    },
    {
      "epoch": 2.0340109289617487,
      "grad_norm": 0.651754558027234,
      "learning_rate": 0.00010208857887269525,
      "loss": 12.1466,
      "step": 11632
    },
    {
      "epoch": 2.0341857923497266,
      "grad_norm": 0.7203662272466553,
      "learning_rate": 0.00010207442105431617,
      "loss": 12.0377,
      "step": 11633
    },
    {
      "epoch": 2.034360655737705,
      "grad_norm": 0.6242166563783274,
      "learning_rate": 0.00010206026319433857,
      "loss": 11.9061,
      "step": 11634
    },
    {
      "epoch": 2.034535519125683,
      "grad_norm": 0.7026133194940366,
      "learning_rate": 0.00010204610529304637,
      "loss": 12.0681,
      "step": 11635
    },
    {
      "epoch": 2.034710382513661,
      "grad_norm": 0.7216309973973754,
      "learning_rate": 0.00010203194735072348,
      "loss": 11.9093,
      "step": 11636
    },
    {
      "epoch": 2.0348852459016395,
      "grad_norm": 0.6213801246218533,
      "learning_rate": 0.0001020177893676538,
      "loss": 12.0374,
      "step": 11637
    },
    {
      "epoch": 2.0350601092896174,
      "grad_norm": 0.7495143520555165,
      "learning_rate": 0.00010200363134412125,
      "loss": 11.747,
      "step": 11638
    },
    {
      "epoch": 2.035234972677596,
      "grad_norm": 0.6384006974992995,
      "learning_rate": 0.00010198947328040972,
      "loss": 11.8526,
      "step": 11639
    },
    {
      "epoch": 2.0354098360655737,
      "grad_norm": 0.6262351496642692,
      "learning_rate": 0.00010197531517680319,
      "loss": 11.7146,
      "step": 11640
    },
    {
      "epoch": 2.035584699453552,
      "grad_norm": 0.657579347225521,
      "learning_rate": 0.00010196115703358548,
      "loss": 11.8486,
      "step": 11641
    },
    {
      "epoch": 2.03575956284153,
      "grad_norm": 0.6666420436953234,
      "learning_rate": 0.00010194699885104058,
      "loss": 11.8965,
      "step": 11642
    },
    {
      "epoch": 2.0359344262295083,
      "grad_norm": 0.653510295378104,
      "learning_rate": 0.00010193284062945236,
      "loss": 11.9584,
      "step": 11643
    },
    {
      "epoch": 2.036109289617486,
      "grad_norm": 0.6450263542911842,
      "learning_rate": 0.00010191868236910475,
      "loss": 11.9126,
      "step": 11644
    },
    {
      "epoch": 2.0362841530054645,
      "grad_norm": 0.6476329020883871,
      "learning_rate": 0.00010190452407028168,
      "loss": 11.7689,
      "step": 11645
    },
    {
      "epoch": 2.036459016393443,
      "grad_norm": 0.6457483160642173,
      "learning_rate": 0.00010189036573326705,
      "loss": 11.9995,
      "step": 11646
    },
    {
      "epoch": 2.0366338797814207,
      "grad_norm": 0.7225854649575004,
      "learning_rate": 0.00010187620735834478,
      "loss": 11.6686,
      "step": 11647
    },
    {
      "epoch": 2.036808743169399,
      "grad_norm": 0.6232802024545495,
      "learning_rate": 0.00010186204894579878,
      "loss": 11.7577,
      "step": 11648
    },
    {
      "epoch": 2.036983606557377,
      "grad_norm": 0.6334133050596652,
      "learning_rate": 0.00010184789049591299,
      "loss": 11.8561,
      "step": 11649
    },
    {
      "epoch": 2.0371584699453553,
      "grad_norm": 0.6491192175506282,
      "learning_rate": 0.00010183373200897133,
      "loss": 11.598,
      "step": 11650
    },
    {
      "epoch": 2.037333333333333,
      "grad_norm": 0.6270221082498156,
      "learning_rate": 0.00010181957348525772,
      "loss": 11.9093,
      "step": 11651
    },
    {
      "epoch": 2.0375081967213116,
      "grad_norm": 0.6486117921155304,
      "learning_rate": 0.00010180541492505604,
      "loss": 11.6909,
      "step": 11652
    },
    {
      "epoch": 2.0376830601092895,
      "grad_norm": 0.6880577837307368,
      "learning_rate": 0.0001017912563286503,
      "loss": 11.911,
      "step": 11653
    },
    {
      "epoch": 2.037857923497268,
      "grad_norm": 0.6271258983825341,
      "learning_rate": 0.00010177709769632431,
      "loss": 11.6848,
      "step": 11654
    },
    {
      "epoch": 2.0380327868852457,
      "grad_norm": 0.6926942505629254,
      "learning_rate": 0.00010176293902836209,
      "loss": 11.6304,
      "step": 11655
    },
    {
      "epoch": 2.038207650273224,
      "grad_norm": 0.6670121472210566,
      "learning_rate": 0.00010174878032504753,
      "loss": 12.0666,
      "step": 11656
    },
    {
      "epoch": 2.0383825136612024,
      "grad_norm": 0.6525090825069932,
      "learning_rate": 0.00010173462158666453,
      "loss": 12.0355,
      "step": 11657
    },
    {
      "epoch": 2.0385573770491803,
      "grad_norm": 0.6367896535809415,
      "learning_rate": 0.0001017204628134971,
      "loss": 11.933,
      "step": 11658
    },
    {
      "epoch": 2.0387322404371586,
      "grad_norm": 0.6898380109776993,
      "learning_rate": 0.00010170630400582903,
      "loss": 12.023,
      "step": 11659
    },
    {
      "epoch": 2.0389071038251365,
      "grad_norm": 0.6496637464353864,
      "learning_rate": 0.00010169214516394441,
      "loss": 11.9506,
      "step": 11660
    },
    {
      "epoch": 2.039081967213115,
      "grad_norm": 0.6504804998319148,
      "learning_rate": 0.00010167798628812701,
      "loss": 11.9122,
      "step": 11661
    },
    {
      "epoch": 2.0392568306010928,
      "grad_norm": 0.6707808932435323,
      "learning_rate": 0.00010166382737866087,
      "loss": 11.7975,
      "step": 11662
    },
    {
      "epoch": 2.039431693989071,
      "grad_norm": 0.6528215275970726,
      "learning_rate": 0.00010164966843582988,
      "loss": 11.9533,
      "step": 11663
    },
    {
      "epoch": 2.039606557377049,
      "grad_norm": 0.6132670469280209,
      "learning_rate": 0.00010163550945991795,
      "loss": 11.9044,
      "step": 11664
    },
    {
      "epoch": 2.0397814207650273,
      "grad_norm": 0.6205128357358586,
      "learning_rate": 0.00010162135045120906,
      "loss": 12.0266,
      "step": 11665
    },
    {
      "epoch": 2.0399562841530057,
      "grad_norm": 0.7204085399472966,
      "learning_rate": 0.0001016071914099871,
      "loss": 11.9002,
      "step": 11666
    },
    {
      "epoch": 2.0401311475409836,
      "grad_norm": 0.6306807840166295,
      "learning_rate": 0.00010159303233653604,
      "loss": 11.9934,
      "step": 11667
    },
    {
      "epoch": 2.040306010928962,
      "grad_norm": 0.6253818810385465,
      "learning_rate": 0.00010157887323113975,
      "loss": 11.7348,
      "step": 11668
    },
    {
      "epoch": 2.04048087431694,
      "grad_norm": 0.6091335168637515,
      "learning_rate": 0.00010156471409408224,
      "loss": 11.9561,
      "step": 11669
    },
    {
      "epoch": 2.040655737704918,
      "grad_norm": 0.6439368262949983,
      "learning_rate": 0.0001015505549256474,
      "loss": 11.9002,
      "step": 11670
    },
    {
      "epoch": 2.040830601092896,
      "grad_norm": 0.7821786747681153,
      "learning_rate": 0.00010153639572611918,
      "loss": 12.1098,
      "step": 11671
    },
    {
      "epoch": 2.0410054644808744,
      "grad_norm": 0.6794254112698237,
      "learning_rate": 0.00010152223649578148,
      "loss": 12.0626,
      "step": 11672
    },
    {
      "epoch": 2.0411803278688523,
      "grad_norm": 0.6351290419839359,
      "learning_rate": 0.00010150807723491827,
      "loss": 12.1468,
      "step": 11673
    },
    {
      "epoch": 2.0413551912568306,
      "grad_norm": 0.6586498591646788,
      "learning_rate": 0.00010149391794381352,
      "loss": 11.7834,
      "step": 11674
    },
    {
      "epoch": 2.0415300546448085,
      "grad_norm": 0.6631394241455554,
      "learning_rate": 0.00010147975862275108,
      "loss": 11.8429,
      "step": 11675
    },
    {
      "epoch": 2.041704918032787,
      "grad_norm": 0.6297847323591259,
      "learning_rate": 0.00010146559927201495,
      "loss": 12.2369,
      "step": 11676
    },
    {
      "epoch": 2.041879781420765,
      "grad_norm": 0.6351235333618233,
      "learning_rate": 0.00010145143989188908,
      "loss": 11.9661,
      "step": 11677
    },
    {
      "epoch": 2.042054644808743,
      "grad_norm": 0.6257905714371192,
      "learning_rate": 0.00010143728048265735,
      "loss": 11.7276,
      "step": 11678
    },
    {
      "epoch": 2.0422295081967214,
      "grad_norm": 0.6685641691543815,
      "learning_rate": 0.00010142312104460374,
      "loss": 11.917,
      "step": 11679
    },
    {
      "epoch": 2.0424043715846993,
      "grad_norm": 0.7043305602043919,
      "learning_rate": 0.0001014089615780122,
      "loss": 12.0474,
      "step": 11680
    },
    {
      "epoch": 2.0425792349726777,
      "grad_norm": 0.678438428565303,
      "learning_rate": 0.00010139480208316665,
      "loss": 12.0705,
      "step": 11681
    },
    {
      "epoch": 2.0427540983606556,
      "grad_norm": 0.6702228255615995,
      "learning_rate": 0.00010138064256035102,
      "loss": 12.025,
      "step": 11682
    },
    {
      "epoch": 2.042928961748634,
      "grad_norm": 0.6325152297906428,
      "learning_rate": 0.00010136648300984927,
      "loss": 12.0654,
      "step": 11683
    },
    {
      "epoch": 2.043103825136612,
      "grad_norm": 0.6822352559067013,
      "learning_rate": 0.00010135232343194536,
      "loss": 11.9817,
      "step": 11684
    },
    {
      "epoch": 2.04327868852459,
      "grad_norm": 0.6474336826249996,
      "learning_rate": 0.0001013381638269232,
      "loss": 11.6203,
      "step": 11685
    },
    {
      "epoch": 2.0434535519125685,
      "grad_norm": 0.6819696997772242,
      "learning_rate": 0.00010132400419506676,
      "loss": 11.9713,
      "step": 11686
    },
    {
      "epoch": 2.0436284153005464,
      "grad_norm": 0.6256246849262822,
      "learning_rate": 0.00010130984453665995,
      "loss": 11.9404,
      "step": 11687
    },
    {
      "epoch": 2.0438032786885247,
      "grad_norm": 0.6441620987785587,
      "learning_rate": 0.00010129568485198675,
      "loss": 12.0589,
      "step": 11688
    },
    {
      "epoch": 2.0439781420765026,
      "grad_norm": 0.6787984578724554,
      "learning_rate": 0.00010128152514133109,
      "loss": 12.0246,
      "step": 11689
    },
    {
      "epoch": 2.044153005464481,
      "grad_norm": 0.6245910736689122,
      "learning_rate": 0.0001012673654049769,
      "loss": 12.0172,
      "step": 11690
    },
    {
      "epoch": 2.044327868852459,
      "grad_norm": 0.7166676736850284,
      "learning_rate": 0.00010125320564320815,
      "loss": 11.9758,
      "step": 11691
    },
    {
      "epoch": 2.044502732240437,
      "grad_norm": 0.6503885764308952,
      "learning_rate": 0.00010123904585630879,
      "loss": 11.6671,
      "step": 11692
    },
    {
      "epoch": 2.044677595628415,
      "grad_norm": 0.6487025406796331,
      "learning_rate": 0.00010122488604456274,
      "loss": 11.7735,
      "step": 11693
    },
    {
      "epoch": 2.0448524590163935,
      "grad_norm": 0.6554844536850413,
      "learning_rate": 0.00010121072620825397,
      "loss": 12.1239,
      "step": 11694
    },
    {
      "epoch": 2.045027322404372,
      "grad_norm": 0.6802334207271977,
      "learning_rate": 0.00010119656634766643,
      "loss": 12.0106,
      "step": 11695
    },
    {
      "epoch": 2.0452021857923497,
      "grad_norm": 0.6549978256115141,
      "learning_rate": 0.00010118240646308405,
      "loss": 12.1244,
      "step": 11696
    },
    {
      "epoch": 2.045377049180328,
      "grad_norm": 0.6859081117547811,
      "learning_rate": 0.0001011682465547908,
      "loss": 12.1067,
      "step": 11697
    },
    {
      "epoch": 2.045551912568306,
      "grad_norm": 0.658278138456595,
      "learning_rate": 0.00010115408662307062,
      "loss": 12.0597,
      "step": 11698
    },
    {
      "epoch": 2.0457267759562843,
      "grad_norm": 0.709558376832944,
      "learning_rate": 0.00010113992666820747,
      "loss": 12.0818,
      "step": 11699
    },
    {
      "epoch": 2.045901639344262,
      "grad_norm": 0.6562077359697149,
      "learning_rate": 0.00010112576669048526,
      "loss": 11.7915,
      "step": 11700
    },
    {
      "epoch": 2.0460765027322405,
      "grad_norm": 0.6514945940275618,
      "learning_rate": 0.000101111606690188,
      "loss": 11.8953,
      "step": 11701
    },
    {
      "epoch": 2.0462513661202184,
      "grad_norm": 0.6425483633234802,
      "learning_rate": 0.00010109744666759957,
      "loss": 12.0046,
      "step": 11702
    },
    {
      "epoch": 2.0464262295081967,
      "grad_norm": 0.6651708767951199,
      "learning_rate": 0.000101083286623004,
      "loss": 11.9882,
      "step": 11703
    },
    {
      "epoch": 2.0466010928961746,
      "grad_norm": 0.6892327480635488,
      "learning_rate": 0.0001010691265566852,
      "loss": 12.1474,
      "step": 11704
    },
    {
      "epoch": 2.046775956284153,
      "grad_norm": 0.7237269685681817,
      "learning_rate": 0.00010105496646892713,
      "loss": 12.2248,
      "step": 11705
    },
    {
      "epoch": 2.0469508196721313,
      "grad_norm": 0.6472609954134135,
      "learning_rate": 0.00010104080636001376,
      "loss": 12.0864,
      "step": 11706
    },
    {
      "epoch": 2.0471256830601092,
      "grad_norm": 0.6638019653314355,
      "learning_rate": 0.00010102664623022899,
      "loss": 11.8841,
      "step": 11707
    },
    {
      "epoch": 2.0473005464480876,
      "grad_norm": 0.6555267964092817,
      "learning_rate": 0.00010101248607985685,
      "loss": 12.0032,
      "step": 11708
    },
    {
      "epoch": 2.0474754098360655,
      "grad_norm": 0.70959244733651,
      "learning_rate": 0.00010099832590918122,
      "loss": 12.0111,
      "step": 11709
    },
    {
      "epoch": 2.047650273224044,
      "grad_norm": 0.6500411476123733,
      "learning_rate": 0.00010098416571848613,
      "loss": 11.9707,
      "step": 11710
    },
    {
      "epoch": 2.0478251366120217,
      "grad_norm": 0.6500387326434347,
      "learning_rate": 0.00010097000550805547,
      "loss": 11.9877,
      "step": 11711
    },
    {
      "epoch": 2.048,
      "grad_norm": 0.6665399032819073,
      "learning_rate": 0.00010095584527817319,
      "loss": 12.2032,
      "step": 11712
    },
    {
      "epoch": 2.048174863387978,
      "grad_norm": 0.646705231491578,
      "learning_rate": 0.00010094168502912333,
      "loss": 11.8723,
      "step": 11713
    },
    {
      "epoch": 2.0483497267759563,
      "grad_norm": 0.6895661852417831,
      "learning_rate": 0.00010092752476118976,
      "loss": 11.9957,
      "step": 11714
    },
    {
      "epoch": 2.0485245901639346,
      "grad_norm": 0.7192444257829382,
      "learning_rate": 0.00010091336447465651,
      "loss": 11.6089,
      "step": 11715
    },
    {
      "epoch": 2.0486994535519125,
      "grad_norm": 0.6768507465537822,
      "learning_rate": 0.00010089920416980746,
      "loss": 12.0679,
      "step": 11716
    },
    {
      "epoch": 2.048874316939891,
      "grad_norm": 0.6910178648866632,
      "learning_rate": 0.00010088504384692663,
      "loss": 12.0587,
      "step": 11717
    },
    {
      "epoch": 2.0490491803278688,
      "grad_norm": 0.7661148334372163,
      "learning_rate": 0.00010087088350629794,
      "loss": 12.0083,
      "step": 11718
    },
    {
      "epoch": 2.049224043715847,
      "grad_norm": 0.6141072474627466,
      "learning_rate": 0.00010085672314820538,
      "loss": 11.8736,
      "step": 11719
    },
    {
      "epoch": 2.049398907103825,
      "grad_norm": 0.828276954386087,
      "learning_rate": 0.00010084256277293286,
      "loss": 12.009,
      "step": 11720
    },
    {
      "epoch": 2.0495737704918033,
      "grad_norm": 0.6890921287824735,
      "learning_rate": 0.00010082840238076436,
      "loss": 11.7538,
      "step": 11721
    },
    {
      "epoch": 2.0497486338797812,
      "grad_norm": 0.7709649587000551,
      "learning_rate": 0.0001008142419719839,
      "loss": 12.0145,
      "step": 11722
    },
    {
      "epoch": 2.0499234972677596,
      "grad_norm": 0.6869586154940835,
      "learning_rate": 0.00010080008154687537,
      "loss": 11.5901,
      "step": 11723
    },
    {
      "epoch": 2.050098360655738,
      "grad_norm": 0.6978046333145901,
      "learning_rate": 0.00010078592110572273,
      "loss": 11.9156,
      "step": 11724
    },
    {
      "epoch": 2.050273224043716,
      "grad_norm": 0.7556260131951416,
      "learning_rate": 0.00010077176064880997,
      "loss": 11.9863,
      "step": 11725
    },
    {
      "epoch": 2.050448087431694,
      "grad_norm": 0.6468870725174743,
      "learning_rate": 0.00010075760017642104,
      "loss": 12.0162,
      "step": 11726
    },
    {
      "epoch": 2.050622950819672,
      "grad_norm": 0.6777135816890337,
      "learning_rate": 0.00010074343968883988,
      "loss": 12.0413,
      "step": 11727
    },
    {
      "epoch": 2.0507978142076504,
      "grad_norm": 0.817176817853863,
      "learning_rate": 0.00010072927918635049,
      "loss": 12.0291,
      "step": 11728
    },
    {
      "epoch": 2.0509726775956283,
      "grad_norm": 0.6308016594026971,
      "learning_rate": 0.00010071511866923682,
      "loss": 11.986,
      "step": 11729
    },
    {
      "epoch": 2.0511475409836066,
      "grad_norm": 0.7854005362206643,
      "learning_rate": 0.00010070095813778281,
      "loss": 11.7827,
      "step": 11730
    },
    {
      "epoch": 2.0513224043715845,
      "grad_norm": 0.6798348187274722,
      "learning_rate": 0.00010068679759227244,
      "loss": 11.7483,
      "step": 11731
    },
    {
      "epoch": 2.051497267759563,
      "grad_norm": 0.6884658462514063,
      "learning_rate": 0.00010067263703298969,
      "loss": 11.9847,
      "step": 11732
    },
    {
      "epoch": 2.0516721311475408,
      "grad_norm": 0.752132351460563,
      "learning_rate": 0.00010065847646021849,
      "loss": 12.0369,
      "step": 11733
    },
    {
      "epoch": 2.051846994535519,
      "grad_norm": 0.6395562300300689,
      "learning_rate": 0.00010064431587424281,
      "loss": 11.8481,
      "step": 11734
    },
    {
      "epoch": 2.0520218579234974,
      "grad_norm": 0.7558066550303053,
      "learning_rate": 0.00010063015527534661,
      "loss": 11.7654,
      "step": 11735
    },
    {
      "epoch": 2.0521967213114753,
      "grad_norm": 0.750670563762473,
      "learning_rate": 0.00010061599466381389,
      "loss": 12.1456,
      "step": 11736
    },
    {
      "epoch": 2.0523715846994537,
      "grad_norm": 0.78693857235553,
      "learning_rate": 0.00010060183403992856,
      "loss": 11.9708,
      "step": 11737
    },
    {
      "epoch": 2.0525464480874316,
      "grad_norm": 0.779015931991475,
      "learning_rate": 0.00010058767340397461,
      "loss": 12.0616,
      "step": 11738
    },
    {
      "epoch": 2.05272131147541,
      "grad_norm": 0.6572008013516831,
      "learning_rate": 0.000100573512756236,
      "loss": 11.722,
      "step": 11739
    },
    {
      "epoch": 2.052896174863388,
      "grad_norm": 0.7387907402420373,
      "learning_rate": 0.00010055935209699671,
      "loss": 12.0476,
      "step": 11740
    },
    {
      "epoch": 2.053071038251366,
      "grad_norm": 0.787223926896884,
      "learning_rate": 0.00010054519142654071,
      "loss": 12.1286,
      "step": 11741
    },
    {
      "epoch": 2.053245901639344,
      "grad_norm": 0.653506543903058,
      "learning_rate": 0.00010053103074515191,
      "loss": 11.8686,
      "step": 11742
    },
    {
      "epoch": 2.0534207650273224,
      "grad_norm": 0.7050767085411505,
      "learning_rate": 0.00010051687005311434,
      "loss": 12.0729,
      "step": 11743
    },
    {
      "epoch": 2.0535956284153007,
      "grad_norm": 0.6660095730590656,
      "learning_rate": 0.00010050270935071193,
      "loss": 11.8117,
      "step": 11744
    },
    {
      "epoch": 2.0537704918032786,
      "grad_norm": 0.7075968301333405,
      "learning_rate": 0.00010048854863822866,
      "loss": 11.9059,
      "step": 11745
    },
    {
      "epoch": 2.053945355191257,
      "grad_norm": 0.6684024190095615,
      "learning_rate": 0.00010047438791594848,
      "loss": 11.9267,
      "step": 11746
    },
    {
      "epoch": 2.054120218579235,
      "grad_norm": 0.6299277287936548,
      "learning_rate": 0.0001004602271841554,
      "loss": 12.0563,
      "step": 11747
    },
    {
      "epoch": 2.054295081967213,
      "grad_norm": 0.7485027074611795,
      "learning_rate": 0.0001004460664431333,
      "loss": 11.9954,
      "step": 11748
    },
    {
      "epoch": 2.054469945355191,
      "grad_norm": 0.6883162466768156,
      "learning_rate": 0.00010043190569316624,
      "loss": 12.1705,
      "step": 11749
    },
    {
      "epoch": 2.0546448087431695,
      "grad_norm": 0.6950930473055532,
      "learning_rate": 0.00010041774493453811,
      "loss": 11.9052,
      "step": 11750
    },
    {
      "epoch": 2.0548196721311474,
      "grad_norm": 0.7839171603562849,
      "learning_rate": 0.00010040358416753293,
      "loss": 11.6058,
      "step": 11751
    },
    {
      "epoch": 2.0549945355191257,
      "grad_norm": 0.6989140560957063,
      "learning_rate": 0.00010038942339243464,
      "loss": 11.7587,
      "step": 11752
    },
    {
      "epoch": 2.055169398907104,
      "grad_norm": 0.7345148155050125,
      "learning_rate": 0.00010037526260952721,
      "loss": 11.835,
      "step": 11753
    },
    {
      "epoch": 2.055344262295082,
      "grad_norm": 0.6363060506808208,
      "learning_rate": 0.00010036110181909466,
      "loss": 11.8647,
      "step": 11754
    },
    {
      "epoch": 2.0555191256830603,
      "grad_norm": 0.6427347017155213,
      "learning_rate": 0.00010034694102142084,
      "loss": 12.0514,
      "step": 11755
    },
    {
      "epoch": 2.055693989071038,
      "grad_norm": 0.7001384184891001,
      "learning_rate": 0.00010033278021678984,
      "loss": 11.8946,
      "step": 11756
    },
    {
      "epoch": 2.0558688524590165,
      "grad_norm": 0.719278355467558,
      "learning_rate": 0.00010031861940548555,
      "loss": 11.9474,
      "step": 11757
    },
    {
      "epoch": 2.0560437158469944,
      "grad_norm": 0.6690238195546281,
      "learning_rate": 0.00010030445858779199,
      "loss": 11.7264,
      "step": 11758
    },
    {
      "epoch": 2.0562185792349728,
      "grad_norm": 0.7712305784992869,
      "learning_rate": 0.00010029029776399306,
      "loss": 11.98,
      "step": 11759
    },
    {
      "epoch": 2.0563934426229507,
      "grad_norm": 0.6430015576203322,
      "learning_rate": 0.00010027613693437276,
      "loss": 12.0269,
      "step": 11760
    },
    {
      "epoch": 2.056568306010929,
      "grad_norm": 0.7133377616625319,
      "learning_rate": 0.00010026197609921512,
      "loss": 11.7661,
      "step": 11761
    },
    {
      "epoch": 2.056743169398907,
      "grad_norm": 0.7300675270243008,
      "learning_rate": 0.000100247815258804,
      "loss": 11.95,
      "step": 11762
    },
    {
      "epoch": 2.0569180327868852,
      "grad_norm": 0.6906025393643185,
      "learning_rate": 0.00010023365441342349,
      "loss": 11.9573,
      "step": 11763
    },
    {
      "epoch": 2.0570928961748636,
      "grad_norm": 0.7300070743322351,
      "learning_rate": 0.00010021949356335743,
      "loss": 12.1154,
      "step": 11764
    },
    {
      "epoch": 2.0572677595628415,
      "grad_norm": 0.6513070211769444,
      "learning_rate": 0.0001002053327088899,
      "loss": 11.9827,
      "step": 11765
    },
    {
      "epoch": 2.05744262295082,
      "grad_norm": 0.6949569864538356,
      "learning_rate": 0.00010019117185030478,
      "loss": 12.0147,
      "step": 11766
    },
    {
      "epoch": 2.0576174863387977,
      "grad_norm": 0.6622829350430084,
      "learning_rate": 0.0001001770109878861,
      "loss": 11.9679,
      "step": 11767
    },
    {
      "epoch": 2.057792349726776,
      "grad_norm": 0.6743597362544483,
      "learning_rate": 0.0001001628501219178,
      "loss": 11.9492,
      "step": 11768
    },
    {
      "epoch": 2.057967213114754,
      "grad_norm": 0.650680000871111,
      "learning_rate": 0.00010014868925268381,
      "loss": 12.0718,
      "step": 11769
    },
    {
      "epoch": 2.0581420765027323,
      "grad_norm": 0.7318551140023672,
      "learning_rate": 0.00010013452838046822,
      "loss": 12.1162,
      "step": 11770
    },
    {
      "epoch": 2.05831693989071,
      "grad_norm": 0.6577471303480303,
      "learning_rate": 0.00010012036750555487,
      "loss": 11.8524,
      "step": 11771
    },
    {
      "epoch": 2.0584918032786885,
      "grad_norm": 0.6682116655570207,
      "learning_rate": 0.0001001062066282278,
      "loss": 11.6732,
      "step": 11772
    },
    {
      "epoch": 2.058666666666667,
      "grad_norm": 0.7002874313948508,
      "learning_rate": 0.00010009204574877099,
      "loss": 12.1354,
      "step": 11773
    },
    {
      "epoch": 2.0588415300546448,
      "grad_norm": 0.6819206090900376,
      "learning_rate": 0.00010007788486746832,
      "loss": 11.763,
      "step": 11774
    },
    {
      "epoch": 2.059016393442623,
      "grad_norm": 0.6884189385838588,
      "learning_rate": 0.00010006372398460387,
      "loss": 12.029,
      "step": 11775
    },
    {
      "epoch": 2.059191256830601,
      "grad_norm": 0.6710546158374754,
      "learning_rate": 0.00010004956310046151,
      "loss": 11.8853,
      "step": 11776
    },
    {
      "epoch": 2.0593661202185793,
      "grad_norm": 0.6696162811022913,
      "learning_rate": 0.0001000354022153253,
      "loss": 11.9509,
      "step": 11777
    },
    {
      "epoch": 2.0595409836065572,
      "grad_norm": 0.6313743402906,
      "learning_rate": 0.00010002124132947917,
      "loss": 11.977,
      "step": 11778
    },
    {
      "epoch": 2.0597158469945356,
      "grad_norm": 0.6942517646003039,
      "learning_rate": 0.00010000708044320706,
      "loss": 12.0192,
      "step": 11779
    },
    {
      "epoch": 2.0598907103825135,
      "grad_norm": 0.7241644271249722,
      "learning_rate": 9.999291955679296e-05,
      "loss": 12.1642,
      "step": 11780
    },
    {
      "epoch": 2.060065573770492,
      "grad_norm": 0.6644624229664372,
      "learning_rate": 9.99787586705209e-05,
      "loss": 11.8525,
      "step": 11781
    },
    {
      "epoch": 2.06024043715847,
      "grad_norm": 0.6493118683551957,
      "learning_rate": 9.99645977846747e-05,
      "loss": 11.9851,
      "step": 11782
    },
    {
      "epoch": 2.060415300546448,
      "grad_norm": 0.6895227903871902,
      "learning_rate": 9.99504368995385e-05,
      "loss": 11.8966,
      "step": 11783
    },
    {
      "epoch": 2.0605901639344264,
      "grad_norm": 0.7315929161992816,
      "learning_rate": 9.993627601539617e-05,
      "loss": 12.1443,
      "step": 11784
    },
    {
      "epoch": 2.0607650273224043,
      "grad_norm": 0.5951220893395717,
      "learning_rate": 9.992211513253169e-05,
      "loss": 11.9441,
      "step": 11785
    },
    {
      "epoch": 2.0609398907103826,
      "grad_norm": 0.6431633364542342,
      "learning_rate": 9.990795425122908e-05,
      "loss": 11.7692,
      "step": 11786
    },
    {
      "epoch": 2.0611147540983605,
      "grad_norm": 0.6274587638844661,
      "learning_rate": 9.98937933717722e-05,
      "loss": 12.018,
      "step": 11787
    },
    {
      "epoch": 2.061289617486339,
      "grad_norm": 0.6323964900210488,
      "learning_rate": 9.987963249444513e-05,
      "loss": 11.9744,
      "step": 11788
    },
    {
      "epoch": 2.0614644808743168,
      "grad_norm": 0.5719997372685361,
      "learning_rate": 9.986547161953182e-05,
      "loss": 11.996,
      "step": 11789
    },
    {
      "epoch": 2.061639344262295,
      "grad_norm": 0.642903312196384,
      "learning_rate": 9.98513107473162e-05,
      "loss": 11.8548,
      "step": 11790
    },
    {
      "epoch": 2.061814207650273,
      "grad_norm": 0.6418403051722376,
      "learning_rate": 9.983714987808226e-05,
      "loss": 12.0426,
      "step": 11791
    },
    {
      "epoch": 2.0619890710382514,
      "grad_norm": 0.5766032618957575,
      "learning_rate": 9.982298901211393e-05,
      "loss": 11.6337,
      "step": 11792
    },
    {
      "epoch": 2.0621639344262297,
      "grad_norm": 0.6818116660810354,
      "learning_rate": 9.980882814969524e-05,
      "loss": 11.7741,
      "step": 11793
    },
    {
      "epoch": 2.0623387978142076,
      "grad_norm": 0.7094855945579931,
      "learning_rate": 9.979466729111013e-05,
      "loss": 11.7663,
      "step": 11794
    },
    {
      "epoch": 2.062513661202186,
      "grad_norm": 0.6639530030675432,
      "learning_rate": 9.978050643664259e-05,
      "loss": 11.8346,
      "step": 11795
    },
    {
      "epoch": 2.062688524590164,
      "grad_norm": 0.6372546822881247,
      "learning_rate": 9.976634558657656e-05,
      "loss": 11.9038,
      "step": 11796
    },
    {
      "epoch": 2.062863387978142,
      "grad_norm": 0.5742656181288611,
      "learning_rate": 9.975218474119599e-05,
      "loss": 11.7016,
      "step": 11797
    },
    {
      "epoch": 2.06303825136612,
      "grad_norm": 0.7066200324956117,
      "learning_rate": 9.973802390078489e-05,
      "loss": 12.0958,
      "step": 11798
    },
    {
      "epoch": 2.0632131147540984,
      "grad_norm": 0.6385784198528113,
      "learning_rate": 9.972386306562726e-05,
      "loss": 12.0131,
      "step": 11799
    },
    {
      "epoch": 2.0633879781420763,
      "grad_norm": 0.6807871836839549,
      "learning_rate": 9.970970223600698e-05,
      "loss": 11.8668,
      "step": 11800
    },
    {
      "epoch": 2.0635628415300546,
      "grad_norm": 0.66607751981615,
      "learning_rate": 9.969554141220808e-05,
      "loss": 11.9316,
      "step": 11801
    },
    {
      "epoch": 2.063737704918033,
      "grad_norm": 0.6781045627856621,
      "learning_rate": 9.968138059451446e-05,
      "loss": 11.8822,
      "step": 11802
    },
    {
      "epoch": 2.063912568306011,
      "grad_norm": 0.6460912133401866,
      "learning_rate": 9.966721978321018e-05,
      "loss": 11.951,
      "step": 11803
    },
    {
      "epoch": 2.0640874316939892,
      "grad_norm": 0.7214874689546624,
      "learning_rate": 9.965305897857918e-05,
      "loss": 11.8827,
      "step": 11804
    },
    {
      "epoch": 2.064262295081967,
      "grad_norm": 0.601302222185756,
      "learning_rate": 9.96388981809054e-05,
      "loss": 11.8003,
      "step": 11805
    },
    {
      "epoch": 2.0644371584699455,
      "grad_norm": 0.6944861625957509,
      "learning_rate": 9.962473739047277e-05,
      "loss": 12.0063,
      "step": 11806
    },
    {
      "epoch": 2.0646120218579234,
      "grad_norm": 0.676423097967237,
      "learning_rate": 9.961057660756537e-05,
      "loss": 11.8354,
      "step": 11807
    },
    {
      "epoch": 2.0647868852459017,
      "grad_norm": 0.6975226873618433,
      "learning_rate": 9.95964158324671e-05,
      "loss": 11.9874,
      "step": 11808
    },
    {
      "epoch": 2.0649617486338796,
      "grad_norm": 0.7270642390341325,
      "learning_rate": 9.958225506546191e-05,
      "loss": 11.8648,
      "step": 11809
    },
    {
      "epoch": 2.065136612021858,
      "grad_norm": 0.6195182666022191,
      "learning_rate": 9.95680943068338e-05,
      "loss": 11.9005,
      "step": 11810
    },
    {
      "epoch": 2.0653114754098363,
      "grad_norm": 0.6749715875038208,
      "learning_rate": 9.955393355686671e-05,
      "loss": 11.8856,
      "step": 11811
    },
    {
      "epoch": 2.065486338797814,
      "grad_norm": 0.6524164471147332,
      "learning_rate": 9.953977281584463e-05,
      "loss": 12.0515,
      "step": 11812
    },
    {
      "epoch": 2.0656612021857925,
      "grad_norm": 0.6888635017137763,
      "learning_rate": 9.952561208405153e-05,
      "loss": 12.1866,
      "step": 11813
    },
    {
      "epoch": 2.0658360655737704,
      "grad_norm": 0.6654856712161128,
      "learning_rate": 9.951145136177137e-05,
      "loss": 11.7845,
      "step": 11814
    },
    {
      "epoch": 2.0660109289617488,
      "grad_norm": 0.6810320289467565,
      "learning_rate": 9.949729064928808e-05,
      "loss": 12.1043,
      "step": 11815
    },
    {
      "epoch": 2.0661857923497267,
      "grad_norm": 0.6230313631584221,
      "learning_rate": 9.948312994688566e-05,
      "loss": 11.9901,
      "step": 11816
    },
    {
      "epoch": 2.066360655737705,
      "grad_norm": 0.6140443325146278,
      "learning_rate": 9.94689692548481e-05,
      "loss": 11.5823,
      "step": 11817
    },
    {
      "epoch": 2.066535519125683,
      "grad_norm": 0.6240613961456787,
      "learning_rate": 9.945480857345933e-05,
      "loss": 12.1085,
      "step": 11818
    },
    {
      "epoch": 2.0667103825136612,
      "grad_norm": 0.6194385983161901,
      "learning_rate": 9.94406479030033e-05,
      "loss": 11.9635,
      "step": 11819
    },
    {
      "epoch": 2.066885245901639,
      "grad_norm": 0.7032488858002789,
      "learning_rate": 9.942648724376403e-05,
      "loss": 11.8889,
      "step": 11820
    },
    {
      "epoch": 2.0670601092896175,
      "grad_norm": 0.6251039804481292,
      "learning_rate": 9.94123265960254e-05,
      "loss": 11.9096,
      "step": 11821
    },
    {
      "epoch": 2.067234972677596,
      "grad_norm": 0.7283616995451214,
      "learning_rate": 9.939816596007146e-05,
      "loss": 11.9024,
      "step": 11822
    },
    {
      "epoch": 2.0674098360655737,
      "grad_norm": 0.6806499810693499,
      "learning_rate": 9.938400533618615e-05,
      "loss": 11.9223,
      "step": 11823
    },
    {
      "epoch": 2.067584699453552,
      "grad_norm": 0.6608497343714834,
      "learning_rate": 9.93698447246534e-05,
      "loss": 11.9118,
      "step": 11824
    },
    {
      "epoch": 2.06775956284153,
      "grad_norm": 0.7010602714359856,
      "learning_rate": 9.935568412575724e-05,
      "loss": 11.6195,
      "step": 11825
    },
    {
      "epoch": 2.0679344262295083,
      "grad_norm": 0.6642741431385625,
      "learning_rate": 9.934152353978152e-05,
      "loss": 12.0307,
      "step": 11826
    },
    {
      "epoch": 2.068109289617486,
      "grad_norm": 0.7102615525186985,
      "learning_rate": 9.932736296701034e-05,
      "loss": 12.0904,
      "step": 11827
    },
    {
      "epoch": 2.0682841530054645,
      "grad_norm": 0.7163371364979111,
      "learning_rate": 9.931320240772757e-05,
      "loss": 11.9558,
      "step": 11828
    },
    {
      "epoch": 2.0684590163934424,
      "grad_norm": 0.6595478579448771,
      "learning_rate": 9.929904186221722e-05,
      "loss": 11.8549,
      "step": 11829
    },
    {
      "epoch": 2.0686338797814208,
      "grad_norm": 0.6419955890072363,
      "learning_rate": 9.928488133076318e-05,
      "loss": 11.9432,
      "step": 11830
    },
    {
      "epoch": 2.068808743169399,
      "grad_norm": 0.6476614684506564,
      "learning_rate": 9.927072081364952e-05,
      "loss": 12.0688,
      "step": 11831
    },
    {
      "epoch": 2.068983606557377,
      "grad_norm": 0.6950669283915917,
      "learning_rate": 9.925656031116013e-05,
      "loss": 11.8134,
      "step": 11832
    },
    {
      "epoch": 2.0691584699453553,
      "grad_norm": 0.6559980324103717,
      "learning_rate": 9.924239982357899e-05,
      "loss": 11.9704,
      "step": 11833
    },
    {
      "epoch": 2.0693333333333332,
      "grad_norm": 0.7085236772664875,
      "learning_rate": 9.922823935119008e-05,
      "loss": 12.0499,
      "step": 11834
    },
    {
      "epoch": 2.0695081967213116,
      "grad_norm": 0.7244638841697434,
      "learning_rate": 9.921407889427728e-05,
      "loss": 12.0568,
      "step": 11835
    },
    {
      "epoch": 2.0696830601092895,
      "grad_norm": 0.6922784524599171,
      "learning_rate": 9.919991845312467e-05,
      "loss": 11.9995,
      "step": 11836
    },
    {
      "epoch": 2.069857923497268,
      "grad_norm": 0.6355949639000781,
      "learning_rate": 9.918575802801612e-05,
      "loss": 11.8881,
      "step": 11837
    },
    {
      "epoch": 2.0700327868852457,
      "grad_norm": 0.6996074221273911,
      "learning_rate": 9.917159761923566e-05,
      "loss": 11.7485,
      "step": 11838
    },
    {
      "epoch": 2.070207650273224,
      "grad_norm": 0.6533114839122962,
      "learning_rate": 9.915743722706719e-05,
      "loss": 11.9914,
      "step": 11839
    },
    {
      "epoch": 2.070382513661202,
      "grad_norm": 0.7537832532452371,
      "learning_rate": 9.914327685179466e-05,
      "loss": 11.9248,
      "step": 11840
    },
    {
      "epoch": 2.0705573770491803,
      "grad_norm": 0.7241509024535933,
      "learning_rate": 9.912911649370207e-05,
      "loss": 11.6321,
      "step": 11841
    },
    {
      "epoch": 2.0707322404371586,
      "grad_norm": 0.7020463168683047,
      "learning_rate": 9.91149561530734e-05,
      "loss": 11.6083,
      "step": 11842
    },
    {
      "epoch": 2.0709071038251365,
      "grad_norm": 0.6835430024432618,
      "learning_rate": 9.910079583019257e-05,
      "loss": 12.0108,
      "step": 11843
    },
    {
      "epoch": 2.071081967213115,
      "grad_norm": 0.667082844072246,
      "learning_rate": 9.908663552534353e-05,
      "loss": 12.068,
      "step": 11844
    },
    {
      "epoch": 2.071256830601093,
      "grad_norm": 0.6285653459826338,
      "learning_rate": 9.907247523881024e-05,
      "loss": 11.9382,
      "step": 11845
    },
    {
      "epoch": 2.071431693989071,
      "grad_norm": 0.6329872065023217,
      "learning_rate": 9.905831497087668e-05,
      "loss": 12.0672,
      "step": 11846
    },
    {
      "epoch": 2.071606557377049,
      "grad_norm": 0.6207415110670071,
      "learning_rate": 9.904415472182682e-05,
      "loss": 12.1458,
      "step": 11847
    },
    {
      "epoch": 2.0717814207650274,
      "grad_norm": 0.642095257104398,
      "learning_rate": 9.902999449194458e-05,
      "loss": 11.8347,
      "step": 11848
    },
    {
      "epoch": 2.0719562841530053,
      "grad_norm": 0.6638795076899987,
      "learning_rate": 9.901583428151392e-05,
      "loss": 12.1039,
      "step": 11849
    },
    {
      "epoch": 2.0721311475409836,
      "grad_norm": 0.6160700828714045,
      "learning_rate": 9.900167409081879e-05,
      "loss": 11.9695,
      "step": 11850
    },
    {
      "epoch": 2.072306010928962,
      "grad_norm": 0.6576952130882938,
      "learning_rate": 9.898751392014317e-05,
      "loss": 11.7327,
      "step": 11851
    },
    {
      "epoch": 2.07248087431694,
      "grad_norm": 0.6491825667074613,
      "learning_rate": 9.897335376977102e-05,
      "loss": 12.0078,
      "step": 11852
    },
    {
      "epoch": 2.072655737704918,
      "grad_norm": 0.601404772982478,
      "learning_rate": 9.895919363998627e-05,
      "loss": 11.9647,
      "step": 11853
    },
    {
      "epoch": 2.072830601092896,
      "grad_norm": 0.6667322476657067,
      "learning_rate": 9.894503353107288e-05,
      "loss": 11.8949,
      "step": 11854
    },
    {
      "epoch": 2.0730054644808744,
      "grad_norm": 0.650336083643857,
      "learning_rate": 9.89308734433148e-05,
      "loss": 11.7885,
      "step": 11855
    },
    {
      "epoch": 2.0731803278688523,
      "grad_norm": 0.6708067562246977,
      "learning_rate": 9.891671337699602e-05,
      "loss": 12.0364,
      "step": 11856
    },
    {
      "epoch": 2.0733551912568307,
      "grad_norm": 0.6507352487966507,
      "learning_rate": 9.890255333240044e-05,
      "loss": 11.829,
      "step": 11857
    },
    {
      "epoch": 2.0735300546448086,
      "grad_norm": 0.6479484315469826,
      "learning_rate": 9.888839330981204e-05,
      "loss": 12.1714,
      "step": 11858
    },
    {
      "epoch": 2.073704918032787,
      "grad_norm": 0.6766669227685159,
      "learning_rate": 9.887423330951475e-05,
      "loss": 11.9258,
      "step": 11859
    },
    {
      "epoch": 2.0738797814207652,
      "grad_norm": 0.6680985381942787,
      "learning_rate": 9.886007333179255e-05,
      "loss": 11.9021,
      "step": 11860
    },
    {
      "epoch": 2.074054644808743,
      "grad_norm": 0.6513403732109331,
      "learning_rate": 9.884591337692941e-05,
      "loss": 11.8368,
      "step": 11861
    },
    {
      "epoch": 2.0742295081967215,
      "grad_norm": 0.6639978064577294,
      "learning_rate": 9.883175344520922e-05,
      "loss": 11.701,
      "step": 11862
    },
    {
      "epoch": 2.0744043715846994,
      "grad_norm": 0.7388457493370812,
      "learning_rate": 9.881759353691596e-05,
      "loss": 12.0945,
      "step": 11863
    },
    {
      "epoch": 2.0745792349726777,
      "grad_norm": 0.6558599450308127,
      "learning_rate": 9.880343365233356e-05,
      "loss": 11.9797,
      "step": 11864
    },
    {
      "epoch": 2.0747540983606556,
      "grad_norm": 0.6434364124823571,
      "learning_rate": 9.878927379174605e-05,
      "loss": 12.0014,
      "step": 11865
    },
    {
      "epoch": 2.074928961748634,
      "grad_norm": 0.7095567555806754,
      "learning_rate": 9.877511395543728e-05,
      "loss": 12.0152,
      "step": 11866
    },
    {
      "epoch": 2.075103825136612,
      "grad_norm": 0.6825690461142142,
      "learning_rate": 9.876095414369124e-05,
      "loss": 11.922,
      "step": 11867
    },
    {
      "epoch": 2.07527868852459,
      "grad_norm": 0.8262697274803784,
      "learning_rate": 9.87467943567919e-05,
      "loss": 12.011,
      "step": 11868
    },
    {
      "epoch": 2.0754535519125685,
      "grad_norm": 0.7218182093720793,
      "learning_rate": 9.87326345950231e-05,
      "loss": 11.8334,
      "step": 11869
    },
    {
      "epoch": 2.0756284153005464,
      "grad_norm": 0.5875669918087266,
      "learning_rate": 9.871847485866895e-05,
      "loss": 12.0741,
      "step": 11870
    },
    {
      "epoch": 2.0758032786885248,
      "grad_norm": 0.7229579588700894,
      "learning_rate": 9.870431514801327e-05,
      "loss": 12.0551,
      "step": 11871
    },
    {
      "epoch": 2.0759781420765027,
      "grad_norm": 0.6580948488741636,
      "learning_rate": 9.869015546334007e-05,
      "loss": 11.9696,
      "step": 11872
    },
    {
      "epoch": 2.076153005464481,
      "grad_norm": 0.6748764159371633,
      "learning_rate": 9.867599580493329e-05,
      "loss": 11.7426,
      "step": 11873
    },
    {
      "epoch": 2.076327868852459,
      "grad_norm": 0.6591137787083832,
      "learning_rate": 9.866183617307682e-05,
      "loss": 12.0388,
      "step": 11874
    },
    {
      "epoch": 2.0765027322404372,
      "grad_norm": 0.6705215228353515,
      "learning_rate": 9.864767656805467e-05,
      "loss": 12.0223,
      "step": 11875
    },
    {
      "epoch": 2.076677595628415,
      "grad_norm": 0.6954224074707429,
      "learning_rate": 9.863351699015074e-05,
      "loss": 12.1521,
      "step": 11876
    },
    {
      "epoch": 2.0768524590163935,
      "grad_norm": 0.669762401328407,
      "learning_rate": 9.861935743964902e-05,
      "loss": 11.978,
      "step": 11877
    },
    {
      "epoch": 2.0770273224043714,
      "grad_norm": 0.6917081334954001,
      "learning_rate": 9.860519791683336e-05,
      "loss": 12.0451,
      "step": 11878
    },
    {
      "epoch": 2.0772021857923497,
      "grad_norm": 0.6876132331224007,
      "learning_rate": 9.859103842198782e-05,
      "loss": 11.9632,
      "step": 11879
    },
    {
      "epoch": 2.077377049180328,
      "grad_norm": 0.6980678732868595,
      "learning_rate": 9.857687895539627e-05,
      "loss": 12.0317,
      "step": 11880
    },
    {
      "epoch": 2.077551912568306,
      "grad_norm": 0.6795444382521534,
      "learning_rate": 9.856271951734268e-05,
      "loss": 12.0114,
      "step": 11881
    },
    {
      "epoch": 2.0777267759562843,
      "grad_norm": 0.6998923217123144,
      "learning_rate": 9.854856010811098e-05,
      "loss": 11.966,
      "step": 11882
    },
    {
      "epoch": 2.077901639344262,
      "grad_norm": 0.6839823414924857,
      "learning_rate": 9.853440072798507e-05,
      "loss": 11.9721,
      "step": 11883
    },
    {
      "epoch": 2.0780765027322405,
      "grad_norm": 0.9000636970157018,
      "learning_rate": 9.852024137724893e-05,
      "loss": 12.1657,
      "step": 11884
    },
    {
      "epoch": 2.0782513661202184,
      "grad_norm": 0.7132938612239694,
      "learning_rate": 9.850608205618652e-05,
      "loss": 11.8856,
      "step": 11885
    },
    {
      "epoch": 2.0784262295081968,
      "grad_norm": 0.7439065914162505,
      "learning_rate": 9.849192276508177e-05,
      "loss": 11.767,
      "step": 11886
    },
    {
      "epoch": 2.0786010928961747,
      "grad_norm": 0.6429830689750428,
      "learning_rate": 9.847776350421857e-05,
      "loss": 11.8573,
      "step": 11887
    },
    {
      "epoch": 2.078775956284153,
      "grad_norm": 0.65633789481898,
      "learning_rate": 9.846360427388086e-05,
      "loss": 12.0768,
      "step": 11888
    },
    {
      "epoch": 2.0789508196721314,
      "grad_norm": 0.7077062112588133,
      "learning_rate": 9.844944507435262e-05,
      "loss": 11.8324,
      "step": 11889
    },
    {
      "epoch": 2.0791256830601093,
      "grad_norm": 0.7101148206309726,
      "learning_rate": 9.843528590591777e-05,
      "loss": 11.8156,
      "step": 11890
    },
    {
      "epoch": 2.0793005464480876,
      "grad_norm": 0.6803983733732748,
      "learning_rate": 9.842112676886027e-05,
      "loss": 11.876,
      "step": 11891
    },
    {
      "epoch": 2.0794754098360655,
      "grad_norm": 0.7774455527589912,
      "learning_rate": 9.840696766346401e-05,
      "loss": 12.0085,
      "step": 11892
    },
    {
      "epoch": 2.079650273224044,
      "grad_norm": 0.6356176376435128,
      "learning_rate": 9.839280859001291e-05,
      "loss": 12.0,
      "step": 11893
    },
    {
      "epoch": 2.0798251366120217,
      "grad_norm": 0.6793539913125686,
      "learning_rate": 9.837864954879096e-05,
      "loss": 11.7355,
      "step": 11894
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.6856643998479927,
      "learning_rate": 9.836449054008207e-05,
      "loss": 11.974,
      "step": 11895
    },
    {
      "epoch": 2.080174863387978,
      "grad_norm": 0.7059605683074888,
      "learning_rate": 9.835033156417017e-05,
      "loss": 12.0849,
      "step": 11896
    },
    {
      "epoch": 2.0803497267759563,
      "grad_norm": 0.7271616726700436,
      "learning_rate": 9.833617262133917e-05,
      "loss": 11.8746,
      "step": 11897
    },
    {
      "epoch": 2.080524590163934,
      "grad_norm": 0.6763031892321218,
      "learning_rate": 9.8322013711873e-05,
      "loss": 11.741,
      "step": 11898
    },
    {
      "epoch": 2.0806994535519125,
      "grad_norm": 0.6586765459902492,
      "learning_rate": 9.830785483605563e-05,
      "loss": 11.9397,
      "step": 11899
    },
    {
      "epoch": 2.080874316939891,
      "grad_norm": 0.6691882939107162,
      "learning_rate": 9.829369599417098e-05,
      "loss": 11.7376,
      "step": 11900
    },
    {
      "epoch": 2.081049180327869,
      "grad_norm": 0.6657081964819319,
      "learning_rate": 9.827953718650295e-05,
      "loss": 11.7818,
      "step": 11901
    },
    {
      "epoch": 2.081224043715847,
      "grad_norm": 0.7294328691317647,
      "learning_rate": 9.826537841333545e-05,
      "loss": 11.883,
      "step": 11902
    },
    {
      "epoch": 2.081398907103825,
      "grad_norm": 0.6807139776814499,
      "learning_rate": 9.825121967495248e-05,
      "loss": 11.7052,
      "step": 11903
    },
    {
      "epoch": 2.0815737704918034,
      "grad_norm": 0.7356955226144244,
      "learning_rate": 9.823706097163792e-05,
      "loss": 12.1297,
      "step": 11904
    },
    {
      "epoch": 2.0817486338797813,
      "grad_norm": 0.6457104782004466,
      "learning_rate": 9.82229023036757e-05,
      "loss": 12.0731,
      "step": 11905
    },
    {
      "epoch": 2.0819234972677596,
      "grad_norm": 0.6277532702821373,
      "learning_rate": 9.820874367134974e-05,
      "loss": 12.0075,
      "step": 11906
    },
    {
      "epoch": 2.0820983606557375,
      "grad_norm": 0.6955652675093186,
      "learning_rate": 9.819458507494394e-05,
      "loss": 11.9417,
      "step": 11907
    },
    {
      "epoch": 2.082273224043716,
      "grad_norm": 0.6555016860573393,
      "learning_rate": 9.818042651474229e-05,
      "loss": 12.0065,
      "step": 11908
    },
    {
      "epoch": 2.082448087431694,
      "grad_norm": 0.6649227059173518,
      "learning_rate": 9.81662679910287e-05,
      "loss": 12.0009,
      "step": 11909
    },
    {
      "epoch": 2.082622950819672,
      "grad_norm": 0.66047049727575,
      "learning_rate": 9.815210950408704e-05,
      "loss": 11.8855,
      "step": 11910
    },
    {
      "epoch": 2.0827978142076504,
      "grad_norm": 0.6264416098130081,
      "learning_rate": 9.813795105420124e-05,
      "loss": 11.9441,
      "step": 11911
    },
    {
      "epoch": 2.0829726775956283,
      "grad_norm": 0.6678409593565958,
      "learning_rate": 9.812379264165522e-05,
      "loss": 11.9536,
      "step": 11912
    },
    {
      "epoch": 2.0831475409836067,
      "grad_norm": 0.6937099276991834,
      "learning_rate": 9.810963426673297e-05,
      "loss": 12.2015,
      "step": 11913
    },
    {
      "epoch": 2.0833224043715846,
      "grad_norm": 0.6419667057180004,
      "learning_rate": 9.809547592971834e-05,
      "loss": 12.1385,
      "step": 11914
    },
    {
      "epoch": 2.083497267759563,
      "grad_norm": 0.6887246069408705,
      "learning_rate": 9.808131763089526e-05,
      "loss": 12.0447,
      "step": 11915
    },
    {
      "epoch": 2.083672131147541,
      "grad_norm": 0.7012937296435565,
      "learning_rate": 9.806715937054768e-05,
      "loss": 11.9334,
      "step": 11916
    },
    {
      "epoch": 2.083846994535519,
      "grad_norm": 0.676343646503632,
      "learning_rate": 9.805300114895942e-05,
      "loss": 11.8773,
      "step": 11917
    },
    {
      "epoch": 2.0840218579234975,
      "grad_norm": 0.6824220732075039,
      "learning_rate": 9.803884296641454e-05,
      "loss": 11.9902,
      "step": 11918
    },
    {
      "epoch": 2.0841967213114754,
      "grad_norm": 0.6155386284190746,
      "learning_rate": 9.802468482319683e-05,
      "loss": 11.7842,
      "step": 11919
    },
    {
      "epoch": 2.0843715846994537,
      "grad_norm": 0.6073373373057018,
      "learning_rate": 9.801052671959028e-05,
      "loss": 11.8184,
      "step": 11920
    },
    {
      "epoch": 2.0845464480874316,
      "grad_norm": 0.7449577526337167,
      "learning_rate": 9.799636865587879e-05,
      "loss": 11.8253,
      "step": 11921
    },
    {
      "epoch": 2.08472131147541,
      "grad_norm": 0.6723413492620428,
      "learning_rate": 9.798221063234623e-05,
      "loss": 11.6974,
      "step": 11922
    },
    {
      "epoch": 2.084896174863388,
      "grad_norm": 0.650948280919005,
      "learning_rate": 9.796805264927655e-05,
      "loss": 11.8579,
      "step": 11923
    },
    {
      "epoch": 2.085071038251366,
      "grad_norm": 0.6854229247684038,
      "learning_rate": 9.795389470695365e-05,
      "loss": 11.8499,
      "step": 11924
    },
    {
      "epoch": 2.085245901639344,
      "grad_norm": 0.6641034784242941,
      "learning_rate": 9.793973680566147e-05,
      "loss": 11.9467,
      "step": 11925
    },
    {
      "epoch": 2.0854207650273224,
      "grad_norm": 0.7220027638178186,
      "learning_rate": 9.792557894568383e-05,
      "loss": 11.901,
      "step": 11926
    },
    {
      "epoch": 2.0855956284153008,
      "grad_norm": 0.6646933360868468,
      "learning_rate": 9.791142112730477e-05,
      "loss": 11.8288,
      "step": 11927
    },
    {
      "epoch": 2.0857704918032787,
      "grad_norm": 0.6834099448313343,
      "learning_rate": 9.78972633508081e-05,
      "loss": 11.977,
      "step": 11928
    },
    {
      "epoch": 2.085945355191257,
      "grad_norm": 0.6766068993723022,
      "learning_rate": 9.788310561647776e-05,
      "loss": 11.8042,
      "step": 11929
    },
    {
      "epoch": 2.086120218579235,
      "grad_norm": 0.7125574682815169,
      "learning_rate": 9.786894792459768e-05,
      "loss": 12.2588,
      "step": 11930
    },
    {
      "epoch": 2.0862950819672132,
      "grad_norm": 0.7193098360330841,
      "learning_rate": 9.785479027545168e-05,
      "loss": 11.8741,
      "step": 11931
    },
    {
      "epoch": 2.086469945355191,
      "grad_norm": 0.6685841106698577,
      "learning_rate": 9.784063266932375e-05,
      "loss": 12.0516,
      "step": 11932
    },
    {
      "epoch": 2.0866448087431695,
      "grad_norm": 0.6804120459846024,
      "learning_rate": 9.782647510649776e-05,
      "loss": 12.0173,
      "step": 11933
    },
    {
      "epoch": 2.0868196721311474,
      "grad_norm": 0.636862691383253,
      "learning_rate": 9.781231758725766e-05,
      "loss": 11.8755,
      "step": 11934
    },
    {
      "epoch": 2.0869945355191257,
      "grad_norm": 0.7053620349735726,
      "learning_rate": 9.779816011188726e-05,
      "loss": 12.0139,
      "step": 11935
    },
    {
      "epoch": 2.0871693989071036,
      "grad_norm": 0.6652137210599235,
      "learning_rate": 9.77840026806705e-05,
      "loss": 11.6715,
      "step": 11936
    },
    {
      "epoch": 2.087344262295082,
      "grad_norm": 0.6911077420050878,
      "learning_rate": 9.776984529389131e-05,
      "loss": 11.9238,
      "step": 11937
    },
    {
      "epoch": 2.0875191256830603,
      "grad_norm": 0.6287052650735151,
      "learning_rate": 9.775568795183359e-05,
      "loss": 11.7989,
      "step": 11938
    },
    {
      "epoch": 2.087693989071038,
      "grad_norm": 0.722393584024324,
      "learning_rate": 9.774153065478121e-05,
      "loss": 12.1138,
      "step": 11939
    },
    {
      "epoch": 2.0878688524590165,
      "grad_norm": 0.6476023623602175,
      "learning_rate": 9.772737340301807e-05,
      "loss": 11.8814,
      "step": 11940
    },
    {
      "epoch": 2.0880437158469944,
      "grad_norm": 0.6709385531341328,
      "learning_rate": 9.771321619682804e-05,
      "loss": 11.9159,
      "step": 11941
    },
    {
      "epoch": 2.088218579234973,
      "grad_norm": 0.6908521309910871,
      "learning_rate": 9.769905903649508e-05,
      "loss": 11.7038,
      "step": 11942
    },
    {
      "epoch": 2.0883934426229507,
      "grad_norm": 0.6484346987232879,
      "learning_rate": 9.768490192230306e-05,
      "loss": 11.9056,
      "step": 11943
    },
    {
      "epoch": 2.088568306010929,
      "grad_norm": 0.6644826597699932,
      "learning_rate": 9.767074485453584e-05,
      "loss": 11.9752,
      "step": 11944
    },
    {
      "epoch": 2.088743169398907,
      "grad_norm": 0.7617131176452423,
      "learning_rate": 9.765658783347736e-05,
      "loss": 11.8805,
      "step": 11945
    },
    {
      "epoch": 2.0889180327868853,
      "grad_norm": 0.8068537527190401,
      "learning_rate": 9.764243085941145e-05,
      "loss": 11.9212,
      "step": 11946
    },
    {
      "epoch": 2.0890928961748636,
      "grad_norm": 0.6175909662782364,
      "learning_rate": 9.762827393262206e-05,
      "loss": 11.7879,
      "step": 11947
    },
    {
      "epoch": 2.0892677595628415,
      "grad_norm": 0.6248545280494304,
      "learning_rate": 9.761411705339309e-05,
      "loss": 12.1876,
      "step": 11948
    },
    {
      "epoch": 2.08944262295082,
      "grad_norm": 0.6541248970582183,
      "learning_rate": 9.759996022200836e-05,
      "loss": 11.869,
      "step": 11949
    },
    {
      "epoch": 2.0896174863387977,
      "grad_norm": 0.6313407169365783,
      "learning_rate": 9.75858034387518e-05,
      "loss": 12.0718,
      "step": 11950
    },
    {
      "epoch": 2.089792349726776,
      "grad_norm": 0.6534268136940574,
      "learning_rate": 9.757164670390731e-05,
      "loss": 12.0227,
      "step": 11951
    },
    {
      "epoch": 2.089967213114754,
      "grad_norm": 0.6453709554491267,
      "learning_rate": 9.755749001775877e-05,
      "loss": 11.8356,
      "step": 11952
    },
    {
      "epoch": 2.0901420765027323,
      "grad_norm": 0.6680360070265278,
      "learning_rate": 9.754333338059005e-05,
      "loss": 11.6008,
      "step": 11953
    },
    {
      "epoch": 2.09031693989071,
      "grad_norm": 0.6814701031930278,
      "learning_rate": 9.752917679268502e-05,
      "loss": 11.7962,
      "step": 11954
    },
    {
      "epoch": 2.0904918032786886,
      "grad_norm": 0.5715439742951505,
      "learning_rate": 9.751502025432756e-05,
      "loss": 12.1301,
      "step": 11955
    },
    {
      "epoch": 2.0906666666666665,
      "grad_norm": 0.6884640108401373,
      "learning_rate": 9.750086376580161e-05,
      "loss": 11.7811,
      "step": 11956
    },
    {
      "epoch": 2.090841530054645,
      "grad_norm": 0.6777007302929258,
      "learning_rate": 9.748670732739103e-05,
      "loss": 12.0594,
      "step": 11957
    },
    {
      "epoch": 2.091016393442623,
      "grad_norm": 0.6305907387241836,
      "learning_rate": 9.747255093937965e-05,
      "loss": 11.7818,
      "step": 11958
    },
    {
      "epoch": 2.091191256830601,
      "grad_norm": 0.6280874405830532,
      "learning_rate": 9.745839460205141e-05,
      "loss": 11.815,
      "step": 11959
    },
    {
      "epoch": 2.0913661202185794,
      "grad_norm": 0.7315132413897873,
      "learning_rate": 9.744423831569012e-05,
      "loss": 11.7191,
      "step": 11960
    },
    {
      "epoch": 2.0915409836065573,
      "grad_norm": 0.722842637529713,
      "learning_rate": 9.743008208057974e-05,
      "loss": 11.8321,
      "step": 11961
    },
    {
      "epoch": 2.0917158469945356,
      "grad_norm": 0.6879067692619716,
      "learning_rate": 9.74159258970041e-05,
      "loss": 12.0741,
      "step": 11962
    },
    {
      "epoch": 2.0918907103825135,
      "grad_norm": 0.6314330435246209,
      "learning_rate": 9.740176976524707e-05,
      "loss": 12.135,
      "step": 11963
    },
    {
      "epoch": 2.092065573770492,
      "grad_norm": 0.6960008927733743,
      "learning_rate": 9.738761368559256e-05,
      "loss": 12.1537,
      "step": 11964
    },
    {
      "epoch": 2.0922404371584697,
      "grad_norm": 0.7015361650950205,
      "learning_rate": 9.737345765832434e-05,
      "loss": 12.1053,
      "step": 11965
    },
    {
      "epoch": 2.092415300546448,
      "grad_norm": 0.7595419771720887,
      "learning_rate": 9.735930168372644e-05,
      "loss": 12.0573,
      "step": 11966
    },
    {
      "epoch": 2.0925901639344264,
      "grad_norm": 0.6825662603350742,
      "learning_rate": 9.734514576208262e-05,
      "loss": 12.0443,
      "step": 11967
    },
    {
      "epoch": 2.0927650273224043,
      "grad_norm": 0.7075236782892613,
      "learning_rate": 9.733098989367677e-05,
      "loss": 11.8146,
      "step": 11968
    },
    {
      "epoch": 2.0929398907103827,
      "grad_norm": 0.7301253100565546,
      "learning_rate": 9.731683407879281e-05,
      "loss": 11.8011,
      "step": 11969
    },
    {
      "epoch": 2.0931147540983606,
      "grad_norm": 0.673318775240326,
      "learning_rate": 9.73026783177145e-05,
      "loss": 11.9141,
      "step": 11970
    },
    {
      "epoch": 2.093289617486339,
      "grad_norm": 0.7684999459639826,
      "learning_rate": 9.728852261072581e-05,
      "loss": 11.9287,
      "step": 11971
    },
    {
      "epoch": 2.093464480874317,
      "grad_norm": 0.6630429947049569,
      "learning_rate": 9.727436695811056e-05,
      "loss": 12.0454,
      "step": 11972
    },
    {
      "epoch": 2.093639344262295,
      "grad_norm": 0.7667092167730407,
      "learning_rate": 9.726021136015265e-05,
      "loss": 12.1499,
      "step": 11973
    },
    {
      "epoch": 2.093814207650273,
      "grad_norm": 0.693126116722811,
      "learning_rate": 9.724605581713584e-05,
      "loss": 12.0076,
      "step": 11974
    },
    {
      "epoch": 2.0939890710382514,
      "grad_norm": 0.6729862101599913,
      "learning_rate": 9.723190032934412e-05,
      "loss": 11.8908,
      "step": 11975
    },
    {
      "epoch": 2.0941639344262297,
      "grad_norm": 0.7541891162089458,
      "learning_rate": 9.72177448970613e-05,
      "loss": 11.9858,
      "step": 11976
    },
    {
      "epoch": 2.0943387978142076,
      "grad_norm": 0.6889379422316795,
      "learning_rate": 9.720358952057123e-05,
      "loss": 11.995,
      "step": 11977
    },
    {
      "epoch": 2.094513661202186,
      "grad_norm": 0.7852094694478298,
      "learning_rate": 9.71894342001578e-05,
      "loss": 11.9389,
      "step": 11978
    },
    {
      "epoch": 2.094688524590164,
      "grad_norm": 0.7071346125286396,
      "learning_rate": 9.717527893610478e-05,
      "loss": 11.9229,
      "step": 11979
    },
    {
      "epoch": 2.094863387978142,
      "grad_norm": 0.7337079849840865,
      "learning_rate": 9.716112372869614e-05,
      "loss": 11.9818,
      "step": 11980
    },
    {
      "epoch": 2.09503825136612,
      "grad_norm": 0.7208607283197799,
      "learning_rate": 9.714696857821567e-05,
      "loss": 12.0167,
      "step": 11981
    },
    {
      "epoch": 2.0952131147540984,
      "grad_norm": 0.7170099916001494,
      "learning_rate": 9.713281348494726e-05,
      "loss": 12.0661,
      "step": 11982
    },
    {
      "epoch": 2.0953879781420763,
      "grad_norm": 0.6388272726321772,
      "learning_rate": 9.711865844917474e-05,
      "loss": 11.9223,
      "step": 11983
    },
    {
      "epoch": 2.0955628415300547,
      "grad_norm": 0.7285378321449278,
      "learning_rate": 9.710450347118192e-05,
      "loss": 11.9838,
      "step": 11984
    },
    {
      "epoch": 2.0957377049180326,
      "grad_norm": 0.7092824539031927,
      "learning_rate": 9.709034855125273e-05,
      "loss": 11.9309,
      "step": 11985
    },
    {
      "epoch": 2.095912568306011,
      "grad_norm": 0.7563593724955447,
      "learning_rate": 9.707619368967098e-05,
      "loss": 12.2093,
      "step": 11986
    },
    {
      "epoch": 2.0960874316939893,
      "grad_norm": 0.7369450391424128,
      "learning_rate": 9.706203888672056e-05,
      "loss": 12.0604,
      "step": 11987
    },
    {
      "epoch": 2.096262295081967,
      "grad_norm": 0.6213575581793357,
      "learning_rate": 9.704788414268525e-05,
      "loss": 11.8016,
      "step": 11988
    },
    {
      "epoch": 2.0964371584699455,
      "grad_norm": 0.7403931693906808,
      "learning_rate": 9.703372945784889e-05,
      "loss": 11.7926,
      "step": 11989
    },
    {
      "epoch": 2.0966120218579234,
      "grad_norm": 0.6955122101004989,
      "learning_rate": 9.701957483249539e-05,
      "loss": 11.7796,
      "step": 11990
    },
    {
      "epoch": 2.0967868852459017,
      "grad_norm": 0.7184384804299504,
      "learning_rate": 9.700542026690859e-05,
      "loss": 11.9615,
      "step": 11991
    },
    {
      "epoch": 2.0969617486338796,
      "grad_norm": 0.666395962642651,
      "learning_rate": 9.699126576137231e-05,
      "loss": 11.865,
      "step": 11992
    },
    {
      "epoch": 2.097136612021858,
      "grad_norm": 0.6706306496942441,
      "learning_rate": 9.697711131617037e-05,
      "loss": 12.0617,
      "step": 11993
    },
    {
      "epoch": 2.097311475409836,
      "grad_norm": 0.7356127251528503,
      "learning_rate": 9.696295693158661e-05,
      "loss": 12.0265,
      "step": 11994
    },
    {
      "epoch": 2.097486338797814,
      "grad_norm": 0.7266051219282893,
      "learning_rate": 9.69488026079049e-05,
      "loss": 12.0911,
      "step": 11995
    },
    {
      "epoch": 2.0976612021857926,
      "grad_norm": 0.6881791135872465,
      "learning_rate": 9.69346483454091e-05,
      "loss": 11.9882,
      "step": 11996
    },
    {
      "epoch": 2.0978360655737704,
      "grad_norm": 0.695035659478405,
      "learning_rate": 9.692049414438299e-05,
      "loss": 12.0642,
      "step": 11997
    },
    {
      "epoch": 2.098010928961749,
      "grad_norm": 0.6126261432648584,
      "learning_rate": 9.69063400051104e-05,
      "loss": 11.6896,
      "step": 11998
    },
    {
      "epoch": 2.0981857923497267,
      "grad_norm": 0.6653726850586389,
      "learning_rate": 9.689218592787521e-05,
      "loss": 12.024,
      "step": 11999
    },
    {
      "epoch": 2.098360655737705,
      "grad_norm": 0.7110946762812899,
      "learning_rate": 9.687803191296126e-05,
      "loss": 11.9808,
      "step": 12000
    },
    {
      "epoch": 2.098535519125683,
      "grad_norm": 0.6777012836828626,
      "learning_rate": 9.686387796065234e-05,
      "loss": 12.1409,
      "step": 12001
    },
    {
      "epoch": 2.0987103825136613,
      "grad_norm": 0.8184864425910283,
      "learning_rate": 9.68497240712323e-05,
      "loss": 12.1321,
      "step": 12002
    },
    {
      "epoch": 2.098885245901639,
      "grad_norm": 0.6327952365391065,
      "learning_rate": 9.683557024498493e-05,
      "loss": 11.9575,
      "step": 12003
    },
    {
      "epoch": 2.0990601092896175,
      "grad_norm": 0.7099972674871419,
      "learning_rate": 9.682141648219411e-05,
      "loss": 11.8236,
      "step": 12004
    },
    {
      "epoch": 2.0992349726775954,
      "grad_norm": 0.7138506820250989,
      "learning_rate": 9.68072627831437e-05,
      "loss": 11.7098,
      "step": 12005
    },
    {
      "epoch": 2.0994098360655737,
      "grad_norm": 0.6722522148590405,
      "learning_rate": 9.679310914811742e-05,
      "loss": 11.6627,
      "step": 12006
    },
    {
      "epoch": 2.099584699453552,
      "grad_norm": 0.8342482717997004,
      "learning_rate": 9.677895557739916e-05,
      "loss": 11.987,
      "step": 12007
    },
    {
      "epoch": 2.09975956284153,
      "grad_norm": 0.6974427065881226,
      "learning_rate": 9.676480207127272e-05,
      "loss": 12.2584,
      "step": 12008
    },
    {
      "epoch": 2.0999344262295083,
      "grad_norm": 0.712977420581952,
      "learning_rate": 9.675064863002196e-05,
      "loss": 12.0874,
      "step": 12009
    },
    {
      "epoch": 2.100109289617486,
      "grad_norm": 0.7242005550582064,
      "learning_rate": 9.673649525393066e-05,
      "loss": 11.9448,
      "step": 12010
    },
    {
      "epoch": 2.1002841530054646,
      "grad_norm": 0.6456956087346492,
      "learning_rate": 9.672234194328265e-05,
      "loss": 12.0921,
      "step": 12011
    },
    {
      "epoch": 2.1004590163934425,
      "grad_norm": 0.6822480202165104,
      "learning_rate": 9.670818869836177e-05,
      "loss": 11.9691,
      "step": 12012
    },
    {
      "epoch": 2.100633879781421,
      "grad_norm": 0.6731269175606728,
      "learning_rate": 9.669403551945173e-05,
      "loss": 11.6361,
      "step": 12013
    },
    {
      "epoch": 2.1008087431693987,
      "grad_norm": 0.7229925881337844,
      "learning_rate": 9.667988240683652e-05,
      "loss": 12.0286,
      "step": 12014
    },
    {
      "epoch": 2.100983606557377,
      "grad_norm": 0.71684281969777,
      "learning_rate": 9.666572936079983e-05,
      "loss": 11.9478,
      "step": 12015
    },
    {
      "epoch": 2.1011584699453554,
      "grad_norm": 0.6739044406480549,
      "learning_rate": 9.66515763816255e-05,
      "loss": 12.0219,
      "step": 12016
    },
    {
      "epoch": 2.1013333333333333,
      "grad_norm": 0.7472439422243683,
      "learning_rate": 9.663742346959736e-05,
      "loss": 11.688,
      "step": 12017
    },
    {
      "epoch": 2.1015081967213116,
      "grad_norm": 0.7212076812235583,
      "learning_rate": 9.662327062499918e-05,
      "loss": 11.6456,
      "step": 12018
    },
    {
      "epoch": 2.1016830601092895,
      "grad_norm": 0.745639665872201,
      "learning_rate": 9.66091178481148e-05,
      "loss": 11.9543,
      "step": 12019
    },
    {
      "epoch": 2.101857923497268,
      "grad_norm": 0.8407836640397314,
      "learning_rate": 9.659496513922801e-05,
      "loss": 11.9121,
      "step": 12020
    },
    {
      "epoch": 2.1020327868852458,
      "grad_norm": 0.6951479501810968,
      "learning_rate": 9.658081249862266e-05,
      "loss": 12.0677,
      "step": 12021
    },
    {
      "epoch": 2.102207650273224,
      "grad_norm": 0.8710578380362431,
      "learning_rate": 9.656665992658245e-05,
      "loss": 11.9071,
      "step": 12022
    },
    {
      "epoch": 2.102382513661202,
      "grad_norm": 0.6452600635894823,
      "learning_rate": 9.655250742339132e-05,
      "loss": 12.0517,
      "step": 12023
    },
    {
      "epoch": 2.1025573770491803,
      "grad_norm": 0.8550250457076615,
      "learning_rate": 9.653835498933296e-05,
      "loss": 12.1596,
      "step": 12024
    },
    {
      "epoch": 2.1027322404371587,
      "grad_norm": 0.7172044618236489,
      "learning_rate": 9.652420262469124e-05,
      "loss": 12.0763,
      "step": 12025
    },
    {
      "epoch": 2.1029071038251366,
      "grad_norm": 0.7056041786774392,
      "learning_rate": 9.651005032974994e-05,
      "loss": 11.9477,
      "step": 12026
    },
    {
      "epoch": 2.103081967213115,
      "grad_norm": 0.8258518888611982,
      "learning_rate": 9.64958981047928e-05,
      "loss": 11.9497,
      "step": 12027
    },
    {
      "epoch": 2.103256830601093,
      "grad_norm": 0.6312660757296019,
      "learning_rate": 9.648174595010368e-05,
      "loss": 12.0667,
      "step": 12028
    },
    {
      "epoch": 2.103431693989071,
      "grad_norm": 0.805878468594948,
      "learning_rate": 9.646759386596635e-05,
      "loss": 11.8094,
      "step": 12029
    },
    {
      "epoch": 2.103606557377049,
      "grad_norm": 0.6428705743120724,
      "learning_rate": 9.645344185266464e-05,
      "loss": 11.7179,
      "step": 12030
    },
    {
      "epoch": 2.1037814207650274,
      "grad_norm": 0.7047286999352127,
      "learning_rate": 9.643928991048229e-05,
      "loss": 11.9594,
      "step": 12031
    },
    {
      "epoch": 2.1039562841530053,
      "grad_norm": 0.6558986331900927,
      "learning_rate": 9.64251380397031e-05,
      "loss": 11.9557,
      "step": 12032
    },
    {
      "epoch": 2.1041311475409836,
      "grad_norm": 0.7394319863451775,
      "learning_rate": 9.641098624061087e-05,
      "loss": 12.0207,
      "step": 12033
    },
    {
      "epoch": 2.104306010928962,
      "grad_norm": 0.6261300269123721,
      "learning_rate": 9.639683451348939e-05,
      "loss": 11.9773,
      "step": 12034
    },
    {
      "epoch": 2.10448087431694,
      "grad_norm": 0.7474843243423533,
      "learning_rate": 9.638268285862246e-05,
      "loss": 11.9479,
      "step": 12035
    },
    {
      "epoch": 2.104655737704918,
      "grad_norm": 0.7127698685450558,
      "learning_rate": 9.636853127629383e-05,
      "loss": 11.713,
      "step": 12036
    },
    {
      "epoch": 2.104830601092896,
      "grad_norm": 0.649877477236501,
      "learning_rate": 9.635437976678728e-05,
      "loss": 11.8669,
      "step": 12037
    },
    {
      "epoch": 2.1050054644808744,
      "grad_norm": 0.6222076603803383,
      "learning_rate": 9.634022833038662e-05,
      "loss": 11.819,
      "step": 12038
    },
    {
      "epoch": 2.1051803278688523,
      "grad_norm": 0.7858619221368385,
      "learning_rate": 9.632607696737564e-05,
      "loss": 11.9763,
      "step": 12039
    },
    {
      "epoch": 2.1053551912568307,
      "grad_norm": 0.6268498189742195,
      "learning_rate": 9.63119256780381e-05,
      "loss": 11.9022,
      "step": 12040
    },
    {
      "epoch": 2.1055300546448086,
      "grad_norm": 0.7984342593716013,
      "learning_rate": 9.629777446265776e-05,
      "loss": 11.7427,
      "step": 12041
    },
    {
      "epoch": 2.105704918032787,
      "grad_norm": 0.6582728467274335,
      "learning_rate": 9.628362332151837e-05,
      "loss": 11.9808,
      "step": 12042
    },
    {
      "epoch": 2.105879781420765,
      "grad_norm": 0.663189239752876,
      "learning_rate": 9.626947225490378e-05,
      "loss": 11.9392,
      "step": 12043
    },
    {
      "epoch": 2.106054644808743,
      "grad_norm": 0.7280257564634521,
      "learning_rate": 9.625532126309775e-05,
      "loss": 11.9911,
      "step": 12044
    },
    {
      "epoch": 2.1062295081967215,
      "grad_norm": 0.6954198091935947,
      "learning_rate": 9.6241170346384e-05,
      "loss": 12.0601,
      "step": 12045
    },
    {
      "epoch": 2.1064043715846994,
      "grad_norm": 0.6236045737992271,
      "learning_rate": 9.62270195050463e-05,
      "loss": 11.9738,
      "step": 12046
    },
    {
      "epoch": 2.1065792349726777,
      "grad_norm": 0.6470579589491454,
      "learning_rate": 9.621286873936848e-05,
      "loss": 11.9118,
      "step": 12047
    },
    {
      "epoch": 2.1067540983606556,
      "grad_norm": 0.6342515681956665,
      "learning_rate": 9.619871804963427e-05,
      "loss": 11.8561,
      "step": 12048
    },
    {
      "epoch": 2.106928961748634,
      "grad_norm": 0.7384171437021174,
      "learning_rate": 9.618456743612742e-05,
      "loss": 11.8744,
      "step": 12049
    },
    {
      "epoch": 2.107103825136612,
      "grad_norm": 0.6806958142498071,
      "learning_rate": 9.617041689913172e-05,
      "loss": 12.1145,
      "step": 12050
    },
    {
      "epoch": 2.10727868852459,
      "grad_norm": 0.6637330205366491,
      "learning_rate": 9.615626643893089e-05,
      "loss": 11.8076,
      "step": 12051
    },
    {
      "epoch": 2.107453551912568,
      "grad_norm": 0.6526580167272948,
      "learning_rate": 9.614211605580874e-05,
      "loss": 12.0373,
      "step": 12052
    },
    {
      "epoch": 2.1076284153005465,
      "grad_norm": 0.6524259157781103,
      "learning_rate": 9.612796575004903e-05,
      "loss": 12.0288,
      "step": 12053
    },
    {
      "epoch": 2.107803278688525,
      "grad_norm": 0.5982753943220364,
      "learning_rate": 9.611381552193548e-05,
      "loss": 11.8352,
      "step": 12054
    },
    {
      "epoch": 2.1079781420765027,
      "grad_norm": 0.6277431229696118,
      "learning_rate": 9.609966537175185e-05,
      "loss": 11.9838,
      "step": 12055
    },
    {
      "epoch": 2.108153005464481,
      "grad_norm": 0.7044237486625377,
      "learning_rate": 9.608551529978189e-05,
      "loss": 11.9331,
      "step": 12056
    },
    {
      "epoch": 2.108327868852459,
      "grad_norm": 0.6653505345422335,
      "learning_rate": 9.607136530630941e-05,
      "loss": 11.9842,
      "step": 12057
    },
    {
      "epoch": 2.1085027322404373,
      "grad_norm": 0.6884282321410075,
      "learning_rate": 9.605721539161811e-05,
      "loss": 11.7342,
      "step": 12058
    },
    {
      "epoch": 2.108677595628415,
      "grad_norm": 0.6590827028084235,
      "learning_rate": 9.604306555599174e-05,
      "loss": 11.9459,
      "step": 12059
    },
    {
      "epoch": 2.1088524590163935,
      "grad_norm": 0.6299322104845716,
      "learning_rate": 9.602891579971405e-05,
      "loss": 11.894,
      "step": 12060
    },
    {
      "epoch": 2.1090273224043714,
      "grad_norm": 0.6680700491444582,
      "learning_rate": 9.601476612306876e-05,
      "loss": 11.9656,
      "step": 12061
    },
    {
      "epoch": 2.1092021857923497,
      "grad_norm": 0.693061979839655,
      "learning_rate": 9.60006165263397e-05,
      "loss": 12.0954,
      "step": 12062
    },
    {
      "epoch": 2.1093770491803276,
      "grad_norm": 0.6345871422974804,
      "learning_rate": 9.598646700981051e-05,
      "loss": 12.0237,
      "step": 12063
    },
    {
      "epoch": 2.109551912568306,
      "grad_norm": 0.7668365175538718,
      "learning_rate": 9.597231757376501e-05,
      "loss": 11.8098,
      "step": 12064
    },
    {
      "epoch": 2.1097267759562843,
      "grad_norm": 0.68363699572093,
      "learning_rate": 9.595816821848693e-05,
      "loss": 12.145,
      "step": 12065
    },
    {
      "epoch": 2.1099016393442622,
      "grad_norm": 0.6779926211445728,
      "learning_rate": 9.594401894425994e-05,
      "loss": 11.6022,
      "step": 12066
    },
    {
      "epoch": 2.1100765027322406,
      "grad_norm": 0.687920682520824,
      "learning_rate": 9.592986975136785e-05,
      "loss": 11.887,
      "step": 12067
    },
    {
      "epoch": 2.1102513661202185,
      "grad_norm": 0.6578945226980278,
      "learning_rate": 9.591572064009436e-05,
      "loss": 12.1945,
      "step": 12068
    },
    {
      "epoch": 2.110426229508197,
      "grad_norm": 0.7082762817405744,
      "learning_rate": 9.590157161072323e-05,
      "loss": 11.9203,
      "step": 12069
    },
    {
      "epoch": 2.1106010928961747,
      "grad_norm": 0.6728183437465027,
      "learning_rate": 9.58874226635381e-05,
      "loss": 12.1895,
      "step": 12070
    },
    {
      "epoch": 2.110775956284153,
      "grad_norm": 0.5934953184002759,
      "learning_rate": 9.587327379882286e-05,
      "loss": 12.0567,
      "step": 12071
    },
    {
      "epoch": 2.110950819672131,
      "grad_norm": 0.6906531914965796,
      "learning_rate": 9.585912501686111e-05,
      "loss": 12.016,
      "step": 12072
    },
    {
      "epoch": 2.1111256830601093,
      "grad_norm": 0.6751501636534306,
      "learning_rate": 9.584497631793664e-05,
      "loss": 12.0522,
      "step": 12073
    },
    {
      "epoch": 2.1113005464480876,
      "grad_norm": 0.5747816999145999,
      "learning_rate": 9.583082770233315e-05,
      "loss": 11.9913,
      "step": 12074
    },
    {
      "epoch": 2.1114754098360655,
      "grad_norm": 0.7497871919071503,
      "learning_rate": 9.581667917033434e-05,
      "loss": 12.0746,
      "step": 12075
    },
    {
      "epoch": 2.111650273224044,
      "grad_norm": 0.7029120698937712,
      "learning_rate": 9.580253072222398e-05,
      "loss": 12.0704,
      "step": 12076
    },
    {
      "epoch": 2.1118251366120218,
      "grad_norm": 0.6448400329661979,
      "learning_rate": 9.578838235828575e-05,
      "loss": 11.8431,
      "step": 12077
    },
    {
      "epoch": 2.112,
      "grad_norm": 0.6294958073639286,
      "learning_rate": 9.577423407880343e-05,
      "loss": 12.0075,
      "step": 12078
    },
    {
      "epoch": 2.112174863387978,
      "grad_norm": 0.7377071734547487,
      "learning_rate": 9.576008588406065e-05,
      "loss": 11.8335,
      "step": 12079
    },
    {
      "epoch": 2.1123497267759563,
      "grad_norm": 0.6709262877049447,
      "learning_rate": 9.574593777434116e-05,
      "loss": 12.1568,
      "step": 12080
    },
    {
      "epoch": 2.1125245901639342,
      "grad_norm": 0.8338334729878601,
      "learning_rate": 9.57317897499287e-05,
      "loss": 12.0925,
      "step": 12081
    },
    {
      "epoch": 2.1126994535519126,
      "grad_norm": 0.6667341251884956,
      "learning_rate": 9.571764181110694e-05,
      "loss": 11.9314,
      "step": 12082
    },
    {
      "epoch": 2.112874316939891,
      "grad_norm": 0.7980546177991882,
      "learning_rate": 9.570349395815964e-05,
      "loss": 12.0475,
      "step": 12083
    },
    {
      "epoch": 2.113049180327869,
      "grad_norm": 0.7069065442712642,
      "learning_rate": 9.568934619137046e-05,
      "loss": 11.9835,
      "step": 12084
    },
    {
      "epoch": 2.113224043715847,
      "grad_norm": 0.645087577880777,
      "learning_rate": 9.567519851102311e-05,
      "loss": 11.9194,
      "step": 12085
    },
    {
      "epoch": 2.113398907103825,
      "grad_norm": 0.867402075755154,
      "learning_rate": 9.566105091740132e-05,
      "loss": 11.6696,
      "step": 12086
    },
    {
      "epoch": 2.1135737704918034,
      "grad_norm": 0.6733716573040307,
      "learning_rate": 9.564690341078882e-05,
      "loss": 11.7673,
      "step": 12087
    },
    {
      "epoch": 2.1137486338797813,
      "grad_norm": 0.7334002032532796,
      "learning_rate": 9.563275599146922e-05,
      "loss": 11.8787,
      "step": 12088
    },
    {
      "epoch": 2.1139234972677596,
      "grad_norm": 0.7884708213026347,
      "learning_rate": 9.561860865972629e-05,
      "loss": 11.7349,
      "step": 12089
    },
    {
      "epoch": 2.1140983606557375,
      "grad_norm": 0.6899202662233521,
      "learning_rate": 9.56044614158437e-05,
      "loss": 11.9902,
      "step": 12090
    },
    {
      "epoch": 2.114273224043716,
      "grad_norm": 0.7795097309732004,
      "learning_rate": 9.559031426010516e-05,
      "loss": 11.9441,
      "step": 12091
    },
    {
      "epoch": 2.114448087431694,
      "grad_norm": 0.7710589088979548,
      "learning_rate": 9.557616719279438e-05,
      "loss": 12.2882,
      "step": 12092
    },
    {
      "epoch": 2.114622950819672,
      "grad_norm": 0.6680612917599522,
      "learning_rate": 9.556202021419501e-05,
      "loss": 11.9296,
      "step": 12093
    },
    {
      "epoch": 2.1147978142076505,
      "grad_norm": 0.858959220681495,
      "learning_rate": 9.554787332459074e-05,
      "loss": 11.823,
      "step": 12094
    },
    {
      "epoch": 2.1149726775956283,
      "grad_norm": 0.7074629210226503,
      "learning_rate": 9.55337265242653e-05,
      "loss": 11.876,
      "step": 12095
    },
    {
      "epoch": 2.1151475409836067,
      "grad_norm": 0.6850201667682501,
      "learning_rate": 9.551957981350239e-05,
      "loss": 11.7806,
      "step": 12096
    },
    {
      "epoch": 2.1153224043715846,
      "grad_norm": 0.7987944166697495,
      "learning_rate": 9.550543319258563e-05,
      "loss": 11.9272,
      "step": 12097
    },
    {
      "epoch": 2.115497267759563,
      "grad_norm": 0.6649547935240752,
      "learning_rate": 9.549128666179873e-05,
      "loss": 11.9946,
      "step": 12098
    },
    {
      "epoch": 2.115672131147541,
      "grad_norm": 0.7272726425725995,
      "learning_rate": 9.547714022142537e-05,
      "loss": 12.0693,
      "step": 12099
    },
    {
      "epoch": 2.115846994535519,
      "grad_norm": 0.7459156098673047,
      "learning_rate": 9.546299387174925e-05,
      "loss": 11.7884,
      "step": 12100
    },
    {
      "epoch": 2.116021857923497,
      "grad_norm": 0.6820301828237688,
      "learning_rate": 9.544884761305404e-05,
      "loss": 11.8827,
      "step": 12101
    },
    {
      "epoch": 2.1161967213114754,
      "grad_norm": 0.7580425348084282,
      "learning_rate": 9.543470144562341e-05,
      "loss": 12.107,
      "step": 12102
    },
    {
      "epoch": 2.1163715846994537,
      "grad_norm": 0.6687325572675555,
      "learning_rate": 9.542055536974103e-05,
      "loss": 11.842,
      "step": 12103
    },
    {
      "epoch": 2.1165464480874316,
      "grad_norm": 0.6319844300716372,
      "learning_rate": 9.540640938569054e-05,
      "loss": 11.9445,
      "step": 12104
    },
    {
      "epoch": 2.11672131147541,
      "grad_norm": 0.6977588184448537,
      "learning_rate": 9.53922634937557e-05,
      "loss": 11.9158,
      "step": 12105
    },
    {
      "epoch": 2.116896174863388,
      "grad_norm": 0.6209527457244023,
      "learning_rate": 9.537811769422009e-05,
      "loss": 12.1098,
      "step": 12106
    },
    {
      "epoch": 2.117071038251366,
      "grad_norm": 0.6872532708146646,
      "learning_rate": 9.536397198736743e-05,
      "loss": 11.9603,
      "step": 12107
    },
    {
      "epoch": 2.117245901639344,
      "grad_norm": 0.6886311414192776,
      "learning_rate": 9.534982637348137e-05,
      "loss": 11.7042,
      "step": 12108
    },
    {
      "epoch": 2.1174207650273225,
      "grad_norm": 0.6556412680624449,
      "learning_rate": 9.533568085284552e-05,
      "loss": 11.938,
      "step": 12109
    },
    {
      "epoch": 2.1175956284153004,
      "grad_norm": 0.6609068509895527,
      "learning_rate": 9.532153542574365e-05,
      "loss": 12.1836,
      "step": 12110
    },
    {
      "epoch": 2.1177704918032787,
      "grad_norm": 0.699831971137092,
      "learning_rate": 9.530739009245934e-05,
      "loss": 11.9397,
      "step": 12111
    },
    {
      "epoch": 2.117945355191257,
      "grad_norm": 0.7387307292031731,
      "learning_rate": 9.529324485327626e-05,
      "loss": 12.0206,
      "step": 12112
    },
    {
      "epoch": 2.118120218579235,
      "grad_norm": 0.6728590851824089,
      "learning_rate": 9.52790997084781e-05,
      "loss": 11.7192,
      "step": 12113
    },
    {
      "epoch": 2.1182950819672133,
      "grad_norm": 0.655103148134238,
      "learning_rate": 9.526495465834844e-05,
      "loss": 11.9873,
      "step": 12114
    },
    {
      "epoch": 2.118469945355191,
      "grad_norm": 0.7143947221896969,
      "learning_rate": 9.525080970317099e-05,
      "loss": 11.8929,
      "step": 12115
    },
    {
      "epoch": 2.1186448087431695,
      "grad_norm": 0.6970292329557423,
      "learning_rate": 9.52366648432294e-05,
      "loss": 12.0989,
      "step": 12116
    },
    {
      "epoch": 2.1188196721311474,
      "grad_norm": 0.5972469217516766,
      "learning_rate": 9.522252007880732e-05,
      "loss": 12.0931,
      "step": 12117
    },
    {
      "epoch": 2.1189945355191258,
      "grad_norm": 0.6846251518537316,
      "learning_rate": 9.520837541018833e-05,
      "loss": 12.1929,
      "step": 12118
    },
    {
      "epoch": 2.1191693989071037,
      "grad_norm": 0.6685464811595517,
      "learning_rate": 9.519423083765617e-05,
      "loss": 12.0205,
      "step": 12119
    },
    {
      "epoch": 2.119344262295082,
      "grad_norm": 0.6618139340644694,
      "learning_rate": 9.518008636149442e-05,
      "loss": 11.6731,
      "step": 12120
    },
    {
      "epoch": 2.11951912568306,
      "grad_norm": 0.754831496220189,
      "learning_rate": 9.516594198198674e-05,
      "loss": 11.8372,
      "step": 12121
    },
    {
      "epoch": 2.1196939890710382,
      "grad_norm": 0.6151370573500755,
      "learning_rate": 9.51517976994168e-05,
      "loss": 11.9921,
      "step": 12122
    },
    {
      "epoch": 2.1198688524590166,
      "grad_norm": 0.7610137807944181,
      "learning_rate": 9.513765351406814e-05,
      "loss": 11.8304,
      "step": 12123
    },
    {
      "epoch": 2.1200437158469945,
      "grad_norm": 0.6974679520426565,
      "learning_rate": 9.512350942622449e-05,
      "loss": 11.9496,
      "step": 12124
    },
    {
      "epoch": 2.120218579234973,
      "grad_norm": 0.6342385267611786,
      "learning_rate": 9.510936543616944e-05,
      "loss": 11.9732,
      "step": 12125
    },
    {
      "epoch": 2.1203934426229507,
      "grad_norm": 0.6879386870318858,
      "learning_rate": 9.509522154418667e-05,
      "loss": 12.136,
      "step": 12126
    },
    {
      "epoch": 2.120568306010929,
      "grad_norm": 0.6658870558520009,
      "learning_rate": 9.508107775055973e-05,
      "loss": 12.1516,
      "step": 12127
    },
    {
      "epoch": 2.120743169398907,
      "grad_norm": 0.6427771520138618,
      "learning_rate": 9.506693405557226e-05,
      "loss": 11.8256,
      "step": 12128
    },
    {
      "epoch": 2.1209180327868853,
      "grad_norm": 0.7607258588373373,
      "learning_rate": 9.505279045950793e-05,
      "loss": 11.7949,
      "step": 12129
    },
    {
      "epoch": 2.121092896174863,
      "grad_norm": 0.6499604282346304,
      "learning_rate": 9.503864696265034e-05,
      "loss": 12.1085,
      "step": 12130
    },
    {
      "epoch": 2.1212677595628415,
      "grad_norm": 0.6800679633732758,
      "learning_rate": 9.502450356528316e-05,
      "loss": 11.9427,
      "step": 12131
    },
    {
      "epoch": 2.12144262295082,
      "grad_norm": 0.6669498348315746,
      "learning_rate": 9.501036026768991e-05,
      "loss": 12.0458,
      "step": 12132
    },
    {
      "epoch": 2.1216174863387978,
      "grad_norm": 0.6402689344641053,
      "learning_rate": 9.499621707015425e-05,
      "loss": 11.9587,
      "step": 12133
    },
    {
      "epoch": 2.121792349726776,
      "grad_norm": 0.6079842304478426,
      "learning_rate": 9.49820739729598e-05,
      "loss": 11.9834,
      "step": 12134
    },
    {
      "epoch": 2.121967213114754,
      "grad_norm": 0.7486485915728063,
      "learning_rate": 9.496793097639022e-05,
      "loss": 12.1345,
      "step": 12135
    },
    {
      "epoch": 2.1221420765027323,
      "grad_norm": 0.6351275374043192,
      "learning_rate": 9.495378808072905e-05,
      "loss": 12.1343,
      "step": 12136
    },
    {
      "epoch": 2.1223169398907102,
      "grad_norm": 0.6676630242241589,
      "learning_rate": 9.493964528625991e-05,
      "loss": 11.872,
      "step": 12137
    },
    {
      "epoch": 2.1224918032786886,
      "grad_norm": 0.6613978965926387,
      "learning_rate": 9.492550259326642e-05,
      "loss": 12.1505,
      "step": 12138
    },
    {
      "epoch": 2.1226666666666665,
      "grad_norm": 0.6477419080620134,
      "learning_rate": 9.491136000203219e-05,
      "loss": 12.0875,
      "step": 12139
    },
    {
      "epoch": 2.122841530054645,
      "grad_norm": 0.6194989059968801,
      "learning_rate": 9.489721751284082e-05,
      "loss": 11.948,
      "step": 12140
    },
    {
      "epoch": 2.123016393442623,
      "grad_norm": 0.6844930604431642,
      "learning_rate": 9.488307512597592e-05,
      "loss": 12.0927,
      "step": 12141
    },
    {
      "epoch": 2.123191256830601,
      "grad_norm": 0.6824937614115364,
      "learning_rate": 9.486893284172102e-05,
      "loss": 11.9901,
      "step": 12142
    },
    {
      "epoch": 2.1233661202185794,
      "grad_norm": 0.6157137386257313,
      "learning_rate": 9.485479066035981e-05,
      "loss": 12.0376,
      "step": 12143
    },
    {
      "epoch": 2.1235409836065573,
      "grad_norm": 0.7285991120741057,
      "learning_rate": 9.484064858217587e-05,
      "loss": 12.0093,
      "step": 12144
    },
    {
      "epoch": 2.1237158469945356,
      "grad_norm": 0.6612607948041246,
      "learning_rate": 9.482650660745274e-05,
      "loss": 11.9514,
      "step": 12145
    },
    {
      "epoch": 2.1238907103825135,
      "grad_norm": 0.7186801986086176,
      "learning_rate": 9.481236473647404e-05,
      "loss": 11.9404,
      "step": 12146
    },
    {
      "epoch": 2.124065573770492,
      "grad_norm": 0.6968711266840745,
      "learning_rate": 9.479822296952335e-05,
      "loss": 11.7707,
      "step": 12147
    },
    {
      "epoch": 2.1242404371584698,
      "grad_norm": 0.6694593783607462,
      "learning_rate": 9.478408130688426e-05,
      "loss": 12.1291,
      "step": 12148
    },
    {
      "epoch": 2.124415300546448,
      "grad_norm": 0.7629402474102364,
      "learning_rate": 9.476993974884038e-05,
      "loss": 11.8072,
      "step": 12149
    },
    {
      "epoch": 2.1245901639344265,
      "grad_norm": 0.7172265459132293,
      "learning_rate": 9.475579829567525e-05,
      "loss": 12.2027,
      "step": 12150
    },
    {
      "epoch": 2.1247650273224044,
      "grad_norm": 0.7107020061675168,
      "learning_rate": 9.474165694767246e-05,
      "loss": 12.0635,
      "step": 12151
    },
    {
      "epoch": 2.1249398907103827,
      "grad_norm": 0.6836870896466465,
      "learning_rate": 9.472751570511556e-05,
      "loss": 11.838,
      "step": 12152
    },
    {
      "epoch": 2.1251147540983606,
      "grad_norm": 0.6421774133794628,
      "learning_rate": 9.471337456828822e-05,
      "loss": 12.1198,
      "step": 12153
    },
    {
      "epoch": 2.125289617486339,
      "grad_norm": 0.6641596769447181,
      "learning_rate": 9.469923353747393e-05,
      "loss": 11.9581,
      "step": 12154
    },
    {
      "epoch": 2.125464480874317,
      "grad_norm": 0.6717007222647046,
      "learning_rate": 9.468509261295628e-05,
      "loss": 11.7437,
      "step": 12155
    },
    {
      "epoch": 2.125639344262295,
      "grad_norm": 0.650582499859069,
      "learning_rate": 9.467095179501887e-05,
      "loss": 12.1174,
      "step": 12156
    },
    {
      "epoch": 2.125814207650273,
      "grad_norm": 0.6953155717917213,
      "learning_rate": 9.465681108394516e-05,
      "loss": 11.8772,
      "step": 12157
    },
    {
      "epoch": 2.1259890710382514,
      "grad_norm": 0.6655245940092709,
      "learning_rate": 9.464267048001887e-05,
      "loss": 11.8348,
      "step": 12158
    },
    {
      "epoch": 2.1261639344262293,
      "grad_norm": 0.686544895619105,
      "learning_rate": 9.462852998352346e-05,
      "loss": 12.0148,
      "step": 12159
    },
    {
      "epoch": 2.1263387978142076,
      "grad_norm": 0.7132019999761002,
      "learning_rate": 9.461438959474252e-05,
      "loss": 11.4952,
      "step": 12160
    },
    {
      "epoch": 2.126513661202186,
      "grad_norm": 0.6239532644931801,
      "learning_rate": 9.460024931395962e-05,
      "loss": 11.6183,
      "step": 12161
    },
    {
      "epoch": 2.126688524590164,
      "grad_norm": 0.5871986948635309,
      "learning_rate": 9.458610914145826e-05,
      "loss": 11.7993,
      "step": 12162
    },
    {
      "epoch": 2.1268633879781422,
      "grad_norm": 0.659194490999742,
      "learning_rate": 9.457196907752204e-05,
      "loss": 11.9937,
      "step": 12163
    },
    {
      "epoch": 2.12703825136612,
      "grad_norm": 0.6450356568761938,
      "learning_rate": 9.455782912243454e-05,
      "loss": 12.0412,
      "step": 12164
    },
    {
      "epoch": 2.1272131147540985,
      "grad_norm": 0.6165559146404029,
      "learning_rate": 9.454368927647927e-05,
      "loss": 11.9445,
      "step": 12165
    },
    {
      "epoch": 2.1273879781420764,
      "grad_norm": 0.6603684217037677,
      "learning_rate": 9.452954953993973e-05,
      "loss": 11.8656,
      "step": 12166
    },
    {
      "epoch": 2.1275628415300547,
      "grad_norm": 0.6194779426766139,
      "learning_rate": 9.451540991309958e-05,
      "loss": 11.7217,
      "step": 12167
    },
    {
      "epoch": 2.1277377049180326,
      "grad_norm": 0.6625610475634529,
      "learning_rate": 9.450127039624228e-05,
      "loss": 11.8645,
      "step": 12168
    },
    {
      "epoch": 2.127912568306011,
      "grad_norm": 0.6442640979427452,
      "learning_rate": 9.44871309896514e-05,
      "loss": 11.821,
      "step": 12169
    },
    {
      "epoch": 2.128087431693989,
      "grad_norm": 0.6856949702979899,
      "learning_rate": 9.447299169361048e-05,
      "loss": 12.0994,
      "step": 12170
    },
    {
      "epoch": 2.128262295081967,
      "grad_norm": 0.676409833965256,
      "learning_rate": 9.4458852508403e-05,
      "loss": 11.9309,
      "step": 12171
    },
    {
      "epoch": 2.1284371584699455,
      "grad_norm": 0.6612405883003969,
      "learning_rate": 9.444471343431258e-05,
      "loss": 12.0672,
      "step": 12172
    },
    {
      "epoch": 2.1286120218579234,
      "grad_norm": 0.640964874297984,
      "learning_rate": 9.44305744716227e-05,
      "loss": 12.1205,
      "step": 12173
    },
    {
      "epoch": 2.1287868852459018,
      "grad_norm": 0.7152916484357126,
      "learning_rate": 9.441643562061693e-05,
      "loss": 11.9726,
      "step": 12174
    },
    {
      "epoch": 2.1289617486338797,
      "grad_norm": 0.639293660483913,
      "learning_rate": 9.440229688157875e-05,
      "loss": 11.62,
      "step": 12175
    },
    {
      "epoch": 2.129136612021858,
      "grad_norm": 0.6703235388834454,
      "learning_rate": 9.438815825479167e-05,
      "loss": 12.0715,
      "step": 12176
    },
    {
      "epoch": 2.129311475409836,
      "grad_norm": 0.6712715168877653,
      "learning_rate": 9.437401974053928e-05,
      "loss": 12.1255,
      "step": 12177
    },
    {
      "epoch": 2.1294863387978142,
      "grad_norm": 0.6284482659772669,
      "learning_rate": 9.435988133910506e-05,
      "loss": 11.7899,
      "step": 12178
    },
    {
      "epoch": 2.129661202185792,
      "grad_norm": 0.7005703459740893,
      "learning_rate": 9.434574305077257e-05,
      "loss": 11.9837,
      "step": 12179
    },
    {
      "epoch": 2.1298360655737705,
      "grad_norm": 0.709171163336269,
      "learning_rate": 9.433160487582526e-05,
      "loss": 11.8505,
      "step": 12180
    },
    {
      "epoch": 2.130010928961749,
      "grad_norm": 0.6890824139389307,
      "learning_rate": 9.431746681454667e-05,
      "loss": 11.6302,
      "step": 12181
    },
    {
      "epoch": 2.1301857923497267,
      "grad_norm": 0.6690667170693355,
      "learning_rate": 9.430332886722032e-05,
      "loss": 12.1823,
      "step": 12182
    },
    {
      "epoch": 2.130360655737705,
      "grad_norm": 0.6835656526597772,
      "learning_rate": 9.428919103412975e-05,
      "loss": 11.9573,
      "step": 12183
    },
    {
      "epoch": 2.130535519125683,
      "grad_norm": 0.7061470708943005,
      "learning_rate": 9.427505331555842e-05,
      "loss": 12.0512,
      "step": 12184
    },
    {
      "epoch": 2.1307103825136613,
      "grad_norm": 0.6415989175572638,
      "learning_rate": 9.426091571178984e-05,
      "loss": 11.868,
      "step": 12185
    },
    {
      "epoch": 2.130885245901639,
      "grad_norm": 0.6792888196886223,
      "learning_rate": 9.424677822310749e-05,
      "loss": 11.9348,
      "step": 12186
    },
    {
      "epoch": 2.1310601092896175,
      "grad_norm": 0.5812997010708003,
      "learning_rate": 9.423264084979493e-05,
      "loss": 11.6391,
      "step": 12187
    },
    {
      "epoch": 2.1312349726775954,
      "grad_norm": 0.7284554089623956,
      "learning_rate": 9.421850359213566e-05,
      "loss": 11.8575,
      "step": 12188
    },
    {
      "epoch": 2.1314098360655738,
      "grad_norm": 0.6715229707167247,
      "learning_rate": 9.420436645041311e-05,
      "loss": 11.8648,
      "step": 12189
    },
    {
      "epoch": 2.131584699453552,
      "grad_norm": 0.6547774141830482,
      "learning_rate": 9.41902294249108e-05,
      "loss": 11.8647,
      "step": 12190
    },
    {
      "epoch": 2.13175956284153,
      "grad_norm": 0.6826092682669168,
      "learning_rate": 9.417609251591222e-05,
      "loss": 11.9432,
      "step": 12191
    },
    {
      "epoch": 2.1319344262295084,
      "grad_norm": 0.6019276994150261,
      "learning_rate": 9.416195572370093e-05,
      "loss": 12.0985,
      "step": 12192
    },
    {
      "epoch": 2.1321092896174862,
      "grad_norm": 0.6938321933770296,
      "learning_rate": 9.41478190485603e-05,
      "loss": 11.8694,
      "step": 12193
    },
    {
      "epoch": 2.1322841530054646,
      "grad_norm": 0.6512350980003583,
      "learning_rate": 9.413368249077387e-05,
      "loss": 11.6158,
      "step": 12194
    },
    {
      "epoch": 2.1324590163934425,
      "grad_norm": 0.6607005925828253,
      "learning_rate": 9.411954605062512e-05,
      "loss": 12.0359,
      "step": 12195
    },
    {
      "epoch": 2.132633879781421,
      "grad_norm": 0.6872054327012731,
      "learning_rate": 9.410540972839753e-05,
      "loss": 11.9333,
      "step": 12196
    },
    {
      "epoch": 2.1328087431693987,
      "grad_norm": 0.6555896654676824,
      "learning_rate": 9.40912735243746e-05,
      "loss": 11.8862,
      "step": 12197
    },
    {
      "epoch": 2.132983606557377,
      "grad_norm": 0.6181486232359737,
      "learning_rate": 9.407713743883976e-05,
      "loss": 11.6902,
      "step": 12198
    },
    {
      "epoch": 2.1331584699453554,
      "grad_norm": 0.7805248260040867,
      "learning_rate": 9.40630014720765e-05,
      "loss": 11.9902,
      "step": 12199
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 0.626476946994905,
      "learning_rate": 9.404886562436825e-05,
      "loss": 11.9114,
      "step": 12200
    },
    {
      "epoch": 2.1335081967213116,
      "grad_norm": 0.7453803947128389,
      "learning_rate": 9.403472989599857e-05,
      "loss": 12.0428,
      "step": 12201
    },
    {
      "epoch": 2.1336830601092895,
      "grad_norm": 0.639863438833595,
      "learning_rate": 9.402059428725085e-05,
      "loss": 12.0815,
      "step": 12202
    },
    {
      "epoch": 2.133857923497268,
      "grad_norm": 0.6906639370731695,
      "learning_rate": 9.400645879840859e-05,
      "loss": 12.155,
      "step": 12203
    },
    {
      "epoch": 2.134032786885246,
      "grad_norm": 0.674264888058854,
      "learning_rate": 9.399232342975523e-05,
      "loss": 11.7304,
      "step": 12204
    },
    {
      "epoch": 2.134207650273224,
      "grad_norm": 0.6363827495378098,
      "learning_rate": 9.397818818157418e-05,
      "loss": 11.9695,
      "step": 12205
    },
    {
      "epoch": 2.134382513661202,
      "grad_norm": 0.7188905756588271,
      "learning_rate": 9.396405305414902e-05,
      "loss": 11.3607,
      "step": 12206
    },
    {
      "epoch": 2.1345573770491804,
      "grad_norm": 0.6942895979348872,
      "learning_rate": 9.394991804776309e-05,
      "loss": 11.8067,
      "step": 12207
    },
    {
      "epoch": 2.1347322404371587,
      "grad_norm": 0.7194618830360923,
      "learning_rate": 9.393578316269991e-05,
      "loss": 12.0729,
      "step": 12208
    },
    {
      "epoch": 2.1349071038251366,
      "grad_norm": 0.7537784418370588,
      "learning_rate": 9.392164839924286e-05,
      "loss": 11.8914,
      "step": 12209
    },
    {
      "epoch": 2.135081967213115,
      "grad_norm": 0.6628343942173831,
      "learning_rate": 9.390751375767542e-05,
      "loss": 11.7778,
      "step": 12210
    },
    {
      "epoch": 2.135256830601093,
      "grad_norm": 0.6739300098761353,
      "learning_rate": 9.389337923828105e-05,
      "loss": 11.7455,
      "step": 12211
    },
    {
      "epoch": 2.135431693989071,
      "grad_norm": 0.6870424002107157,
      "learning_rate": 9.387924484134318e-05,
      "loss": 11.9903,
      "step": 12212
    },
    {
      "epoch": 2.135606557377049,
      "grad_norm": 0.6492804304622787,
      "learning_rate": 9.386511056714527e-05,
      "loss": 11.7592,
      "step": 12213
    },
    {
      "epoch": 2.1357814207650274,
      "grad_norm": 0.8229511955315975,
      "learning_rate": 9.385097641597065e-05,
      "loss": 11.9549,
      "step": 12214
    },
    {
      "epoch": 2.1359562841530053,
      "grad_norm": 0.7880368530908156,
      "learning_rate": 9.38368423881029e-05,
      "loss": 11.9582,
      "step": 12215
    },
    {
      "epoch": 2.1361311475409837,
      "grad_norm": 0.8676343874580549,
      "learning_rate": 9.382270848382537e-05,
      "loss": 12.0287,
      "step": 12216
    },
    {
      "epoch": 2.1363060109289616,
      "grad_norm": 0.8527854016545164,
      "learning_rate": 9.380857470342152e-05,
      "loss": 12.0859,
      "step": 12217
    },
    {
      "epoch": 2.13648087431694,
      "grad_norm": 0.8913148773647749,
      "learning_rate": 9.379444104717474e-05,
      "loss": 11.931,
      "step": 12218
    },
    {
      "epoch": 2.1366557377049182,
      "grad_norm": 0.7271656282795104,
      "learning_rate": 9.378030751536843e-05,
      "loss": 11.5777,
      "step": 12219
    },
    {
      "epoch": 2.136830601092896,
      "grad_norm": 0.6360143844301336,
      "learning_rate": 9.37661741082861e-05,
      "loss": 11.6636,
      "step": 12220
    },
    {
      "epoch": 2.1370054644808745,
      "grad_norm": 0.765917003962393,
      "learning_rate": 9.375204082621111e-05,
      "loss": 11.9611,
      "step": 12221
    },
    {
      "epoch": 2.1371803278688524,
      "grad_norm": 0.6907079764970804,
      "learning_rate": 9.373790766942691e-05,
      "loss": 11.9613,
      "step": 12222
    },
    {
      "epoch": 2.1373551912568307,
      "grad_norm": 0.619765812954385,
      "learning_rate": 9.372377463821685e-05,
      "loss": 11.6608,
      "step": 12223
    },
    {
      "epoch": 2.1375300546448086,
      "grad_norm": 0.6887002095944528,
      "learning_rate": 9.370964173286438e-05,
      "loss": 12.0053,
      "step": 12224
    },
    {
      "epoch": 2.137704918032787,
      "grad_norm": 0.7201954817316218,
      "learning_rate": 9.369550895365291e-05,
      "loss": 11.9058,
      "step": 12225
    },
    {
      "epoch": 2.137879781420765,
      "grad_norm": 0.6709234914380803,
      "learning_rate": 9.368137630086585e-05,
      "loss": 11.9135,
      "step": 12226
    },
    {
      "epoch": 2.138054644808743,
      "grad_norm": 0.7606376077243984,
      "learning_rate": 9.366724377478662e-05,
      "loss": 12.07,
      "step": 12227
    },
    {
      "epoch": 2.138229508196721,
      "grad_norm": 0.675986162051054,
      "learning_rate": 9.365311137569856e-05,
      "loss": 12.0055,
      "step": 12228
    },
    {
      "epoch": 2.1384043715846994,
      "grad_norm": 0.6570807947306284,
      "learning_rate": 9.36389791038851e-05,
      "loss": 11.8772,
      "step": 12229
    },
    {
      "epoch": 2.1385792349726778,
      "grad_norm": 0.8066993501334477,
      "learning_rate": 9.362484695962966e-05,
      "loss": 12.1751,
      "step": 12230
    },
    {
      "epoch": 2.1387540983606557,
      "grad_norm": 0.6364364526668229,
      "learning_rate": 9.361071494321564e-05,
      "loss": 12.0673,
      "step": 12231
    },
    {
      "epoch": 2.138928961748634,
      "grad_norm": 0.7076443482947058,
      "learning_rate": 9.359658305492636e-05,
      "loss": 11.9353,
      "step": 12232
    },
    {
      "epoch": 2.139103825136612,
      "grad_norm": 0.6659732000601282,
      "learning_rate": 9.358245129504527e-05,
      "loss": 11.6708,
      "step": 12233
    },
    {
      "epoch": 2.1392786885245902,
      "grad_norm": 0.7386837926301744,
      "learning_rate": 9.356831966385571e-05,
      "loss": 12.1499,
      "step": 12234
    },
    {
      "epoch": 2.139453551912568,
      "grad_norm": 0.7517445236891257,
      "learning_rate": 9.355418816164111e-05,
      "loss": 11.953,
      "step": 12235
    },
    {
      "epoch": 2.1396284153005465,
      "grad_norm": 0.6489067939885628,
      "learning_rate": 9.354005678868485e-05,
      "loss": 12.0613,
      "step": 12236
    },
    {
      "epoch": 2.1398032786885244,
      "grad_norm": 0.688448627715474,
      "learning_rate": 9.352592554527025e-05,
      "loss": 11.8914,
      "step": 12237
    },
    {
      "epoch": 2.1399781420765027,
      "grad_norm": 0.7042652500568541,
      "learning_rate": 9.351179443168072e-05,
      "loss": 11.9216,
      "step": 12238
    },
    {
      "epoch": 2.140153005464481,
      "grad_norm": 0.6221633009878631,
      "learning_rate": 9.349766344819965e-05,
      "loss": 11.954,
      "step": 12239
    },
    {
      "epoch": 2.140327868852459,
      "grad_norm": 0.7419137758567478,
      "learning_rate": 9.34835325951104e-05,
      "loss": 11.6789,
      "step": 12240
    },
    {
      "epoch": 2.1405027322404373,
      "grad_norm": 0.644055405095753,
      "learning_rate": 9.346940187269633e-05,
      "loss": 12.0404,
      "step": 12241
    },
    {
      "epoch": 2.140677595628415,
      "grad_norm": 0.6145517257408157,
      "learning_rate": 9.345527128124079e-05,
      "loss": 12.065,
      "step": 12242
    },
    {
      "epoch": 2.1408524590163935,
      "grad_norm": 0.6311045920590828,
      "learning_rate": 9.344114082102712e-05,
      "loss": 11.874,
      "step": 12243
    },
    {
      "epoch": 2.1410273224043714,
      "grad_norm": 0.6532866898047806,
      "learning_rate": 9.342701049233877e-05,
      "loss": 11.7807,
      "step": 12244
    },
    {
      "epoch": 2.1412021857923498,
      "grad_norm": 0.6428860926728028,
      "learning_rate": 9.341288029545905e-05,
      "loss": 12.0537,
      "step": 12245
    },
    {
      "epoch": 2.1413770491803277,
      "grad_norm": 0.7371655018114788,
      "learning_rate": 9.339875023067128e-05,
      "loss": 11.8336,
      "step": 12246
    },
    {
      "epoch": 2.141551912568306,
      "grad_norm": 0.6227147402500575,
      "learning_rate": 9.338462029825886e-05,
      "loss": 12.0695,
      "step": 12247
    },
    {
      "epoch": 2.1417267759562844,
      "grad_norm": 0.6650179451317201,
      "learning_rate": 9.337049049850503e-05,
      "loss": 11.5277,
      "step": 12248
    },
    {
      "epoch": 2.1419016393442623,
      "grad_norm": 0.7234340711921042,
      "learning_rate": 9.33563608316933e-05,
      "loss": 11.9695,
      "step": 12249
    },
    {
      "epoch": 2.1420765027322406,
      "grad_norm": 0.6156194700589571,
      "learning_rate": 9.334223129810693e-05,
      "loss": 11.8848,
      "step": 12250
    },
    {
      "epoch": 2.1422513661202185,
      "grad_norm": 0.7100411846872144,
      "learning_rate": 9.332810189802925e-05,
      "loss": 12.0366,
      "step": 12251
    },
    {
      "epoch": 2.142426229508197,
      "grad_norm": 0.6911005567491109,
      "learning_rate": 9.331397263174364e-05,
      "loss": 12.0669,
      "step": 12252
    },
    {
      "epoch": 2.1426010928961747,
      "grad_norm": 0.6971186820640689,
      "learning_rate": 9.329984349953334e-05,
      "loss": 12.2692,
      "step": 12253
    },
    {
      "epoch": 2.142775956284153,
      "grad_norm": 0.768599740501114,
      "learning_rate": 9.32857145016818e-05,
      "loss": 11.7871,
      "step": 12254
    },
    {
      "epoch": 2.142950819672131,
      "grad_norm": 0.5955485039715475,
      "learning_rate": 9.32715856384723e-05,
      "loss": 12.0746,
      "step": 12255
    },
    {
      "epoch": 2.1431256830601093,
      "grad_norm": 0.8081034507066325,
      "learning_rate": 9.325745691018818e-05,
      "loss": 11.9787,
      "step": 12256
    },
    {
      "epoch": 2.1433005464480877,
      "grad_norm": 0.6243679902974143,
      "learning_rate": 9.324332831711272e-05,
      "loss": 11.8421,
      "step": 12257
    },
    {
      "epoch": 2.1434754098360655,
      "grad_norm": 0.6426356643296002,
      "learning_rate": 9.322919985952926e-05,
      "loss": 11.9776,
      "step": 12258
    },
    {
      "epoch": 2.143650273224044,
      "grad_norm": 0.6569102804808825,
      "learning_rate": 9.321507153772115e-05,
      "loss": 11.9645,
      "step": 12259
    },
    {
      "epoch": 2.143825136612022,
      "grad_norm": 0.6589750946609195,
      "learning_rate": 9.32009433519717e-05,
      "loss": 12.0152,
      "step": 12260
    },
    {
      "epoch": 2.144,
      "grad_norm": 0.7292313476676537,
      "learning_rate": 9.318681530256423e-05,
      "loss": 12.0991,
      "step": 12261
    },
    {
      "epoch": 2.144174863387978,
      "grad_norm": 0.6454724882970038,
      "learning_rate": 9.317268738978196e-05,
      "loss": 11.9475,
      "step": 12262
    },
    {
      "epoch": 2.1443497267759564,
      "grad_norm": 0.6847743240574943,
      "learning_rate": 9.315855961390833e-05,
      "loss": 11.6719,
      "step": 12263
    },
    {
      "epoch": 2.1445245901639343,
      "grad_norm": 0.6906839218889567,
      "learning_rate": 9.314443197522657e-05,
      "loss": 12.1862,
      "step": 12264
    },
    {
      "epoch": 2.1446994535519126,
      "grad_norm": 0.7727383713743533,
      "learning_rate": 9.313030447402002e-05,
      "loss": 11.9082,
      "step": 12265
    },
    {
      "epoch": 2.144874316939891,
      "grad_norm": 0.6921625157345016,
      "learning_rate": 9.311617711057194e-05,
      "loss": 11.9448,
      "step": 12266
    },
    {
      "epoch": 2.145049180327869,
      "grad_norm": 0.6412404756319786,
      "learning_rate": 9.310204988516562e-05,
      "loss": 12.12,
      "step": 12267
    },
    {
      "epoch": 2.145224043715847,
      "grad_norm": 0.6683048835827766,
      "learning_rate": 9.30879227980844e-05,
      "loss": 11.8767,
      "step": 12268
    },
    {
      "epoch": 2.145398907103825,
      "grad_norm": 0.7911327576666192,
      "learning_rate": 9.307379584961156e-05,
      "loss": 11.9621,
      "step": 12269
    },
    {
      "epoch": 2.1455737704918034,
      "grad_norm": 0.6299727618920512,
      "learning_rate": 9.30596690400304e-05,
      "loss": 12.1063,
      "step": 12270
    },
    {
      "epoch": 2.1457486338797813,
      "grad_norm": 0.6789646702889819,
      "learning_rate": 9.304554236962415e-05,
      "loss": 11.9897,
      "step": 12271
    },
    {
      "epoch": 2.1459234972677597,
      "grad_norm": 0.6209365528777552,
      "learning_rate": 9.303141583867611e-05,
      "loss": 11.7415,
      "step": 12272
    },
    {
      "epoch": 2.1460983606557376,
      "grad_norm": 0.620062023595651,
      "learning_rate": 9.301728944746961e-05,
      "loss": 11.7774,
      "step": 12273
    },
    {
      "epoch": 2.146273224043716,
      "grad_norm": 0.6736729705641114,
      "learning_rate": 9.300316319628791e-05,
      "loss": 12.087,
      "step": 12274
    },
    {
      "epoch": 2.146448087431694,
      "grad_norm": 0.7004109122147412,
      "learning_rate": 9.298903708541425e-05,
      "loss": 12.0152,
      "step": 12275
    },
    {
      "epoch": 2.146622950819672,
      "grad_norm": 0.6441330831856124,
      "learning_rate": 9.297491111513193e-05,
      "loss": 11.9202,
      "step": 12276
    },
    {
      "epoch": 2.1467978142076505,
      "grad_norm": 0.6613843685540082,
      "learning_rate": 9.296078528572419e-05,
      "loss": 12.0283,
      "step": 12277
    },
    {
      "epoch": 2.1469726775956284,
      "grad_norm": 0.6649092213267801,
      "learning_rate": 9.294665959747434e-05,
      "loss": 11.8855,
      "step": 12278
    },
    {
      "epoch": 2.1471475409836067,
      "grad_norm": 0.6737129419235103,
      "learning_rate": 9.293253405066563e-05,
      "loss": 12.1513,
      "step": 12279
    },
    {
      "epoch": 2.1473224043715846,
      "grad_norm": 0.7039730273784115,
      "learning_rate": 9.29184086455813e-05,
      "loss": 11.7628,
      "step": 12280
    },
    {
      "epoch": 2.147497267759563,
      "grad_norm": 0.6593337187643502,
      "learning_rate": 9.290428338250461e-05,
      "loss": 11.9685,
      "step": 12281
    },
    {
      "epoch": 2.147672131147541,
      "grad_norm": 0.6544999191688478,
      "learning_rate": 9.289015826171883e-05,
      "loss": 11.6649,
      "step": 12282
    },
    {
      "epoch": 2.147846994535519,
      "grad_norm": 0.6578610043977388,
      "learning_rate": 9.287603328350723e-05,
      "loss": 11.9312,
      "step": 12283
    },
    {
      "epoch": 2.148021857923497,
      "grad_norm": 0.7216784594596056,
      "learning_rate": 9.286190844815301e-05,
      "loss": 12.124,
      "step": 12284
    },
    {
      "epoch": 2.1481967213114754,
      "grad_norm": 0.6608887237665221,
      "learning_rate": 9.284778375593945e-05,
      "loss": 11.7541,
      "step": 12285
    },
    {
      "epoch": 2.1483715846994533,
      "grad_norm": 0.701680636832786,
      "learning_rate": 9.283365920714977e-05,
      "loss": 12.1042,
      "step": 12286
    },
    {
      "epoch": 2.1485464480874317,
      "grad_norm": 0.8589912453598453,
      "learning_rate": 9.281953480206725e-05,
      "loss": 11.8837,
      "step": 12287
    },
    {
      "epoch": 2.14872131147541,
      "grad_norm": 0.6719003110223553,
      "learning_rate": 9.28054105409751e-05,
      "loss": 11.7073,
      "step": 12288
    },
    {
      "epoch": 2.148896174863388,
      "grad_norm": 0.8767214945131251,
      "learning_rate": 9.279128642415657e-05,
      "loss": 11.7452,
      "step": 12289
    },
    {
      "epoch": 2.1490710382513663,
      "grad_norm": 0.8256177414636328,
      "learning_rate": 9.277716245189486e-05,
      "loss": 12.2222,
      "step": 12290
    },
    {
      "epoch": 2.149245901639344,
      "grad_norm": 0.7264020760866446,
      "learning_rate": 9.276303862447321e-05,
      "loss": 12.1328,
      "step": 12291
    },
    {
      "epoch": 2.1494207650273225,
      "grad_norm": 0.7962939901186606,
      "learning_rate": 9.27489149421749e-05,
      "loss": 11.6891,
      "step": 12292
    },
    {
      "epoch": 2.1495956284153004,
      "grad_norm": 0.6018742006011171,
      "learning_rate": 9.273479140528307e-05,
      "loss": 11.7313,
      "step": 12293
    },
    {
      "epoch": 2.1497704918032787,
      "grad_norm": 0.6929998994159133,
      "learning_rate": 9.272066801408099e-05,
      "loss": 11.8626,
      "step": 12294
    },
    {
      "epoch": 2.1499453551912566,
      "grad_norm": 0.738514200243885,
      "learning_rate": 9.270654476885189e-05,
      "loss": 11.9815,
      "step": 12295
    },
    {
      "epoch": 2.150120218579235,
      "grad_norm": 0.730637481482677,
      "learning_rate": 9.269242166987889e-05,
      "loss": 11.6916,
      "step": 12296
    },
    {
      "epoch": 2.1502950819672133,
      "grad_norm": 0.7079599794786807,
      "learning_rate": 9.267829871744536e-05,
      "loss": 11.8835,
      "step": 12297
    },
    {
      "epoch": 2.150469945355191,
      "grad_norm": 0.6399506437670013,
      "learning_rate": 9.266417591183437e-05,
      "loss": 12.0097,
      "step": 12298
    },
    {
      "epoch": 2.1506448087431695,
      "grad_norm": 0.6695059264390145,
      "learning_rate": 9.26500532533292e-05,
      "loss": 11.9683,
      "step": 12299
    },
    {
      "epoch": 2.1508196721311474,
      "grad_norm": 0.6341010133768569,
      "learning_rate": 9.263593074221303e-05,
      "loss": 12.0433,
      "step": 12300
    },
    {
      "epoch": 2.150994535519126,
      "grad_norm": 0.6563124700119334,
      "learning_rate": 9.2621808378769e-05,
      "loss": 12.0413,
      "step": 12301
    },
    {
      "epoch": 2.1511693989071037,
      "grad_norm": 0.6608831135990082,
      "learning_rate": 9.260768616328045e-05,
      "loss": 11.9539,
      "step": 12302
    },
    {
      "epoch": 2.151344262295082,
      "grad_norm": 0.6569527878506991,
      "learning_rate": 9.259356409603045e-05,
      "loss": 11.85,
      "step": 12303
    },
    {
      "epoch": 2.15151912568306,
      "grad_norm": 0.6374253286755356,
      "learning_rate": 9.257944217730226e-05,
      "loss": 11.8887,
      "step": 12304
    },
    {
      "epoch": 2.1516939890710383,
      "grad_norm": 0.7903473699755511,
      "learning_rate": 9.2565320407379e-05,
      "loss": 11.988,
      "step": 12305
    },
    {
      "epoch": 2.1518688524590166,
      "grad_norm": 0.6141623630692677,
      "learning_rate": 9.25511987865439e-05,
      "loss": 12.1044,
      "step": 12306
    },
    {
      "epoch": 2.1520437158469945,
      "grad_norm": 0.6021896715948426,
      "learning_rate": 9.253707731508013e-05,
      "loss": 11.9849,
      "step": 12307
    },
    {
      "epoch": 2.152218579234973,
      "grad_norm": 0.6718049464038836,
      "learning_rate": 9.25229559932709e-05,
      "loss": 12.0929,
      "step": 12308
    },
    {
      "epoch": 2.1523934426229507,
      "grad_norm": 0.6574788001451192,
      "learning_rate": 9.250883482139936e-05,
      "loss": 11.5924,
      "step": 12309
    },
    {
      "epoch": 2.152568306010929,
      "grad_norm": 0.6545616270804526,
      "learning_rate": 9.249471379974862e-05,
      "loss": 11.8893,
      "step": 12310
    },
    {
      "epoch": 2.152743169398907,
      "grad_norm": 0.6880144998280707,
      "learning_rate": 9.248059292860198e-05,
      "loss": 11.9644,
      "step": 12311
    },
    {
      "epoch": 2.1529180327868853,
      "grad_norm": 0.6661856396803946,
      "learning_rate": 9.246647220824252e-05,
      "loss": 12.1243,
      "step": 12312
    },
    {
      "epoch": 2.153092896174863,
      "grad_norm": 0.6139288746035657,
      "learning_rate": 9.245235163895344e-05,
      "loss": 11.6082,
      "step": 12313
    },
    {
      "epoch": 2.1532677595628416,
      "grad_norm": 0.6592056184584402,
      "learning_rate": 9.243823122101788e-05,
      "loss": 12.2126,
      "step": 12314
    },
    {
      "epoch": 2.15344262295082,
      "grad_norm": 0.7189223067252671,
      "learning_rate": 9.242411095471897e-05,
      "loss": 11.8384,
      "step": 12315
    },
    {
      "epoch": 2.153617486338798,
      "grad_norm": 0.7368051979121034,
      "learning_rate": 9.240999084033991e-05,
      "loss": 11.8866,
      "step": 12316
    },
    {
      "epoch": 2.153792349726776,
      "grad_norm": 0.6766892408710755,
      "learning_rate": 9.239587087816385e-05,
      "loss": 11.7958,
      "step": 12317
    },
    {
      "epoch": 2.153967213114754,
      "grad_norm": 0.6506230135727653,
      "learning_rate": 9.238175106847394e-05,
      "loss": 11.9257,
      "step": 12318
    },
    {
      "epoch": 2.1541420765027324,
      "grad_norm": 0.7225264729012995,
      "learning_rate": 9.23676314115533e-05,
      "loss": 11.9865,
      "step": 12319
    },
    {
      "epoch": 2.1543169398907103,
      "grad_norm": 0.6735906247111945,
      "learning_rate": 9.235351190768504e-05,
      "loss": 12.0104,
      "step": 12320
    },
    {
      "epoch": 2.1544918032786886,
      "grad_norm": 0.6309094439078566,
      "learning_rate": 9.233939255715239e-05,
      "loss": 11.9385,
      "step": 12321
    },
    {
      "epoch": 2.1546666666666665,
      "grad_norm": 0.6788943651921506,
      "learning_rate": 9.232527336023845e-05,
      "loss": 12.0815,
      "step": 12322
    },
    {
      "epoch": 2.154841530054645,
      "grad_norm": 0.6723915539071303,
      "learning_rate": 9.231115431722632e-05,
      "loss": 11.9753,
      "step": 12323
    },
    {
      "epoch": 2.1550163934426227,
      "grad_norm": 0.733807380484441,
      "learning_rate": 9.229703542839917e-05,
      "loss": 11.8622,
      "step": 12324
    },
    {
      "epoch": 2.155191256830601,
      "grad_norm": 0.7190006361060871,
      "learning_rate": 9.228291669404007e-05,
      "loss": 11.5986,
      "step": 12325
    },
    {
      "epoch": 2.1553661202185794,
      "grad_norm": 0.6956182718167512,
      "learning_rate": 9.226879811443221e-05,
      "loss": 12.0492,
      "step": 12326
    },
    {
      "epoch": 2.1555409836065573,
      "grad_norm": 0.7381406951781911,
      "learning_rate": 9.225467968985872e-05,
      "loss": 12.0715,
      "step": 12327
    },
    {
      "epoch": 2.1557158469945357,
      "grad_norm": 0.6955626580910802,
      "learning_rate": 9.224056142060266e-05,
      "loss": 11.6431,
      "step": 12328
    },
    {
      "epoch": 2.1558907103825136,
      "grad_norm": 0.6574905258467216,
      "learning_rate": 9.222644330694716e-05,
      "loss": 11.6772,
      "step": 12329
    },
    {
      "epoch": 2.156065573770492,
      "grad_norm": 0.6907775100553983,
      "learning_rate": 9.221232534917533e-05,
      "loss": 11.549,
      "step": 12330
    },
    {
      "epoch": 2.15624043715847,
      "grad_norm": 0.7171886077755539,
      "learning_rate": 9.219820754757032e-05,
      "loss": 12.1646,
      "step": 12331
    },
    {
      "epoch": 2.156415300546448,
      "grad_norm": 0.725201797945081,
      "learning_rate": 9.218408990241517e-05,
      "loss": 11.9488,
      "step": 12332
    },
    {
      "epoch": 2.156590163934426,
      "grad_norm": 0.6711080237308313,
      "learning_rate": 9.216997241399303e-05,
      "loss": 11.9036,
      "step": 12333
    },
    {
      "epoch": 2.1567650273224044,
      "grad_norm": 0.6160859247905545,
      "learning_rate": 9.215585508258696e-05,
      "loss": 11.8738,
      "step": 12334
    },
    {
      "epoch": 2.1569398907103823,
      "grad_norm": 0.6736315349289886,
      "learning_rate": 9.214173790848011e-05,
      "loss": 12.1199,
      "step": 12335
    },
    {
      "epoch": 2.1571147540983606,
      "grad_norm": 0.6330320352229647,
      "learning_rate": 9.212762089195556e-05,
      "loss": 12.0937,
      "step": 12336
    },
    {
      "epoch": 2.157289617486339,
      "grad_norm": 0.7027486637950372,
      "learning_rate": 9.211350403329635e-05,
      "loss": 11.9431,
      "step": 12337
    },
    {
      "epoch": 2.157464480874317,
      "grad_norm": 0.6762793625898845,
      "learning_rate": 9.209938733278561e-05,
      "loss": 12.0325,
      "step": 12338
    },
    {
      "epoch": 2.157639344262295,
      "grad_norm": 0.6936412383830932,
      "learning_rate": 9.20852707907064e-05,
      "loss": 11.8056,
      "step": 12339
    },
    {
      "epoch": 2.157814207650273,
      "grad_norm": 0.6436936731330015,
      "learning_rate": 9.207115440734183e-05,
      "loss": 12.0861,
      "step": 12340
    },
    {
      "epoch": 2.1579890710382514,
      "grad_norm": 0.6910200132483921,
      "learning_rate": 9.205703818297495e-05,
      "loss": 12.094,
      "step": 12341
    },
    {
      "epoch": 2.1581639344262293,
      "grad_norm": 0.6908863817092825,
      "learning_rate": 9.204292211788884e-05,
      "loss": 12.1656,
      "step": 12342
    },
    {
      "epoch": 2.1583387978142077,
      "grad_norm": 0.6074078932706108,
      "learning_rate": 9.202880621236661e-05,
      "loss": 11.8259,
      "step": 12343
    },
    {
      "epoch": 2.1585136612021856,
      "grad_norm": 0.6420289205417209,
      "learning_rate": 9.20146904666912e-05,
      "loss": 11.9445,
      "step": 12344
    },
    {
      "epoch": 2.158688524590164,
      "grad_norm": 0.6818259652018133,
      "learning_rate": 9.200057488114585e-05,
      "loss": 11.8934,
      "step": 12345
    },
    {
      "epoch": 2.1588633879781423,
      "grad_norm": 0.7246582941228797,
      "learning_rate": 9.19864594560135e-05,
      "loss": 12.1338,
      "step": 12346
    },
    {
      "epoch": 2.15903825136612,
      "grad_norm": 0.6797123735881224,
      "learning_rate": 9.197234419157725e-05,
      "loss": 12.0261,
      "step": 12347
    },
    {
      "epoch": 2.1592131147540985,
      "grad_norm": 0.642098624785695,
      "learning_rate": 9.195822908812017e-05,
      "loss": 12.1965,
      "step": 12348
    },
    {
      "epoch": 2.1593879781420764,
      "grad_norm": 0.6690558956384556,
      "learning_rate": 9.194411414592524e-05,
      "loss": 12.1505,
      "step": 12349
    },
    {
      "epoch": 2.1595628415300547,
      "grad_norm": 0.7584020600108693,
      "learning_rate": 9.192999936527557e-05,
      "loss": 12.1123,
      "step": 12350
    },
    {
      "epoch": 2.1597377049180326,
      "grad_norm": 0.6001919254676127,
      "learning_rate": 9.19158847464542e-05,
      "loss": 12.0159,
      "step": 12351
    },
    {
      "epoch": 2.159912568306011,
      "grad_norm": 0.6629223463383743,
      "learning_rate": 9.19017702897442e-05,
      "loss": 11.8986,
      "step": 12352
    },
    {
      "epoch": 2.160087431693989,
      "grad_norm": 0.6952774738283126,
      "learning_rate": 9.188765599542853e-05,
      "loss": 11.6081,
      "step": 12353
    },
    {
      "epoch": 2.160262295081967,
      "grad_norm": 0.6336150642832288,
      "learning_rate": 9.187354186379025e-05,
      "loss": 11.9247,
      "step": 12354
    },
    {
      "epoch": 2.1604371584699456,
      "grad_norm": 0.694301428290758,
      "learning_rate": 9.185942789511243e-05,
      "loss": 11.8307,
      "step": 12355
    },
    {
      "epoch": 2.1606120218579234,
      "grad_norm": 0.6720380282606492,
      "learning_rate": 9.184531408967807e-05,
      "loss": 11.8027,
      "step": 12356
    },
    {
      "epoch": 2.160786885245902,
      "grad_norm": 0.6255925439342257,
      "learning_rate": 9.183120044777022e-05,
      "loss": 11.9201,
      "step": 12357
    },
    {
      "epoch": 2.1609617486338797,
      "grad_norm": 0.7710055105413156,
      "learning_rate": 9.181708696967184e-05,
      "loss": 12.0696,
      "step": 12358
    },
    {
      "epoch": 2.161136612021858,
      "grad_norm": 0.622468379647245,
      "learning_rate": 9.180297365566602e-05,
      "loss": 11.9069,
      "step": 12359
    },
    {
      "epoch": 2.161311475409836,
      "grad_norm": 0.6748955569695483,
      "learning_rate": 9.178886050603574e-05,
      "loss": 11.9401,
      "step": 12360
    },
    {
      "epoch": 2.1614863387978143,
      "grad_norm": 0.6669997367357053,
      "learning_rate": 9.177474752106404e-05,
      "loss": 11.9531,
      "step": 12361
    },
    {
      "epoch": 2.161661202185792,
      "grad_norm": 0.5954414593371989,
      "learning_rate": 9.176063470103389e-05,
      "loss": 11.7271,
      "step": 12362
    },
    {
      "epoch": 2.1618360655737705,
      "grad_norm": 0.6911779502007127,
      "learning_rate": 9.17465220462283e-05,
      "loss": 11.9037,
      "step": 12363
    },
    {
      "epoch": 2.162010928961749,
      "grad_norm": 0.6488618545804857,
      "learning_rate": 9.173240955693029e-05,
      "loss": 11.9724,
      "step": 12364
    },
    {
      "epoch": 2.1621857923497267,
      "grad_norm": 0.6182597040603951,
      "learning_rate": 9.171829723342286e-05,
      "loss": 11.9469,
      "step": 12365
    },
    {
      "epoch": 2.162360655737705,
      "grad_norm": 0.6715980295426401,
      "learning_rate": 9.170418507598901e-05,
      "loss": 11.8198,
      "step": 12366
    },
    {
      "epoch": 2.162535519125683,
      "grad_norm": 0.6615957516347818,
      "learning_rate": 9.169007308491171e-05,
      "loss": 12.1745,
      "step": 12367
    },
    {
      "epoch": 2.1627103825136613,
      "grad_norm": 0.6049374970079487,
      "learning_rate": 9.167596126047394e-05,
      "loss": 11.9597,
      "step": 12368
    },
    {
      "epoch": 2.162885245901639,
      "grad_norm": 0.6711871911393361,
      "learning_rate": 9.166184960295872e-05,
      "loss": 11.823,
      "step": 12369
    },
    {
      "epoch": 2.1630601092896176,
      "grad_norm": 0.6923187043631215,
      "learning_rate": 9.164773811264904e-05,
      "loss": 11.5231,
      "step": 12370
    },
    {
      "epoch": 2.1632349726775955,
      "grad_norm": 0.6576829040409435,
      "learning_rate": 9.163362678982783e-05,
      "loss": 11.8098,
      "step": 12371
    },
    {
      "epoch": 2.163409836065574,
      "grad_norm": 0.6446705016849561,
      "learning_rate": 9.16195156347781e-05,
      "loss": 11.8514,
      "step": 12372
    },
    {
      "epoch": 2.163584699453552,
      "grad_norm": 0.5875423048177887,
      "learning_rate": 9.160540464778277e-05,
      "loss": 12.0126,
      "step": 12373
    },
    {
      "epoch": 2.16375956284153,
      "grad_norm": 0.6594086769755004,
      "learning_rate": 9.15912938291249e-05,
      "loss": 12.047,
      "step": 12374
    },
    {
      "epoch": 2.1639344262295084,
      "grad_norm": 0.6391515565428026,
      "learning_rate": 9.157718317908743e-05,
      "loss": 12.0134,
      "step": 12375
    },
    {
      "epoch": 2.1641092896174863,
      "grad_norm": 0.658105153174936,
      "learning_rate": 9.156307269795325e-05,
      "loss": 11.9917,
      "step": 12376
    },
    {
      "epoch": 2.1642841530054646,
      "grad_norm": 0.6408984599150283,
      "learning_rate": 9.154896238600539e-05,
      "loss": 12.0515,
      "step": 12377
    },
    {
      "epoch": 2.1644590163934425,
      "grad_norm": 0.5895247827212091,
      "learning_rate": 9.153485224352675e-05,
      "loss": 11.9574,
      "step": 12378
    },
    {
      "epoch": 2.164633879781421,
      "grad_norm": 0.6602571462143825,
      "learning_rate": 9.152074227080036e-05,
      "loss": 11.9037,
      "step": 12379
    },
    {
      "epoch": 2.1648087431693988,
      "grad_norm": 0.7146307054175225,
      "learning_rate": 9.15066324681091e-05,
      "loss": 11.881,
      "step": 12380
    },
    {
      "epoch": 2.164983606557377,
      "grad_norm": 0.6732712714261521,
      "learning_rate": 9.149252283573594e-05,
      "loss": 12.2173,
      "step": 12381
    },
    {
      "epoch": 2.165158469945355,
      "grad_norm": 0.6463100182606829,
      "learning_rate": 9.14784133739638e-05,
      "loss": 11.8241,
      "step": 12382
    },
    {
      "epoch": 2.1653333333333333,
      "grad_norm": 0.6574009199716082,
      "learning_rate": 9.146430408307565e-05,
      "loss": 11.5984,
      "step": 12383
    },
    {
      "epoch": 2.1655081967213117,
      "grad_norm": 0.7045751776288449,
      "learning_rate": 9.145019496335443e-05,
      "loss": 11.9115,
      "step": 12384
    },
    {
      "epoch": 2.1656830601092896,
      "grad_norm": 0.6591047913161834,
      "learning_rate": 9.143608601508305e-05,
      "loss": 11.6815,
      "step": 12385
    },
    {
      "epoch": 2.165857923497268,
      "grad_norm": 0.7623968262804708,
      "learning_rate": 9.142197723854442e-05,
      "loss": 11.9772,
      "step": 12386
    },
    {
      "epoch": 2.166032786885246,
      "grad_norm": 0.6368342879025696,
      "learning_rate": 9.140786863402147e-05,
      "loss": 11.7938,
      "step": 12387
    },
    {
      "epoch": 2.166207650273224,
      "grad_norm": 0.791048553283201,
      "learning_rate": 9.139376020179718e-05,
      "loss": 12.0754,
      "step": 12388
    },
    {
      "epoch": 2.166382513661202,
      "grad_norm": 0.6095441965656856,
      "learning_rate": 9.13796519421544e-05,
      "loss": 11.4591,
      "step": 12389
    },
    {
      "epoch": 2.1665573770491804,
      "grad_norm": 0.6394865897354278,
      "learning_rate": 9.136554385537606e-05,
      "loss": 12.0031,
      "step": 12390
    },
    {
      "epoch": 2.1667322404371583,
      "grad_norm": 0.745614444439484,
      "learning_rate": 9.13514359417451e-05,
      "loss": 11.9457,
      "step": 12391
    },
    {
      "epoch": 2.1669071038251366,
      "grad_norm": 0.7021838071258669,
      "learning_rate": 9.133732820154435e-05,
      "loss": 11.7442,
      "step": 12392
    },
    {
      "epoch": 2.1670819672131145,
      "grad_norm": 0.7177406897490823,
      "learning_rate": 9.132322063505682e-05,
      "loss": 11.8053,
      "step": 12393
    },
    {
      "epoch": 2.167256830601093,
      "grad_norm": 0.6951342341640669,
      "learning_rate": 9.130911324256535e-05,
      "loss": 11.9494,
      "step": 12394
    },
    {
      "epoch": 2.167431693989071,
      "grad_norm": 0.6639773588493285,
      "learning_rate": 9.129500602435284e-05,
      "loss": 12.115,
      "step": 12395
    },
    {
      "epoch": 2.167606557377049,
      "grad_norm": 0.6928216438060201,
      "learning_rate": 9.12808989807022e-05,
      "loss": 11.6929,
      "step": 12396
    },
    {
      "epoch": 2.1677814207650274,
      "grad_norm": 0.6494283674627342,
      "learning_rate": 9.126679211189626e-05,
      "loss": 11.9622,
      "step": 12397
    },
    {
      "epoch": 2.1679562841530053,
      "grad_norm": 0.6828335892965957,
      "learning_rate": 9.125268541821799e-05,
      "loss": 11.6703,
      "step": 12398
    },
    {
      "epoch": 2.1681311475409837,
      "grad_norm": 0.6466394405945747,
      "learning_rate": 9.123857889995022e-05,
      "loss": 11.8361,
      "step": 12399
    },
    {
      "epoch": 2.1683060109289616,
      "grad_norm": 0.7465512830280734,
      "learning_rate": 9.122447255737587e-05,
      "loss": 11.86,
      "step": 12400
    },
    {
      "epoch": 2.16848087431694,
      "grad_norm": 0.6399264772159928,
      "learning_rate": 9.121036639077777e-05,
      "loss": 12.1636,
      "step": 12401
    },
    {
      "epoch": 2.168655737704918,
      "grad_norm": 0.6280062166885083,
      "learning_rate": 9.119626040043879e-05,
      "loss": 12.0287,
      "step": 12402
    },
    {
      "epoch": 2.168830601092896,
      "grad_norm": 0.701928633955586,
      "learning_rate": 9.118215458664185e-05,
      "loss": 12.1841,
      "step": 12403
    },
    {
      "epoch": 2.1690054644808745,
      "grad_norm": 0.6752527837043532,
      "learning_rate": 9.116804894966978e-05,
      "loss": 12.0679,
      "step": 12404
    },
    {
      "epoch": 2.1691803278688524,
      "grad_norm": 0.6126233460400298,
      "learning_rate": 9.115394348980546e-05,
      "loss": 12.1009,
      "step": 12405
    },
    {
      "epoch": 2.1693551912568307,
      "grad_norm": 0.6978602771840763,
      "learning_rate": 9.113983820733169e-05,
      "loss": 11.6989,
      "step": 12406
    },
    {
      "epoch": 2.1695300546448086,
      "grad_norm": 0.6237389105278203,
      "learning_rate": 9.112573310253139e-05,
      "loss": 12.0146,
      "step": 12407
    },
    {
      "epoch": 2.169704918032787,
      "grad_norm": 0.6599653598785018,
      "learning_rate": 9.11116281756874e-05,
      "loss": 11.9599,
      "step": 12408
    },
    {
      "epoch": 2.169879781420765,
      "grad_norm": 0.6549099051396255,
      "learning_rate": 9.109752342708257e-05,
      "loss": 12.0397,
      "step": 12409
    },
    {
      "epoch": 2.170054644808743,
      "grad_norm": 0.7873814945752354,
      "learning_rate": 9.108341885699971e-05,
      "loss": 11.9525,
      "step": 12410
    },
    {
      "epoch": 2.170229508196721,
      "grad_norm": 0.6357614261524975,
      "learning_rate": 9.106931446572167e-05,
      "loss": 11.8996,
      "step": 12411
    },
    {
      "epoch": 2.1704043715846995,
      "grad_norm": 0.5947412939940496,
      "learning_rate": 9.10552102535313e-05,
      "loss": 12.1134,
      "step": 12412
    },
    {
      "epoch": 2.170579234972678,
      "grad_norm": 0.714839108863264,
      "learning_rate": 9.104110622071142e-05,
      "loss": 11.9601,
      "step": 12413
    },
    {
      "epoch": 2.1707540983606557,
      "grad_norm": 0.6761892660964698,
      "learning_rate": 9.102700236754492e-05,
      "loss": 12.1045,
      "step": 12414
    },
    {
      "epoch": 2.170928961748634,
      "grad_norm": 0.7559583092565907,
      "learning_rate": 9.101289869431452e-05,
      "loss": 11.9398,
      "step": 12415
    },
    {
      "epoch": 2.171103825136612,
      "grad_norm": 0.5935338185776924,
      "learning_rate": 9.09987952013031e-05,
      "loss": 11.5896,
      "step": 12416
    },
    {
      "epoch": 2.1712786885245903,
      "grad_norm": 0.7126932839917745,
      "learning_rate": 9.098469188879349e-05,
      "loss": 12.0883,
      "step": 12417
    },
    {
      "epoch": 2.171453551912568,
      "grad_norm": 0.6143651241494148,
      "learning_rate": 9.097058875706848e-05,
      "loss": 11.9602,
      "step": 12418
    },
    {
      "epoch": 2.1716284153005465,
      "grad_norm": 0.610252755706883,
      "learning_rate": 9.09564858064109e-05,
      "loss": 11.8876,
      "step": 12419
    },
    {
      "epoch": 2.1718032786885244,
      "grad_norm": 0.6782533282284753,
      "learning_rate": 9.094238303710354e-05,
      "loss": 11.8417,
      "step": 12420
    },
    {
      "epoch": 2.1719781420765027,
      "grad_norm": 0.6573385134207052,
      "learning_rate": 9.092828044942918e-05,
      "loss": 11.8037,
      "step": 12421
    },
    {
      "epoch": 2.172153005464481,
      "grad_norm": 0.6846053003010191,
      "learning_rate": 9.091417804367069e-05,
      "loss": 12.0358,
      "step": 12422
    },
    {
      "epoch": 2.172327868852459,
      "grad_norm": 0.6670105889240932,
      "learning_rate": 9.090007582011082e-05,
      "loss": 11.7935,
      "step": 12423
    },
    {
      "epoch": 2.1725027322404373,
      "grad_norm": 0.6355190156118953,
      "learning_rate": 9.088597377903238e-05,
      "loss": 12.0062,
      "step": 12424
    },
    {
      "epoch": 2.1726775956284152,
      "grad_norm": 0.7018453612029688,
      "learning_rate": 9.087187192071813e-05,
      "loss": 12.0368,
      "step": 12425
    },
    {
      "epoch": 2.1728524590163936,
      "grad_norm": 0.6580549249367369,
      "learning_rate": 9.085777024545086e-05,
      "loss": 11.8828,
      "step": 12426
    },
    {
      "epoch": 2.1730273224043715,
      "grad_norm": 0.6644681065760814,
      "learning_rate": 9.08436687535134e-05,
      "loss": 12.101,
      "step": 12427
    },
    {
      "epoch": 2.17320218579235,
      "grad_norm": 0.6930606634513966,
      "learning_rate": 9.082956744518849e-05,
      "loss": 11.9454,
      "step": 12428
    },
    {
      "epoch": 2.1733770491803277,
      "grad_norm": 0.7042188453997632,
      "learning_rate": 9.081546632075891e-05,
      "loss": 12.0413,
      "step": 12429
    },
    {
      "epoch": 2.173551912568306,
      "grad_norm": 0.6075810401881482,
      "learning_rate": 9.08013653805074e-05,
      "loss": 12.0642,
      "step": 12430
    },
    {
      "epoch": 2.1737267759562844,
      "grad_norm": 0.6514004166800624,
      "learning_rate": 9.078726462471679e-05,
      "loss": 11.9583,
      "step": 12431
    },
    {
      "epoch": 2.1739016393442623,
      "grad_norm": 0.666583812826492,
      "learning_rate": 9.077316405366981e-05,
      "loss": 11.7744,
      "step": 12432
    },
    {
      "epoch": 2.1740765027322406,
      "grad_norm": 0.663315701428543,
      "learning_rate": 9.075906366764921e-05,
      "loss": 11.9828,
      "step": 12433
    },
    {
      "epoch": 2.1742513661202185,
      "grad_norm": 0.641899860034376,
      "learning_rate": 9.074496346693777e-05,
      "loss": 11.8857,
      "step": 12434
    },
    {
      "epoch": 2.174426229508197,
      "grad_norm": 0.75930587348934,
      "learning_rate": 9.073086345181819e-05,
      "loss": 12.0786,
      "step": 12435
    },
    {
      "epoch": 2.1746010928961748,
      "grad_norm": 0.6665800658567703,
      "learning_rate": 9.071676362257331e-05,
      "loss": 11.9891,
      "step": 12436
    },
    {
      "epoch": 2.174775956284153,
      "grad_norm": 0.7163824338384863,
      "learning_rate": 9.07026639794858e-05,
      "loss": 11.9483,
      "step": 12437
    },
    {
      "epoch": 2.174950819672131,
      "grad_norm": 0.7842393940165764,
      "learning_rate": 9.068856452283842e-05,
      "loss": 12.0939,
      "step": 12438
    },
    {
      "epoch": 2.1751256830601093,
      "grad_norm": 0.715900036302062,
      "learning_rate": 9.067446525291395e-05,
      "loss": 11.7408,
      "step": 12439
    },
    {
      "epoch": 2.1753005464480872,
      "grad_norm": 0.7871202095426206,
      "learning_rate": 9.066036616999499e-05,
      "loss": 11.8296,
      "step": 12440
    },
    {
      "epoch": 2.1754754098360656,
      "grad_norm": 0.6371804315937771,
      "learning_rate": 9.064626727436445e-05,
      "loss": 12.0371,
      "step": 12441
    },
    {
      "epoch": 2.175650273224044,
      "grad_norm": 0.7544560637594528,
      "learning_rate": 9.063216856630494e-05,
      "loss": 12.1245,
      "step": 12442
    },
    {
      "epoch": 2.175825136612022,
      "grad_norm": 0.6901710499438009,
      "learning_rate": 9.061807004609923e-05,
      "loss": 12.043,
      "step": 12443
    },
    {
      "epoch": 2.176,
      "grad_norm": 0.6058743032670684,
      "learning_rate": 9.060397171403003e-05,
      "loss": 11.6513,
      "step": 12444
    },
    {
      "epoch": 2.176174863387978,
      "grad_norm": 0.7303026978351255,
      "learning_rate": 9.058987357038e-05,
      "loss": 11.9229,
      "step": 12445
    },
    {
      "epoch": 2.1763497267759564,
      "grad_norm": 0.702337689304443,
      "learning_rate": 9.057577561543193e-05,
      "loss": 11.8601,
      "step": 12446
    },
    {
      "epoch": 2.1765245901639343,
      "grad_norm": 0.7111172226000004,
      "learning_rate": 9.056167784946848e-05,
      "loss": 11.9027,
      "step": 12447
    },
    {
      "epoch": 2.1766994535519126,
      "grad_norm": 0.640258375492408,
      "learning_rate": 9.05475802727724e-05,
      "loss": 11.7163,
      "step": 12448
    },
    {
      "epoch": 2.1768743169398905,
      "grad_norm": 0.6354584822737837,
      "learning_rate": 9.053348288562634e-05,
      "loss": 11.5203,
      "step": 12449
    },
    {
      "epoch": 2.177049180327869,
      "grad_norm": 0.654253710737197,
      "learning_rate": 9.051938568831298e-05,
      "loss": 11.8181,
      "step": 12450
    },
    {
      "epoch": 2.1772240437158468,
      "grad_norm": 0.6256844262807457,
      "learning_rate": 9.050528868111508e-05,
      "loss": 12.0908,
      "step": 12451
    },
    {
      "epoch": 2.177398907103825,
      "grad_norm": 0.6955689677074922,
      "learning_rate": 9.04911918643153e-05,
      "loss": 12.1117,
      "step": 12452
    },
    {
      "epoch": 2.1775737704918035,
      "grad_norm": 0.642076018738675,
      "learning_rate": 9.047709523819632e-05,
      "loss": 12.0192,
      "step": 12453
    },
    {
      "epoch": 2.1777486338797813,
      "grad_norm": 0.6605069870178296,
      "learning_rate": 9.04629988030408e-05,
      "loss": 11.7574,
      "step": 12454
    },
    {
      "epoch": 2.1779234972677597,
      "grad_norm": 0.6030826319192033,
      "learning_rate": 9.044890255913145e-05,
      "loss": 11.8543,
      "step": 12455
    },
    {
      "epoch": 2.1780983606557376,
      "grad_norm": 0.674731810112305,
      "learning_rate": 9.043480650675092e-05,
      "loss": 11.6882,
      "step": 12456
    },
    {
      "epoch": 2.178273224043716,
      "grad_norm": 0.6236148223704538,
      "learning_rate": 9.042071064618192e-05,
      "loss": 12.0798,
      "step": 12457
    },
    {
      "epoch": 2.178448087431694,
      "grad_norm": 0.6276003964444219,
      "learning_rate": 9.040661497770707e-05,
      "loss": 12.2164,
      "step": 12458
    },
    {
      "epoch": 2.178622950819672,
      "grad_norm": 0.61463472931169,
      "learning_rate": 9.039251950160899e-05,
      "loss": 11.9225,
      "step": 12459
    },
    {
      "epoch": 2.17879781420765,
      "grad_norm": 0.6403640146002676,
      "learning_rate": 9.037842421817045e-05,
      "loss": 12.067,
      "step": 12460
    },
    {
      "epoch": 2.1789726775956284,
      "grad_norm": 0.6485159514272617,
      "learning_rate": 9.036432912767403e-05,
      "loss": 11.7748,
      "step": 12461
    },
    {
      "epoch": 2.1791475409836067,
      "grad_norm": 0.62611249351827,
      "learning_rate": 9.035023423040242e-05,
      "loss": 11.7418,
      "step": 12462
    },
    {
      "epoch": 2.1793224043715846,
      "grad_norm": 0.6266812662980994,
      "learning_rate": 9.033613952663822e-05,
      "loss": 11.7226,
      "step": 12463
    },
    {
      "epoch": 2.179497267759563,
      "grad_norm": 0.6509800082401126,
      "learning_rate": 9.032204501666407e-05,
      "loss": 11.824,
      "step": 12464
    },
    {
      "epoch": 2.179672131147541,
      "grad_norm": 0.6674575854125173,
      "learning_rate": 9.030795070076265e-05,
      "loss": 11.7046,
      "step": 12465
    },
    {
      "epoch": 2.1798469945355192,
      "grad_norm": 0.6818473968602908,
      "learning_rate": 9.029385657921661e-05,
      "loss": 11.8728,
      "step": 12466
    },
    {
      "epoch": 2.180021857923497,
      "grad_norm": 0.7603704632313324,
      "learning_rate": 9.027976265230853e-05,
      "loss": 11.7657,
      "step": 12467
    },
    {
      "epoch": 2.1801967213114755,
      "grad_norm": 0.6270631354591232,
      "learning_rate": 9.026566892032105e-05,
      "loss": 11.7737,
      "step": 12468
    },
    {
      "epoch": 2.1803715846994534,
      "grad_norm": 0.7197075362435128,
      "learning_rate": 9.025157538353677e-05,
      "loss": 12.0231,
      "step": 12469
    },
    {
      "epoch": 2.1805464480874317,
      "grad_norm": 0.6242592660416875,
      "learning_rate": 9.023748204223836e-05,
      "loss": 11.9121,
      "step": 12470
    },
    {
      "epoch": 2.18072131147541,
      "grad_norm": 0.6843733139326598,
      "learning_rate": 9.022338889670842e-05,
      "loss": 12.0224,
      "step": 12471
    },
    {
      "epoch": 2.180896174863388,
      "grad_norm": 0.6252981401039656,
      "learning_rate": 9.020929594722955e-05,
      "loss": 12.1032,
      "step": 12472
    },
    {
      "epoch": 2.1810710382513663,
      "grad_norm": 0.6658775261094764,
      "learning_rate": 9.019520319408435e-05,
      "loss": 11.9972,
      "step": 12473
    },
    {
      "epoch": 2.181245901639344,
      "grad_norm": 0.6310925206978365,
      "learning_rate": 9.01811106375554e-05,
      "loss": 11.9357,
      "step": 12474
    },
    {
      "epoch": 2.1814207650273225,
      "grad_norm": 0.6461281708982457,
      "learning_rate": 9.016701827792538e-05,
      "loss": 11.6606,
      "step": 12475
    },
    {
      "epoch": 2.1815956284153004,
      "grad_norm": 0.6892929915095232,
      "learning_rate": 9.015292611547681e-05,
      "loss": 11.9742,
      "step": 12476
    },
    {
      "epoch": 2.1817704918032788,
      "grad_norm": 0.7669758083724979,
      "learning_rate": 9.01388341504923e-05,
      "loss": 12.0379,
      "step": 12477
    },
    {
      "epoch": 2.1819453551912567,
      "grad_norm": 0.6595185653197684,
      "learning_rate": 9.012474238325443e-05,
      "loss": 11.9605,
      "step": 12478
    },
    {
      "epoch": 2.182120218579235,
      "grad_norm": 0.6758668640511497,
      "learning_rate": 9.011065081404582e-05,
      "loss": 12.1175,
      "step": 12479
    },
    {
      "epoch": 2.1822950819672133,
      "grad_norm": 0.6463842580088254,
      "learning_rate": 9.009655944314905e-05,
      "loss": 11.8542,
      "step": 12480
    },
    {
      "epoch": 2.1824699453551912,
      "grad_norm": 0.6935093871640459,
      "learning_rate": 9.008246827084664e-05,
      "loss": 11.945,
      "step": 12481
    },
    {
      "epoch": 2.1826448087431696,
      "grad_norm": 0.6266908618740543,
      "learning_rate": 9.006837729742119e-05,
      "loss": 11.6703,
      "step": 12482
    },
    {
      "epoch": 2.1828196721311475,
      "grad_norm": 0.6085950121996482,
      "learning_rate": 9.005428652315525e-05,
      "loss": 11.9925,
      "step": 12483
    },
    {
      "epoch": 2.182994535519126,
      "grad_norm": 0.5977222275712897,
      "learning_rate": 9.004019594833145e-05,
      "loss": 11.8451,
      "step": 12484
    },
    {
      "epoch": 2.1831693989071037,
      "grad_norm": 0.6026844542230494,
      "learning_rate": 9.002610557323228e-05,
      "loss": 11.6326,
      "step": 12485
    },
    {
      "epoch": 2.183344262295082,
      "grad_norm": 0.6578375799527075,
      "learning_rate": 9.001201539814031e-05,
      "loss": 11.7491,
      "step": 12486
    },
    {
      "epoch": 2.18351912568306,
      "grad_norm": 0.7082059351273485,
      "learning_rate": 8.999792542333814e-05,
      "loss": 11.6813,
      "step": 12487
    },
    {
      "epoch": 2.1836939890710383,
      "grad_norm": 0.6980840656230557,
      "learning_rate": 8.998383564910819e-05,
      "loss": 11.9462,
      "step": 12488
    },
    {
      "epoch": 2.183868852459016,
      "grad_norm": 0.6210592927882347,
      "learning_rate": 8.996974607573317e-05,
      "loss": 11.867,
      "step": 12489
    },
    {
      "epoch": 2.1840437158469945,
      "grad_norm": 0.6365069532222642,
      "learning_rate": 8.99556567034955e-05,
      "loss": 11.7038,
      "step": 12490
    },
    {
      "epoch": 2.184218579234973,
      "grad_norm": 0.6863438530696725,
      "learning_rate": 8.994156753267777e-05,
      "loss": 11.8983,
      "step": 12491
    },
    {
      "epoch": 2.1843934426229508,
      "grad_norm": 0.7504676698219112,
      "learning_rate": 8.99274785635625e-05,
      "loss": 11.8455,
      "step": 12492
    },
    {
      "epoch": 2.184568306010929,
      "grad_norm": 0.6501449642754813,
      "learning_rate": 8.991338979643217e-05,
      "loss": 12.0068,
      "step": 12493
    },
    {
      "epoch": 2.184743169398907,
      "grad_norm": 0.6935598958141272,
      "learning_rate": 8.989930123156937e-05,
      "loss": 11.8443,
      "step": 12494
    },
    {
      "epoch": 2.1849180327868853,
      "grad_norm": 0.6645233726894013,
      "learning_rate": 8.98852128692566e-05,
      "loss": 11.9079,
      "step": 12495
    },
    {
      "epoch": 2.1850928961748632,
      "grad_norm": 0.6357776445674428,
      "learning_rate": 8.987112470977638e-05,
      "loss": 11.9912,
      "step": 12496
    },
    {
      "epoch": 2.1852677595628416,
      "grad_norm": 0.6487239037171059,
      "learning_rate": 8.985703675341119e-05,
      "loss": 11.7401,
      "step": 12497
    },
    {
      "epoch": 2.1854426229508195,
      "grad_norm": 0.5829108259097452,
      "learning_rate": 8.984294900044355e-05,
      "loss": 11.8436,
      "step": 12498
    },
    {
      "epoch": 2.185617486338798,
      "grad_norm": 0.6731158844915099,
      "learning_rate": 8.982886145115597e-05,
      "loss": 11.7289,
      "step": 12499
    },
    {
      "epoch": 2.185792349726776,
      "grad_norm": 0.6948546985921982,
      "learning_rate": 8.981477410583094e-05,
      "loss": 11.7355,
      "step": 12500
    },
    {
      "epoch": 2.185967213114754,
      "grad_norm": 0.6711621948023545,
      "learning_rate": 8.980068696475101e-05,
      "loss": 12.0738,
      "step": 12501
    },
    {
      "epoch": 2.1861420765027324,
      "grad_norm": 0.6766359426163393,
      "learning_rate": 8.978660002819857e-05,
      "loss": 11.7622,
      "step": 12502
    },
    {
      "epoch": 2.1863169398907103,
      "grad_norm": 0.7230252424463026,
      "learning_rate": 8.977251329645617e-05,
      "loss": 11.7752,
      "step": 12503
    },
    {
      "epoch": 2.1864918032786886,
      "grad_norm": 0.5785303588104943,
      "learning_rate": 8.975842676980629e-05,
      "loss": 12.0766,
      "step": 12504
    },
    {
      "epoch": 2.1866666666666665,
      "grad_norm": 0.7534639924335546,
      "learning_rate": 8.974434044853141e-05,
      "loss": 11.7951,
      "step": 12505
    },
    {
      "epoch": 2.186841530054645,
      "grad_norm": 0.6208155737088896,
      "learning_rate": 8.973025433291399e-05,
      "loss": 12.0079,
      "step": 12506
    },
    {
      "epoch": 2.1870163934426228,
      "grad_norm": 0.6577132465770797,
      "learning_rate": 8.971616842323646e-05,
      "loss": 11.937,
      "step": 12507
    },
    {
      "epoch": 2.187191256830601,
      "grad_norm": 0.6672611243142115,
      "learning_rate": 8.970208271978137e-05,
      "loss": 11.7685,
      "step": 12508
    },
    {
      "epoch": 2.187366120218579,
      "grad_norm": 0.6242819737012648,
      "learning_rate": 8.968799722283114e-05,
      "loss": 11.8238,
      "step": 12509
    },
    {
      "epoch": 2.1875409836065574,
      "grad_norm": 0.6860049511348615,
      "learning_rate": 8.967391193266823e-05,
      "loss": 11.9644,
      "step": 12510
    },
    {
      "epoch": 2.1877158469945357,
      "grad_norm": 0.6335127109906776,
      "learning_rate": 8.965982684957509e-05,
      "loss": 11.9475,
      "step": 12511
    },
    {
      "epoch": 2.1878907103825136,
      "grad_norm": 0.7271336744447076,
      "learning_rate": 8.964574197383413e-05,
      "loss": 12.0027,
      "step": 12512
    },
    {
      "epoch": 2.188065573770492,
      "grad_norm": 0.7665948239269237,
      "learning_rate": 8.963165730572787e-05,
      "loss": 12.1047,
      "step": 12513
    },
    {
      "epoch": 2.18824043715847,
      "grad_norm": 0.7096155361457945,
      "learning_rate": 8.961757284553874e-05,
      "loss": 11.7646,
      "step": 12514
    },
    {
      "epoch": 2.188415300546448,
      "grad_norm": 0.7279273575713296,
      "learning_rate": 8.960348859354913e-05,
      "loss": 12.1355,
      "step": 12515
    },
    {
      "epoch": 2.188590163934426,
      "grad_norm": 0.6286415310400547,
      "learning_rate": 8.958940455004148e-05,
      "loss": 12.0345,
      "step": 12516
    },
    {
      "epoch": 2.1887650273224044,
      "grad_norm": 0.6423113410225934,
      "learning_rate": 8.957532071529821e-05,
      "loss": 11.931,
      "step": 12517
    },
    {
      "epoch": 2.1889398907103823,
      "grad_norm": 0.7197512366670404,
      "learning_rate": 8.956123708960181e-05,
      "loss": 12.0488,
      "step": 12518
    },
    {
      "epoch": 2.1891147540983606,
      "grad_norm": 0.6755607001210039,
      "learning_rate": 8.954715367323468e-05,
      "loss": 12.1134,
      "step": 12519
    },
    {
      "epoch": 2.189289617486339,
      "grad_norm": 0.6438148349077567,
      "learning_rate": 8.953307046647918e-05,
      "loss": 11.7077,
      "step": 12520
    },
    {
      "epoch": 2.189464480874317,
      "grad_norm": 0.704956889395777,
      "learning_rate": 8.951898746961776e-05,
      "loss": 12.0301,
      "step": 12521
    },
    {
      "epoch": 2.1896393442622952,
      "grad_norm": 0.6612645584748231,
      "learning_rate": 8.950490468293279e-05,
      "loss": 11.9537,
      "step": 12522
    },
    {
      "epoch": 2.189814207650273,
      "grad_norm": 0.7102625004693203,
      "learning_rate": 8.949082210670675e-05,
      "loss": 11.6626,
      "step": 12523
    },
    {
      "epoch": 2.1899890710382515,
      "grad_norm": 0.6538477308007755,
      "learning_rate": 8.947673974122198e-05,
      "loss": 11.9304,
      "step": 12524
    },
    {
      "epoch": 2.1901639344262294,
      "grad_norm": 0.7145508200777573,
      "learning_rate": 8.946265758676089e-05,
      "loss": 11.9363,
      "step": 12525
    },
    {
      "epoch": 2.1903387978142077,
      "grad_norm": 0.6176896402381793,
      "learning_rate": 8.944857564360585e-05,
      "loss": 11.7842,
      "step": 12526
    },
    {
      "epoch": 2.1905136612021856,
      "grad_norm": 0.6211635704446127,
      "learning_rate": 8.943449391203928e-05,
      "loss": 12.0403,
      "step": 12527
    },
    {
      "epoch": 2.190688524590164,
      "grad_norm": 0.6373416494970955,
      "learning_rate": 8.942041239234356e-05,
      "loss": 12.0564,
      "step": 12528
    },
    {
      "epoch": 2.1908633879781423,
      "grad_norm": 0.7059922210820557,
      "learning_rate": 8.940633108480106e-05,
      "loss": 12.0884,
      "step": 12529
    },
    {
      "epoch": 2.19103825136612,
      "grad_norm": 0.6477983613243269,
      "learning_rate": 8.939224998969412e-05,
      "loss": 12.0071,
      "step": 12530
    },
    {
      "epoch": 2.1912131147540985,
      "grad_norm": 0.7307429287236993,
      "learning_rate": 8.937816910730513e-05,
      "loss": 12.054,
      "step": 12531
    },
    {
      "epoch": 2.1913879781420764,
      "grad_norm": 0.7300265314825611,
      "learning_rate": 8.93640884379165e-05,
      "loss": 12.0092,
      "step": 12532
    },
    {
      "epoch": 2.1915628415300548,
      "grad_norm": 0.6319536773869863,
      "learning_rate": 8.935000798181054e-05,
      "loss": 11.9713,
      "step": 12533
    },
    {
      "epoch": 2.1917377049180327,
      "grad_norm": 0.6215677170907934,
      "learning_rate": 8.933592773926961e-05,
      "loss": 11.7639,
      "step": 12534
    },
    {
      "epoch": 2.191912568306011,
      "grad_norm": 0.6386114817318747,
      "learning_rate": 8.932184771057611e-05,
      "loss": 11.8009,
      "step": 12535
    },
    {
      "epoch": 2.192087431693989,
      "grad_norm": 0.6261893643351994,
      "learning_rate": 8.930776789601227e-05,
      "loss": 11.8417,
      "step": 12536
    },
    {
      "epoch": 2.1922622950819672,
      "grad_norm": 0.6596163372060875,
      "learning_rate": 8.929368829586059e-05,
      "loss": 12.104,
      "step": 12537
    },
    {
      "epoch": 2.1924371584699456,
      "grad_norm": 0.6804824395148107,
      "learning_rate": 8.92796089104033e-05,
      "loss": 11.8109,
      "step": 12538
    },
    {
      "epoch": 2.1926120218579235,
      "grad_norm": 0.6853602116488051,
      "learning_rate": 8.926552973992277e-05,
      "loss": 11.7067,
      "step": 12539
    },
    {
      "epoch": 2.192786885245902,
      "grad_norm": 0.7703642666324548,
      "learning_rate": 8.925145078470135e-05,
      "loss": 11.9465,
      "step": 12540
    },
    {
      "epoch": 2.1929617486338797,
      "grad_norm": 0.7557538589284682,
      "learning_rate": 8.923737204502131e-05,
      "loss": 11.8242,
      "step": 12541
    },
    {
      "epoch": 2.193136612021858,
      "grad_norm": 0.6397897601967024,
      "learning_rate": 8.922329352116501e-05,
      "loss": 12.0377,
      "step": 12542
    },
    {
      "epoch": 2.193311475409836,
      "grad_norm": 0.6472476100324834,
      "learning_rate": 8.920921521341478e-05,
      "loss": 11.7877,
      "step": 12543
    },
    {
      "epoch": 2.1934863387978143,
      "grad_norm": 0.6889337885126701,
      "learning_rate": 8.919513712205293e-05,
      "loss": 12.0738,
      "step": 12544
    },
    {
      "epoch": 2.193661202185792,
      "grad_norm": 0.6316727990619809,
      "learning_rate": 8.918105924736174e-05,
      "loss": 11.7057,
      "step": 12545
    },
    {
      "epoch": 2.1938360655737705,
      "grad_norm": 0.660850399939988,
      "learning_rate": 8.91669815896235e-05,
      "loss": 11.8712,
      "step": 12546
    },
    {
      "epoch": 2.1940109289617484,
      "grad_norm": 0.6167510421869586,
      "learning_rate": 8.915290414912057e-05,
      "loss": 11.9056,
      "step": 12547
    },
    {
      "epoch": 2.1941857923497268,
      "grad_norm": 0.6435063345606007,
      "learning_rate": 8.91388269261352e-05,
      "loss": 11.9874,
      "step": 12548
    },
    {
      "epoch": 2.194360655737705,
      "grad_norm": 0.6331316967407608,
      "learning_rate": 8.912474992094974e-05,
      "loss": 11.8846,
      "step": 12549
    },
    {
      "epoch": 2.194535519125683,
      "grad_norm": 0.7620989767059017,
      "learning_rate": 8.911067313384638e-05,
      "loss": 11.9846,
      "step": 12550
    },
    {
      "epoch": 2.1947103825136614,
      "grad_norm": 0.6220672612323838,
      "learning_rate": 8.90965965651075e-05,
      "loss": 12.0885,
      "step": 12551
    },
    {
      "epoch": 2.1948852459016392,
      "grad_norm": 0.6181691390489393,
      "learning_rate": 8.908252021501532e-05,
      "loss": 11.9857,
      "step": 12552
    },
    {
      "epoch": 2.1950601092896176,
      "grad_norm": 0.6328967226538635,
      "learning_rate": 8.906844408385215e-05,
      "loss": 11.947,
      "step": 12553
    },
    {
      "epoch": 2.1952349726775955,
      "grad_norm": 0.6248819508630731,
      "learning_rate": 8.905436817190023e-05,
      "loss": 11.8834,
      "step": 12554
    },
    {
      "epoch": 2.195409836065574,
      "grad_norm": 0.6050848941484787,
      "learning_rate": 8.90402924794418e-05,
      "loss": 11.6726,
      "step": 12555
    },
    {
      "epoch": 2.1955846994535517,
      "grad_norm": 0.6326425787814014,
      "learning_rate": 8.90262170067592e-05,
      "loss": 11.936,
      "step": 12556
    },
    {
      "epoch": 2.19575956284153,
      "grad_norm": 0.6899509750081223,
      "learning_rate": 8.901214175413462e-05,
      "loss": 12.04,
      "step": 12557
    },
    {
      "epoch": 2.195934426229508,
      "grad_norm": 0.6967824652581375,
      "learning_rate": 8.899806672185037e-05,
      "loss": 11.9713,
      "step": 12558
    },
    {
      "epoch": 2.1961092896174863,
      "grad_norm": 0.6773049323590206,
      "learning_rate": 8.898399191018864e-05,
      "loss": 11.997,
      "step": 12559
    },
    {
      "epoch": 2.1962841530054646,
      "grad_norm": 0.6901027837303126,
      "learning_rate": 8.896991731943166e-05,
      "loss": 11.8902,
      "step": 12560
    },
    {
      "epoch": 2.1964590163934425,
      "grad_norm": 0.6986270444441472,
      "learning_rate": 8.895584294986175e-05,
      "loss": 11.9946,
      "step": 12561
    },
    {
      "epoch": 2.196633879781421,
      "grad_norm": 0.6749093747761701,
      "learning_rate": 8.894176880176109e-05,
      "loss": 11.8429,
      "step": 12562
    },
    {
      "epoch": 2.196808743169399,
      "grad_norm": 0.6449947240919952,
      "learning_rate": 8.892769487541193e-05,
      "loss": 12.0911,
      "step": 12563
    },
    {
      "epoch": 2.196983606557377,
      "grad_norm": 0.7203149377427062,
      "learning_rate": 8.891362117109646e-05,
      "loss": 11.9268,
      "step": 12564
    },
    {
      "epoch": 2.197158469945355,
      "grad_norm": 0.6374678249629107,
      "learning_rate": 8.889954768909692e-05,
      "loss": 11.8005,
      "step": 12565
    },
    {
      "epoch": 2.1973333333333334,
      "grad_norm": 0.644506915144701,
      "learning_rate": 8.888547442969554e-05,
      "loss": 11.9285,
      "step": 12566
    },
    {
      "epoch": 2.1975081967213113,
      "grad_norm": 0.6240183298500909,
      "learning_rate": 8.887140139317454e-05,
      "loss": 12.0331,
      "step": 12567
    },
    {
      "epoch": 2.1976830601092896,
      "grad_norm": 0.6503253411097911,
      "learning_rate": 8.885732857981611e-05,
      "loss": 12.0046,
      "step": 12568
    },
    {
      "epoch": 2.197857923497268,
      "grad_norm": 0.681656017138683,
      "learning_rate": 8.884325598990244e-05,
      "loss": 12.2003,
      "step": 12569
    },
    {
      "epoch": 2.198032786885246,
      "grad_norm": 0.7176314855307016,
      "learning_rate": 8.882918362371573e-05,
      "loss": 12.0935,
      "step": 12570
    },
    {
      "epoch": 2.198207650273224,
      "grad_norm": 0.7000419556541425,
      "learning_rate": 8.881511148153823e-05,
      "loss": 11.925,
      "step": 12571
    },
    {
      "epoch": 2.198382513661202,
      "grad_norm": 0.7292245423786134,
      "learning_rate": 8.880103956365207e-05,
      "loss": 11.9885,
      "step": 12572
    },
    {
      "epoch": 2.1985573770491804,
      "grad_norm": 0.69978322550671,
      "learning_rate": 8.878696787033943e-05,
      "loss": 12.0561,
      "step": 12573
    },
    {
      "epoch": 2.1987322404371583,
      "grad_norm": 0.6163119145882988,
      "learning_rate": 8.877289640188251e-05,
      "loss": 11.7532,
      "step": 12574
    },
    {
      "epoch": 2.1989071038251367,
      "grad_norm": 0.6939030492644211,
      "learning_rate": 8.875882515856351e-05,
      "loss": 11.7938,
      "step": 12575
    },
    {
      "epoch": 2.1990819672131146,
      "grad_norm": 0.6201292138786689,
      "learning_rate": 8.87447541406646e-05,
      "loss": 11.8214,
      "step": 12576
    },
    {
      "epoch": 2.199256830601093,
      "grad_norm": 0.6665183762492454,
      "learning_rate": 8.87306833484679e-05,
      "loss": 11.9815,
      "step": 12577
    },
    {
      "epoch": 2.1994316939890712,
      "grad_norm": 0.6467273468204295,
      "learning_rate": 8.87166127822556e-05,
      "loss": 12.1195,
      "step": 12578
    },
    {
      "epoch": 2.199606557377049,
      "grad_norm": 0.6365031171924433,
      "learning_rate": 8.870254244230984e-05,
      "loss": 11.8603,
      "step": 12579
    },
    {
      "epoch": 2.1997814207650275,
      "grad_norm": 0.6259438927983829,
      "learning_rate": 8.868847232891284e-05,
      "loss": 11.7276,
      "step": 12580
    },
    {
      "epoch": 2.1999562841530054,
      "grad_norm": 0.6661003181065771,
      "learning_rate": 8.867440244234668e-05,
      "loss": 11.7319,
      "step": 12581
    },
    {
      "epoch": 2.2001311475409837,
      "grad_norm": 0.6595490571203783,
      "learning_rate": 8.866033278289352e-05,
      "loss": 11.9605,
      "step": 12582
    },
    {
      "epoch": 2.2003060109289616,
      "grad_norm": 0.6377337556082127,
      "learning_rate": 8.864626335083553e-05,
      "loss": 11.9407,
      "step": 12583
    },
    {
      "epoch": 2.20048087431694,
      "grad_norm": 0.6534087470077422,
      "learning_rate": 8.863219414645475e-05,
      "loss": 11.712,
      "step": 12584
    },
    {
      "epoch": 2.200655737704918,
      "grad_norm": 0.7548179056992772,
      "learning_rate": 8.861812517003345e-05,
      "loss": 11.9061,
      "step": 12585
    },
    {
      "epoch": 2.200830601092896,
      "grad_norm": 0.651020993405007,
      "learning_rate": 8.860405642185366e-05,
      "loss": 12.0027,
      "step": 12586
    },
    {
      "epoch": 2.2010054644808745,
      "grad_norm": 0.74853088925123,
      "learning_rate": 8.858998790219753e-05,
      "loss": 11.9995,
      "step": 12587
    },
    {
      "epoch": 2.2011803278688524,
      "grad_norm": 0.6132263215029491,
      "learning_rate": 8.857591961134721e-05,
      "loss": 11.9036,
      "step": 12588
    },
    {
      "epoch": 2.2013551912568308,
      "grad_norm": 0.6082724177084062,
      "learning_rate": 8.856185154958472e-05,
      "loss": 11.4403,
      "step": 12589
    },
    {
      "epoch": 2.2015300546448087,
      "grad_norm": 0.6943110120996963,
      "learning_rate": 8.854778371719226e-05,
      "loss": 12.1027,
      "step": 12590
    },
    {
      "epoch": 2.201704918032787,
      "grad_norm": 0.7127391604675998,
      "learning_rate": 8.853371611445189e-05,
      "loss": 12.0674,
      "step": 12591
    },
    {
      "epoch": 2.201879781420765,
      "grad_norm": 0.6546991495155595,
      "learning_rate": 8.851964874164574e-05,
      "loss": 11.9856,
      "step": 12592
    },
    {
      "epoch": 2.2020546448087432,
      "grad_norm": 0.6946842507251836,
      "learning_rate": 8.850558159905587e-05,
      "loss": 11.7868,
      "step": 12593
    },
    {
      "epoch": 2.202229508196721,
      "grad_norm": 0.7155636421285866,
      "learning_rate": 8.849151468696434e-05,
      "loss": 11.9184,
      "step": 12594
    },
    {
      "epoch": 2.2024043715846995,
      "grad_norm": 0.6317934866755042,
      "learning_rate": 8.847744800565331e-05,
      "loss": 11.688,
      "step": 12595
    },
    {
      "epoch": 2.202579234972678,
      "grad_norm": 0.6528268893284,
      "learning_rate": 8.846338155540484e-05,
      "loss": 11.8325,
      "step": 12596
    },
    {
      "epoch": 2.2027540983606557,
      "grad_norm": 0.6692646503131284,
      "learning_rate": 8.844931533650099e-05,
      "loss": 12.0375,
      "step": 12597
    },
    {
      "epoch": 2.202928961748634,
      "grad_norm": 0.678942561155214,
      "learning_rate": 8.843524934922381e-05,
      "loss": 11.902,
      "step": 12598
    },
    {
      "epoch": 2.203103825136612,
      "grad_norm": 0.7566158493144303,
      "learning_rate": 8.84211835938554e-05,
      "loss": 11.939,
      "step": 12599
    },
    {
      "epoch": 2.2032786885245903,
      "grad_norm": 0.7258976056174147,
      "learning_rate": 8.840711807067781e-05,
      "loss": 11.9101,
      "step": 12600
    },
    {
      "epoch": 2.203453551912568,
      "grad_norm": 0.6441491840452342,
      "learning_rate": 8.839305277997312e-05,
      "loss": 11.7783,
      "step": 12601
    },
    {
      "epoch": 2.2036284153005465,
      "grad_norm": 0.6355077333433344,
      "learning_rate": 8.837898772202333e-05,
      "loss": 11.9057,
      "step": 12602
    },
    {
      "epoch": 2.2038032786885244,
      "grad_norm": 0.5768702966055322,
      "learning_rate": 8.836492289711051e-05,
      "loss": 11.8607,
      "step": 12603
    },
    {
      "epoch": 2.2039781420765028,
      "grad_norm": 0.6640836480654886,
      "learning_rate": 8.835085830551673e-05,
      "loss": 11.8742,
      "step": 12604
    },
    {
      "epoch": 2.2041530054644807,
      "grad_norm": 0.6466406989812807,
      "learning_rate": 8.833679394752399e-05,
      "loss": 11.8574,
      "step": 12605
    },
    {
      "epoch": 2.204327868852459,
      "grad_norm": 0.6320951291957083,
      "learning_rate": 8.832272982341439e-05,
      "loss": 11.9805,
      "step": 12606
    },
    {
      "epoch": 2.2045027322404374,
      "grad_norm": 0.6223162931572245,
      "learning_rate": 8.830866593346988e-05,
      "loss": 11.8755,
      "step": 12607
    },
    {
      "epoch": 2.2046775956284153,
      "grad_norm": 0.6483383500656695,
      "learning_rate": 8.82946022779725e-05,
      "loss": 11.9994,
      "step": 12608
    },
    {
      "epoch": 2.2048524590163936,
      "grad_norm": 0.6328705599071179,
      "learning_rate": 8.828053885720429e-05,
      "loss": 12.1043,
      "step": 12609
    },
    {
      "epoch": 2.2050273224043715,
      "grad_norm": 0.6526732989801504,
      "learning_rate": 8.82664756714473e-05,
      "loss": 11.9005,
      "step": 12610
    },
    {
      "epoch": 2.20520218579235,
      "grad_norm": 0.6358282532466204,
      "learning_rate": 8.825241272098346e-05,
      "loss": 11.6358,
      "step": 12611
    },
    {
      "epoch": 2.2053770491803277,
      "grad_norm": 0.6417030037185005,
      "learning_rate": 8.823835000609482e-05,
      "loss": 11.9483,
      "step": 12612
    },
    {
      "epoch": 2.205551912568306,
      "grad_norm": 0.6365505939224543,
      "learning_rate": 8.822428752706336e-05,
      "loss": 11.8139,
      "step": 12613
    },
    {
      "epoch": 2.205726775956284,
      "grad_norm": 0.6201561154476956,
      "learning_rate": 8.82102252841711e-05,
      "loss": 12.051,
      "step": 12614
    },
    {
      "epoch": 2.2059016393442623,
      "grad_norm": 0.6812167206657801,
      "learning_rate": 8.819616327770006e-05,
      "loss": 12.0646,
      "step": 12615
    },
    {
      "epoch": 2.20607650273224,
      "grad_norm": 0.66792193261601,
      "learning_rate": 8.818210150793215e-05,
      "loss": 11.9463,
      "step": 12616
    },
    {
      "epoch": 2.2062513661202185,
      "grad_norm": 0.68873166428662,
      "learning_rate": 8.816803997514943e-05,
      "loss": 12.155,
      "step": 12617
    },
    {
      "epoch": 2.206426229508197,
      "grad_norm": 0.7183238830636484,
      "learning_rate": 8.815397867963375e-05,
      "loss": 11.9597,
      "step": 12618
    },
    {
      "epoch": 2.206601092896175,
      "grad_norm": 0.6327495060322484,
      "learning_rate": 8.813991762166724e-05,
      "loss": 12.0138,
      "step": 12619
    },
    {
      "epoch": 2.206775956284153,
      "grad_norm": 0.6176539031854132,
      "learning_rate": 8.81258568015318e-05,
      "loss": 11.9225,
      "step": 12620
    },
    {
      "epoch": 2.206950819672131,
      "grad_norm": 0.6632133080919049,
      "learning_rate": 8.811179621950936e-05,
      "loss": 11.6508,
      "step": 12621
    },
    {
      "epoch": 2.2071256830601094,
      "grad_norm": 0.7124837939476331,
      "learning_rate": 8.80977358758819e-05,
      "loss": 12.017,
      "step": 12622
    },
    {
      "epoch": 2.2073005464480873,
      "grad_norm": 0.7160669799369911,
      "learning_rate": 8.80836757709314e-05,
      "loss": 11.4962,
      "step": 12623
    },
    {
      "epoch": 2.2074754098360656,
      "grad_norm": 0.682852741899579,
      "learning_rate": 8.806961590493981e-05,
      "loss": 11.8733,
      "step": 12624
    },
    {
      "epoch": 2.2076502732240435,
      "grad_norm": 0.7056460507793905,
      "learning_rate": 8.805555627818905e-05,
      "loss": 12.0701,
      "step": 12625
    },
    {
      "epoch": 2.207825136612022,
      "grad_norm": 0.7728172204633034,
      "learning_rate": 8.804149689096103e-05,
      "loss": 11.8749,
      "step": 12626
    },
    {
      "epoch": 2.208,
      "grad_norm": 0.6197177192760084,
      "learning_rate": 8.802743774353771e-05,
      "loss": 11.6565,
      "step": 12627
    },
    {
      "epoch": 2.208174863387978,
      "grad_norm": 0.6168359348172123,
      "learning_rate": 8.801337883620107e-05,
      "loss": 12.0241,
      "step": 12628
    },
    {
      "epoch": 2.2083497267759564,
      "grad_norm": 0.6908266927737637,
      "learning_rate": 8.799932016923295e-05,
      "loss": 11.9371,
      "step": 12629
    },
    {
      "epoch": 2.2085245901639343,
      "grad_norm": 0.6521391947297352,
      "learning_rate": 8.798526174291531e-05,
      "loss": 12.0215,
      "step": 12630
    },
    {
      "epoch": 2.2086994535519127,
      "grad_norm": 0.7412620270845175,
      "learning_rate": 8.79712035575301e-05,
      "loss": 11.972,
      "step": 12631
    },
    {
      "epoch": 2.2088743169398906,
      "grad_norm": 0.7370452057845779,
      "learning_rate": 8.795714561335911e-05,
      "loss": 12.1368,
      "step": 12632
    },
    {
      "epoch": 2.209049180327869,
      "grad_norm": 0.6530952263041035,
      "learning_rate": 8.794308791068441e-05,
      "loss": 12.0621,
      "step": 12633
    },
    {
      "epoch": 2.209224043715847,
      "grad_norm": 0.764262135893395,
      "learning_rate": 8.792903044978777e-05,
      "loss": 12.0761,
      "step": 12634
    },
    {
      "epoch": 2.209398907103825,
      "grad_norm": 0.6427070863417738,
      "learning_rate": 8.791497323095116e-05,
      "loss": 11.7009,
      "step": 12635
    },
    {
      "epoch": 2.2095737704918035,
      "grad_norm": 0.6269094008560159,
      "learning_rate": 8.790091625445644e-05,
      "loss": 11.7555,
      "step": 12636
    },
    {
      "epoch": 2.2097486338797814,
      "grad_norm": 0.7811497087737318,
      "learning_rate": 8.788685952058548e-05,
      "loss": 11.9755,
      "step": 12637
    },
    {
      "epoch": 2.2099234972677597,
      "grad_norm": 0.6435990634810366,
      "learning_rate": 8.787280302962018e-05,
      "loss": 11.8296,
      "step": 12638
    },
    {
      "epoch": 2.2100983606557376,
      "grad_norm": 0.7138842727776531,
      "learning_rate": 8.785874678184242e-05,
      "loss": 11.9614,
      "step": 12639
    },
    {
      "epoch": 2.210273224043716,
      "grad_norm": 0.5952377180180155,
      "learning_rate": 8.784469077753409e-05,
      "loss": 11.9346,
      "step": 12640
    },
    {
      "epoch": 2.210448087431694,
      "grad_norm": 0.6626087052119538,
      "learning_rate": 8.783063501697701e-05,
      "loss": 11.8097,
      "step": 12641
    },
    {
      "epoch": 2.210622950819672,
      "grad_norm": 0.6935143474360119,
      "learning_rate": 8.781657950045304e-05,
      "loss": 11.9535,
      "step": 12642
    },
    {
      "epoch": 2.21079781420765,
      "grad_norm": 0.6319467522115696,
      "learning_rate": 8.780252422824408e-05,
      "loss": 11.7283,
      "step": 12643
    },
    {
      "epoch": 2.2109726775956284,
      "grad_norm": 0.6108189509566853,
      "learning_rate": 8.778846920063197e-05,
      "loss": 11.4416,
      "step": 12644
    },
    {
      "epoch": 2.2111475409836068,
      "grad_norm": 0.7023922931850133,
      "learning_rate": 8.777441441789855e-05,
      "loss": 11.9985,
      "step": 12645
    },
    {
      "epoch": 2.2113224043715847,
      "grad_norm": 0.7226528474216752,
      "learning_rate": 8.776035988032563e-05,
      "loss": 11.9632,
      "step": 12646
    },
    {
      "epoch": 2.211497267759563,
      "grad_norm": 0.6513506646168504,
      "learning_rate": 8.774630558819507e-05,
      "loss": 12.089,
      "step": 12647
    },
    {
      "epoch": 2.211672131147541,
      "grad_norm": 0.6554928925960601,
      "learning_rate": 8.773225154178873e-05,
      "loss": 12.067,
      "step": 12648
    },
    {
      "epoch": 2.2118469945355193,
      "grad_norm": 0.6558011695806727,
      "learning_rate": 8.771819774138844e-05,
      "loss": 11.9697,
      "step": 12649
    },
    {
      "epoch": 2.212021857923497,
      "grad_norm": 0.6146450132396932,
      "learning_rate": 8.770414418727597e-05,
      "loss": 11.7061,
      "step": 12650
    },
    {
      "epoch": 2.2121967213114755,
      "grad_norm": 0.6722580800070624,
      "learning_rate": 8.769009087973314e-05,
      "loss": 11.6579,
      "step": 12651
    },
    {
      "epoch": 2.2123715846994534,
      "grad_norm": 0.663156746600246,
      "learning_rate": 8.76760378190418e-05,
      "loss": 11.9973,
      "step": 12652
    },
    {
      "epoch": 2.2125464480874317,
      "grad_norm": 0.6329186513574412,
      "learning_rate": 8.766198500548375e-05,
      "loss": 11.9018,
      "step": 12653
    },
    {
      "epoch": 2.21272131147541,
      "grad_norm": 0.6769720355646833,
      "learning_rate": 8.76479324393408e-05,
      "loss": 11.8305,
      "step": 12654
    },
    {
      "epoch": 2.212896174863388,
      "grad_norm": 0.6589766292466536,
      "learning_rate": 8.76338801208947e-05,
      "loss": 11.9326,
      "step": 12655
    },
    {
      "epoch": 2.2130710382513663,
      "grad_norm": 0.7052475902182813,
      "learning_rate": 8.761982805042726e-05,
      "loss": 11.8992,
      "step": 12656
    },
    {
      "epoch": 2.213245901639344,
      "grad_norm": 0.6728950535373084,
      "learning_rate": 8.76057762282203e-05,
      "loss": 12.0802,
      "step": 12657
    },
    {
      "epoch": 2.2134207650273225,
      "grad_norm": 0.6664264198585006,
      "learning_rate": 8.759172465455561e-05,
      "loss": 12.0768,
      "step": 12658
    },
    {
      "epoch": 2.2135956284153004,
      "grad_norm": 0.6205604088721337,
      "learning_rate": 8.75776733297149e-05,
      "loss": 11.796,
      "step": 12659
    },
    {
      "epoch": 2.213770491803279,
      "grad_norm": 0.6784592458028467,
      "learning_rate": 8.756362225397999e-05,
      "loss": 11.9701,
      "step": 12660
    },
    {
      "epoch": 2.2139453551912567,
      "grad_norm": 0.6326590126071568,
      "learning_rate": 8.754957142763263e-05,
      "loss": 11.9396,
      "step": 12661
    },
    {
      "epoch": 2.214120218579235,
      "grad_norm": 0.64822767025942,
      "learning_rate": 8.75355208509546e-05,
      "loss": 12.0809,
      "step": 12662
    },
    {
      "epoch": 2.214295081967213,
      "grad_norm": 0.6688472425879004,
      "learning_rate": 8.752147052422767e-05,
      "loss": 11.8594,
      "step": 12663
    },
    {
      "epoch": 2.2144699453551913,
      "grad_norm": 0.6783323707426138,
      "learning_rate": 8.750742044773354e-05,
      "loss": 11.7891,
      "step": 12664
    },
    {
      "epoch": 2.2146448087431696,
      "grad_norm": 0.6423613113477022,
      "learning_rate": 8.749337062175403e-05,
      "loss": 12.0348,
      "step": 12665
    },
    {
      "epoch": 2.2148196721311475,
      "grad_norm": 0.6882894898634404,
      "learning_rate": 8.747932104657076e-05,
      "loss": 12.0221,
      "step": 12666
    },
    {
      "epoch": 2.214994535519126,
      "grad_norm": 0.627602797855594,
      "learning_rate": 8.746527172246562e-05,
      "loss": 11.9966,
      "step": 12667
    },
    {
      "epoch": 2.2151693989071037,
      "grad_norm": 0.650323899271686,
      "learning_rate": 8.745122264972023e-05,
      "loss": 12.1099,
      "step": 12668
    },
    {
      "epoch": 2.215344262295082,
      "grad_norm": 0.6124155954158126,
      "learning_rate": 8.743717382861639e-05,
      "loss": 11.8354,
      "step": 12669
    },
    {
      "epoch": 2.21551912568306,
      "grad_norm": 0.7304882807063585,
      "learning_rate": 8.742312525943572e-05,
      "loss": 11.9575,
      "step": 12670
    },
    {
      "epoch": 2.2156939890710383,
      "grad_norm": 0.6878319771157054,
      "learning_rate": 8.740907694246006e-05,
      "loss": 11.9511,
      "step": 12671
    },
    {
      "epoch": 2.215868852459016,
      "grad_norm": 0.6054802381212282,
      "learning_rate": 8.739502887797107e-05,
      "loss": 12.0361,
      "step": 12672
    },
    {
      "epoch": 2.2160437158469946,
      "grad_norm": 0.674971325672521,
      "learning_rate": 8.738098106625045e-05,
      "loss": 11.9466,
      "step": 12673
    },
    {
      "epoch": 2.2162185792349725,
      "grad_norm": 0.6252368442644686,
      "learning_rate": 8.736693350757992e-05,
      "loss": 11.841,
      "step": 12674
    },
    {
      "epoch": 2.216393442622951,
      "grad_norm": 0.645414719538429,
      "learning_rate": 8.73528862022411e-05,
      "loss": 11.8522,
      "step": 12675
    },
    {
      "epoch": 2.216568306010929,
      "grad_norm": 0.7026536751373109,
      "learning_rate": 8.733883915051578e-05,
      "loss": 12.1187,
      "step": 12676
    },
    {
      "epoch": 2.216743169398907,
      "grad_norm": 0.680901900561159,
      "learning_rate": 8.732479235268561e-05,
      "loss": 11.9921,
      "step": 12677
    },
    {
      "epoch": 2.2169180327868854,
      "grad_norm": 0.6338368652255919,
      "learning_rate": 8.731074580903228e-05,
      "loss": 11.9036,
      "step": 12678
    },
    {
      "epoch": 2.2170928961748633,
      "grad_norm": 0.5923544786883379,
      "learning_rate": 8.729669951983747e-05,
      "loss": 11.9665,
      "step": 12679
    },
    {
      "epoch": 2.2172677595628416,
      "grad_norm": 0.7363308825410465,
      "learning_rate": 8.728265348538277e-05,
      "loss": 12.046,
      "step": 12680
    },
    {
      "epoch": 2.2174426229508195,
      "grad_norm": 0.6879716845135739,
      "learning_rate": 8.726860770594998e-05,
      "loss": 12.0665,
      "step": 12681
    },
    {
      "epoch": 2.217617486338798,
      "grad_norm": 0.6579383572187161,
      "learning_rate": 8.725456218182068e-05,
      "loss": 11.9404,
      "step": 12682
    },
    {
      "epoch": 2.2177923497267757,
      "grad_norm": 0.7543468350333976,
      "learning_rate": 8.724051691327657e-05,
      "loss": 11.9747,
      "step": 12683
    },
    {
      "epoch": 2.217967213114754,
      "grad_norm": 0.6892494555622731,
      "learning_rate": 8.722647190059924e-05,
      "loss": 11.9063,
      "step": 12684
    },
    {
      "epoch": 2.2181420765027324,
      "grad_norm": 0.7217494407475614,
      "learning_rate": 8.721242714407035e-05,
      "loss": 12.0881,
      "step": 12685
    },
    {
      "epoch": 2.2183169398907103,
      "grad_norm": 0.7161774777974518,
      "learning_rate": 8.719838264397158e-05,
      "loss": 11.9762,
      "step": 12686
    },
    {
      "epoch": 2.2184918032786887,
      "grad_norm": 0.616303023256918,
      "learning_rate": 8.718433840058454e-05,
      "loss": 11.4163,
      "step": 12687
    },
    {
      "epoch": 2.2186666666666666,
      "grad_norm": 0.6916671625696378,
      "learning_rate": 8.717029441419088e-05,
      "loss": 12.0137,
      "step": 12688
    },
    {
      "epoch": 2.218841530054645,
      "grad_norm": 0.735978067566617,
      "learning_rate": 8.71562506850722e-05,
      "loss": 11.8969,
      "step": 12689
    },
    {
      "epoch": 2.219016393442623,
      "grad_norm": 0.6546188189142912,
      "learning_rate": 8.714220721351012e-05,
      "loss": 11.8915,
      "step": 12690
    },
    {
      "epoch": 2.219191256830601,
      "grad_norm": 0.8142506033748761,
      "learning_rate": 8.712816399978627e-05,
      "loss": 11.8789,
      "step": 12691
    },
    {
      "epoch": 2.219366120218579,
      "grad_norm": 0.6532987398363169,
      "learning_rate": 8.711412104418228e-05,
      "loss": 11.6375,
      "step": 12692
    },
    {
      "epoch": 2.2195409836065574,
      "grad_norm": 0.6936294983644956,
      "learning_rate": 8.710007834697969e-05,
      "loss": 11.9933,
      "step": 12693
    },
    {
      "epoch": 2.2197158469945357,
      "grad_norm": 0.6548741082150494,
      "learning_rate": 8.708603590846014e-05,
      "loss": 11.9624,
      "step": 12694
    },
    {
      "epoch": 2.2198907103825136,
      "grad_norm": 0.675884314792811,
      "learning_rate": 8.707199372890523e-05,
      "loss": 11.9346,
      "step": 12695
    },
    {
      "epoch": 2.220065573770492,
      "grad_norm": 0.64961012868439,
      "learning_rate": 8.705795180859654e-05,
      "loss": 11.9385,
      "step": 12696
    },
    {
      "epoch": 2.22024043715847,
      "grad_norm": 0.6564679360504811,
      "learning_rate": 8.704391014781567e-05,
      "loss": 11.9303,
      "step": 12697
    },
    {
      "epoch": 2.220415300546448,
      "grad_norm": 0.7127656010682201,
      "learning_rate": 8.702986874684417e-05,
      "loss": 12.1571,
      "step": 12698
    },
    {
      "epoch": 2.220590163934426,
      "grad_norm": 0.6914639275059655,
      "learning_rate": 8.70158276059636e-05,
      "loss": 11.9819,
      "step": 12699
    },
    {
      "epoch": 2.2207650273224044,
      "grad_norm": 0.6516246911787091,
      "learning_rate": 8.700178672545556e-05,
      "loss": 12.0812,
      "step": 12700
    },
    {
      "epoch": 2.2209398907103823,
      "grad_norm": 0.6286254798132268,
      "learning_rate": 8.698774610560163e-05,
      "loss": 11.6345,
      "step": 12701
    },
    {
      "epoch": 2.2211147540983607,
      "grad_norm": 0.7378778340105159,
      "learning_rate": 8.697370574668335e-05,
      "loss": 11.7968,
      "step": 12702
    },
    {
      "epoch": 2.221289617486339,
      "grad_norm": 0.6316581697508021,
      "learning_rate": 8.695966564898226e-05,
      "loss": 11.7634,
      "step": 12703
    },
    {
      "epoch": 2.221464480874317,
      "grad_norm": 0.6553664323081922,
      "learning_rate": 8.694562581277987e-05,
      "loss": 11.7722,
      "step": 12704
    },
    {
      "epoch": 2.2216393442622953,
      "grad_norm": 0.6936835192092468,
      "learning_rate": 8.693158623835781e-05,
      "loss": 11.8692,
      "step": 12705
    },
    {
      "epoch": 2.221814207650273,
      "grad_norm": 0.6587402691873832,
      "learning_rate": 8.691754692599758e-05,
      "loss": 11.755,
      "step": 12706
    },
    {
      "epoch": 2.2219890710382515,
      "grad_norm": 0.6949135053074387,
      "learning_rate": 8.690350787598067e-05,
      "loss": 12.0929,
      "step": 12707
    },
    {
      "epoch": 2.2221639344262294,
      "grad_norm": 0.611748018782154,
      "learning_rate": 8.688946908858867e-05,
      "loss": 11.9123,
      "step": 12708
    },
    {
      "epoch": 2.2223387978142077,
      "grad_norm": 0.6485250868200263,
      "learning_rate": 8.6875430564103e-05,
      "loss": 11.6465,
      "step": 12709
    },
    {
      "epoch": 2.2225136612021856,
      "grad_norm": 0.6483415533254695,
      "learning_rate": 8.68613923028053e-05,
      "loss": 12.024,
      "step": 12710
    },
    {
      "epoch": 2.222688524590164,
      "grad_norm": 0.6831164959454679,
      "learning_rate": 8.684735430497704e-05,
      "loss": 12.1723,
      "step": 12711
    },
    {
      "epoch": 2.222863387978142,
      "grad_norm": 0.6668791304580113,
      "learning_rate": 8.68333165708997e-05,
      "loss": 11.8857,
      "step": 12712
    },
    {
      "epoch": 2.22303825136612,
      "grad_norm": 0.6473830377588692,
      "learning_rate": 8.68192791008548e-05,
      "loss": 11.947,
      "step": 12713
    },
    {
      "epoch": 2.2232131147540986,
      "grad_norm": 0.7122607288470352,
      "learning_rate": 8.680524189512376e-05,
      "loss": 11.7864,
      "step": 12714
    },
    {
      "epoch": 2.2233879781420764,
      "grad_norm": 0.6399197299547079,
      "learning_rate": 8.679120495398819e-05,
      "loss": 11.8854,
      "step": 12715
    },
    {
      "epoch": 2.223562841530055,
      "grad_norm": 0.6730249339010577,
      "learning_rate": 8.67771682777295e-05,
      "loss": 12.0049,
      "step": 12716
    },
    {
      "epoch": 2.2237377049180327,
      "grad_norm": 0.659216877532175,
      "learning_rate": 8.67631318666292e-05,
      "loss": 11.715,
      "step": 12717
    },
    {
      "epoch": 2.223912568306011,
      "grad_norm": 0.6959822778588687,
      "learning_rate": 8.674909572096871e-05,
      "loss": 11.8519,
      "step": 12718
    },
    {
      "epoch": 2.224087431693989,
      "grad_norm": 0.6045807299146778,
      "learning_rate": 8.673505984102957e-05,
      "loss": 11.8425,
      "step": 12719
    },
    {
      "epoch": 2.2242622950819673,
      "grad_norm": 0.6395897057763079,
      "learning_rate": 8.672102422709323e-05,
      "loss": 11.9246,
      "step": 12720
    },
    {
      "epoch": 2.224437158469945,
      "grad_norm": 0.671458171480757,
      "learning_rate": 8.670698887944108e-05,
      "loss": 11.8152,
      "step": 12721
    },
    {
      "epoch": 2.2246120218579235,
      "grad_norm": 0.6253111255228745,
      "learning_rate": 8.669295379835467e-05,
      "loss": 12.0518,
      "step": 12722
    },
    {
      "epoch": 2.224786885245902,
      "grad_norm": 0.6448118368464314,
      "learning_rate": 8.667891898411531e-05,
      "loss": 12.0986,
      "step": 12723
    },
    {
      "epoch": 2.2249617486338797,
      "grad_norm": 0.6627648246334645,
      "learning_rate": 8.66648844370046e-05,
      "loss": 11.9891,
      "step": 12724
    },
    {
      "epoch": 2.225136612021858,
      "grad_norm": 0.6427976911999496,
      "learning_rate": 8.665085015730388e-05,
      "loss": 11.8102,
      "step": 12725
    },
    {
      "epoch": 2.225311475409836,
      "grad_norm": 0.6324178808405513,
      "learning_rate": 8.663681614529459e-05,
      "loss": 12.1502,
      "step": 12726
    },
    {
      "epoch": 2.2254863387978143,
      "grad_norm": 0.7524081627745448,
      "learning_rate": 8.66227824012582e-05,
      "loss": 12.0831,
      "step": 12727
    },
    {
      "epoch": 2.225661202185792,
      "grad_norm": 0.6302627460253153,
      "learning_rate": 8.660874892547603e-05,
      "loss": 11.6812,
      "step": 12728
    },
    {
      "epoch": 2.2258360655737706,
      "grad_norm": 0.6976179340081939,
      "learning_rate": 8.659471571822964e-05,
      "loss": 11.8213,
      "step": 12729
    },
    {
      "epoch": 2.2260109289617485,
      "grad_norm": 0.6867912820701254,
      "learning_rate": 8.658068277980031e-05,
      "loss": 11.8061,
      "step": 12730
    },
    {
      "epoch": 2.226185792349727,
      "grad_norm": 0.6767777102270969,
      "learning_rate": 8.656665011046952e-05,
      "loss": 11.9608,
      "step": 12731
    },
    {
      "epoch": 2.2263606557377047,
      "grad_norm": 0.6928330561823056,
      "learning_rate": 8.655261771051863e-05,
      "loss": 11.8473,
      "step": 12732
    },
    {
      "epoch": 2.226535519125683,
      "grad_norm": 0.626385140732259,
      "learning_rate": 8.653858558022903e-05,
      "loss": 11.9691,
      "step": 12733
    },
    {
      "epoch": 2.2267103825136614,
      "grad_norm": 0.6062535566653373,
      "learning_rate": 8.652455371988213e-05,
      "loss": 11.8849,
      "step": 12734
    },
    {
      "epoch": 2.2268852459016393,
      "grad_norm": 0.670714709916714,
      "learning_rate": 8.651052212975929e-05,
      "loss": 11.7028,
      "step": 12735
    },
    {
      "epoch": 2.2270601092896176,
      "grad_norm": 0.6781574107069379,
      "learning_rate": 8.649649081014195e-05,
      "loss": 12.0668,
      "step": 12736
    },
    {
      "epoch": 2.2272349726775955,
      "grad_norm": 0.6771500305882205,
      "learning_rate": 8.64824597613114e-05,
      "loss": 11.8649,
      "step": 12737
    },
    {
      "epoch": 2.227409836065574,
      "grad_norm": 0.6671212932923851,
      "learning_rate": 8.6468428983549e-05,
      "loss": 11.7748,
      "step": 12738
    },
    {
      "epoch": 2.2275846994535518,
      "grad_norm": 0.6643386000221989,
      "learning_rate": 8.645439847713618e-05,
      "loss": 11.8549,
      "step": 12739
    },
    {
      "epoch": 2.22775956284153,
      "grad_norm": 0.6440376912471473,
      "learning_rate": 8.644036824235427e-05,
      "loss": 11.819,
      "step": 12740
    },
    {
      "epoch": 2.227934426229508,
      "grad_norm": 0.6599402843554746,
      "learning_rate": 8.64263382794846e-05,
      "loss": 11.7843,
      "step": 12741
    },
    {
      "epoch": 2.2281092896174863,
      "grad_norm": 0.6898574289574815,
      "learning_rate": 8.64123085888085e-05,
      "loss": 11.8481,
      "step": 12742
    },
    {
      "epoch": 2.2282841530054647,
      "grad_norm": 0.5894248011029385,
      "learning_rate": 8.639827917060735e-05,
      "loss": 11.7441,
      "step": 12743
    },
    {
      "epoch": 2.2284590163934426,
      "grad_norm": 0.6123549480867103,
      "learning_rate": 8.638425002516247e-05,
      "loss": 11.7423,
      "step": 12744
    },
    {
      "epoch": 2.228633879781421,
      "grad_norm": 0.6737710139006453,
      "learning_rate": 8.637022115275521e-05,
      "loss": 11.7936,
      "step": 12745
    },
    {
      "epoch": 2.228808743169399,
      "grad_norm": 0.7216219639612474,
      "learning_rate": 8.635619255366684e-05,
      "loss": 11.9651,
      "step": 12746
    },
    {
      "epoch": 2.228983606557377,
      "grad_norm": 0.6766204244615424,
      "learning_rate": 8.634216422817867e-05,
      "loss": 12.0495,
      "step": 12747
    },
    {
      "epoch": 2.229158469945355,
      "grad_norm": 0.6787142868420015,
      "learning_rate": 8.632813617657207e-05,
      "loss": 12.0165,
      "step": 12748
    },
    {
      "epoch": 2.2293333333333334,
      "grad_norm": 0.6242370535409238,
      "learning_rate": 8.631410839912835e-05,
      "loss": 11.9723,
      "step": 12749
    },
    {
      "epoch": 2.2295081967213113,
      "grad_norm": 0.7144350247565915,
      "learning_rate": 8.630008089612874e-05,
      "loss": 12.0377,
      "step": 12750
    },
    {
      "epoch": 2.2296830601092896,
      "grad_norm": 0.6250612004472722,
      "learning_rate": 8.628605366785458e-05,
      "loss": 11.5992,
      "step": 12751
    },
    {
      "epoch": 2.229857923497268,
      "grad_norm": 0.7143862788240113,
      "learning_rate": 8.627202671458713e-05,
      "loss": 11.749,
      "step": 12752
    },
    {
      "epoch": 2.230032786885246,
      "grad_norm": 0.645974199492688,
      "learning_rate": 8.62580000366077e-05,
      "loss": 11.8692,
      "step": 12753
    },
    {
      "epoch": 2.230207650273224,
      "grad_norm": 0.7172524806341244,
      "learning_rate": 8.62439736341976e-05,
      "loss": 11.7297,
      "step": 12754
    },
    {
      "epoch": 2.230382513661202,
      "grad_norm": 0.6514827734715585,
      "learning_rate": 8.622994750763805e-05,
      "loss": 11.9753,
      "step": 12755
    },
    {
      "epoch": 2.2305573770491804,
      "grad_norm": 0.699878419625104,
      "learning_rate": 8.621592165721034e-05,
      "loss": 11.8952,
      "step": 12756
    },
    {
      "epoch": 2.2307322404371583,
      "grad_norm": 0.6582297388123026,
      "learning_rate": 8.620189608319568e-05,
      "loss": 12.058,
      "step": 12757
    },
    {
      "epoch": 2.2309071038251367,
      "grad_norm": 0.6548259439775754,
      "learning_rate": 8.61878707858754e-05,
      "loss": 11.9713,
      "step": 12758
    },
    {
      "epoch": 2.2310819672131146,
      "grad_norm": 0.6698911423273721,
      "learning_rate": 8.617384576553073e-05,
      "loss": 12.0864,
      "step": 12759
    },
    {
      "epoch": 2.231256830601093,
      "grad_norm": 0.6908447814991158,
      "learning_rate": 8.615982102244289e-05,
      "loss": 11.8835,
      "step": 12760
    },
    {
      "epoch": 2.2314316939890713,
      "grad_norm": 0.660953831206281,
      "learning_rate": 8.614579655689317e-05,
      "loss": 11.8796,
      "step": 12761
    },
    {
      "epoch": 2.231606557377049,
      "grad_norm": 0.6891740874543824,
      "learning_rate": 8.613177236916268e-05,
      "loss": 12.1508,
      "step": 12762
    },
    {
      "epoch": 2.2317814207650275,
      "grad_norm": 0.6797782873366303,
      "learning_rate": 8.611774845953281e-05,
      "loss": 11.8253,
      "step": 12763
    },
    {
      "epoch": 2.2319562841530054,
      "grad_norm": 0.6744339348965671,
      "learning_rate": 8.610372482828469e-05,
      "loss": 12.0975,
      "step": 12764
    },
    {
      "epoch": 2.2321311475409837,
      "grad_norm": 0.6644255373750684,
      "learning_rate": 8.608970147569954e-05,
      "loss": 11.5745,
      "step": 12765
    },
    {
      "epoch": 2.2323060109289616,
      "grad_norm": 0.6945969212954546,
      "learning_rate": 8.607567840205857e-05,
      "loss": 11.9823,
      "step": 12766
    },
    {
      "epoch": 2.23248087431694,
      "grad_norm": 0.65173489561297,
      "learning_rate": 8.606165560764306e-05,
      "loss": 11.8775,
      "step": 12767
    },
    {
      "epoch": 2.232655737704918,
      "grad_norm": 0.6456615625579943,
      "learning_rate": 8.604763309273412e-05,
      "loss": 12.1251,
      "step": 12768
    },
    {
      "epoch": 2.232830601092896,
      "grad_norm": 0.6595506811053397,
      "learning_rate": 8.603361085761297e-05,
      "loss": 11.9362,
      "step": 12769
    },
    {
      "epoch": 2.233005464480874,
      "grad_norm": 0.6109559780127607,
      "learning_rate": 8.601958890256082e-05,
      "loss": 11.8915,
      "step": 12770
    },
    {
      "epoch": 2.2331803278688525,
      "grad_norm": 0.6206106320348085,
      "learning_rate": 8.600556722785879e-05,
      "loss": 11.6089,
      "step": 12771
    },
    {
      "epoch": 2.233355191256831,
      "grad_norm": 0.612022913784969,
      "learning_rate": 8.599154583378816e-05,
      "loss": 11.7818,
      "step": 12772
    },
    {
      "epoch": 2.2335300546448087,
      "grad_norm": 0.716455319825564,
      "learning_rate": 8.597752472063003e-05,
      "loss": 11.8378,
      "step": 12773
    },
    {
      "epoch": 2.233704918032787,
      "grad_norm": 0.7056056848463725,
      "learning_rate": 8.596350388866558e-05,
      "loss": 11.8473,
      "step": 12774
    },
    {
      "epoch": 2.233879781420765,
      "grad_norm": 0.7412591539802561,
      "learning_rate": 8.5949483338176e-05,
      "loss": 11.9766,
      "step": 12775
    },
    {
      "epoch": 2.2340546448087433,
      "grad_norm": 0.7580725029322397,
      "learning_rate": 8.593546306944235e-05,
      "loss": 11.6266,
      "step": 12776
    },
    {
      "epoch": 2.234229508196721,
      "grad_norm": 0.6666504616479673,
      "learning_rate": 8.592144308274593e-05,
      "loss": 11.5224,
      "step": 12777
    },
    {
      "epoch": 2.2344043715846995,
      "grad_norm": 0.7845609856528075,
      "learning_rate": 8.590742337836775e-05,
      "loss": 11.7788,
      "step": 12778
    },
    {
      "epoch": 2.2345792349726774,
      "grad_norm": 0.6306915532460683,
      "learning_rate": 8.589340395658906e-05,
      "loss": 11.8383,
      "step": 12779
    },
    {
      "epoch": 2.2347540983606557,
      "grad_norm": 0.6999938761315306,
      "learning_rate": 8.587938481769089e-05,
      "loss": 11.8849,
      "step": 12780
    },
    {
      "epoch": 2.2349289617486336,
      "grad_norm": 0.6399642292600005,
      "learning_rate": 8.586536596195439e-05,
      "loss": 11.9284,
      "step": 12781
    },
    {
      "epoch": 2.235103825136612,
      "grad_norm": 0.6709450498671986,
      "learning_rate": 8.585134738966073e-05,
      "loss": 12.0344,
      "step": 12782
    },
    {
      "epoch": 2.2352786885245903,
      "grad_norm": 0.6416393072614911,
      "learning_rate": 8.5837329101091e-05,
      "loss": 11.757,
      "step": 12783
    },
    {
      "epoch": 2.2354535519125682,
      "grad_norm": 0.642547853830352,
      "learning_rate": 8.582331109652631e-05,
      "loss": 12.0271,
      "step": 12784
    },
    {
      "epoch": 2.2356284153005466,
      "grad_norm": 0.6563915578761533,
      "learning_rate": 8.580929337624775e-05,
      "loss": 11.3945,
      "step": 12785
    },
    {
      "epoch": 2.2358032786885245,
      "grad_norm": 0.6828934576297169,
      "learning_rate": 8.579527594053641e-05,
      "loss": 11.9937,
      "step": 12786
    },
    {
      "epoch": 2.235978142076503,
      "grad_norm": 0.786152683749663,
      "learning_rate": 8.578125878967342e-05,
      "loss": 12.088,
      "step": 12787
    },
    {
      "epoch": 2.2361530054644807,
      "grad_norm": 0.5979182894878525,
      "learning_rate": 8.576724192393987e-05,
      "loss": 11.7874,
      "step": 12788
    },
    {
      "epoch": 2.236327868852459,
      "grad_norm": 0.7328474076184605,
      "learning_rate": 8.57532253436168e-05,
      "loss": 11.9351,
      "step": 12789
    },
    {
      "epoch": 2.236502732240437,
      "grad_norm": 0.7073937924871355,
      "learning_rate": 8.573920904898529e-05,
      "loss": 11.7644,
      "step": 12790
    },
    {
      "epoch": 2.2366775956284153,
      "grad_norm": 0.6976624125375165,
      "learning_rate": 8.572519304032644e-05,
      "loss": 11.9985,
      "step": 12791
    },
    {
      "epoch": 2.2368524590163936,
      "grad_norm": 0.7175558567803887,
      "learning_rate": 8.57111773179213e-05,
      "loss": 11.4956,
      "step": 12792
    },
    {
      "epoch": 2.2370273224043715,
      "grad_norm": 0.6622696231140618,
      "learning_rate": 8.569716188205095e-05,
      "loss": 11.8759,
      "step": 12793
    },
    {
      "epoch": 2.23720218579235,
      "grad_norm": 0.6447576596297335,
      "learning_rate": 8.568314673299639e-05,
      "loss": 11.8824,
      "step": 12794
    },
    {
      "epoch": 2.2373770491803278,
      "grad_norm": 0.6626680212621313,
      "learning_rate": 8.566913187103868e-05,
      "loss": 11.7642,
      "step": 12795
    },
    {
      "epoch": 2.237551912568306,
      "grad_norm": 0.6207914197172969,
      "learning_rate": 8.56551172964589e-05,
      "loss": 12.0063,
      "step": 12796
    },
    {
      "epoch": 2.237726775956284,
      "grad_norm": 0.6257393704024612,
      "learning_rate": 8.564110300953808e-05,
      "loss": 11.9779,
      "step": 12797
    },
    {
      "epoch": 2.2379016393442623,
      "grad_norm": 0.6145040286489947,
      "learning_rate": 8.562708901055722e-05,
      "loss": 11.8954,
      "step": 12798
    },
    {
      "epoch": 2.2380765027322402,
      "grad_norm": 0.7050419060645432,
      "learning_rate": 8.561307529979736e-05,
      "loss": 11.891,
      "step": 12799
    },
    {
      "epoch": 2.2382513661202186,
      "grad_norm": 0.6642139124722208,
      "learning_rate": 8.559906187753947e-05,
      "loss": 11.8037,
      "step": 12800
    },
    {
      "epoch": 2.238426229508197,
      "grad_norm": 0.6651822925895813,
      "learning_rate": 8.558504874406464e-05,
      "loss": 12.1134,
      "step": 12801
    },
    {
      "epoch": 2.238601092896175,
      "grad_norm": 0.6298325399259073,
      "learning_rate": 8.557103589965387e-05,
      "loss": 11.6024,
      "step": 12802
    },
    {
      "epoch": 2.238775956284153,
      "grad_norm": 0.6365343752510815,
      "learning_rate": 8.555702334458811e-05,
      "loss": 12.1992,
      "step": 12803
    },
    {
      "epoch": 2.238950819672131,
      "grad_norm": 0.5988896320311486,
      "learning_rate": 8.554301107914837e-05,
      "loss": 11.8914,
      "step": 12804
    },
    {
      "epoch": 2.2391256830601094,
      "grad_norm": 0.6069848545321078,
      "learning_rate": 8.552899910361562e-05,
      "loss": 12.0435,
      "step": 12805
    },
    {
      "epoch": 2.2393005464480873,
      "grad_norm": 0.6741522565893383,
      "learning_rate": 8.551498741827094e-05,
      "loss": 11.9867,
      "step": 12806
    },
    {
      "epoch": 2.2394754098360656,
      "grad_norm": 0.6590958110101739,
      "learning_rate": 8.550097602339518e-05,
      "loss": 11.7349,
      "step": 12807
    },
    {
      "epoch": 2.2396502732240435,
      "grad_norm": 0.6735379372207002,
      "learning_rate": 8.548696491926939e-05,
      "loss": 11.9055,
      "step": 12808
    },
    {
      "epoch": 2.239825136612022,
      "grad_norm": 0.6987862663544011,
      "learning_rate": 8.547295410617453e-05,
      "loss": 12.0314,
      "step": 12809
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.706629291054325,
      "learning_rate": 8.545894358439148e-05,
      "loss": 12.0215,
      "step": 12810
    },
    {
      "epoch": 2.240174863387978,
      "grad_norm": 0.6760476653392928,
      "learning_rate": 8.544493335420133e-05,
      "loss": 11.9269,
      "step": 12811
    },
    {
      "epoch": 2.2403497267759565,
      "grad_norm": 0.7786413922355074,
      "learning_rate": 8.543092341588492e-05,
      "loss": 11.9553,
      "step": 12812
    },
    {
      "epoch": 2.2405245901639343,
      "grad_norm": 0.6417527408043628,
      "learning_rate": 8.541691376972325e-05,
      "loss": 11.8098,
      "step": 12813
    },
    {
      "epoch": 2.2406994535519127,
      "grad_norm": 0.7165806413789465,
      "learning_rate": 8.540290441599719e-05,
      "loss": 11.8682,
      "step": 12814
    },
    {
      "epoch": 2.2408743169398906,
      "grad_norm": 0.6400758824834324,
      "learning_rate": 8.538889535498777e-05,
      "loss": 11.8878,
      "step": 12815
    },
    {
      "epoch": 2.241049180327869,
      "grad_norm": 0.7672488037373366,
      "learning_rate": 8.537488658697584e-05,
      "loss": 11.9935,
      "step": 12816
    },
    {
      "epoch": 2.241224043715847,
      "grad_norm": 0.7449922799644404,
      "learning_rate": 8.536087811224234e-05,
      "loss": 11.9108,
      "step": 12817
    },
    {
      "epoch": 2.241398907103825,
      "grad_norm": 0.6969999764718567,
      "learning_rate": 8.53468699310682e-05,
      "loss": 12.0088,
      "step": 12818
    },
    {
      "epoch": 2.2415737704918035,
      "grad_norm": 0.7284289879327512,
      "learning_rate": 8.533286204373424e-05,
      "loss": 11.8004,
      "step": 12819
    },
    {
      "epoch": 2.2417486338797814,
      "grad_norm": 0.5845175073239458,
      "learning_rate": 8.53188544505215e-05,
      "loss": 11.861,
      "step": 12820
    },
    {
      "epoch": 2.2419234972677597,
      "grad_norm": 0.6402725222790215,
      "learning_rate": 8.530484715171078e-05,
      "loss": 11.9528,
      "step": 12821
    },
    {
      "epoch": 2.2420983606557376,
      "grad_norm": 0.6767686000964196,
      "learning_rate": 8.529084014758299e-05,
      "loss": 11.9838,
      "step": 12822
    },
    {
      "epoch": 2.242273224043716,
      "grad_norm": 0.7284989898752557,
      "learning_rate": 8.527683343841906e-05,
      "loss": 11.6725,
      "step": 12823
    },
    {
      "epoch": 2.242448087431694,
      "grad_norm": 0.6980990418627016,
      "learning_rate": 8.526282702449977e-05,
      "loss": 11.6022,
      "step": 12824
    },
    {
      "epoch": 2.2426229508196722,
      "grad_norm": 0.6433942438429492,
      "learning_rate": 8.524882090610607e-05,
      "loss": 12.0141,
      "step": 12825
    },
    {
      "epoch": 2.24279781420765,
      "grad_norm": 0.5934987294845767,
      "learning_rate": 8.523481508351879e-05,
      "loss": 12.006,
      "step": 12826
    },
    {
      "epoch": 2.2429726775956285,
      "grad_norm": 0.6490193893613645,
      "learning_rate": 8.522080955701883e-05,
      "loss": 11.8054,
      "step": 12827
    },
    {
      "epoch": 2.2431475409836064,
      "grad_norm": 0.6638663132648622,
      "learning_rate": 8.520680432688702e-05,
      "loss": 12.0317,
      "step": 12828
    },
    {
      "epoch": 2.2433224043715847,
      "grad_norm": 0.6185112228521866,
      "learning_rate": 8.519279939340414e-05,
      "loss": 11.6939,
      "step": 12829
    },
    {
      "epoch": 2.243497267759563,
      "grad_norm": 0.7049708071161148,
      "learning_rate": 8.517879475685115e-05,
      "loss": 12.0937,
      "step": 12830
    },
    {
      "epoch": 2.243672131147541,
      "grad_norm": 0.6148385630829368,
      "learning_rate": 8.516479041750882e-05,
      "loss": 11.9838,
      "step": 12831
    },
    {
      "epoch": 2.2438469945355193,
      "grad_norm": 0.7106426424057461,
      "learning_rate": 8.515078637565801e-05,
      "loss": 11.8633,
      "step": 12832
    },
    {
      "epoch": 2.244021857923497,
      "grad_norm": 0.6947020157274757,
      "learning_rate": 8.513678263157952e-05,
      "loss": 12.0418,
      "step": 12833
    },
    {
      "epoch": 2.2441967213114755,
      "grad_norm": 0.6388198291717538,
      "learning_rate": 8.512277918555415e-05,
      "loss": 11.9445,
      "step": 12834
    },
    {
      "epoch": 2.2443715846994534,
      "grad_norm": 0.634251912839842,
      "learning_rate": 8.510877603786275e-05,
      "loss": 11.9192,
      "step": 12835
    },
    {
      "epoch": 2.2445464480874318,
      "grad_norm": 0.6277519714179507,
      "learning_rate": 8.509477318878613e-05,
      "loss": 11.8944,
      "step": 12836
    },
    {
      "epoch": 2.2447213114754097,
      "grad_norm": 0.6476647441356309,
      "learning_rate": 8.508077063860506e-05,
      "loss": 11.7992,
      "step": 12837
    },
    {
      "epoch": 2.244896174863388,
      "grad_norm": 0.704675401385058,
      "learning_rate": 8.506676838760032e-05,
      "loss": 11.9071,
      "step": 12838
    },
    {
      "epoch": 2.245071038251366,
      "grad_norm": 0.6900746713810103,
      "learning_rate": 8.505276643605275e-05,
      "loss": 11.8036,
      "step": 12839
    },
    {
      "epoch": 2.2452459016393442,
      "grad_norm": 0.7184774955357949,
      "learning_rate": 8.50387647842431e-05,
      "loss": 11.9536,
      "step": 12840
    },
    {
      "epoch": 2.2454207650273226,
      "grad_norm": 0.7390935933949163,
      "learning_rate": 8.502476343245217e-05,
      "loss": 11.7501,
      "step": 12841
    },
    {
      "epoch": 2.2455956284153005,
      "grad_norm": 0.786415027485568,
      "learning_rate": 8.501076238096072e-05,
      "loss": 11.6499,
      "step": 12842
    },
    {
      "epoch": 2.245770491803279,
      "grad_norm": 0.7070726115509582,
      "learning_rate": 8.499676163004946e-05,
      "loss": 11.9079,
      "step": 12843
    },
    {
      "epoch": 2.2459453551912567,
      "grad_norm": 0.6867508160432928,
      "learning_rate": 8.498276117999922e-05,
      "loss": 11.8245,
      "step": 12844
    },
    {
      "epoch": 2.246120218579235,
      "grad_norm": 0.7282460945744517,
      "learning_rate": 8.496876103109076e-05,
      "loss": 11.7655,
      "step": 12845
    },
    {
      "epoch": 2.246295081967213,
      "grad_norm": 0.7293564149812172,
      "learning_rate": 8.495476118360477e-05,
      "loss": 11.8626,
      "step": 12846
    },
    {
      "epoch": 2.2464699453551913,
      "grad_norm": 0.7158332456244406,
      "learning_rate": 8.494076163782201e-05,
      "loss": 12.0081,
      "step": 12847
    },
    {
      "epoch": 2.246644808743169,
      "grad_norm": 0.7948365900126403,
      "learning_rate": 8.492676239402319e-05,
      "loss": 11.7436,
      "step": 12848
    },
    {
      "epoch": 2.2468196721311475,
      "grad_norm": 0.646811141312975,
      "learning_rate": 8.491276345248908e-05,
      "loss": 11.8701,
      "step": 12849
    },
    {
      "epoch": 2.246994535519126,
      "grad_norm": 0.7689010940644949,
      "learning_rate": 8.489876481350043e-05,
      "loss": 12.0838,
      "step": 12850
    },
    {
      "epoch": 2.2471693989071038,
      "grad_norm": 0.8546424560111039,
      "learning_rate": 8.488476647733788e-05,
      "loss": 11.9183,
      "step": 12851
    },
    {
      "epoch": 2.247344262295082,
      "grad_norm": 0.6351104939734813,
      "learning_rate": 8.487076844428217e-05,
      "loss": 12.0573,
      "step": 12852
    },
    {
      "epoch": 2.24751912568306,
      "grad_norm": 0.7218406094233442,
      "learning_rate": 8.485677071461398e-05,
      "loss": 12.0757,
      "step": 12853
    },
    {
      "epoch": 2.2476939890710383,
      "grad_norm": 0.6518763407236815,
      "learning_rate": 8.484277328861409e-05,
      "loss": 12.0086,
      "step": 12854
    },
    {
      "epoch": 2.2478688524590162,
      "grad_norm": 0.611014295864099,
      "learning_rate": 8.48287761665631e-05,
      "loss": 11.511,
      "step": 12855
    },
    {
      "epoch": 2.2480437158469946,
      "grad_norm": 0.6215930790885612,
      "learning_rate": 8.481477934874173e-05,
      "loss": 11.6074,
      "step": 12856
    },
    {
      "epoch": 2.2482185792349725,
      "grad_norm": 0.7256332482689796,
      "learning_rate": 8.480078283543069e-05,
      "loss": 11.7631,
      "step": 12857
    },
    {
      "epoch": 2.248393442622951,
      "grad_norm": 0.67041531631559,
      "learning_rate": 8.478678662691055e-05,
      "loss": 11.9808,
      "step": 12858
    },
    {
      "epoch": 2.248568306010929,
      "grad_norm": 0.763759899635597,
      "learning_rate": 8.477279072346211e-05,
      "loss": 12.0813,
      "step": 12859
    },
    {
      "epoch": 2.248743169398907,
      "grad_norm": 0.6680335006329534,
      "learning_rate": 8.475879512536596e-05,
      "loss": 11.5576,
      "step": 12860
    },
    {
      "epoch": 2.2489180327868854,
      "grad_norm": 0.733844009611437,
      "learning_rate": 8.474479983290275e-05,
      "loss": 12.11,
      "step": 12861
    },
    {
      "epoch": 2.2490928961748633,
      "grad_norm": 0.6705557321822603,
      "learning_rate": 8.473080484635314e-05,
      "loss": 11.9186,
      "step": 12862
    },
    {
      "epoch": 2.2492677595628416,
      "grad_norm": 0.692524420996452,
      "learning_rate": 8.47168101659978e-05,
      "loss": 11.6375,
      "step": 12863
    },
    {
      "epoch": 2.2494426229508195,
      "grad_norm": 0.7694563385125216,
      "learning_rate": 8.470281579211733e-05,
      "loss": 11.9,
      "step": 12864
    },
    {
      "epoch": 2.249617486338798,
      "grad_norm": 0.7008531507505197,
      "learning_rate": 8.468882172499237e-05,
      "loss": 11.9645,
      "step": 12865
    },
    {
      "epoch": 2.2497923497267758,
      "grad_norm": 0.619669071942225,
      "learning_rate": 8.467482796490357e-05,
      "loss": 11.6577,
      "step": 12866
    },
    {
      "epoch": 2.249967213114754,
      "grad_norm": 0.7912449766368769,
      "learning_rate": 8.466083451213144e-05,
      "loss": 11.8452,
      "step": 12867
    },
    {
      "epoch": 2.2501420765027325,
      "grad_norm": 0.7118754944287327,
      "learning_rate": 8.464684136695677e-05,
      "loss": 11.9546,
      "step": 12868
    },
    {
      "epoch": 2.2503169398907104,
      "grad_norm": 0.6836106374920784,
      "learning_rate": 8.463284852966004e-05,
      "loss": 12.0789,
      "step": 12869
    },
    {
      "epoch": 2.2504918032786887,
      "grad_norm": 0.7746249838040454,
      "learning_rate": 8.461885600052186e-05,
      "loss": 11.9517,
      "step": 12870
    },
    {
      "epoch": 2.2506666666666666,
      "grad_norm": 0.6432704111076268,
      "learning_rate": 8.460486377982289e-05,
      "loss": 12.1615,
      "step": 12871
    },
    {
      "epoch": 2.250841530054645,
      "grad_norm": 0.7152096761510796,
      "learning_rate": 8.459087186784363e-05,
      "loss": 12.0408,
      "step": 12872
    },
    {
      "epoch": 2.251016393442623,
      "grad_norm": 0.6495429338303458,
      "learning_rate": 8.45768802648647e-05,
      "loss": 11.7123,
      "step": 12873
    },
    {
      "epoch": 2.251191256830601,
      "grad_norm": 0.6142722877913197,
      "learning_rate": 8.456288897116669e-05,
      "loss": 11.6684,
      "step": 12874
    },
    {
      "epoch": 2.251366120218579,
      "grad_norm": 0.6473562130550282,
      "learning_rate": 8.454889798703016e-05,
      "loss": 11.7942,
      "step": 12875
    },
    {
      "epoch": 2.2515409836065574,
      "grad_norm": 0.623709269552421,
      "learning_rate": 8.453490731273567e-05,
      "loss": 11.7631,
      "step": 12876
    },
    {
      "epoch": 2.2517158469945358,
      "grad_norm": 0.6129992226255268,
      "learning_rate": 8.452091694856373e-05,
      "loss": 11.84,
      "step": 12877
    },
    {
      "epoch": 2.2518907103825136,
      "grad_norm": 0.6208516771474533,
      "learning_rate": 8.450692689479497e-05,
      "loss": 11.7139,
      "step": 12878
    },
    {
      "epoch": 2.252065573770492,
      "grad_norm": 0.6813386604903862,
      "learning_rate": 8.449293715170989e-05,
      "loss": 11.9535,
      "step": 12879
    },
    {
      "epoch": 2.25224043715847,
      "grad_norm": 0.6490278942841279,
      "learning_rate": 8.447894771958906e-05,
      "loss": 12.0144,
      "step": 12880
    },
    {
      "epoch": 2.2524153005464482,
      "grad_norm": 0.6186380615629712,
      "learning_rate": 8.446495859871294e-05,
      "loss": 11.638,
      "step": 12881
    },
    {
      "epoch": 2.252590163934426,
      "grad_norm": 0.6960402509555472,
      "learning_rate": 8.44509697893621e-05,
      "loss": 11.7873,
      "step": 12882
    },
    {
      "epoch": 2.2527650273224045,
      "grad_norm": 0.690568540887886,
      "learning_rate": 8.443698129181708e-05,
      "loss": 12.0887,
      "step": 12883
    },
    {
      "epoch": 2.2529398907103824,
      "grad_norm": 0.6977243819198408,
      "learning_rate": 8.442299310635838e-05,
      "loss": 12.0265,
      "step": 12884
    },
    {
      "epoch": 2.2531147540983607,
      "grad_norm": 0.6801144158073077,
      "learning_rate": 8.440900523326647e-05,
      "loss": 11.974,
      "step": 12885
    },
    {
      "epoch": 2.253289617486339,
      "grad_norm": 0.6580131979032329,
      "learning_rate": 8.439501767282186e-05,
      "loss": 11.7148,
      "step": 12886
    },
    {
      "epoch": 2.253464480874317,
      "grad_norm": 0.7125296840142278,
      "learning_rate": 8.438103042530509e-05,
      "loss": 11.9886,
      "step": 12887
    },
    {
      "epoch": 2.253639344262295,
      "grad_norm": 0.8932434553578751,
      "learning_rate": 8.43670434909966e-05,
      "loss": 11.9293,
      "step": 12888
    },
    {
      "epoch": 2.253814207650273,
      "grad_norm": 0.6860418298619353,
      "learning_rate": 8.435305687017691e-05,
      "loss": 11.7061,
      "step": 12889
    },
    {
      "epoch": 2.2539890710382515,
      "grad_norm": 0.862514127236325,
      "learning_rate": 8.433907056312646e-05,
      "loss": 11.9453,
      "step": 12890
    },
    {
      "epoch": 2.2541639344262294,
      "grad_norm": 0.667648930737012,
      "learning_rate": 8.432508457012571e-05,
      "loss": 11.8545,
      "step": 12891
    },
    {
      "epoch": 2.2543387978142078,
      "grad_norm": 0.7231736693326754,
      "learning_rate": 8.431109889145516e-05,
      "loss": 12.0276,
      "step": 12892
    },
    {
      "epoch": 2.2545136612021857,
      "grad_norm": 0.7836400605889691,
      "learning_rate": 8.429711352739527e-05,
      "loss": 12.0668,
      "step": 12893
    },
    {
      "epoch": 2.254688524590164,
      "grad_norm": 0.6608487851314746,
      "learning_rate": 8.428312847822646e-05,
      "loss": 11.8495,
      "step": 12894
    },
    {
      "epoch": 2.254863387978142,
      "grad_norm": 0.7463735392974081,
      "learning_rate": 8.426914374422917e-05,
      "loss": 12.0426,
      "step": 12895
    },
    {
      "epoch": 2.2550382513661202,
      "grad_norm": 0.6142821528100586,
      "learning_rate": 8.425515932568382e-05,
      "loss": 11.9873,
      "step": 12896
    },
    {
      "epoch": 2.255213114754098,
      "grad_norm": 0.7084436106007417,
      "learning_rate": 8.424117522287089e-05,
      "loss": 11.8131,
      "step": 12897
    },
    {
      "epoch": 2.2553879781420765,
      "grad_norm": 0.6861283170863884,
      "learning_rate": 8.422719143607082e-05,
      "loss": 11.8045,
      "step": 12898
    },
    {
      "epoch": 2.255562841530055,
      "grad_norm": 0.651911530307453,
      "learning_rate": 8.421320796556396e-05,
      "loss": 11.9567,
      "step": 12899
    },
    {
      "epoch": 2.2557377049180327,
      "grad_norm": 0.73664471057771,
      "learning_rate": 8.419922481163075e-05,
      "loss": 12.0801,
      "step": 12900
    },
    {
      "epoch": 2.255912568306011,
      "grad_norm": 0.6391465275274141,
      "learning_rate": 8.418524197455158e-05,
      "loss": 12.0635,
      "step": 12901
    },
    {
      "epoch": 2.256087431693989,
      "grad_norm": 0.6993566474708857,
      "learning_rate": 8.41712594546069e-05,
      "loss": 11.9576,
      "step": 12902
    },
    {
      "epoch": 2.2562622950819673,
      "grad_norm": 0.7314003440121017,
      "learning_rate": 8.415727725207706e-05,
      "loss": 12.0511,
      "step": 12903
    },
    {
      "epoch": 2.256437158469945,
      "grad_norm": 0.7121904333795039,
      "learning_rate": 8.414329536724245e-05,
      "loss": 11.973,
      "step": 12904
    },
    {
      "epoch": 2.2566120218579235,
      "grad_norm": 0.8872024084624411,
      "learning_rate": 8.412931380038347e-05,
      "loss": 11.9344,
      "step": 12905
    },
    {
      "epoch": 2.2567868852459014,
      "grad_norm": 0.68804111371826,
      "learning_rate": 8.411533255178041e-05,
      "loss": 11.7457,
      "step": 12906
    },
    {
      "epoch": 2.2569617486338798,
      "grad_norm": 0.6591644990227385,
      "learning_rate": 8.410135162171379e-05,
      "loss": 11.974,
      "step": 12907
    },
    {
      "epoch": 2.257136612021858,
      "grad_norm": 0.7291900385118307,
      "learning_rate": 8.408737101046384e-05,
      "loss": 11.7971,
      "step": 12908
    },
    {
      "epoch": 2.257311475409836,
      "grad_norm": 0.6289128284637275,
      "learning_rate": 8.407339071831097e-05,
      "loss": 11.7942,
      "step": 12909
    },
    {
      "epoch": 2.2574863387978144,
      "grad_norm": 0.6353941094341873,
      "learning_rate": 8.40594107455355e-05,
      "loss": 12.0766,
      "step": 12910
    },
    {
      "epoch": 2.2576612021857922,
      "grad_norm": 0.6847667975420999,
      "learning_rate": 8.404543109241782e-05,
      "loss": 11.8435,
      "step": 12911
    },
    {
      "epoch": 2.2578360655737706,
      "grad_norm": 0.6153850902397306,
      "learning_rate": 8.403145175923822e-05,
      "loss": 11.8727,
      "step": 12912
    },
    {
      "epoch": 2.2580109289617485,
      "grad_norm": 0.6957780963644559,
      "learning_rate": 8.401747274627705e-05,
      "loss": 11.845,
      "step": 12913
    },
    {
      "epoch": 2.258185792349727,
      "grad_norm": 0.6605729969924533,
      "learning_rate": 8.400349405381464e-05,
      "loss": 11.9755,
      "step": 12914
    },
    {
      "epoch": 2.2583606557377047,
      "grad_norm": 0.6962878679142533,
      "learning_rate": 8.398951568213122e-05,
      "loss": 11.9273,
      "step": 12915
    },
    {
      "epoch": 2.258535519125683,
      "grad_norm": 0.6445374520224163,
      "learning_rate": 8.397553763150723e-05,
      "loss": 11.94,
      "step": 12916
    },
    {
      "epoch": 2.2587103825136614,
      "grad_norm": 0.665883554473628,
      "learning_rate": 8.39615599022229e-05,
      "loss": 12.033,
      "step": 12917
    },
    {
      "epoch": 2.2588852459016393,
      "grad_norm": 0.6287216945033061,
      "learning_rate": 8.394758249455853e-05,
      "loss": 11.8571,
      "step": 12918
    },
    {
      "epoch": 2.2590601092896176,
      "grad_norm": 0.6823748583946353,
      "learning_rate": 8.393360540879445e-05,
      "loss": 11.4788,
      "step": 12919
    },
    {
      "epoch": 2.2592349726775955,
      "grad_norm": 0.7311260804779179,
      "learning_rate": 8.391962864521086e-05,
      "loss": 11.7808,
      "step": 12920
    },
    {
      "epoch": 2.259409836065574,
      "grad_norm": 0.6452143350849769,
      "learning_rate": 8.390565220408812e-05,
      "loss": 11.8388,
      "step": 12921
    },
    {
      "epoch": 2.259584699453552,
      "grad_norm": 0.6349164445822238,
      "learning_rate": 8.389167608570645e-05,
      "loss": 12.0434,
      "step": 12922
    },
    {
      "epoch": 2.25975956284153,
      "grad_norm": 0.6718772762957652,
      "learning_rate": 8.387770029034618e-05,
      "loss": 11.8662,
      "step": 12923
    },
    {
      "epoch": 2.259934426229508,
      "grad_norm": 0.6367523847894855,
      "learning_rate": 8.386372481828748e-05,
      "loss": 11.9789,
      "step": 12924
    },
    {
      "epoch": 2.2601092896174864,
      "grad_norm": 0.7285403204529122,
      "learning_rate": 8.384974966981063e-05,
      "loss": 11.9588,
      "step": 12925
    },
    {
      "epoch": 2.2602841530054647,
      "grad_norm": 0.825097083135792,
      "learning_rate": 8.383577484519591e-05,
      "loss": 11.8946,
      "step": 12926
    },
    {
      "epoch": 2.2604590163934426,
      "grad_norm": 0.667930791263118,
      "learning_rate": 8.382180034472353e-05,
      "loss": 11.8161,
      "step": 12927
    },
    {
      "epoch": 2.260633879781421,
      "grad_norm": 0.6251224660147181,
      "learning_rate": 8.380782616867375e-05,
      "loss": 11.7781,
      "step": 12928
    },
    {
      "epoch": 2.260808743169399,
      "grad_norm": 0.6748691640870466,
      "learning_rate": 8.379385231732674e-05,
      "loss": 11.9807,
      "step": 12929
    },
    {
      "epoch": 2.260983606557377,
      "grad_norm": 0.7233804502083414,
      "learning_rate": 8.377987879096275e-05,
      "loss": 11.823,
      "step": 12930
    },
    {
      "epoch": 2.261158469945355,
      "grad_norm": 0.7144187577038146,
      "learning_rate": 8.3765905589862e-05,
      "loss": 11.672,
      "step": 12931
    },
    {
      "epoch": 2.2613333333333334,
      "grad_norm": 0.6697910224048366,
      "learning_rate": 8.375193271430472e-05,
      "loss": 11.8456,
      "step": 12932
    },
    {
      "epoch": 2.2615081967213113,
      "grad_norm": 0.6648959384277682,
      "learning_rate": 8.373796016457105e-05,
      "loss": 11.4497,
      "step": 12933
    },
    {
      "epoch": 2.2616830601092897,
      "grad_norm": 0.7262986663284201,
      "learning_rate": 8.372398794094118e-05,
      "loss": 11.9185,
      "step": 12934
    },
    {
      "epoch": 2.261857923497268,
      "grad_norm": 0.6840787265861445,
      "learning_rate": 8.371001604369537e-05,
      "loss": 11.9152,
      "step": 12935
    },
    {
      "epoch": 2.262032786885246,
      "grad_norm": 0.6943918084401617,
      "learning_rate": 8.369604447311373e-05,
      "loss": 12.0227,
      "step": 12936
    },
    {
      "epoch": 2.2622076502732242,
      "grad_norm": 0.6760752208136802,
      "learning_rate": 8.368207322947648e-05,
      "loss": 11.9663,
      "step": 12937
    },
    {
      "epoch": 2.262382513661202,
      "grad_norm": 0.6325426998319756,
      "learning_rate": 8.366810231306377e-05,
      "loss": 11.8908,
      "step": 12938
    },
    {
      "epoch": 2.2625573770491805,
      "grad_norm": 0.6524370290777871,
      "learning_rate": 8.365413172415571e-05,
      "loss": 11.8259,
      "step": 12939
    },
    {
      "epoch": 2.2627322404371584,
      "grad_norm": 0.6689461252124508,
      "learning_rate": 8.364016146303251e-05,
      "loss": 11.6439,
      "step": 12940
    },
    {
      "epoch": 2.2629071038251367,
      "grad_norm": 0.6680844466082192,
      "learning_rate": 8.362619152997435e-05,
      "loss": 11.6832,
      "step": 12941
    },
    {
      "epoch": 2.2630819672131146,
      "grad_norm": 0.6016739438410572,
      "learning_rate": 8.361222192526129e-05,
      "loss": 11.8589,
      "step": 12942
    },
    {
      "epoch": 2.263256830601093,
      "grad_norm": 0.6356320409533323,
      "learning_rate": 8.359825264917351e-05,
      "loss": 11.7875,
      "step": 12943
    },
    {
      "epoch": 2.263431693989071,
      "grad_norm": 0.6584523896881066,
      "learning_rate": 8.358428370199109e-05,
      "loss": 11.8999,
      "step": 12944
    },
    {
      "epoch": 2.263606557377049,
      "grad_norm": 0.6511952218541429,
      "learning_rate": 8.35703150839942e-05,
      "loss": 11.9199,
      "step": 12945
    },
    {
      "epoch": 2.263781420765027,
      "grad_norm": 0.6952782067605975,
      "learning_rate": 8.355634679546296e-05,
      "loss": 11.7332,
      "step": 12946
    },
    {
      "epoch": 2.2639562841530054,
      "grad_norm": 0.6716477415237638,
      "learning_rate": 8.354237883667745e-05,
      "loss": 11.6525,
      "step": 12947
    },
    {
      "epoch": 2.2641311475409838,
      "grad_norm": 0.7518839074405327,
      "learning_rate": 8.352841120791776e-05,
      "loss": 11.6632,
      "step": 12948
    },
    {
      "epoch": 2.2643060109289617,
      "grad_norm": 0.6611780139613156,
      "learning_rate": 8.3514443909464e-05,
      "loss": 11.8378,
      "step": 12949
    },
    {
      "epoch": 2.26448087431694,
      "grad_norm": 0.6762253711504426,
      "learning_rate": 8.350047694159628e-05,
      "loss": 11.7777,
      "step": 12950
    },
    {
      "epoch": 2.264655737704918,
      "grad_norm": 0.6987804318715464,
      "learning_rate": 8.348651030459465e-05,
      "loss": 12.0802,
      "step": 12951
    },
    {
      "epoch": 2.2648306010928962,
      "grad_norm": 0.6429607002752251,
      "learning_rate": 8.347254399873917e-05,
      "loss": 11.9465,
      "step": 12952
    },
    {
      "epoch": 2.265005464480874,
      "grad_norm": 0.7159056080626424,
      "learning_rate": 8.345857802430996e-05,
      "loss": 11.477,
      "step": 12953
    },
    {
      "epoch": 2.2651803278688525,
      "grad_norm": 0.6447197905287839,
      "learning_rate": 8.344461238158699e-05,
      "loss": 11.8558,
      "step": 12954
    },
    {
      "epoch": 2.2653551912568304,
      "grad_norm": 0.6542757222120665,
      "learning_rate": 8.343064707085044e-05,
      "loss": 11.6647,
      "step": 12955
    },
    {
      "epoch": 2.2655300546448087,
      "grad_norm": 0.6908194659769307,
      "learning_rate": 8.341668209238027e-05,
      "loss": 11.764,
      "step": 12956
    },
    {
      "epoch": 2.265704918032787,
      "grad_norm": 0.7281873398603448,
      "learning_rate": 8.340271744645653e-05,
      "loss": 11.7591,
      "step": 12957
    },
    {
      "epoch": 2.265879781420765,
      "grad_norm": 0.7034562519458992,
      "learning_rate": 8.338875313335926e-05,
      "loss": 12.0087,
      "step": 12958
    },
    {
      "epoch": 2.2660546448087433,
      "grad_norm": 0.6289822970866009,
      "learning_rate": 8.337478915336853e-05,
      "loss": 12.0739,
      "step": 12959
    },
    {
      "epoch": 2.266229508196721,
      "grad_norm": 0.5870037179281778,
      "learning_rate": 8.336082550676429e-05,
      "loss": 11.744,
      "step": 12960
    },
    {
      "epoch": 2.2664043715846995,
      "grad_norm": 0.6932310995520415,
      "learning_rate": 8.33468621938266e-05,
      "loss": 11.8216,
      "step": 12961
    },
    {
      "epoch": 2.2665792349726774,
      "grad_norm": 0.7653373294566302,
      "learning_rate": 8.333289921483547e-05,
      "loss": 12.0792,
      "step": 12962
    },
    {
      "epoch": 2.266754098360656,
      "grad_norm": 0.6200153671779917,
      "learning_rate": 8.331893657007082e-05,
      "loss": 11.9454,
      "step": 12963
    },
    {
      "epoch": 2.2669289617486337,
      "grad_norm": 0.6716184638188041,
      "learning_rate": 8.330497425981279e-05,
      "loss": 11.9779,
      "step": 12964
    },
    {
      "epoch": 2.267103825136612,
      "grad_norm": 0.6927146626627306,
      "learning_rate": 8.329101228434124e-05,
      "loss": 11.6494,
      "step": 12965
    },
    {
      "epoch": 2.2672786885245904,
      "grad_norm": 0.7169389589552219,
      "learning_rate": 8.327705064393621e-05,
      "loss": 12.0976,
      "step": 12966
    },
    {
      "epoch": 2.2674535519125683,
      "grad_norm": 0.7417742002261916,
      "learning_rate": 8.326308933887769e-05,
      "loss": 11.7568,
      "step": 12967
    },
    {
      "epoch": 2.2676284153005466,
      "grad_norm": 0.6461683185786578,
      "learning_rate": 8.324912836944556e-05,
      "loss": 11.8255,
      "step": 12968
    },
    {
      "epoch": 2.2678032786885245,
      "grad_norm": 0.6703286230027821,
      "learning_rate": 8.323516773591988e-05,
      "loss": 11.9752,
      "step": 12969
    },
    {
      "epoch": 2.267978142076503,
      "grad_norm": 0.7104044602878956,
      "learning_rate": 8.322120743858054e-05,
      "loss": 11.7448,
      "step": 12970
    },
    {
      "epoch": 2.2681530054644807,
      "grad_norm": 0.6890178187001905,
      "learning_rate": 8.320724747770753e-05,
      "loss": 11.6967,
      "step": 12971
    },
    {
      "epoch": 2.268327868852459,
      "grad_norm": 0.761964072541549,
      "learning_rate": 8.319328785358078e-05,
      "loss": 11.8532,
      "step": 12972
    },
    {
      "epoch": 2.268502732240437,
      "grad_norm": 0.7964587717502749,
      "learning_rate": 8.317932856648015e-05,
      "loss": 11.9817,
      "step": 12973
    },
    {
      "epoch": 2.2686775956284153,
      "grad_norm": 0.6428584957926315,
      "learning_rate": 8.316536961668567e-05,
      "loss": 12.0633,
      "step": 12974
    },
    {
      "epoch": 2.2688524590163937,
      "grad_norm": 0.8162304516760271,
      "learning_rate": 8.315141100447721e-05,
      "loss": 12.1602,
      "step": 12975
    },
    {
      "epoch": 2.2690273224043715,
      "grad_norm": 0.7266735714199689,
      "learning_rate": 8.313745273013473e-05,
      "loss": 11.5022,
      "step": 12976
    },
    {
      "epoch": 2.26920218579235,
      "grad_norm": 0.638313270558629,
      "learning_rate": 8.312349479393805e-05,
      "loss": 11.8474,
      "step": 12977
    },
    {
      "epoch": 2.269377049180328,
      "grad_norm": 0.6865946079561902,
      "learning_rate": 8.310953719616711e-05,
      "loss": 11.9207,
      "step": 12978
    },
    {
      "epoch": 2.269551912568306,
      "grad_norm": 0.7271999207463787,
      "learning_rate": 8.309557993710183e-05,
      "loss": 12.1012,
      "step": 12979
    },
    {
      "epoch": 2.269726775956284,
      "grad_norm": 0.678340436288341,
      "learning_rate": 8.30816230170221e-05,
      "loss": 11.7868,
      "step": 12980
    },
    {
      "epoch": 2.2699016393442624,
      "grad_norm": 0.738262566057061,
      "learning_rate": 8.306766643620774e-05,
      "loss": 12.0896,
      "step": 12981
    },
    {
      "epoch": 2.2700765027322403,
      "grad_norm": 0.7844740839120022,
      "learning_rate": 8.305371019493866e-05,
      "loss": 11.8482,
      "step": 12982
    },
    {
      "epoch": 2.2702513661202186,
      "grad_norm": 0.6852152602886301,
      "learning_rate": 8.303975429349473e-05,
      "loss": 11.7876,
      "step": 12983
    },
    {
      "epoch": 2.270426229508197,
      "grad_norm": 0.7613559833489445,
      "learning_rate": 8.30257987321558e-05,
      "loss": 11.9372,
      "step": 12984
    },
    {
      "epoch": 2.270601092896175,
      "grad_norm": 0.8154125901443441,
      "learning_rate": 8.301184351120175e-05,
      "loss": 12.1962,
      "step": 12985
    },
    {
      "epoch": 2.270775956284153,
      "grad_norm": 0.7672216961629809,
      "learning_rate": 8.299788863091236e-05,
      "loss": 11.8082,
      "step": 12986
    },
    {
      "epoch": 2.270950819672131,
      "grad_norm": 0.6730196813927078,
      "learning_rate": 8.298393409156751e-05,
      "loss": 11.9263,
      "step": 12987
    },
    {
      "epoch": 2.2711256830601094,
      "grad_norm": 0.6338037732733587,
      "learning_rate": 8.296997989344704e-05,
      "loss": 12.0699,
      "step": 12988
    },
    {
      "epoch": 2.2713005464480873,
      "grad_norm": 0.7705507962416338,
      "learning_rate": 8.295602603683077e-05,
      "loss": 11.9716,
      "step": 12989
    },
    {
      "epoch": 2.2714754098360657,
      "grad_norm": 0.6631273617988891,
      "learning_rate": 8.29420725219985e-05,
      "loss": 11.912,
      "step": 12990
    },
    {
      "epoch": 2.2716502732240436,
      "grad_norm": 0.6492031670979228,
      "learning_rate": 8.292811934923005e-05,
      "loss": 11.8569,
      "step": 12991
    },
    {
      "epoch": 2.271825136612022,
      "grad_norm": 0.6929075899025108,
      "learning_rate": 8.29141665188052e-05,
      "loss": 11.9328,
      "step": 12992
    },
    {
      "epoch": 2.2720000000000002,
      "grad_norm": 0.6490813665398668,
      "learning_rate": 8.29002140310038e-05,
      "loss": 12.0155,
      "step": 12993
    },
    {
      "epoch": 2.272174863387978,
      "grad_norm": 0.6239455399174898,
      "learning_rate": 8.288626188610562e-05,
      "loss": 11.9673,
      "step": 12994
    },
    {
      "epoch": 2.272349726775956,
      "grad_norm": 0.8349079040339455,
      "learning_rate": 8.287231008439043e-05,
      "loss": 11.714,
      "step": 12995
    },
    {
      "epoch": 2.2725245901639344,
      "grad_norm": 0.7256586864529393,
      "learning_rate": 8.2858358626138e-05,
      "loss": 11.9994,
      "step": 12996
    },
    {
      "epoch": 2.2726994535519127,
      "grad_norm": 0.7034463462155829,
      "learning_rate": 8.28444075116281e-05,
      "loss": 11.8407,
      "step": 12997
    },
    {
      "epoch": 2.2728743169398906,
      "grad_norm": 0.6751082221381328,
      "learning_rate": 8.283045674114053e-05,
      "loss": 11.7586,
      "step": 12998
    },
    {
      "epoch": 2.273049180327869,
      "grad_norm": 0.7253306871551908,
      "learning_rate": 8.281650631495501e-05,
      "loss": 11.5779,
      "step": 12999
    },
    {
      "epoch": 2.273224043715847,
      "grad_norm": 0.7052780497963762,
      "learning_rate": 8.28025562333513e-05,
      "loss": 11.9193,
      "step": 13000
    },
    {
      "epoch": 2.273398907103825,
      "grad_norm": 0.6653272932213351,
      "learning_rate": 8.278860649660915e-05,
      "loss": 11.591,
      "step": 13001
    },
    {
      "epoch": 2.273573770491803,
      "grad_norm": 0.7303292986074567,
      "learning_rate": 8.277465710500824e-05,
      "loss": 12.0686,
      "step": 13002
    },
    {
      "epoch": 2.2737486338797814,
      "grad_norm": 0.6558247117277085,
      "learning_rate": 8.276070805882841e-05,
      "loss": 12.0025,
      "step": 13003
    },
    {
      "epoch": 2.2739234972677593,
      "grad_norm": 0.7110765834177646,
      "learning_rate": 8.274675935834928e-05,
      "loss": 11.9927,
      "step": 13004
    },
    {
      "epoch": 2.2740983606557377,
      "grad_norm": 0.6601434378108402,
      "learning_rate": 8.273281100385059e-05,
      "loss": 11.921,
      "step": 13005
    },
    {
      "epoch": 2.274273224043716,
      "grad_norm": 0.6912427237826997,
      "learning_rate": 8.271886299561205e-05,
      "loss": 11.9685,
      "step": 13006
    },
    {
      "epoch": 2.274448087431694,
      "grad_norm": 0.7068981900325255,
      "learning_rate": 8.270491533391341e-05,
      "loss": 11.8023,
      "step": 13007
    },
    {
      "epoch": 2.2746229508196723,
      "grad_norm": 0.7097118413194486,
      "learning_rate": 8.26909680190343e-05,
      "loss": 11.9979,
      "step": 13008
    },
    {
      "epoch": 2.27479781420765,
      "grad_norm": 0.6925025353520856,
      "learning_rate": 8.267702105125442e-05,
      "loss": 12.1262,
      "step": 13009
    },
    {
      "epoch": 2.2749726775956285,
      "grad_norm": 0.7247642894048953,
      "learning_rate": 8.266307443085348e-05,
      "loss": 11.9513,
      "step": 13010
    },
    {
      "epoch": 2.2751475409836064,
      "grad_norm": 0.6647808611722829,
      "learning_rate": 8.264912815811106e-05,
      "loss": 12.09,
      "step": 13011
    },
    {
      "epoch": 2.2753224043715847,
      "grad_norm": 0.7672030967404709,
      "learning_rate": 8.263518223330697e-05,
      "loss": 12.022,
      "step": 13012
    },
    {
      "epoch": 2.2754972677595626,
      "grad_norm": 0.6654877598202885,
      "learning_rate": 8.262123665672077e-05,
      "loss": 11.6104,
      "step": 13013
    },
    {
      "epoch": 2.275672131147541,
      "grad_norm": 0.6884271852472474,
      "learning_rate": 8.260729142863213e-05,
      "loss": 11.7975,
      "step": 13014
    },
    {
      "epoch": 2.2758469945355193,
      "grad_norm": 0.6958271155144067,
      "learning_rate": 8.259334654932072e-05,
      "loss": 12.0838,
      "step": 13015
    },
    {
      "epoch": 2.276021857923497,
      "grad_norm": 0.6166206631889788,
      "learning_rate": 8.257940201906612e-05,
      "loss": 12.0175,
      "step": 13016
    },
    {
      "epoch": 2.2761967213114755,
      "grad_norm": 0.6265056698289418,
      "learning_rate": 8.256545783814802e-05,
      "loss": 11.6938,
      "step": 13017
    },
    {
      "epoch": 2.2763715846994534,
      "grad_norm": 0.6540813463328086,
      "learning_rate": 8.255151400684601e-05,
      "loss": 11.678,
      "step": 13018
    },
    {
      "epoch": 2.276546448087432,
      "grad_norm": 0.6537303119919176,
      "learning_rate": 8.253757052543976e-05,
      "loss": 11.8905,
      "step": 13019
    },
    {
      "epoch": 2.2767213114754097,
      "grad_norm": 0.7010836844992498,
      "learning_rate": 8.252362739420879e-05,
      "loss": 11.6573,
      "step": 13020
    },
    {
      "epoch": 2.276896174863388,
      "grad_norm": 0.7474279369544043,
      "learning_rate": 8.250968461343275e-05,
      "loss": 11.7056,
      "step": 13021
    },
    {
      "epoch": 2.277071038251366,
      "grad_norm": 0.7148515667813555,
      "learning_rate": 8.249574218339124e-05,
      "loss": 11.8377,
      "step": 13022
    },
    {
      "epoch": 2.2772459016393443,
      "grad_norm": 0.6779585855300139,
      "learning_rate": 8.248180010436386e-05,
      "loss": 11.8249,
      "step": 13023
    },
    {
      "epoch": 2.2774207650273226,
      "grad_norm": 0.7221742901808165,
      "learning_rate": 8.246785837663019e-05,
      "loss": 11.8133,
      "step": 13024
    },
    {
      "epoch": 2.2775956284153005,
      "grad_norm": 0.6664291426238845,
      "learning_rate": 8.245391700046976e-05,
      "loss": 12.0801,
      "step": 13025
    },
    {
      "epoch": 2.277770491803279,
      "grad_norm": 0.6810340498698865,
      "learning_rate": 8.243997597616217e-05,
      "loss": 12.1123,
      "step": 13026
    },
    {
      "epoch": 2.2779453551912567,
      "grad_norm": 0.6911625490203545,
      "learning_rate": 8.242603530398699e-05,
      "loss": 11.7916,
      "step": 13027
    },
    {
      "epoch": 2.278120218579235,
      "grad_norm": 0.6516663154079411,
      "learning_rate": 8.241209498422377e-05,
      "loss": 12.1194,
      "step": 13028
    },
    {
      "epoch": 2.278295081967213,
      "grad_norm": 0.6236690606352389,
      "learning_rate": 8.239815501715202e-05,
      "loss": 11.9237,
      "step": 13029
    },
    {
      "epoch": 2.2784699453551913,
      "grad_norm": 0.6555499510563895,
      "learning_rate": 8.23842154030513e-05,
      "loss": 12.1444,
      "step": 13030
    },
    {
      "epoch": 2.278644808743169,
      "grad_norm": 0.7088914169464523,
      "learning_rate": 8.237027614220117e-05,
      "loss": 11.8181,
      "step": 13031
    },
    {
      "epoch": 2.2788196721311476,
      "grad_norm": 0.7456679414929575,
      "learning_rate": 8.235633723488114e-05,
      "loss": 12.143,
      "step": 13032
    },
    {
      "epoch": 2.278994535519126,
      "grad_norm": 0.6844866896699553,
      "learning_rate": 8.234239868137073e-05,
      "loss": 12.0336,
      "step": 13033
    },
    {
      "epoch": 2.279169398907104,
      "grad_norm": 0.6957653495018511,
      "learning_rate": 8.232846048194943e-05,
      "loss": 11.7944,
      "step": 13034
    },
    {
      "epoch": 2.279344262295082,
      "grad_norm": 0.810468149921482,
      "learning_rate": 8.231452263689674e-05,
      "loss": 11.791,
      "step": 13035
    },
    {
      "epoch": 2.27951912568306,
      "grad_norm": 0.7592643833358784,
      "learning_rate": 8.230058514649218e-05,
      "loss": 11.4564,
      "step": 13036
    },
    {
      "epoch": 2.2796939890710384,
      "grad_norm": 0.6862703096420605,
      "learning_rate": 8.228664801101527e-05,
      "loss": 11.9151,
      "step": 13037
    },
    {
      "epoch": 2.2798688524590163,
      "grad_norm": 0.7059446053550582,
      "learning_rate": 8.227271123074541e-05,
      "loss": 12.111,
      "step": 13038
    },
    {
      "epoch": 2.2800437158469946,
      "grad_norm": 0.7144211954266122,
      "learning_rate": 8.225877480596213e-05,
      "loss": 11.9792,
      "step": 13039
    },
    {
      "epoch": 2.2802185792349725,
      "grad_norm": 0.6896999011931502,
      "learning_rate": 8.224483873694487e-05,
      "loss": 12.0378,
      "step": 13040
    },
    {
      "epoch": 2.280393442622951,
      "grad_norm": 0.8181900193568756,
      "learning_rate": 8.223090302397313e-05,
      "loss": 12.0021,
      "step": 13041
    },
    {
      "epoch": 2.280568306010929,
      "grad_norm": 0.8028558324866231,
      "learning_rate": 8.221696766732635e-05,
      "loss": 11.9104,
      "step": 13042
    },
    {
      "epoch": 2.280743169398907,
      "grad_norm": 0.6967391895015209,
      "learning_rate": 8.220303266728397e-05,
      "loss": 11.8475,
      "step": 13043
    },
    {
      "epoch": 2.2809180327868854,
      "grad_norm": 0.7862512099518484,
      "learning_rate": 8.218909802412542e-05,
      "loss": 11.9846,
      "step": 13044
    },
    {
      "epoch": 2.2810928961748633,
      "grad_norm": 0.7186543978312744,
      "learning_rate": 8.217516373813013e-05,
      "loss": 11.8377,
      "step": 13045
    },
    {
      "epoch": 2.2812677595628417,
      "grad_norm": 0.7500064924789117,
      "learning_rate": 8.216122980957757e-05,
      "loss": 11.8517,
      "step": 13046
    },
    {
      "epoch": 2.2814426229508196,
      "grad_norm": 0.6673870382492265,
      "learning_rate": 8.21472962387471e-05,
      "loss": 11.9256,
      "step": 13047
    },
    {
      "epoch": 2.281617486338798,
      "grad_norm": 0.7154490819662463,
      "learning_rate": 8.213336302591816e-05,
      "loss": 11.8184,
      "step": 13048
    },
    {
      "epoch": 2.281792349726776,
      "grad_norm": 0.6283663368459675,
      "learning_rate": 8.211943017137015e-05,
      "loss": 11.7275,
      "step": 13049
    },
    {
      "epoch": 2.281967213114754,
      "grad_norm": 0.6952401585591893,
      "learning_rate": 8.210549767538243e-05,
      "loss": 11.6716,
      "step": 13050
    },
    {
      "epoch": 2.2821420765027325,
      "grad_norm": 0.6598050931952265,
      "learning_rate": 8.209156553823449e-05,
      "loss": 12.0101,
      "step": 13051
    },
    {
      "epoch": 2.2823169398907104,
      "grad_norm": 0.720574922870259,
      "learning_rate": 8.20776337602056e-05,
      "loss": 12.0509,
      "step": 13052
    },
    {
      "epoch": 2.2824918032786883,
      "grad_norm": 0.6588701387455679,
      "learning_rate": 8.20637023415752e-05,
      "loss": 12.0061,
      "step": 13053
    },
    {
      "epoch": 2.2826666666666666,
      "grad_norm": 0.686066961135884,
      "learning_rate": 8.204977128262261e-05,
      "loss": 11.6975,
      "step": 13054
    },
    {
      "epoch": 2.282841530054645,
      "grad_norm": 0.6237576129284377,
      "learning_rate": 8.203584058362728e-05,
      "loss": 12.0561,
      "step": 13055
    },
    {
      "epoch": 2.283016393442623,
      "grad_norm": 0.675732497854846,
      "learning_rate": 8.202191024486846e-05,
      "loss": 11.7517,
      "step": 13056
    },
    {
      "epoch": 2.283191256830601,
      "grad_norm": 0.6143123121044896,
      "learning_rate": 8.200798026662555e-05,
      "loss": 11.7317,
      "step": 13057
    },
    {
      "epoch": 2.283366120218579,
      "grad_norm": 0.6689523385122425,
      "learning_rate": 8.19940506491779e-05,
      "loss": 12.1142,
      "step": 13058
    },
    {
      "epoch": 2.2835409836065574,
      "grad_norm": 0.6460938186780545,
      "learning_rate": 8.198012139280476e-05,
      "loss": 11.7464,
      "step": 13059
    },
    {
      "epoch": 2.2837158469945353,
      "grad_norm": 0.6597613372375538,
      "learning_rate": 8.196619249778557e-05,
      "loss": 11.711,
      "step": 13060
    },
    {
      "epoch": 2.2838907103825137,
      "grad_norm": 0.6660338468642958,
      "learning_rate": 8.195226396439958e-05,
      "loss": 11.7387,
      "step": 13061
    },
    {
      "epoch": 2.2840655737704916,
      "grad_norm": 0.6163517143630429,
      "learning_rate": 8.19383357929261e-05,
      "loss": 11.9779,
      "step": 13062
    },
    {
      "epoch": 2.28424043715847,
      "grad_norm": 0.6610439745012691,
      "learning_rate": 8.192440798364446e-05,
      "loss": 12.0178,
      "step": 13063
    },
    {
      "epoch": 2.2844153005464483,
      "grad_norm": 0.6748729041224211,
      "learning_rate": 8.191048053683392e-05,
      "loss": 12.2343,
      "step": 13064
    },
    {
      "epoch": 2.284590163934426,
      "grad_norm": 0.6158165966915802,
      "learning_rate": 8.18965534527738e-05,
      "loss": 11.6816,
      "step": 13065
    },
    {
      "epoch": 2.2847650273224045,
      "grad_norm": 0.6447439835452448,
      "learning_rate": 8.188262673174337e-05,
      "loss": 12.0033,
      "step": 13066
    },
    {
      "epoch": 2.2849398907103824,
      "grad_norm": 0.657309192545808,
      "learning_rate": 8.18687003740219e-05,
      "loss": 11.7437,
      "step": 13067
    },
    {
      "epoch": 2.2851147540983607,
      "grad_norm": 0.7021559596411484,
      "learning_rate": 8.185477437988866e-05,
      "loss": 11.6709,
      "step": 13068
    },
    {
      "epoch": 2.2852896174863386,
      "grad_norm": 0.6811405522732257,
      "learning_rate": 8.184084874962287e-05,
      "loss": 11.7555,
      "step": 13069
    },
    {
      "epoch": 2.285464480874317,
      "grad_norm": 0.6474952651625312,
      "learning_rate": 8.182692348350385e-05,
      "loss": 11.9021,
      "step": 13070
    },
    {
      "epoch": 2.285639344262295,
      "grad_norm": 0.7078844470026522,
      "learning_rate": 8.181299858181082e-05,
      "loss": 11.8207,
      "step": 13071
    },
    {
      "epoch": 2.285814207650273,
      "grad_norm": 0.7023195492209694,
      "learning_rate": 8.179907404482301e-05,
      "loss": 12.0873,
      "step": 13072
    },
    {
      "epoch": 2.2859890710382516,
      "grad_norm": 0.6781676878419208,
      "learning_rate": 8.178514987281964e-05,
      "loss": 12.0776,
      "step": 13073
    },
    {
      "epoch": 2.2861639344262294,
      "grad_norm": 0.6679464414554497,
      "learning_rate": 8.177122606607993e-05,
      "loss": 12.0647,
      "step": 13074
    },
    {
      "epoch": 2.286338797814208,
      "grad_norm": 0.6526667432475157,
      "learning_rate": 8.175730262488313e-05,
      "loss": 11.8384,
      "step": 13075
    },
    {
      "epoch": 2.2865136612021857,
      "grad_norm": 0.659621968461623,
      "learning_rate": 8.174337954950842e-05,
      "loss": 12.226,
      "step": 13076
    },
    {
      "epoch": 2.286688524590164,
      "grad_norm": 0.6268393891800702,
      "learning_rate": 8.172945684023501e-05,
      "loss": 11.9481,
      "step": 13077
    },
    {
      "epoch": 2.286863387978142,
      "grad_norm": 0.6114764665346569,
      "learning_rate": 8.171553449734206e-05,
      "loss": 11.8564,
      "step": 13078
    },
    {
      "epoch": 2.2870382513661203,
      "grad_norm": 0.6910537915275983,
      "learning_rate": 8.17016125211088e-05,
      "loss": 11.7736,
      "step": 13079
    },
    {
      "epoch": 2.287213114754098,
      "grad_norm": 0.6057224824365852,
      "learning_rate": 8.168769091181438e-05,
      "loss": 11.68,
      "step": 13080
    },
    {
      "epoch": 2.2873879781420765,
      "grad_norm": 0.6457542751762063,
      "learning_rate": 8.167376966973803e-05,
      "loss": 11.9511,
      "step": 13081
    },
    {
      "epoch": 2.287562841530055,
      "grad_norm": 0.6554936418460034,
      "learning_rate": 8.165984879515883e-05,
      "loss": 11.9535,
      "step": 13082
    },
    {
      "epoch": 2.2877377049180327,
      "grad_norm": 0.6847639792230087,
      "learning_rate": 8.164592828835596e-05,
      "loss": 11.9205,
      "step": 13083
    },
    {
      "epoch": 2.287912568306011,
      "grad_norm": 0.6554441820156989,
      "learning_rate": 8.163200814960861e-05,
      "loss": 11.8689,
      "step": 13084
    },
    {
      "epoch": 2.288087431693989,
      "grad_norm": 0.6265038045134865,
      "learning_rate": 8.16180883791959e-05,
      "loss": 11.9214,
      "step": 13085
    },
    {
      "epoch": 2.2882622950819673,
      "grad_norm": 0.6952174929331777,
      "learning_rate": 8.160416897739693e-05,
      "loss": 11.9776,
      "step": 13086
    },
    {
      "epoch": 2.288437158469945,
      "grad_norm": 0.6337273790030225,
      "learning_rate": 8.159024994449087e-05,
      "loss": 11.7477,
      "step": 13087
    },
    {
      "epoch": 2.2886120218579236,
      "grad_norm": 0.6932979708233182,
      "learning_rate": 8.157633128075681e-05,
      "loss": 11.7349,
      "step": 13088
    },
    {
      "epoch": 2.2887868852459015,
      "grad_norm": 0.6688601953090342,
      "learning_rate": 8.156241298647388e-05,
      "loss": 12.0346,
      "step": 13089
    },
    {
      "epoch": 2.28896174863388,
      "grad_norm": 0.6659653116160306,
      "learning_rate": 8.154849506192121e-05,
      "loss": 11.8712,
      "step": 13090
    },
    {
      "epoch": 2.289136612021858,
      "grad_norm": 0.6574479562189974,
      "learning_rate": 8.153457750737785e-05,
      "loss": 11.9907,
      "step": 13091
    },
    {
      "epoch": 2.289311475409836,
      "grad_norm": 0.7078024286003712,
      "learning_rate": 8.152066032312293e-05,
      "loss": 11.917,
      "step": 13092
    },
    {
      "epoch": 2.2894863387978144,
      "grad_norm": 0.6437791667753319,
      "learning_rate": 8.150674350943542e-05,
      "loss": 11.9195,
      "step": 13093
    },
    {
      "epoch": 2.2896612021857923,
      "grad_norm": 0.7879742247275623,
      "learning_rate": 8.149282706659458e-05,
      "loss": 11.6232,
      "step": 13094
    },
    {
      "epoch": 2.2898360655737706,
      "grad_norm": 0.7072843243795858,
      "learning_rate": 8.147891099487934e-05,
      "loss": 11.8385,
      "step": 13095
    },
    {
      "epoch": 2.2900109289617485,
      "grad_norm": 0.6558647370739134,
      "learning_rate": 8.146499529456881e-05,
      "loss": 12.008,
      "step": 13096
    },
    {
      "epoch": 2.290185792349727,
      "grad_norm": 0.7800088061413957,
      "learning_rate": 8.145107996594206e-05,
      "loss": 11.7308,
      "step": 13097
    },
    {
      "epoch": 2.2903606557377048,
      "grad_norm": 0.6263491961721765,
      "learning_rate": 8.143716500927804e-05,
      "loss": 11.8161,
      "step": 13098
    },
    {
      "epoch": 2.290535519125683,
      "grad_norm": 0.666848717207824,
      "learning_rate": 8.142325042485592e-05,
      "loss": 11.9323,
      "step": 13099
    },
    {
      "epoch": 2.2907103825136614,
      "grad_norm": 0.7410255053937399,
      "learning_rate": 8.140933621295465e-05,
      "loss": 12.0118,
      "step": 13100
    },
    {
      "epoch": 2.2908852459016393,
      "grad_norm": 0.6754733360378078,
      "learning_rate": 8.139542237385328e-05,
      "loss": 12.1759,
      "step": 13101
    },
    {
      "epoch": 2.2910601092896177,
      "grad_norm": 0.751529400437084,
      "learning_rate": 8.138150890783077e-05,
      "loss": 12.0847,
      "step": 13102
    },
    {
      "epoch": 2.2912349726775956,
      "grad_norm": 0.6503015919737921,
      "learning_rate": 8.136759581516622e-05,
      "loss": 11.9561,
      "step": 13103
    },
    {
      "epoch": 2.291409836065574,
      "grad_norm": 0.7346316873902545,
      "learning_rate": 8.135368309613856e-05,
      "loss": 12.1402,
      "step": 13104
    },
    {
      "epoch": 2.291584699453552,
      "grad_norm": 0.6765060305232347,
      "learning_rate": 8.133977075102682e-05,
      "loss": 12.1206,
      "step": 13105
    },
    {
      "epoch": 2.29175956284153,
      "grad_norm": 0.6496001477501664,
      "learning_rate": 8.132585878010998e-05,
      "loss": 11.7502,
      "step": 13106
    },
    {
      "epoch": 2.291934426229508,
      "grad_norm": 0.6679114141412373,
      "learning_rate": 8.131194718366696e-05,
      "loss": 11.9531,
      "step": 13107
    },
    {
      "epoch": 2.2921092896174864,
      "grad_norm": 0.7183459389662088,
      "learning_rate": 8.129803596197682e-05,
      "loss": 11.9744,
      "step": 13108
    },
    {
      "epoch": 2.2922841530054643,
      "grad_norm": 0.592305613884343,
      "learning_rate": 8.128412511531849e-05,
      "loss": 11.7669,
      "step": 13109
    },
    {
      "epoch": 2.2924590163934426,
      "grad_norm": 0.7466599987757979,
      "learning_rate": 8.127021464397089e-05,
      "loss": 11.9367,
      "step": 13110
    },
    {
      "epoch": 2.2926338797814205,
      "grad_norm": 0.5897240693491508,
      "learning_rate": 8.125630454821302e-05,
      "loss": 11.8129,
      "step": 13111
    },
    {
      "epoch": 2.292808743169399,
      "grad_norm": 0.6392162345617656,
      "learning_rate": 8.124239482832377e-05,
      "loss": 11.9531,
      "step": 13112
    },
    {
      "epoch": 2.292983606557377,
      "grad_norm": 0.6792538323155598,
      "learning_rate": 8.12284854845821e-05,
      "loss": 11.6895,
      "step": 13113
    },
    {
      "epoch": 2.293158469945355,
      "grad_norm": 0.7337012432627734,
      "learning_rate": 8.121457651726694e-05,
      "loss": 11.8122,
      "step": 13114
    },
    {
      "epoch": 2.2933333333333334,
      "grad_norm": 0.7863073575893215,
      "learning_rate": 8.120066792665722e-05,
      "loss": 11.8954,
      "step": 13115
    },
    {
      "epoch": 2.2935081967213113,
      "grad_norm": 0.6711677945380525,
      "learning_rate": 8.11867597130318e-05,
      "loss": 11.9433,
      "step": 13116
    },
    {
      "epoch": 2.2936830601092897,
      "grad_norm": 0.76501649492228,
      "learning_rate": 8.11728518766696e-05,
      "loss": 11.8683,
      "step": 13117
    },
    {
      "epoch": 2.2938579234972676,
      "grad_norm": 0.6594577368731428,
      "learning_rate": 8.115894441784955e-05,
      "loss": 11.8853,
      "step": 13118
    },
    {
      "epoch": 2.294032786885246,
      "grad_norm": 0.7183373560027461,
      "learning_rate": 8.11450373368505e-05,
      "loss": 11.5019,
      "step": 13119
    },
    {
      "epoch": 2.294207650273224,
      "grad_norm": 0.653355119877172,
      "learning_rate": 8.113113063395138e-05,
      "loss": 11.8117,
      "step": 13120
    },
    {
      "epoch": 2.294382513661202,
      "grad_norm": 0.6891508845562238,
      "learning_rate": 8.1117224309431e-05,
      "loss": 11.7413,
      "step": 13121
    },
    {
      "epoch": 2.2945573770491805,
      "grad_norm": 0.6958508172008818,
      "learning_rate": 8.110331836356824e-05,
      "loss": 12.0731,
      "step": 13122
    },
    {
      "epoch": 2.2947322404371584,
      "grad_norm": 0.6290295901530938,
      "learning_rate": 8.108941279664197e-05,
      "loss": 11.9737,
      "step": 13123
    },
    {
      "epoch": 2.2949071038251367,
      "grad_norm": 0.674068210678229,
      "learning_rate": 8.107550760893105e-05,
      "loss": 12.1156,
      "step": 13124
    },
    {
      "epoch": 2.2950819672131146,
      "grad_norm": 0.7066000850050405,
      "learning_rate": 8.106160280071431e-05,
      "loss": 11.996,
      "step": 13125
    },
    {
      "epoch": 2.295256830601093,
      "grad_norm": 0.6123083271658178,
      "learning_rate": 8.104769837227055e-05,
      "loss": 12.0764,
      "step": 13126
    },
    {
      "epoch": 2.295431693989071,
      "grad_norm": 0.6790616821482848,
      "learning_rate": 8.103379432387865e-05,
      "loss": 11.7417,
      "step": 13127
    },
    {
      "epoch": 2.295606557377049,
      "grad_norm": 0.6327600806124447,
      "learning_rate": 8.101989065581743e-05,
      "loss": 11.8297,
      "step": 13128
    },
    {
      "epoch": 2.295781420765027,
      "grad_norm": 0.7041362790272575,
      "learning_rate": 8.100598736836567e-05,
      "loss": 11.9968,
      "step": 13129
    },
    {
      "epoch": 2.2959562841530055,
      "grad_norm": 0.6701788965673539,
      "learning_rate": 8.099208446180218e-05,
      "loss": 11.9683,
      "step": 13130
    },
    {
      "epoch": 2.296131147540984,
      "grad_norm": 0.6250014588249267,
      "learning_rate": 8.097818193640572e-05,
      "loss": 12.0605,
      "step": 13131
    },
    {
      "epoch": 2.2963060109289617,
      "grad_norm": 0.6427818938750739,
      "learning_rate": 8.096427979245516e-05,
      "loss": 11.8062,
      "step": 13132
    },
    {
      "epoch": 2.29648087431694,
      "grad_norm": 0.6841482658504967,
      "learning_rate": 8.095037803022923e-05,
      "loss": 12.1049,
      "step": 13133
    },
    {
      "epoch": 2.296655737704918,
      "grad_norm": 0.7548909055167318,
      "learning_rate": 8.093647665000672e-05,
      "loss": 11.7156,
      "step": 13134
    },
    {
      "epoch": 2.2968306010928963,
      "grad_norm": 0.6982581154490217,
      "learning_rate": 8.092257565206637e-05,
      "loss": 11.9465,
      "step": 13135
    },
    {
      "epoch": 2.297005464480874,
      "grad_norm": 0.7455141975930661,
      "learning_rate": 8.090867503668693e-05,
      "loss": 12.0925,
      "step": 13136
    },
    {
      "epoch": 2.2971803278688525,
      "grad_norm": 0.6638640896346025,
      "learning_rate": 8.08947748041472e-05,
      "loss": 12.001,
      "step": 13137
    },
    {
      "epoch": 2.2973551912568304,
      "grad_norm": 0.6777820960845106,
      "learning_rate": 8.088087495472591e-05,
      "loss": 12.0358,
      "step": 13138
    },
    {
      "epoch": 2.2975300546448088,
      "grad_norm": 0.7979142380596218,
      "learning_rate": 8.086697548870177e-05,
      "loss": 11.89,
      "step": 13139
    },
    {
      "epoch": 2.297704918032787,
      "grad_norm": 0.6396646641340197,
      "learning_rate": 8.085307640635353e-05,
      "loss": 11.9312,
      "step": 13140
    },
    {
      "epoch": 2.297879781420765,
      "grad_norm": 0.8318677027765252,
      "learning_rate": 8.083917770795983e-05,
      "loss": 11.9278,
      "step": 13141
    },
    {
      "epoch": 2.2980546448087433,
      "grad_norm": 0.7368885805410063,
      "learning_rate": 8.082527939379953e-05,
      "loss": 12.0656,
      "step": 13142
    },
    {
      "epoch": 2.2982295081967212,
      "grad_norm": 0.7306653087182313,
      "learning_rate": 8.081138146415121e-05,
      "loss": 11.6443,
      "step": 13143
    },
    {
      "epoch": 2.2984043715846996,
      "grad_norm": 0.8256887540056881,
      "learning_rate": 8.079748391929363e-05,
      "loss": 11.9033,
      "step": 13144
    },
    {
      "epoch": 2.2985792349726775,
      "grad_norm": 0.7136632480324506,
      "learning_rate": 8.078358675950546e-05,
      "loss": 11.9811,
      "step": 13145
    },
    {
      "epoch": 2.298754098360656,
      "grad_norm": 0.7931617102621847,
      "learning_rate": 8.076968998506532e-05,
      "loss": 11.9647,
      "step": 13146
    },
    {
      "epoch": 2.2989289617486337,
      "grad_norm": 0.6894111368029461,
      "learning_rate": 8.075579359625201e-05,
      "loss": 11.9321,
      "step": 13147
    },
    {
      "epoch": 2.299103825136612,
      "grad_norm": 0.6344156062840488,
      "learning_rate": 8.07418975933441e-05,
      "loss": 11.9348,
      "step": 13148
    },
    {
      "epoch": 2.2992786885245904,
      "grad_norm": 0.7883379382906456,
      "learning_rate": 8.072800197662029e-05,
      "loss": 11.8341,
      "step": 13149
    },
    {
      "epoch": 2.2994535519125683,
      "grad_norm": 0.7459828801813101,
      "learning_rate": 8.071410674635915e-05,
      "loss": 11.8421,
      "step": 13150
    },
    {
      "epoch": 2.2996284153005466,
      "grad_norm": 0.6663048976617394,
      "learning_rate": 8.070021190283946e-05,
      "loss": 11.8364,
      "step": 13151
    },
    {
      "epoch": 2.2998032786885245,
      "grad_norm": 0.6565067010883532,
      "learning_rate": 8.068631744633976e-05,
      "loss": 11.9167,
      "step": 13152
    },
    {
      "epoch": 2.299978142076503,
      "grad_norm": 0.7029912726236989,
      "learning_rate": 8.067242337713868e-05,
      "loss": 11.7709,
      "step": 13153
    },
    {
      "epoch": 2.3001530054644808,
      "grad_norm": 0.6835295633400156,
      "learning_rate": 8.065852969551488e-05,
      "loss": 11.9778,
      "step": 13154
    },
    {
      "epoch": 2.300327868852459,
      "grad_norm": 0.7313392731099722,
      "learning_rate": 8.064463640174689e-05,
      "loss": 11.8546,
      "step": 13155
    },
    {
      "epoch": 2.300502732240437,
      "grad_norm": 0.6924133269147585,
      "learning_rate": 8.063074349611342e-05,
      "loss": 11.9149,
      "step": 13156
    },
    {
      "epoch": 2.3006775956284153,
      "grad_norm": 0.6956622934441405,
      "learning_rate": 8.0616850978893e-05,
      "loss": 11.7074,
      "step": 13157
    },
    {
      "epoch": 2.3008524590163937,
      "grad_norm": 0.6871096879486137,
      "learning_rate": 8.060295885036426e-05,
      "loss": 11.8281,
      "step": 13158
    },
    {
      "epoch": 2.3010273224043716,
      "grad_norm": 0.6528567659326167,
      "learning_rate": 8.058906711080573e-05,
      "loss": 11.9899,
      "step": 13159
    },
    {
      "epoch": 2.30120218579235,
      "grad_norm": 0.7692363590242486,
      "learning_rate": 8.057517576049597e-05,
      "loss": 11.9146,
      "step": 13160
    },
    {
      "epoch": 2.301377049180328,
      "grad_norm": 0.7159460664130245,
      "learning_rate": 8.056128479971361e-05,
      "loss": 11.3769,
      "step": 13161
    },
    {
      "epoch": 2.301551912568306,
      "grad_norm": 0.6351376538278731,
      "learning_rate": 8.054739422873717e-05,
      "loss": 11.965,
      "step": 13162
    },
    {
      "epoch": 2.301726775956284,
      "grad_norm": 0.7176767866228039,
      "learning_rate": 8.053350404784522e-05,
      "loss": 12.0013,
      "step": 13163
    },
    {
      "epoch": 2.3019016393442624,
      "grad_norm": 0.6824017223619661,
      "learning_rate": 8.051961425731627e-05,
      "loss": 12.076,
      "step": 13164
    },
    {
      "epoch": 2.3020765027322403,
      "grad_norm": 0.6116846480861344,
      "learning_rate": 8.050572485742884e-05,
      "loss": 12.1195,
      "step": 13165
    },
    {
      "epoch": 2.3022513661202186,
      "grad_norm": 0.6694176679250091,
      "learning_rate": 8.04918358484615e-05,
      "loss": 11.4633,
      "step": 13166
    },
    {
      "epoch": 2.3024262295081965,
      "grad_norm": 0.6848170879849,
      "learning_rate": 8.047794723069279e-05,
      "loss": 11.8399,
      "step": 13167
    },
    {
      "epoch": 2.302601092896175,
      "grad_norm": 0.728534314392715,
      "learning_rate": 8.046405900440113e-05,
      "loss": 11.9115,
      "step": 13168
    },
    {
      "epoch": 2.3027759562841528,
      "grad_norm": 0.9443369422157962,
      "learning_rate": 8.04501711698651e-05,
      "loss": 12.1242,
      "step": 13169
    },
    {
      "epoch": 2.302950819672131,
      "grad_norm": 0.6549504053757996,
      "learning_rate": 8.043628372736311e-05,
      "loss": 11.9926,
      "step": 13170
    },
    {
      "epoch": 2.3031256830601095,
      "grad_norm": 0.6129857027778083,
      "learning_rate": 8.042239667717374e-05,
      "loss": 12.0068,
      "step": 13171
    },
    {
      "epoch": 2.3033005464480873,
      "grad_norm": 0.7114866672008693,
      "learning_rate": 8.040851001957546e-05,
      "loss": 11.7206,
      "step": 13172
    },
    {
      "epoch": 2.3034754098360657,
      "grad_norm": 0.6084415723600606,
      "learning_rate": 8.039462375484666e-05,
      "loss": 11.7357,
      "step": 13173
    },
    {
      "epoch": 2.3036502732240436,
      "grad_norm": 0.7547761848611327,
      "learning_rate": 8.038073788326585e-05,
      "loss": 11.9392,
      "step": 13174
    },
    {
      "epoch": 2.303825136612022,
      "grad_norm": 0.6181643738889431,
      "learning_rate": 8.03668524051115e-05,
      "loss": 11.7198,
      "step": 13175
    },
    {
      "epoch": 2.304,
      "grad_norm": 0.6368164156410054,
      "learning_rate": 8.035296732066204e-05,
      "loss": 11.9639,
      "step": 13176
    },
    {
      "epoch": 2.304174863387978,
      "grad_norm": 0.7363931881011749,
      "learning_rate": 8.033908263019593e-05,
      "loss": 11.5102,
      "step": 13177
    },
    {
      "epoch": 2.304349726775956,
      "grad_norm": 0.7221824446284113,
      "learning_rate": 8.032519833399155e-05,
      "loss": 11.7034,
      "step": 13178
    },
    {
      "epoch": 2.3045245901639344,
      "grad_norm": 0.6298567247768996,
      "learning_rate": 8.031131443232734e-05,
      "loss": 11.8709,
      "step": 13179
    },
    {
      "epoch": 2.3046994535519127,
      "grad_norm": 0.7421907205830315,
      "learning_rate": 8.029743092548175e-05,
      "loss": 11.7551,
      "step": 13180
    },
    {
      "epoch": 2.3048743169398906,
      "grad_norm": 0.6553991391824869,
      "learning_rate": 8.02835478137332e-05,
      "loss": 11.9684,
      "step": 13181
    },
    {
      "epoch": 2.305049180327869,
      "grad_norm": 0.6347643046524357,
      "learning_rate": 8.026966509736001e-05,
      "loss": 11.724,
      "step": 13182
    },
    {
      "epoch": 2.305224043715847,
      "grad_norm": 0.6245495823127722,
      "learning_rate": 8.025578277664062e-05,
      "loss": 11.7991,
      "step": 13183
    },
    {
      "epoch": 2.3053989071038252,
      "grad_norm": 0.6109482605035106,
      "learning_rate": 8.024190085185339e-05,
      "loss": 11.6653,
      "step": 13184
    },
    {
      "epoch": 2.305573770491803,
      "grad_norm": 0.7018270154124594,
      "learning_rate": 8.022801932327673e-05,
      "loss": 11.7807,
      "step": 13185
    },
    {
      "epoch": 2.3057486338797815,
      "grad_norm": 0.7216486335732646,
      "learning_rate": 8.021413819118903e-05,
      "loss": 11.8595,
      "step": 13186
    },
    {
      "epoch": 2.3059234972677594,
      "grad_norm": 0.6293324483241656,
      "learning_rate": 8.020025745586857e-05,
      "loss": 11.9653,
      "step": 13187
    },
    {
      "epoch": 2.3060983606557377,
      "grad_norm": 0.7124508121543499,
      "learning_rate": 8.018637711759377e-05,
      "loss": 11.5883,
      "step": 13188
    },
    {
      "epoch": 2.306273224043716,
      "grad_norm": 0.7079459035902758,
      "learning_rate": 8.017249717664288e-05,
      "loss": 11.8613,
      "step": 13189
    },
    {
      "epoch": 2.306448087431694,
      "grad_norm": 0.6578831593810941,
      "learning_rate": 8.015861763329435e-05,
      "loss": 11.8557,
      "step": 13190
    },
    {
      "epoch": 2.3066229508196723,
      "grad_norm": 0.7442756919537508,
      "learning_rate": 8.014473848782646e-05,
      "loss": 11.5944,
      "step": 13191
    },
    {
      "epoch": 2.30679781420765,
      "grad_norm": 0.719359658659029,
      "learning_rate": 8.01308597405175e-05,
      "loss": 11.8504,
      "step": 13192
    },
    {
      "epoch": 2.3069726775956285,
      "grad_norm": 0.6293327431894757,
      "learning_rate": 8.011698139164585e-05,
      "loss": 12.0323,
      "step": 13193
    },
    {
      "epoch": 2.3071475409836064,
      "grad_norm": 0.7204698820153836,
      "learning_rate": 8.01031034414897e-05,
      "loss": 11.76,
      "step": 13194
    },
    {
      "epoch": 2.3073224043715848,
      "grad_norm": 0.6402124530833211,
      "learning_rate": 8.008922589032749e-05,
      "loss": 11.8762,
      "step": 13195
    },
    {
      "epoch": 2.3074972677595627,
      "grad_norm": 0.6867609353640831,
      "learning_rate": 8.00753487384374e-05,
      "loss": 11.9333,
      "step": 13196
    },
    {
      "epoch": 2.307672131147541,
      "grad_norm": 0.7350632956734032,
      "learning_rate": 8.006147198609778e-05,
      "loss": 11.6812,
      "step": 13197
    },
    {
      "epoch": 2.3078469945355193,
      "grad_norm": 0.6538750638200322,
      "learning_rate": 8.004759563358679e-05,
      "loss": 11.8659,
      "step": 13198
    },
    {
      "epoch": 2.3080218579234972,
      "grad_norm": 0.6848175966183417,
      "learning_rate": 8.003371968118285e-05,
      "loss": 12.0192,
      "step": 13199
    },
    {
      "epoch": 2.3081967213114756,
      "grad_norm": 0.7065772105416315,
      "learning_rate": 8.00198441291641e-05,
      "loss": 11.8183,
      "step": 13200
    },
    {
      "epoch": 2.3083715846994535,
      "grad_norm": 0.7137715395761468,
      "learning_rate": 8.000596897780883e-05,
      "loss": 11.9502,
      "step": 13201
    },
    {
      "epoch": 2.308546448087432,
      "grad_norm": 0.6820395919804206,
      "learning_rate": 7.999209422739528e-05,
      "loss": 11.9178,
      "step": 13202
    },
    {
      "epoch": 2.3087213114754097,
      "grad_norm": 0.6930718974557025,
      "learning_rate": 7.997821987820162e-05,
      "loss": 11.7848,
      "step": 13203
    },
    {
      "epoch": 2.308896174863388,
      "grad_norm": 0.6448662628722425,
      "learning_rate": 7.996434593050619e-05,
      "loss": 11.7948,
      "step": 13204
    },
    {
      "epoch": 2.309071038251366,
      "grad_norm": 0.6680842278152448,
      "learning_rate": 7.995047238458711e-05,
      "loss": 12.0453,
      "step": 13205
    },
    {
      "epoch": 2.3092459016393443,
      "grad_norm": 0.6233627776859999,
      "learning_rate": 7.993659924072265e-05,
      "loss": 12.027,
      "step": 13206
    },
    {
      "epoch": 2.3094207650273226,
      "grad_norm": 0.6638443566940364,
      "learning_rate": 7.992272649919096e-05,
      "loss": 11.9702,
      "step": 13207
    },
    {
      "epoch": 2.3095956284153005,
      "grad_norm": 0.6219773688352827,
      "learning_rate": 7.990885416027023e-05,
      "loss": 12.0686,
      "step": 13208
    },
    {
      "epoch": 2.309770491803279,
      "grad_norm": 0.6737464627014674,
      "learning_rate": 7.989498222423866e-05,
      "loss": 11.5988,
      "step": 13209
    },
    {
      "epoch": 2.3099453551912568,
      "grad_norm": 0.6178069691184924,
      "learning_rate": 7.988111069137445e-05,
      "loss": 11.9396,
      "step": 13210
    },
    {
      "epoch": 2.310120218579235,
      "grad_norm": 0.6759151471163606,
      "learning_rate": 7.986723956195576e-05,
      "loss": 11.9715,
      "step": 13211
    },
    {
      "epoch": 2.310295081967213,
      "grad_norm": 0.644328504006939,
      "learning_rate": 7.985336883626071e-05,
      "loss": 12.0544,
      "step": 13212
    },
    {
      "epoch": 2.3104699453551913,
      "grad_norm": 0.6996305252865643,
      "learning_rate": 7.983949851456744e-05,
      "loss": 11.916,
      "step": 13213
    },
    {
      "epoch": 2.3106448087431692,
      "grad_norm": 0.71019232233566,
      "learning_rate": 7.982562859715416e-05,
      "loss": 11.9074,
      "step": 13214
    },
    {
      "epoch": 2.3108196721311476,
      "grad_norm": 0.6825032704751607,
      "learning_rate": 7.9811759084299e-05,
      "loss": 11.8221,
      "step": 13215
    },
    {
      "epoch": 2.310994535519126,
      "grad_norm": 0.6453786754696813,
      "learning_rate": 7.979788997628002e-05,
      "loss": 11.9586,
      "step": 13216
    },
    {
      "epoch": 2.311169398907104,
      "grad_norm": 0.6612337828962989,
      "learning_rate": 7.978402127337539e-05,
      "loss": 12.057,
      "step": 13217
    },
    {
      "epoch": 2.3113442622950817,
      "grad_norm": 0.7098008333138567,
      "learning_rate": 7.977015297586317e-05,
      "loss": 11.9576,
      "step": 13218
    },
    {
      "epoch": 2.31151912568306,
      "grad_norm": 0.6263183038918919,
      "learning_rate": 7.975628508402151e-05,
      "loss": 12.1251,
      "step": 13219
    },
    {
      "epoch": 2.3116939890710384,
      "grad_norm": 0.6464649559834587,
      "learning_rate": 7.974241759812853e-05,
      "loss": 11.9788,
      "step": 13220
    },
    {
      "epoch": 2.3118688524590163,
      "grad_norm": 0.6382056413713854,
      "learning_rate": 7.972855051846224e-05,
      "loss": 11.7677,
      "step": 13221
    },
    {
      "epoch": 2.3120437158469946,
      "grad_norm": 0.6611015199327676,
      "learning_rate": 7.971468384530074e-05,
      "loss": 11.9108,
      "step": 13222
    },
    {
      "epoch": 2.3122185792349725,
      "grad_norm": 0.6247632148576265,
      "learning_rate": 7.970081757892214e-05,
      "loss": 12.0492,
      "step": 13223
    },
    {
      "epoch": 2.312393442622951,
      "grad_norm": 0.683230396542041,
      "learning_rate": 7.968695171960449e-05,
      "loss": 11.8328,
      "step": 13224
    },
    {
      "epoch": 2.3125683060109288,
      "grad_norm": 0.6599010097361419,
      "learning_rate": 7.96730862676258e-05,
      "loss": 11.9723,
      "step": 13225
    },
    {
      "epoch": 2.312743169398907,
      "grad_norm": 0.6131493851787304,
      "learning_rate": 7.965922122326414e-05,
      "loss": 12.0152,
      "step": 13226
    },
    {
      "epoch": 2.312918032786885,
      "grad_norm": 0.6699198685774317,
      "learning_rate": 7.964535658679754e-05,
      "loss": 11.8171,
      "step": 13227
    },
    {
      "epoch": 2.3130928961748634,
      "grad_norm": 0.6963019309655617,
      "learning_rate": 7.963149235850405e-05,
      "loss": 12.0407,
      "step": 13228
    },
    {
      "epoch": 2.3132677595628417,
      "grad_norm": 0.631881715981184,
      "learning_rate": 7.96176285386617e-05,
      "loss": 12.0534,
      "step": 13229
    },
    {
      "epoch": 2.3134426229508196,
      "grad_norm": 0.6939055877813468,
      "learning_rate": 7.960376512754847e-05,
      "loss": 12.0546,
      "step": 13230
    },
    {
      "epoch": 2.313617486338798,
      "grad_norm": 0.6133920239460846,
      "learning_rate": 7.958990212544237e-05,
      "loss": 11.8646,
      "step": 13231
    },
    {
      "epoch": 2.313792349726776,
      "grad_norm": 0.6225658068747305,
      "learning_rate": 7.957603953262136e-05,
      "loss": 11.701,
      "step": 13232
    },
    {
      "epoch": 2.313967213114754,
      "grad_norm": 0.6745954299543792,
      "learning_rate": 7.956217734936353e-05,
      "loss": 12.0935,
      "step": 13233
    },
    {
      "epoch": 2.314142076502732,
      "grad_norm": 0.6506933899043783,
      "learning_rate": 7.954831557594677e-05,
      "loss": 11.6212,
      "step": 13234
    },
    {
      "epoch": 2.3143169398907104,
      "grad_norm": 0.7016022908948273,
      "learning_rate": 7.953445421264906e-05,
      "loss": 11.7206,
      "step": 13235
    },
    {
      "epoch": 2.3144918032786883,
      "grad_norm": 0.6529669560545698,
      "learning_rate": 7.952059325974843e-05,
      "loss": 11.8422,
      "step": 13236
    },
    {
      "epoch": 2.3146666666666667,
      "grad_norm": 0.6260268788745756,
      "learning_rate": 7.95067327175227e-05,
      "loss": 11.8309,
      "step": 13237
    },
    {
      "epoch": 2.314841530054645,
      "grad_norm": 0.6635737812467141,
      "learning_rate": 7.949287258624998e-05,
      "loss": 11.8455,
      "step": 13238
    },
    {
      "epoch": 2.315016393442623,
      "grad_norm": 0.6278041162159751,
      "learning_rate": 7.94790128662081e-05,
      "loss": 12.1305,
      "step": 13239
    },
    {
      "epoch": 2.3151912568306012,
      "grad_norm": 0.6818514956168562,
      "learning_rate": 7.946515355767504e-05,
      "loss": 11.9107,
      "step": 13240
    },
    {
      "epoch": 2.315366120218579,
      "grad_norm": 0.5924324409352622,
      "learning_rate": 7.945129466092868e-05,
      "loss": 11.8081,
      "step": 13241
    },
    {
      "epoch": 2.3155409836065575,
      "grad_norm": 0.6278573835996065,
      "learning_rate": 7.943743617624695e-05,
      "loss": 12.0309,
      "step": 13242
    },
    {
      "epoch": 2.3157158469945354,
      "grad_norm": 0.7592227142981144,
      "learning_rate": 7.942357810390776e-05,
      "loss": 11.8802,
      "step": 13243
    },
    {
      "epoch": 2.3158907103825137,
      "grad_norm": 0.6597480165237338,
      "learning_rate": 7.940972044418902e-05,
      "loss": 11.9703,
      "step": 13244
    },
    {
      "epoch": 2.3160655737704916,
      "grad_norm": 0.7216416347320764,
      "learning_rate": 7.939586319736863e-05,
      "loss": 11.9632,
      "step": 13245
    },
    {
      "epoch": 2.31624043715847,
      "grad_norm": 0.7831783991974518,
      "learning_rate": 7.938200636372437e-05,
      "loss": 11.8369,
      "step": 13246
    },
    {
      "epoch": 2.3164153005464483,
      "grad_norm": 0.6420726008870055,
      "learning_rate": 7.936814994353425e-05,
      "loss": 11.9315,
      "step": 13247
    },
    {
      "epoch": 2.316590163934426,
      "grad_norm": 0.7168567416924222,
      "learning_rate": 7.935429393707607e-05,
      "loss": 12.0293,
      "step": 13248
    },
    {
      "epoch": 2.3167650273224045,
      "grad_norm": 0.7499832344304079,
      "learning_rate": 7.934043834462767e-05,
      "loss": 11.8337,
      "step": 13249
    },
    {
      "epoch": 2.3169398907103824,
      "grad_norm": 0.6609970665759887,
      "learning_rate": 7.932658316646695e-05,
      "loss": 12.091,
      "step": 13250
    },
    {
      "epoch": 2.3171147540983608,
      "grad_norm": 0.699429773844677,
      "learning_rate": 7.931272840287165e-05,
      "loss": 12.0118,
      "step": 13251
    },
    {
      "epoch": 2.3172896174863387,
      "grad_norm": 0.6254204239751195,
      "learning_rate": 7.929887405411969e-05,
      "loss": 11.996,
      "step": 13252
    },
    {
      "epoch": 2.317464480874317,
      "grad_norm": 0.6945237640268098,
      "learning_rate": 7.928502012048888e-05,
      "loss": 11.7667,
      "step": 13253
    },
    {
      "epoch": 2.317639344262295,
      "grad_norm": 0.618536363239973,
      "learning_rate": 7.927116660225704e-05,
      "loss": 11.821,
      "step": 13254
    },
    {
      "epoch": 2.3178142076502732,
      "grad_norm": 0.6889993962060594,
      "learning_rate": 7.925731349970191e-05,
      "loss": 11.9296,
      "step": 13255
    },
    {
      "epoch": 2.3179890710382516,
      "grad_norm": 0.6218820691690733,
      "learning_rate": 7.924346081310133e-05,
      "loss": 11.8743,
      "step": 13256
    },
    {
      "epoch": 2.3181639344262295,
      "grad_norm": 0.7226367023644066,
      "learning_rate": 7.922960854273311e-05,
      "loss": 12.0045,
      "step": 13257
    },
    {
      "epoch": 2.318338797814208,
      "grad_norm": 0.6412411055086993,
      "learning_rate": 7.921575668887502e-05,
      "loss": 12.0442,
      "step": 13258
    },
    {
      "epoch": 2.3185136612021857,
      "grad_norm": 0.6438331844809936,
      "learning_rate": 7.920190525180485e-05,
      "loss": 11.7848,
      "step": 13259
    },
    {
      "epoch": 2.318688524590164,
      "grad_norm": 0.5937531435597369,
      "learning_rate": 7.918805423180029e-05,
      "loss": 11.8638,
      "step": 13260
    },
    {
      "epoch": 2.318863387978142,
      "grad_norm": 0.7956893696930956,
      "learning_rate": 7.917420362913914e-05,
      "loss": 12.1588,
      "step": 13261
    },
    {
      "epoch": 2.3190382513661203,
      "grad_norm": 0.7009783025634223,
      "learning_rate": 7.916035344409918e-05,
      "loss": 11.9261,
      "step": 13262
    },
    {
      "epoch": 2.319213114754098,
      "grad_norm": 0.653676071992657,
      "learning_rate": 7.914650367695812e-05,
      "loss": 12.0149,
      "step": 13263
    },
    {
      "epoch": 2.3193879781420765,
      "grad_norm": 0.6337351874676826,
      "learning_rate": 7.913265432799368e-05,
      "loss": 11.8228,
      "step": 13264
    },
    {
      "epoch": 2.319562841530055,
      "grad_norm": 0.7114500043572805,
      "learning_rate": 7.911880539748359e-05,
      "loss": 11.9237,
      "step": 13265
    },
    {
      "epoch": 2.3197377049180328,
      "grad_norm": 0.6447944294322739,
      "learning_rate": 7.910495688570555e-05,
      "loss": 11.7823,
      "step": 13266
    },
    {
      "epoch": 2.319912568306011,
      "grad_norm": 0.622079293036694,
      "learning_rate": 7.909110879293731e-05,
      "loss": 11.914,
      "step": 13267
    },
    {
      "epoch": 2.320087431693989,
      "grad_norm": 0.6852394879326806,
      "learning_rate": 7.907726111945653e-05,
      "loss": 11.7789,
      "step": 13268
    },
    {
      "epoch": 2.3202622950819674,
      "grad_norm": 0.6688294004986832,
      "learning_rate": 7.90634138655409e-05,
      "loss": 11.7669,
      "step": 13269
    },
    {
      "epoch": 2.3204371584699452,
      "grad_norm": 0.7060669413692974,
      "learning_rate": 7.904956703146809e-05,
      "loss": 12.1067,
      "step": 13270
    },
    {
      "epoch": 2.3206120218579236,
      "grad_norm": 0.6836797445452759,
      "learning_rate": 7.90357206175158e-05,
      "loss": 11.8288,
      "step": 13271
    },
    {
      "epoch": 2.3207868852459015,
      "grad_norm": 0.6505384482666782,
      "learning_rate": 7.902187462396172e-05,
      "loss": 11.8223,
      "step": 13272
    },
    {
      "epoch": 2.32096174863388,
      "grad_norm": 0.6403188188451946,
      "learning_rate": 7.900802905108342e-05,
      "loss": 11.9171,
      "step": 13273
    },
    {
      "epoch": 2.321136612021858,
      "grad_norm": 0.6473984562010399,
      "learning_rate": 7.89941838991586e-05,
      "loss": 12.0687,
      "step": 13274
    },
    {
      "epoch": 2.321311475409836,
      "grad_norm": 0.5905198577236415,
      "learning_rate": 7.898033916846487e-05,
      "loss": 12.0333,
      "step": 13275
    },
    {
      "epoch": 2.321486338797814,
      "grad_norm": 0.6870588340806388,
      "learning_rate": 7.89664948592799e-05,
      "loss": 11.7197,
      "step": 13276
    },
    {
      "epoch": 2.3216612021857923,
      "grad_norm": 0.6972819649350005,
      "learning_rate": 7.895265097188131e-05,
      "loss": 11.9099,
      "step": 13277
    },
    {
      "epoch": 2.3218360655737706,
      "grad_norm": 0.7063886898694851,
      "learning_rate": 7.893880750654668e-05,
      "loss": 11.9693,
      "step": 13278
    },
    {
      "epoch": 2.3220109289617485,
      "grad_norm": 0.7212984239734404,
      "learning_rate": 7.892496446355362e-05,
      "loss": 11.7425,
      "step": 13279
    },
    {
      "epoch": 2.322185792349727,
      "grad_norm": 0.6876797513391492,
      "learning_rate": 7.89111218431797e-05,
      "loss": 11.8885,
      "step": 13280
    },
    {
      "epoch": 2.322360655737705,
      "grad_norm": 0.6873213913175193,
      "learning_rate": 7.889727964570258e-05,
      "loss": 11.8904,
      "step": 13281
    },
    {
      "epoch": 2.322535519125683,
      "grad_norm": 0.700532547273111,
      "learning_rate": 7.88834378713998e-05,
      "loss": 11.8178,
      "step": 13282
    },
    {
      "epoch": 2.322710382513661,
      "grad_norm": 0.6944756514789144,
      "learning_rate": 7.88695965205489e-05,
      "loss": 11.9598,
      "step": 13283
    },
    {
      "epoch": 2.3228852459016394,
      "grad_norm": 0.6198194858916032,
      "learning_rate": 7.885575559342751e-05,
      "loss": 11.9687,
      "step": 13284
    },
    {
      "epoch": 2.3230601092896173,
      "grad_norm": 0.7078950839400565,
      "learning_rate": 7.884191509031305e-05,
      "loss": 11.7446,
      "step": 13285
    },
    {
      "epoch": 2.3232349726775956,
      "grad_norm": 0.7003661068753279,
      "learning_rate": 7.882807501148324e-05,
      "loss": 12.0478,
      "step": 13286
    },
    {
      "epoch": 2.323409836065574,
      "grad_norm": 0.69667446701389,
      "learning_rate": 7.881423535721551e-05,
      "loss": 12.0506,
      "step": 13287
    },
    {
      "epoch": 2.323584699453552,
      "grad_norm": 0.6523740080747258,
      "learning_rate": 7.880039612778741e-05,
      "loss": 11.9308,
      "step": 13288
    },
    {
      "epoch": 2.32375956284153,
      "grad_norm": 0.6433038201817262,
      "learning_rate": 7.878655732347649e-05,
      "loss": 11.9446,
      "step": 13289
    },
    {
      "epoch": 2.323934426229508,
      "grad_norm": 0.6421246333242651,
      "learning_rate": 7.877271894456016e-05,
      "loss": 11.8072,
      "step": 13290
    },
    {
      "epoch": 2.3241092896174864,
      "grad_norm": 0.6323203604084652,
      "learning_rate": 7.875888099131603e-05,
      "loss": 12.1212,
      "step": 13291
    },
    {
      "epoch": 2.3242841530054643,
      "grad_norm": 0.6312461806691588,
      "learning_rate": 7.874504346402156e-05,
      "loss": 12.0044,
      "step": 13292
    },
    {
      "epoch": 2.3244590163934427,
      "grad_norm": 0.6140657468393509,
      "learning_rate": 7.873120636295425e-05,
      "loss": 11.8819,
      "step": 13293
    },
    {
      "epoch": 2.3246338797814206,
      "grad_norm": 0.6345383283721425,
      "learning_rate": 7.871736968839147e-05,
      "loss": 12.1012,
      "step": 13294
    },
    {
      "epoch": 2.324808743169399,
      "grad_norm": 0.6442554270044709,
      "learning_rate": 7.870353344061087e-05,
      "loss": 11.5958,
      "step": 13295
    },
    {
      "epoch": 2.3249836065573772,
      "grad_norm": 0.6994690353511193,
      "learning_rate": 7.868969761988978e-05,
      "loss": 11.8597,
      "step": 13296
    },
    {
      "epoch": 2.325158469945355,
      "grad_norm": 0.6537789917901372,
      "learning_rate": 7.867586222650568e-05,
      "loss": 11.5611,
      "step": 13297
    },
    {
      "epoch": 2.3253333333333335,
      "grad_norm": 0.7010444500594964,
      "learning_rate": 7.866202726073604e-05,
      "loss": 11.6006,
      "step": 13298
    },
    {
      "epoch": 2.3255081967213114,
      "grad_norm": 0.6871456819748436,
      "learning_rate": 7.864819272285822e-05,
      "loss": 11.8995,
      "step": 13299
    },
    {
      "epoch": 2.3256830601092897,
      "grad_norm": 0.5620997554316338,
      "learning_rate": 7.863435861314972e-05,
      "loss": 11.8316,
      "step": 13300
    },
    {
      "epoch": 2.3258579234972676,
      "grad_norm": 0.7012718060837548,
      "learning_rate": 7.862052493188793e-05,
      "loss": 11.8822,
      "step": 13301
    },
    {
      "epoch": 2.326032786885246,
      "grad_norm": 0.6538121154810157,
      "learning_rate": 7.860669167935028e-05,
      "loss": 11.4977,
      "step": 13302
    },
    {
      "epoch": 2.326207650273224,
      "grad_norm": 0.6577383958071923,
      "learning_rate": 7.859285885581413e-05,
      "loss": 11.771,
      "step": 13303
    },
    {
      "epoch": 2.326382513661202,
      "grad_norm": 0.7621137905137018,
      "learning_rate": 7.857902646155687e-05,
      "loss": 12.0714,
      "step": 13304
    },
    {
      "epoch": 2.3265573770491805,
      "grad_norm": 0.6296442099426843,
      "learning_rate": 7.856519449685591e-05,
      "loss": 11.8518,
      "step": 13305
    },
    {
      "epoch": 2.3267322404371584,
      "grad_norm": 0.7166329415399756,
      "learning_rate": 7.855136296198863e-05,
      "loss": 11.7932,
      "step": 13306
    },
    {
      "epoch": 2.3269071038251368,
      "grad_norm": 0.6909865430527634,
      "learning_rate": 7.853753185723238e-05,
      "loss": 11.8505,
      "step": 13307
    },
    {
      "epoch": 2.3270819672131147,
      "grad_norm": 0.6955665860154243,
      "learning_rate": 7.852370118286452e-05,
      "loss": 12.0469,
      "step": 13308
    },
    {
      "epoch": 2.327256830601093,
      "grad_norm": 0.6222417206076539,
      "learning_rate": 7.850987093916235e-05,
      "loss": 12.021,
      "step": 13309
    },
    {
      "epoch": 2.327431693989071,
      "grad_norm": 0.622844456603069,
      "learning_rate": 7.849604112640327e-05,
      "loss": 12.0785,
      "step": 13310
    },
    {
      "epoch": 2.3276065573770492,
      "grad_norm": 0.6168513738102088,
      "learning_rate": 7.848221174486463e-05,
      "loss": 12.0725,
      "step": 13311
    },
    {
      "epoch": 2.327781420765027,
      "grad_norm": 0.6990958235530824,
      "learning_rate": 7.846838279482368e-05,
      "loss": 12.0667,
      "step": 13312
    },
    {
      "epoch": 2.3279562841530055,
      "grad_norm": 0.6566001645599874,
      "learning_rate": 7.845455427655776e-05,
      "loss": 11.8307,
      "step": 13313
    },
    {
      "epoch": 2.328131147540984,
      "grad_norm": 0.6656395529238247,
      "learning_rate": 7.844072619034417e-05,
      "loss": 11.8212,
      "step": 13314
    },
    {
      "epoch": 2.3283060109289617,
      "grad_norm": 0.6921559915288455,
      "learning_rate": 7.842689853646024e-05,
      "loss": 12.0544,
      "step": 13315
    },
    {
      "epoch": 2.32848087431694,
      "grad_norm": 0.6261992528570915,
      "learning_rate": 7.841307131518323e-05,
      "loss": 11.6411,
      "step": 13316
    },
    {
      "epoch": 2.328655737704918,
      "grad_norm": 0.6689968513573482,
      "learning_rate": 7.839924452679042e-05,
      "loss": 11.9551,
      "step": 13317
    },
    {
      "epoch": 2.3288306010928963,
      "grad_norm": 0.7112328669576433,
      "learning_rate": 7.838541817155907e-05,
      "loss": 11.9634,
      "step": 13318
    },
    {
      "epoch": 2.329005464480874,
      "grad_norm": 0.7122747837425079,
      "learning_rate": 7.837159224976645e-05,
      "loss": 11.8958,
      "step": 13319
    },
    {
      "epoch": 2.3291803278688525,
      "grad_norm": 0.7616969894759156,
      "learning_rate": 7.835776676168985e-05,
      "loss": 11.9467,
      "step": 13320
    },
    {
      "epoch": 2.3293551912568304,
      "grad_norm": 0.6914416398265304,
      "learning_rate": 7.834394170760644e-05,
      "loss": 11.8425,
      "step": 13321
    },
    {
      "epoch": 2.329530054644809,
      "grad_norm": 0.638693577986285,
      "learning_rate": 7.83301170877935e-05,
      "loss": 11.86,
      "step": 13322
    },
    {
      "epoch": 2.329704918032787,
      "grad_norm": 0.7284351308865451,
      "learning_rate": 7.831629290252823e-05,
      "loss": 11.8258,
      "step": 13323
    },
    {
      "epoch": 2.329879781420765,
      "grad_norm": 0.6931216975760816,
      "learning_rate": 7.830246915208788e-05,
      "loss": 11.602,
      "step": 13324
    },
    {
      "epoch": 2.3300546448087434,
      "grad_norm": 0.6450427106538706,
      "learning_rate": 7.828864583674966e-05,
      "loss": 11.949,
      "step": 13325
    },
    {
      "epoch": 2.3302295081967213,
      "grad_norm": 0.7977026534136424,
      "learning_rate": 7.827482295679076e-05,
      "loss": 11.7766,
      "step": 13326
    },
    {
      "epoch": 2.3304043715846996,
      "grad_norm": 0.7025700720864221,
      "learning_rate": 7.826100051248834e-05,
      "loss": 11.6766,
      "step": 13327
    },
    {
      "epoch": 2.3305792349726775,
      "grad_norm": 0.70232379424004,
      "learning_rate": 7.824717850411958e-05,
      "loss": 11.7173,
      "step": 13328
    },
    {
      "epoch": 2.330754098360656,
      "grad_norm": 0.6737439724608456,
      "learning_rate": 7.823335693196174e-05,
      "loss": 11.8305,
      "step": 13329
    },
    {
      "epoch": 2.3309289617486337,
      "grad_norm": 0.6658405093469723,
      "learning_rate": 7.82195357962919e-05,
      "loss": 11.9367,
      "step": 13330
    },
    {
      "epoch": 2.331103825136612,
      "grad_norm": 0.7447411980097597,
      "learning_rate": 7.820571509738723e-05,
      "loss": 11.9733,
      "step": 13331
    },
    {
      "epoch": 2.33127868852459,
      "grad_norm": 0.6883229337865678,
      "learning_rate": 7.819189483552493e-05,
      "loss": 11.7192,
      "step": 13332
    },
    {
      "epoch": 2.3314535519125683,
      "grad_norm": 0.6440172546477145,
      "learning_rate": 7.817807501098203e-05,
      "loss": 11.9938,
      "step": 13333
    },
    {
      "epoch": 2.331628415300546,
      "grad_norm": 0.6474968128309885,
      "learning_rate": 7.816425562403578e-05,
      "loss": 12.1683,
      "step": 13334
    },
    {
      "epoch": 2.3318032786885245,
      "grad_norm": 0.7363005499645948,
      "learning_rate": 7.815043667496324e-05,
      "loss": 11.8328,
      "step": 13335
    },
    {
      "epoch": 2.331978142076503,
      "grad_norm": 0.6383922608507366,
      "learning_rate": 7.813661816404152e-05,
      "loss": 12.0232,
      "step": 13336
    },
    {
      "epoch": 2.332153005464481,
      "grad_norm": 0.6422949479204235,
      "learning_rate": 7.812280009154776e-05,
      "loss": 11.8839,
      "step": 13337
    },
    {
      "epoch": 2.332327868852459,
      "grad_norm": 0.7772237907902777,
      "learning_rate": 7.810898245775899e-05,
      "loss": 11.6562,
      "step": 13338
    },
    {
      "epoch": 2.332502732240437,
      "grad_norm": 0.6574654814460013,
      "learning_rate": 7.809516526295236e-05,
      "loss": 12.0116,
      "step": 13339
    },
    {
      "epoch": 2.3326775956284154,
      "grad_norm": 0.7352181478827086,
      "learning_rate": 7.808134850740493e-05,
      "loss": 12.0483,
      "step": 13340
    },
    {
      "epoch": 2.3328524590163933,
      "grad_norm": 0.7300754069520963,
      "learning_rate": 7.806753219139377e-05,
      "loss": 12.0802,
      "step": 13341
    },
    {
      "epoch": 2.3330273224043716,
      "grad_norm": 0.6087636951515182,
      "learning_rate": 7.805371631519587e-05,
      "loss": 12.0033,
      "step": 13342
    },
    {
      "epoch": 2.3332021857923495,
      "grad_norm": 0.6432907829426172,
      "learning_rate": 7.803990087908841e-05,
      "loss": 12.1855,
      "step": 13343
    },
    {
      "epoch": 2.333377049180328,
      "grad_norm": 0.6528310268202752,
      "learning_rate": 7.802608588334834e-05,
      "loss": 12.0429,
      "step": 13344
    },
    {
      "epoch": 2.333551912568306,
      "grad_norm": 0.6343206507665934,
      "learning_rate": 7.801227132825272e-05,
      "loss": 12.0658,
      "step": 13345
    },
    {
      "epoch": 2.333726775956284,
      "grad_norm": 0.7328377978853865,
      "learning_rate": 7.799845721407859e-05,
      "loss": 12.0039,
      "step": 13346
    },
    {
      "epoch": 2.3339016393442624,
      "grad_norm": 0.6942942230412502,
      "learning_rate": 7.79846435411029e-05,
      "loss": 11.7114,
      "step": 13347
    },
    {
      "epoch": 2.3340765027322403,
      "grad_norm": 0.6247203701088058,
      "learning_rate": 7.797083030960273e-05,
      "loss": 11.7284,
      "step": 13348
    },
    {
      "epoch": 2.3342513661202187,
      "grad_norm": 0.6103247449344018,
      "learning_rate": 7.795701751985505e-05,
      "loss": 11.6134,
      "step": 13349
    },
    {
      "epoch": 2.3344262295081966,
      "grad_norm": 0.6334684824312324,
      "learning_rate": 7.794320517213687e-05,
      "loss": 11.9834,
      "step": 13350
    },
    {
      "epoch": 2.334601092896175,
      "grad_norm": 0.6465137519045024,
      "learning_rate": 7.792939326672513e-05,
      "loss": 11.6984,
      "step": 13351
    },
    {
      "epoch": 2.334775956284153,
      "grad_norm": 0.6936658992235475,
      "learning_rate": 7.791558180389678e-05,
      "loss": 11.8451,
      "step": 13352
    },
    {
      "epoch": 2.334950819672131,
      "grad_norm": 0.6151063363826672,
      "learning_rate": 7.790177078392887e-05,
      "loss": 11.7506,
      "step": 13353
    },
    {
      "epoch": 2.3351256830601095,
      "grad_norm": 0.7101815996488837,
      "learning_rate": 7.78879602070983e-05,
      "loss": 11.8846,
      "step": 13354
    },
    {
      "epoch": 2.3353005464480874,
      "grad_norm": 0.6467191020880381,
      "learning_rate": 7.787415007368204e-05,
      "loss": 12.0263,
      "step": 13355
    },
    {
      "epoch": 2.3354754098360657,
      "grad_norm": 0.6513180559750373,
      "learning_rate": 7.786034038395697e-05,
      "loss": 11.5055,
      "step": 13356
    },
    {
      "epoch": 2.3356502732240436,
      "grad_norm": 0.6158458324680458,
      "learning_rate": 7.784653113820006e-05,
      "loss": 11.6685,
      "step": 13357
    },
    {
      "epoch": 2.335825136612022,
      "grad_norm": 0.6550648962458473,
      "learning_rate": 7.783272233668821e-05,
      "loss": 12.0126,
      "step": 13358
    },
    {
      "epoch": 2.336,
      "grad_norm": 0.6126579083539981,
      "learning_rate": 7.781891397969838e-05,
      "loss": 12.0422,
      "step": 13359
    },
    {
      "epoch": 2.336174863387978,
      "grad_norm": 0.6845514800370396,
      "learning_rate": 7.780510606750742e-05,
      "loss": 11.6943,
      "step": 13360
    },
    {
      "epoch": 2.336349726775956,
      "grad_norm": 0.672456549860395,
      "learning_rate": 7.779129860039222e-05,
      "loss": 11.9137,
      "step": 13361
    },
    {
      "epoch": 2.3365245901639344,
      "grad_norm": 0.6216323940981512,
      "learning_rate": 7.777749157862963e-05,
      "loss": 11.9582,
      "step": 13362
    },
    {
      "epoch": 2.3366994535519128,
      "grad_norm": 0.7254629054633509,
      "learning_rate": 7.776368500249663e-05,
      "loss": 12.0226,
      "step": 13363
    },
    {
      "epoch": 2.3368743169398907,
      "grad_norm": 0.6114550421690393,
      "learning_rate": 7.774987887227001e-05,
      "loss": 12.0389,
      "step": 13364
    },
    {
      "epoch": 2.337049180327869,
      "grad_norm": 0.6518218190821096,
      "learning_rate": 7.773607318822663e-05,
      "loss": 11.9715,
      "step": 13365
    },
    {
      "epoch": 2.337224043715847,
      "grad_norm": 0.650857941431602,
      "learning_rate": 7.772226795064331e-05,
      "loss": 12.0578,
      "step": 13366
    },
    {
      "epoch": 2.3373989071038253,
      "grad_norm": 0.6608435235694896,
      "learning_rate": 7.770846315979693e-05,
      "loss": 12.0213,
      "step": 13367
    },
    {
      "epoch": 2.337573770491803,
      "grad_norm": 0.6168821628216198,
      "learning_rate": 7.769465881596434e-05,
      "loss": 11.8693,
      "step": 13368
    },
    {
      "epoch": 2.3377486338797815,
      "grad_norm": 0.6818477333697299,
      "learning_rate": 7.768085491942231e-05,
      "loss": 11.9418,
      "step": 13369
    },
    {
      "epoch": 2.3379234972677594,
      "grad_norm": 0.7015081164069398,
      "learning_rate": 7.766705147044766e-05,
      "loss": 11.9624,
      "step": 13370
    },
    {
      "epoch": 2.3380983606557377,
      "grad_norm": 0.6879425974784453,
      "learning_rate": 7.765324846931718e-05,
      "loss": 11.9608,
      "step": 13371
    },
    {
      "epoch": 2.338273224043716,
      "grad_norm": 0.6819964856807749,
      "learning_rate": 7.763944591630772e-05,
      "loss": 11.7528,
      "step": 13372
    },
    {
      "epoch": 2.338448087431694,
      "grad_norm": 0.6672313673042042,
      "learning_rate": 7.762564381169603e-05,
      "loss": 11.6501,
      "step": 13373
    },
    {
      "epoch": 2.3386229508196723,
      "grad_norm": 0.7110093091202283,
      "learning_rate": 7.761184215575887e-05,
      "loss": 11.8389,
      "step": 13374
    },
    {
      "epoch": 2.33879781420765,
      "grad_norm": 0.6573355356203785,
      "learning_rate": 7.7598040948773e-05,
      "loss": 11.9989,
      "step": 13375
    },
    {
      "epoch": 2.3389726775956285,
      "grad_norm": 0.7888958386156772,
      "learning_rate": 7.758424019101518e-05,
      "loss": 11.9929,
      "step": 13376
    },
    {
      "epoch": 2.3391475409836064,
      "grad_norm": 0.7197563214184913,
      "learning_rate": 7.75704398827622e-05,
      "loss": 12.1404,
      "step": 13377
    },
    {
      "epoch": 2.339322404371585,
      "grad_norm": 0.6711706067907268,
      "learning_rate": 7.755664002429075e-05,
      "loss": 12.0456,
      "step": 13378
    },
    {
      "epoch": 2.3394972677595627,
      "grad_norm": 0.7091436545871839,
      "learning_rate": 7.75428406158776e-05,
      "loss": 11.7286,
      "step": 13379
    },
    {
      "epoch": 2.339672131147541,
      "grad_norm": 0.7072474903073486,
      "learning_rate": 7.752904165779945e-05,
      "loss": 12.006,
      "step": 13380
    },
    {
      "epoch": 2.3398469945355194,
      "grad_norm": 0.6311194334959355,
      "learning_rate": 7.751524315033294e-05,
      "loss": 11.8943,
      "step": 13381
    },
    {
      "epoch": 2.3400218579234973,
      "grad_norm": 0.7072134370403634,
      "learning_rate": 7.750144509375492e-05,
      "loss": 12.0532,
      "step": 13382
    },
    {
      "epoch": 2.3401967213114756,
      "grad_norm": 0.714284120310602,
      "learning_rate": 7.748764748834198e-05,
      "loss": 12.0141,
      "step": 13383
    },
    {
      "epoch": 2.3403715846994535,
      "grad_norm": 0.7351844606949064,
      "learning_rate": 7.747385033437082e-05,
      "loss": 11.9931,
      "step": 13384
    },
    {
      "epoch": 2.340546448087432,
      "grad_norm": 0.6610611148622503,
      "learning_rate": 7.746005363211816e-05,
      "loss": 11.7918,
      "step": 13385
    },
    {
      "epoch": 2.3407213114754097,
      "grad_norm": 0.7526946816906802,
      "learning_rate": 7.744625738186059e-05,
      "loss": 12.0073,
      "step": 13386
    },
    {
      "epoch": 2.340896174863388,
      "grad_norm": 0.6698529635321263,
      "learning_rate": 7.743246158387483e-05,
      "loss": 11.8609,
      "step": 13387
    },
    {
      "epoch": 2.341071038251366,
      "grad_norm": 0.7703761617408463,
      "learning_rate": 7.74186662384375e-05,
      "loss": 11.7845,
      "step": 13388
    },
    {
      "epoch": 2.3412459016393443,
      "grad_norm": 0.7431863087387022,
      "learning_rate": 7.740487134582525e-05,
      "loss": 12.0675,
      "step": 13389
    },
    {
      "epoch": 2.341420765027322,
      "grad_norm": 0.6906420790730456,
      "learning_rate": 7.739107690631467e-05,
      "loss": 11.6853,
      "step": 13390
    },
    {
      "epoch": 2.3415956284153006,
      "grad_norm": 0.7451647275632568,
      "learning_rate": 7.737728292018247e-05,
      "loss": 11.9842,
      "step": 13391
    },
    {
      "epoch": 2.3417704918032785,
      "grad_norm": 0.6774657073551041,
      "learning_rate": 7.736348938770519e-05,
      "loss": 11.7534,
      "step": 13392
    },
    {
      "epoch": 2.341945355191257,
      "grad_norm": 0.7606112601810068,
      "learning_rate": 7.734969630915944e-05,
      "loss": 11.7737,
      "step": 13393
    },
    {
      "epoch": 2.342120218579235,
      "grad_norm": 0.7273682625852194,
      "learning_rate": 7.733590368482187e-05,
      "loss": 12.0129,
      "step": 13394
    },
    {
      "epoch": 2.342295081967213,
      "grad_norm": 0.651609626583321,
      "learning_rate": 7.732211151496895e-05,
      "loss": 11.6544,
      "step": 13395
    },
    {
      "epoch": 2.3424699453551914,
      "grad_norm": 0.7761604939258907,
      "learning_rate": 7.730831979987736e-05,
      "loss": 11.981,
      "step": 13396
    },
    {
      "epoch": 2.3426448087431693,
      "grad_norm": 0.7941035118484997,
      "learning_rate": 7.729452853982362e-05,
      "loss": 11.7466,
      "step": 13397
    },
    {
      "epoch": 2.3428196721311476,
      "grad_norm": 0.723523219357611,
      "learning_rate": 7.728073773508433e-05,
      "loss": 11.8268,
      "step": 13398
    },
    {
      "epoch": 2.3429945355191255,
      "grad_norm": 0.7844592065041618,
      "learning_rate": 7.726694738593598e-05,
      "loss": 11.7243,
      "step": 13399
    },
    {
      "epoch": 2.343169398907104,
      "grad_norm": 0.7740578977572341,
      "learning_rate": 7.725315749265512e-05,
      "loss": 11.9395,
      "step": 13400
    },
    {
      "epoch": 2.3433442622950817,
      "grad_norm": 0.6634433985628634,
      "learning_rate": 7.723936805551831e-05,
      "loss": 12.0989,
      "step": 13401
    },
    {
      "epoch": 2.34351912568306,
      "grad_norm": 0.8184220309653405,
      "learning_rate": 7.722557907480206e-05,
      "loss": 12.1334,
      "step": 13402
    },
    {
      "epoch": 2.3436939890710384,
      "grad_norm": 0.7307741616444069,
      "learning_rate": 7.72117905507829e-05,
      "loss": 11.9631,
      "step": 13403
    },
    {
      "epoch": 2.3438688524590163,
      "grad_norm": 0.5926541008581685,
      "learning_rate": 7.719800248373726e-05,
      "loss": 11.8344,
      "step": 13404
    },
    {
      "epoch": 2.3440437158469947,
      "grad_norm": 0.7741615352771662,
      "learning_rate": 7.718421487394169e-05,
      "loss": 11.9549,
      "step": 13405
    },
    {
      "epoch": 2.3442185792349726,
      "grad_norm": 0.7419591868604647,
      "learning_rate": 7.717042772167266e-05,
      "loss": 11.6284,
      "step": 13406
    },
    {
      "epoch": 2.344393442622951,
      "grad_norm": 0.6611124005716997,
      "learning_rate": 7.71566410272067e-05,
      "loss": 11.7735,
      "step": 13407
    },
    {
      "epoch": 2.344568306010929,
      "grad_norm": 0.7019728683228974,
      "learning_rate": 7.714285479082019e-05,
      "loss": 11.8982,
      "step": 13408
    },
    {
      "epoch": 2.344743169398907,
      "grad_norm": 0.7531063395971332,
      "learning_rate": 7.712906901278962e-05,
      "loss": 11.97,
      "step": 13409
    },
    {
      "epoch": 2.344918032786885,
      "grad_norm": 0.6977837105000431,
      "learning_rate": 7.711528369339143e-05,
      "loss": 11.9665,
      "step": 13410
    },
    {
      "epoch": 2.3450928961748634,
      "grad_norm": 0.6470927145687608,
      "learning_rate": 7.710149883290207e-05,
      "loss": 12.0106,
      "step": 13411
    },
    {
      "epoch": 2.3452677595628417,
      "grad_norm": 0.6651194887300136,
      "learning_rate": 7.7087714431598e-05,
      "loss": 11.7389,
      "step": 13412
    },
    {
      "epoch": 2.3454426229508196,
      "grad_norm": 0.6200774220325994,
      "learning_rate": 7.707393048975558e-05,
      "loss": 11.7488,
      "step": 13413
    },
    {
      "epoch": 2.345617486338798,
      "grad_norm": 0.6713431201273277,
      "learning_rate": 7.706014700765123e-05,
      "loss": 11.8052,
      "step": 13414
    },
    {
      "epoch": 2.345792349726776,
      "grad_norm": 0.7116925309982199,
      "learning_rate": 7.704636398556138e-05,
      "loss": 11.8096,
      "step": 13415
    },
    {
      "epoch": 2.345967213114754,
      "grad_norm": 0.6459594246986736,
      "learning_rate": 7.703258142376244e-05,
      "loss": 11.8536,
      "step": 13416
    },
    {
      "epoch": 2.346142076502732,
      "grad_norm": 0.6401479277987329,
      "learning_rate": 7.701879932253073e-05,
      "loss": 11.8774,
      "step": 13417
    },
    {
      "epoch": 2.3463169398907104,
      "grad_norm": 0.7816643546487044,
      "learning_rate": 7.700501768214267e-05,
      "loss": 11.7244,
      "step": 13418
    },
    {
      "epoch": 2.3464918032786883,
      "grad_norm": 0.7238838324181143,
      "learning_rate": 7.699123650287457e-05,
      "loss": 11.8192,
      "step": 13419
    },
    {
      "epoch": 2.3466666666666667,
      "grad_norm": 0.6747149486219346,
      "learning_rate": 7.697745578500286e-05,
      "loss": 11.9501,
      "step": 13420
    },
    {
      "epoch": 2.346841530054645,
      "grad_norm": 0.699747257642589,
      "learning_rate": 7.696367552880386e-05,
      "loss": 11.527,
      "step": 13421
    },
    {
      "epoch": 2.347016393442623,
      "grad_norm": 0.6006851821947898,
      "learning_rate": 7.694989573455388e-05,
      "loss": 11.9625,
      "step": 13422
    },
    {
      "epoch": 2.3471912568306013,
      "grad_norm": 0.6476067187406018,
      "learning_rate": 7.693611640252927e-05,
      "loss": 12.0194,
      "step": 13423
    },
    {
      "epoch": 2.347366120218579,
      "grad_norm": 0.6809419408159861,
      "learning_rate": 7.692233753300631e-05,
      "loss": 11.7937,
      "step": 13424
    },
    {
      "epoch": 2.3475409836065575,
      "grad_norm": 0.6256566973893803,
      "learning_rate": 7.690855912626139e-05,
      "loss": 11.4966,
      "step": 13425
    },
    {
      "epoch": 2.3477158469945354,
      "grad_norm": 0.7123322512604962,
      "learning_rate": 7.689478118257074e-05,
      "loss": 12.0898,
      "step": 13426
    },
    {
      "epoch": 2.3478907103825137,
      "grad_norm": 0.6214680383237181,
      "learning_rate": 7.688100370221067e-05,
      "loss": 11.9603,
      "step": 13427
    },
    {
      "epoch": 2.3480655737704916,
      "grad_norm": 0.6405844027788161,
      "learning_rate": 7.686722668545747e-05,
      "loss": 11.8798,
      "step": 13428
    },
    {
      "epoch": 2.34824043715847,
      "grad_norm": 0.8010020607673962,
      "learning_rate": 7.685345013258735e-05,
      "loss": 11.8107,
      "step": 13429
    },
    {
      "epoch": 2.3484153005464483,
      "grad_norm": 0.6444211641079882,
      "learning_rate": 7.683967404387669e-05,
      "loss": 11.828,
      "step": 13430
    },
    {
      "epoch": 2.348590163934426,
      "grad_norm": 0.6828751107864088,
      "learning_rate": 7.682589841960164e-05,
      "loss": 11.446,
      "step": 13431
    },
    {
      "epoch": 2.3487650273224046,
      "grad_norm": 0.783742336956799,
      "learning_rate": 7.681212326003848e-05,
      "loss": 11.9142,
      "step": 13432
    },
    {
      "epoch": 2.3489398907103824,
      "grad_norm": 0.7094066754252706,
      "learning_rate": 7.679834856546349e-05,
      "loss": 11.9282,
      "step": 13433
    },
    {
      "epoch": 2.349114754098361,
      "grad_norm": 0.6525263990537704,
      "learning_rate": 7.678457433615278e-05,
      "loss": 11.5794,
      "step": 13434
    },
    {
      "epoch": 2.3492896174863387,
      "grad_norm": 0.6045372461581033,
      "learning_rate": 7.677080057238268e-05,
      "loss": 11.9639,
      "step": 13435
    },
    {
      "epoch": 2.349464480874317,
      "grad_norm": 0.6119193006335848,
      "learning_rate": 7.675702727442933e-05,
      "loss": 12.0779,
      "step": 13436
    },
    {
      "epoch": 2.349639344262295,
      "grad_norm": 0.6856987351743293,
      "learning_rate": 7.674325444256899e-05,
      "loss": 12.211,
      "step": 13437
    },
    {
      "epoch": 2.3498142076502733,
      "grad_norm": 0.6612712457379436,
      "learning_rate": 7.672948207707772e-05,
      "loss": 11.8655,
      "step": 13438
    },
    {
      "epoch": 2.3499890710382516,
      "grad_norm": 0.7410576252906598,
      "learning_rate": 7.671571017823187e-05,
      "loss": 11.7868,
      "step": 13439
    },
    {
      "epoch": 2.3501639344262295,
      "grad_norm": 0.6252141957661155,
      "learning_rate": 7.670193874630749e-05,
      "loss": 11.8959,
      "step": 13440
    },
    {
      "epoch": 2.3503387978142074,
      "grad_norm": 0.7072754016521271,
      "learning_rate": 7.668816778158078e-05,
      "loss": 11.6326,
      "step": 13441
    },
    {
      "epoch": 2.3505136612021857,
      "grad_norm": 0.6644758207209193,
      "learning_rate": 7.667439728432789e-05,
      "loss": 11.7237,
      "step": 13442
    },
    {
      "epoch": 2.350688524590164,
      "grad_norm": 0.6384158201744867,
      "learning_rate": 7.666062725482492e-05,
      "loss": 11.6096,
      "step": 13443
    },
    {
      "epoch": 2.350863387978142,
      "grad_norm": 0.6901340357629282,
      "learning_rate": 7.664685769334806e-05,
      "loss": 11.7196,
      "step": 13444
    },
    {
      "epoch": 2.3510382513661203,
      "grad_norm": 0.6884375305795596,
      "learning_rate": 7.66330886001734e-05,
      "loss": 11.8704,
      "step": 13445
    },
    {
      "epoch": 2.351213114754098,
      "grad_norm": 0.6356295814688363,
      "learning_rate": 7.661931997557709e-05,
      "loss": 11.8978,
      "step": 13446
    },
    {
      "epoch": 2.3513879781420766,
      "grad_norm": 0.6654182070078032,
      "learning_rate": 7.660555181983518e-05,
      "loss": 11.8435,
      "step": 13447
    },
    {
      "epoch": 2.3515628415300545,
      "grad_norm": 0.7011476119287335,
      "learning_rate": 7.659178413322375e-05,
      "loss": 11.9032,
      "step": 13448
    },
    {
      "epoch": 2.351737704918033,
      "grad_norm": 0.6739934646263427,
      "learning_rate": 7.657801691601896e-05,
      "loss": 12.1254,
      "step": 13449
    },
    {
      "epoch": 2.3519125683060107,
      "grad_norm": 0.7055047594794394,
      "learning_rate": 7.656425016849683e-05,
      "loss": 11.9612,
      "step": 13450
    },
    {
      "epoch": 2.352087431693989,
      "grad_norm": 0.6581648281240687,
      "learning_rate": 7.655048389093347e-05,
      "loss": 11.6658,
      "step": 13451
    },
    {
      "epoch": 2.3522622950819674,
      "grad_norm": 0.6081762786107094,
      "learning_rate": 7.653671808360488e-05,
      "loss": 11.8696,
      "step": 13452
    },
    {
      "epoch": 2.3524371584699453,
      "grad_norm": 0.6722663484573108,
      "learning_rate": 7.652295274678714e-05,
      "loss": 11.8996,
      "step": 13453
    },
    {
      "epoch": 2.3526120218579236,
      "grad_norm": 0.6842329701200159,
      "learning_rate": 7.650918788075628e-05,
      "loss": 11.9486,
      "step": 13454
    },
    {
      "epoch": 2.3527868852459015,
      "grad_norm": 0.6370529522760199,
      "learning_rate": 7.649542348578834e-05,
      "loss": 11.7824,
      "step": 13455
    },
    {
      "epoch": 2.35296174863388,
      "grad_norm": 0.7184134113451643,
      "learning_rate": 7.648165956215933e-05,
      "loss": 11.8911,
      "step": 13456
    },
    {
      "epoch": 2.3531366120218578,
      "grad_norm": 0.6948903500464302,
      "learning_rate": 7.646789611014525e-05,
      "loss": 11.9807,
      "step": 13457
    },
    {
      "epoch": 2.353311475409836,
      "grad_norm": 0.6308554167580304,
      "learning_rate": 7.645413313002207e-05,
      "loss": 11.7814,
      "step": 13458
    },
    {
      "epoch": 2.353486338797814,
      "grad_norm": 0.7291324008702942,
      "learning_rate": 7.644037062206585e-05,
      "loss": 12.0302,
      "step": 13459
    },
    {
      "epoch": 2.3536612021857923,
      "grad_norm": 0.5784245978701782,
      "learning_rate": 7.642660858655255e-05,
      "loss": 11.9064,
      "step": 13460
    },
    {
      "epoch": 2.3538360655737707,
      "grad_norm": 0.7527949769121537,
      "learning_rate": 7.641284702375811e-05,
      "loss": 12.0031,
      "step": 13461
    },
    {
      "epoch": 2.3540109289617486,
      "grad_norm": 0.644348770548093,
      "learning_rate": 7.639908593395849e-05,
      "loss": 12.0432,
      "step": 13462
    },
    {
      "epoch": 2.354185792349727,
      "grad_norm": 0.6785291268962415,
      "learning_rate": 7.638532531742969e-05,
      "loss": 11.7249,
      "step": 13463
    },
    {
      "epoch": 2.354360655737705,
      "grad_norm": 0.6833251822263293,
      "learning_rate": 7.637156517444764e-05,
      "loss": 12.1648,
      "step": 13464
    },
    {
      "epoch": 2.354535519125683,
      "grad_norm": 0.654774473097664,
      "learning_rate": 7.635780550528823e-05,
      "loss": 11.6154,
      "step": 13465
    },
    {
      "epoch": 2.354710382513661,
      "grad_norm": 0.7191404122459103,
      "learning_rate": 7.634404631022741e-05,
      "loss": 12.0521,
      "step": 13466
    },
    {
      "epoch": 2.3548852459016394,
      "grad_norm": 0.8573155626564857,
      "learning_rate": 7.633028758954109e-05,
      "loss": 11.9088,
      "step": 13467
    },
    {
      "epoch": 2.3550601092896173,
      "grad_norm": 0.6899187971133318,
      "learning_rate": 7.631652934350518e-05,
      "loss": 12.0093,
      "step": 13468
    },
    {
      "epoch": 2.3552349726775956,
      "grad_norm": 0.6893598825333782,
      "learning_rate": 7.63027715723956e-05,
      "loss": 11.9792,
      "step": 13469
    },
    {
      "epoch": 2.355409836065574,
      "grad_norm": 0.7300889779169261,
      "learning_rate": 7.62890142764882e-05,
      "loss": 11.8886,
      "step": 13470
    },
    {
      "epoch": 2.355584699453552,
      "grad_norm": 0.6810153852634062,
      "learning_rate": 7.627525745605886e-05,
      "loss": 11.9019,
      "step": 13471
    },
    {
      "epoch": 2.35575956284153,
      "grad_norm": 0.6804301857699143,
      "learning_rate": 7.626150111138342e-05,
      "loss": 11.7763,
      "step": 13472
    },
    {
      "epoch": 2.355934426229508,
      "grad_norm": 0.728791946168863,
      "learning_rate": 7.624774524273783e-05,
      "loss": 11.6583,
      "step": 13473
    },
    {
      "epoch": 2.3561092896174864,
      "grad_norm": 0.6223442022354227,
      "learning_rate": 7.623398985039784e-05,
      "loss": 11.9435,
      "step": 13474
    },
    {
      "epoch": 2.3562841530054643,
      "grad_norm": 0.6758025791009526,
      "learning_rate": 7.622023493463933e-05,
      "loss": 11.9728,
      "step": 13475
    },
    {
      "epoch": 2.3564590163934427,
      "grad_norm": 0.7040427076849316,
      "learning_rate": 7.620648049573815e-05,
      "loss": 11.8057,
      "step": 13476
    },
    {
      "epoch": 2.3566338797814206,
      "grad_norm": 0.6532565708830693,
      "learning_rate": 7.619272653397002e-05,
      "loss": 11.8847,
      "step": 13477
    },
    {
      "epoch": 2.356808743169399,
      "grad_norm": 0.6332215073152807,
      "learning_rate": 7.61789730496109e-05,
      "loss": 11.8674,
      "step": 13478
    },
    {
      "epoch": 2.3569836065573773,
      "grad_norm": 0.7199470731517896,
      "learning_rate": 7.616522004293648e-05,
      "loss": 11.6894,
      "step": 13479
    },
    {
      "epoch": 2.357158469945355,
      "grad_norm": 0.605043375423893,
      "learning_rate": 7.615146751422258e-05,
      "loss": 12.0222,
      "step": 13480
    },
    {
      "epoch": 2.3573333333333335,
      "grad_norm": 0.6871842269362273,
      "learning_rate": 7.613771546374499e-05,
      "loss": 11.8628,
      "step": 13481
    },
    {
      "epoch": 2.3575081967213114,
      "grad_norm": 0.6604595331576274,
      "learning_rate": 7.612396389177945e-05,
      "loss": 11.8073,
      "step": 13482
    },
    {
      "epoch": 2.3576830601092897,
      "grad_norm": 0.6199473792276802,
      "learning_rate": 7.611021279860175e-05,
      "loss": 11.9733,
      "step": 13483
    },
    {
      "epoch": 2.3578579234972676,
      "grad_norm": 0.7020235310715104,
      "learning_rate": 7.609646218448765e-05,
      "loss": 11.5589,
      "step": 13484
    },
    {
      "epoch": 2.358032786885246,
      "grad_norm": 0.704517544420229,
      "learning_rate": 7.608271204971287e-05,
      "loss": 12.0249,
      "step": 13485
    },
    {
      "epoch": 2.358207650273224,
      "grad_norm": 0.7226697327603839,
      "learning_rate": 7.60689623945531e-05,
      "loss": 11.9877,
      "step": 13486
    },
    {
      "epoch": 2.358382513661202,
      "grad_norm": 0.7062354390907233,
      "learning_rate": 7.605521321928419e-05,
      "loss": 11.8239,
      "step": 13487
    },
    {
      "epoch": 2.3585573770491806,
      "grad_norm": 0.6749454414852901,
      "learning_rate": 7.604146452418175e-05,
      "loss": 11.8827,
      "step": 13488
    },
    {
      "epoch": 2.3587322404371585,
      "grad_norm": 0.6659006983752909,
      "learning_rate": 7.60277163095215e-05,
      "loss": 11.998,
      "step": 13489
    },
    {
      "epoch": 2.358907103825137,
      "grad_norm": 0.6980471657816424,
      "learning_rate": 7.601396857557916e-05,
      "loss": 11.8888,
      "step": 13490
    },
    {
      "epoch": 2.3590819672131147,
      "grad_norm": 0.6474474597322284,
      "learning_rate": 7.600022132263037e-05,
      "loss": 11.9971,
      "step": 13491
    },
    {
      "epoch": 2.359256830601093,
      "grad_norm": 0.637959812572952,
      "learning_rate": 7.598647455095084e-05,
      "loss": 11.9581,
      "step": 13492
    },
    {
      "epoch": 2.359431693989071,
      "grad_norm": 0.6804148072507966,
      "learning_rate": 7.597272826081623e-05,
      "loss": 12.0051,
      "step": 13493
    },
    {
      "epoch": 2.3596065573770493,
      "grad_norm": 0.6942855476409863,
      "learning_rate": 7.59589824525022e-05,
      "loss": 11.9792,
      "step": 13494
    },
    {
      "epoch": 2.359781420765027,
      "grad_norm": 0.6957683695170348,
      "learning_rate": 7.594523712628438e-05,
      "loss": 11.7852,
      "step": 13495
    },
    {
      "epoch": 2.3599562841530055,
      "grad_norm": 0.6589138813236074,
      "learning_rate": 7.59314922824384e-05,
      "loss": 11.8369,
      "step": 13496
    },
    {
      "epoch": 2.360131147540984,
      "grad_norm": 0.6835192485304495,
      "learning_rate": 7.59177479212399e-05,
      "loss": 11.7891,
      "step": 13497
    },
    {
      "epoch": 2.3603060109289618,
      "grad_norm": 0.72707768819758,
      "learning_rate": 7.590400404296451e-05,
      "loss": 11.8522,
      "step": 13498
    },
    {
      "epoch": 2.3604808743169396,
      "grad_norm": 0.602598514945065,
      "learning_rate": 7.589026064788784e-05,
      "loss": 11.7041,
      "step": 13499
    },
    {
      "epoch": 2.360655737704918,
      "grad_norm": 0.6678760148279598,
      "learning_rate": 7.587651773628546e-05,
      "loss": 11.9674,
      "step": 13500
    },
    {
      "epoch": 2.3608306010928963,
      "grad_norm": 0.6509336596928607,
      "learning_rate": 7.586277530843293e-05,
      "loss": 11.8757,
      "step": 13501
    },
    {
      "epoch": 2.3610054644808742,
      "grad_norm": 0.6234833263495254,
      "learning_rate": 7.584903336460589e-05,
      "loss": 11.8972,
      "step": 13502
    },
    {
      "epoch": 2.3611803278688526,
      "grad_norm": 0.6268475030165495,
      "learning_rate": 7.583529190507992e-05,
      "loss": 12.0171,
      "step": 13503
    },
    {
      "epoch": 2.3613551912568305,
      "grad_norm": 0.7661575732916609,
      "learning_rate": 7.582155093013051e-05,
      "loss": 11.8958,
      "step": 13504
    },
    {
      "epoch": 2.361530054644809,
      "grad_norm": 0.675712981525013,
      "learning_rate": 7.580781044003324e-05,
      "loss": 12.0309,
      "step": 13505
    },
    {
      "epoch": 2.3617049180327867,
      "grad_norm": 0.6912366877105385,
      "learning_rate": 7.579407043506364e-05,
      "loss": 11.8305,
      "step": 13506
    },
    {
      "epoch": 2.361879781420765,
      "grad_norm": 0.6552319705946836,
      "learning_rate": 7.578033091549727e-05,
      "loss": 11.8889,
      "step": 13507
    },
    {
      "epoch": 2.362054644808743,
      "grad_norm": 0.7810744669739744,
      "learning_rate": 7.576659188160963e-05,
      "loss": 12.2661,
      "step": 13508
    },
    {
      "epoch": 2.3622295081967213,
      "grad_norm": 0.6446394246842074,
      "learning_rate": 7.575285333367624e-05,
      "loss": 12.0527,
      "step": 13509
    },
    {
      "epoch": 2.3624043715846996,
      "grad_norm": 0.6692638018483958,
      "learning_rate": 7.573911527197255e-05,
      "loss": 11.8986,
      "step": 13510
    },
    {
      "epoch": 2.3625792349726775,
      "grad_norm": 0.5977013625740246,
      "learning_rate": 7.572537769677411e-05,
      "loss": 11.9721,
      "step": 13511
    },
    {
      "epoch": 2.362754098360656,
      "grad_norm": 0.6831123878151473,
      "learning_rate": 7.571164060835641e-05,
      "loss": 11.6859,
      "step": 13512
    },
    {
      "epoch": 2.3629289617486338,
      "grad_norm": 0.6515669813456935,
      "learning_rate": 7.569790400699485e-05,
      "loss": 11.7772,
      "step": 13513
    },
    {
      "epoch": 2.363103825136612,
      "grad_norm": 0.6231418733637545,
      "learning_rate": 7.568416789296494e-05,
      "loss": 11.9165,
      "step": 13514
    },
    {
      "epoch": 2.36327868852459,
      "grad_norm": 0.6769777945399785,
      "learning_rate": 7.567043226654211e-05,
      "loss": 11.9512,
      "step": 13515
    },
    {
      "epoch": 2.3634535519125683,
      "grad_norm": 0.6787289073868347,
      "learning_rate": 7.565669712800183e-05,
      "loss": 11.7519,
      "step": 13516
    },
    {
      "epoch": 2.3636284153005462,
      "grad_norm": 0.710625151940548,
      "learning_rate": 7.564296247761954e-05,
      "loss": 11.8868,
      "step": 13517
    },
    {
      "epoch": 2.3638032786885246,
      "grad_norm": 0.6321788724207885,
      "learning_rate": 7.562922831567061e-05,
      "loss": 12.0723,
      "step": 13518
    },
    {
      "epoch": 2.363978142076503,
      "grad_norm": 0.6567709156604743,
      "learning_rate": 7.561549464243049e-05,
      "loss": 12.0538,
      "step": 13519
    },
    {
      "epoch": 2.364153005464481,
      "grad_norm": 0.641233330283685,
      "learning_rate": 7.560176145817456e-05,
      "loss": 11.9273,
      "step": 13520
    },
    {
      "epoch": 2.364327868852459,
      "grad_norm": 0.6426818933865279,
      "learning_rate": 7.558802876317825e-05,
      "loss": 11.7705,
      "step": 13521
    },
    {
      "epoch": 2.364502732240437,
      "grad_norm": 0.711907879618146,
      "learning_rate": 7.55742965577169e-05,
      "loss": 11.954,
      "step": 13522
    },
    {
      "epoch": 2.3646775956284154,
      "grad_norm": 0.6350898140851406,
      "learning_rate": 7.55605648420659e-05,
      "loss": 12.0067,
      "step": 13523
    },
    {
      "epoch": 2.3648524590163933,
      "grad_norm": 0.6916969382619125,
      "learning_rate": 7.554683361650064e-05,
      "loss": 12.0421,
      "step": 13524
    },
    {
      "epoch": 2.3650273224043716,
      "grad_norm": 0.6191409399553368,
      "learning_rate": 7.553310288129639e-05,
      "loss": 11.977,
      "step": 13525
    },
    {
      "epoch": 2.3652021857923495,
      "grad_norm": 0.7191010164278624,
      "learning_rate": 7.55193726367286e-05,
      "loss": 11.9645,
      "step": 13526
    },
    {
      "epoch": 2.365377049180328,
      "grad_norm": 0.676043714615962,
      "learning_rate": 7.550564288307254e-05,
      "loss": 11.6058,
      "step": 13527
    },
    {
      "epoch": 2.365551912568306,
      "grad_norm": 0.6912909008904993,
      "learning_rate": 7.549191362060355e-05,
      "loss": 11.9757,
      "step": 13528
    },
    {
      "epoch": 2.365726775956284,
      "grad_norm": 0.6182539457106863,
      "learning_rate": 7.547818484959696e-05,
      "loss": 12.1733,
      "step": 13529
    },
    {
      "epoch": 2.3659016393442625,
      "grad_norm": 0.6546623126944596,
      "learning_rate": 7.546445657032801e-05,
      "loss": 11.7037,
      "step": 13530
    },
    {
      "epoch": 2.3660765027322403,
      "grad_norm": 0.6245182154742117,
      "learning_rate": 7.545072878307206e-05,
      "loss": 12.0925,
      "step": 13531
    },
    {
      "epoch": 2.3662513661202187,
      "grad_norm": 0.6095478178024556,
      "learning_rate": 7.543700148810439e-05,
      "loss": 11.9434,
      "step": 13532
    },
    {
      "epoch": 2.3664262295081966,
      "grad_norm": 0.7377902312259775,
      "learning_rate": 7.542327468570026e-05,
      "loss": 11.9915,
      "step": 13533
    },
    {
      "epoch": 2.366601092896175,
      "grad_norm": 0.6442903875911083,
      "learning_rate": 7.540954837613488e-05,
      "loss": 11.8361,
      "step": 13534
    },
    {
      "epoch": 2.366775956284153,
      "grad_norm": 0.6028637306712679,
      "learning_rate": 7.539582255968362e-05,
      "loss": 11.7052,
      "step": 13535
    },
    {
      "epoch": 2.366950819672131,
      "grad_norm": 0.6882026768672325,
      "learning_rate": 7.538209723662165e-05,
      "loss": 11.9135,
      "step": 13536
    },
    {
      "epoch": 2.3671256830601095,
      "grad_norm": 0.6118796046486275,
      "learning_rate": 7.536837240722419e-05,
      "loss": 11.9716,
      "step": 13537
    },
    {
      "epoch": 2.3673005464480874,
      "grad_norm": 0.6605435852043932,
      "learning_rate": 7.535464807176654e-05,
      "loss": 12.0406,
      "step": 13538
    },
    {
      "epoch": 2.3674754098360657,
      "grad_norm": 0.7116002355126557,
      "learning_rate": 7.534092423052381e-05,
      "loss": 11.7515,
      "step": 13539
    },
    {
      "epoch": 2.3676502732240436,
      "grad_norm": 0.6192780799313202,
      "learning_rate": 7.532720088377129e-05,
      "loss": 11.8857,
      "step": 13540
    },
    {
      "epoch": 2.367825136612022,
      "grad_norm": 0.6952140047116062,
      "learning_rate": 7.531347803178415e-05,
      "loss": 11.8689,
      "step": 13541
    },
    {
      "epoch": 2.368,
      "grad_norm": 0.6778021938416833,
      "learning_rate": 7.529975567483758e-05,
      "loss": 11.9642,
      "step": 13542
    },
    {
      "epoch": 2.3681748633879782,
      "grad_norm": 0.6916943836624024,
      "learning_rate": 7.528603381320673e-05,
      "loss": 11.7148,
      "step": 13543
    },
    {
      "epoch": 2.368349726775956,
      "grad_norm": 0.678011342590073,
      "learning_rate": 7.527231244716678e-05,
      "loss": 11.8482,
      "step": 13544
    },
    {
      "epoch": 2.3685245901639345,
      "grad_norm": 0.7110632192477794,
      "learning_rate": 7.525859157699291e-05,
      "loss": 11.7072,
      "step": 13545
    },
    {
      "epoch": 2.368699453551913,
      "grad_norm": 0.6397454083647601,
      "learning_rate": 7.524487120296024e-05,
      "loss": 11.7303,
      "step": 13546
    },
    {
      "epoch": 2.3688743169398907,
      "grad_norm": 0.6496631671340223,
      "learning_rate": 7.523115132534392e-05,
      "loss": 11.78,
      "step": 13547
    },
    {
      "epoch": 2.369049180327869,
      "grad_norm": 0.6545340476949314,
      "learning_rate": 7.521743194441904e-05,
      "loss": 12.0477,
      "step": 13548
    },
    {
      "epoch": 2.369224043715847,
      "grad_norm": 0.656398448865471,
      "learning_rate": 7.520371306046074e-05,
      "loss": 11.484,
      "step": 13549
    },
    {
      "epoch": 2.3693989071038253,
      "grad_norm": 0.679033205546624,
      "learning_rate": 7.518999467374414e-05,
      "loss": 11.9737,
      "step": 13550
    },
    {
      "epoch": 2.369573770491803,
      "grad_norm": 0.7284039752335498,
      "learning_rate": 7.517627678454435e-05,
      "loss": 11.9317,
      "step": 13551
    },
    {
      "epoch": 2.3697486338797815,
      "grad_norm": 0.6972079970336887,
      "learning_rate": 7.516255939313641e-05,
      "loss": 11.9522,
      "step": 13552
    },
    {
      "epoch": 2.3699234972677594,
      "grad_norm": 0.7425026775352961,
      "learning_rate": 7.514884249979541e-05,
      "loss": 11.9766,
      "step": 13553
    },
    {
      "epoch": 2.3700983606557378,
      "grad_norm": 0.7371188183079421,
      "learning_rate": 7.513512610479641e-05,
      "loss": 11.8167,
      "step": 13554
    },
    {
      "epoch": 2.3702732240437157,
      "grad_norm": 0.6660374989831622,
      "learning_rate": 7.512141020841448e-05,
      "loss": 11.706,
      "step": 13555
    },
    {
      "epoch": 2.370448087431694,
      "grad_norm": 0.6785162819620992,
      "learning_rate": 7.51076948109247e-05,
      "loss": 12.0344,
      "step": 13556
    },
    {
      "epoch": 2.370622950819672,
      "grad_norm": 0.6568689732318523,
      "learning_rate": 7.509397991260202e-05,
      "loss": 11.927,
      "step": 13557
    },
    {
      "epoch": 2.3707978142076502,
      "grad_norm": 0.616746625574742,
      "learning_rate": 7.508026551372153e-05,
      "loss": 11.8165,
      "step": 13558
    },
    {
      "epoch": 2.3709726775956286,
      "grad_norm": 0.6508337097596869,
      "learning_rate": 7.506655161455822e-05,
      "loss": 11.5788,
      "step": 13559
    },
    {
      "epoch": 2.3711475409836065,
      "grad_norm": 0.6555991572601569,
      "learning_rate": 7.505283821538715e-05,
      "loss": 11.861,
      "step": 13560
    },
    {
      "epoch": 2.371322404371585,
      "grad_norm": 0.68710459754257,
      "learning_rate": 7.503912531648324e-05,
      "loss": 11.7031,
      "step": 13561
    },
    {
      "epoch": 2.3714972677595627,
      "grad_norm": 0.6808585217124887,
      "learning_rate": 7.502541291812151e-05,
      "loss": 11.7517,
      "step": 13562
    },
    {
      "epoch": 2.371672131147541,
      "grad_norm": 0.5785640722815231,
      "learning_rate": 7.50117010205769e-05,
      "loss": 11.9837,
      "step": 13563
    },
    {
      "epoch": 2.371846994535519,
      "grad_norm": 0.7028992938281964,
      "learning_rate": 7.499798962412446e-05,
      "loss": 11.6196,
      "step": 13564
    },
    {
      "epoch": 2.3720218579234973,
      "grad_norm": 0.6433316885508229,
      "learning_rate": 7.498427872903908e-05,
      "loss": 11.9051,
      "step": 13565
    },
    {
      "epoch": 2.372196721311475,
      "grad_norm": 0.6543819581065884,
      "learning_rate": 7.497056833559573e-05,
      "loss": 11.5234,
      "step": 13566
    },
    {
      "epoch": 2.3723715846994535,
      "grad_norm": 0.6407698970416232,
      "learning_rate": 7.495685844406933e-05,
      "loss": 11.9918,
      "step": 13567
    },
    {
      "epoch": 2.372546448087432,
      "grad_norm": 0.6606020061135048,
      "learning_rate": 7.494314905473476e-05,
      "loss": 11.7742,
      "step": 13568
    },
    {
      "epoch": 2.3727213114754098,
      "grad_norm": 0.6654774062041408,
      "learning_rate": 7.492944016786705e-05,
      "loss": 11.7485,
      "step": 13569
    },
    {
      "epoch": 2.372896174863388,
      "grad_norm": 0.6595299097840834,
      "learning_rate": 7.491573178374102e-05,
      "loss": 11.8476,
      "step": 13570
    },
    {
      "epoch": 2.373071038251366,
      "grad_norm": 0.6520376076807386,
      "learning_rate": 7.490202390263159e-05,
      "loss": 11.9662,
      "step": 13571
    },
    {
      "epoch": 2.3732459016393443,
      "grad_norm": 0.6233696423786591,
      "learning_rate": 7.488831652481365e-05,
      "loss": 11.9659,
      "step": 13572
    },
    {
      "epoch": 2.3734207650273222,
      "grad_norm": 0.6949335628473059,
      "learning_rate": 7.487460965056201e-05,
      "loss": 11.9037,
      "step": 13573
    },
    {
      "epoch": 2.3735956284153006,
      "grad_norm": 0.6912502808298344,
      "learning_rate": 7.486090328015166e-05,
      "loss": 11.9858,
      "step": 13574
    },
    {
      "epoch": 2.3737704918032785,
      "grad_norm": 0.7120178902388793,
      "learning_rate": 7.484719741385735e-05,
      "loss": 11.8445,
      "step": 13575
    },
    {
      "epoch": 2.373945355191257,
      "grad_norm": 0.6471534994495111,
      "learning_rate": 7.4833492051954e-05,
      "loss": 11.6919,
      "step": 13576
    },
    {
      "epoch": 2.374120218579235,
      "grad_norm": 0.6477439279768838,
      "learning_rate": 7.481978719471638e-05,
      "loss": 11.8641,
      "step": 13577
    },
    {
      "epoch": 2.374295081967213,
      "grad_norm": 0.6609344762033271,
      "learning_rate": 7.480608284241933e-05,
      "loss": 11.5904,
      "step": 13578
    },
    {
      "epoch": 2.3744699453551914,
      "grad_norm": 0.675734334905356,
      "learning_rate": 7.479237899533766e-05,
      "loss": 11.9575,
      "step": 13579
    },
    {
      "epoch": 2.3746448087431693,
      "grad_norm": 0.7100204288230192,
      "learning_rate": 7.477867565374622e-05,
      "loss": 11.8348,
      "step": 13580
    },
    {
      "epoch": 2.3748196721311476,
      "grad_norm": 0.5932655333502288,
      "learning_rate": 7.476497281791978e-05,
      "loss": 11.8466,
      "step": 13581
    },
    {
      "epoch": 2.3749945355191255,
      "grad_norm": 0.6616690169948828,
      "learning_rate": 7.475127048813306e-05,
      "loss": 11.9896,
      "step": 13582
    },
    {
      "epoch": 2.375169398907104,
      "grad_norm": 0.7512307220591963,
      "learning_rate": 7.473756866466095e-05,
      "loss": 11.6492,
      "step": 13583
    },
    {
      "epoch": 2.3753442622950818,
      "grad_norm": 0.7493921404276677,
      "learning_rate": 7.472386734777814e-05,
      "loss": 11.7587,
      "step": 13584
    },
    {
      "epoch": 2.37551912568306,
      "grad_norm": 0.8226502926430972,
      "learning_rate": 7.47101665377594e-05,
      "loss": 11.8377,
      "step": 13585
    },
    {
      "epoch": 2.3756939890710385,
      "grad_norm": 0.6708692562089031,
      "learning_rate": 7.469646623487948e-05,
      "loss": 11.9751,
      "step": 13586
    },
    {
      "epoch": 2.3758688524590164,
      "grad_norm": 0.6779024795321966,
      "learning_rate": 7.468276643941306e-05,
      "loss": 11.9908,
      "step": 13587
    },
    {
      "epoch": 2.3760437158469947,
      "grad_norm": 0.7218661465405168,
      "learning_rate": 7.466906715163495e-05,
      "loss": 11.8835,
      "step": 13588
    },
    {
      "epoch": 2.3762185792349726,
      "grad_norm": 0.6955025762707142,
      "learning_rate": 7.465536837181979e-05,
      "loss": 12.198,
      "step": 13589
    },
    {
      "epoch": 2.376393442622951,
      "grad_norm": 0.784601874423983,
      "learning_rate": 7.464167010024235e-05,
      "loss": 11.7172,
      "step": 13590
    },
    {
      "epoch": 2.376568306010929,
      "grad_norm": 0.7377954019720017,
      "learning_rate": 7.462797233717726e-05,
      "loss": 12.1009,
      "step": 13591
    },
    {
      "epoch": 2.376743169398907,
      "grad_norm": 0.672894642534143,
      "learning_rate": 7.461427508289922e-05,
      "loss": 11.8513,
      "step": 13592
    },
    {
      "epoch": 2.376918032786885,
      "grad_norm": 0.7471094282773832,
      "learning_rate": 7.460057833768292e-05,
      "loss": 11.6283,
      "step": 13593
    },
    {
      "epoch": 2.3770928961748634,
      "grad_norm": 0.8619663739792348,
      "learning_rate": 7.4586882101803e-05,
      "loss": 11.766,
      "step": 13594
    },
    {
      "epoch": 2.3772677595628418,
      "grad_norm": 0.662262091848667,
      "learning_rate": 7.457318637553415e-05,
      "loss": 11.8588,
      "step": 13595
    },
    {
      "epoch": 2.3774426229508197,
      "grad_norm": 0.8097770680299827,
      "learning_rate": 7.455949115915098e-05,
      "loss": 11.5174,
      "step": 13596
    },
    {
      "epoch": 2.377617486338798,
      "grad_norm": 0.6559271075064508,
      "learning_rate": 7.454579645292808e-05,
      "loss": 11.7445,
      "step": 13597
    },
    {
      "epoch": 2.377792349726776,
      "grad_norm": 0.6519558054645874,
      "learning_rate": 7.453210225714013e-05,
      "loss": 11.9026,
      "step": 13598
    },
    {
      "epoch": 2.3779672131147542,
      "grad_norm": 0.8077050754821604,
      "learning_rate": 7.451840857206178e-05,
      "loss": 11.8333,
      "step": 13599
    },
    {
      "epoch": 2.378142076502732,
      "grad_norm": 0.7414840692007355,
      "learning_rate": 7.450471539796753e-05,
      "loss": 11.9682,
      "step": 13600
    },
    {
      "epoch": 2.3783169398907105,
      "grad_norm": 0.6259067880327553,
      "learning_rate": 7.449102273513202e-05,
      "loss": 11.9291,
      "step": 13601
    },
    {
      "epoch": 2.3784918032786884,
      "grad_norm": 0.7945089514145642,
      "learning_rate": 7.447733058382981e-05,
      "loss": 12.015,
      "step": 13602
    },
    {
      "epoch": 2.3786666666666667,
      "grad_norm": 0.7875800068614273,
      "learning_rate": 7.446363894433552e-05,
      "loss": 11.9114,
      "step": 13603
    },
    {
      "epoch": 2.378841530054645,
      "grad_norm": 0.6188322296206676,
      "learning_rate": 7.444994781692369e-05,
      "loss": 12.0234,
      "step": 13604
    },
    {
      "epoch": 2.379016393442623,
      "grad_norm": 0.9404357274404902,
      "learning_rate": 7.443625720186882e-05,
      "loss": 11.8104,
      "step": 13605
    },
    {
      "epoch": 2.3791912568306013,
      "grad_norm": 0.7539808086848252,
      "learning_rate": 7.442256709944549e-05,
      "loss": 11.9929,
      "step": 13606
    },
    {
      "epoch": 2.379366120218579,
      "grad_norm": 0.6471795986643871,
      "learning_rate": 7.440887750992822e-05,
      "loss": 11.5615,
      "step": 13607
    },
    {
      "epoch": 2.3795409836065575,
      "grad_norm": 0.9774100190735434,
      "learning_rate": 7.439518843359156e-05,
      "loss": 12.1034,
      "step": 13608
    },
    {
      "epoch": 2.3797158469945354,
      "grad_norm": 0.653181819820652,
      "learning_rate": 7.438149987070997e-05,
      "loss": 11.7816,
      "step": 13609
    },
    {
      "epoch": 2.3798907103825138,
      "grad_norm": 0.6705414127368193,
      "learning_rate": 7.436781182155796e-05,
      "loss": 12.0065,
      "step": 13610
    },
    {
      "epoch": 2.3800655737704917,
      "grad_norm": 0.7492435701290587,
      "learning_rate": 7.435412428641001e-05,
      "loss": 11.795,
      "step": 13611
    },
    {
      "epoch": 2.38024043715847,
      "grad_norm": 0.7111028875003684,
      "learning_rate": 7.434043726554063e-05,
      "loss": 11.8371,
      "step": 13612
    },
    {
      "epoch": 2.380415300546448,
      "grad_norm": 0.6963885277147072,
      "learning_rate": 7.432675075922429e-05,
      "loss": 11.6976,
      "step": 13613
    },
    {
      "epoch": 2.3805901639344262,
      "grad_norm": 0.7414467239048392,
      "learning_rate": 7.431306476773541e-05,
      "loss": 11.9902,
      "step": 13614
    },
    {
      "epoch": 2.380765027322404,
      "grad_norm": 0.6524199827380994,
      "learning_rate": 7.429937929134845e-05,
      "loss": 11.779,
      "step": 13615
    },
    {
      "epoch": 2.3809398907103825,
      "grad_norm": 0.6666154363084171,
      "learning_rate": 7.428569433033781e-05,
      "loss": 12.0377,
      "step": 13616
    },
    {
      "epoch": 2.381114754098361,
      "grad_norm": 0.8077607383597877,
      "learning_rate": 7.4272009884978e-05,
      "loss": 11.7854,
      "step": 13617
    },
    {
      "epoch": 2.3812896174863387,
      "grad_norm": 0.7217217256526378,
      "learning_rate": 7.425832595554337e-05,
      "loss": 11.8169,
      "step": 13618
    },
    {
      "epoch": 2.381464480874317,
      "grad_norm": 0.6708593172562058,
      "learning_rate": 7.424464254230834e-05,
      "loss": 11.7951,
      "step": 13619
    },
    {
      "epoch": 2.381639344262295,
      "grad_norm": 0.7325430861934158,
      "learning_rate": 7.423095964554731e-05,
      "loss": 11.9311,
      "step": 13620
    },
    {
      "epoch": 2.3818142076502733,
      "grad_norm": 0.6455792379299404,
      "learning_rate": 7.421727726553463e-05,
      "loss": 11.9627,
      "step": 13621
    },
    {
      "epoch": 2.381989071038251,
      "grad_norm": 0.7206722533121818,
      "learning_rate": 7.420359540254474e-05,
      "loss": 11.947,
      "step": 13622
    },
    {
      "epoch": 2.3821639344262295,
      "grad_norm": 0.6880388923557861,
      "learning_rate": 7.418991405685197e-05,
      "loss": 11.8582,
      "step": 13623
    },
    {
      "epoch": 2.3823387978142074,
      "grad_norm": 0.620375325926491,
      "learning_rate": 7.417623322873068e-05,
      "loss": 11.8957,
      "step": 13624
    },
    {
      "epoch": 2.3825136612021858,
      "grad_norm": 0.7418927983698842,
      "learning_rate": 7.416255291845517e-05,
      "loss": 11.8815,
      "step": 13625
    },
    {
      "epoch": 2.382688524590164,
      "grad_norm": 0.6325432421084819,
      "learning_rate": 7.414887312629981e-05,
      "loss": 11.954,
      "step": 13626
    },
    {
      "epoch": 2.382863387978142,
      "grad_norm": 0.5854629379570586,
      "learning_rate": 7.413519385253891e-05,
      "loss": 11.725,
      "step": 13627
    },
    {
      "epoch": 2.3830382513661204,
      "grad_norm": 0.6250959099130163,
      "learning_rate": 7.412151509744681e-05,
      "loss": 11.5438,
      "step": 13628
    },
    {
      "epoch": 2.3832131147540982,
      "grad_norm": 0.6966210329493491,
      "learning_rate": 7.41078368612978e-05,
      "loss": 11.7243,
      "step": 13629
    },
    {
      "epoch": 2.3833879781420766,
      "grad_norm": 0.6679431662485953,
      "learning_rate": 7.40941591443661e-05,
      "loss": 12.0298,
      "step": 13630
    },
    {
      "epoch": 2.3835628415300545,
      "grad_norm": 0.6580035257214178,
      "learning_rate": 7.408048194692613e-05,
      "loss": 11.8098,
      "step": 13631
    },
    {
      "epoch": 2.383737704918033,
      "grad_norm": 0.6572216196180996,
      "learning_rate": 7.406680526925204e-05,
      "loss": 11.7576,
      "step": 13632
    },
    {
      "epoch": 2.3839125683060107,
      "grad_norm": 0.6399847120363992,
      "learning_rate": 7.405312911161816e-05,
      "loss": 11.8151,
      "step": 13633
    },
    {
      "epoch": 2.384087431693989,
      "grad_norm": 0.6659872570298517,
      "learning_rate": 7.403945347429868e-05,
      "loss": 11.6569,
      "step": 13634
    },
    {
      "epoch": 2.3842622950819674,
      "grad_norm": 0.6980558263752131,
      "learning_rate": 7.402577835756785e-05,
      "loss": 11.9109,
      "step": 13635
    },
    {
      "epoch": 2.3844371584699453,
      "grad_norm": 0.6984902984424296,
      "learning_rate": 7.401210376169994e-05,
      "loss": 11.7639,
      "step": 13636
    },
    {
      "epoch": 2.3846120218579236,
      "grad_norm": 0.6458380180985772,
      "learning_rate": 7.399842968696913e-05,
      "loss": 12.0007,
      "step": 13637
    },
    {
      "epoch": 2.3847868852459015,
      "grad_norm": 0.6729924932659209,
      "learning_rate": 7.398475613364969e-05,
      "loss": 11.8824,
      "step": 13638
    },
    {
      "epoch": 2.38496174863388,
      "grad_norm": 0.6684551036394603,
      "learning_rate": 7.397108310201571e-05,
      "loss": 11.9918,
      "step": 13639
    },
    {
      "epoch": 2.385136612021858,
      "grad_norm": 0.6300610154635852,
      "learning_rate": 7.395741059234143e-05,
      "loss": 12.0182,
      "step": 13640
    },
    {
      "epoch": 2.385311475409836,
      "grad_norm": 0.6327644720756466,
      "learning_rate": 7.394373860490104e-05,
      "loss": 11.7926,
      "step": 13641
    },
    {
      "epoch": 2.385486338797814,
      "grad_norm": 0.64795502560989,
      "learning_rate": 7.393006713996873e-05,
      "loss": 11.7305,
      "step": 13642
    },
    {
      "epoch": 2.3856612021857924,
      "grad_norm": 0.6326577474972178,
      "learning_rate": 7.391639619781859e-05,
      "loss": 11.8222,
      "step": 13643
    },
    {
      "epoch": 2.3858360655737707,
      "grad_norm": 0.6555947092346753,
      "learning_rate": 7.39027257787248e-05,
      "loss": 11.9475,
      "step": 13644
    },
    {
      "epoch": 2.3860109289617486,
      "grad_norm": 0.661476895574631,
      "learning_rate": 7.388905588296145e-05,
      "loss": 11.9426,
      "step": 13645
    },
    {
      "epoch": 2.386185792349727,
      "grad_norm": 0.6480541745510332,
      "learning_rate": 7.387538651080273e-05,
      "loss": 11.9522,
      "step": 13646
    },
    {
      "epoch": 2.386360655737705,
      "grad_norm": 0.6444387154125245,
      "learning_rate": 7.386171766252274e-05,
      "loss": 12.0883,
      "step": 13647
    },
    {
      "epoch": 2.386535519125683,
      "grad_norm": 0.6560002903235398,
      "learning_rate": 7.384804933839555e-05,
      "loss": 11.9052,
      "step": 13648
    },
    {
      "epoch": 2.386710382513661,
      "grad_norm": 0.649199857939037,
      "learning_rate": 7.383438153869529e-05,
      "loss": 11.8038,
      "step": 13649
    },
    {
      "epoch": 2.3868852459016394,
      "grad_norm": 0.6167745548251126,
      "learning_rate": 7.382071426369597e-05,
      "loss": 11.9267,
      "step": 13650
    },
    {
      "epoch": 2.3870601092896173,
      "grad_norm": 0.6792611176397664,
      "learning_rate": 7.380704751367176e-05,
      "loss": 11.8636,
      "step": 13651
    },
    {
      "epoch": 2.3872349726775957,
      "grad_norm": 0.6462550523961649,
      "learning_rate": 7.379338128889665e-05,
      "loss": 11.7918,
      "step": 13652
    },
    {
      "epoch": 2.387409836065574,
      "grad_norm": 0.6412543319336363,
      "learning_rate": 7.377971558964471e-05,
      "loss": 11.9723,
      "step": 13653
    },
    {
      "epoch": 2.387584699453552,
      "grad_norm": 0.6206670570412774,
      "learning_rate": 7.376605041618995e-05,
      "loss": 11.5579,
      "step": 13654
    },
    {
      "epoch": 2.3877595628415302,
      "grad_norm": 0.6494352354549884,
      "learning_rate": 7.375238576880645e-05,
      "loss": 11.8647,
      "step": 13655
    },
    {
      "epoch": 2.387934426229508,
      "grad_norm": 0.640551917993113,
      "learning_rate": 7.373872164776824e-05,
      "loss": 11.9319,
      "step": 13656
    },
    {
      "epoch": 2.3881092896174865,
      "grad_norm": 0.6490806162450763,
      "learning_rate": 7.372505805334927e-05,
      "loss": 11.8967,
      "step": 13657
    },
    {
      "epoch": 2.3882841530054644,
      "grad_norm": 0.6599340710933466,
      "learning_rate": 7.371139498582356e-05,
      "loss": 12.0396,
      "step": 13658
    },
    {
      "epoch": 2.3884590163934427,
      "grad_norm": 0.6553553134444505,
      "learning_rate": 7.369773244546506e-05,
      "loss": 11.8132,
      "step": 13659
    },
    {
      "epoch": 2.3886338797814206,
      "grad_norm": 0.7031086512560728,
      "learning_rate": 7.368407043254781e-05,
      "loss": 11.7607,
      "step": 13660
    },
    {
      "epoch": 2.388808743169399,
      "grad_norm": 0.7241694180209041,
      "learning_rate": 7.367040894734579e-05,
      "loss": 11.99,
      "step": 13661
    },
    {
      "epoch": 2.3889836065573773,
      "grad_norm": 0.6686797823116513,
      "learning_rate": 7.365674799013287e-05,
      "loss": 11.7942,
      "step": 13662
    },
    {
      "epoch": 2.389158469945355,
      "grad_norm": 0.6957686793069454,
      "learning_rate": 7.364308756118308e-05,
      "loss": 11.7824,
      "step": 13663
    },
    {
      "epoch": 2.389333333333333,
      "grad_norm": 0.6536097023100618,
      "learning_rate": 7.362942766077022e-05,
      "loss": 11.695,
      "step": 13664
    },
    {
      "epoch": 2.3895081967213114,
      "grad_norm": 0.7459617312069008,
      "learning_rate": 7.361576828916839e-05,
      "loss": 11.9944,
      "step": 13665
    },
    {
      "epoch": 2.3896830601092898,
      "grad_norm": 0.6891712228671846,
      "learning_rate": 7.360210944665139e-05,
      "loss": 11.6583,
      "step": 13666
    },
    {
      "epoch": 2.3898579234972677,
      "grad_norm": 0.6972237739085516,
      "learning_rate": 7.358845113349313e-05,
      "loss": 11.7633,
      "step": 13667
    },
    {
      "epoch": 2.390032786885246,
      "grad_norm": 0.7242923696097808,
      "learning_rate": 7.357479334996755e-05,
      "loss": 11.9314,
      "step": 13668
    },
    {
      "epoch": 2.390207650273224,
      "grad_norm": 0.7463474730964441,
      "learning_rate": 7.356113609634845e-05,
      "loss": 11.8082,
      "step": 13669
    },
    {
      "epoch": 2.3903825136612022,
      "grad_norm": 0.6239431565098705,
      "learning_rate": 7.35474793729098e-05,
      "loss": 12.0246,
      "step": 13670
    },
    {
      "epoch": 2.39055737704918,
      "grad_norm": 0.7889241470581302,
      "learning_rate": 7.353382317992539e-05,
      "loss": 11.5494,
      "step": 13671
    },
    {
      "epoch": 2.3907322404371585,
      "grad_norm": 0.7337623738232257,
      "learning_rate": 7.35201675176691e-05,
      "loss": 11.9133,
      "step": 13672
    },
    {
      "epoch": 2.3909071038251364,
      "grad_norm": 0.6539612668517442,
      "learning_rate": 7.350651238641475e-05,
      "loss": 11.9392,
      "step": 13673
    },
    {
      "epoch": 2.3910819672131147,
      "grad_norm": 0.7248532436669823,
      "learning_rate": 7.349285778643614e-05,
      "loss": 11.7026,
      "step": 13674
    },
    {
      "epoch": 2.391256830601093,
      "grad_norm": 0.7999084605907848,
      "learning_rate": 7.347920371800714e-05,
      "loss": 11.9532,
      "step": 13675
    },
    {
      "epoch": 2.391431693989071,
      "grad_norm": 0.6547652306536275,
      "learning_rate": 7.346555018140154e-05,
      "loss": 11.5908,
      "step": 13676
    },
    {
      "epoch": 2.3916065573770493,
      "grad_norm": 0.7378749651326129,
      "learning_rate": 7.345189717689316e-05,
      "loss": 11.8348,
      "step": 13677
    },
    {
      "epoch": 2.391781420765027,
      "grad_norm": 0.6597106402694886,
      "learning_rate": 7.343824470475567e-05,
      "loss": 11.8074,
      "step": 13678
    },
    {
      "epoch": 2.3919562841530055,
      "grad_norm": 0.6621775389583617,
      "learning_rate": 7.342459276526302e-05,
      "loss": 11.8752,
      "step": 13679
    },
    {
      "epoch": 2.3921311475409834,
      "grad_norm": 0.6818918910123047,
      "learning_rate": 7.341094135868882e-05,
      "loss": 11.8217,
      "step": 13680
    },
    {
      "epoch": 2.392306010928962,
      "grad_norm": 0.6981197515106092,
      "learning_rate": 7.339729048530695e-05,
      "loss": 11.9065,
      "step": 13681
    },
    {
      "epoch": 2.3924808743169397,
      "grad_norm": 0.710918341076492,
      "learning_rate": 7.338364014539103e-05,
      "loss": 12.2094,
      "step": 13682
    },
    {
      "epoch": 2.392655737704918,
      "grad_norm": 0.6323788469826114,
      "learning_rate": 7.336999033921486e-05,
      "loss": 11.6614,
      "step": 13683
    },
    {
      "epoch": 2.3928306010928964,
      "grad_norm": 0.7394055260032788,
      "learning_rate": 7.335634106705213e-05,
      "loss": 11.7882,
      "step": 13684
    },
    {
      "epoch": 2.3930054644808743,
      "grad_norm": 0.5933068511438638,
      "learning_rate": 7.33426923291766e-05,
      "loss": 12.0848,
      "step": 13685
    },
    {
      "epoch": 2.3931803278688526,
      "grad_norm": 0.6949620671558296,
      "learning_rate": 7.332904412586194e-05,
      "loss": 11.8241,
      "step": 13686
    },
    {
      "epoch": 2.3933551912568305,
      "grad_norm": 0.755276453789919,
      "learning_rate": 7.331539645738183e-05,
      "loss": 11.9732,
      "step": 13687
    },
    {
      "epoch": 2.393530054644809,
      "grad_norm": 0.6362956164789307,
      "learning_rate": 7.330174932400993e-05,
      "loss": 11.8918,
      "step": 13688
    },
    {
      "epoch": 2.3937049180327867,
      "grad_norm": 0.6185212944392698,
      "learning_rate": 7.328810272601996e-05,
      "loss": 11.9792,
      "step": 13689
    },
    {
      "epoch": 2.393879781420765,
      "grad_norm": 0.6969380517876206,
      "learning_rate": 7.327445666368555e-05,
      "loss": 12.0339,
      "step": 13690
    },
    {
      "epoch": 2.394054644808743,
      "grad_norm": 0.6572682465019518,
      "learning_rate": 7.326081113728032e-05,
      "loss": 11.7845,
      "step": 13691
    },
    {
      "epoch": 2.3942295081967213,
      "grad_norm": 0.6822296293989933,
      "learning_rate": 7.324716614707793e-05,
      "loss": 11.8941,
      "step": 13692
    },
    {
      "epoch": 2.3944043715846997,
      "grad_norm": 0.6841901090513184,
      "learning_rate": 7.323352169335199e-05,
      "loss": 11.8395,
      "step": 13693
    },
    {
      "epoch": 2.3945792349726776,
      "grad_norm": 0.6860793971864567,
      "learning_rate": 7.321987777637614e-05,
      "loss": 11.7318,
      "step": 13694
    },
    {
      "epoch": 2.394754098360656,
      "grad_norm": 0.6973582282125528,
      "learning_rate": 7.320623439642399e-05,
      "loss": 11.8406,
      "step": 13695
    },
    {
      "epoch": 2.394928961748634,
      "grad_norm": 0.6739026907619927,
      "learning_rate": 7.319259155376908e-05,
      "loss": 11.78,
      "step": 13696
    },
    {
      "epoch": 2.395103825136612,
      "grad_norm": 0.7837170656243104,
      "learning_rate": 7.317894924868502e-05,
      "loss": 11.9446,
      "step": 13697
    },
    {
      "epoch": 2.39527868852459,
      "grad_norm": 0.683508185840493,
      "learning_rate": 7.316530748144537e-05,
      "loss": 11.8722,
      "step": 13698
    },
    {
      "epoch": 2.3954535519125684,
      "grad_norm": 0.6280200096449423,
      "learning_rate": 7.315166625232373e-05,
      "loss": 11.9591,
      "step": 13699
    },
    {
      "epoch": 2.3956284153005463,
      "grad_norm": 0.6715379446802738,
      "learning_rate": 7.313802556159359e-05,
      "loss": 11.8066,
      "step": 13700
    },
    {
      "epoch": 2.3958032786885246,
      "grad_norm": 0.6777174460057234,
      "learning_rate": 7.312438540952852e-05,
      "loss": 12.0492,
      "step": 13701
    },
    {
      "epoch": 2.395978142076503,
      "grad_norm": 0.6754699955553967,
      "learning_rate": 7.311074579640202e-05,
      "loss": 11.8718,
      "step": 13702
    },
    {
      "epoch": 2.396153005464481,
      "grad_norm": 0.7257830346697257,
      "learning_rate": 7.309710672248765e-05,
      "loss": 11.4765,
      "step": 13703
    },
    {
      "epoch": 2.396327868852459,
      "grad_norm": 0.7683451170984809,
      "learning_rate": 7.308346818805891e-05,
      "loss": 11.8563,
      "step": 13704
    },
    {
      "epoch": 2.396502732240437,
      "grad_norm": 0.6431939729193343,
      "learning_rate": 7.306983019338927e-05,
      "loss": 12.0174,
      "step": 13705
    },
    {
      "epoch": 2.3966775956284154,
      "grad_norm": 0.6147644861580788,
      "learning_rate": 7.30561927387522e-05,
      "loss": 11.8056,
      "step": 13706
    },
    {
      "epoch": 2.3968524590163933,
      "grad_norm": 0.7527930369704111,
      "learning_rate": 7.30425558244212e-05,
      "loss": 11.7023,
      "step": 13707
    },
    {
      "epoch": 2.3970273224043717,
      "grad_norm": 0.731810207423438,
      "learning_rate": 7.302891945066974e-05,
      "loss": 11.9826,
      "step": 13708
    },
    {
      "epoch": 2.3972021857923496,
      "grad_norm": 0.6769036500277402,
      "learning_rate": 7.301528361777125e-05,
      "loss": 11.67,
      "step": 13709
    },
    {
      "epoch": 2.397377049180328,
      "grad_norm": 0.7441619026120972,
      "learning_rate": 7.300164832599917e-05,
      "loss": 11.5625,
      "step": 13710
    },
    {
      "epoch": 2.3975519125683062,
      "grad_norm": 0.6519237823437275,
      "learning_rate": 7.298801357562696e-05,
      "loss": 11.8507,
      "step": 13711
    },
    {
      "epoch": 2.397726775956284,
      "grad_norm": 0.7288300936931733,
      "learning_rate": 7.297437936692796e-05,
      "loss": 12.0324,
      "step": 13712
    },
    {
      "epoch": 2.3979016393442625,
      "grad_norm": 0.6932855514576498,
      "learning_rate": 7.29607457001757e-05,
      "loss": 11.9163,
      "step": 13713
    },
    {
      "epoch": 2.3980765027322404,
      "grad_norm": 0.6136387351551829,
      "learning_rate": 7.294711257564349e-05,
      "loss": 11.7589,
      "step": 13714
    },
    {
      "epoch": 2.3982513661202187,
      "grad_norm": 0.6618407906831236,
      "learning_rate": 7.293347999360472e-05,
      "loss": 11.8883,
      "step": 13715
    },
    {
      "epoch": 2.3984262295081966,
      "grad_norm": 0.7646943438065624,
      "learning_rate": 7.291984795433279e-05,
      "loss": 12.093,
      "step": 13716
    },
    {
      "epoch": 2.398601092896175,
      "grad_norm": 0.6497944650670727,
      "learning_rate": 7.290621645810105e-05,
      "loss": 12.0111,
      "step": 13717
    },
    {
      "epoch": 2.398775956284153,
      "grad_norm": 0.7264882239997268,
      "learning_rate": 7.289258550518286e-05,
      "loss": 11.763,
      "step": 13718
    },
    {
      "epoch": 2.398950819672131,
      "grad_norm": 0.7085209824805742,
      "learning_rate": 7.287895509585156e-05,
      "loss": 11.7992,
      "step": 13719
    },
    {
      "epoch": 2.3991256830601095,
      "grad_norm": 0.7089297064841262,
      "learning_rate": 7.286532523038052e-05,
      "loss": 11.8572,
      "step": 13720
    },
    {
      "epoch": 2.3993005464480874,
      "grad_norm": 0.7097602500219757,
      "learning_rate": 7.2851695909043e-05,
      "loss": 11.6741,
      "step": 13721
    },
    {
      "epoch": 2.3994754098360653,
      "grad_norm": 0.6573847696777411,
      "learning_rate": 7.283806713211228e-05,
      "loss": 11.5205,
      "step": 13722
    },
    {
      "epoch": 2.3996502732240437,
      "grad_norm": 0.6881190294626651,
      "learning_rate": 7.282443889986177e-05,
      "loss": 11.9287,
      "step": 13723
    },
    {
      "epoch": 2.399825136612022,
      "grad_norm": 0.638909586245582,
      "learning_rate": 7.281081121256468e-05,
      "loss": 11.8115,
      "step": 13724
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.6387806140724237,
      "learning_rate": 7.279718407049434e-05,
      "loss": 11.7442,
      "step": 13725
    },
    {
      "epoch": 2.4001748633879783,
      "grad_norm": 0.6986971862152849,
      "learning_rate": 7.278355747392394e-05,
      "loss": 11.7844,
      "step": 13726
    },
    {
      "epoch": 2.400349726775956,
      "grad_norm": 0.590691574456923,
      "learning_rate": 7.276993142312679e-05,
      "loss": 11.9683,
      "step": 13727
    },
    {
      "epoch": 2.4005245901639345,
      "grad_norm": 0.5663081406454011,
      "learning_rate": 7.275630591837613e-05,
      "loss": 12.0142,
      "step": 13728
    },
    {
      "epoch": 2.4006994535519124,
      "grad_norm": 0.6261945530671245,
      "learning_rate": 7.27426809599452e-05,
      "loss": 11.9129,
      "step": 13729
    },
    {
      "epoch": 2.4008743169398907,
      "grad_norm": 0.640383322211533,
      "learning_rate": 7.272905654810718e-05,
      "loss": 11.6979,
      "step": 13730
    },
    {
      "epoch": 2.4010491803278686,
      "grad_norm": 0.6914981409859678,
      "learning_rate": 7.27154326831353e-05,
      "loss": 12.0087,
      "step": 13731
    },
    {
      "epoch": 2.401224043715847,
      "grad_norm": 0.7111407056235172,
      "learning_rate": 7.27018093653028e-05,
      "loss": 12.0701,
      "step": 13732
    },
    {
      "epoch": 2.4013989071038253,
      "grad_norm": 0.6627034050198598,
      "learning_rate": 7.268818659488281e-05,
      "loss": 12.0932,
      "step": 13733
    },
    {
      "epoch": 2.401573770491803,
      "grad_norm": 0.6815820844219332,
      "learning_rate": 7.267456437214859e-05,
      "loss": 11.8288,
      "step": 13734
    },
    {
      "epoch": 2.4017486338797815,
      "grad_norm": 0.7614606229375755,
      "learning_rate": 7.266094269737322e-05,
      "loss": 11.5679,
      "step": 13735
    },
    {
      "epoch": 2.4019234972677594,
      "grad_norm": 0.6916665324572345,
      "learning_rate": 7.264732157082986e-05,
      "loss": 11.9552,
      "step": 13736
    },
    {
      "epoch": 2.402098360655738,
      "grad_norm": 0.7551137572953208,
      "learning_rate": 7.263370099279172e-05,
      "loss": 11.662,
      "step": 13737
    },
    {
      "epoch": 2.4022732240437157,
      "grad_norm": 0.6798868796035059,
      "learning_rate": 7.26200809635319e-05,
      "loss": 11.6356,
      "step": 13738
    },
    {
      "epoch": 2.402448087431694,
      "grad_norm": 0.6626338696003364,
      "learning_rate": 7.260646148332354e-05,
      "loss": 12.0393,
      "step": 13739
    },
    {
      "epoch": 2.402622950819672,
      "grad_norm": 0.7168731286990596,
      "learning_rate": 7.259284255243971e-05,
      "loss": 11.5246,
      "step": 13740
    },
    {
      "epoch": 2.4027978142076503,
      "grad_norm": 0.6741944080838825,
      "learning_rate": 7.257922417115352e-05,
      "loss": 11.9668,
      "step": 13741
    },
    {
      "epoch": 2.4029726775956286,
      "grad_norm": 0.7041617472933239,
      "learning_rate": 7.25656063397381e-05,
      "loss": 11.8396,
      "step": 13742
    },
    {
      "epoch": 2.4031475409836065,
      "grad_norm": 0.6466968493811732,
      "learning_rate": 7.255198905846655e-05,
      "loss": 11.8661,
      "step": 13743
    },
    {
      "epoch": 2.403322404371585,
      "grad_norm": 0.7159438940778415,
      "learning_rate": 7.253837232761185e-05,
      "loss": 11.7802,
      "step": 13744
    },
    {
      "epoch": 2.4034972677595627,
      "grad_norm": 0.646800978612612,
      "learning_rate": 7.252475614744709e-05,
      "loss": 11.8518,
      "step": 13745
    },
    {
      "epoch": 2.403672131147541,
      "grad_norm": 0.7051465885125967,
      "learning_rate": 7.251114051824534e-05,
      "loss": 11.7303,
      "step": 13746
    },
    {
      "epoch": 2.403846994535519,
      "grad_norm": 0.6859138150034056,
      "learning_rate": 7.249752544027966e-05,
      "loss": 11.7476,
      "step": 13747
    },
    {
      "epoch": 2.4040218579234973,
      "grad_norm": 0.6200338753849257,
      "learning_rate": 7.2483910913823e-05,
      "loss": 12.0219,
      "step": 13748
    },
    {
      "epoch": 2.404196721311475,
      "grad_norm": 0.6988689095213393,
      "learning_rate": 7.247029693914842e-05,
      "loss": 12.0605,
      "step": 13749
    },
    {
      "epoch": 2.4043715846994536,
      "grad_norm": 0.7144401092755458,
      "learning_rate": 7.24566835165289e-05,
      "loss": 11.9857,
      "step": 13750
    },
    {
      "epoch": 2.404546448087432,
      "grad_norm": 0.6609094774798605,
      "learning_rate": 7.244307064623746e-05,
      "loss": 11.682,
      "step": 13751
    },
    {
      "epoch": 2.40472131147541,
      "grad_norm": 0.6709764704137691,
      "learning_rate": 7.242945832854708e-05,
      "loss": 11.7644,
      "step": 13752
    },
    {
      "epoch": 2.404896174863388,
      "grad_norm": 0.6085367343094882,
      "learning_rate": 7.24158465637307e-05,
      "loss": 11.9357,
      "step": 13753
    },
    {
      "epoch": 2.405071038251366,
      "grad_norm": 0.656598175709133,
      "learning_rate": 7.24022353520613e-05,
      "loss": 11.8628,
      "step": 13754
    },
    {
      "epoch": 2.4052459016393444,
      "grad_norm": 0.7262005314629446,
      "learning_rate": 7.238862469381177e-05,
      "loss": 11.9175,
      "step": 13755
    },
    {
      "epoch": 2.4054207650273223,
      "grad_norm": 0.6957404964499035,
      "learning_rate": 7.237501458925518e-05,
      "loss": 12.1451,
      "step": 13756
    },
    {
      "epoch": 2.4055956284153006,
      "grad_norm": 0.6356094305840456,
      "learning_rate": 7.23614050386643e-05,
      "loss": 11.8695,
      "step": 13757
    },
    {
      "epoch": 2.4057704918032785,
      "grad_norm": 0.6425931654670114,
      "learning_rate": 7.234779604231214e-05,
      "loss": 12.06,
      "step": 13758
    },
    {
      "epoch": 2.405945355191257,
      "grad_norm": 0.633704203552587,
      "learning_rate": 7.233418760047156e-05,
      "loss": 11.7263,
      "step": 13759
    },
    {
      "epoch": 2.406120218579235,
      "grad_norm": 0.6897549355376489,
      "learning_rate": 7.232057971341544e-05,
      "loss": 11.8914,
      "step": 13760
    },
    {
      "epoch": 2.406295081967213,
      "grad_norm": 0.6043364457944199,
      "learning_rate": 7.230697238141673e-05,
      "loss": 11.9125,
      "step": 13761
    },
    {
      "epoch": 2.4064699453551914,
      "grad_norm": 0.7056497599657404,
      "learning_rate": 7.229336560474822e-05,
      "loss": 11.6523,
      "step": 13762
    },
    {
      "epoch": 2.4066448087431693,
      "grad_norm": 0.5891634690658656,
      "learning_rate": 7.227975938368282e-05,
      "loss": 12.0068,
      "step": 13763
    },
    {
      "epoch": 2.4068196721311477,
      "grad_norm": 0.6842969823388337,
      "learning_rate": 7.226615371849337e-05,
      "loss": 11.773,
      "step": 13764
    },
    {
      "epoch": 2.4069945355191256,
      "grad_norm": 0.6553326894818352,
      "learning_rate": 7.225254860945265e-05,
      "loss": 11.846,
      "step": 13765
    },
    {
      "epoch": 2.407169398907104,
      "grad_norm": 0.7107228546732501,
      "learning_rate": 7.223894405683354e-05,
      "loss": 11.9405,
      "step": 13766
    },
    {
      "epoch": 2.407344262295082,
      "grad_norm": 0.61998249950626,
      "learning_rate": 7.222534006090886e-05,
      "loss": 11.566,
      "step": 13767
    },
    {
      "epoch": 2.40751912568306,
      "grad_norm": 0.7385800466577359,
      "learning_rate": 7.221173662195139e-05,
      "loss": 11.7583,
      "step": 13768
    },
    {
      "epoch": 2.4076939890710385,
      "grad_norm": 0.6041608146975416,
      "learning_rate": 7.219813374023391e-05,
      "loss": 11.9042,
      "step": 13769
    },
    {
      "epoch": 2.4078688524590164,
      "grad_norm": 0.6548458304314156,
      "learning_rate": 7.21845314160292e-05,
      "loss": 12.0024,
      "step": 13770
    },
    {
      "epoch": 2.4080437158469947,
      "grad_norm": 0.6155181769208339,
      "learning_rate": 7.217092964961003e-05,
      "loss": 12.0348,
      "step": 13771
    },
    {
      "epoch": 2.4082185792349726,
      "grad_norm": 0.6808103434368257,
      "learning_rate": 7.21573284412492e-05,
      "loss": 11.9429,
      "step": 13772
    },
    {
      "epoch": 2.408393442622951,
      "grad_norm": 0.8122078960605423,
      "learning_rate": 7.214372779121942e-05,
      "loss": 11.9449,
      "step": 13773
    },
    {
      "epoch": 2.408568306010929,
      "grad_norm": 0.621443321942328,
      "learning_rate": 7.21301276997934e-05,
      "loss": 11.7398,
      "step": 13774
    },
    {
      "epoch": 2.408743169398907,
      "grad_norm": 0.7208169113217336,
      "learning_rate": 7.211652816724391e-05,
      "loss": 11.5272,
      "step": 13775
    },
    {
      "epoch": 2.408918032786885,
      "grad_norm": 0.7256618589556266,
      "learning_rate": 7.210292919384365e-05,
      "loss": 11.7767,
      "step": 13776
    },
    {
      "epoch": 2.4090928961748634,
      "grad_norm": 0.6523360935260157,
      "learning_rate": 7.208933077986532e-05,
      "loss": 11.7948,
      "step": 13777
    },
    {
      "epoch": 2.4092677595628413,
      "grad_norm": 0.9155680428496953,
      "learning_rate": 7.20757329255816e-05,
      "loss": 11.9097,
      "step": 13778
    },
    {
      "epoch": 2.4094426229508197,
      "grad_norm": 0.6798424802557315,
      "learning_rate": 7.206213563126513e-05,
      "loss": 12.1742,
      "step": 13779
    },
    {
      "epoch": 2.4096174863387976,
      "grad_norm": 0.6213829564316797,
      "learning_rate": 7.204853889718867e-05,
      "loss": 11.9113,
      "step": 13780
    },
    {
      "epoch": 2.409792349726776,
      "grad_norm": 0.7010489803902046,
      "learning_rate": 7.203494272362481e-05,
      "loss": 11.7483,
      "step": 13781
    },
    {
      "epoch": 2.4099672131147543,
      "grad_norm": 0.7746840690063023,
      "learning_rate": 7.202134711084624e-05,
      "loss": 11.8723,
      "step": 13782
    },
    {
      "epoch": 2.410142076502732,
      "grad_norm": 0.6408324161308381,
      "learning_rate": 7.200775205912555e-05,
      "loss": 12.1008,
      "step": 13783
    },
    {
      "epoch": 2.4103169398907105,
      "grad_norm": 0.7542169796090676,
      "learning_rate": 7.199415756873534e-05,
      "loss": 11.8574,
      "step": 13784
    },
    {
      "epoch": 2.4104918032786884,
      "grad_norm": 0.7138404792964581,
      "learning_rate": 7.19805636399483e-05,
      "loss": 12.0922,
      "step": 13785
    },
    {
      "epoch": 2.4106666666666667,
      "grad_norm": 0.6371119113199482,
      "learning_rate": 7.1966970273037e-05,
      "loss": 12.0416,
      "step": 13786
    },
    {
      "epoch": 2.4108415300546446,
      "grad_norm": 0.7184358527656826,
      "learning_rate": 7.1953377468274e-05,
      "loss": 11.8981,
      "step": 13787
    },
    {
      "epoch": 2.411016393442623,
      "grad_norm": 0.677948386207093,
      "learning_rate": 7.19397852259319e-05,
      "loss": 11.6688,
      "step": 13788
    },
    {
      "epoch": 2.411191256830601,
      "grad_norm": 0.6965230044669958,
      "learning_rate": 7.192619354628322e-05,
      "loss": 12.1899,
      "step": 13789
    },
    {
      "epoch": 2.411366120218579,
      "grad_norm": 0.6793768409600466,
      "learning_rate": 7.19126024296006e-05,
      "loss": 12.067,
      "step": 13790
    },
    {
      "epoch": 2.4115409836065576,
      "grad_norm": 0.7038792659926274,
      "learning_rate": 7.189901187615658e-05,
      "loss": 11.934,
      "step": 13791
    },
    {
      "epoch": 2.4117158469945355,
      "grad_norm": 0.7167668992037486,
      "learning_rate": 7.188542188622361e-05,
      "loss": 11.8409,
      "step": 13792
    },
    {
      "epoch": 2.411890710382514,
      "grad_norm": 0.6745912543871182,
      "learning_rate": 7.187183246007425e-05,
      "loss": 11.8647,
      "step": 13793
    },
    {
      "epoch": 2.4120655737704917,
      "grad_norm": 0.6297206494334233,
      "learning_rate": 7.1858243597981e-05,
      "loss": 11.6782,
      "step": 13794
    },
    {
      "epoch": 2.41224043715847,
      "grad_norm": 0.6513960026348449,
      "learning_rate": 7.18446553002164e-05,
      "loss": 11.7095,
      "step": 13795
    },
    {
      "epoch": 2.412415300546448,
      "grad_norm": 0.6666464030295275,
      "learning_rate": 7.183106756705292e-05,
      "loss": 11.7833,
      "step": 13796
    },
    {
      "epoch": 2.4125901639344263,
      "grad_norm": 0.7374334470061936,
      "learning_rate": 7.1817480398763e-05,
      "loss": 11.9292,
      "step": 13797
    },
    {
      "epoch": 2.412765027322404,
      "grad_norm": 0.7283013651956716,
      "learning_rate": 7.18038937956191e-05,
      "loss": 12.1242,
      "step": 13798
    },
    {
      "epoch": 2.4129398907103825,
      "grad_norm": 0.6483772281883743,
      "learning_rate": 7.179030775789374e-05,
      "loss": 11.7704,
      "step": 13799
    },
    {
      "epoch": 2.413114754098361,
      "grad_norm": 0.6937103103942595,
      "learning_rate": 7.177672228585935e-05,
      "loss": 11.9482,
      "step": 13800
    },
    {
      "epoch": 2.4132896174863387,
      "grad_norm": 0.6639806323619353,
      "learning_rate": 7.17631373797883e-05,
      "loss": 11.9814,
      "step": 13801
    },
    {
      "epoch": 2.413464480874317,
      "grad_norm": 0.8004099632140957,
      "learning_rate": 7.174955303995306e-05,
      "loss": 12.2162,
      "step": 13802
    },
    {
      "epoch": 2.413639344262295,
      "grad_norm": 0.6643189383194376,
      "learning_rate": 7.173596926662599e-05,
      "loss": 12.0781,
      "step": 13803
    },
    {
      "epoch": 2.4138142076502733,
      "grad_norm": 0.6229577073939757,
      "learning_rate": 7.172238606007958e-05,
      "loss": 11.9462,
      "step": 13804
    },
    {
      "epoch": 2.413989071038251,
      "grad_norm": 0.620124572235537,
      "learning_rate": 7.170880342058611e-05,
      "loss": 11.8727,
      "step": 13805
    },
    {
      "epoch": 2.4141639344262296,
      "grad_norm": 0.7044061090503421,
      "learning_rate": 7.1695221348418e-05,
      "loss": 11.8072,
      "step": 13806
    },
    {
      "epoch": 2.4143387978142075,
      "grad_norm": 0.6763583092307243,
      "learning_rate": 7.168163984384763e-05,
      "loss": 11.7846,
      "step": 13807
    },
    {
      "epoch": 2.414513661202186,
      "grad_norm": 0.6961689169139476,
      "learning_rate": 7.166805890714726e-05,
      "loss": 11.5466,
      "step": 13808
    },
    {
      "epoch": 2.414688524590164,
      "grad_norm": 0.701291700097299,
      "learning_rate": 7.165447853858937e-05,
      "loss": 11.8634,
      "step": 13809
    },
    {
      "epoch": 2.414863387978142,
      "grad_norm": 0.7396171322814974,
      "learning_rate": 7.16408987384462e-05,
      "loss": 11.7572,
      "step": 13810
    },
    {
      "epoch": 2.4150382513661204,
      "grad_norm": 0.7300383901033947,
      "learning_rate": 7.162731950699009e-05,
      "loss": 11.8873,
      "step": 13811
    },
    {
      "epoch": 2.4152131147540983,
      "grad_norm": 0.6527246157163431,
      "learning_rate": 7.161374084449334e-05,
      "loss": 11.5944,
      "step": 13812
    },
    {
      "epoch": 2.4153879781420766,
      "grad_norm": 0.7393946006640285,
      "learning_rate": 7.160016275122823e-05,
      "loss": 11.926,
      "step": 13813
    },
    {
      "epoch": 2.4155628415300545,
      "grad_norm": 0.7286687227707687,
      "learning_rate": 7.158658522746706e-05,
      "loss": 11.7274,
      "step": 13814
    },
    {
      "epoch": 2.415737704918033,
      "grad_norm": 0.6559185884497131,
      "learning_rate": 7.157300827348209e-05,
      "loss": 11.8902,
      "step": 13815
    },
    {
      "epoch": 2.4159125683060108,
      "grad_norm": 0.6679151816237435,
      "learning_rate": 7.155943188954561e-05,
      "loss": 11.9203,
      "step": 13816
    },
    {
      "epoch": 2.416087431693989,
      "grad_norm": 0.6641255158183522,
      "learning_rate": 7.154585607592982e-05,
      "loss": 11.9735,
      "step": 13817
    },
    {
      "epoch": 2.4162622950819674,
      "grad_norm": 0.7223552229764383,
      "learning_rate": 7.153228083290698e-05,
      "loss": 11.9201,
      "step": 13818
    },
    {
      "epoch": 2.4164371584699453,
      "grad_norm": 0.6442382412916995,
      "learning_rate": 7.151870616074934e-05,
      "loss": 11.5212,
      "step": 13819
    },
    {
      "epoch": 2.4166120218579237,
      "grad_norm": 0.6406801902119916,
      "learning_rate": 7.150513205972907e-05,
      "loss": 11.681,
      "step": 13820
    },
    {
      "epoch": 2.4167868852459016,
      "grad_norm": 0.6771442815846114,
      "learning_rate": 7.149155853011842e-05,
      "loss": 12.1099,
      "step": 13821
    },
    {
      "epoch": 2.41696174863388,
      "grad_norm": 0.7224625912730809,
      "learning_rate": 7.147798557218953e-05,
      "loss": 11.7761,
      "step": 13822
    },
    {
      "epoch": 2.417136612021858,
      "grad_norm": 0.6365526569865547,
      "learning_rate": 7.146441318621459e-05,
      "loss": 11.8089,
      "step": 13823
    },
    {
      "epoch": 2.417311475409836,
      "grad_norm": 0.6314761946762879,
      "learning_rate": 7.14508413724658e-05,
      "loss": 12.1469,
      "step": 13824
    },
    {
      "epoch": 2.417486338797814,
      "grad_norm": 0.7840173534246361,
      "learning_rate": 7.143727013121532e-05,
      "loss": 11.759,
      "step": 13825
    },
    {
      "epoch": 2.4176612021857924,
      "grad_norm": 0.6195704472959735,
      "learning_rate": 7.142369946273525e-05,
      "loss": 11.8925,
      "step": 13826
    },
    {
      "epoch": 2.4178360655737707,
      "grad_norm": 0.6798113806280719,
      "learning_rate": 7.141012936729771e-05,
      "loss": 11.804,
      "step": 13827
    },
    {
      "epoch": 2.4180109289617486,
      "grad_norm": 0.669567396466901,
      "learning_rate": 7.13965598451749e-05,
      "loss": 11.6554,
      "step": 13828
    },
    {
      "epoch": 2.418185792349727,
      "grad_norm": 0.6411612512570662,
      "learning_rate": 7.138299089663889e-05,
      "loss": 11.985,
      "step": 13829
    },
    {
      "epoch": 2.418360655737705,
      "grad_norm": 0.6275252692875266,
      "learning_rate": 7.13694225219618e-05,
      "loss": 11.9512,
      "step": 13830
    },
    {
      "epoch": 2.418535519125683,
      "grad_norm": 0.6629376033948499,
      "learning_rate": 7.135585472141566e-05,
      "loss": 11.7232,
      "step": 13831
    },
    {
      "epoch": 2.418710382513661,
      "grad_norm": 0.6699213978834248,
      "learning_rate": 7.134228749527258e-05,
      "loss": 11.9287,
      "step": 13832
    },
    {
      "epoch": 2.4188852459016394,
      "grad_norm": 0.6084542345686498,
      "learning_rate": 7.132872084380463e-05,
      "loss": 11.9666,
      "step": 13833
    },
    {
      "epoch": 2.4190601092896173,
      "grad_norm": 0.6580093289141203,
      "learning_rate": 7.131515476728389e-05,
      "loss": 11.2253,
      "step": 13834
    },
    {
      "epoch": 2.4192349726775957,
      "grad_norm": 0.6470087256670236,
      "learning_rate": 7.130158926598234e-05,
      "loss": 11.9946,
      "step": 13835
    },
    {
      "epoch": 2.4194098360655736,
      "grad_norm": 0.6638152327999282,
      "learning_rate": 7.128802434017205e-05,
      "loss": 11.9121,
      "step": 13836
    },
    {
      "epoch": 2.419584699453552,
      "grad_norm": 0.6377273860422477,
      "learning_rate": 7.1274459990125e-05,
      "loss": 11.6337,
      "step": 13837
    },
    {
      "epoch": 2.41975956284153,
      "grad_norm": 0.6659380448416065,
      "learning_rate": 7.126089621611326e-05,
      "loss": 11.9452,
      "step": 13838
    },
    {
      "epoch": 2.419934426229508,
      "grad_norm": 0.6841214629703094,
      "learning_rate": 7.12473330184088e-05,
      "loss": 12.1265,
      "step": 13839
    },
    {
      "epoch": 2.4201092896174865,
      "grad_norm": 0.740589686140627,
      "learning_rate": 7.123377039728358e-05,
      "loss": 11.9084,
      "step": 13840
    },
    {
      "epoch": 2.4202841530054644,
      "grad_norm": 0.6953421646212381,
      "learning_rate": 7.12202083530096e-05,
      "loss": 11.8585,
      "step": 13841
    },
    {
      "epoch": 2.4204590163934427,
      "grad_norm": 0.7328552517166012,
      "learning_rate": 7.120664688585876e-05,
      "loss": 12.146,
      "step": 13842
    },
    {
      "epoch": 2.4206338797814206,
      "grad_norm": 0.6339544893711644,
      "learning_rate": 7.119308599610312e-05,
      "loss": 12.0267,
      "step": 13843
    },
    {
      "epoch": 2.420808743169399,
      "grad_norm": 0.6839767582948892,
      "learning_rate": 7.117952568401454e-05,
      "loss": 11.9259,
      "step": 13844
    },
    {
      "epoch": 2.420983606557377,
      "grad_norm": 0.7345130150223225,
      "learning_rate": 7.116596594986494e-05,
      "loss": 11.9454,
      "step": 13845
    },
    {
      "epoch": 2.421158469945355,
      "grad_norm": 0.7103258345736841,
      "learning_rate": 7.115240679392623e-05,
      "loss": 11.6462,
      "step": 13846
    },
    {
      "epoch": 2.421333333333333,
      "grad_norm": 0.7068600743316881,
      "learning_rate": 7.113884821647039e-05,
      "loss": 11.7577,
      "step": 13847
    },
    {
      "epoch": 2.4215081967213115,
      "grad_norm": 0.6215265743801123,
      "learning_rate": 7.112529021776926e-05,
      "loss": 11.9044,
      "step": 13848
    },
    {
      "epoch": 2.42168306010929,
      "grad_norm": 0.6523834042989965,
      "learning_rate": 7.111173279809468e-05,
      "loss": 11.6691,
      "step": 13849
    },
    {
      "epoch": 2.4218579234972677,
      "grad_norm": 0.7114484109880213,
      "learning_rate": 7.109817595771859e-05,
      "loss": 12.0981,
      "step": 13850
    },
    {
      "epoch": 2.422032786885246,
      "grad_norm": 0.6846179323603435,
      "learning_rate": 7.108461969691278e-05,
      "loss": 11.777,
      "step": 13851
    },
    {
      "epoch": 2.422207650273224,
      "grad_norm": 0.6498925146525576,
      "learning_rate": 7.107106401594916e-05,
      "loss": 11.9614,
      "step": 13852
    },
    {
      "epoch": 2.4223825136612023,
      "grad_norm": 0.8029746751106414,
      "learning_rate": 7.10575089150995e-05,
      "loss": 11.922,
      "step": 13853
    },
    {
      "epoch": 2.42255737704918,
      "grad_norm": 0.7082908794463473,
      "learning_rate": 7.104395439463567e-05,
      "loss": 12.1104,
      "step": 13854
    },
    {
      "epoch": 2.4227322404371585,
      "grad_norm": 0.7139706485839399,
      "learning_rate": 7.103040045482947e-05,
      "loss": 11.9515,
      "step": 13855
    },
    {
      "epoch": 2.4229071038251364,
      "grad_norm": 0.769156805690202,
      "learning_rate": 7.101684709595264e-05,
      "loss": 11.7675,
      "step": 13856
    },
    {
      "epoch": 2.4230819672131148,
      "grad_norm": 0.6831290057166665,
      "learning_rate": 7.100329431827706e-05,
      "loss": 11.7688,
      "step": 13857
    },
    {
      "epoch": 2.423256830601093,
      "grad_norm": 0.7900108705536222,
      "learning_rate": 7.098974212207445e-05,
      "loss": 11.7975,
      "step": 13858
    },
    {
      "epoch": 2.423431693989071,
      "grad_norm": 0.7397178986953136,
      "learning_rate": 7.097619050761659e-05,
      "loss": 11.8045,
      "step": 13859
    },
    {
      "epoch": 2.4236065573770493,
      "grad_norm": 0.6356612084431215,
      "learning_rate": 7.096263947517523e-05,
      "loss": 12.0232,
      "step": 13860
    },
    {
      "epoch": 2.4237814207650272,
      "grad_norm": 0.6723482141874731,
      "learning_rate": 7.094908902502208e-05,
      "loss": 11.6531,
      "step": 13861
    },
    {
      "epoch": 2.4239562841530056,
      "grad_norm": 0.6297501017257054,
      "learning_rate": 7.09355391574289e-05,
      "loss": 11.7819,
      "step": 13862
    },
    {
      "epoch": 2.4241311475409835,
      "grad_norm": 0.7030586872789458,
      "learning_rate": 7.092198987266742e-05,
      "loss": 11.8574,
      "step": 13863
    },
    {
      "epoch": 2.424306010928962,
      "grad_norm": 0.6906870247721346,
      "learning_rate": 7.090844117100933e-05,
      "loss": 11.7349,
      "step": 13864
    },
    {
      "epoch": 2.4244808743169397,
      "grad_norm": 0.6608676350917054,
      "learning_rate": 7.08948930527263e-05,
      "loss": 11.8615,
      "step": 13865
    },
    {
      "epoch": 2.424655737704918,
      "grad_norm": 0.649332567730964,
      "learning_rate": 7.088134551809e-05,
      "loss": 11.7888,
      "step": 13866
    },
    {
      "epoch": 2.4248306010928964,
      "grad_norm": 0.7496220615630445,
      "learning_rate": 7.086779856737215e-05,
      "loss": 11.8365,
      "step": 13867
    },
    {
      "epoch": 2.4250054644808743,
      "grad_norm": 0.6985218909114406,
      "learning_rate": 7.08542522008444e-05,
      "loss": 11.9845,
      "step": 13868
    },
    {
      "epoch": 2.4251803278688526,
      "grad_norm": 0.6805404065739779,
      "learning_rate": 7.08407064187784e-05,
      "loss": 11.9537,
      "step": 13869
    },
    {
      "epoch": 2.4253551912568305,
      "grad_norm": 0.6552727178917227,
      "learning_rate": 7.082716122144572e-05,
      "loss": 12.0127,
      "step": 13870
    },
    {
      "epoch": 2.425530054644809,
      "grad_norm": 0.6333164042168475,
      "learning_rate": 7.081361660911805e-05,
      "loss": 12.1109,
      "step": 13871
    },
    {
      "epoch": 2.4257049180327868,
      "grad_norm": 0.7020019034419631,
      "learning_rate": 7.080007258206698e-05,
      "loss": 11.7515,
      "step": 13872
    },
    {
      "epoch": 2.425879781420765,
      "grad_norm": 0.6452023240948739,
      "learning_rate": 7.078652914056414e-05,
      "loss": 11.6695,
      "step": 13873
    },
    {
      "epoch": 2.426054644808743,
      "grad_norm": 0.7234279248875205,
      "learning_rate": 7.077298628488104e-05,
      "loss": 12.0403,
      "step": 13874
    },
    {
      "epoch": 2.4262295081967213,
      "grad_norm": 0.6205385622628556,
      "learning_rate": 7.075944401528932e-05,
      "loss": 12.0617,
      "step": 13875
    },
    {
      "epoch": 2.4264043715846997,
      "grad_norm": 0.6999380893224638,
      "learning_rate": 7.074590233206053e-05,
      "loss": 11.8738,
      "step": 13876
    },
    {
      "epoch": 2.4265792349726776,
      "grad_norm": 0.7124429212421164,
      "learning_rate": 7.073236123546622e-05,
      "loss": 11.8727,
      "step": 13877
    },
    {
      "epoch": 2.426754098360656,
      "grad_norm": 0.7005835069911206,
      "learning_rate": 7.071882072577796e-05,
      "loss": 11.9705,
      "step": 13878
    },
    {
      "epoch": 2.426928961748634,
      "grad_norm": 0.6237314513012369,
      "learning_rate": 7.070528080326722e-05,
      "loss": 11.7295,
      "step": 13879
    },
    {
      "epoch": 2.427103825136612,
      "grad_norm": 0.7662948252436891,
      "learning_rate": 7.069174146820552e-05,
      "loss": 11.7917,
      "step": 13880
    },
    {
      "epoch": 2.42727868852459,
      "grad_norm": 0.6742094892646353,
      "learning_rate": 7.067820272086443e-05,
      "loss": 11.7508,
      "step": 13881
    },
    {
      "epoch": 2.4274535519125684,
      "grad_norm": 0.6270130461437268,
      "learning_rate": 7.066466456151541e-05,
      "loss": 11.9349,
      "step": 13882
    },
    {
      "epoch": 2.4276284153005463,
      "grad_norm": 0.6985795834144602,
      "learning_rate": 7.065112699042993e-05,
      "loss": 11.9487,
      "step": 13883
    },
    {
      "epoch": 2.4278032786885246,
      "grad_norm": 0.6499139689279919,
      "learning_rate": 7.063759000787945e-05,
      "loss": 11.9578,
      "step": 13884
    },
    {
      "epoch": 2.427978142076503,
      "grad_norm": 0.6718946198897817,
      "learning_rate": 7.062405361413545e-05,
      "loss": 12.0576,
      "step": 13885
    },
    {
      "epoch": 2.428153005464481,
      "grad_norm": 0.6583203885593792,
      "learning_rate": 7.061051780946936e-05,
      "loss": 12.003,
      "step": 13886
    },
    {
      "epoch": 2.4283278688524588,
      "grad_norm": 0.5787370201030146,
      "learning_rate": 7.059698259415266e-05,
      "loss": 12.0724,
      "step": 13887
    },
    {
      "epoch": 2.428502732240437,
      "grad_norm": 0.6560613262017936,
      "learning_rate": 7.058344796845673e-05,
      "loss": 11.7234,
      "step": 13888
    },
    {
      "epoch": 2.4286775956284155,
      "grad_norm": 0.7104301094764379,
      "learning_rate": 7.056991393265298e-05,
      "loss": 11.9495,
      "step": 13889
    },
    {
      "epoch": 2.4288524590163934,
      "grad_norm": 0.6526770717313247,
      "learning_rate": 7.055638048701278e-05,
      "loss": 11.8467,
      "step": 13890
    },
    {
      "epoch": 2.4290273224043717,
      "grad_norm": 0.8338379164842166,
      "learning_rate": 7.05428476318076e-05,
      "loss": 12.0086,
      "step": 13891
    },
    {
      "epoch": 2.4292021857923496,
      "grad_norm": 0.6475918132085033,
      "learning_rate": 7.052931536730878e-05,
      "loss": 11.9108,
      "step": 13892
    },
    {
      "epoch": 2.429377049180328,
      "grad_norm": 0.6926034651766979,
      "learning_rate": 7.051578369378764e-05,
      "loss": 11.9104,
      "step": 13893
    },
    {
      "epoch": 2.429551912568306,
      "grad_norm": 0.7029710711839376,
      "learning_rate": 7.050225261151558e-05,
      "loss": 12.1287,
      "step": 13894
    },
    {
      "epoch": 2.429726775956284,
      "grad_norm": 0.6594084643995964,
      "learning_rate": 7.048872212076392e-05,
      "loss": 11.941,
      "step": 13895
    },
    {
      "epoch": 2.429901639344262,
      "grad_norm": 0.6384277134070162,
      "learning_rate": 7.0475192221804e-05,
      "loss": 11.816,
      "step": 13896
    },
    {
      "epoch": 2.4300765027322404,
      "grad_norm": 0.652274651854183,
      "learning_rate": 7.046166291490715e-05,
      "loss": 11.8573,
      "step": 13897
    },
    {
      "epoch": 2.4302513661202187,
      "grad_norm": 0.6474952105212277,
      "learning_rate": 7.044813420034462e-05,
      "loss": 11.7374,
      "step": 13898
    },
    {
      "epoch": 2.4304262295081966,
      "grad_norm": 0.6599827640234969,
      "learning_rate": 7.043460607838772e-05,
      "loss": 11.7529,
      "step": 13899
    },
    {
      "epoch": 2.430601092896175,
      "grad_norm": 0.6802907167178901,
      "learning_rate": 7.04210785493078e-05,
      "loss": 11.5363,
      "step": 13900
    },
    {
      "epoch": 2.430775956284153,
      "grad_norm": 0.6694457708062803,
      "learning_rate": 7.040755161337605e-05,
      "loss": 11.9775,
      "step": 13901
    },
    {
      "epoch": 2.4309508196721312,
      "grad_norm": 0.6674904694805001,
      "learning_rate": 7.039402527086373e-05,
      "loss": 11.8707,
      "step": 13902
    },
    {
      "epoch": 2.431125683060109,
      "grad_norm": 0.709288599187284,
      "learning_rate": 7.038049952204215e-05,
      "loss": 11.731,
      "step": 13903
    },
    {
      "epoch": 2.4313005464480875,
      "grad_norm": 0.6935487091448725,
      "learning_rate": 7.036697436718242e-05,
      "loss": 12.0669,
      "step": 13904
    },
    {
      "epoch": 2.4314754098360654,
      "grad_norm": 0.6407778887937499,
      "learning_rate": 7.035344980655592e-05,
      "loss": 11.7186,
      "step": 13905
    },
    {
      "epoch": 2.4316502732240437,
      "grad_norm": 0.6590480543642598,
      "learning_rate": 7.033992584043374e-05,
      "loss": 11.7496,
      "step": 13906
    },
    {
      "epoch": 2.431825136612022,
      "grad_norm": 0.6354410620681576,
      "learning_rate": 7.032640246908713e-05,
      "loss": 11.6115,
      "step": 13907
    },
    {
      "epoch": 2.432,
      "grad_norm": 0.6835388621423695,
      "learning_rate": 7.031287969278728e-05,
      "loss": 11.9069,
      "step": 13908
    },
    {
      "epoch": 2.4321748633879783,
      "grad_norm": 0.6753631931877783,
      "learning_rate": 7.02993575118053e-05,
      "loss": 11.6426,
      "step": 13909
    },
    {
      "epoch": 2.432349726775956,
      "grad_norm": 0.7537249163044215,
      "learning_rate": 7.02858359264124e-05,
      "loss": 12.1271,
      "step": 13910
    },
    {
      "epoch": 2.4325245901639345,
      "grad_norm": 0.6693725068332161,
      "learning_rate": 7.027231493687974e-05,
      "loss": 12.046,
      "step": 13911
    },
    {
      "epoch": 2.4326994535519124,
      "grad_norm": 0.6355662030610274,
      "learning_rate": 7.025879454347846e-05,
      "loss": 12.0585,
      "step": 13912
    },
    {
      "epoch": 2.4328743169398908,
      "grad_norm": 0.6542502984918057,
      "learning_rate": 7.024527474647964e-05,
      "loss": 12.0705,
      "step": 13913
    },
    {
      "epoch": 2.4330491803278687,
      "grad_norm": 0.6200301039833437,
      "learning_rate": 7.023175554615441e-05,
      "loss": 12.0415,
      "step": 13914
    },
    {
      "epoch": 2.433224043715847,
      "grad_norm": 0.6500807098537192,
      "learning_rate": 7.02182369427739e-05,
      "loss": 11.8951,
      "step": 13915
    },
    {
      "epoch": 2.4333989071038253,
      "grad_norm": 0.6350580469748062,
      "learning_rate": 7.020471893660917e-05,
      "loss": 11.667,
      "step": 13916
    },
    {
      "epoch": 2.4335737704918032,
      "grad_norm": 0.6750314301687411,
      "learning_rate": 7.019120152793135e-05,
      "loss": 11.9052,
      "step": 13917
    },
    {
      "epoch": 2.4337486338797816,
      "grad_norm": 0.6715459036930824,
      "learning_rate": 7.01776847170114e-05,
      "loss": 11.7082,
      "step": 13918
    },
    {
      "epoch": 2.4339234972677595,
      "grad_norm": 0.6635605889116803,
      "learning_rate": 7.016416850412046e-05,
      "loss": 11.9115,
      "step": 13919
    },
    {
      "epoch": 2.434098360655738,
      "grad_norm": 0.6413660430546424,
      "learning_rate": 7.015065288952957e-05,
      "loss": 11.928,
      "step": 13920
    },
    {
      "epoch": 2.4342732240437157,
      "grad_norm": 0.6769174079399213,
      "learning_rate": 7.013713787350975e-05,
      "loss": 12.0116,
      "step": 13921
    },
    {
      "epoch": 2.434448087431694,
      "grad_norm": 0.7394596732724394,
      "learning_rate": 7.012362345633199e-05,
      "loss": 11.955,
      "step": 13922
    },
    {
      "epoch": 2.434622950819672,
      "grad_norm": 0.6066695355743332,
      "learning_rate": 7.011010963826727e-05,
      "loss": 11.6735,
      "step": 13923
    },
    {
      "epoch": 2.4347978142076503,
      "grad_norm": 0.6808379947856806,
      "learning_rate": 7.009659641958668e-05,
      "loss": 11.9857,
      "step": 13924
    },
    {
      "epoch": 2.4349726775956286,
      "grad_norm": 0.6365785336774855,
      "learning_rate": 7.008308380056112e-05,
      "loss": 12.0299,
      "step": 13925
    },
    {
      "epoch": 2.4351475409836065,
      "grad_norm": 0.677591317135453,
      "learning_rate": 7.006957178146162e-05,
      "loss": 11.9654,
      "step": 13926
    },
    {
      "epoch": 2.435322404371585,
      "grad_norm": 0.6262279972746797,
      "learning_rate": 7.005606036255909e-05,
      "loss": 12.0806,
      "step": 13927
    },
    {
      "epoch": 2.4354972677595628,
      "grad_norm": 0.6275798980141947,
      "learning_rate": 7.004254954412446e-05,
      "loss": 11.5719,
      "step": 13928
    },
    {
      "epoch": 2.435672131147541,
      "grad_norm": 0.6740907531195276,
      "learning_rate": 7.00290393264287e-05,
      "loss": 11.9,
      "step": 13929
    },
    {
      "epoch": 2.435846994535519,
      "grad_norm": 0.6846485202107483,
      "learning_rate": 7.001552970974277e-05,
      "loss": 11.8684,
      "step": 13930
    },
    {
      "epoch": 2.4360218579234973,
      "grad_norm": 0.6193861238236462,
      "learning_rate": 7.00020206943375e-05,
      "loss": 12.0051,
      "step": 13931
    },
    {
      "epoch": 2.4361967213114752,
      "grad_norm": 0.6208422472406023,
      "learning_rate": 6.998851228048382e-05,
      "loss": 11.6225,
      "step": 13932
    },
    {
      "epoch": 2.4363715846994536,
      "grad_norm": 0.6903251453574716,
      "learning_rate": 6.997500446845259e-05,
      "loss": 12.0806,
      "step": 13933
    },
    {
      "epoch": 2.436546448087432,
      "grad_norm": 0.6597499585150182,
      "learning_rate": 6.996149725851475e-05,
      "loss": 11.9651,
      "step": 13934
    },
    {
      "epoch": 2.43672131147541,
      "grad_norm": 0.6040582882554278,
      "learning_rate": 6.994799065094113e-05,
      "loss": 11.8923,
      "step": 13935
    },
    {
      "epoch": 2.436896174863388,
      "grad_norm": 0.6197718540839149,
      "learning_rate": 6.993448464600254e-05,
      "loss": 11.6357,
      "step": 13936
    },
    {
      "epoch": 2.437071038251366,
      "grad_norm": 0.6348470843275305,
      "learning_rate": 6.992097924396986e-05,
      "loss": 12.0992,
      "step": 13937
    },
    {
      "epoch": 2.4372459016393444,
      "grad_norm": 0.6609372245886289,
      "learning_rate": 6.990747444511387e-05,
      "loss": 11.8687,
      "step": 13938
    },
    {
      "epoch": 2.4374207650273223,
      "grad_norm": 0.6349129952954786,
      "learning_rate": 6.989397024970545e-05,
      "loss": 11.9008,
      "step": 13939
    },
    {
      "epoch": 2.4375956284153006,
      "grad_norm": 0.6572715258199837,
      "learning_rate": 6.988046665801536e-05,
      "loss": 11.846,
      "step": 13940
    },
    {
      "epoch": 2.4377704918032785,
      "grad_norm": 0.6150466838065342,
      "learning_rate": 6.986696367031439e-05,
      "loss": 11.7578,
      "step": 13941
    },
    {
      "epoch": 2.437945355191257,
      "grad_norm": 0.6320716577313535,
      "learning_rate": 6.985346128687329e-05,
      "loss": 12.0445,
      "step": 13942
    },
    {
      "epoch": 2.438120218579235,
      "grad_norm": 0.7125338496482918,
      "learning_rate": 6.983995950796288e-05,
      "loss": 11.9314,
      "step": 13943
    },
    {
      "epoch": 2.438295081967213,
      "grad_norm": 0.623580358479271,
      "learning_rate": 6.982645833385391e-05,
      "loss": 11.8273,
      "step": 13944
    },
    {
      "epoch": 2.438469945355191,
      "grad_norm": 0.6756581252766998,
      "learning_rate": 6.981295776481707e-05,
      "loss": 12.0561,
      "step": 13945
    },
    {
      "epoch": 2.4386448087431694,
      "grad_norm": 0.7139152560958105,
      "learning_rate": 6.97994578011231e-05,
      "loss": 12.0754,
      "step": 13946
    },
    {
      "epoch": 2.4388196721311477,
      "grad_norm": 0.6514980101667597,
      "learning_rate": 6.978595844304271e-05,
      "loss": 11.9348,
      "step": 13947
    },
    {
      "epoch": 2.4389945355191256,
      "grad_norm": 0.6871883226422012,
      "learning_rate": 6.977245969084669e-05,
      "loss": 12.0166,
      "step": 13948
    },
    {
      "epoch": 2.439169398907104,
      "grad_norm": 0.6851824056984152,
      "learning_rate": 6.975896154480563e-05,
      "loss": 11.7134,
      "step": 13949
    },
    {
      "epoch": 2.439344262295082,
      "grad_norm": 0.5980947649896949,
      "learning_rate": 6.974546400519023e-05,
      "loss": 11.5754,
      "step": 13950
    },
    {
      "epoch": 2.43951912568306,
      "grad_norm": 0.7547553878939677,
      "learning_rate": 6.973196707227121e-05,
      "loss": 12.1292,
      "step": 13951
    },
    {
      "epoch": 2.439693989071038,
      "grad_norm": 0.6960170324238985,
      "learning_rate": 6.971847074631911e-05,
      "loss": 11.8245,
      "step": 13952
    },
    {
      "epoch": 2.4398688524590164,
      "grad_norm": 0.6614963096240288,
      "learning_rate": 6.970497502760471e-05,
      "loss": 11.8019,
      "step": 13953
    },
    {
      "epoch": 2.4400437158469943,
      "grad_norm": 0.7327045370838089,
      "learning_rate": 6.969147991639857e-05,
      "loss": 11.9881,
      "step": 13954
    },
    {
      "epoch": 2.4402185792349727,
      "grad_norm": 0.6467497511756409,
      "learning_rate": 6.96779854129713e-05,
      "loss": 11.7768,
      "step": 13955
    },
    {
      "epoch": 2.440393442622951,
      "grad_norm": 0.6387224747910042,
      "learning_rate": 6.966449151759356e-05,
      "loss": 11.9662,
      "step": 13956
    },
    {
      "epoch": 2.440568306010929,
      "grad_norm": 0.6903283476236343,
      "learning_rate": 6.965099823053584e-05,
      "loss": 11.7398,
      "step": 13957
    },
    {
      "epoch": 2.4407431693989072,
      "grad_norm": 0.6939730733373508,
      "learning_rate": 6.963750555206885e-05,
      "loss": 12.0278,
      "step": 13958
    },
    {
      "epoch": 2.440918032786885,
      "grad_norm": 0.6799732114686492,
      "learning_rate": 6.962401348246308e-05,
      "loss": 12.0456,
      "step": 13959
    },
    {
      "epoch": 2.4410928961748635,
      "grad_norm": 0.7179494853747117,
      "learning_rate": 6.961052202198912e-05,
      "loss": 11.8511,
      "step": 13960
    },
    {
      "epoch": 2.4412677595628414,
      "grad_norm": 0.896664228359917,
      "learning_rate": 6.959703117091748e-05,
      "loss": 11.7675,
      "step": 13961
    },
    {
      "epoch": 2.4414426229508197,
      "grad_norm": 0.6649151408662635,
      "learning_rate": 6.95835409295187e-05,
      "loss": 11.866,
      "step": 13962
    },
    {
      "epoch": 2.4416174863387976,
      "grad_norm": 0.7847002007138206,
      "learning_rate": 6.957005129806333e-05,
      "loss": 12.0595,
      "step": 13963
    },
    {
      "epoch": 2.441792349726776,
      "grad_norm": 0.7063432369810692,
      "learning_rate": 6.955656227682187e-05,
      "loss": 11.9166,
      "step": 13964
    },
    {
      "epoch": 2.4419672131147543,
      "grad_norm": 0.8013956917606483,
      "learning_rate": 6.954307386606483e-05,
      "loss": 11.7809,
      "step": 13965
    },
    {
      "epoch": 2.442142076502732,
      "grad_norm": 0.7515302462771736,
      "learning_rate": 6.952958606606264e-05,
      "loss": 11.8695,
      "step": 13966
    },
    {
      "epoch": 2.4423169398907105,
      "grad_norm": 0.5851257131597393,
      "learning_rate": 6.951609887708582e-05,
      "loss": 11.8118,
      "step": 13967
    },
    {
      "epoch": 2.4424918032786884,
      "grad_norm": 0.6417953193103364,
      "learning_rate": 6.95026122994048e-05,
      "loss": 11.8673,
      "step": 13968
    },
    {
      "epoch": 2.4426666666666668,
      "grad_norm": 0.6365458269957398,
      "learning_rate": 6.948912633329007e-05,
      "loss": 11.8116,
      "step": 13969
    },
    {
      "epoch": 2.4428415300546447,
      "grad_norm": 0.6967839263800372,
      "learning_rate": 6.947564097901205e-05,
      "loss": 12.1398,
      "step": 13970
    },
    {
      "epoch": 2.443016393442623,
      "grad_norm": 0.650935876830943,
      "learning_rate": 6.94621562368411e-05,
      "loss": 11.8574,
      "step": 13971
    },
    {
      "epoch": 2.443191256830601,
      "grad_norm": 0.6246553548725803,
      "learning_rate": 6.94486721070477e-05,
      "loss": 11.9206,
      "step": 13972
    },
    {
      "epoch": 2.4433661202185792,
      "grad_norm": 0.6517019670771796,
      "learning_rate": 6.943518858990223e-05,
      "loss": 11.7968,
      "step": 13973
    },
    {
      "epoch": 2.4435409836065576,
      "grad_norm": 0.693387991196263,
      "learning_rate": 6.94217056856751e-05,
      "loss": 11.9036,
      "step": 13974
    },
    {
      "epoch": 2.4437158469945355,
      "grad_norm": 0.6710179139364587,
      "learning_rate": 6.940822339463663e-05,
      "loss": 11.9293,
      "step": 13975
    },
    {
      "epoch": 2.443890710382514,
      "grad_norm": 0.6617059836584682,
      "learning_rate": 6.93947417170572e-05,
      "loss": 11.9618,
      "step": 13976
    },
    {
      "epoch": 2.4440655737704917,
      "grad_norm": 0.6613882996185293,
      "learning_rate": 6.938126065320721e-05,
      "loss": 11.9853,
      "step": 13977
    },
    {
      "epoch": 2.44424043715847,
      "grad_norm": 0.6133139506029968,
      "learning_rate": 6.936778020335695e-05,
      "loss": 12.1207,
      "step": 13978
    },
    {
      "epoch": 2.444415300546448,
      "grad_norm": 0.7414241533494187,
      "learning_rate": 6.935430036777673e-05,
      "loss": 11.7895,
      "step": 13979
    },
    {
      "epoch": 2.4445901639344263,
      "grad_norm": 0.635945267902119,
      "learning_rate": 6.934082114673688e-05,
      "loss": 11.5709,
      "step": 13980
    },
    {
      "epoch": 2.444765027322404,
      "grad_norm": 0.6121772893772676,
      "learning_rate": 6.932734254050769e-05,
      "loss": 11.5697,
      "step": 13981
    },
    {
      "epoch": 2.4449398907103825,
      "grad_norm": 0.614625467179982,
      "learning_rate": 6.931386454935948e-05,
      "loss": 12.0735,
      "step": 13982
    },
    {
      "epoch": 2.445114754098361,
      "grad_norm": 0.7093746038884831,
      "learning_rate": 6.930038717356252e-05,
      "loss": 11.9515,
      "step": 13983
    },
    {
      "epoch": 2.4452896174863388,
      "grad_norm": 0.6222128044401033,
      "learning_rate": 6.928691041338704e-05,
      "loss": 11.8752,
      "step": 13984
    },
    {
      "epoch": 2.445464480874317,
      "grad_norm": 0.5900603473111848,
      "learning_rate": 6.927343426910332e-05,
      "loss": 11.9758,
      "step": 13985
    },
    {
      "epoch": 2.445639344262295,
      "grad_norm": 0.5970265082487417,
      "learning_rate": 6.925995874098155e-05,
      "loss": 11.9851,
      "step": 13986
    },
    {
      "epoch": 2.4458142076502734,
      "grad_norm": 0.6025651510771706,
      "learning_rate": 6.924648382929201e-05,
      "loss": 11.957,
      "step": 13987
    },
    {
      "epoch": 2.4459890710382513,
      "grad_norm": 0.6157469084557583,
      "learning_rate": 6.923300953430491e-05,
      "loss": 11.667,
      "step": 13988
    },
    {
      "epoch": 2.4461639344262296,
      "grad_norm": 0.6290578306309609,
      "learning_rate": 6.921953585629043e-05,
      "loss": 11.841,
      "step": 13989
    },
    {
      "epoch": 2.4463387978142075,
      "grad_norm": 0.6358278749091008,
      "learning_rate": 6.920606279551873e-05,
      "loss": 11.8563,
      "step": 13990
    },
    {
      "epoch": 2.446513661202186,
      "grad_norm": 0.6079785002952379,
      "learning_rate": 6.919259035226006e-05,
      "loss": 11.7858,
      "step": 13991
    },
    {
      "epoch": 2.446688524590164,
      "grad_norm": 0.6681834696932814,
      "learning_rate": 6.917911852678456e-05,
      "loss": 11.8782,
      "step": 13992
    },
    {
      "epoch": 2.446863387978142,
      "grad_norm": 0.6618327103559655,
      "learning_rate": 6.916564731936233e-05,
      "loss": 11.9458,
      "step": 13993
    },
    {
      "epoch": 2.4470382513661204,
      "grad_norm": 0.6338160618614522,
      "learning_rate": 6.915217673026357e-05,
      "loss": 11.8439,
      "step": 13994
    },
    {
      "epoch": 2.4472131147540983,
      "grad_norm": 0.6449785433336878,
      "learning_rate": 6.913870675975834e-05,
      "loss": 11.8459,
      "step": 13995
    },
    {
      "epoch": 2.4473879781420766,
      "grad_norm": 0.7268911317356459,
      "learning_rate": 6.912523740811685e-05,
      "loss": 12.039,
      "step": 13996
    },
    {
      "epoch": 2.4475628415300545,
      "grad_norm": 0.6148538682803538,
      "learning_rate": 6.911176867560911e-05,
      "loss": 11.8275,
      "step": 13997
    },
    {
      "epoch": 2.447737704918033,
      "grad_norm": 0.6358316441556234,
      "learning_rate": 6.909830056250527e-05,
      "loss": 11.7641,
      "step": 13998
    },
    {
      "epoch": 2.447912568306011,
      "grad_norm": 0.6590992699907039,
      "learning_rate": 6.90848330690754e-05,
      "loss": 11.8025,
      "step": 13999
    },
    {
      "epoch": 2.448087431693989,
      "grad_norm": 0.6431997230093676,
      "learning_rate": 6.907136619558947e-05,
      "loss": 12.0145,
      "step": 14000
    },
    {
      "epoch": 2.448262295081967,
      "grad_norm": 0.663683700375629,
      "learning_rate": 6.905789994231769e-05,
      "loss": 11.4181,
      "step": 14001
    },
    {
      "epoch": 2.4484371584699454,
      "grad_norm": 0.688723712087997,
      "learning_rate": 6.904443430952999e-05,
      "loss": 11.7718,
      "step": 14002
    },
    {
      "epoch": 2.4486120218579233,
      "grad_norm": 0.6524240622567288,
      "learning_rate": 6.903096929749644e-05,
      "loss": 11.8034,
      "step": 14003
    },
    {
      "epoch": 2.4487868852459016,
      "grad_norm": 0.6262198636255522,
      "learning_rate": 6.901750490648707e-05,
      "loss": 11.6862,
      "step": 14004
    },
    {
      "epoch": 2.44896174863388,
      "grad_norm": 0.630453569948301,
      "learning_rate": 6.90040411367718e-05,
      "loss": 11.9543,
      "step": 14005
    },
    {
      "epoch": 2.449136612021858,
      "grad_norm": 0.613981787906437,
      "learning_rate": 6.899057798862067e-05,
      "loss": 11.6556,
      "step": 14006
    },
    {
      "epoch": 2.449311475409836,
      "grad_norm": 0.6061198267074803,
      "learning_rate": 6.89771154623037e-05,
      "loss": 11.929,
      "step": 14007
    },
    {
      "epoch": 2.449486338797814,
      "grad_norm": 0.6374864483169359,
      "learning_rate": 6.896365355809082e-05,
      "loss": 12.0468,
      "step": 14008
    },
    {
      "epoch": 2.4496612021857924,
      "grad_norm": 0.6039963642213121,
      "learning_rate": 6.895019227625198e-05,
      "loss": 11.6855,
      "step": 14009
    },
    {
      "epoch": 2.4498360655737703,
      "grad_norm": 0.6345117286437801,
      "learning_rate": 6.893673161705709e-05,
      "loss": 11.9849,
      "step": 14010
    },
    {
      "epoch": 2.4500109289617487,
      "grad_norm": 0.6275850622619802,
      "learning_rate": 6.892327158077611e-05,
      "loss": 11.8036,
      "step": 14011
    },
    {
      "epoch": 2.4501857923497266,
      "grad_norm": 0.6757011810712387,
      "learning_rate": 6.890981216767898e-05,
      "loss": 11.7896,
      "step": 14012
    },
    {
      "epoch": 2.450360655737705,
      "grad_norm": 0.7017224222568503,
      "learning_rate": 6.88963533780356e-05,
      "loss": 11.8523,
      "step": 14013
    },
    {
      "epoch": 2.4505355191256832,
      "grad_norm": 0.6615467990680074,
      "learning_rate": 6.888289521211578e-05,
      "loss": 11.7725,
      "step": 14014
    },
    {
      "epoch": 2.450710382513661,
      "grad_norm": 0.6791001976990692,
      "learning_rate": 6.886943767018945e-05,
      "loss": 12.0683,
      "step": 14015
    },
    {
      "epoch": 2.4508852459016395,
      "grad_norm": 0.7162387941909759,
      "learning_rate": 6.88559807525265e-05,
      "loss": 11.9573,
      "step": 14016
    },
    {
      "epoch": 2.4510601092896174,
      "grad_norm": 0.6450162604373162,
      "learning_rate": 6.884252445939679e-05,
      "loss": 12.024,
      "step": 14017
    },
    {
      "epoch": 2.4512349726775957,
      "grad_norm": 0.6310663331220839,
      "learning_rate": 6.88290687910701e-05,
      "loss": 11.5341,
      "step": 14018
    },
    {
      "epoch": 2.4514098360655736,
      "grad_norm": 0.6927425236816513,
      "learning_rate": 6.881561374781627e-05,
      "loss": 11.8265,
      "step": 14019
    },
    {
      "epoch": 2.451584699453552,
      "grad_norm": 0.7043377751172222,
      "learning_rate": 6.880215932990515e-05,
      "loss": 11.8975,
      "step": 14020
    },
    {
      "epoch": 2.45175956284153,
      "grad_norm": 0.6986610759721101,
      "learning_rate": 6.878870553760653e-05,
      "loss": 11.982,
      "step": 14021
    },
    {
      "epoch": 2.451934426229508,
      "grad_norm": 0.6432468975965122,
      "learning_rate": 6.877525237119022e-05,
      "loss": 11.945,
      "step": 14022
    },
    {
      "epoch": 2.4521092896174865,
      "grad_norm": 0.6459243540065839,
      "learning_rate": 6.876179983092594e-05,
      "loss": 11.8468,
      "step": 14023
    },
    {
      "epoch": 2.4522841530054644,
      "grad_norm": 0.7016366378044402,
      "learning_rate": 6.874834791708348e-05,
      "loss": 11.6118,
      "step": 14024
    },
    {
      "epoch": 2.4524590163934428,
      "grad_norm": 0.7334559101420682,
      "learning_rate": 6.873489662993261e-05,
      "loss": 11.9179,
      "step": 14025
    },
    {
      "epoch": 2.4526338797814207,
      "grad_norm": 0.7593946286992794,
      "learning_rate": 6.872144596974308e-05,
      "loss": 12.149,
      "step": 14026
    },
    {
      "epoch": 2.452808743169399,
      "grad_norm": 0.7373274019426983,
      "learning_rate": 6.870799593678459e-05,
      "loss": 11.9469,
      "step": 14027
    },
    {
      "epoch": 2.452983606557377,
      "grad_norm": 0.7145255529455704,
      "learning_rate": 6.869454653132684e-05,
      "loss": 11.7121,
      "step": 14028
    },
    {
      "epoch": 2.4531584699453552,
      "grad_norm": 0.701612723872864,
      "learning_rate": 6.868109775363954e-05,
      "loss": 11.8403,
      "step": 14029
    },
    {
      "epoch": 2.453333333333333,
      "grad_norm": 0.7761360675674557,
      "learning_rate": 6.866764960399241e-05,
      "loss": 11.6659,
      "step": 14030
    },
    {
      "epoch": 2.4535081967213115,
      "grad_norm": 0.6653049939781764,
      "learning_rate": 6.865420208265514e-05,
      "loss": 11.9442,
      "step": 14031
    },
    {
      "epoch": 2.45368306010929,
      "grad_norm": 0.693220275663366,
      "learning_rate": 6.864075518989733e-05,
      "loss": 12.0401,
      "step": 14032
    },
    {
      "epoch": 2.4538579234972677,
      "grad_norm": 0.7775443631646869,
      "learning_rate": 6.862730892598869e-05,
      "loss": 11.746,
      "step": 14033
    },
    {
      "epoch": 2.454032786885246,
      "grad_norm": 0.6575429974348481,
      "learning_rate": 6.861386329119877e-05,
      "loss": 11.965,
      "step": 14034
    },
    {
      "epoch": 2.454207650273224,
      "grad_norm": 0.6810357832655517,
      "learning_rate": 6.860041828579732e-05,
      "loss": 11.9349,
      "step": 14035
    },
    {
      "epoch": 2.4543825136612023,
      "grad_norm": 0.6641344621382407,
      "learning_rate": 6.858697391005386e-05,
      "loss": 11.9236,
      "step": 14036
    },
    {
      "epoch": 2.45455737704918,
      "grad_norm": 0.5874495125751601,
      "learning_rate": 6.857353016423804e-05,
      "loss": 11.9388,
      "step": 14037
    },
    {
      "epoch": 2.4547322404371585,
      "grad_norm": 0.6784222252597246,
      "learning_rate": 6.856008704861942e-05,
      "loss": 11.9106,
      "step": 14038
    },
    {
      "epoch": 2.4549071038251364,
      "grad_norm": 0.6774230726531459,
      "learning_rate": 6.854664456346759e-05,
      "loss": 11.8767,
      "step": 14039
    },
    {
      "epoch": 2.455081967213115,
      "grad_norm": 0.6386534515827244,
      "learning_rate": 6.853320270905213e-05,
      "loss": 11.9318,
      "step": 14040
    },
    {
      "epoch": 2.455256830601093,
      "grad_norm": 0.729307323371052,
      "learning_rate": 6.851976148564257e-05,
      "loss": 12.0294,
      "step": 14041
    },
    {
      "epoch": 2.455431693989071,
      "grad_norm": 0.6707258232299199,
      "learning_rate": 6.850632089350846e-05,
      "loss": 11.8826,
      "step": 14042
    },
    {
      "epoch": 2.4556065573770494,
      "grad_norm": 0.6443809943734856,
      "learning_rate": 6.849288093291924e-05,
      "loss": 11.798,
      "step": 14043
    },
    {
      "epoch": 2.4557814207650273,
      "grad_norm": 0.6529521591431062,
      "learning_rate": 6.847944160414456e-05,
      "loss": 11.646,
      "step": 14044
    },
    {
      "epoch": 2.4559562841530056,
      "grad_norm": 0.6336620608146928,
      "learning_rate": 6.846600290745386e-05,
      "loss": 11.7671,
      "step": 14045
    },
    {
      "epoch": 2.4561311475409835,
      "grad_norm": 0.713727811063279,
      "learning_rate": 6.845256484311662e-05,
      "loss": 12.1921,
      "step": 14046
    },
    {
      "epoch": 2.456306010928962,
      "grad_norm": 0.6910401725988315,
      "learning_rate": 6.843912741140232e-05,
      "loss": 11.9521,
      "step": 14047
    },
    {
      "epoch": 2.4564808743169397,
      "grad_norm": 0.6705052079998064,
      "learning_rate": 6.842569061258037e-05,
      "loss": 11.9657,
      "step": 14048
    },
    {
      "epoch": 2.456655737704918,
      "grad_norm": 0.6622496378634312,
      "learning_rate": 6.841225444692032e-05,
      "loss": 11.9227,
      "step": 14049
    },
    {
      "epoch": 2.4568306010928964,
      "grad_norm": 0.6837919840475464,
      "learning_rate": 6.839881891469156e-05,
      "loss": 11.8539,
      "step": 14050
    },
    {
      "epoch": 2.4570054644808743,
      "grad_norm": 0.5754159382225503,
      "learning_rate": 6.83853840161635e-05,
      "loss": 11.9862,
      "step": 14051
    },
    {
      "epoch": 2.457180327868852,
      "grad_norm": 0.6208649949461534,
      "learning_rate": 6.837194975160554e-05,
      "loss": 11.8843,
      "step": 14052
    },
    {
      "epoch": 2.4573551912568306,
      "grad_norm": 0.5657011507769049,
      "learning_rate": 6.835851612128709e-05,
      "loss": 11.7622,
      "step": 14053
    },
    {
      "epoch": 2.457530054644809,
      "grad_norm": 0.6050231403302504,
      "learning_rate": 6.834508312547757e-05,
      "loss": 11.7661,
      "step": 14054
    },
    {
      "epoch": 2.457704918032787,
      "grad_norm": 0.603163087620878,
      "learning_rate": 6.833165076444633e-05,
      "loss": 11.8386,
      "step": 14055
    },
    {
      "epoch": 2.457879781420765,
      "grad_norm": 0.6850628217767866,
      "learning_rate": 6.831821903846273e-05,
      "loss": 11.9609,
      "step": 14056
    },
    {
      "epoch": 2.458054644808743,
      "grad_norm": 0.6275931674747238,
      "learning_rate": 6.83047879477961e-05,
      "loss": 11.7721,
      "step": 14057
    },
    {
      "epoch": 2.4582295081967214,
      "grad_norm": 0.6229155921615366,
      "learning_rate": 6.829135749271578e-05,
      "loss": 11.5791,
      "step": 14058
    },
    {
      "epoch": 2.4584043715846993,
      "grad_norm": 0.7144535527267986,
      "learning_rate": 6.827792767349111e-05,
      "loss": 11.9921,
      "step": 14059
    },
    {
      "epoch": 2.4585792349726776,
      "grad_norm": 0.6650292874737614,
      "learning_rate": 6.826449849039138e-05,
      "loss": 12.0647,
      "step": 14060
    },
    {
      "epoch": 2.4587540983606555,
      "grad_norm": 0.594164461002552,
      "learning_rate": 6.825106994368593e-05,
      "loss": 11.6192,
      "step": 14061
    },
    {
      "epoch": 2.458928961748634,
      "grad_norm": 0.6426099664194584,
      "learning_rate": 6.823764203364397e-05,
      "loss": 12.1083,
      "step": 14062
    },
    {
      "epoch": 2.459103825136612,
      "grad_norm": 0.6661585424550438,
      "learning_rate": 6.822421476053483e-05,
      "loss": 12.0025,
      "step": 14063
    },
    {
      "epoch": 2.45927868852459,
      "grad_norm": 0.6340750424012084,
      "learning_rate": 6.821078812462773e-05,
      "loss": 11.9137,
      "step": 14064
    },
    {
      "epoch": 2.4594535519125684,
      "grad_norm": 0.631195583335007,
      "learning_rate": 6.819736212619197e-05,
      "loss": 11.8487,
      "step": 14065
    },
    {
      "epoch": 2.4596284153005463,
      "grad_norm": 0.6625210210685741,
      "learning_rate": 6.818393676549672e-05,
      "loss": 11.9575,
      "step": 14066
    },
    {
      "epoch": 2.4598032786885247,
      "grad_norm": 0.6176334970674285,
      "learning_rate": 6.817051204281122e-05,
      "loss": 11.7422,
      "step": 14067
    },
    {
      "epoch": 2.4599781420765026,
      "grad_norm": 0.6805292204488765,
      "learning_rate": 6.81570879584047e-05,
      "loss": 11.9503,
      "step": 14068
    },
    {
      "epoch": 2.460153005464481,
      "grad_norm": 0.6398046609340083,
      "learning_rate": 6.814366451254632e-05,
      "loss": 12.1143,
      "step": 14069
    },
    {
      "epoch": 2.460327868852459,
      "grad_norm": 0.6329999202289329,
      "learning_rate": 6.813024170550531e-05,
      "loss": 11.8391,
      "step": 14070
    },
    {
      "epoch": 2.460502732240437,
      "grad_norm": 0.6739078159989238,
      "learning_rate": 6.811681953755079e-05,
      "loss": 11.7681,
      "step": 14071
    },
    {
      "epoch": 2.4606775956284155,
      "grad_norm": 0.6268032540487284,
      "learning_rate": 6.810339800895191e-05,
      "loss": 11.9391,
      "step": 14072
    },
    {
      "epoch": 2.4608524590163934,
      "grad_norm": 0.646621841796286,
      "learning_rate": 6.808997711997786e-05,
      "loss": 11.9653,
      "step": 14073
    },
    {
      "epoch": 2.4610273224043717,
      "grad_norm": 0.6394886022479656,
      "learning_rate": 6.807655687089776e-05,
      "loss": 11.7592,
      "step": 14074
    },
    {
      "epoch": 2.4612021857923496,
      "grad_norm": 0.6611637698546082,
      "learning_rate": 6.806313726198069e-05,
      "loss": 11.6039,
      "step": 14075
    },
    {
      "epoch": 2.461377049180328,
      "grad_norm": 0.688461306890388,
      "learning_rate": 6.804971829349578e-05,
      "loss": 11.9775,
      "step": 14076
    },
    {
      "epoch": 2.461551912568306,
      "grad_norm": 0.62718971012087,
      "learning_rate": 6.80362999657121e-05,
      "loss": 11.8007,
      "step": 14077
    },
    {
      "epoch": 2.461726775956284,
      "grad_norm": 0.5969833275967369,
      "learning_rate": 6.802288227889876e-05,
      "loss": 11.7232,
      "step": 14078
    },
    {
      "epoch": 2.461901639344262,
      "grad_norm": 0.7393600454557497,
      "learning_rate": 6.800946523332484e-05,
      "loss": 11.9966,
      "step": 14079
    },
    {
      "epoch": 2.4620765027322404,
      "grad_norm": 0.632245616306904,
      "learning_rate": 6.799604882925934e-05,
      "loss": 11.8599,
      "step": 14080
    },
    {
      "epoch": 2.4622513661202188,
      "grad_norm": 0.7606432932264042,
      "learning_rate": 6.798263306697135e-05,
      "loss": 11.9038,
      "step": 14081
    },
    {
      "epoch": 2.4624262295081967,
      "grad_norm": 0.7027149005510791,
      "learning_rate": 6.796921794672981e-05,
      "loss": 11.8537,
      "step": 14082
    },
    {
      "epoch": 2.462601092896175,
      "grad_norm": 0.6026767686526738,
      "learning_rate": 6.795580346880385e-05,
      "loss": 11.8763,
      "step": 14083
    },
    {
      "epoch": 2.462775956284153,
      "grad_norm": 0.63734162399892,
      "learning_rate": 6.794238963346242e-05,
      "loss": 11.9902,
      "step": 14084
    },
    {
      "epoch": 2.4629508196721313,
      "grad_norm": 0.6454775206748675,
      "learning_rate": 6.792897644097451e-05,
      "loss": 12.1514,
      "step": 14085
    },
    {
      "epoch": 2.463125683060109,
      "grad_norm": 0.6956348009178003,
      "learning_rate": 6.791556389160904e-05,
      "loss": 11.9206,
      "step": 14086
    },
    {
      "epoch": 2.4633005464480875,
      "grad_norm": 0.6355880192866753,
      "learning_rate": 6.790215198563507e-05,
      "loss": 11.9538,
      "step": 14087
    },
    {
      "epoch": 2.4634754098360654,
      "grad_norm": 0.617833386038852,
      "learning_rate": 6.788874072332152e-05,
      "loss": 12.0732,
      "step": 14088
    },
    {
      "epoch": 2.4636502732240437,
      "grad_norm": 0.6423511853399643,
      "learning_rate": 6.787533010493729e-05,
      "loss": 11.7946,
      "step": 14089
    },
    {
      "epoch": 2.463825136612022,
      "grad_norm": 0.7027229108524468,
      "learning_rate": 6.786192013075137e-05,
      "loss": 11.9939,
      "step": 14090
    },
    {
      "epoch": 2.464,
      "grad_norm": 0.625599558314447,
      "learning_rate": 6.784851080103254e-05,
      "loss": 11.6511,
      "step": 14091
    },
    {
      "epoch": 2.4641748633879783,
      "grad_norm": 0.616193387694963,
      "learning_rate": 6.783510211604986e-05,
      "loss": 11.7193,
      "step": 14092
    },
    {
      "epoch": 2.464349726775956,
      "grad_norm": 0.6550165774353645,
      "learning_rate": 6.782169407607213e-05,
      "loss": 11.9873,
      "step": 14093
    },
    {
      "epoch": 2.4645245901639345,
      "grad_norm": 0.7151263640908149,
      "learning_rate": 6.780828668136822e-05,
      "loss": 11.8643,
      "step": 14094
    },
    {
      "epoch": 2.4646994535519124,
      "grad_norm": 0.6275864201024597,
      "learning_rate": 6.779487993220702e-05,
      "loss": 11.9573,
      "step": 14095
    },
    {
      "epoch": 2.464874316939891,
      "grad_norm": 0.644757250096537,
      "learning_rate": 6.778147382885732e-05,
      "loss": 11.6995,
      "step": 14096
    },
    {
      "epoch": 2.4650491803278687,
      "grad_norm": 0.671885840536996,
      "learning_rate": 6.776806837158802e-05,
      "loss": 11.9979,
      "step": 14097
    },
    {
      "epoch": 2.465224043715847,
      "grad_norm": 0.6717186365511448,
      "learning_rate": 6.775466356066794e-05,
      "loss": 12.0347,
      "step": 14098
    },
    {
      "epoch": 2.4653989071038254,
      "grad_norm": 0.677155268359359,
      "learning_rate": 6.774125939636584e-05,
      "loss": 12.004,
      "step": 14099
    },
    {
      "epoch": 2.4655737704918033,
      "grad_norm": 0.6179933988391048,
      "learning_rate": 6.772785587895054e-05,
      "loss": 11.9921,
      "step": 14100
    },
    {
      "epoch": 2.4657486338797816,
      "grad_norm": 0.8070166501572474,
      "learning_rate": 6.771445300869079e-05,
      "loss": 11.9551,
      "step": 14101
    },
    {
      "epoch": 2.4659234972677595,
      "grad_norm": 0.748529749203194,
      "learning_rate": 6.77010507858554e-05,
      "loss": 11.8605,
      "step": 14102
    },
    {
      "epoch": 2.466098360655738,
      "grad_norm": 0.6853964206967524,
      "learning_rate": 6.768764921071312e-05,
      "loss": 11.8175,
      "step": 14103
    },
    {
      "epoch": 2.4662732240437157,
      "grad_norm": 0.7192234423500365,
      "learning_rate": 6.76742482835327e-05,
      "loss": 12.0249,
      "step": 14104
    },
    {
      "epoch": 2.466448087431694,
      "grad_norm": 0.6740642672541199,
      "learning_rate": 6.766084800458285e-05,
      "loss": 11.6731,
      "step": 14105
    },
    {
      "epoch": 2.466622950819672,
      "grad_norm": 0.710619656563819,
      "learning_rate": 6.764744837413225e-05,
      "loss": 11.9199,
      "step": 14106
    },
    {
      "epoch": 2.4667978142076503,
      "grad_norm": 0.6730377320342508,
      "learning_rate": 6.763404939244968e-05,
      "loss": 12.015,
      "step": 14107
    },
    {
      "epoch": 2.4669726775956287,
      "grad_norm": 0.6333991969252201,
      "learning_rate": 6.762065105980381e-05,
      "loss": 11.9648,
      "step": 14108
    },
    {
      "epoch": 2.4671475409836066,
      "grad_norm": 0.6447218908836038,
      "learning_rate": 6.76072533764633e-05,
      "loss": 11.9152,
      "step": 14109
    },
    {
      "epoch": 2.4673224043715845,
      "grad_norm": 0.6915463383411716,
      "learning_rate": 6.759385634269679e-05,
      "loss": 11.7952,
      "step": 14110
    },
    {
      "epoch": 2.467497267759563,
      "grad_norm": 0.6817260473910594,
      "learning_rate": 6.758045995877297e-05,
      "loss": 11.9746,
      "step": 14111
    },
    {
      "epoch": 2.467672131147541,
      "grad_norm": 0.6424186988188596,
      "learning_rate": 6.756706422496049e-05,
      "loss": 12.0417,
      "step": 14112
    },
    {
      "epoch": 2.467846994535519,
      "grad_norm": 0.6578271104997667,
      "learning_rate": 6.755366914152795e-05,
      "loss": 12.0509,
      "step": 14113
    },
    {
      "epoch": 2.4680218579234974,
      "grad_norm": 0.6453851027264185,
      "learning_rate": 6.754027470874396e-05,
      "loss": 11.7752,
      "step": 14114
    },
    {
      "epoch": 2.4681967213114753,
      "grad_norm": 0.653913556531626,
      "learning_rate": 6.75268809268771e-05,
      "loss": 11.9141,
      "step": 14115
    },
    {
      "epoch": 2.4683715846994536,
      "grad_norm": 0.7196482792101261,
      "learning_rate": 6.7513487796196e-05,
      "loss": 11.8624,
      "step": 14116
    },
    {
      "epoch": 2.4685464480874315,
      "grad_norm": 0.6783469780877773,
      "learning_rate": 6.750009531696926e-05,
      "loss": 11.531,
      "step": 14117
    },
    {
      "epoch": 2.46872131147541,
      "grad_norm": 0.7159912411024112,
      "learning_rate": 6.748670348946535e-05,
      "loss": 12.0443,
      "step": 14118
    },
    {
      "epoch": 2.4688961748633877,
      "grad_norm": 0.7284704115325387,
      "learning_rate": 6.747331231395288e-05,
      "loss": 11.9183,
      "step": 14119
    },
    {
      "epoch": 2.469071038251366,
      "grad_norm": 0.6547010987734843,
      "learning_rate": 6.745992179070034e-05,
      "loss": 12.0484,
      "step": 14120
    },
    {
      "epoch": 2.4692459016393444,
      "grad_norm": 0.6817057505646509,
      "learning_rate": 6.744653191997629e-05,
      "loss": 11.9279,
      "step": 14121
    },
    {
      "epoch": 2.4694207650273223,
      "grad_norm": 0.6505483440845056,
      "learning_rate": 6.743314270204926e-05,
      "loss": 11.9901,
      "step": 14122
    },
    {
      "epoch": 2.4695956284153007,
      "grad_norm": 0.6257623529040711,
      "learning_rate": 6.74197541371877e-05,
      "loss": 12.0231,
      "step": 14123
    },
    {
      "epoch": 2.4697704918032786,
      "grad_norm": 0.6893520292647118,
      "learning_rate": 6.74063662256601e-05,
      "loss": 11.968,
      "step": 14124
    },
    {
      "epoch": 2.469945355191257,
      "grad_norm": 0.6775573135676658,
      "learning_rate": 6.73929789677349e-05,
      "loss": 12.0361,
      "step": 14125
    },
    {
      "epoch": 2.470120218579235,
      "grad_norm": 0.6831358599391676,
      "learning_rate": 6.737959236368063e-05,
      "loss": 11.7839,
      "step": 14126
    },
    {
      "epoch": 2.470295081967213,
      "grad_norm": 0.7042767739238045,
      "learning_rate": 6.736620641376568e-05,
      "loss": 11.9402,
      "step": 14127
    },
    {
      "epoch": 2.470469945355191,
      "grad_norm": 0.7340385430039887,
      "learning_rate": 6.735282111825849e-05,
      "loss": 11.9167,
      "step": 14128
    },
    {
      "epoch": 2.4706448087431694,
      "grad_norm": 0.8309836069790625,
      "learning_rate": 6.73394364774275e-05,
      "loss": 11.9606,
      "step": 14129
    },
    {
      "epoch": 2.4708196721311477,
      "grad_norm": 0.6583552279225462,
      "learning_rate": 6.732605249154102e-05,
      "loss": 11.8075,
      "step": 14130
    },
    {
      "epoch": 2.4709945355191256,
      "grad_norm": 0.7487210340021166,
      "learning_rate": 6.731266916086757e-05,
      "loss": 11.9104,
      "step": 14131
    },
    {
      "epoch": 2.471169398907104,
      "grad_norm": 0.7227451045332517,
      "learning_rate": 6.729928648567545e-05,
      "loss": 12.0308,
      "step": 14132
    },
    {
      "epoch": 2.471344262295082,
      "grad_norm": 0.6770796378066981,
      "learning_rate": 6.728590446623305e-05,
      "loss": 11.7514,
      "step": 14133
    },
    {
      "epoch": 2.47151912568306,
      "grad_norm": 0.6989250808916333,
      "learning_rate": 6.727252310280867e-05,
      "loss": 11.8619,
      "step": 14134
    },
    {
      "epoch": 2.471693989071038,
      "grad_norm": 0.6564406658350402,
      "learning_rate": 6.725914239567072e-05,
      "loss": 11.8426,
      "step": 14135
    },
    {
      "epoch": 2.4718688524590164,
      "grad_norm": 0.724558836053279,
      "learning_rate": 6.724576234508751e-05,
      "loss": 11.9449,
      "step": 14136
    },
    {
      "epoch": 2.4720437158469943,
      "grad_norm": 0.6279504032112205,
      "learning_rate": 6.723238295132732e-05,
      "loss": 11.7237,
      "step": 14137
    },
    {
      "epoch": 2.4722185792349727,
      "grad_norm": 0.638330495931262,
      "learning_rate": 6.721900421465847e-05,
      "loss": 11.9298,
      "step": 14138
    },
    {
      "epoch": 2.472393442622951,
      "grad_norm": 0.6471515653227098,
      "learning_rate": 6.72056261353492e-05,
      "loss": 11.8646,
      "step": 14139
    },
    {
      "epoch": 2.472568306010929,
      "grad_norm": 0.6938533670376353,
      "learning_rate": 6.719224871366786e-05,
      "loss": 11.7652,
      "step": 14140
    },
    {
      "epoch": 2.4727431693989073,
      "grad_norm": 0.6934665292318951,
      "learning_rate": 6.717887194988264e-05,
      "loss": 12.0731,
      "step": 14141
    },
    {
      "epoch": 2.472918032786885,
      "grad_norm": 0.6119922524467823,
      "learning_rate": 6.716549584426182e-05,
      "loss": 11.9361,
      "step": 14142
    },
    {
      "epoch": 2.4730928961748635,
      "grad_norm": 0.6553324058227863,
      "learning_rate": 6.715212039707364e-05,
      "loss": 11.9348,
      "step": 14143
    },
    {
      "epoch": 2.4732677595628414,
      "grad_norm": 0.6441057642270849,
      "learning_rate": 6.713874560858625e-05,
      "loss": 11.9681,
      "step": 14144
    },
    {
      "epoch": 2.4734426229508197,
      "grad_norm": 0.6150665413502171,
      "learning_rate": 6.712537147906797e-05,
      "loss": 11.8565,
      "step": 14145
    },
    {
      "epoch": 2.4736174863387976,
      "grad_norm": 0.6324010335470039,
      "learning_rate": 6.711199800878689e-05,
      "loss": 11.7256,
      "step": 14146
    },
    {
      "epoch": 2.473792349726776,
      "grad_norm": 0.6855582250226234,
      "learning_rate": 6.709862519801127e-05,
      "loss": 12.087,
      "step": 14147
    },
    {
      "epoch": 2.4739672131147543,
      "grad_norm": 0.6570048069995919,
      "learning_rate": 6.708525304700921e-05,
      "loss": 11.9774,
      "step": 14148
    },
    {
      "epoch": 2.474142076502732,
      "grad_norm": 0.6577896238657501,
      "learning_rate": 6.707188155604887e-05,
      "loss": 11.9858,
      "step": 14149
    },
    {
      "epoch": 2.4743169398907106,
      "grad_norm": 0.6831717431038375,
      "learning_rate": 6.705851072539842e-05,
      "loss": 11.8216,
      "step": 14150
    },
    {
      "epoch": 2.4744918032786885,
      "grad_norm": 0.6947419094341757,
      "learning_rate": 6.704514055532597e-05,
      "loss": 11.7877,
      "step": 14151
    },
    {
      "epoch": 2.474666666666667,
      "grad_norm": 0.7387531165231034,
      "learning_rate": 6.703177104609966e-05,
      "loss": 11.5153,
      "step": 14152
    },
    {
      "epoch": 2.4748415300546447,
      "grad_norm": 0.7377986581159046,
      "learning_rate": 6.701840219798756e-05,
      "loss": 11.8265,
      "step": 14153
    },
    {
      "epoch": 2.475016393442623,
      "grad_norm": 0.6272513597755289,
      "learning_rate": 6.700503401125773e-05,
      "loss": 11.9387,
      "step": 14154
    },
    {
      "epoch": 2.475191256830601,
      "grad_norm": 0.7448269818123157,
      "learning_rate": 6.69916664861783e-05,
      "loss": 12.0093,
      "step": 14155
    },
    {
      "epoch": 2.4753661202185793,
      "grad_norm": 0.7367536112202078,
      "learning_rate": 6.697829962301731e-05,
      "loss": 11.7001,
      "step": 14156
    },
    {
      "epoch": 2.4755409836065576,
      "grad_norm": 0.6865407393532234,
      "learning_rate": 6.69649334220428e-05,
      "loss": 11.9665,
      "step": 14157
    },
    {
      "epoch": 2.4757158469945355,
      "grad_norm": 0.6798058236217263,
      "learning_rate": 6.695156788352279e-05,
      "loss": 12.0223,
      "step": 14158
    },
    {
      "epoch": 2.475890710382514,
      "grad_norm": 0.7177058166404506,
      "learning_rate": 6.693820300772534e-05,
      "loss": 11.7674,
      "step": 14159
    },
    {
      "epoch": 2.4760655737704917,
      "grad_norm": 0.6554669015787957,
      "learning_rate": 6.692483879491841e-05,
      "loss": 11.7701,
      "step": 14160
    },
    {
      "epoch": 2.47624043715847,
      "grad_norm": 0.6892391593610583,
      "learning_rate": 6.691147524537005e-05,
      "loss": 11.9991,
      "step": 14161
    },
    {
      "epoch": 2.476415300546448,
      "grad_norm": 0.6918663545295147,
      "learning_rate": 6.689811235934818e-05,
      "loss": 11.8501,
      "step": 14162
    },
    {
      "epoch": 2.4765901639344263,
      "grad_norm": 0.636844856187623,
      "learning_rate": 6.688475013712077e-05,
      "loss": 11.9876,
      "step": 14163
    },
    {
      "epoch": 2.4767650273224042,
      "grad_norm": 0.6411958797771907,
      "learning_rate": 6.687138857895583e-05,
      "loss": 11.5558,
      "step": 14164
    },
    {
      "epoch": 2.4769398907103826,
      "grad_norm": 0.7062074242815399,
      "learning_rate": 6.685802768512129e-05,
      "loss": 12.0644,
      "step": 14165
    },
    {
      "epoch": 2.477114754098361,
      "grad_norm": 0.6893527876430883,
      "learning_rate": 6.684466745588502e-05,
      "loss": 11.8291,
      "step": 14166
    },
    {
      "epoch": 2.477289617486339,
      "grad_norm": 0.6787193225432483,
      "learning_rate": 6.683130789151497e-05,
      "loss": 12.0117,
      "step": 14167
    },
    {
      "epoch": 2.4774644808743167,
      "grad_norm": 0.6689408548297244,
      "learning_rate": 6.681794899227903e-05,
      "loss": 12.0616,
      "step": 14168
    },
    {
      "epoch": 2.477639344262295,
      "grad_norm": 0.6712861574735658,
      "learning_rate": 6.68045907584451e-05,
      "loss": 12.0365,
      "step": 14169
    },
    {
      "epoch": 2.4778142076502734,
      "grad_norm": 0.6635397819149805,
      "learning_rate": 6.679123319028108e-05,
      "loss": 11.9739,
      "step": 14170
    },
    {
      "epoch": 2.4779890710382513,
      "grad_norm": 0.6845255752574583,
      "learning_rate": 6.677787628805476e-05,
      "loss": 11.6059,
      "step": 14171
    },
    {
      "epoch": 2.4781639344262296,
      "grad_norm": 0.695959701844703,
      "learning_rate": 6.676452005203406e-05,
      "loss": 11.7761,
      "step": 14172
    },
    {
      "epoch": 2.4783387978142075,
      "grad_norm": 0.5975428623348715,
      "learning_rate": 6.675116448248672e-05,
      "loss": 11.8803,
      "step": 14173
    },
    {
      "epoch": 2.478513661202186,
      "grad_norm": 0.6525354793282436,
      "learning_rate": 6.673780957968069e-05,
      "loss": 11.9136,
      "step": 14174
    },
    {
      "epoch": 2.4786885245901638,
      "grad_norm": 0.6296083165177251,
      "learning_rate": 6.672445534388367e-05,
      "loss": 11.7995,
      "step": 14175
    },
    {
      "epoch": 2.478863387978142,
      "grad_norm": 0.6358937142236718,
      "learning_rate": 6.671110177536349e-05,
      "loss": 11.7403,
      "step": 14176
    },
    {
      "epoch": 2.47903825136612,
      "grad_norm": 0.5855687984786743,
      "learning_rate": 6.669774887438796e-05,
      "loss": 11.9494,
      "step": 14177
    },
    {
      "epoch": 2.4792131147540983,
      "grad_norm": 0.6364047603622495,
      "learning_rate": 6.668439664122475e-05,
      "loss": 11.839,
      "step": 14178
    },
    {
      "epoch": 2.4793879781420767,
      "grad_norm": 0.7260182260203623,
      "learning_rate": 6.667104507614175e-05,
      "loss": 11.8318,
      "step": 14179
    },
    {
      "epoch": 2.4795628415300546,
      "grad_norm": 0.7239183304678448,
      "learning_rate": 6.66576941794066e-05,
      "loss": 11.8116,
      "step": 14180
    },
    {
      "epoch": 2.479737704918033,
      "grad_norm": 0.6466812332082158,
      "learning_rate": 6.664434395128704e-05,
      "loss": 11.6604,
      "step": 14181
    },
    {
      "epoch": 2.479912568306011,
      "grad_norm": 0.6395780233044899,
      "learning_rate": 6.66309943920508e-05,
      "loss": 11.938,
      "step": 14182
    },
    {
      "epoch": 2.480087431693989,
      "grad_norm": 0.7355175127443533,
      "learning_rate": 6.66176455019656e-05,
      "loss": 11.8406,
      "step": 14183
    },
    {
      "epoch": 2.480262295081967,
      "grad_norm": 0.6283169102334285,
      "learning_rate": 6.660429728129911e-05,
      "loss": 11.8717,
      "step": 14184
    },
    {
      "epoch": 2.4804371584699454,
      "grad_norm": 0.6469074290879995,
      "learning_rate": 6.659094973031899e-05,
      "loss": 11.8054,
      "step": 14185
    },
    {
      "epoch": 2.4806120218579233,
      "grad_norm": 0.7081307088397137,
      "learning_rate": 6.657760284929293e-05,
      "loss": 12.077,
      "step": 14186
    },
    {
      "epoch": 2.4807868852459016,
      "grad_norm": 0.6101588673172984,
      "learning_rate": 6.656425663848848e-05,
      "loss": 12.0252,
      "step": 14187
    },
    {
      "epoch": 2.48096174863388,
      "grad_norm": 0.7799848955968129,
      "learning_rate": 6.655091109817342e-05,
      "loss": 11.7647,
      "step": 14188
    },
    {
      "epoch": 2.481136612021858,
      "grad_norm": 0.608833448583788,
      "learning_rate": 6.653756622861525e-05,
      "loss": 11.942,
      "step": 14189
    },
    {
      "epoch": 2.481311475409836,
      "grad_norm": 0.6449563174022227,
      "learning_rate": 6.652422203008165e-05,
      "loss": 11.9638,
      "step": 14190
    },
    {
      "epoch": 2.481486338797814,
      "grad_norm": 0.7086569813747565,
      "learning_rate": 6.651087850284019e-05,
      "loss": 12.0144,
      "step": 14191
    },
    {
      "epoch": 2.4816612021857924,
      "grad_norm": 0.6792859126316573,
      "learning_rate": 6.64975356471584e-05,
      "loss": 11.9079,
      "step": 14192
    },
    {
      "epoch": 2.4818360655737703,
      "grad_norm": 0.6547188359060098,
      "learning_rate": 6.64841934633039e-05,
      "loss": 11.9644,
      "step": 14193
    },
    {
      "epoch": 2.4820109289617487,
      "grad_norm": 0.7298247000404345,
      "learning_rate": 6.647085195154423e-05,
      "loss": 11.7772,
      "step": 14194
    },
    {
      "epoch": 2.4821857923497266,
      "grad_norm": 0.6138856641401971,
      "learning_rate": 6.645751111214695e-05,
      "loss": 12.0241,
      "step": 14195
    },
    {
      "epoch": 2.482360655737705,
      "grad_norm": 0.726384604053508,
      "learning_rate": 6.644417094537956e-05,
      "loss": 11.6836,
      "step": 14196
    },
    {
      "epoch": 2.4825355191256833,
      "grad_norm": 0.7615647870493009,
      "learning_rate": 6.643083145150953e-05,
      "loss": 11.9334,
      "step": 14197
    },
    {
      "epoch": 2.482710382513661,
      "grad_norm": 0.6825771351737335,
      "learning_rate": 6.641749263080442e-05,
      "loss": 11.907,
      "step": 14198
    },
    {
      "epoch": 2.4828852459016395,
      "grad_norm": 0.7839072351900914,
      "learning_rate": 6.64041544835317e-05,
      "loss": 11.6338,
      "step": 14199
    },
    {
      "epoch": 2.4830601092896174,
      "grad_norm": 0.7479086565970291,
      "learning_rate": 6.639081700995887e-05,
      "loss": 11.744,
      "step": 14200
    },
    {
      "epoch": 2.4832349726775957,
      "grad_norm": 0.7233041003566919,
      "learning_rate": 6.63774802103533e-05,
      "loss": 11.5961,
      "step": 14201
    },
    {
      "epoch": 2.4834098360655736,
      "grad_norm": 0.7008221161485859,
      "learning_rate": 6.636414408498249e-05,
      "loss": 11.904,
      "step": 14202
    },
    {
      "epoch": 2.483584699453552,
      "grad_norm": 0.6843033874374741,
      "learning_rate": 6.635080863411388e-05,
      "loss": 11.7552,
      "step": 14203
    },
    {
      "epoch": 2.48375956284153,
      "grad_norm": 0.6846817627739558,
      "learning_rate": 6.63374738580149e-05,
      "loss": 11.8555,
      "step": 14204
    },
    {
      "epoch": 2.483934426229508,
      "grad_norm": 0.6680822402698035,
      "learning_rate": 6.63241397569529e-05,
      "loss": 11.8409,
      "step": 14205
    },
    {
      "epoch": 2.4841092896174866,
      "grad_norm": 0.6799899249575843,
      "learning_rate": 6.631080633119527e-05,
      "loss": 12.072,
      "step": 14206
    },
    {
      "epoch": 2.4842841530054645,
      "grad_norm": 0.6915093838696484,
      "learning_rate": 6.629747358100944e-05,
      "loss": 11.9327,
      "step": 14207
    },
    {
      "epoch": 2.484459016393443,
      "grad_norm": 0.6421216481469267,
      "learning_rate": 6.628414150666276e-05,
      "loss": 12.0516,
      "step": 14208
    },
    {
      "epoch": 2.4846338797814207,
      "grad_norm": 0.639627817263961,
      "learning_rate": 6.627081010842255e-05,
      "loss": 11.8126,
      "step": 14209
    },
    {
      "epoch": 2.484808743169399,
      "grad_norm": 0.7400886153243984,
      "learning_rate": 6.625747938655618e-05,
      "loss": 11.8854,
      "step": 14210
    },
    {
      "epoch": 2.484983606557377,
      "grad_norm": 0.6907583263880065,
      "learning_rate": 6.62441493413309e-05,
      "loss": 11.8845,
      "step": 14211
    },
    {
      "epoch": 2.4851584699453553,
      "grad_norm": 0.6881058738164837,
      "learning_rate": 6.62308199730141e-05,
      "loss": 12.0035,
      "step": 14212
    },
    {
      "epoch": 2.485333333333333,
      "grad_norm": 0.7044946027676807,
      "learning_rate": 6.621749128187307e-05,
      "loss": 11.7043,
      "step": 14213
    },
    {
      "epoch": 2.4855081967213115,
      "grad_norm": 0.6995704984918588,
      "learning_rate": 6.620416326817504e-05,
      "loss": 11.8611,
      "step": 14214
    },
    {
      "epoch": 2.48568306010929,
      "grad_norm": 0.655327131060553,
      "learning_rate": 6.61908359321873e-05,
      "loss": 11.966,
      "step": 14215
    },
    {
      "epoch": 2.4858579234972678,
      "grad_norm": 0.8589696000626913,
      "learning_rate": 6.61775092741771e-05,
      "loss": 11.9586,
      "step": 14216
    },
    {
      "epoch": 2.486032786885246,
      "grad_norm": 0.6704584420337806,
      "learning_rate": 6.616418329441168e-05,
      "loss": 11.9059,
      "step": 14217
    },
    {
      "epoch": 2.486207650273224,
      "grad_norm": 0.744570335003329,
      "learning_rate": 6.615085799315833e-05,
      "loss": 11.9937,
      "step": 14218
    },
    {
      "epoch": 2.4863825136612023,
      "grad_norm": 0.7525227688366336,
      "learning_rate": 6.613753337068416e-05,
      "loss": 11.8589,
      "step": 14219
    },
    {
      "epoch": 2.4865573770491802,
      "grad_norm": 0.6797384863056015,
      "learning_rate": 6.612420942725644e-05,
      "loss": 11.9719,
      "step": 14220
    },
    {
      "epoch": 2.4867322404371586,
      "grad_norm": 0.6617878827683029,
      "learning_rate": 6.61108861631423e-05,
      "loss": 12.0163,
      "step": 14221
    },
    {
      "epoch": 2.4869071038251365,
      "grad_norm": 0.786659063548536,
      "learning_rate": 6.6097563578609e-05,
      "loss": 12.0174,
      "step": 14222
    },
    {
      "epoch": 2.487081967213115,
      "grad_norm": 0.6990684753326258,
      "learning_rate": 6.60842416739236e-05,
      "loss": 11.8744,
      "step": 14223
    },
    {
      "epoch": 2.4872568306010927,
      "grad_norm": 0.6882910131363458,
      "learning_rate": 6.60709204493533e-05,
      "loss": 11.8964,
      "step": 14224
    },
    {
      "epoch": 2.487431693989071,
      "grad_norm": 0.6956685537879761,
      "learning_rate": 6.605759990516525e-05,
      "loss": 11.8615,
      "step": 14225
    },
    {
      "epoch": 2.487606557377049,
      "grad_norm": 0.6269411330449733,
      "learning_rate": 6.604428004162646e-05,
      "loss": 11.5985,
      "step": 14226
    },
    {
      "epoch": 2.4877814207650273,
      "grad_norm": 0.7259296567930198,
      "learning_rate": 6.603096085900418e-05,
      "loss": 11.7543,
      "step": 14227
    },
    {
      "epoch": 2.4879562841530056,
      "grad_norm": 0.6411418104282,
      "learning_rate": 6.601764235756543e-05,
      "loss": 11.9611,
      "step": 14228
    },
    {
      "epoch": 2.4881311475409835,
      "grad_norm": 0.6356765115524855,
      "learning_rate": 6.600432453757727e-05,
      "loss": 11.805,
      "step": 14229
    },
    {
      "epoch": 2.488306010928962,
      "grad_norm": 0.6795980702528205,
      "learning_rate": 6.599100739930677e-05,
      "loss": 11.7047,
      "step": 14230
    },
    {
      "epoch": 2.4884808743169398,
      "grad_norm": 0.6660109864723774,
      "learning_rate": 6.597769094302103e-05,
      "loss": 11.6974,
      "step": 14231
    },
    {
      "epoch": 2.488655737704918,
      "grad_norm": 0.6222995298709892,
      "learning_rate": 6.596437516898703e-05,
      "loss": 11.8239,
      "step": 14232
    },
    {
      "epoch": 2.488830601092896,
      "grad_norm": 0.783056657180629,
      "learning_rate": 6.595106007747183e-05,
      "loss": 11.9371,
      "step": 14233
    },
    {
      "epoch": 2.4890054644808743,
      "grad_norm": 0.7107304612114903,
      "learning_rate": 6.593774566874241e-05,
      "loss": 11.7895,
      "step": 14234
    },
    {
      "epoch": 2.4891803278688522,
      "grad_norm": 0.7012767707145546,
      "learning_rate": 6.592443194306572e-05,
      "loss": 11.8686,
      "step": 14235
    },
    {
      "epoch": 2.4893551912568306,
      "grad_norm": 0.6884757355903375,
      "learning_rate": 6.591111890070887e-05,
      "loss": 12.0754,
      "step": 14236
    },
    {
      "epoch": 2.489530054644809,
      "grad_norm": 0.6433583229758617,
      "learning_rate": 6.589780654193873e-05,
      "loss": 11.8669,
      "step": 14237
    },
    {
      "epoch": 2.489704918032787,
      "grad_norm": 0.5980296087566311,
      "learning_rate": 6.588449486702226e-05,
      "loss": 12.0157,
      "step": 14238
    },
    {
      "epoch": 2.489879781420765,
      "grad_norm": 0.6867754200103536,
      "learning_rate": 6.587118387622647e-05,
      "loss": 11.7731,
      "step": 14239
    },
    {
      "epoch": 2.490054644808743,
      "grad_norm": 0.7156707691719375,
      "learning_rate": 6.585787356981816e-05,
      "loss": 11.8913,
      "step": 14240
    },
    {
      "epoch": 2.4902295081967214,
      "grad_norm": 0.7545275906889839,
      "learning_rate": 6.584456394806434e-05,
      "loss": 11.796,
      "step": 14241
    },
    {
      "epoch": 2.4904043715846993,
      "grad_norm": 0.7139204377375112,
      "learning_rate": 6.583125501123189e-05,
      "loss": 11.6445,
      "step": 14242
    },
    {
      "epoch": 2.4905792349726776,
      "grad_norm": 0.6387941514213974,
      "learning_rate": 6.581794675958772e-05,
      "loss": 11.6972,
      "step": 14243
    },
    {
      "epoch": 2.4907540983606555,
      "grad_norm": 0.6523671555231235,
      "learning_rate": 6.580463919339863e-05,
      "loss": 12.0228,
      "step": 14244
    },
    {
      "epoch": 2.490928961748634,
      "grad_norm": 0.722802907427597,
      "learning_rate": 6.57913323129315e-05,
      "loss": 11.8487,
      "step": 14245
    },
    {
      "epoch": 2.491103825136612,
      "grad_norm": 0.6723376615718727,
      "learning_rate": 6.577802611845323e-05,
      "loss": 11.8342,
      "step": 14246
    },
    {
      "epoch": 2.49127868852459,
      "grad_norm": 0.6394416226723153,
      "learning_rate": 6.576472061023059e-05,
      "loss": 11.6355,
      "step": 14247
    },
    {
      "epoch": 2.4914535519125685,
      "grad_norm": 0.7049270292306483,
      "learning_rate": 6.575141578853045e-05,
      "loss": 11.7244,
      "step": 14248
    },
    {
      "epoch": 2.4916284153005464,
      "grad_norm": 0.7092092836330545,
      "learning_rate": 6.573811165361956e-05,
      "loss": 11.8727,
      "step": 14249
    },
    {
      "epoch": 2.4918032786885247,
      "grad_norm": 0.6357354907111775,
      "learning_rate": 6.572480820576469e-05,
      "loss": 11.7635,
      "step": 14250
    },
    {
      "epoch": 2.4919781420765026,
      "grad_norm": 0.6726749314725885,
      "learning_rate": 6.57115054452327e-05,
      "loss": 11.7425,
      "step": 14251
    },
    {
      "epoch": 2.492153005464481,
      "grad_norm": 0.6558644118497237,
      "learning_rate": 6.569820337229033e-05,
      "loss": 12.0454,
      "step": 14252
    },
    {
      "epoch": 2.492327868852459,
      "grad_norm": 0.6226472646886227,
      "learning_rate": 6.568490198720426e-05,
      "loss": 11.9617,
      "step": 14253
    },
    {
      "epoch": 2.492502732240437,
      "grad_norm": 0.6596025620454851,
      "learning_rate": 6.567160129024126e-05,
      "loss": 11.8234,
      "step": 14254
    },
    {
      "epoch": 2.4926775956284155,
      "grad_norm": 0.6789728134479681,
      "learning_rate": 6.565830128166807e-05,
      "loss": 11.6891,
      "step": 14255
    },
    {
      "epoch": 2.4928524590163934,
      "grad_norm": 0.6646015692523319,
      "learning_rate": 6.564500196175138e-05,
      "loss": 11.6908,
      "step": 14256
    },
    {
      "epoch": 2.4930273224043717,
      "grad_norm": 0.6449605598953619,
      "learning_rate": 6.563170333075791e-05,
      "loss": 11.96,
      "step": 14257
    },
    {
      "epoch": 2.4932021857923496,
      "grad_norm": 0.6656761621657246,
      "learning_rate": 6.561840538895429e-05,
      "loss": 11.9644,
      "step": 14258
    },
    {
      "epoch": 2.493377049180328,
      "grad_norm": 0.5931429519175204,
      "learning_rate": 6.560510813660719e-05,
      "loss": 11.9874,
      "step": 14259
    },
    {
      "epoch": 2.493551912568306,
      "grad_norm": 0.741536246250961,
      "learning_rate": 6.559181157398329e-05,
      "loss": 11.98,
      "step": 14260
    },
    {
      "epoch": 2.4937267759562842,
      "grad_norm": 0.6828350425081202,
      "learning_rate": 6.557851570134925e-05,
      "loss": 11.7122,
      "step": 14261
    },
    {
      "epoch": 2.493901639344262,
      "grad_norm": 0.6664846100261447,
      "learning_rate": 6.556522051897163e-05,
      "loss": 11.7132,
      "step": 14262
    },
    {
      "epoch": 2.4940765027322405,
      "grad_norm": 0.7720515808261763,
      "learning_rate": 6.555192602711706e-05,
      "loss": 11.9951,
      "step": 14263
    },
    {
      "epoch": 2.494251366120219,
      "grad_norm": 0.698230343728968,
      "learning_rate": 6.553863222605212e-05,
      "loss": 11.8258,
      "step": 14264
    },
    {
      "epoch": 2.4944262295081967,
      "grad_norm": 0.7615404232534402,
      "learning_rate": 6.552533911604345e-05,
      "loss": 11.926,
      "step": 14265
    },
    {
      "epoch": 2.494601092896175,
      "grad_norm": 0.783335647192449,
      "learning_rate": 6.55120466973576e-05,
      "loss": 11.8986,
      "step": 14266
    },
    {
      "epoch": 2.494775956284153,
      "grad_norm": 0.7778081810546106,
      "learning_rate": 6.549875497026107e-05,
      "loss": 11.9156,
      "step": 14267
    },
    {
      "epoch": 2.4949508196721313,
      "grad_norm": 0.7573447628403208,
      "learning_rate": 6.548546393502045e-05,
      "loss": 11.6861,
      "step": 14268
    },
    {
      "epoch": 2.495125683060109,
      "grad_norm": 0.8390057610493041,
      "learning_rate": 6.547217359190223e-05,
      "loss": 11.6774,
      "step": 14269
    },
    {
      "epoch": 2.4953005464480875,
      "grad_norm": 0.7462721662318179,
      "learning_rate": 6.545888394117299e-05,
      "loss": 11.7743,
      "step": 14270
    },
    {
      "epoch": 2.4954754098360654,
      "grad_norm": 0.6877078525453996,
      "learning_rate": 6.544559498309915e-05,
      "loss": 11.907,
      "step": 14271
    },
    {
      "epoch": 2.4956502732240438,
      "grad_norm": 0.8577321944645959,
      "learning_rate": 6.543230671794722e-05,
      "loss": 12.1561,
      "step": 14272
    },
    {
      "epoch": 2.495825136612022,
      "grad_norm": 0.6778096859927714,
      "learning_rate": 6.541901914598371e-05,
      "loss": 11.6911,
      "step": 14273
    },
    {
      "epoch": 2.496,
      "grad_norm": 0.6932832875433244,
      "learning_rate": 6.540573226747498e-05,
      "loss": 12.0119,
      "step": 14274
    },
    {
      "epoch": 2.496174863387978,
      "grad_norm": 0.7653146616841904,
      "learning_rate": 6.53924460826876e-05,
      "loss": 11.7591,
      "step": 14275
    },
    {
      "epoch": 2.4963497267759562,
      "grad_norm": 0.7040049475572216,
      "learning_rate": 6.537916059188792e-05,
      "loss": 11.7368,
      "step": 14276
    },
    {
      "epoch": 2.4965245901639346,
      "grad_norm": 0.6773535665943329,
      "learning_rate": 6.536587579534236e-05,
      "loss": 11.8027,
      "step": 14277
    },
    {
      "epoch": 2.4966994535519125,
      "grad_norm": 0.6084975264428976,
      "learning_rate": 6.535259169331731e-05,
      "loss": 12.0222,
      "step": 14278
    },
    {
      "epoch": 2.496874316939891,
      "grad_norm": 0.6394241694746214,
      "learning_rate": 6.533930828607922e-05,
      "loss": 11.746,
      "step": 14279
    },
    {
      "epoch": 2.4970491803278687,
      "grad_norm": 0.6237537585460337,
      "learning_rate": 6.53260255738944e-05,
      "loss": 12.0234,
      "step": 14280
    },
    {
      "epoch": 2.497224043715847,
      "grad_norm": 0.6885190410099615,
      "learning_rate": 6.531274355702924e-05,
      "loss": 11.5842,
      "step": 14281
    },
    {
      "epoch": 2.497398907103825,
      "grad_norm": 0.7367255885456047,
      "learning_rate": 6.529946223575008e-05,
      "loss": 12.0747,
      "step": 14282
    },
    {
      "epoch": 2.4975737704918033,
      "grad_norm": 0.6009074888684851,
      "learning_rate": 6.52861816103232e-05,
      "loss": 11.557,
      "step": 14283
    },
    {
      "epoch": 2.497748633879781,
      "grad_norm": 0.6757463219264925,
      "learning_rate": 6.527290168101502e-05,
      "loss": 11.8849,
      "step": 14284
    },
    {
      "epoch": 2.4979234972677595,
      "grad_norm": 0.7157996722402467,
      "learning_rate": 6.525962244809176e-05,
      "loss": 11.852,
      "step": 14285
    },
    {
      "epoch": 2.498098360655738,
      "grad_norm": 0.6271682513641128,
      "learning_rate": 6.524634391181975e-05,
      "loss": 11.8377,
      "step": 14286
    },
    {
      "epoch": 2.4982732240437158,
      "grad_norm": 0.6067407813120218,
      "learning_rate": 6.523306607246527e-05,
      "loss": 11.5447,
      "step": 14287
    },
    {
      "epoch": 2.498448087431694,
      "grad_norm": 0.7524832355816524,
      "learning_rate": 6.521978893029452e-05,
      "loss": 12.0465,
      "step": 14288
    },
    {
      "epoch": 2.498622950819672,
      "grad_norm": 0.6523990560432938,
      "learning_rate": 6.52065124855738e-05,
      "loss": 11.8025,
      "step": 14289
    },
    {
      "epoch": 2.4987978142076503,
      "grad_norm": 0.6588148170550334,
      "learning_rate": 6.519323673856936e-05,
      "loss": 12.0886,
      "step": 14290
    },
    {
      "epoch": 2.4989726775956282,
      "grad_norm": 0.7137642164022184,
      "learning_rate": 6.51799616895474e-05,
      "loss": 11.9234,
      "step": 14291
    },
    {
      "epoch": 2.4991475409836066,
      "grad_norm": 0.6915049830820644,
      "learning_rate": 6.516668733877409e-05,
      "loss": 11.7869,
      "step": 14292
    },
    {
      "epoch": 2.4993224043715845,
      "grad_norm": 0.6895329039728104,
      "learning_rate": 6.515341368651564e-05,
      "loss": 11.6571,
      "step": 14293
    },
    {
      "epoch": 2.499497267759563,
      "grad_norm": 0.6100528204275982,
      "learning_rate": 6.514014073303826e-05,
      "loss": 11.7648,
      "step": 14294
    },
    {
      "epoch": 2.499672131147541,
      "grad_norm": 0.6753617132809088,
      "learning_rate": 6.51268684786081e-05,
      "loss": 11.785,
      "step": 14295
    },
    {
      "epoch": 2.499846994535519,
      "grad_norm": 0.657404764033412,
      "learning_rate": 6.511359692349132e-05,
      "loss": 12.0303,
      "step": 14296
    },
    {
      "epoch": 2.5000218579234974,
      "grad_norm": 0.6892544562297858,
      "learning_rate": 6.510032606795399e-05,
      "loss": 11.8618,
      "step": 14297
    },
    {
      "epoch": 2.5001967213114753,
      "grad_norm": 0.5939803050910446,
      "learning_rate": 6.50870559122623e-05,
      "loss": 12.0275,
      "step": 14298
    },
    {
      "epoch": 2.5003715846994536,
      "grad_norm": 0.7166288931182901,
      "learning_rate": 6.507378645668234e-05,
      "loss": 11.9077,
      "step": 14299
    },
    {
      "epoch": 2.5005464480874315,
      "grad_norm": 0.6700931809051893,
      "learning_rate": 6.50605177014802e-05,
      "loss": 11.6745,
      "step": 14300
    },
    {
      "epoch": 2.50072131147541,
      "grad_norm": 0.6929623373125078,
      "learning_rate": 6.504724964692195e-05,
      "loss": 11.8713,
      "step": 14301
    },
    {
      "epoch": 2.5008961748633878,
      "grad_norm": 0.6389017990468437,
      "learning_rate": 6.503398229327364e-05,
      "loss": 11.8912,
      "step": 14302
    },
    {
      "epoch": 2.501071038251366,
      "grad_norm": 0.7059046485874717,
      "learning_rate": 6.502071564080138e-05,
      "loss": 11.584,
      "step": 14303
    },
    {
      "epoch": 2.5012459016393445,
      "grad_norm": 0.7071097492196579,
      "learning_rate": 6.500744968977116e-05,
      "loss": 11.957,
      "step": 14304
    },
    {
      "epoch": 2.5014207650273224,
      "grad_norm": 0.7203724508807222,
      "learning_rate": 6.499418444044905e-05,
      "loss": 12.0122,
      "step": 14305
    },
    {
      "epoch": 2.5015956284153007,
      "grad_norm": 0.6729675535358131,
      "learning_rate": 6.498091989310097e-05,
      "loss": 11.9172,
      "step": 14306
    },
    {
      "epoch": 2.5017704918032786,
      "grad_norm": 0.7186516048860162,
      "learning_rate": 6.496765604799298e-05,
      "loss": 11.8544,
      "step": 14307
    },
    {
      "epoch": 2.501945355191257,
      "grad_norm": 0.7113374149590364,
      "learning_rate": 6.495439290539106e-05,
      "loss": 11.7654,
      "step": 14308
    },
    {
      "epoch": 2.502120218579235,
      "grad_norm": 0.6775972125640892,
      "learning_rate": 6.494113046556118e-05,
      "loss": 11.8,
      "step": 14309
    },
    {
      "epoch": 2.502295081967213,
      "grad_norm": 0.8093802263664592,
      "learning_rate": 6.492786872876927e-05,
      "loss": 12.0677,
      "step": 14310
    },
    {
      "epoch": 2.502469945355191,
      "grad_norm": 0.6854699327335921,
      "learning_rate": 6.491460769528129e-05,
      "loss": 11.7966,
      "step": 14311
    },
    {
      "epoch": 2.5026448087431694,
      "grad_norm": 0.7229991141342862,
      "learning_rate": 6.49013473653631e-05,
      "loss": 11.5852,
      "step": 14312
    },
    {
      "epoch": 2.5028196721311478,
      "grad_norm": 0.6236542787081729,
      "learning_rate": 6.48880877392807e-05,
      "loss": 11.7007,
      "step": 14313
    },
    {
      "epoch": 2.5029945355191257,
      "grad_norm": 0.6401117325324888,
      "learning_rate": 6.487482881729996e-05,
      "loss": 11.6031,
      "step": 14314
    },
    {
      "epoch": 2.5031693989071035,
      "grad_norm": 0.7620090851415671,
      "learning_rate": 6.486157059968674e-05,
      "loss": 11.8249,
      "step": 14315
    },
    {
      "epoch": 2.503344262295082,
      "grad_norm": 0.6313535748829622,
      "learning_rate": 6.484831308670691e-05,
      "loss": 11.92,
      "step": 14316
    },
    {
      "epoch": 2.5035191256830602,
      "grad_norm": 0.670366257688551,
      "learning_rate": 6.483505627862632e-05,
      "loss": 11.8176,
      "step": 14317
    },
    {
      "epoch": 2.503693989071038,
      "grad_norm": 0.6155748928399313,
      "learning_rate": 6.482180017571087e-05,
      "loss": 11.8884,
      "step": 14318
    },
    {
      "epoch": 2.5038688524590165,
      "grad_norm": 0.6332644413812084,
      "learning_rate": 6.48085447782263e-05,
      "loss": 11.7519,
      "step": 14319
    },
    {
      "epoch": 2.5040437158469944,
      "grad_norm": 0.7056190261681317,
      "learning_rate": 6.479529008643846e-05,
      "loss": 11.9189,
      "step": 14320
    },
    {
      "epoch": 2.5042185792349727,
      "grad_norm": 0.7002391338015971,
      "learning_rate": 6.478203610061317e-05,
      "loss": 11.7634,
      "step": 14321
    },
    {
      "epoch": 2.504393442622951,
      "grad_norm": 0.6581398654502896,
      "learning_rate": 6.476878282101614e-05,
      "loss": 11.9842,
      "step": 14322
    },
    {
      "epoch": 2.504568306010929,
      "grad_norm": 0.6996318869873761,
      "learning_rate": 6.475553024791324e-05,
      "loss": 11.7914,
      "step": 14323
    },
    {
      "epoch": 2.504743169398907,
      "grad_norm": 0.6334142631221639,
      "learning_rate": 6.474227838157015e-05,
      "loss": 11.4994,
      "step": 14324
    },
    {
      "epoch": 2.504918032786885,
      "grad_norm": 0.6627139810886528,
      "learning_rate": 6.472902722225265e-05,
      "loss": 11.6973,
      "step": 14325
    },
    {
      "epoch": 2.5050928961748635,
      "grad_norm": 0.6645116870037898,
      "learning_rate": 6.471577677022641e-05,
      "loss": 11.8747,
      "step": 14326
    },
    {
      "epoch": 2.5052677595628414,
      "grad_norm": 0.676657050627423,
      "learning_rate": 6.470252702575723e-05,
      "loss": 11.5731,
      "step": 14327
    },
    {
      "epoch": 2.5054426229508198,
      "grad_norm": 0.6460060171372148,
      "learning_rate": 6.468927798911075e-05,
      "loss": 12.0197,
      "step": 14328
    },
    {
      "epoch": 2.5056174863387977,
      "grad_norm": 0.6412672091955001,
      "learning_rate": 6.467602966055265e-05,
      "loss": 11.89,
      "step": 14329
    },
    {
      "epoch": 2.505792349726776,
      "grad_norm": 0.6264426892534061,
      "learning_rate": 6.466278204034863e-05,
      "loss": 11.771,
      "step": 14330
    },
    {
      "epoch": 2.5059672131147543,
      "grad_norm": 0.6805342089131707,
      "learning_rate": 6.46495351287643e-05,
      "loss": 11.7931,
      "step": 14331
    },
    {
      "epoch": 2.5061420765027322,
      "grad_norm": 0.649527658869758,
      "learning_rate": 6.463628892606537e-05,
      "loss": 12.0176,
      "step": 14332
    },
    {
      "epoch": 2.50631693989071,
      "grad_norm": 0.7040690329115179,
      "learning_rate": 6.46230434325174e-05,
      "loss": 11.8358,
      "step": 14333
    },
    {
      "epoch": 2.5064918032786885,
      "grad_norm": 0.6206016703341302,
      "learning_rate": 6.460979864838604e-05,
      "loss": 11.9658,
      "step": 14334
    },
    {
      "epoch": 2.506666666666667,
      "grad_norm": 0.6196715926816099,
      "learning_rate": 6.459655457393689e-05,
      "loss": 11.9267,
      "step": 14335
    },
    {
      "epoch": 2.5068415300546447,
      "grad_norm": 0.6074212172080664,
      "learning_rate": 6.45833112094355e-05,
      "loss": 12.0691,
      "step": 14336
    },
    {
      "epoch": 2.507016393442623,
      "grad_norm": 0.6136415204891893,
      "learning_rate": 6.457006855514748e-05,
      "loss": 11.6524,
      "step": 14337
    },
    {
      "epoch": 2.507191256830601,
      "grad_norm": 0.6105968786943712,
      "learning_rate": 6.455682661133836e-05,
      "loss": 11.8422,
      "step": 14338
    },
    {
      "epoch": 2.5073661202185793,
      "grad_norm": 0.6153309913257384,
      "learning_rate": 6.454358537827372e-05,
      "loss": 11.9249,
      "step": 14339
    },
    {
      "epoch": 2.5075409836065576,
      "grad_norm": 0.616734362433964,
      "learning_rate": 6.453034485621904e-05,
      "loss": 11.9097,
      "step": 14340
    },
    {
      "epoch": 2.5077158469945355,
      "grad_norm": 0.637006381223961,
      "learning_rate": 6.451710504543981e-05,
      "loss": 11.9061,
      "step": 14341
    },
    {
      "epoch": 2.5078907103825134,
      "grad_norm": 0.6584722851591249,
      "learning_rate": 6.450386594620161e-05,
      "loss": 12.0121,
      "step": 14342
    },
    {
      "epoch": 2.5080655737704918,
      "grad_norm": 0.6345822397040858,
      "learning_rate": 6.449062755876988e-05,
      "loss": 11.6491,
      "step": 14343
    },
    {
      "epoch": 2.50824043715847,
      "grad_norm": 0.583626624998533,
      "learning_rate": 6.447738988341013e-05,
      "loss": 12.0961,
      "step": 14344
    },
    {
      "epoch": 2.508415300546448,
      "grad_norm": 0.6097985050447005,
      "learning_rate": 6.446415292038775e-05,
      "loss": 11.7516,
      "step": 14345
    },
    {
      "epoch": 2.5085901639344264,
      "grad_norm": 0.6001401657759856,
      "learning_rate": 6.44509166699682e-05,
      "loss": 11.9489,
      "step": 14346
    },
    {
      "epoch": 2.5087650273224043,
      "grad_norm": 0.6173328271355191,
      "learning_rate": 6.443768113241693e-05,
      "loss": 11.7933,
      "step": 14347
    },
    {
      "epoch": 2.5089398907103826,
      "grad_norm": 0.6183662314073906,
      "learning_rate": 6.442444630799938e-05,
      "loss": 11.7627,
      "step": 14348
    },
    {
      "epoch": 2.509114754098361,
      "grad_norm": 0.6132080453827214,
      "learning_rate": 6.441121219698087e-05,
      "loss": 11.7934,
      "step": 14349
    },
    {
      "epoch": 2.509289617486339,
      "grad_norm": 0.626785376259208,
      "learning_rate": 6.439797879962681e-05,
      "loss": 11.6211,
      "step": 14350
    },
    {
      "epoch": 2.5094644808743167,
      "grad_norm": 0.694342147247149,
      "learning_rate": 6.438474611620262e-05,
      "loss": 12.1264,
      "step": 14351
    },
    {
      "epoch": 2.509639344262295,
      "grad_norm": 0.68994319853651,
      "learning_rate": 6.437151414697362e-05,
      "loss": 12.0903,
      "step": 14352
    },
    {
      "epoch": 2.5098142076502734,
      "grad_norm": 0.6676635241681346,
      "learning_rate": 6.435828289220518e-05,
      "loss": 11.8262,
      "step": 14353
    },
    {
      "epoch": 2.5099890710382513,
      "grad_norm": 0.6645192923629639,
      "learning_rate": 6.434505235216257e-05,
      "loss": 12.113,
      "step": 14354
    },
    {
      "epoch": 2.5101639344262296,
      "grad_norm": 0.6188819760588218,
      "learning_rate": 6.433182252711112e-05,
      "loss": 11.9414,
      "step": 14355
    },
    {
      "epoch": 2.5103387978142075,
      "grad_norm": 0.6028549081386,
      "learning_rate": 6.431859341731617e-05,
      "loss": 11.9671,
      "step": 14356
    },
    {
      "epoch": 2.510513661202186,
      "grad_norm": 0.7082963550817609,
      "learning_rate": 6.4305365023043e-05,
      "loss": 11.6875,
      "step": 14357
    },
    {
      "epoch": 2.510688524590164,
      "grad_norm": 0.6406228017996632,
      "learning_rate": 6.429213734455683e-05,
      "loss": 11.7702,
      "step": 14358
    },
    {
      "epoch": 2.510863387978142,
      "grad_norm": 0.6617943676071393,
      "learning_rate": 6.427891038212294e-05,
      "loss": 11.6462,
      "step": 14359
    },
    {
      "epoch": 2.51103825136612,
      "grad_norm": 0.7902337917356034,
      "learning_rate": 6.426568413600658e-05,
      "loss": 11.6698,
      "step": 14360
    },
    {
      "epoch": 2.5112131147540984,
      "grad_norm": 0.6768200334844684,
      "learning_rate": 6.425245860647298e-05,
      "loss": 11.8281,
      "step": 14361
    },
    {
      "epoch": 2.5113879781420767,
      "grad_norm": 0.6247167314053597,
      "learning_rate": 6.423923379378736e-05,
      "loss": 11.4754,
      "step": 14362
    },
    {
      "epoch": 2.5115628415300546,
      "grad_norm": 0.6653668878174439,
      "learning_rate": 6.422600969821487e-05,
      "loss": 11.7042,
      "step": 14363
    },
    {
      "epoch": 2.511737704918033,
      "grad_norm": 0.6767251265478846,
      "learning_rate": 6.421278632002073e-05,
      "loss": 11.888,
      "step": 14364
    },
    {
      "epoch": 2.511912568306011,
      "grad_norm": 0.698352020900902,
      "learning_rate": 6.419956365947009e-05,
      "loss": 11.954,
      "step": 14365
    },
    {
      "epoch": 2.512087431693989,
      "grad_norm": 0.7008361839491888,
      "learning_rate": 6.418634171682817e-05,
      "loss": 11.8071,
      "step": 14366
    },
    {
      "epoch": 2.512262295081967,
      "grad_norm": 0.6699494056715211,
      "learning_rate": 6.417312049236004e-05,
      "loss": 12.0237,
      "step": 14367
    },
    {
      "epoch": 2.5124371584699454,
      "grad_norm": 0.7304829870399262,
      "learning_rate": 6.415989998633083e-05,
      "loss": 11.8586,
      "step": 14368
    },
    {
      "epoch": 2.5126120218579233,
      "grad_norm": 0.660697863447798,
      "learning_rate": 6.41466801990057e-05,
      "loss": 11.8597,
      "step": 14369
    },
    {
      "epoch": 2.5127868852459017,
      "grad_norm": 0.6851630100295015,
      "learning_rate": 6.413346113064966e-05,
      "loss": 12.0845,
      "step": 14370
    },
    {
      "epoch": 2.51296174863388,
      "grad_norm": 0.8882114040889932,
      "learning_rate": 6.41202427815279e-05,
      "loss": 12.0445,
      "step": 14371
    },
    {
      "epoch": 2.513136612021858,
      "grad_norm": 0.7642327225973785,
      "learning_rate": 6.410702515190543e-05,
      "loss": 11.8384,
      "step": 14372
    },
    {
      "epoch": 2.513311475409836,
      "grad_norm": 0.6589791598566528,
      "learning_rate": 6.40938082420473e-05,
      "loss": 11.8985,
      "step": 14373
    },
    {
      "epoch": 2.513486338797814,
      "grad_norm": 0.700842366287596,
      "learning_rate": 6.408059205221854e-05,
      "loss": 11.9966,
      "step": 14374
    },
    {
      "epoch": 2.5136612021857925,
      "grad_norm": 0.6804246896220609,
      "learning_rate": 6.406737658268425e-05,
      "loss": 11.7076,
      "step": 14375
    },
    {
      "epoch": 2.5138360655737704,
      "grad_norm": 0.6670343632140766,
      "learning_rate": 6.405416183370936e-05,
      "loss": 11.4006,
      "step": 14376
    },
    {
      "epoch": 2.5140109289617487,
      "grad_norm": 0.6682329007688019,
      "learning_rate": 6.40409478055589e-05,
      "loss": 11.8333,
      "step": 14377
    },
    {
      "epoch": 2.5141857923497266,
      "grad_norm": 0.6594468964832062,
      "learning_rate": 6.402773449849785e-05,
      "loss": 11.6083,
      "step": 14378
    },
    {
      "epoch": 2.514360655737705,
      "grad_norm": 0.6666357108720058,
      "learning_rate": 6.401452191279112e-05,
      "loss": 11.9995,
      "step": 14379
    },
    {
      "epoch": 2.5145355191256833,
      "grad_norm": 0.6214645731499226,
      "learning_rate": 6.400131004870378e-05,
      "loss": 11.8945,
      "step": 14380
    },
    {
      "epoch": 2.514710382513661,
      "grad_norm": 0.6639736030739699,
      "learning_rate": 6.398809890650068e-05,
      "loss": 11.8699,
      "step": 14381
    },
    {
      "epoch": 2.514885245901639,
      "grad_norm": 0.6188064639433347,
      "learning_rate": 6.397488848644676e-05,
      "loss": 12.0355,
      "step": 14382
    },
    {
      "epoch": 2.5150601092896174,
      "grad_norm": 0.7087946730386121,
      "learning_rate": 6.396167878880696e-05,
      "loss": 11.601,
      "step": 14383
    },
    {
      "epoch": 2.5152349726775958,
      "grad_norm": 0.6394764672651364,
      "learning_rate": 6.394846981384613e-05,
      "loss": 11.6303,
      "step": 14384
    },
    {
      "epoch": 2.5154098360655737,
      "grad_norm": 0.5658114065124495,
      "learning_rate": 6.393526156182918e-05,
      "loss": 11.7371,
      "step": 14385
    },
    {
      "epoch": 2.515584699453552,
      "grad_norm": 0.6365517551410813,
      "learning_rate": 6.392205403302097e-05,
      "loss": 11.9503,
      "step": 14386
    },
    {
      "epoch": 2.51575956284153,
      "grad_norm": 0.6947567844676534,
      "learning_rate": 6.390884722768636e-05,
      "loss": 11.745,
      "step": 14387
    },
    {
      "epoch": 2.5159344262295082,
      "grad_norm": 0.6925547228010778,
      "learning_rate": 6.389564114609017e-05,
      "loss": 11.6162,
      "step": 14388
    },
    {
      "epoch": 2.5161092896174866,
      "grad_norm": 0.6520872603170814,
      "learning_rate": 6.38824357884972e-05,
      "loss": 11.8192,
      "step": 14389
    },
    {
      "epoch": 2.5162841530054645,
      "grad_norm": 0.640818191241075,
      "learning_rate": 6.386923115517232e-05,
      "loss": 11.9723,
      "step": 14390
    },
    {
      "epoch": 2.5164590163934424,
      "grad_norm": 0.7178566725130203,
      "learning_rate": 6.385602724638029e-05,
      "loss": 11.9238,
      "step": 14391
    },
    {
      "epoch": 2.5166338797814207,
      "grad_norm": 0.6215220401332324,
      "learning_rate": 6.38428240623859e-05,
      "loss": 11.8982,
      "step": 14392
    },
    {
      "epoch": 2.516808743169399,
      "grad_norm": 0.6567876487014768,
      "learning_rate": 6.382962160345389e-05,
      "loss": 11.8764,
      "step": 14393
    },
    {
      "epoch": 2.516983606557377,
      "grad_norm": 0.8012201598585531,
      "learning_rate": 6.381641986984901e-05,
      "loss": 11.8557,
      "step": 14394
    },
    {
      "epoch": 2.5171584699453553,
      "grad_norm": 0.8147605734700112,
      "learning_rate": 6.380321886183603e-05,
      "loss": 12.0455,
      "step": 14395
    },
    {
      "epoch": 2.517333333333333,
      "grad_norm": 0.716392686453325,
      "learning_rate": 6.379001857967966e-05,
      "loss": 11.6826,
      "step": 14396
    },
    {
      "epoch": 2.5175081967213115,
      "grad_norm": 0.9948145887029456,
      "learning_rate": 6.377681902364458e-05,
      "loss": 11.937,
      "step": 14397
    },
    {
      "epoch": 2.51768306010929,
      "grad_norm": 0.7843757648166909,
      "learning_rate": 6.376362019399549e-05,
      "loss": 11.9101,
      "step": 14398
    },
    {
      "epoch": 2.517857923497268,
      "grad_norm": 0.6779098428538112,
      "learning_rate": 6.375042209099709e-05,
      "loss": 11.7352,
      "step": 14399
    },
    {
      "epoch": 2.5180327868852457,
      "grad_norm": 0.8087222795952086,
      "learning_rate": 6.373722471491402e-05,
      "loss": 12.0333,
      "step": 14400
    },
    {
      "epoch": 2.518207650273224,
      "grad_norm": 0.7502496086259288,
      "learning_rate": 6.372402806601097e-05,
      "loss": 12.0122,
      "step": 14401
    },
    {
      "epoch": 2.5183825136612024,
      "grad_norm": 0.6261659048547703,
      "learning_rate": 6.371083214455251e-05,
      "loss": 11.9827,
      "step": 14402
    },
    {
      "epoch": 2.5185573770491803,
      "grad_norm": 0.7629320391510803,
      "learning_rate": 6.369763695080327e-05,
      "loss": 12.0849,
      "step": 14403
    },
    {
      "epoch": 2.5187322404371586,
      "grad_norm": 0.7154401311145993,
      "learning_rate": 6.368444248502789e-05,
      "loss": 11.9811,
      "step": 14404
    },
    {
      "epoch": 2.5189071038251365,
      "grad_norm": 0.6370205086878262,
      "learning_rate": 6.367124874749096e-05,
      "loss": 11.4774,
      "step": 14405
    },
    {
      "epoch": 2.519081967213115,
      "grad_norm": 0.7351401704888287,
      "learning_rate": 6.365805573845701e-05,
      "loss": 11.8797,
      "step": 14406
    },
    {
      "epoch": 2.5192568306010927,
      "grad_norm": 0.6165606237535076,
      "learning_rate": 6.364486345819064e-05,
      "loss": 11.973,
      "step": 14407
    },
    {
      "epoch": 2.519431693989071,
      "grad_norm": 0.6584354592453072,
      "learning_rate": 6.363167190695636e-05,
      "loss": 11.8103,
      "step": 14408
    },
    {
      "epoch": 2.519606557377049,
      "grad_norm": 0.6972001183147832,
      "learning_rate": 6.361848108501873e-05,
      "loss": 11.9267,
      "step": 14409
    },
    {
      "epoch": 2.5197814207650273,
      "grad_norm": 0.6169782497181853,
      "learning_rate": 6.36052909926423e-05,
      "loss": 11.7467,
      "step": 14410
    },
    {
      "epoch": 2.5199562841530057,
      "grad_norm": 0.6597825621372874,
      "learning_rate": 6.35921016300915e-05,
      "loss": 11.9239,
      "step": 14411
    },
    {
      "epoch": 2.5201311475409836,
      "grad_norm": 0.6454885056705988,
      "learning_rate": 6.357891299763086e-05,
      "loss": 11.8305,
      "step": 14412
    },
    {
      "epoch": 2.520306010928962,
      "grad_norm": 0.6980654535741657,
      "learning_rate": 6.35657250955248e-05,
      "loss": 11.964,
      "step": 14413
    },
    {
      "epoch": 2.52048087431694,
      "grad_norm": 0.635551017944573,
      "learning_rate": 6.355253792403788e-05,
      "loss": 11.8461,
      "step": 14414
    },
    {
      "epoch": 2.520655737704918,
      "grad_norm": 0.6435945129679886,
      "learning_rate": 6.353935148343446e-05,
      "loss": 11.7972,
      "step": 14415
    },
    {
      "epoch": 2.520830601092896,
      "grad_norm": 0.6644561383408568,
      "learning_rate": 6.352616577397899e-05,
      "loss": 11.8558,
      "step": 14416
    },
    {
      "epoch": 2.5210054644808744,
      "grad_norm": 0.7084602714010435,
      "learning_rate": 6.351298079593591e-05,
      "loss": 11.7695,
      "step": 14417
    },
    {
      "epoch": 2.5211803278688523,
      "grad_norm": 0.6881193118831606,
      "learning_rate": 6.349979654956952e-05,
      "loss": 11.802,
      "step": 14418
    },
    {
      "epoch": 2.5213551912568306,
      "grad_norm": 0.720262285764757,
      "learning_rate": 6.348661303514436e-05,
      "loss": 11.7194,
      "step": 14419
    },
    {
      "epoch": 2.521530054644809,
      "grad_norm": 0.6312476883674158,
      "learning_rate": 6.347343025292469e-05,
      "loss": 11.6748,
      "step": 14420
    },
    {
      "epoch": 2.521704918032787,
      "grad_norm": 0.7044535215501144,
      "learning_rate": 6.346024820317488e-05,
      "loss": 11.7779,
      "step": 14421
    },
    {
      "epoch": 2.521879781420765,
      "grad_norm": 0.6552769113062694,
      "learning_rate": 6.344706688615928e-05,
      "loss": 11.903,
      "step": 14422
    },
    {
      "epoch": 2.522054644808743,
      "grad_norm": 0.668662776453353,
      "learning_rate": 6.343388630214226e-05,
      "loss": 11.6535,
      "step": 14423
    },
    {
      "epoch": 2.5222295081967214,
      "grad_norm": 0.6460363340939987,
      "learning_rate": 6.342070645138807e-05,
      "loss": 11.7734,
      "step": 14424
    },
    {
      "epoch": 2.5224043715846993,
      "grad_norm": 0.7341200259012173,
      "learning_rate": 6.340752733416104e-05,
      "loss": 11.8683,
      "step": 14425
    },
    {
      "epoch": 2.5225792349726777,
      "grad_norm": 0.6855499140076804,
      "learning_rate": 6.339434895072545e-05,
      "loss": 11.735,
      "step": 14426
    },
    {
      "epoch": 2.5227540983606556,
      "grad_norm": 0.7288315342102741,
      "learning_rate": 6.338117130134549e-05,
      "loss": 11.8802,
      "step": 14427
    },
    {
      "epoch": 2.522928961748634,
      "grad_norm": 0.6640782083756809,
      "learning_rate": 6.336799438628556e-05,
      "loss": 11.849,
      "step": 14428
    },
    {
      "epoch": 2.5231038251366122,
      "grad_norm": 0.7126558231210158,
      "learning_rate": 6.335481820580978e-05,
      "loss": 11.8611,
      "step": 14429
    },
    {
      "epoch": 2.52327868852459,
      "grad_norm": 0.6778381121544009,
      "learning_rate": 6.334164276018242e-05,
      "loss": 11.5903,
      "step": 14430
    },
    {
      "epoch": 2.523453551912568,
      "grad_norm": 0.7997159913934928,
      "learning_rate": 6.33284680496677e-05,
      "loss": 11.9253,
      "step": 14431
    },
    {
      "epoch": 2.5236284153005464,
      "grad_norm": 0.725172948182083,
      "learning_rate": 6.331529407452974e-05,
      "loss": 11.8272,
      "step": 14432
    },
    {
      "epoch": 2.5238032786885247,
      "grad_norm": 0.6266498017540005,
      "learning_rate": 6.33021208350328e-05,
      "loss": 11.8867,
      "step": 14433
    },
    {
      "epoch": 2.5239781420765026,
      "grad_norm": 0.7526115989304253,
      "learning_rate": 6.328894833144101e-05,
      "loss": 11.802,
      "step": 14434
    },
    {
      "epoch": 2.524153005464481,
      "grad_norm": 0.8157443138039268,
      "learning_rate": 6.327577656401855e-05,
      "loss": 12.0618,
      "step": 14435
    },
    {
      "epoch": 2.524327868852459,
      "grad_norm": 0.7313609122451418,
      "learning_rate": 6.32626055330295e-05,
      "loss": 11.9188,
      "step": 14436
    },
    {
      "epoch": 2.524502732240437,
      "grad_norm": 0.7569399828487399,
      "learning_rate": 6.324943523873798e-05,
      "loss": 11.5064,
      "step": 14437
    },
    {
      "epoch": 2.5246775956284155,
      "grad_norm": 0.6923982557501057,
      "learning_rate": 6.323626568140816e-05,
      "loss": 11.8502,
      "step": 14438
    },
    {
      "epoch": 2.5248524590163934,
      "grad_norm": 0.6819756543218731,
      "learning_rate": 6.32230968613041e-05,
      "loss": 11.6151,
      "step": 14439
    },
    {
      "epoch": 2.5250273224043713,
      "grad_norm": 0.701438680707712,
      "learning_rate": 6.320992877868987e-05,
      "loss": 11.6772,
      "step": 14440
    },
    {
      "epoch": 2.5252021857923497,
      "grad_norm": 0.7240077473204323,
      "learning_rate": 6.31967614338295e-05,
      "loss": 11.5691,
      "step": 14441
    },
    {
      "epoch": 2.525377049180328,
      "grad_norm": 0.6796513532729057,
      "learning_rate": 6.318359482698709e-05,
      "loss": 11.9472,
      "step": 14442
    },
    {
      "epoch": 2.525551912568306,
      "grad_norm": 0.6926478411679106,
      "learning_rate": 6.317042895842662e-05,
      "loss": 11.7164,
      "step": 14443
    },
    {
      "epoch": 2.5257267759562843,
      "grad_norm": 0.6968884491018571,
      "learning_rate": 6.315726382841217e-05,
      "loss": 11.747,
      "step": 14444
    },
    {
      "epoch": 2.525901639344262,
      "grad_norm": 0.6938749095812807,
      "learning_rate": 6.314409943720767e-05,
      "loss": 12.0867,
      "step": 14445
    },
    {
      "epoch": 2.5260765027322405,
      "grad_norm": 0.6934955421570611,
      "learning_rate": 6.313093578507711e-05,
      "loss": 11.9078,
      "step": 14446
    },
    {
      "epoch": 2.526251366120219,
      "grad_norm": 0.6927866612235812,
      "learning_rate": 6.311777287228454e-05,
      "loss": 11.7993,
      "step": 14447
    },
    {
      "epoch": 2.5264262295081967,
      "grad_norm": 0.6880516475987268,
      "learning_rate": 6.310461069909384e-05,
      "loss": 11.9959,
      "step": 14448
    },
    {
      "epoch": 2.5266010928961746,
      "grad_norm": 0.6680772659106506,
      "learning_rate": 6.309144926576902e-05,
      "loss": 11.8169,
      "step": 14449
    },
    {
      "epoch": 2.526775956284153,
      "grad_norm": 0.656886545053804,
      "learning_rate": 6.307828857257394e-05,
      "loss": 11.9197,
      "step": 14450
    },
    {
      "epoch": 2.5269508196721313,
      "grad_norm": 0.6233512141686437,
      "learning_rate": 6.306512861977251e-05,
      "loss": 11.7329,
      "step": 14451
    },
    {
      "epoch": 2.527125683060109,
      "grad_norm": 0.6316737627202784,
      "learning_rate": 6.305196940762869e-05,
      "loss": 11.883,
      "step": 14452
    },
    {
      "epoch": 2.5273005464480875,
      "grad_norm": 0.7063402584724301,
      "learning_rate": 6.303881093640634e-05,
      "loss": 11.8844,
      "step": 14453
    },
    {
      "epoch": 2.5274754098360654,
      "grad_norm": 0.7277817322597999,
      "learning_rate": 6.30256532063693e-05,
      "loss": 11.7763,
      "step": 14454
    },
    {
      "epoch": 2.527650273224044,
      "grad_norm": 0.7257856858420149,
      "learning_rate": 6.301249621778142e-05,
      "loss": 12.151,
      "step": 14455
    },
    {
      "epoch": 2.527825136612022,
      "grad_norm": 0.6823996150238332,
      "learning_rate": 6.299933997090657e-05,
      "loss": 11.9647,
      "step": 14456
    },
    {
      "epoch": 2.528,
      "grad_norm": 0.6323662503010069,
      "learning_rate": 6.298618446600856e-05,
      "loss": 11.8678,
      "step": 14457
    },
    {
      "epoch": 2.528174863387978,
      "grad_norm": 0.7057741717335044,
      "learning_rate": 6.297302970335121e-05,
      "loss": 12.0003,
      "step": 14458
    },
    {
      "epoch": 2.5283497267759563,
      "grad_norm": 0.6850946686508183,
      "learning_rate": 6.295987568319831e-05,
      "loss": 11.7665,
      "step": 14459
    },
    {
      "epoch": 2.5285245901639346,
      "grad_norm": 0.7085928660496964,
      "learning_rate": 6.294672240581361e-05,
      "loss": 11.7945,
      "step": 14460
    },
    {
      "epoch": 2.5286994535519125,
      "grad_norm": 0.6655288262582765,
      "learning_rate": 6.293356987146088e-05,
      "loss": 11.8233,
      "step": 14461
    },
    {
      "epoch": 2.528874316939891,
      "grad_norm": 0.7407936371742396,
      "learning_rate": 6.292041808040393e-05,
      "loss": 11.9541,
      "step": 14462
    },
    {
      "epoch": 2.5290491803278687,
      "grad_norm": 0.6999458292703199,
      "learning_rate": 6.290726703290642e-05,
      "loss": 11.8236,
      "step": 14463
    },
    {
      "epoch": 2.529224043715847,
      "grad_norm": 0.5786186401889735,
      "learning_rate": 6.28941167292321e-05,
      "loss": 12.025,
      "step": 14464
    },
    {
      "epoch": 2.529398907103825,
      "grad_norm": 0.7297266551681637,
      "learning_rate": 6.288096716964469e-05,
      "loss": 12.0632,
      "step": 14465
    },
    {
      "epoch": 2.5295737704918033,
      "grad_norm": 0.6386039565742563,
      "learning_rate": 6.286781835440778e-05,
      "loss": 12.0049,
      "step": 14466
    },
    {
      "epoch": 2.529748633879781,
      "grad_norm": 0.5893410339983479,
      "learning_rate": 6.28546702837852e-05,
      "loss": 11.6359,
      "step": 14467
    },
    {
      "epoch": 2.5299234972677596,
      "grad_norm": 0.6559675039901139,
      "learning_rate": 6.284152295804051e-05,
      "loss": 11.8205,
      "step": 14468
    },
    {
      "epoch": 2.530098360655738,
      "grad_norm": 0.6473298599676612,
      "learning_rate": 6.282837637743738e-05,
      "loss": 12.0098,
      "step": 14469
    },
    {
      "epoch": 2.530273224043716,
      "grad_norm": 0.638900013077077,
      "learning_rate": 6.28152305422394e-05,
      "loss": 11.9918,
      "step": 14470
    },
    {
      "epoch": 2.530448087431694,
      "grad_norm": 0.6954346266960245,
      "learning_rate": 6.28020854527103e-05,
      "loss": 12.0519,
      "step": 14471
    },
    {
      "epoch": 2.530622950819672,
      "grad_norm": 0.6905335268341909,
      "learning_rate": 6.278894110911353e-05,
      "loss": 11.685,
      "step": 14472
    },
    {
      "epoch": 2.5307978142076504,
      "grad_norm": 0.6437686832704212,
      "learning_rate": 6.277579751171276e-05,
      "loss": 11.7836,
      "step": 14473
    },
    {
      "epoch": 2.5309726775956283,
      "grad_norm": 0.6448710119951377,
      "learning_rate": 6.276265466077158e-05,
      "loss": 11.9594,
      "step": 14474
    },
    {
      "epoch": 2.5311475409836066,
      "grad_norm": 0.6845322967504297,
      "learning_rate": 6.274951255655344e-05,
      "loss": 11.8083,
      "step": 14475
    },
    {
      "epoch": 2.5313224043715845,
      "grad_norm": 0.6675023399335386,
      "learning_rate": 6.273637119932199e-05,
      "loss": 12.0774,
      "step": 14476
    },
    {
      "epoch": 2.531497267759563,
      "grad_norm": 0.6359304279838331,
      "learning_rate": 6.272323058934071e-05,
      "loss": 11.8765,
      "step": 14477
    },
    {
      "epoch": 2.531672131147541,
      "grad_norm": 0.6137877224083362,
      "learning_rate": 6.27100907268731e-05,
      "loss": 11.7252,
      "step": 14478
    },
    {
      "epoch": 2.531846994535519,
      "grad_norm": 0.6309796890068967,
      "learning_rate": 6.26969516121827e-05,
      "loss": 11.755,
      "step": 14479
    },
    {
      "epoch": 2.5320218579234974,
      "grad_norm": 0.6262190710549496,
      "learning_rate": 6.26838132455329e-05,
      "loss": 11.5933,
      "step": 14480
    },
    {
      "epoch": 2.5321967213114753,
      "grad_norm": 0.6414846665838225,
      "learning_rate": 6.267067562718725e-05,
      "loss": 11.779,
      "step": 14481
    },
    {
      "epoch": 2.5323715846994537,
      "grad_norm": 0.6709545177509172,
      "learning_rate": 6.265753875740916e-05,
      "loss": 12.0124,
      "step": 14482
    },
    {
      "epoch": 2.5325464480874316,
      "grad_norm": 0.6262485764732955,
      "learning_rate": 6.26444026364621e-05,
      "loss": 11.8957,
      "step": 14483
    },
    {
      "epoch": 2.53272131147541,
      "grad_norm": 0.643554845034864,
      "learning_rate": 6.263126726460945e-05,
      "loss": 12.1061,
      "step": 14484
    },
    {
      "epoch": 2.532896174863388,
      "grad_norm": 0.7107500918677744,
      "learning_rate": 6.261813264211461e-05,
      "loss": 11.7533,
      "step": 14485
    },
    {
      "epoch": 2.533071038251366,
      "grad_norm": 0.6813200036935066,
      "learning_rate": 6.2604998769241e-05,
      "loss": 11.6169,
      "step": 14486
    },
    {
      "epoch": 2.5332459016393445,
      "grad_norm": 0.7237551656224125,
      "learning_rate": 6.259186564625197e-05,
      "loss": 11.8213,
      "step": 14487
    },
    {
      "epoch": 2.5334207650273224,
      "grad_norm": 0.7137419873564703,
      "learning_rate": 6.257873327341093e-05,
      "loss": 12.0509,
      "step": 14488
    },
    {
      "epoch": 2.5335956284153003,
      "grad_norm": 0.6238311706328433,
      "learning_rate": 6.256560165098116e-05,
      "loss": 12.0884,
      "step": 14489
    },
    {
      "epoch": 2.5337704918032786,
      "grad_norm": 0.5839835463909488,
      "learning_rate": 6.255247077922601e-05,
      "loss": 12.0672,
      "step": 14490
    },
    {
      "epoch": 2.533945355191257,
      "grad_norm": 0.6695819766624099,
      "learning_rate": 6.25393406584088e-05,
      "loss": 12.0836,
      "step": 14491
    },
    {
      "epoch": 2.534120218579235,
      "grad_norm": 0.6260196995509036,
      "learning_rate": 6.252621128879286e-05,
      "loss": 11.8133,
      "step": 14492
    },
    {
      "epoch": 2.534295081967213,
      "grad_norm": 0.6632487322431008,
      "learning_rate": 6.251308267064143e-05,
      "loss": 11.8747,
      "step": 14493
    },
    {
      "epoch": 2.534469945355191,
      "grad_norm": 0.6135045234656304,
      "learning_rate": 6.249995480421776e-05,
      "loss": 11.8773,
      "step": 14494
    },
    {
      "epoch": 2.5346448087431694,
      "grad_norm": 0.6241826822256799,
      "learning_rate": 6.248682768978514e-05,
      "loss": 11.7171,
      "step": 14495
    },
    {
      "epoch": 2.534819672131148,
      "grad_norm": 0.6430610582806733,
      "learning_rate": 6.247370132760683e-05,
      "loss": 11.9934,
      "step": 14496
    },
    {
      "epoch": 2.5349945355191257,
      "grad_norm": 0.6528090424553885,
      "learning_rate": 6.246057571794604e-05,
      "loss": 12.0342,
      "step": 14497
    },
    {
      "epoch": 2.5351693989071036,
      "grad_norm": 0.605218451664377,
      "learning_rate": 6.244745086106596e-05,
      "loss": 11.9284,
      "step": 14498
    },
    {
      "epoch": 2.535344262295082,
      "grad_norm": 0.615388589314074,
      "learning_rate": 6.243432675722976e-05,
      "loss": 11.7632,
      "step": 14499
    },
    {
      "epoch": 2.5355191256830603,
      "grad_norm": 0.6652273154836373,
      "learning_rate": 6.242120340670067e-05,
      "loss": 11.7845,
      "step": 14500
    },
    {
      "epoch": 2.535693989071038,
      "grad_norm": 0.6441834095776507,
      "learning_rate": 6.240808080974185e-05,
      "loss": 11.6172,
      "step": 14501
    },
    {
      "epoch": 2.5358688524590165,
      "grad_norm": 0.6122601600505417,
      "learning_rate": 6.239495896661643e-05,
      "loss": 11.8785,
      "step": 14502
    },
    {
      "epoch": 2.5360437158469944,
      "grad_norm": 0.6640346184633606,
      "learning_rate": 6.238183787758754e-05,
      "loss": 11.8982,
      "step": 14503
    },
    {
      "epoch": 2.5362185792349727,
      "grad_norm": 0.6636033062720366,
      "learning_rate": 6.236871754291827e-05,
      "loss": 11.7974,
      "step": 14504
    },
    {
      "epoch": 2.536393442622951,
      "grad_norm": 0.6687888371933879,
      "learning_rate": 6.235559796287179e-05,
      "loss": 11.7306,
      "step": 14505
    },
    {
      "epoch": 2.536568306010929,
      "grad_norm": 0.6717636558809739,
      "learning_rate": 6.234247913771116e-05,
      "loss": 11.6514,
      "step": 14506
    },
    {
      "epoch": 2.536743169398907,
      "grad_norm": 0.7225743866580258,
      "learning_rate": 6.232936106769944e-05,
      "loss": 12.1096,
      "step": 14507
    },
    {
      "epoch": 2.536918032786885,
      "grad_norm": 0.7393124598834223,
      "learning_rate": 6.23162437530997e-05,
      "loss": 11.777,
      "step": 14508
    },
    {
      "epoch": 2.5370928961748636,
      "grad_norm": 0.6658409711152251,
      "learning_rate": 6.230312719417493e-05,
      "loss": 11.7467,
      "step": 14509
    },
    {
      "epoch": 2.5372677595628415,
      "grad_norm": 0.7109557614111613,
      "learning_rate": 6.229001139118826e-05,
      "loss": 11.7892,
      "step": 14510
    },
    {
      "epoch": 2.53744262295082,
      "grad_norm": 0.7151073664051624,
      "learning_rate": 6.227689634440263e-05,
      "loss": 11.8139,
      "step": 14511
    },
    {
      "epoch": 2.5376174863387977,
      "grad_norm": 0.6311994495920793,
      "learning_rate": 6.226378205408105e-05,
      "loss": 11.8877,
      "step": 14512
    },
    {
      "epoch": 2.537792349726776,
      "grad_norm": 0.6945908368169471,
      "learning_rate": 6.225066852048654e-05,
      "loss": 11.7346,
      "step": 14513
    },
    {
      "epoch": 2.5379672131147544,
      "grad_norm": 0.7911145808766237,
      "learning_rate": 6.223755574388195e-05,
      "loss": 11.9563,
      "step": 14514
    },
    {
      "epoch": 2.5381420765027323,
      "grad_norm": 0.6462564428789036,
      "learning_rate": 6.222444372453038e-05,
      "loss": 11.7458,
      "step": 14515
    },
    {
      "epoch": 2.53831693989071,
      "grad_norm": 0.663972673654223,
      "learning_rate": 6.221133246269468e-05,
      "loss": 11.6271,
      "step": 14516
    },
    {
      "epoch": 2.5384918032786885,
      "grad_norm": 0.7752801543487099,
      "learning_rate": 6.219822195863782e-05,
      "loss": 11.7332,
      "step": 14517
    },
    {
      "epoch": 2.538666666666667,
      "grad_norm": 0.7067955628802801,
      "learning_rate": 6.21851122126226e-05,
      "loss": 11.6136,
      "step": 14518
    },
    {
      "epoch": 2.5388415300546447,
      "grad_norm": 0.696976362376782,
      "learning_rate": 6.217200322491205e-05,
      "loss": 11.8892,
      "step": 14519
    },
    {
      "epoch": 2.539016393442623,
      "grad_norm": 0.6942797104132918,
      "learning_rate": 6.215889499576898e-05,
      "loss": 12.0315,
      "step": 14520
    },
    {
      "epoch": 2.539191256830601,
      "grad_norm": 0.6398337129722469,
      "learning_rate": 6.214578752545625e-05,
      "loss": 11.7962,
      "step": 14521
    },
    {
      "epoch": 2.5393661202185793,
      "grad_norm": 0.621256464049392,
      "learning_rate": 6.213268081423671e-05,
      "loss": 12.0644,
      "step": 14522
    },
    {
      "epoch": 2.5395409836065572,
      "grad_norm": 0.7641802364532826,
      "learning_rate": 6.211957486237314e-05,
      "loss": 12.1599,
      "step": 14523
    },
    {
      "epoch": 2.5397158469945356,
      "grad_norm": 0.6851622057102748,
      "learning_rate": 6.210646967012845e-05,
      "loss": 12.1141,
      "step": 14524
    },
    {
      "epoch": 2.5398907103825135,
      "grad_norm": 0.6287794824418296,
      "learning_rate": 6.209336523776538e-05,
      "loss": 11.9574,
      "step": 14525
    },
    {
      "epoch": 2.540065573770492,
      "grad_norm": 0.7131627744511843,
      "learning_rate": 6.208026156554676e-05,
      "loss": 11.8303,
      "step": 14526
    },
    {
      "epoch": 2.54024043715847,
      "grad_norm": 0.5962558319163206,
      "learning_rate": 6.206715865373529e-05,
      "loss": 11.8016,
      "step": 14527
    },
    {
      "epoch": 2.540415300546448,
      "grad_norm": 0.7243551091053478,
      "learning_rate": 6.205405650259373e-05,
      "loss": 11.9334,
      "step": 14528
    },
    {
      "epoch": 2.5405901639344264,
      "grad_norm": 0.8204801056889036,
      "learning_rate": 6.204095511238487e-05,
      "loss": 11.7335,
      "step": 14529
    },
    {
      "epoch": 2.5407650273224043,
      "grad_norm": 0.6407289565853442,
      "learning_rate": 6.202785448337142e-05,
      "loss": 11.8307,
      "step": 14530
    },
    {
      "epoch": 2.5409398907103826,
      "grad_norm": 0.7226599460822816,
      "learning_rate": 6.201475461581609e-05,
      "loss": 12.0425,
      "step": 14531
    },
    {
      "epoch": 2.5411147540983605,
      "grad_norm": 0.6957172755216375,
      "learning_rate": 6.200165550998153e-05,
      "loss": 11.7933,
      "step": 14532
    },
    {
      "epoch": 2.541289617486339,
      "grad_norm": 0.6793935483084652,
      "learning_rate": 6.198855716613046e-05,
      "loss": 11.9233,
      "step": 14533
    },
    {
      "epoch": 2.5414644808743168,
      "grad_norm": 0.6478024144963024,
      "learning_rate": 6.197545958452552e-05,
      "loss": 11.855,
      "step": 14534
    },
    {
      "epoch": 2.541639344262295,
      "grad_norm": 0.6133062152699681,
      "learning_rate": 6.196236276542938e-05,
      "loss": 11.8917,
      "step": 14535
    },
    {
      "epoch": 2.5418142076502734,
      "grad_norm": 0.6375918299027786,
      "learning_rate": 6.194926670910467e-05,
      "loss": 12.0556,
      "step": 14536
    },
    {
      "epoch": 2.5419890710382513,
      "grad_norm": 0.6631575378708757,
      "learning_rate": 6.193617141581398e-05,
      "loss": 11.8629,
      "step": 14537
    },
    {
      "epoch": 2.5421639344262292,
      "grad_norm": 0.6277052480792198,
      "learning_rate": 6.192307688581989e-05,
      "loss": 11.7998,
      "step": 14538
    },
    {
      "epoch": 2.5423387978142076,
      "grad_norm": 0.6312418053623894,
      "learning_rate": 6.190998311938505e-05,
      "loss": 11.9285,
      "step": 14539
    },
    {
      "epoch": 2.542513661202186,
      "grad_norm": 0.6071689676721128,
      "learning_rate": 6.189689011677202e-05,
      "loss": 11.7605,
      "step": 14540
    },
    {
      "epoch": 2.542688524590164,
      "grad_norm": 0.6063502162505041,
      "learning_rate": 6.188379787824332e-05,
      "loss": 11.8549,
      "step": 14541
    },
    {
      "epoch": 2.542863387978142,
      "grad_norm": 0.5810986374565766,
      "learning_rate": 6.187070640406147e-05,
      "loss": 11.8072,
      "step": 14542
    },
    {
      "epoch": 2.54303825136612,
      "grad_norm": 0.6161454149673873,
      "learning_rate": 6.185761569448907e-05,
      "loss": 11.9528,
      "step": 14543
    },
    {
      "epoch": 2.5432131147540984,
      "grad_norm": 0.6147561931006735,
      "learning_rate": 6.184452574978858e-05,
      "loss": 11.712,
      "step": 14544
    },
    {
      "epoch": 2.5433879781420767,
      "grad_norm": 0.6123637386893521,
      "learning_rate": 6.183143657022252e-05,
      "loss": 11.9544,
      "step": 14545
    },
    {
      "epoch": 2.5435628415300546,
      "grad_norm": 0.6470642757805382,
      "learning_rate": 6.181834815605331e-05,
      "loss": 11.8295,
      "step": 14546
    },
    {
      "epoch": 2.5437377049180325,
      "grad_norm": 0.6398060246400944,
      "learning_rate": 6.180526050754346e-05,
      "loss": 11.9336,
      "step": 14547
    },
    {
      "epoch": 2.543912568306011,
      "grad_norm": 0.678788374906942,
      "learning_rate": 6.179217362495542e-05,
      "loss": 12.0194,
      "step": 14548
    },
    {
      "epoch": 2.544087431693989,
      "grad_norm": 0.6980728267944136,
      "learning_rate": 6.177908750855164e-05,
      "loss": 11.9231,
      "step": 14549
    },
    {
      "epoch": 2.544262295081967,
      "grad_norm": 0.6054418979429542,
      "learning_rate": 6.176600215859447e-05,
      "loss": 11.5431,
      "step": 14550
    },
    {
      "epoch": 2.5444371584699454,
      "grad_norm": 0.6982235134361875,
      "learning_rate": 6.175291757534635e-05,
      "loss": 11.9772,
      "step": 14551
    },
    {
      "epoch": 2.5446120218579233,
      "grad_norm": 0.6382767265396985,
      "learning_rate": 6.173983375906966e-05,
      "loss": 12.0357,
      "step": 14552
    },
    {
      "epoch": 2.5447868852459017,
      "grad_norm": 0.6244113709721669,
      "learning_rate": 6.17267507100268e-05,
      "loss": 11.726,
      "step": 14553
    },
    {
      "epoch": 2.54496174863388,
      "grad_norm": 0.7035482370187376,
      "learning_rate": 6.171366842848012e-05,
      "loss": 11.7099,
      "step": 14554
    },
    {
      "epoch": 2.545136612021858,
      "grad_norm": 0.636521408013411,
      "learning_rate": 6.170058691469193e-05,
      "loss": 12.0222,
      "step": 14555
    },
    {
      "epoch": 2.545311475409836,
      "grad_norm": 0.6487113807563029,
      "learning_rate": 6.168750616892459e-05,
      "loss": 11.8633,
      "step": 14556
    },
    {
      "epoch": 2.545486338797814,
      "grad_norm": 0.6500470787217237,
      "learning_rate": 6.16744261914403e-05,
      "loss": 11.8974,
      "step": 14557
    },
    {
      "epoch": 2.5456612021857925,
      "grad_norm": 0.6192041474456685,
      "learning_rate": 6.166134698250152e-05,
      "loss": 11.6434,
      "step": 14558
    },
    {
      "epoch": 2.5458360655737704,
      "grad_norm": 0.6432839887464546,
      "learning_rate": 6.164826854237042e-05,
      "loss": 11.8062,
      "step": 14559
    },
    {
      "epoch": 2.5460109289617487,
      "grad_norm": 0.6339476051356446,
      "learning_rate": 6.163519087130929e-05,
      "loss": 11.9199,
      "step": 14560
    },
    {
      "epoch": 2.5461857923497266,
      "grad_norm": 0.6317214014484818,
      "learning_rate": 6.16221139695804e-05,
      "loss": 11.9199,
      "step": 14561
    },
    {
      "epoch": 2.546360655737705,
      "grad_norm": 0.7126374619928539,
      "learning_rate": 6.16090378374459e-05,
      "loss": 11.7952,
      "step": 14562
    },
    {
      "epoch": 2.5465355191256833,
      "grad_norm": 0.6260569603091394,
      "learning_rate": 6.159596247516814e-05,
      "loss": 11.8259,
      "step": 14563
    },
    {
      "epoch": 2.546710382513661,
      "grad_norm": 0.6108113709298535,
      "learning_rate": 6.158288788300922e-05,
      "loss": 11.8843,
      "step": 14564
    },
    {
      "epoch": 2.546885245901639,
      "grad_norm": 0.6412771358203616,
      "learning_rate": 6.156981406123137e-05,
      "loss": 11.9385,
      "step": 14565
    },
    {
      "epoch": 2.5470601092896175,
      "grad_norm": 0.7171614485471505,
      "learning_rate": 6.155674101009667e-05,
      "loss": 11.8751,
      "step": 14566
    },
    {
      "epoch": 2.547234972677596,
      "grad_norm": 0.6962262372370299,
      "learning_rate": 6.154366872986744e-05,
      "loss": 11.8581,
      "step": 14567
    },
    {
      "epoch": 2.5474098360655737,
      "grad_norm": 0.6916319559436302,
      "learning_rate": 6.153059722080569e-05,
      "loss": 11.9795,
      "step": 14568
    },
    {
      "epoch": 2.547584699453552,
      "grad_norm": 0.6481898480263782,
      "learning_rate": 6.151752648317357e-05,
      "loss": 11.9258,
      "step": 14569
    },
    {
      "epoch": 2.54775956284153,
      "grad_norm": 0.6101352830822662,
      "learning_rate": 6.150445651723325e-05,
      "loss": 11.8789,
      "step": 14570
    },
    {
      "epoch": 2.5479344262295083,
      "grad_norm": 0.6339902382309455,
      "learning_rate": 6.149138732324669e-05,
      "loss": 11.6962,
      "step": 14571
    },
    {
      "epoch": 2.548109289617486,
      "grad_norm": 0.6461236556722169,
      "learning_rate": 6.147831890147613e-05,
      "loss": 12.0422,
      "step": 14572
    },
    {
      "epoch": 2.5482841530054645,
      "grad_norm": 0.6390153208001842,
      "learning_rate": 6.146525125218352e-05,
      "loss": 11.82,
      "step": 14573
    },
    {
      "epoch": 2.5484590163934424,
      "grad_norm": 0.663957306309619,
      "learning_rate": 6.145218437563097e-05,
      "loss": 11.7756,
      "step": 14574
    },
    {
      "epoch": 2.5486338797814208,
      "grad_norm": 1.8136587651622709,
      "learning_rate": 6.143911827208045e-05,
      "loss": 11.9825,
      "step": 14575
    },
    {
      "epoch": 2.548808743169399,
      "grad_norm": 0.6460929404802299,
      "learning_rate": 6.142605294179399e-05,
      "loss": 11.9664,
      "step": 14576
    },
    {
      "epoch": 2.548983606557377,
      "grad_norm": 0.6402709511517976,
      "learning_rate": 6.141298838503362e-05,
      "loss": 11.7351,
      "step": 14577
    },
    {
      "epoch": 2.5491584699453553,
      "grad_norm": 0.6057060770110683,
      "learning_rate": 6.139992460206132e-05,
      "loss": 11.6476,
      "step": 14578
    },
    {
      "epoch": 2.5493333333333332,
      "grad_norm": 0.6507392672436557,
      "learning_rate": 6.138686159313906e-05,
      "loss": 11.755,
      "step": 14579
    },
    {
      "epoch": 2.5495081967213116,
      "grad_norm": 0.6381813917544574,
      "learning_rate": 6.137379935852879e-05,
      "loss": 11.6339,
      "step": 14580
    },
    {
      "epoch": 2.5496830601092895,
      "grad_norm": 0.6201254597929432,
      "learning_rate": 6.13607378984924e-05,
      "loss": 11.7428,
      "step": 14581
    },
    {
      "epoch": 2.549857923497268,
      "grad_norm": 0.6488717488340235,
      "learning_rate": 6.134767721329187e-05,
      "loss": 11.7207,
      "step": 14582
    },
    {
      "epoch": 2.5500327868852457,
      "grad_norm": 0.6462933081718124,
      "learning_rate": 6.133461730318911e-05,
      "loss": 11.7714,
      "step": 14583
    },
    {
      "epoch": 2.550207650273224,
      "grad_norm": 0.7087559317387654,
      "learning_rate": 6.132155816844597e-05,
      "loss": 11.962,
      "step": 14584
    },
    {
      "epoch": 2.5503825136612024,
      "grad_norm": 0.7080495954762975,
      "learning_rate": 6.130849980932434e-05,
      "loss": 11.9108,
      "step": 14585
    },
    {
      "epoch": 2.5505573770491803,
      "grad_norm": 0.647618721947358,
      "learning_rate": 6.12954422260861e-05,
      "loss": 11.9353,
      "step": 14586
    },
    {
      "epoch": 2.5507322404371586,
      "grad_norm": 0.5708242318575654,
      "learning_rate": 6.128238541899307e-05,
      "loss": 11.3079,
      "step": 14587
    },
    {
      "epoch": 2.5509071038251365,
      "grad_norm": 0.6958198790168798,
      "learning_rate": 6.126932938830712e-05,
      "loss": 11.6509,
      "step": 14588
    },
    {
      "epoch": 2.551081967213115,
      "grad_norm": 0.7618919044188479,
      "learning_rate": 6.125627413429003e-05,
      "loss": 11.7801,
      "step": 14589
    },
    {
      "epoch": 2.5512568306010928,
      "grad_norm": 0.6965093630413679,
      "learning_rate": 6.124321965720357e-05,
      "loss": 11.9694,
      "step": 14590
    },
    {
      "epoch": 2.551431693989071,
      "grad_norm": 0.6932253829275454,
      "learning_rate": 6.123016595730958e-05,
      "loss": 11.7941,
      "step": 14591
    },
    {
      "epoch": 2.551606557377049,
      "grad_norm": 0.6514231924824689,
      "learning_rate": 6.12171130348698e-05,
      "loss": 12.0388,
      "step": 14592
    },
    {
      "epoch": 2.5517814207650273,
      "grad_norm": 0.7150640982026023,
      "learning_rate": 6.120406089014599e-05,
      "loss": 11.8018,
      "step": 14593
    },
    {
      "epoch": 2.5519562841530057,
      "grad_norm": 0.7752044769664471,
      "learning_rate": 6.119100952339985e-05,
      "loss": 12.0153,
      "step": 14594
    },
    {
      "epoch": 2.5521311475409836,
      "grad_norm": 0.7312527496635524,
      "learning_rate": 6.117795893489313e-05,
      "loss": 11.8765,
      "step": 14595
    },
    {
      "epoch": 2.5523060109289615,
      "grad_norm": 0.736813286539694,
      "learning_rate": 6.116490912488755e-05,
      "loss": 11.8855,
      "step": 14596
    },
    {
      "epoch": 2.55248087431694,
      "grad_norm": 0.7360103927076138,
      "learning_rate": 6.11518600936448e-05,
      "loss": 11.8378,
      "step": 14597
    },
    {
      "epoch": 2.552655737704918,
      "grad_norm": 0.6971216703698143,
      "learning_rate": 6.113881184142652e-05,
      "loss": 11.7297,
      "step": 14598
    },
    {
      "epoch": 2.552830601092896,
      "grad_norm": 0.6355471369377022,
      "learning_rate": 6.112576436849437e-05,
      "loss": 11.6997,
      "step": 14599
    },
    {
      "epoch": 2.5530054644808744,
      "grad_norm": 0.6731597055816968,
      "learning_rate": 6.111271767510998e-05,
      "loss": 11.954,
      "step": 14600
    },
    {
      "epoch": 2.5531803278688523,
      "grad_norm": 0.6419899310617166,
      "learning_rate": 6.109967176153506e-05,
      "loss": 11.7541,
      "step": 14601
    },
    {
      "epoch": 2.5533551912568306,
      "grad_norm": 0.7220420692642526,
      "learning_rate": 6.108662662803113e-05,
      "loss": 12.0963,
      "step": 14602
    },
    {
      "epoch": 2.553530054644809,
      "grad_norm": 0.635676981116985,
      "learning_rate": 6.107358227485981e-05,
      "loss": 11.6733,
      "step": 14603
    },
    {
      "epoch": 2.553704918032787,
      "grad_norm": 1.3923514143867268,
      "learning_rate": 6.106053870228272e-05,
      "loss": 11.9283,
      "step": 14604
    },
    {
      "epoch": 2.5538797814207648,
      "grad_norm": 0.7521989107258195,
      "learning_rate": 6.104749591056131e-05,
      "loss": 11.6595,
      "step": 14605
    },
    {
      "epoch": 2.554054644808743,
      "grad_norm": 0.6382998197345187,
      "learning_rate": 6.103445389995729e-05,
      "loss": 11.959,
      "step": 14606
    },
    {
      "epoch": 2.5542295081967215,
      "grad_norm": 0.5811727125597195,
      "learning_rate": 6.102141267073207e-05,
      "loss": 11.9007,
      "step": 14607
    },
    {
      "epoch": 2.5544043715846994,
      "grad_norm": 0.6971122665339865,
      "learning_rate": 6.100837222314721e-05,
      "loss": 11.9294,
      "step": 14608
    },
    {
      "epoch": 2.5545792349726777,
      "grad_norm": 0.6911455585849985,
      "learning_rate": 6.0995332557464234e-05,
      "loss": 11.9565,
      "step": 14609
    },
    {
      "epoch": 2.5547540983606556,
      "grad_norm": 0.6101732153380385,
      "learning_rate": 6.0982293673944544e-05,
      "loss": 11.5437,
      "step": 14610
    },
    {
      "epoch": 2.554928961748634,
      "grad_norm": 0.6947397533860931,
      "learning_rate": 6.096925557284971e-05,
      "loss": 12.0473,
      "step": 14611
    },
    {
      "epoch": 2.5551038251366123,
      "grad_norm": 0.6648471428283348,
      "learning_rate": 6.0956218254441135e-05,
      "loss": 11.8612,
      "step": 14612
    },
    {
      "epoch": 2.55527868852459,
      "grad_norm": 0.6763020642257705,
      "learning_rate": 6.094318171898028e-05,
      "loss": 11.7305,
      "step": 14613
    },
    {
      "epoch": 2.555453551912568,
      "grad_norm": 0.6624470014800757,
      "learning_rate": 6.093014596672849e-05,
      "loss": 11.7312,
      "step": 14614
    },
    {
      "epoch": 2.5556284153005464,
      "grad_norm": 0.6795026393471716,
      "learning_rate": 6.0917110997947304e-05,
      "loss": 11.8194,
      "step": 14615
    },
    {
      "epoch": 2.5558032786885247,
      "grad_norm": 0.6394706322343718,
      "learning_rate": 6.0904076812898e-05,
      "loss": 11.848,
      "step": 14616
    },
    {
      "epoch": 2.5559781420765026,
      "grad_norm": 0.6608410978103859,
      "learning_rate": 6.0891043411842016e-05,
      "loss": 11.7565,
      "step": 14617
    },
    {
      "epoch": 2.556153005464481,
      "grad_norm": 0.6057328600556435,
      "learning_rate": 6.08780107950407e-05,
      "loss": 11.8871,
      "step": 14618
    },
    {
      "epoch": 2.556327868852459,
      "grad_norm": 0.6722402294426891,
      "learning_rate": 6.0864978962755335e-05,
      "loss": 11.8191,
      "step": 14619
    },
    {
      "epoch": 2.5565027322404372,
      "grad_norm": 0.7255616294158245,
      "learning_rate": 6.085194791524735e-05,
      "loss": 11.552,
      "step": 14620
    },
    {
      "epoch": 2.5566775956284156,
      "grad_norm": 0.6828209011729267,
      "learning_rate": 6.083891765277801e-05,
      "loss": 11.9539,
      "step": 14621
    },
    {
      "epoch": 2.5568524590163935,
      "grad_norm": 0.6293372192463867,
      "learning_rate": 6.082588817560862e-05,
      "loss": 11.8676,
      "step": 14622
    },
    {
      "epoch": 2.5570273224043714,
      "grad_norm": 0.6517262954470017,
      "learning_rate": 6.0812859484000436e-05,
      "loss": 11.8768,
      "step": 14623
    },
    {
      "epoch": 2.5572021857923497,
      "grad_norm": 0.6090518992696426,
      "learning_rate": 6.0799831578214714e-05,
      "loss": 11.8083,
      "step": 14624
    },
    {
      "epoch": 2.557377049180328,
      "grad_norm": 0.6239745268205878,
      "learning_rate": 6.078680445851276e-05,
      "loss": 11.4073,
      "step": 14625
    },
    {
      "epoch": 2.557551912568306,
      "grad_norm": 0.7035903717590744,
      "learning_rate": 6.077377812515577e-05,
      "loss": 12.003,
      "step": 14626
    },
    {
      "epoch": 2.5577267759562843,
      "grad_norm": 0.685593180477828,
      "learning_rate": 6.076075257840499e-05,
      "loss": 11.8022,
      "step": 14627
    },
    {
      "epoch": 2.557901639344262,
      "grad_norm": 0.6447386743873488,
      "learning_rate": 6.074772781852158e-05,
      "loss": 12.0037,
      "step": 14628
    },
    {
      "epoch": 2.5580765027322405,
      "grad_norm": 0.6583305997649805,
      "learning_rate": 6.0734703845766716e-05,
      "loss": 12.0711,
      "step": 14629
    },
    {
      "epoch": 2.5582513661202184,
      "grad_norm": 0.6495323729469633,
      "learning_rate": 6.0721680660401635e-05,
      "loss": 11.5533,
      "step": 14630
    },
    {
      "epoch": 2.5584262295081968,
      "grad_norm": 0.6615592246070012,
      "learning_rate": 6.070865826268748e-05,
      "loss": 11.7284,
      "step": 14631
    },
    {
      "epoch": 2.5586010928961747,
      "grad_norm": 0.6274877157009379,
      "learning_rate": 6.069563665288533e-05,
      "loss": 11.5965,
      "step": 14632
    },
    {
      "epoch": 2.558775956284153,
      "grad_norm": 0.6559144056107575,
      "learning_rate": 6.0682615831256364e-05,
      "loss": 11.7866,
      "step": 14633
    },
    {
      "epoch": 2.5589508196721313,
      "grad_norm": 0.6918438933588332,
      "learning_rate": 6.066959579806164e-05,
      "loss": 11.6727,
      "step": 14634
    },
    {
      "epoch": 2.5591256830601092,
      "grad_norm": 0.6214621565224037,
      "learning_rate": 6.0656576553562295e-05,
      "loss": 11.765,
      "step": 14635
    },
    {
      "epoch": 2.5593005464480876,
      "grad_norm": 0.6303263551339872,
      "learning_rate": 6.064355809801943e-05,
      "loss": 11.8179,
      "step": 14636
    },
    {
      "epoch": 2.5594754098360655,
      "grad_norm": 0.6154608863919058,
      "learning_rate": 6.0630540431694026e-05,
      "loss": 11.8987,
      "step": 14637
    },
    {
      "epoch": 2.559650273224044,
      "grad_norm": 0.7169320301670464,
      "learning_rate": 6.061752355484715e-05,
      "loss": 11.4443,
      "step": 14638
    },
    {
      "epoch": 2.5598251366120217,
      "grad_norm": 0.6418467126940594,
      "learning_rate": 6.060450746773988e-05,
      "loss": 11.9767,
      "step": 14639
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.6107192026314088,
      "learning_rate": 6.0591492170633204e-05,
      "loss": 12.0122,
      "step": 14640
    },
    {
      "epoch": 2.560174863387978,
      "grad_norm": 0.6320541210904848,
      "learning_rate": 6.057847766378809e-05,
      "loss": 11.5406,
      "step": 14641
    },
    {
      "epoch": 2.5603497267759563,
      "grad_norm": 0.6580443119053979,
      "learning_rate": 6.056546394746554e-05,
      "loss": 11.7888,
      "step": 14642
    },
    {
      "epoch": 2.5605245901639346,
      "grad_norm": 0.683536781694075,
      "learning_rate": 6.0552451021926504e-05,
      "loss": 11.6145,
      "step": 14643
    },
    {
      "epoch": 2.5606994535519125,
      "grad_norm": 0.6694091142272058,
      "learning_rate": 6.053943888743196e-05,
      "loss": 11.8501,
      "step": 14644
    },
    {
      "epoch": 2.560874316939891,
      "grad_norm": 0.631169920666632,
      "learning_rate": 6.052642754424285e-05,
      "loss": 11.8583,
      "step": 14645
    },
    {
      "epoch": 2.5610491803278688,
      "grad_norm": 0.7365779182051281,
      "learning_rate": 6.051341699262003e-05,
      "loss": 12.0413,
      "step": 14646
    },
    {
      "epoch": 2.561224043715847,
      "grad_norm": 0.6628187169793307,
      "learning_rate": 6.050040723282446e-05,
      "loss": 11.7789,
      "step": 14647
    },
    {
      "epoch": 2.561398907103825,
      "grad_norm": 0.6213572276082471,
      "learning_rate": 6.0487398265116976e-05,
      "loss": 11.748,
      "step": 14648
    },
    {
      "epoch": 2.5615737704918033,
      "grad_norm": 0.6484440454000509,
      "learning_rate": 6.0474390089758514e-05,
      "loss": 11.5099,
      "step": 14649
    },
    {
      "epoch": 2.5617486338797812,
      "grad_norm": 0.6642936782156651,
      "learning_rate": 6.046138270700987e-05,
      "loss": 11.8276,
      "step": 14650
    },
    {
      "epoch": 2.5619234972677596,
      "grad_norm": 0.6870595136165222,
      "learning_rate": 6.0448376117131914e-05,
      "loss": 11.8325,
      "step": 14651
    },
    {
      "epoch": 2.562098360655738,
      "grad_norm": 0.6786971304026352,
      "learning_rate": 6.043537032038547e-05,
      "loss": 12.0746,
      "step": 14652
    },
    {
      "epoch": 2.562273224043716,
      "grad_norm": 0.6581855862851352,
      "learning_rate": 6.042236531703127e-05,
      "loss": 11.6141,
      "step": 14653
    },
    {
      "epoch": 2.5624480874316937,
      "grad_norm": 0.6497603770340526,
      "learning_rate": 6.040936110733022e-05,
      "loss": 12.0132,
      "step": 14654
    },
    {
      "epoch": 2.562622950819672,
      "grad_norm": 0.6588241852971671,
      "learning_rate": 6.039635769154301e-05,
      "loss": 11.636,
      "step": 14655
    },
    {
      "epoch": 2.5627978142076504,
      "grad_norm": 0.6488559562595592,
      "learning_rate": 6.038335506993045e-05,
      "loss": 11.8629,
      "step": 14656
    },
    {
      "epoch": 2.5629726775956283,
      "grad_norm": 0.6621129204599189,
      "learning_rate": 6.0370353242753265e-05,
      "loss": 11.8568,
      "step": 14657
    },
    {
      "epoch": 2.5631475409836066,
      "grad_norm": 0.7005886489941426,
      "learning_rate": 6.035735221027214e-05,
      "loss": 11.789,
      "step": 14658
    },
    {
      "epoch": 2.5633224043715845,
      "grad_norm": 0.6657080205548994,
      "learning_rate": 6.0344351972747846e-05,
      "loss": 11.6832,
      "step": 14659
    },
    {
      "epoch": 2.563497267759563,
      "grad_norm": 0.7159473713664601,
      "learning_rate": 6.033135253044105e-05,
      "loss": 11.5487,
      "step": 14660
    },
    {
      "epoch": 2.5636721311475412,
      "grad_norm": 0.6856984721165102,
      "learning_rate": 6.0318353883612465e-05,
      "loss": 11.907,
      "step": 14661
    },
    {
      "epoch": 2.563846994535519,
      "grad_norm": 0.6888474107136904,
      "learning_rate": 6.030535603252265e-05,
      "loss": 11.7467,
      "step": 14662
    },
    {
      "epoch": 2.564021857923497,
      "grad_norm": 0.7235374127035022,
      "learning_rate": 6.029235897743239e-05,
      "loss": 11.9288,
      "step": 14663
    },
    {
      "epoch": 2.5641967213114754,
      "grad_norm": 0.6200415462872674,
      "learning_rate": 6.027936271860223e-05,
      "loss": 11.893,
      "step": 14664
    },
    {
      "epoch": 2.5643715846994537,
      "grad_norm": 0.7791166824364524,
      "learning_rate": 6.02663672562928e-05,
      "loss": 11.9735,
      "step": 14665
    },
    {
      "epoch": 2.5645464480874316,
      "grad_norm": 0.6658497714156365,
      "learning_rate": 6.025337259076473e-05,
      "loss": 11.8412,
      "step": 14666
    },
    {
      "epoch": 2.56472131147541,
      "grad_norm": 0.629218868814465,
      "learning_rate": 6.0240378722278547e-05,
      "loss": 11.8282,
      "step": 14667
    },
    {
      "epoch": 2.564896174863388,
      "grad_norm": 0.7456396185686851,
      "learning_rate": 6.022738565109485e-05,
      "loss": 11.8734,
      "step": 14668
    },
    {
      "epoch": 2.565071038251366,
      "grad_norm": 0.721519125122637,
      "learning_rate": 6.021439337747421e-05,
      "loss": 11.7671,
      "step": 14669
    },
    {
      "epoch": 2.5652459016393445,
      "grad_norm": 0.6402154657713595,
      "learning_rate": 6.020140190167715e-05,
      "loss": 11.7007,
      "step": 14670
    },
    {
      "epoch": 2.5654207650273224,
      "grad_norm": 0.6892322416291685,
      "learning_rate": 6.018841122396415e-05,
      "loss": 11.8578,
      "step": 14671
    },
    {
      "epoch": 2.5655956284153003,
      "grad_norm": 0.7105313486427592,
      "learning_rate": 6.017542134459573e-05,
      "loss": 11.8194,
      "step": 14672
    },
    {
      "epoch": 2.5657704918032787,
      "grad_norm": 0.6827930575209579,
      "learning_rate": 6.016243226383241e-05,
      "loss": 11.1605,
      "step": 14673
    },
    {
      "epoch": 2.565945355191257,
      "grad_norm": 0.5999661559188136,
      "learning_rate": 6.014944398193465e-05,
      "loss": 11.964,
      "step": 14674
    },
    {
      "epoch": 2.566120218579235,
      "grad_norm": 0.6808042276667865,
      "learning_rate": 6.013645649916291e-05,
      "loss": 12.0107,
      "step": 14675
    },
    {
      "epoch": 2.5662950819672132,
      "grad_norm": 0.6449304925995143,
      "learning_rate": 6.012346981577759e-05,
      "loss": 11.8406,
      "step": 14676
    },
    {
      "epoch": 2.566469945355191,
      "grad_norm": 0.644079352283675,
      "learning_rate": 6.011048393203911e-05,
      "loss": 11.7752,
      "step": 14677
    },
    {
      "epoch": 2.5666448087431695,
      "grad_norm": 0.6183717245199953,
      "learning_rate": 6.009749884820793e-05,
      "loss": 11.7486,
      "step": 14678
    },
    {
      "epoch": 2.566819672131148,
      "grad_norm": 0.6895378538114113,
      "learning_rate": 6.0084514564544446e-05,
      "loss": 11.9786,
      "step": 14679
    },
    {
      "epoch": 2.5669945355191257,
      "grad_norm": 0.6214771762378788,
      "learning_rate": 6.007153108130896e-05,
      "loss": 11.8284,
      "step": 14680
    },
    {
      "epoch": 2.5671693989071036,
      "grad_norm": 0.6311662020307306,
      "learning_rate": 6.0058548398761894e-05,
      "loss": 11.9669,
      "step": 14681
    },
    {
      "epoch": 2.567344262295082,
      "grad_norm": 0.6262305054612486,
      "learning_rate": 6.004556651716354e-05,
      "loss": 11.7791,
      "step": 14682
    },
    {
      "epoch": 2.5675191256830603,
      "grad_norm": 0.6196876456696023,
      "learning_rate": 6.003258543677428e-05,
      "loss": 11.7582,
      "step": 14683
    },
    {
      "epoch": 2.567693989071038,
      "grad_norm": 0.7086237299878778,
      "learning_rate": 6.0019605157854406e-05,
      "loss": 11.895,
      "step": 14684
    },
    {
      "epoch": 2.5678688524590165,
      "grad_norm": 0.6844556386198511,
      "learning_rate": 6.0006625680664195e-05,
      "loss": 12.1023,
      "step": 14685
    },
    {
      "epoch": 2.5680437158469944,
      "grad_norm": 0.6451758027737533,
      "learning_rate": 5.999364700546391e-05,
      "loss": 11.9117,
      "step": 14686
    },
    {
      "epoch": 2.5682185792349728,
      "grad_norm": 0.6719564708616131,
      "learning_rate": 5.9980669132513854e-05,
      "loss": 11.8761,
      "step": 14687
    },
    {
      "epoch": 2.5683934426229507,
      "grad_norm": 0.628131883870098,
      "learning_rate": 5.9967692062074284e-05,
      "loss": 11.9241,
      "step": 14688
    },
    {
      "epoch": 2.568568306010929,
      "grad_norm": 0.578523155744705,
      "learning_rate": 5.995471579440538e-05,
      "loss": 11.6766,
      "step": 14689
    },
    {
      "epoch": 2.568743169398907,
      "grad_norm": 0.6352455726801877,
      "learning_rate": 5.994174032976738e-05,
      "loss": 11.8109,
      "step": 14690
    },
    {
      "epoch": 2.5689180327868852,
      "grad_norm": 0.678548832914041,
      "learning_rate": 5.992876566842047e-05,
      "loss": 11.8118,
      "step": 14691
    },
    {
      "epoch": 2.5690928961748636,
      "grad_norm": 0.5903697547661608,
      "learning_rate": 5.991579181062486e-05,
      "loss": 11.7708,
      "step": 14692
    },
    {
      "epoch": 2.5692677595628415,
      "grad_norm": 0.6447880396093465,
      "learning_rate": 5.99028187566407e-05,
      "loss": 11.6122,
      "step": 14693
    },
    {
      "epoch": 2.56944262295082,
      "grad_norm": 0.6248177122993789,
      "learning_rate": 5.988984650672813e-05,
      "loss": 11.6583,
      "step": 14694
    },
    {
      "epoch": 2.5696174863387977,
      "grad_norm": 0.7158735028468276,
      "learning_rate": 5.9876875061147296e-05,
      "loss": 12.1478,
      "step": 14695
    },
    {
      "epoch": 2.569792349726776,
      "grad_norm": 0.6455010678511778,
      "learning_rate": 5.98639044201583e-05,
      "loss": 11.4825,
      "step": 14696
    },
    {
      "epoch": 2.569967213114754,
      "grad_norm": 0.6657837984577334,
      "learning_rate": 5.985093458402129e-05,
      "loss": 11.8858,
      "step": 14697
    },
    {
      "epoch": 2.5701420765027323,
      "grad_norm": 0.6834080304831316,
      "learning_rate": 5.983796555299628e-05,
      "loss": 11.9074,
      "step": 14698
    },
    {
      "epoch": 2.57031693989071,
      "grad_norm": 0.645907537562242,
      "learning_rate": 5.982499732734339e-05,
      "loss": 11.5821,
      "step": 14699
    },
    {
      "epoch": 2.5704918032786885,
      "grad_norm": 0.6154362865087729,
      "learning_rate": 5.981202990732267e-05,
      "loss": 11.8151,
      "step": 14700
    },
    {
      "epoch": 2.570666666666667,
      "grad_norm": 0.7042089134238196,
      "learning_rate": 5.979906329319408e-05,
      "loss": 11.5967,
      "step": 14701
    },
    {
      "epoch": 2.5708415300546448,
      "grad_norm": 0.7168762244027337,
      "learning_rate": 5.9786097485217774e-05,
      "loss": 11.8485,
      "step": 14702
    },
    {
      "epoch": 2.571016393442623,
      "grad_norm": 0.6662140533295791,
      "learning_rate": 5.977313248365366e-05,
      "loss": 11.874,
      "step": 14703
    },
    {
      "epoch": 2.571191256830601,
      "grad_norm": 0.7031605409234238,
      "learning_rate": 5.9760168288761755e-05,
      "loss": 11.4619,
      "step": 14704
    },
    {
      "epoch": 2.5713661202185794,
      "grad_norm": 0.6413469520368633,
      "learning_rate": 5.9747204900802044e-05,
      "loss": 11.8982,
      "step": 14705
    },
    {
      "epoch": 2.5715409836065573,
      "grad_norm": 0.6403556161077673,
      "learning_rate": 5.973424232003443e-05,
      "loss": 11.5194,
      "step": 14706
    },
    {
      "epoch": 2.5717158469945356,
      "grad_norm": 0.6705412267798841,
      "learning_rate": 5.9721280546718905e-05,
      "loss": 11.8985,
      "step": 14707
    },
    {
      "epoch": 2.5718907103825135,
      "grad_norm": 0.6022601350194989,
      "learning_rate": 5.970831958111538e-05,
      "loss": 11.482,
      "step": 14708
    },
    {
      "epoch": 2.572065573770492,
      "grad_norm": 0.7001718468584903,
      "learning_rate": 5.969535942348379e-05,
      "loss": 11.956,
      "step": 14709
    },
    {
      "epoch": 2.57224043715847,
      "grad_norm": 0.6328774494141804,
      "learning_rate": 5.9682400074083924e-05,
      "loss": 11.8166,
      "step": 14710
    },
    {
      "epoch": 2.572415300546448,
      "grad_norm": 0.6728899085144734,
      "learning_rate": 5.9669441533175785e-05,
      "loss": 11.7135,
      "step": 14711
    },
    {
      "epoch": 2.572590163934426,
      "grad_norm": 0.6570031515844925,
      "learning_rate": 5.965648380101916e-05,
      "loss": 11.5065,
      "step": 14712
    },
    {
      "epoch": 2.5727650273224043,
      "grad_norm": 0.7008148285920179,
      "learning_rate": 5.9643526877873915e-05,
      "loss": 11.7959,
      "step": 14713
    },
    {
      "epoch": 2.5729398907103826,
      "grad_norm": 0.6549808469443736,
      "learning_rate": 5.963057076399987e-05,
      "loss": 11.7624,
      "step": 14714
    },
    {
      "epoch": 2.5731147540983605,
      "grad_norm": 0.6973270729322829,
      "learning_rate": 5.961761545965679e-05,
      "loss": 11.89,
      "step": 14715
    },
    {
      "epoch": 2.573289617486339,
      "grad_norm": 0.6480206314372154,
      "learning_rate": 5.960466096510454e-05,
      "loss": 11.8428,
      "step": 14716
    },
    {
      "epoch": 2.573464480874317,
      "grad_norm": 0.6502931453475851,
      "learning_rate": 5.959170728060286e-05,
      "loss": 11.5231,
      "step": 14717
    },
    {
      "epoch": 2.573639344262295,
      "grad_norm": 0.6791065688267357,
      "learning_rate": 5.957875440641155e-05,
      "loss": 12.0937,
      "step": 14718
    },
    {
      "epoch": 2.5738142076502735,
      "grad_norm": 0.6784277068677628,
      "learning_rate": 5.95658023427903e-05,
      "loss": 11.6659,
      "step": 14719
    },
    {
      "epoch": 2.5739890710382514,
      "grad_norm": 0.6666541250216383,
      "learning_rate": 5.955285108999884e-05,
      "loss": 11.6035,
      "step": 14720
    },
    {
      "epoch": 2.5741639344262293,
      "grad_norm": 0.6260181426666288,
      "learning_rate": 5.953990064829693e-05,
      "loss": 11.6353,
      "step": 14721
    },
    {
      "epoch": 2.5743387978142076,
      "grad_norm": 0.5973036403670813,
      "learning_rate": 5.952695101794423e-05,
      "loss": 11.736,
      "step": 14722
    },
    {
      "epoch": 2.574513661202186,
      "grad_norm": 0.6558607271491157,
      "learning_rate": 5.951400219920046e-05,
      "loss": 11.9898,
      "step": 14723
    },
    {
      "epoch": 2.574688524590164,
      "grad_norm": 0.6256737149694921,
      "learning_rate": 5.9501054192325234e-05,
      "loss": 11.8905,
      "step": 14724
    },
    {
      "epoch": 2.574863387978142,
      "grad_norm": 0.658989986372411,
      "learning_rate": 5.948810699757819e-05,
      "loss": 11.9537,
      "step": 14725
    },
    {
      "epoch": 2.57503825136612,
      "grad_norm": 0.6750129909759068,
      "learning_rate": 5.947516061521902e-05,
      "loss": 11.9412,
      "step": 14726
    },
    {
      "epoch": 2.5752131147540984,
      "grad_norm": 0.6244765393785499,
      "learning_rate": 5.946221504550732e-05,
      "loss": 11.812,
      "step": 14727
    },
    {
      "epoch": 2.5753879781420768,
      "grad_norm": 0.6948897036760172,
      "learning_rate": 5.944927028870266e-05,
      "loss": 12.0722,
      "step": 14728
    },
    {
      "epoch": 2.5755628415300547,
      "grad_norm": 0.6160582817788003,
      "learning_rate": 5.943632634506463e-05,
      "loss": 11.9071,
      "step": 14729
    },
    {
      "epoch": 2.5757377049180326,
      "grad_norm": 0.6760097640711664,
      "learning_rate": 5.942338321485279e-05,
      "loss": 11.6546,
      "step": 14730
    },
    {
      "epoch": 2.575912568306011,
      "grad_norm": 0.6616248089684786,
      "learning_rate": 5.941044089832673e-05,
      "loss": 11.6954,
      "step": 14731
    },
    {
      "epoch": 2.5760874316939892,
      "grad_norm": 0.7195053950027973,
      "learning_rate": 5.939749939574596e-05,
      "loss": 11.9465,
      "step": 14732
    },
    {
      "epoch": 2.576262295081967,
      "grad_norm": 0.6753553461389501,
      "learning_rate": 5.9384558707369987e-05,
      "loss": 11.704,
      "step": 14733
    },
    {
      "epoch": 2.5764371584699455,
      "grad_norm": 0.5971054496173092,
      "learning_rate": 5.9371618833458284e-05,
      "loss": 11.8462,
      "step": 14734
    },
    {
      "epoch": 2.5766120218579234,
      "grad_norm": 0.6483466544209062,
      "learning_rate": 5.935867977427041e-05,
      "loss": 11.8739,
      "step": 14735
    },
    {
      "epoch": 2.5767868852459017,
      "grad_norm": 0.6113216916671534,
      "learning_rate": 5.934574153006579e-05,
      "loss": 11.8299,
      "step": 14736
    },
    {
      "epoch": 2.57696174863388,
      "grad_norm": 0.6419919473181904,
      "learning_rate": 5.9332804101103865e-05,
      "loss": 11.8404,
      "step": 14737
    },
    {
      "epoch": 2.577136612021858,
      "grad_norm": 0.7602814189813087,
      "learning_rate": 5.931986748764409e-05,
      "loss": 11.8768,
      "step": 14738
    },
    {
      "epoch": 2.577311475409836,
      "grad_norm": 0.6737374720017746,
      "learning_rate": 5.9306931689945853e-05,
      "loss": 11.84,
      "step": 14739
    },
    {
      "epoch": 2.577486338797814,
      "grad_norm": 0.624430692977195,
      "learning_rate": 5.929399670826858e-05,
      "loss": 11.8599,
      "step": 14740
    },
    {
      "epoch": 2.5776612021857925,
      "grad_norm": 0.6289358789539974,
      "learning_rate": 5.928106254287171e-05,
      "loss": 11.691,
      "step": 14741
    },
    {
      "epoch": 2.5778360655737704,
      "grad_norm": 0.6012254845213966,
      "learning_rate": 5.926812919401452e-05,
      "loss": 11.9309,
      "step": 14742
    },
    {
      "epoch": 2.5780109289617488,
      "grad_norm": 0.6333766362392569,
      "learning_rate": 5.92551966619564e-05,
      "loss": 11.871,
      "step": 14743
    },
    {
      "epoch": 2.5781857923497267,
      "grad_norm": 0.6810190291006945,
      "learning_rate": 5.924226494695668e-05,
      "loss": 12.0449,
      "step": 14744
    },
    {
      "epoch": 2.578360655737705,
      "grad_norm": 0.7310768929948039,
      "learning_rate": 5.922933404927473e-05,
      "loss": 11.8914,
      "step": 14745
    },
    {
      "epoch": 2.578535519125683,
      "grad_norm": 0.6356706713395579,
      "learning_rate": 5.9216403969169785e-05,
      "loss": 11.9806,
      "step": 14746
    },
    {
      "epoch": 2.5787103825136612,
      "grad_norm": 0.6187693100931921,
      "learning_rate": 5.920347470690118e-05,
      "loss": 11.9067,
      "step": 14747
    },
    {
      "epoch": 2.578885245901639,
      "grad_norm": 0.6417133400902922,
      "learning_rate": 5.9190546262728174e-05,
      "loss": 11.7737,
      "step": 14748
    },
    {
      "epoch": 2.5790601092896175,
      "grad_norm": 0.6719762926645192,
      "learning_rate": 5.917761863690996e-05,
      "loss": 12.0785,
      "step": 14749
    },
    {
      "epoch": 2.579234972677596,
      "grad_norm": 0.7076408107392648,
      "learning_rate": 5.91646918297059e-05,
      "loss": 12.1384,
      "step": 14750
    },
    {
      "epoch": 2.5794098360655737,
      "grad_norm": 0.6758902143493999,
      "learning_rate": 5.9151765841375115e-05,
      "loss": 11.9477,
      "step": 14751
    },
    {
      "epoch": 2.579584699453552,
      "grad_norm": 0.6480564360813802,
      "learning_rate": 5.913884067217685e-05,
      "loss": 11.9027,
      "step": 14752
    },
    {
      "epoch": 2.57975956284153,
      "grad_norm": 0.6968179936195809,
      "learning_rate": 5.9125916322370324e-05,
      "loss": 11.7737,
      "step": 14753
    },
    {
      "epoch": 2.5799344262295083,
      "grad_norm": 0.6916452272485389,
      "learning_rate": 5.911299279221463e-05,
      "loss": 11.6602,
      "step": 14754
    },
    {
      "epoch": 2.580109289617486,
      "grad_norm": 0.6354745132374762,
      "learning_rate": 5.910007008196898e-05,
      "loss": 11.844,
      "step": 14755
    },
    {
      "epoch": 2.5802841530054645,
      "grad_norm": 0.7280888336283194,
      "learning_rate": 5.908714819189252e-05,
      "loss": 11.8143,
      "step": 14756
    },
    {
      "epoch": 2.5804590163934424,
      "grad_norm": 0.6739257594538037,
      "learning_rate": 5.9074227122244356e-05,
      "loss": 12.0599,
      "step": 14757
    },
    {
      "epoch": 2.580633879781421,
      "grad_norm": 0.7125861860684068,
      "learning_rate": 5.906130687328355e-05,
      "loss": 11.8158,
      "step": 14758
    },
    {
      "epoch": 2.580808743169399,
      "grad_norm": 0.6776724813968431,
      "learning_rate": 5.90483874452693e-05,
      "loss": 11.4821,
      "step": 14759
    },
    {
      "epoch": 2.580983606557377,
      "grad_norm": 0.6203360425222244,
      "learning_rate": 5.9035468838460594e-05,
      "loss": 11.5718,
      "step": 14760
    },
    {
      "epoch": 2.581158469945355,
      "grad_norm": 0.6402570726456944,
      "learning_rate": 5.9022551053116513e-05,
      "loss": 11.7653,
      "step": 14761
    },
    {
      "epoch": 2.5813333333333333,
      "grad_norm": 0.6559066097154296,
      "learning_rate": 5.900963408949614e-05,
      "loss": 11.8555,
      "step": 14762
    },
    {
      "epoch": 2.5815081967213116,
      "grad_norm": 0.6932100678554117,
      "learning_rate": 5.899671794785839e-05,
      "loss": 11.8242,
      "step": 14763
    },
    {
      "epoch": 2.5816830601092895,
      "grad_norm": 0.6687192417385316,
      "learning_rate": 5.898380262846239e-05,
      "loss": 12.0407,
      "step": 14764
    },
    {
      "epoch": 2.581857923497268,
      "grad_norm": 0.6626172579848916,
      "learning_rate": 5.897088813156707e-05,
      "loss": 11.8445,
      "step": 14765
    },
    {
      "epoch": 2.5820327868852457,
      "grad_norm": 0.7051231942128117,
      "learning_rate": 5.895797445743145e-05,
      "loss": 11.5688,
      "step": 14766
    },
    {
      "epoch": 2.582207650273224,
      "grad_norm": 0.6322282454612158,
      "learning_rate": 5.894506160631442e-05,
      "loss": 11.6141,
      "step": 14767
    },
    {
      "epoch": 2.5823825136612024,
      "grad_norm": 0.5811573796661592,
      "learning_rate": 5.893214957847495e-05,
      "loss": 11.9592,
      "step": 14768
    },
    {
      "epoch": 2.5825573770491803,
      "grad_norm": 0.6512623185017129,
      "learning_rate": 5.8919238374171994e-05,
      "loss": 11.9163,
      "step": 14769
    },
    {
      "epoch": 2.582732240437158,
      "grad_norm": 0.6372616067750444,
      "learning_rate": 5.890632799366444e-05,
      "loss": 11.8472,
      "step": 14770
    },
    {
      "epoch": 2.5829071038251366,
      "grad_norm": 0.6697666063915599,
      "learning_rate": 5.889341843721121e-05,
      "loss": 11.8185,
      "step": 14771
    },
    {
      "epoch": 2.583081967213115,
      "grad_norm": 0.638166506595419,
      "learning_rate": 5.888050970507114e-05,
      "loss": 11.9151,
      "step": 14772
    },
    {
      "epoch": 2.583256830601093,
      "grad_norm": 0.641871592626562,
      "learning_rate": 5.886760179750308e-05,
      "loss": 11.8976,
      "step": 14773
    },
    {
      "epoch": 2.583431693989071,
      "grad_norm": 0.6977738372849284,
      "learning_rate": 5.88546947147659e-05,
      "loss": 11.9787,
      "step": 14774
    },
    {
      "epoch": 2.583606557377049,
      "grad_norm": 0.7477787118025436,
      "learning_rate": 5.8841788457118464e-05,
      "loss": 11.665,
      "step": 14775
    },
    {
      "epoch": 2.5837814207650274,
      "grad_norm": 0.710720587812521,
      "learning_rate": 5.882888302481952e-05,
      "loss": 11.7777,
      "step": 14776
    },
    {
      "epoch": 2.5839562841530057,
      "grad_norm": 0.7461583949618911,
      "learning_rate": 5.881597841812787e-05,
      "loss": 11.604,
      "step": 14777
    },
    {
      "epoch": 2.5841311475409836,
      "grad_norm": 0.6529342332877853,
      "learning_rate": 5.88030746373023e-05,
      "loss": 11.9984,
      "step": 14778
    },
    {
      "epoch": 2.5843060109289615,
      "grad_norm": 0.7946788809086844,
      "learning_rate": 5.879017168260158e-05,
      "loss": 11.6277,
      "step": 14779
    },
    {
      "epoch": 2.58448087431694,
      "grad_norm": 0.7198594939969748,
      "learning_rate": 5.8777269554284475e-05,
      "loss": 11.9132,
      "step": 14780
    },
    {
      "epoch": 2.584655737704918,
      "grad_norm": 0.7143266353406295,
      "learning_rate": 5.876436825260967e-05,
      "loss": 11.9425,
      "step": 14781
    },
    {
      "epoch": 2.584830601092896,
      "grad_norm": 0.7636972377039721,
      "learning_rate": 5.875146777783587e-05,
      "loss": 11.9608,
      "step": 14782
    },
    {
      "epoch": 2.5850054644808744,
      "grad_norm": 0.6932748493019498,
      "learning_rate": 5.873856813022181e-05,
      "loss": 12.0,
      "step": 14783
    },
    {
      "epoch": 2.5851803278688523,
      "grad_norm": 0.6153429614147043,
      "learning_rate": 5.872566931002617e-05,
      "loss": 11.5539,
      "step": 14784
    },
    {
      "epoch": 2.5853551912568307,
      "grad_norm": 0.6696539286228248,
      "learning_rate": 5.8712771317507555e-05,
      "loss": 11.4872,
      "step": 14785
    },
    {
      "epoch": 2.585530054644809,
      "grad_norm": 0.7535237390628083,
      "learning_rate": 5.869987415292466e-05,
      "loss": 11.95,
      "step": 14786
    },
    {
      "epoch": 2.585704918032787,
      "grad_norm": 0.6752551413716015,
      "learning_rate": 5.868697781653607e-05,
      "loss": 11.6628,
      "step": 14787
    },
    {
      "epoch": 2.585879781420765,
      "grad_norm": 0.7285140500124584,
      "learning_rate": 5.8674082308600434e-05,
      "loss": 11.9673,
      "step": 14788
    },
    {
      "epoch": 2.586054644808743,
      "grad_norm": 0.6555523278387364,
      "learning_rate": 5.866118762937637e-05,
      "loss": 11.7245,
      "step": 14789
    },
    {
      "epoch": 2.5862295081967215,
      "grad_norm": 0.653642423029489,
      "learning_rate": 5.86482937791224e-05,
      "loss": 11.7478,
      "step": 14790
    },
    {
      "epoch": 2.5864043715846994,
      "grad_norm": 0.6180022940352361,
      "learning_rate": 5.8635400758097094e-05,
      "loss": 11.7375,
      "step": 14791
    },
    {
      "epoch": 2.5865792349726777,
      "grad_norm": 0.631017207376954,
      "learning_rate": 5.8622508566559e-05,
      "loss": 11.8931,
      "step": 14792
    },
    {
      "epoch": 2.5867540983606556,
      "grad_norm": 0.602152811159909,
      "learning_rate": 5.8609617204766686e-05,
      "loss": 11.9094,
      "step": 14793
    },
    {
      "epoch": 2.586928961748634,
      "grad_norm": 0.6990621267349818,
      "learning_rate": 5.859672667297862e-05,
      "loss": 11.8455,
      "step": 14794
    },
    {
      "epoch": 2.587103825136612,
      "grad_norm": 0.6223193072287926,
      "learning_rate": 5.858383697145331e-05,
      "loss": 12.0103,
      "step": 14795
    },
    {
      "epoch": 2.58727868852459,
      "grad_norm": 0.6385824438158955,
      "learning_rate": 5.857094810044924e-05,
      "loss": 11.8761,
      "step": 14796
    },
    {
      "epoch": 2.587453551912568,
      "grad_norm": 0.7474284586617315,
      "learning_rate": 5.8558060060224817e-05,
      "loss": 11.8889,
      "step": 14797
    },
    {
      "epoch": 2.5876284153005464,
      "grad_norm": 0.6001232926874064,
      "learning_rate": 5.8545172851038586e-05,
      "loss": 11.7192,
      "step": 14798
    },
    {
      "epoch": 2.5878032786885248,
      "grad_norm": 0.8268786423503065,
      "learning_rate": 5.85322864731489e-05,
      "loss": 11.8672,
      "step": 14799
    },
    {
      "epoch": 2.5879781420765027,
      "grad_norm": 0.6912198760412895,
      "learning_rate": 5.85194009268142e-05,
      "loss": 11.8681,
      "step": 14800
    },
    {
      "epoch": 2.588153005464481,
      "grad_norm": 0.597693803773549,
      "learning_rate": 5.8506516212292894e-05,
      "loss": 11.8646,
      "step": 14801
    },
    {
      "epoch": 2.588327868852459,
      "grad_norm": 0.7341213067272556,
      "learning_rate": 5.849363232984331e-05,
      "loss": 11.844,
      "step": 14802
    },
    {
      "epoch": 2.5885027322404373,
      "grad_norm": 0.6085413022821476,
      "learning_rate": 5.848074927972385e-05,
      "loss": 11.7036,
      "step": 14803
    },
    {
      "epoch": 2.588677595628415,
      "grad_norm": 0.6383088140814512,
      "learning_rate": 5.8467867062192824e-05,
      "loss": 11.8142,
      "step": 14804
    },
    {
      "epoch": 2.5888524590163935,
      "grad_norm": 0.6127357648666217,
      "learning_rate": 5.8454985677508635e-05,
      "loss": 11.8331,
      "step": 14805
    },
    {
      "epoch": 2.5890273224043714,
      "grad_norm": 0.6686752003971801,
      "learning_rate": 5.844210512592947e-05,
      "loss": 12.0018,
      "step": 14806
    },
    {
      "epoch": 2.5892021857923497,
      "grad_norm": 0.6295085441939335,
      "learning_rate": 5.842922540771376e-05,
      "loss": 11.8217,
      "step": 14807
    },
    {
      "epoch": 2.589377049180328,
      "grad_norm": 0.6497814594389353,
      "learning_rate": 5.841634652311969e-05,
      "loss": 11.8488,
      "step": 14808
    },
    {
      "epoch": 2.589551912568306,
      "grad_norm": 0.6745949087187152,
      "learning_rate": 5.8403468472405565e-05,
      "loss": 11.9551,
      "step": 14809
    },
    {
      "epoch": 2.5897267759562843,
      "grad_norm": 0.6653957534659556,
      "learning_rate": 5.8390591255829644e-05,
      "loss": 11.864,
      "step": 14810
    },
    {
      "epoch": 2.589901639344262,
      "grad_norm": 0.7170477227075949,
      "learning_rate": 5.837771487365005e-05,
      "loss": 11.8975,
      "step": 14811
    },
    {
      "epoch": 2.5900765027322405,
      "grad_norm": 0.7042404971572177,
      "learning_rate": 5.836483932612511e-05,
      "loss": 11.7892,
      "step": 14812
    },
    {
      "epoch": 2.5902513661202184,
      "grad_norm": 0.6603391171964472,
      "learning_rate": 5.835196461351302e-05,
      "loss": 11.7038,
      "step": 14813
    },
    {
      "epoch": 2.590426229508197,
      "grad_norm": 0.7149220171729858,
      "learning_rate": 5.8339090736071866e-05,
      "loss": 11.8478,
      "step": 14814
    },
    {
      "epoch": 2.5906010928961747,
      "grad_norm": 0.707224879281045,
      "learning_rate": 5.832621769405987e-05,
      "loss": 11.8964,
      "step": 14815
    },
    {
      "epoch": 2.590775956284153,
      "grad_norm": 0.6446741186893535,
      "learning_rate": 5.831334548773513e-05,
      "loss": 11.8272,
      "step": 14816
    },
    {
      "epoch": 2.5909508196721314,
      "grad_norm": 0.7478016863173005,
      "learning_rate": 5.830047411735588e-05,
      "loss": 11.8212,
      "step": 14817
    },
    {
      "epoch": 2.5911256830601093,
      "grad_norm": 0.7608511547090093,
      "learning_rate": 5.828760358318012e-05,
      "loss": 11.6635,
      "step": 14818
    },
    {
      "epoch": 2.591300546448087,
      "grad_norm": 0.6205265100343275,
      "learning_rate": 5.827473388546598e-05,
      "loss": 11.9755,
      "step": 14819
    },
    {
      "epoch": 2.5914754098360655,
      "grad_norm": 0.7575651172089268,
      "learning_rate": 5.826186502447153e-05,
      "loss": 11.629,
      "step": 14820
    },
    {
      "epoch": 2.591650273224044,
      "grad_norm": 0.7457701760367988,
      "learning_rate": 5.8248997000454805e-05,
      "loss": 11.6714,
      "step": 14821
    },
    {
      "epoch": 2.5918251366120217,
      "grad_norm": 0.621003549107355,
      "learning_rate": 5.823612981367398e-05,
      "loss": 11.8252,
      "step": 14822
    },
    {
      "epoch": 2.592,
      "grad_norm": 0.6293966588336801,
      "learning_rate": 5.822326346438693e-05,
      "loss": 11.9949,
      "step": 14823
    },
    {
      "epoch": 2.592174863387978,
      "grad_norm": 0.7426009331215088,
      "learning_rate": 5.821039795285172e-05,
      "loss": 12.0207,
      "step": 14824
    },
    {
      "epoch": 2.5923497267759563,
      "grad_norm": 0.6901306190305713,
      "learning_rate": 5.819753327932634e-05,
      "loss": 11.8786,
      "step": 14825
    },
    {
      "epoch": 2.5925245901639347,
      "grad_norm": 0.7395076611909255,
      "learning_rate": 5.818466944406877e-05,
      "loss": 11.7822,
      "step": 14826
    },
    {
      "epoch": 2.5926994535519126,
      "grad_norm": 0.7259834792465503,
      "learning_rate": 5.817180644733697e-05,
      "loss": 11.6876,
      "step": 14827
    },
    {
      "epoch": 2.5928743169398905,
      "grad_norm": 0.639783874133822,
      "learning_rate": 5.8158944289388885e-05,
      "loss": 11.8981,
      "step": 14828
    },
    {
      "epoch": 2.593049180327869,
      "grad_norm": 0.7397575568043064,
      "learning_rate": 5.814608297048243e-05,
      "loss": 12.0059,
      "step": 14829
    },
    {
      "epoch": 2.593224043715847,
      "grad_norm": 0.6990712705467376,
      "learning_rate": 5.813322249087547e-05,
      "loss": 11.7997,
      "step": 14830
    },
    {
      "epoch": 2.593398907103825,
      "grad_norm": 0.6154617305196407,
      "learning_rate": 5.812036285082604e-05,
      "loss": 11.989,
      "step": 14831
    },
    {
      "epoch": 2.5935737704918034,
      "grad_norm": 0.7148827761349227,
      "learning_rate": 5.810750405059188e-05,
      "loss": 11.7794,
      "step": 14832
    },
    {
      "epoch": 2.5937486338797813,
      "grad_norm": 0.7625027799725962,
      "learning_rate": 5.809464609043088e-05,
      "loss": 11.8462,
      "step": 14833
    },
    {
      "epoch": 2.5939234972677596,
      "grad_norm": 0.7109965298855074,
      "learning_rate": 5.80817889706009e-05,
      "loss": 11.829,
      "step": 14834
    },
    {
      "epoch": 2.594098360655738,
      "grad_norm": 0.6981014643201288,
      "learning_rate": 5.8068932691359753e-05,
      "loss": 11.9157,
      "step": 14835
    },
    {
      "epoch": 2.594273224043716,
      "grad_norm": 0.757310594513065,
      "learning_rate": 5.805607725296527e-05,
      "loss": 11.8711,
      "step": 14836
    },
    {
      "epoch": 2.5944480874316938,
      "grad_norm": 0.7102066434909501,
      "learning_rate": 5.80432226556752e-05,
      "loss": 11.8279,
      "step": 14837
    },
    {
      "epoch": 2.594622950819672,
      "grad_norm": 0.6635917164100756,
      "learning_rate": 5.803036889974736e-05,
      "loss": 11.9154,
      "step": 14838
    },
    {
      "epoch": 2.5947978142076504,
      "grad_norm": 0.6712715716243349,
      "learning_rate": 5.8017515985439465e-05,
      "loss": 12.0583,
      "step": 14839
    },
    {
      "epoch": 2.5949726775956283,
      "grad_norm": 0.6978080253076524,
      "learning_rate": 5.8004663913009295e-05,
      "loss": 12.0303,
      "step": 14840
    },
    {
      "epoch": 2.5951475409836067,
      "grad_norm": 0.6093326985879207,
      "learning_rate": 5.799181268271454e-05,
      "loss": 11.7545,
      "step": 14841
    },
    {
      "epoch": 2.5953224043715846,
      "grad_norm": 0.7290875275778075,
      "learning_rate": 5.797896229481293e-05,
      "loss": 11.8258,
      "step": 14842
    },
    {
      "epoch": 2.595497267759563,
      "grad_norm": 0.6633462340225915,
      "learning_rate": 5.796611274956216e-05,
      "loss": 11.633,
      "step": 14843
    },
    {
      "epoch": 2.5956721311475413,
      "grad_norm": 0.7059115776756505,
      "learning_rate": 5.795326404721988e-05,
      "loss": 11.92,
      "step": 14844
    },
    {
      "epoch": 2.595846994535519,
      "grad_norm": 0.7031319126243045,
      "learning_rate": 5.7940416188043756e-05,
      "loss": 11.8399,
      "step": 14845
    },
    {
      "epoch": 2.596021857923497,
      "grad_norm": 0.6980619195821424,
      "learning_rate": 5.792756917229143e-05,
      "loss": 12.0129,
      "step": 14846
    },
    {
      "epoch": 2.5961967213114754,
      "grad_norm": 0.6577068540037231,
      "learning_rate": 5.791472300022052e-05,
      "loss": 11.846,
      "step": 14847
    },
    {
      "epoch": 2.5963715846994537,
      "grad_norm": 0.7472152870190326,
      "learning_rate": 5.7901877672088625e-05,
      "loss": 11.9539,
      "step": 14848
    },
    {
      "epoch": 2.5965464480874316,
      "grad_norm": 0.6435456645228704,
      "learning_rate": 5.788903318815335e-05,
      "loss": 11.6382,
      "step": 14849
    },
    {
      "epoch": 2.59672131147541,
      "grad_norm": 0.696297042591123,
      "learning_rate": 5.787618954867224e-05,
      "loss": 11.8845,
      "step": 14850
    },
    {
      "epoch": 2.596896174863388,
      "grad_norm": 0.650131271430085,
      "learning_rate": 5.786334675390289e-05,
      "loss": 11.82,
      "step": 14851
    },
    {
      "epoch": 2.597071038251366,
      "grad_norm": 0.6389060309282744,
      "learning_rate": 5.785050480410281e-05,
      "loss": 11.6142,
      "step": 14852
    },
    {
      "epoch": 2.597245901639344,
      "grad_norm": 0.6374541604149949,
      "learning_rate": 5.783766369952952e-05,
      "loss": 11.7597,
      "step": 14853
    },
    {
      "epoch": 2.5974207650273224,
      "grad_norm": 0.6351147172896502,
      "learning_rate": 5.7824823440440536e-05,
      "loss": 11.8392,
      "step": 14854
    },
    {
      "epoch": 2.5975956284153003,
      "grad_norm": 0.733709069502063,
      "learning_rate": 5.7811984027093315e-05,
      "loss": 11.8344,
      "step": 14855
    },
    {
      "epoch": 2.5977704918032787,
      "grad_norm": 0.6162434565727205,
      "learning_rate": 5.7799145459745364e-05,
      "loss": 11.9602,
      "step": 14856
    },
    {
      "epoch": 2.597945355191257,
      "grad_norm": 0.6217652833754284,
      "learning_rate": 5.7786307738654125e-05,
      "loss": 11.8951,
      "step": 14857
    },
    {
      "epoch": 2.598120218579235,
      "grad_norm": 0.6106397227541707,
      "learning_rate": 5.777347086407703e-05,
      "loss": 11.9267,
      "step": 14858
    },
    {
      "epoch": 2.5982950819672133,
      "grad_norm": 0.6601093817255567,
      "learning_rate": 5.776063483627148e-05,
      "loss": 11.9046,
      "step": 14859
    },
    {
      "epoch": 2.598469945355191,
      "grad_norm": 0.6168595036788287,
      "learning_rate": 5.7747799655494905e-05,
      "loss": 11.9861,
      "step": 14860
    },
    {
      "epoch": 2.5986448087431695,
      "grad_norm": 0.656207230165068,
      "learning_rate": 5.773496532200468e-05,
      "loss": 11.9249,
      "step": 14861
    },
    {
      "epoch": 2.5988196721311474,
      "grad_norm": 0.6524152902823149,
      "learning_rate": 5.772213183605817e-05,
      "loss": 11.9017,
      "step": 14862
    },
    {
      "epoch": 2.5989945355191257,
      "grad_norm": 0.7271567431239155,
      "learning_rate": 5.7709299197912715e-05,
      "loss": 11.8936,
      "step": 14863
    },
    {
      "epoch": 2.5991693989071036,
      "grad_norm": 0.6071493982389209,
      "learning_rate": 5.769646740782566e-05,
      "loss": 11.725,
      "step": 14864
    },
    {
      "epoch": 2.599344262295082,
      "grad_norm": 0.6174641203313724,
      "learning_rate": 5.768363646605435e-05,
      "loss": 12.0337,
      "step": 14865
    },
    {
      "epoch": 2.5995191256830603,
      "grad_norm": 0.5592669175587895,
      "learning_rate": 5.767080637285602e-05,
      "loss": 11.5912,
      "step": 14866
    },
    {
      "epoch": 2.599693989071038,
      "grad_norm": 0.6385110883639254,
      "learning_rate": 5.765797712848802e-05,
      "loss": 11.7052,
      "step": 14867
    },
    {
      "epoch": 2.5998688524590166,
      "grad_norm": 0.62820122054096,
      "learning_rate": 5.764514873320761e-05,
      "loss": 11.8163,
      "step": 14868
    },
    {
      "epoch": 2.6000437158469945,
      "grad_norm": 0.6783108884743478,
      "learning_rate": 5.763232118727194e-05,
      "loss": 11.7474,
      "step": 14869
    },
    {
      "epoch": 2.600218579234973,
      "grad_norm": 0.6672113569584227,
      "learning_rate": 5.7619494490938355e-05,
      "loss": 11.86,
      "step": 14870
    },
    {
      "epoch": 2.6003934426229507,
      "grad_norm": 0.6410674109031473,
      "learning_rate": 5.760666864446403e-05,
      "loss": 11.6691,
      "step": 14871
    },
    {
      "epoch": 2.600568306010929,
      "grad_norm": 0.5942607706780025,
      "learning_rate": 5.7593843648106164e-05,
      "loss": 11.992,
      "step": 14872
    },
    {
      "epoch": 2.600743169398907,
      "grad_norm": 0.6012740217531425,
      "learning_rate": 5.758101950212198e-05,
      "loss": 11.9399,
      "step": 14873
    },
    {
      "epoch": 2.6009180327868853,
      "grad_norm": 0.6163849301455542,
      "learning_rate": 5.756819620676852e-05,
      "loss": 11.6578,
      "step": 14874
    },
    {
      "epoch": 2.6010928961748636,
      "grad_norm": 0.5936739030959928,
      "learning_rate": 5.755537376230304e-05,
      "loss": 11.4804,
      "step": 14875
    },
    {
      "epoch": 2.6012677595628415,
      "grad_norm": 0.6023475503433465,
      "learning_rate": 5.7542552168982635e-05,
      "loss": 11.6888,
      "step": 14876
    },
    {
      "epoch": 2.6014426229508194,
      "grad_norm": 0.670237804318615,
      "learning_rate": 5.752973142706447e-05,
      "loss": 11.8572,
      "step": 14877
    },
    {
      "epoch": 2.6016174863387977,
      "grad_norm": 0.7356453335151805,
      "learning_rate": 5.751691153680552e-05,
      "loss": 11.9242,
      "step": 14878
    },
    {
      "epoch": 2.601792349726776,
      "grad_norm": 0.672007973156792,
      "learning_rate": 5.750409249846296e-05,
      "loss": 11.7172,
      "step": 14879
    },
    {
      "epoch": 2.601967213114754,
      "grad_norm": 0.6788306117062888,
      "learning_rate": 5.7491274312293816e-05,
      "loss": 11.559,
      "step": 14880
    },
    {
      "epoch": 2.6021420765027323,
      "grad_norm": 0.6467332705841978,
      "learning_rate": 5.747845697855516e-05,
      "loss": 11.9949,
      "step": 14881
    },
    {
      "epoch": 2.6023169398907102,
      "grad_norm": 0.6909292114049086,
      "learning_rate": 5.7465640497504026e-05,
      "loss": 11.7289,
      "step": 14882
    },
    {
      "epoch": 2.6024918032786886,
      "grad_norm": 0.714363253407396,
      "learning_rate": 5.7452824869397326e-05,
      "loss": 12.0269,
      "step": 14883
    },
    {
      "epoch": 2.602666666666667,
      "grad_norm": 0.7128538441323775,
      "learning_rate": 5.7440010094492156e-05,
      "loss": 12.015,
      "step": 14884
    },
    {
      "epoch": 2.602841530054645,
      "grad_norm": 0.6702433875901751,
      "learning_rate": 5.742719617304547e-05,
      "loss": 11.7787,
      "step": 14885
    },
    {
      "epoch": 2.6030163934426227,
      "grad_norm": 0.6553079666228501,
      "learning_rate": 5.741438310531426e-05,
      "loss": 11.6124,
      "step": 14886
    },
    {
      "epoch": 2.603191256830601,
      "grad_norm": 0.6540293632100527,
      "learning_rate": 5.740157089155538e-05,
      "loss": 11.9301,
      "step": 14887
    },
    {
      "epoch": 2.6033661202185794,
      "grad_norm": 0.643074371596912,
      "learning_rate": 5.738875953202574e-05,
      "loss": 11.819,
      "step": 14888
    },
    {
      "epoch": 2.6035409836065573,
      "grad_norm": 0.6797218994219757,
      "learning_rate": 5.7375949026982365e-05,
      "loss": 11.6927,
      "step": 14889
    },
    {
      "epoch": 2.6037158469945356,
      "grad_norm": 0.7213157075959173,
      "learning_rate": 5.736313937668207e-05,
      "loss": 11.5682,
      "step": 14890
    },
    {
      "epoch": 2.6038907103825135,
      "grad_norm": 0.644373458801331,
      "learning_rate": 5.735033058138179e-05,
      "loss": 11.6239,
      "step": 14891
    },
    {
      "epoch": 2.604065573770492,
      "grad_norm": 0.6742430128825453,
      "learning_rate": 5.733752264133831e-05,
      "loss": 11.837,
      "step": 14892
    },
    {
      "epoch": 2.60424043715847,
      "grad_norm": 0.6653417953785771,
      "learning_rate": 5.732471555680843e-05,
      "loss": 11.8594,
      "step": 14893
    },
    {
      "epoch": 2.604415300546448,
      "grad_norm": 0.6677593869619532,
      "learning_rate": 5.7311909328049085e-05,
      "loss": 11.993,
      "step": 14894
    },
    {
      "epoch": 2.604590163934426,
      "grad_norm": 0.9396619660853353,
      "learning_rate": 5.729910395531706e-05,
      "loss": 12.0287,
      "step": 14895
    },
    {
      "epoch": 2.6047650273224043,
      "grad_norm": 0.663967122805235,
      "learning_rate": 5.728629943886907e-05,
      "loss": 11.6607,
      "step": 14896
    },
    {
      "epoch": 2.6049398907103827,
      "grad_norm": 0.6867444249386342,
      "learning_rate": 5.727349577896194e-05,
      "loss": 12.0312,
      "step": 14897
    },
    {
      "epoch": 2.6051147540983606,
      "grad_norm": 0.6441113257959278,
      "learning_rate": 5.726069297585235e-05,
      "loss": 11.7954,
      "step": 14898
    },
    {
      "epoch": 2.605289617486339,
      "grad_norm": 0.6310155063612751,
      "learning_rate": 5.724789102979714e-05,
      "loss": 11.6125,
      "step": 14899
    },
    {
      "epoch": 2.605464480874317,
      "grad_norm": 0.6264057622494826,
      "learning_rate": 5.723508994105302e-05,
      "loss": 11.7685,
      "step": 14900
    },
    {
      "epoch": 2.605639344262295,
      "grad_norm": 0.6679318103623371,
      "learning_rate": 5.722228970987661e-05,
      "loss": 11.9852,
      "step": 14901
    },
    {
      "epoch": 2.6058142076502735,
      "grad_norm": 0.6278137100228978,
      "learning_rate": 5.72094903365246e-05,
      "loss": 11.7207,
      "step": 14902
    },
    {
      "epoch": 2.6059890710382514,
      "grad_norm": 0.6083556784048697,
      "learning_rate": 5.719669182125373e-05,
      "loss": 11.9828,
      "step": 14903
    },
    {
      "epoch": 2.6061639344262293,
      "grad_norm": 0.6637616189765774,
      "learning_rate": 5.718389416432065e-05,
      "loss": 11.9326,
      "step": 14904
    },
    {
      "epoch": 2.6063387978142076,
      "grad_norm": 0.626575561658117,
      "learning_rate": 5.717109736598192e-05,
      "loss": 11.9794,
      "step": 14905
    },
    {
      "epoch": 2.606513661202186,
      "grad_norm": 0.6475548788475078,
      "learning_rate": 5.715830142649419e-05,
      "loss": 11.5951,
      "step": 14906
    },
    {
      "epoch": 2.606688524590164,
      "grad_norm": 0.7209585455708688,
      "learning_rate": 5.714550634611401e-05,
      "loss": 11.9084,
      "step": 14907
    },
    {
      "epoch": 2.606863387978142,
      "grad_norm": 0.7221636246365595,
      "learning_rate": 5.7132712125098054e-05,
      "loss": 12.0103,
      "step": 14908
    },
    {
      "epoch": 2.60703825136612,
      "grad_norm": 0.597894472801754,
      "learning_rate": 5.711991876370289e-05,
      "loss": 11.771,
      "step": 14909
    },
    {
      "epoch": 2.6072131147540984,
      "grad_norm": 0.6773697059066802,
      "learning_rate": 5.710712626218495e-05,
      "loss": 11.5149,
      "step": 14910
    },
    {
      "epoch": 2.6073879781420763,
      "grad_norm": 0.675058655733404,
      "learning_rate": 5.7094334620800835e-05,
      "loss": 11.8615,
      "step": 14911
    },
    {
      "epoch": 2.6075628415300547,
      "grad_norm": 0.6693616813611899,
      "learning_rate": 5.7081543839807014e-05,
      "loss": 11.911,
      "step": 14912
    },
    {
      "epoch": 2.6077377049180326,
      "grad_norm": 0.6414710715445738,
      "learning_rate": 5.7068753919460095e-05,
      "loss": 11.9225,
      "step": 14913
    },
    {
      "epoch": 2.607912568306011,
      "grad_norm": 0.6041264012925321,
      "learning_rate": 5.705596486001644e-05,
      "loss": 11.9409,
      "step": 14914
    },
    {
      "epoch": 2.6080874316939893,
      "grad_norm": 0.6659969879749522,
      "learning_rate": 5.704317666173255e-05,
      "loss": 11.7045,
      "step": 14915
    },
    {
      "epoch": 2.608262295081967,
      "grad_norm": 0.6597914988179504,
      "learning_rate": 5.703038932486484e-05,
      "loss": 11.7104,
      "step": 14916
    },
    {
      "epoch": 2.6084371584699455,
      "grad_norm": 0.7167261872627546,
      "learning_rate": 5.701760284966974e-05,
      "loss": 11.7824,
      "step": 14917
    },
    {
      "epoch": 2.6086120218579234,
      "grad_norm": 0.7064712965503676,
      "learning_rate": 5.700481723640374e-05,
      "loss": 11.9333,
      "step": 14918
    },
    {
      "epoch": 2.6087868852459017,
      "grad_norm": 0.6340372148023897,
      "learning_rate": 5.6992032485323146e-05,
      "loss": 11.7958,
      "step": 14919
    },
    {
      "epoch": 2.6089617486338796,
      "grad_norm": 0.7744916015137919,
      "learning_rate": 5.697924859668433e-05,
      "loss": 11.9928,
      "step": 14920
    },
    {
      "epoch": 2.609136612021858,
      "grad_norm": 0.6797289941251856,
      "learning_rate": 5.696646557074369e-05,
      "loss": 11.907,
      "step": 14921
    },
    {
      "epoch": 2.609311475409836,
      "grad_norm": 0.7767196174768222,
      "learning_rate": 5.6953683407757544e-05,
      "loss": 11.8046,
      "step": 14922
    },
    {
      "epoch": 2.609486338797814,
      "grad_norm": 0.6183198723813443,
      "learning_rate": 5.6940902107982194e-05,
      "loss": 11.8516,
      "step": 14923
    },
    {
      "epoch": 2.6096612021857926,
      "grad_norm": 0.6672926803748201,
      "learning_rate": 5.6928121671673985e-05,
      "loss": 11.3917,
      "step": 14924
    },
    {
      "epoch": 2.6098360655737705,
      "grad_norm": 0.6358932135152601,
      "learning_rate": 5.691534209908919e-05,
      "loss": 11.9917,
      "step": 14925
    },
    {
      "epoch": 2.610010928961749,
      "grad_norm": 0.6701079393295047,
      "learning_rate": 5.6902563390484023e-05,
      "loss": 11.9811,
      "step": 14926
    },
    {
      "epoch": 2.6101857923497267,
      "grad_norm": 0.6092793546346689,
      "learning_rate": 5.688978554611486e-05,
      "loss": 11.7083,
      "step": 14927
    },
    {
      "epoch": 2.610360655737705,
      "grad_norm": 0.6174053253636199,
      "learning_rate": 5.687700856623783e-05,
      "loss": 11.6879,
      "step": 14928
    },
    {
      "epoch": 2.610535519125683,
      "grad_norm": 0.691458724724524,
      "learning_rate": 5.6864232451109186e-05,
      "loss": 11.8658,
      "step": 14929
    },
    {
      "epoch": 2.6107103825136613,
      "grad_norm": 0.6969545448081556,
      "learning_rate": 5.685145720098513e-05,
      "loss": 11.7044,
      "step": 14930
    },
    {
      "epoch": 2.610885245901639,
      "grad_norm": 0.6983848640697711,
      "learning_rate": 5.683868281612183e-05,
      "loss": 11.5491,
      "step": 14931
    },
    {
      "epoch": 2.6110601092896175,
      "grad_norm": 0.6687939575305462,
      "learning_rate": 5.682590929677546e-05,
      "loss": 11.9548,
      "step": 14932
    },
    {
      "epoch": 2.611234972677596,
      "grad_norm": 0.7494714256215249,
      "learning_rate": 5.681313664320218e-05,
      "loss": 11.9388,
      "step": 14933
    },
    {
      "epoch": 2.6114098360655738,
      "grad_norm": 0.6483955924773713,
      "learning_rate": 5.680036485565811e-05,
      "loss": 11.5839,
      "step": 14934
    },
    {
      "epoch": 2.6115846994535517,
      "grad_norm": 0.7375464054960783,
      "learning_rate": 5.678759393439937e-05,
      "loss": 11.6843,
      "step": 14935
    },
    {
      "epoch": 2.61175956284153,
      "grad_norm": 0.6561426295060618,
      "learning_rate": 5.677482387968203e-05,
      "loss": 11.7514,
      "step": 14936
    },
    {
      "epoch": 2.6119344262295083,
      "grad_norm": 0.6533489404024165,
      "learning_rate": 5.6762054691762203e-05,
      "loss": 11.6786,
      "step": 14937
    },
    {
      "epoch": 2.6121092896174862,
      "grad_norm": 0.7274841552425475,
      "learning_rate": 5.674928637089594e-05,
      "loss": 11.459,
      "step": 14938
    },
    {
      "epoch": 2.6122841530054646,
      "grad_norm": 0.718040094490131,
      "learning_rate": 5.6736518917339274e-05,
      "loss": 12.0662,
      "step": 14939
    },
    {
      "epoch": 2.6124590163934425,
      "grad_norm": 0.6462461656411159,
      "learning_rate": 5.6723752331348246e-05,
      "loss": 11.5218,
      "step": 14940
    },
    {
      "epoch": 2.612633879781421,
      "grad_norm": 0.7511255900788313,
      "learning_rate": 5.671098661317886e-05,
      "loss": 11.8528,
      "step": 14941
    },
    {
      "epoch": 2.612808743169399,
      "grad_norm": 0.7120539800354413,
      "learning_rate": 5.66982217630871e-05,
      "loss": 11.8529,
      "step": 14942
    },
    {
      "epoch": 2.612983606557377,
      "grad_norm": 0.6837835615175516,
      "learning_rate": 5.668545778132897e-05,
      "loss": 11.898,
      "step": 14943
    },
    {
      "epoch": 2.613158469945355,
      "grad_norm": 0.6706349279837426,
      "learning_rate": 5.6672694668160386e-05,
      "loss": 11.7615,
      "step": 14944
    },
    {
      "epoch": 2.6133333333333333,
      "grad_norm": 0.7093875616801978,
      "learning_rate": 5.6659932423837306e-05,
      "loss": 11.7482,
      "step": 14945
    },
    {
      "epoch": 2.6135081967213116,
      "grad_norm": 0.7460834911994004,
      "learning_rate": 5.6647171048615655e-05,
      "loss": 11.9121,
      "step": 14946
    },
    {
      "epoch": 2.6136830601092895,
      "grad_norm": 0.7060814815198145,
      "learning_rate": 5.663441054275134e-05,
      "loss": 11.7567,
      "step": 14947
    },
    {
      "epoch": 2.613857923497268,
      "grad_norm": 0.7470660084386791,
      "learning_rate": 5.662165090650022e-05,
      "loss": 11.7768,
      "step": 14948
    },
    {
      "epoch": 2.6140327868852458,
      "grad_norm": 0.6479414999651195,
      "learning_rate": 5.660889214011822e-05,
      "loss": 11.8934,
      "step": 14949
    },
    {
      "epoch": 2.614207650273224,
      "grad_norm": 0.6972138007119454,
      "learning_rate": 5.659613424386114e-05,
      "loss": 11.7945,
      "step": 14950
    },
    {
      "epoch": 2.6143825136612024,
      "grad_norm": 0.749470602467332,
      "learning_rate": 5.6583377217984845e-05,
      "loss": 11.9726,
      "step": 14951
    },
    {
      "epoch": 2.6145573770491803,
      "grad_norm": 0.6433880668648204,
      "learning_rate": 5.6570621062745146e-05,
      "loss": 11.7452,
      "step": 14952
    },
    {
      "epoch": 2.6147322404371582,
      "grad_norm": 0.6648189396007054,
      "learning_rate": 5.6557865778397834e-05,
      "loss": 11.7597,
      "step": 14953
    },
    {
      "epoch": 2.6149071038251366,
      "grad_norm": 0.6963716231361394,
      "learning_rate": 5.65451113651987e-05,
      "loss": 11.8904,
      "step": 14954
    },
    {
      "epoch": 2.615081967213115,
      "grad_norm": 0.6480521085735157,
      "learning_rate": 5.6532357823403517e-05,
      "loss": 11.9744,
      "step": 14955
    },
    {
      "epoch": 2.615256830601093,
      "grad_norm": 0.6704009608590485,
      "learning_rate": 5.651960515326801e-05,
      "loss": 11.9883,
      "step": 14956
    },
    {
      "epoch": 2.615431693989071,
      "grad_norm": 0.6991864598999179,
      "learning_rate": 5.650685335504793e-05,
      "loss": 11.702,
      "step": 14957
    },
    {
      "epoch": 2.615606557377049,
      "grad_norm": 0.6598727184086043,
      "learning_rate": 5.6494102428998976e-05,
      "loss": 11.8523,
      "step": 14958
    },
    {
      "epoch": 2.6157814207650274,
      "grad_norm": 0.712489174846939,
      "learning_rate": 5.648135237537688e-05,
      "loss": 11.9112,
      "step": 14959
    },
    {
      "epoch": 2.6159562841530057,
      "grad_norm": 0.6080434952692713,
      "learning_rate": 5.646860319443722e-05,
      "loss": 11.9643,
      "step": 14960
    },
    {
      "epoch": 2.6161311475409836,
      "grad_norm": 0.6305261767782875,
      "learning_rate": 5.6455854886435765e-05,
      "loss": 11.9985,
      "step": 14961
    },
    {
      "epoch": 2.6163060109289615,
      "grad_norm": 0.6182474182027016,
      "learning_rate": 5.644310745162811e-05,
      "loss": 11.7263,
      "step": 14962
    },
    {
      "epoch": 2.61648087431694,
      "grad_norm": 0.6745609975562586,
      "learning_rate": 5.64303608902699e-05,
      "loss": 11.9903,
      "step": 14963
    },
    {
      "epoch": 2.616655737704918,
      "grad_norm": 0.6457820110402273,
      "learning_rate": 5.641761520261677e-05,
      "loss": 11.9011,
      "step": 14964
    },
    {
      "epoch": 2.616830601092896,
      "grad_norm": 0.6481666510937911,
      "learning_rate": 5.640487038892418e-05,
      "loss": 11.7756,
      "step": 14965
    },
    {
      "epoch": 2.6170054644808745,
      "grad_norm": 0.6615966083559276,
      "learning_rate": 5.6392126449447835e-05,
      "loss": 11.8427,
      "step": 14966
    },
    {
      "epoch": 2.6171803278688524,
      "grad_norm": 0.620285911818172,
      "learning_rate": 5.6379383384443255e-05,
      "loss": 11.8266,
      "step": 14967
    },
    {
      "epoch": 2.6173551912568307,
      "grad_norm": 0.6387206387688839,
      "learning_rate": 5.6366641194166015e-05,
      "loss": 11.842,
      "step": 14968
    },
    {
      "epoch": 2.6175300546448086,
      "grad_norm": 0.6636349261932359,
      "learning_rate": 5.635389987887154e-05,
      "loss": 11.9661,
      "step": 14969
    },
    {
      "epoch": 2.617704918032787,
      "grad_norm": 0.644282329458038,
      "learning_rate": 5.634115943881535e-05,
      "loss": 11.8088,
      "step": 14970
    },
    {
      "epoch": 2.617879781420765,
      "grad_norm": 0.6306532711855088,
      "learning_rate": 5.6328419874252994e-05,
      "loss": 11.5851,
      "step": 14971
    },
    {
      "epoch": 2.618054644808743,
      "grad_norm": 0.6996932308456776,
      "learning_rate": 5.6315681185439915e-05,
      "loss": 11.6012,
      "step": 14972
    },
    {
      "epoch": 2.6182295081967215,
      "grad_norm": 0.7297165427778962,
      "learning_rate": 5.63029433726316e-05,
      "loss": 11.8039,
      "step": 14973
    },
    {
      "epoch": 2.6184043715846994,
      "grad_norm": 0.7009443088421926,
      "learning_rate": 5.629020643608336e-05,
      "loss": 11.731,
      "step": 14974
    },
    {
      "epoch": 2.6185792349726777,
      "grad_norm": 0.7520030963886601,
      "learning_rate": 5.6277470376050715e-05,
      "loss": 11.7213,
      "step": 14975
    },
    {
      "epoch": 2.6187540983606556,
      "grad_norm": 0.6662610686651684,
      "learning_rate": 5.6264735192789055e-05,
      "loss": 11.779,
      "step": 14976
    },
    {
      "epoch": 2.618928961748634,
      "grad_norm": 0.7151382174660995,
      "learning_rate": 5.625200088655371e-05,
      "loss": 11.7113,
      "step": 14977
    },
    {
      "epoch": 2.619103825136612,
      "grad_norm": 0.6241064472312801,
      "learning_rate": 5.623926745760013e-05,
      "loss": 11.7533,
      "step": 14978
    },
    {
      "epoch": 2.6192786885245902,
      "grad_norm": 0.7134408481746779,
      "learning_rate": 5.622653490618353e-05,
      "loss": 11.4376,
      "step": 14979
    },
    {
      "epoch": 2.619453551912568,
      "grad_norm": 0.7030586162380482,
      "learning_rate": 5.6213803232559335e-05,
      "loss": 12.0171,
      "step": 14980
    },
    {
      "epoch": 2.6196284153005465,
      "grad_norm": 0.6668174536807245,
      "learning_rate": 5.620107243698282e-05,
      "loss": 11.8719,
      "step": 14981
    },
    {
      "epoch": 2.619803278688525,
      "grad_norm": 0.6659896445705575,
      "learning_rate": 5.618834251970934e-05,
      "loss": 11.8577,
      "step": 14982
    },
    {
      "epoch": 2.6199781420765027,
      "grad_norm": 0.675958397911534,
      "learning_rate": 5.617561348099406e-05,
      "loss": 11.9061,
      "step": 14983
    },
    {
      "epoch": 2.6201530054644806,
      "grad_norm": 0.6642973830892963,
      "learning_rate": 5.616288532109225e-05,
      "loss": 11.9482,
      "step": 14984
    },
    {
      "epoch": 2.620327868852459,
      "grad_norm": 0.7093543300486463,
      "learning_rate": 5.615015804025923e-05,
      "loss": 11.9741,
      "step": 14985
    },
    {
      "epoch": 2.6205027322404373,
      "grad_norm": 0.6120406750600726,
      "learning_rate": 5.613743163875018e-05,
      "loss": 11.9311,
      "step": 14986
    },
    {
      "epoch": 2.620677595628415,
      "grad_norm": 0.6208399030395617,
      "learning_rate": 5.612470611682032e-05,
      "loss": 11.7506,
      "step": 14987
    },
    {
      "epoch": 2.6208524590163935,
      "grad_norm": 0.6875097377665539,
      "learning_rate": 5.611198147472481e-05,
      "loss": 11.8443,
      "step": 14988
    },
    {
      "epoch": 2.6210273224043714,
      "grad_norm": 0.7035272198484734,
      "learning_rate": 5.6099257712718766e-05,
      "loss": 12.0338,
      "step": 14989
    },
    {
      "epoch": 2.6212021857923498,
      "grad_norm": 0.6848835589080675,
      "learning_rate": 5.608653483105744e-05,
      "loss": 11.4941,
      "step": 14990
    },
    {
      "epoch": 2.621377049180328,
      "grad_norm": 0.6148467795084882,
      "learning_rate": 5.607381282999596e-05,
      "loss": 11.5428,
      "step": 14991
    },
    {
      "epoch": 2.621551912568306,
      "grad_norm": 0.6974389948620822,
      "learning_rate": 5.606109170978937e-05,
      "loss": 11.9325,
      "step": 14992
    },
    {
      "epoch": 2.621726775956284,
      "grad_norm": 0.637841138249112,
      "learning_rate": 5.6048371470692805e-05,
      "loss": 11.7664,
      "step": 14993
    },
    {
      "epoch": 2.6219016393442622,
      "grad_norm": 0.6024217868989717,
      "learning_rate": 5.603565211296129e-05,
      "loss": 11.7036,
      "step": 14994
    },
    {
      "epoch": 2.6220765027322406,
      "grad_norm": 0.6615226065396076,
      "learning_rate": 5.602293363684998e-05,
      "loss": 11.6274,
      "step": 14995
    },
    {
      "epoch": 2.6222513661202185,
      "grad_norm": 0.7098818903960054,
      "learning_rate": 5.601021604261394e-05,
      "loss": 11.6225,
      "step": 14996
    },
    {
      "epoch": 2.622426229508197,
      "grad_norm": 0.6457523034496322,
      "learning_rate": 5.5997499330508066e-05,
      "loss": 11.8792,
      "step": 14997
    },
    {
      "epoch": 2.6226010928961747,
      "grad_norm": 0.6632519259398117,
      "learning_rate": 5.598478350078742e-05,
      "loss": 11.8842,
      "step": 14998
    },
    {
      "epoch": 2.622775956284153,
      "grad_norm": 0.6841248634496746,
      "learning_rate": 5.597206855370702e-05,
      "loss": 11.8813,
      "step": 14999
    },
    {
      "epoch": 2.6229508196721314,
      "grad_norm": 0.6779571532052191,
      "learning_rate": 5.5959354489521895e-05,
      "loss": 12.0207,
      "step": 15000
    },
    {
      "epoch": 2.6231256830601093,
      "grad_norm": 0.580278007770792,
      "learning_rate": 5.594664130848688e-05,
      "loss": 11.9314,
      "step": 15001
    },
    {
      "epoch": 2.623300546448087,
      "grad_norm": 0.6294611347127993,
      "learning_rate": 5.593392901085698e-05,
      "loss": 11.9109,
      "step": 15002
    },
    {
      "epoch": 2.6234754098360655,
      "grad_norm": 0.6418222877650857,
      "learning_rate": 5.592121759688706e-05,
      "loss": 11.975,
      "step": 15003
    },
    {
      "epoch": 2.623650273224044,
      "grad_norm": 0.6539039961159373,
      "learning_rate": 5.590850706683211e-05,
      "loss": 11.7557,
      "step": 15004
    },
    {
      "epoch": 2.6238251366120218,
      "grad_norm": 0.6411219055778113,
      "learning_rate": 5.5895797420946995e-05,
      "loss": 11.6738,
      "step": 15005
    },
    {
      "epoch": 2.624,
      "grad_norm": 0.6609841168901588,
      "learning_rate": 5.5883088659486525e-05,
      "loss": 11.747,
      "step": 15006
    },
    {
      "epoch": 2.624174863387978,
      "grad_norm": 0.6430964298461782,
      "learning_rate": 5.587038078270559e-05,
      "loss": 12.0204,
      "step": 15007
    },
    {
      "epoch": 2.6243497267759563,
      "grad_norm": 0.7040151163480102,
      "learning_rate": 5.585767379085898e-05,
      "loss": 11.9894,
      "step": 15008
    },
    {
      "epoch": 2.6245245901639347,
      "grad_norm": 0.7421772274293305,
      "learning_rate": 5.584496768420162e-05,
      "loss": 11.9111,
      "step": 15009
    },
    {
      "epoch": 2.6246994535519126,
      "grad_norm": 0.6718851535149872,
      "learning_rate": 5.5832262462988196e-05,
      "loss": 11.9429,
      "step": 15010
    },
    {
      "epoch": 2.6248743169398905,
      "grad_norm": 0.7185080149798108,
      "learning_rate": 5.5819558127473514e-05,
      "loss": 11.6267,
      "step": 15011
    },
    {
      "epoch": 2.625049180327869,
      "grad_norm": 0.6593937288254558,
      "learning_rate": 5.5806854677912354e-05,
      "loss": 12.019,
      "step": 15012
    },
    {
      "epoch": 2.625224043715847,
      "grad_norm": 0.6237280570741629,
      "learning_rate": 5.579415211455941e-05,
      "loss": 11.8418,
      "step": 15013
    },
    {
      "epoch": 2.625398907103825,
      "grad_norm": 0.6829810220132646,
      "learning_rate": 5.578145043766951e-05,
      "loss": 11.5878,
      "step": 15014
    },
    {
      "epoch": 2.6255737704918034,
      "grad_norm": 0.6749987749162392,
      "learning_rate": 5.576874964749727e-05,
      "loss": 11.7342,
      "step": 15015
    },
    {
      "epoch": 2.6257486338797813,
      "grad_norm": 0.6179896218437659,
      "learning_rate": 5.57560497442974e-05,
      "loss": 11.7979,
      "step": 15016
    },
    {
      "epoch": 2.6259234972677596,
      "grad_norm": 0.619692242631239,
      "learning_rate": 5.574335072832458e-05,
      "loss": 11.9217,
      "step": 15017
    },
    {
      "epoch": 2.6260983606557375,
      "grad_norm": 0.7152423633206464,
      "learning_rate": 5.573065259983346e-05,
      "loss": 12.0104,
      "step": 15018
    },
    {
      "epoch": 2.626273224043716,
      "grad_norm": 0.6176536808643049,
      "learning_rate": 5.571795535907869e-05,
      "loss": 11.3554,
      "step": 15019
    },
    {
      "epoch": 2.626448087431694,
      "grad_norm": 0.6516242526801755,
      "learning_rate": 5.570525900631488e-05,
      "loss": 11.9427,
      "step": 15020
    },
    {
      "epoch": 2.626622950819672,
      "grad_norm": 0.6454260117938976,
      "learning_rate": 5.569256354179663e-05,
      "loss": 11.9049,
      "step": 15021
    },
    {
      "epoch": 2.6267978142076505,
      "grad_norm": 0.6262473484983929,
      "learning_rate": 5.5679868965778484e-05,
      "loss": 11.952,
      "step": 15022
    },
    {
      "epoch": 2.6269726775956284,
      "grad_norm": 0.6156353846534416,
      "learning_rate": 5.5667175278515105e-05,
      "loss": 11.5994,
      "step": 15023
    },
    {
      "epoch": 2.6271475409836067,
      "grad_norm": 0.6831945214248655,
      "learning_rate": 5.5654482480260964e-05,
      "loss": 11.8997,
      "step": 15024
    },
    {
      "epoch": 2.6273224043715846,
      "grad_norm": 0.613276154586668,
      "learning_rate": 5.564179057127059e-05,
      "loss": 11.8278,
      "step": 15025
    },
    {
      "epoch": 2.627497267759563,
      "grad_norm": 0.6683106506060004,
      "learning_rate": 5.562909955179854e-05,
      "loss": 11.846,
      "step": 15026
    },
    {
      "epoch": 2.627672131147541,
      "grad_norm": 0.741372488061035,
      "learning_rate": 5.561640942209925e-05,
      "loss": 11.8879,
      "step": 15027
    },
    {
      "epoch": 2.627846994535519,
      "grad_norm": 0.6273577707172899,
      "learning_rate": 5.560372018242725e-05,
      "loss": 11.8872,
      "step": 15028
    },
    {
      "epoch": 2.628021857923497,
      "grad_norm": 0.6944730811597604,
      "learning_rate": 5.5591031833036956e-05,
      "loss": 11.6352,
      "step": 15029
    },
    {
      "epoch": 2.6281967213114754,
      "grad_norm": 0.6564211668325093,
      "learning_rate": 5.557834437418283e-05,
      "loss": 11.8708,
      "step": 15030
    },
    {
      "epoch": 2.6283715846994538,
      "grad_norm": 0.695744089966459,
      "learning_rate": 5.556565780611931e-05,
      "loss": 11.989,
      "step": 15031
    },
    {
      "epoch": 2.6285464480874317,
      "grad_norm": 0.6841305398101918,
      "learning_rate": 5.555297212910077e-05,
      "loss": 11.8869,
      "step": 15032
    },
    {
      "epoch": 2.62872131147541,
      "grad_norm": 0.6637000058046645,
      "learning_rate": 5.5540287343381606e-05,
      "loss": 11.7716,
      "step": 15033
    },
    {
      "epoch": 2.628896174863388,
      "grad_norm": 0.6440758947038101,
      "learning_rate": 5.5527603449216194e-05,
      "loss": 11.7402,
      "step": 15034
    },
    {
      "epoch": 2.6290710382513662,
      "grad_norm": 0.6547840679160404,
      "learning_rate": 5.551492044685887e-05,
      "loss": 11.7149,
      "step": 15035
    },
    {
      "epoch": 2.629245901639344,
      "grad_norm": 0.6380266473490731,
      "learning_rate": 5.5502238336564e-05,
      "loss": 11.8547,
      "step": 15036
    },
    {
      "epoch": 2.6294207650273225,
      "grad_norm": 0.6586113951986206,
      "learning_rate": 5.548955711858585e-05,
      "loss": 11.627,
      "step": 15037
    },
    {
      "epoch": 2.6295956284153004,
      "grad_norm": 0.675944916006992,
      "learning_rate": 5.547687679317877e-05,
      "loss": 12.0571,
      "step": 15038
    },
    {
      "epoch": 2.6297704918032787,
      "grad_norm": 0.623406181256228,
      "learning_rate": 5.5464197360597005e-05,
      "loss": 12.0196,
      "step": 15039
    },
    {
      "epoch": 2.629945355191257,
      "grad_norm": 0.6777801481858146,
      "learning_rate": 5.5451518821094825e-05,
      "loss": 12.0474,
      "step": 15040
    },
    {
      "epoch": 2.630120218579235,
      "grad_norm": 0.6258844004636799,
      "learning_rate": 5.543884117492648e-05,
      "loss": 11.8101,
      "step": 15041
    },
    {
      "epoch": 2.630295081967213,
      "grad_norm": 0.6841098754771407,
      "learning_rate": 5.542616442234618e-05,
      "loss": 11.649,
      "step": 15042
    },
    {
      "epoch": 2.630469945355191,
      "grad_norm": 0.664732602924074,
      "learning_rate": 5.541348856360815e-05,
      "loss": 11.7267,
      "step": 15043
    },
    {
      "epoch": 2.6306448087431695,
      "grad_norm": 0.6536560807161933,
      "learning_rate": 5.540081359896657e-05,
      "loss": 11.8366,
      "step": 15044
    },
    {
      "epoch": 2.6308196721311474,
      "grad_norm": 0.6312245221657652,
      "learning_rate": 5.5388139528675606e-05,
      "loss": 11.741,
      "step": 15045
    },
    {
      "epoch": 2.6309945355191258,
      "grad_norm": 0.6015019076462644,
      "learning_rate": 5.5375466352989447e-05,
      "loss": 12.0047,
      "step": 15046
    },
    {
      "epoch": 2.6311693989071037,
      "grad_norm": 0.6407397683576471,
      "learning_rate": 5.536279407216218e-05,
      "loss": 11.8067,
      "step": 15047
    },
    {
      "epoch": 2.631344262295082,
      "grad_norm": 0.6050041041423733,
      "learning_rate": 5.5350122686447946e-05,
      "loss": 11.6535,
      "step": 15048
    },
    {
      "epoch": 2.6315191256830603,
      "grad_norm": 0.6488854667079526,
      "learning_rate": 5.533745219610086e-05,
      "loss": 11.8618,
      "step": 15049
    },
    {
      "epoch": 2.6316939890710382,
      "grad_norm": 0.6939661484320611,
      "learning_rate": 5.532478260137498e-05,
      "loss": 11.9284,
      "step": 15050
    },
    {
      "epoch": 2.631868852459016,
      "grad_norm": 0.6693647776767858,
      "learning_rate": 5.531211390252438e-05,
      "loss": 11.682,
      "step": 15051
    },
    {
      "epoch": 2.6320437158469945,
      "grad_norm": 0.6142164685331641,
      "learning_rate": 5.529944609980311e-05,
      "loss": 11.92,
      "step": 15052
    },
    {
      "epoch": 2.632218579234973,
      "grad_norm": 0.7244584096364475,
      "learning_rate": 5.528677919346519e-05,
      "loss": 11.5928,
      "step": 15053
    },
    {
      "epoch": 2.6323934426229507,
      "grad_norm": 0.6565401408845791,
      "learning_rate": 5.5274113183764634e-05,
      "loss": 11.7745,
      "step": 15054
    },
    {
      "epoch": 2.632568306010929,
      "grad_norm": 0.6426742799536633,
      "learning_rate": 5.526144807095548e-05,
      "loss": 11.9087,
      "step": 15055
    },
    {
      "epoch": 2.632743169398907,
      "grad_norm": 0.6384301944304603,
      "learning_rate": 5.524878385529157e-05,
      "loss": 11.4986,
      "step": 15056
    },
    {
      "epoch": 2.6329180327868853,
      "grad_norm": 0.6107117278350854,
      "learning_rate": 5.5236120537027006e-05,
      "loss": 11.9094,
      "step": 15057
    },
    {
      "epoch": 2.6330928961748636,
      "grad_norm": 0.672238991083503,
      "learning_rate": 5.522345811641565e-05,
      "loss": 11.93,
      "step": 15058
    },
    {
      "epoch": 2.6332677595628415,
      "grad_norm": 0.6458739435753336,
      "learning_rate": 5.5210796593711447e-05,
      "loss": 11.8091,
      "step": 15059
    },
    {
      "epoch": 2.6334426229508194,
      "grad_norm": 0.6302038809154955,
      "learning_rate": 5.519813596916833e-05,
      "loss": 11.8844,
      "step": 15060
    },
    {
      "epoch": 2.6336174863387978,
      "grad_norm": 0.6971887969917797,
      "learning_rate": 5.518547624304007e-05,
      "loss": 11.9458,
      "step": 15061
    },
    {
      "epoch": 2.633792349726776,
      "grad_norm": 0.7508746205292675,
      "learning_rate": 5.5172817415580643e-05,
      "loss": 11.9037,
      "step": 15062
    },
    {
      "epoch": 2.633967213114754,
      "grad_norm": 0.6659551959081333,
      "learning_rate": 5.516015948704387e-05,
      "loss": 11.7614,
      "step": 15063
    },
    {
      "epoch": 2.6341420765027324,
      "grad_norm": 0.6620982222768377,
      "learning_rate": 5.514750245768362e-05,
      "loss": 12.0092,
      "step": 15064
    },
    {
      "epoch": 2.6343169398907103,
      "grad_norm": 0.7336538160311022,
      "learning_rate": 5.513484632775362e-05,
      "loss": 11.918,
      "step": 15065
    },
    {
      "epoch": 2.6344918032786886,
      "grad_norm": 0.6812816580563522,
      "learning_rate": 5.5122191097507656e-05,
      "loss": 11.8158,
      "step": 15066
    },
    {
      "epoch": 2.634666666666667,
      "grad_norm": 0.6891642131154195,
      "learning_rate": 5.510953676719961e-05,
      "loss": 11.9134,
      "step": 15067
    },
    {
      "epoch": 2.634841530054645,
      "grad_norm": 0.6458239633751731,
      "learning_rate": 5.509688333708316e-05,
      "loss": 11.966,
      "step": 15068
    },
    {
      "epoch": 2.6350163934426227,
      "grad_norm": 0.65507948272226,
      "learning_rate": 5.5084230807412126e-05,
      "loss": 11.994,
      "step": 15069
    },
    {
      "epoch": 2.635191256830601,
      "grad_norm": 0.6286179153119936,
      "learning_rate": 5.50715791784401e-05,
      "loss": 11.5124,
      "step": 15070
    },
    {
      "epoch": 2.6353661202185794,
      "grad_norm": 0.6509293657235506,
      "learning_rate": 5.505892845042089e-05,
      "loss": 11.6876,
      "step": 15071
    },
    {
      "epoch": 2.6355409836065573,
      "grad_norm": 0.7553758656313184,
      "learning_rate": 5.504627862360815e-05,
      "loss": 11.88,
      "step": 15072
    },
    {
      "epoch": 2.6357158469945356,
      "grad_norm": 0.6849577449281397,
      "learning_rate": 5.503362969825558e-05,
      "loss": 11.8579,
      "step": 15073
    },
    {
      "epoch": 2.6358907103825135,
      "grad_norm": 0.6567315234015209,
      "learning_rate": 5.502098167461678e-05,
      "loss": 11.9251,
      "step": 15074
    },
    {
      "epoch": 2.636065573770492,
      "grad_norm": 0.7212479103005648,
      "learning_rate": 5.5008334552945364e-05,
      "loss": 11.896,
      "step": 15075
    },
    {
      "epoch": 2.63624043715847,
      "grad_norm": 0.6309915168135858,
      "learning_rate": 5.4995688333495e-05,
      "loss": 11.7725,
      "step": 15076
    },
    {
      "epoch": 2.636415300546448,
      "grad_norm": 0.6328705016007915,
      "learning_rate": 5.498304301651929e-05,
      "loss": 11.5089,
      "step": 15077
    },
    {
      "epoch": 2.636590163934426,
      "grad_norm": 0.7120961397425724,
      "learning_rate": 5.497039860227181e-05,
      "loss": 11.2742,
      "step": 15078
    },
    {
      "epoch": 2.6367650273224044,
      "grad_norm": 0.643251813388598,
      "learning_rate": 5.495775509100607e-05,
      "loss": 11.8448,
      "step": 15079
    },
    {
      "epoch": 2.6369398907103827,
      "grad_norm": 0.6364926065192432,
      "learning_rate": 5.494511248297558e-05,
      "loss": 11.7136,
      "step": 15080
    },
    {
      "epoch": 2.6371147540983606,
      "grad_norm": 0.6514793673490161,
      "learning_rate": 5.493247077843397e-05,
      "loss": 11.8661,
      "step": 15081
    },
    {
      "epoch": 2.637289617486339,
      "grad_norm": 0.6354021204339837,
      "learning_rate": 5.491982997763474e-05,
      "loss": 11.893,
      "step": 15082
    },
    {
      "epoch": 2.637464480874317,
      "grad_norm": 0.6161937838692744,
      "learning_rate": 5.490719008083129e-05,
      "loss": 11.9435,
      "step": 15083
    },
    {
      "epoch": 2.637639344262295,
      "grad_norm": 0.6609946476618012,
      "learning_rate": 5.4894551088277144e-05,
      "loss": 11.7571,
      "step": 15084
    },
    {
      "epoch": 2.637814207650273,
      "grad_norm": 0.6472623176682293,
      "learning_rate": 5.488191300022569e-05,
      "loss": 11.6668,
      "step": 15085
    },
    {
      "epoch": 2.6379890710382514,
      "grad_norm": 0.6990248043859608,
      "learning_rate": 5.4869275816930445e-05,
      "loss": 11.8459,
      "step": 15086
    },
    {
      "epoch": 2.6381639344262293,
      "grad_norm": 0.7618223902554113,
      "learning_rate": 5.485663953864484e-05,
      "loss": 12.0425,
      "step": 15087
    },
    {
      "epoch": 2.6383387978142077,
      "grad_norm": 0.6344727043939175,
      "learning_rate": 5.484400416562218e-05,
      "loss": 11.8471,
      "step": 15088
    },
    {
      "epoch": 2.638513661202186,
      "grad_norm": 0.6717699369611642,
      "learning_rate": 5.4831369698115875e-05,
      "loss": 11.7255,
      "step": 15089
    },
    {
      "epoch": 2.638688524590164,
      "grad_norm": 0.6221282371104235,
      "learning_rate": 5.4818736136379266e-05,
      "loss": 11.8327,
      "step": 15090
    },
    {
      "epoch": 2.6388633879781422,
      "grad_norm": 0.6480590272777627,
      "learning_rate": 5.480610348066579e-05,
      "loss": 11.9129,
      "step": 15091
    },
    {
      "epoch": 2.63903825136612,
      "grad_norm": 0.6314821138541471,
      "learning_rate": 5.479347173122866e-05,
      "loss": 11.7104,
      "step": 15092
    },
    {
      "epoch": 2.6392131147540985,
      "grad_norm": 0.6605766415892751,
      "learning_rate": 5.478084088832124e-05,
      "loss": 11.6982,
      "step": 15093
    },
    {
      "epoch": 2.6393879781420764,
      "grad_norm": 0.6737363529020792,
      "learning_rate": 5.476821095219675e-05,
      "loss": 11.649,
      "step": 15094
    },
    {
      "epoch": 2.6395628415300547,
      "grad_norm": 0.6612082273898334,
      "learning_rate": 5.475558192310856e-05,
      "loss": 11.8004,
      "step": 15095
    },
    {
      "epoch": 2.6397377049180326,
      "grad_norm": 0.7239274750194883,
      "learning_rate": 5.474295380130989e-05,
      "loss": 11.9191,
      "step": 15096
    },
    {
      "epoch": 2.639912568306011,
      "grad_norm": 0.648477594462304,
      "learning_rate": 5.4730326587053936e-05,
      "loss": 11.9672,
      "step": 15097
    },
    {
      "epoch": 2.6400874316939893,
      "grad_norm": 0.6476506891723958,
      "learning_rate": 5.4717700280593934e-05,
      "loss": 11.8241,
      "step": 15098
    },
    {
      "epoch": 2.640262295081967,
      "grad_norm": 0.6892135539592105,
      "learning_rate": 5.4705074882183013e-05,
      "loss": 11.9859,
      "step": 15099
    },
    {
      "epoch": 2.640437158469945,
      "grad_norm": 0.6582184377199509,
      "learning_rate": 5.469245039207451e-05,
      "loss": 12.0402,
      "step": 15100
    },
    {
      "epoch": 2.6406120218579234,
      "grad_norm": 0.7039547300545871,
      "learning_rate": 5.467982681052144e-05,
      "loss": 11.6971,
      "step": 15101
    },
    {
      "epoch": 2.6407868852459018,
      "grad_norm": 0.7095648503300651,
      "learning_rate": 5.466720413777699e-05,
      "loss": 11.728,
      "step": 15102
    },
    {
      "epoch": 2.6409617486338797,
      "grad_norm": 0.6908507627737311,
      "learning_rate": 5.465458237409431e-05,
      "loss": 11.8893,
      "step": 15103
    },
    {
      "epoch": 2.641136612021858,
      "grad_norm": 0.6960435354593875,
      "learning_rate": 5.464196151972642e-05,
      "loss": 11.8884,
      "step": 15104
    },
    {
      "epoch": 2.641311475409836,
      "grad_norm": 0.7025140025296848,
      "learning_rate": 5.462934157492656e-05,
      "loss": 11.5782,
      "step": 15105
    },
    {
      "epoch": 2.6414863387978142,
      "grad_norm": 0.6509817166179883,
      "learning_rate": 5.461672253994766e-05,
      "loss": 11.988,
      "step": 15106
    },
    {
      "epoch": 2.6416612021857926,
      "grad_norm": 0.6546866679944827,
      "learning_rate": 5.460410441504282e-05,
      "loss": 11.9003,
      "step": 15107
    },
    {
      "epoch": 2.6418360655737705,
      "grad_norm": 0.6774365452978087,
      "learning_rate": 5.459148720046508e-05,
      "loss": 11.913,
      "step": 15108
    },
    {
      "epoch": 2.6420109289617484,
      "grad_norm": 0.6548283682495188,
      "learning_rate": 5.457887089646743e-05,
      "loss": 11.862,
      "step": 15109
    },
    {
      "epoch": 2.6421857923497267,
      "grad_norm": 0.6622150949209009,
      "learning_rate": 5.45662555033029e-05,
      "loss": 11.8636,
      "step": 15110
    },
    {
      "epoch": 2.642360655737705,
      "grad_norm": 0.620796338395246,
      "learning_rate": 5.455364102122442e-05,
      "loss": 11.903,
      "step": 15111
    },
    {
      "epoch": 2.642535519125683,
      "grad_norm": 0.6043958484178722,
      "learning_rate": 5.454102745048499e-05,
      "loss": 11.8718,
      "step": 15112
    },
    {
      "epoch": 2.6427103825136613,
      "grad_norm": 0.6254028811239105,
      "learning_rate": 5.4528414791337546e-05,
      "loss": 11.7173,
      "step": 15113
    },
    {
      "epoch": 2.642885245901639,
      "grad_norm": 0.6201230436450136,
      "learning_rate": 5.4515803044034985e-05,
      "loss": 12.022,
      "step": 15114
    },
    {
      "epoch": 2.6430601092896175,
      "grad_norm": 0.6485594900558849,
      "learning_rate": 5.4503192208830235e-05,
      "loss": 11.9807,
      "step": 15115
    },
    {
      "epoch": 2.643234972677596,
      "grad_norm": 0.6970192881274352,
      "learning_rate": 5.449058228597618e-05,
      "loss": 11.2607,
      "step": 15116
    },
    {
      "epoch": 2.643409836065574,
      "grad_norm": 0.6168571137545475,
      "learning_rate": 5.4477973275725666e-05,
      "loss": 11.7914,
      "step": 15117
    },
    {
      "epoch": 2.6435846994535517,
      "grad_norm": 0.6359925531574552,
      "learning_rate": 5.446536517833157e-05,
      "loss": 11.7039,
      "step": 15118
    },
    {
      "epoch": 2.64375956284153,
      "grad_norm": 0.7543238276502473,
      "learning_rate": 5.445275799404672e-05,
      "loss": 12.183,
      "step": 15119
    },
    {
      "epoch": 2.6439344262295084,
      "grad_norm": 0.6298258405799392,
      "learning_rate": 5.444015172312391e-05,
      "loss": 11.7679,
      "step": 15120
    },
    {
      "epoch": 2.6441092896174863,
      "grad_norm": 0.6400950679981297,
      "learning_rate": 5.442754636581595e-05,
      "loss": 11.712,
      "step": 15121
    },
    {
      "epoch": 2.6442841530054646,
      "grad_norm": 0.694546270908394,
      "learning_rate": 5.4414941922375605e-05,
      "loss": 11.813,
      "step": 15122
    },
    {
      "epoch": 2.6444590163934425,
      "grad_norm": 0.7046043133618429,
      "learning_rate": 5.440233839305564e-05,
      "loss": 11.8311,
      "step": 15123
    },
    {
      "epoch": 2.644633879781421,
      "grad_norm": 0.6561435415203919,
      "learning_rate": 5.43897357781088e-05,
      "loss": 11.9011,
      "step": 15124
    },
    {
      "epoch": 2.644808743169399,
      "grad_norm": 0.7953212326847766,
      "learning_rate": 5.43771340777878e-05,
      "loss": 11.9262,
      "step": 15125
    },
    {
      "epoch": 2.644983606557377,
      "grad_norm": 0.6247411694159049,
      "learning_rate": 5.436453329234533e-05,
      "loss": 11.8033,
      "step": 15126
    },
    {
      "epoch": 2.645158469945355,
      "grad_norm": 0.6509287758819241,
      "learning_rate": 5.43519334220341e-05,
      "loss": 11.874,
      "step": 15127
    },
    {
      "epoch": 2.6453333333333333,
      "grad_norm": 0.6777453910265068,
      "learning_rate": 5.433933446710675e-05,
      "loss": 11.6328,
      "step": 15128
    },
    {
      "epoch": 2.6455081967213117,
      "grad_norm": 0.6344228983504931,
      "learning_rate": 5.4326736427815946e-05,
      "loss": 11.6553,
      "step": 15129
    },
    {
      "epoch": 2.6456830601092896,
      "grad_norm": 0.6257354075807544,
      "learning_rate": 5.431413930441431e-05,
      "loss": 11.7327,
      "step": 15130
    },
    {
      "epoch": 2.645857923497268,
      "grad_norm": 0.6423053461595467,
      "learning_rate": 5.4301543097154475e-05,
      "loss": 11.8662,
      "step": 15131
    },
    {
      "epoch": 2.646032786885246,
      "grad_norm": 0.7521314992095391,
      "learning_rate": 5.428894780628899e-05,
      "loss": 11.8632,
      "step": 15132
    },
    {
      "epoch": 2.646207650273224,
      "grad_norm": 0.6891247535424896,
      "learning_rate": 5.427635343207046e-05,
      "loss": 11.8177,
      "step": 15133
    },
    {
      "epoch": 2.646382513661202,
      "grad_norm": 0.6624920549098932,
      "learning_rate": 5.4263759974751436e-05,
      "loss": 11.6926,
      "step": 15134
    },
    {
      "epoch": 2.6465573770491804,
      "grad_norm": 0.6294995492272751,
      "learning_rate": 5.4251167434584446e-05,
      "loss": 11.5963,
      "step": 15135
    },
    {
      "epoch": 2.6467322404371583,
      "grad_norm": 0.6379188564903242,
      "learning_rate": 5.423857581182202e-05,
      "loss": 11.7543,
      "step": 15136
    },
    {
      "epoch": 2.6469071038251366,
      "grad_norm": 0.5690370281012223,
      "learning_rate": 5.422598510671666e-05,
      "loss": 11.9163,
      "step": 15137
    },
    {
      "epoch": 2.647081967213115,
      "grad_norm": 0.6493032535093475,
      "learning_rate": 5.421339531952084e-05,
      "loss": 11.5328,
      "step": 15138
    },
    {
      "epoch": 2.647256830601093,
      "grad_norm": 0.6658940446172245,
      "learning_rate": 5.420080645048703e-05,
      "loss": 11.8996,
      "step": 15139
    },
    {
      "epoch": 2.647431693989071,
      "grad_norm": 0.689081154783546,
      "learning_rate": 5.418821849986766e-05,
      "loss": 11.9598,
      "step": 15140
    },
    {
      "epoch": 2.647606557377049,
      "grad_norm": 0.636320650895814,
      "learning_rate": 5.417563146791519e-05,
      "loss": 11.8154,
      "step": 15141
    },
    {
      "epoch": 2.6477814207650274,
      "grad_norm": 0.608488783193556,
      "learning_rate": 5.416304535488199e-05,
      "loss": 11.7995,
      "step": 15142
    },
    {
      "epoch": 2.6479562841530053,
      "grad_norm": 0.6306949935399186,
      "learning_rate": 5.4150460161020475e-05,
      "loss": 11.7113,
      "step": 15143
    },
    {
      "epoch": 2.6481311475409837,
      "grad_norm": 0.6448561193044262,
      "learning_rate": 5.413787588658301e-05,
      "loss": 11.8298,
      "step": 15144
    },
    {
      "epoch": 2.6483060109289616,
      "grad_norm": 0.6211894268805932,
      "learning_rate": 5.412529253182196e-05,
      "loss": 11.8229,
      "step": 15145
    },
    {
      "epoch": 2.64848087431694,
      "grad_norm": 0.6164428941755379,
      "learning_rate": 5.411271009698964e-05,
      "loss": 12.0949,
      "step": 15146
    },
    {
      "epoch": 2.6486557377049182,
      "grad_norm": 0.6337693046850339,
      "learning_rate": 5.4100128582338374e-05,
      "loss": 11.7119,
      "step": 15147
    },
    {
      "epoch": 2.648830601092896,
      "grad_norm": 0.6363047481463812,
      "learning_rate": 5.4087547988120456e-05,
      "loss": 11.6832,
      "step": 15148
    },
    {
      "epoch": 2.6490054644808745,
      "grad_norm": 0.6792767133857153,
      "learning_rate": 5.4074968314588185e-05,
      "loss": 11.9501,
      "step": 15149
    },
    {
      "epoch": 2.6491803278688524,
      "grad_norm": 0.6690663581471529,
      "learning_rate": 5.4062389561993786e-05,
      "loss": 11.9715,
      "step": 15150
    },
    {
      "epoch": 2.6493551912568307,
      "grad_norm": 0.6542651091679762,
      "learning_rate": 5.404981173058958e-05,
      "loss": 12.0004,
      "step": 15151
    },
    {
      "epoch": 2.6495300546448086,
      "grad_norm": 0.6824973983860851,
      "learning_rate": 5.4037234820627655e-05,
      "loss": 11.6676,
      "step": 15152
    },
    {
      "epoch": 2.649704918032787,
      "grad_norm": 0.6266188798823257,
      "learning_rate": 5.402465883236032e-05,
      "loss": 11.7548,
      "step": 15153
    },
    {
      "epoch": 2.649879781420765,
      "grad_norm": 0.6767430628042918,
      "learning_rate": 5.4012083766039756e-05,
      "loss": 11.9537,
      "step": 15154
    },
    {
      "epoch": 2.650054644808743,
      "grad_norm": 0.6492257798772204,
      "learning_rate": 5.399950962191811e-05,
      "loss": 11.9598,
      "step": 15155
    },
    {
      "epoch": 2.6502295081967215,
      "grad_norm": 0.65017206373837,
      "learning_rate": 5.3986936400247565e-05,
      "loss": 11.5455,
      "step": 15156
    },
    {
      "epoch": 2.6504043715846994,
      "grad_norm": 0.7001720534851343,
      "learning_rate": 5.397436410128016e-05,
      "loss": 11.8872,
      "step": 15157
    },
    {
      "epoch": 2.6505792349726773,
      "grad_norm": 0.5904147034939093,
      "learning_rate": 5.39617927252681e-05,
      "loss": 11.818,
      "step": 15158
    },
    {
      "epoch": 2.6507540983606557,
      "grad_norm": 0.6579268718303628,
      "learning_rate": 5.3949222272463464e-05,
      "loss": 12.0068,
      "step": 15159
    },
    {
      "epoch": 2.650928961748634,
      "grad_norm": 0.6953540039812728,
      "learning_rate": 5.3936652743118346e-05,
      "loss": 11.7293,
      "step": 15160
    },
    {
      "epoch": 2.651103825136612,
      "grad_norm": 0.7341089372062035,
      "learning_rate": 5.3924084137484734e-05,
      "loss": 11.9442,
      "step": 15161
    },
    {
      "epoch": 2.6512786885245903,
      "grad_norm": 0.6517359420192308,
      "learning_rate": 5.391151645581467e-05,
      "loss": 11.8195,
      "step": 15162
    },
    {
      "epoch": 2.651453551912568,
      "grad_norm": 0.6917922011992113,
      "learning_rate": 5.389894969836025e-05,
      "loss": 11.9849,
      "step": 15163
    },
    {
      "epoch": 2.6516284153005465,
      "grad_norm": 0.6449548566587812,
      "learning_rate": 5.388638386537343e-05,
      "loss": 11.9967,
      "step": 15164
    },
    {
      "epoch": 2.651803278688525,
      "grad_norm": 0.6172009746908419,
      "learning_rate": 5.387381895710625e-05,
      "loss": 11.9863,
      "step": 15165
    },
    {
      "epoch": 2.6519781420765027,
      "grad_norm": 0.6883583652756953,
      "learning_rate": 5.3861254973810537e-05,
      "loss": 11.5621,
      "step": 15166
    },
    {
      "epoch": 2.6521530054644806,
      "grad_norm": 0.6874391087612328,
      "learning_rate": 5.384869191573838e-05,
      "loss": 12.0024,
      "step": 15167
    },
    {
      "epoch": 2.652327868852459,
      "grad_norm": 0.6153993652220121,
      "learning_rate": 5.383612978314164e-05,
      "loss": 11.925,
      "step": 15168
    },
    {
      "epoch": 2.6525027322404373,
      "grad_norm": 0.7171999879116253,
      "learning_rate": 5.382356857627229e-05,
      "loss": 11.8122,
      "step": 15169
    },
    {
      "epoch": 2.652677595628415,
      "grad_norm": 0.7728914516815996,
      "learning_rate": 5.3811008295382126e-05,
      "loss": 11.637,
      "step": 15170
    },
    {
      "epoch": 2.6528524590163935,
      "grad_norm": 0.6309819860821947,
      "learning_rate": 5.3798448940723025e-05,
      "loss": 11.9239,
      "step": 15171
    },
    {
      "epoch": 2.6530273224043714,
      "grad_norm": 0.717425387323058,
      "learning_rate": 5.378589051254692e-05,
      "loss": 11.6543,
      "step": 15172
    },
    {
      "epoch": 2.65320218579235,
      "grad_norm": 0.8271254105527638,
      "learning_rate": 5.377333301110561e-05,
      "loss": 11.6943,
      "step": 15173
    },
    {
      "epoch": 2.653377049180328,
      "grad_norm": 0.7378628952516755,
      "learning_rate": 5.376077643665094e-05,
      "loss": 11.8234,
      "step": 15174
    },
    {
      "epoch": 2.653551912568306,
      "grad_norm": 0.7422863757589704,
      "learning_rate": 5.3748220789434635e-05,
      "loss": 11.8266,
      "step": 15175
    },
    {
      "epoch": 2.653726775956284,
      "grad_norm": 0.6629714150173707,
      "learning_rate": 5.373566606970848e-05,
      "loss": 11.7348,
      "step": 15176
    },
    {
      "epoch": 2.6539016393442623,
      "grad_norm": 0.6288846184866901,
      "learning_rate": 5.372311227772431e-05,
      "loss": 12.0175,
      "step": 15177
    },
    {
      "epoch": 2.6540765027322406,
      "grad_norm": 0.6433484093866939,
      "learning_rate": 5.371055941373386e-05,
      "loss": 11.884,
      "step": 15178
    },
    {
      "epoch": 2.6542513661202185,
      "grad_norm": 0.6540558223715562,
      "learning_rate": 5.369800747798879e-05,
      "loss": 11.7789,
      "step": 15179
    },
    {
      "epoch": 2.654426229508197,
      "grad_norm": 0.6338853651947467,
      "learning_rate": 5.3685456470740836e-05,
      "loss": 11.9748,
      "step": 15180
    },
    {
      "epoch": 2.6546010928961747,
      "grad_norm": 0.6724337882079485,
      "learning_rate": 5.367290639224165e-05,
      "loss": 12.1136,
      "step": 15181
    },
    {
      "epoch": 2.654775956284153,
      "grad_norm": 0.65591744237739,
      "learning_rate": 5.366035724274297e-05,
      "loss": 11.8272,
      "step": 15182
    },
    {
      "epoch": 2.6549508196721314,
      "grad_norm": 0.648649762168064,
      "learning_rate": 5.364780902249643e-05,
      "loss": 11.7279,
      "step": 15183
    },
    {
      "epoch": 2.6551256830601093,
      "grad_norm": 0.6177475889144292,
      "learning_rate": 5.363526173175362e-05,
      "loss": 11.7866,
      "step": 15184
    },
    {
      "epoch": 2.655300546448087,
      "grad_norm": 0.7193872098032397,
      "learning_rate": 5.3622715370766154e-05,
      "loss": 11.868,
      "step": 15185
    },
    {
      "epoch": 2.6554754098360656,
      "grad_norm": 0.6154814527373977,
      "learning_rate": 5.3610169939785615e-05,
      "loss": 11.8303,
      "step": 15186
    },
    {
      "epoch": 2.655650273224044,
      "grad_norm": 0.6416631325644387,
      "learning_rate": 5.359762543906368e-05,
      "loss": 11.8715,
      "step": 15187
    },
    {
      "epoch": 2.655825136612022,
      "grad_norm": 0.6032095526813038,
      "learning_rate": 5.358508186885178e-05,
      "loss": 11.8028,
      "step": 15188
    },
    {
      "epoch": 2.656,
      "grad_norm": 0.6528706237117993,
      "learning_rate": 5.357253922940152e-05,
      "loss": 11.5736,
      "step": 15189
    },
    {
      "epoch": 2.656174863387978,
      "grad_norm": 0.6068592504885885,
      "learning_rate": 5.355999752096434e-05,
      "loss": 11.9327,
      "step": 15190
    },
    {
      "epoch": 2.6563497267759564,
      "grad_norm": 0.6144274888393301,
      "learning_rate": 5.354745674379187e-05,
      "loss": 11.9478,
      "step": 15191
    },
    {
      "epoch": 2.6565245901639343,
      "grad_norm": 0.6532517797273026,
      "learning_rate": 5.353491689813552e-05,
      "loss": 11.794,
      "step": 15192
    },
    {
      "epoch": 2.6566994535519126,
      "grad_norm": 0.6105345814672368,
      "learning_rate": 5.352237798424674e-05,
      "loss": 11.9101,
      "step": 15193
    },
    {
      "epoch": 2.6568743169398905,
      "grad_norm": 0.6610161006116813,
      "learning_rate": 5.3509840002376974e-05,
      "loss": 11.7678,
      "step": 15194
    },
    {
      "epoch": 2.657049180327869,
      "grad_norm": 0.6135800629934834,
      "learning_rate": 5.349730295277764e-05,
      "loss": 11.7902,
      "step": 15195
    },
    {
      "epoch": 2.657224043715847,
      "grad_norm": 0.6516197277511211,
      "learning_rate": 5.348476683570023e-05,
      "loss": 11.8837,
      "step": 15196
    },
    {
      "epoch": 2.657398907103825,
      "grad_norm": 0.6488478911127089,
      "learning_rate": 5.347223165139601e-05,
      "loss": 12.0261,
      "step": 15197
    },
    {
      "epoch": 2.6575737704918034,
      "grad_norm": 0.6312275084620909,
      "learning_rate": 5.345969740011643e-05,
      "loss": 11.9266,
      "step": 15198
    },
    {
      "epoch": 2.6577486338797813,
      "grad_norm": 0.6350067888541857,
      "learning_rate": 5.344716408211279e-05,
      "loss": 11.7722,
      "step": 15199
    },
    {
      "epoch": 2.6579234972677597,
      "grad_norm": 0.5874728364045951,
      "learning_rate": 5.343463169763643e-05,
      "loss": 11.777,
      "step": 15200
    },
    {
      "epoch": 2.6580983606557376,
      "grad_norm": 0.6225483492921543,
      "learning_rate": 5.3422100246938745e-05,
      "loss": 11.8164,
      "step": 15201
    },
    {
      "epoch": 2.658273224043716,
      "grad_norm": 0.6067894552359231,
      "learning_rate": 5.3409569730270925e-05,
      "loss": 11.9726,
      "step": 15202
    },
    {
      "epoch": 2.658448087431694,
      "grad_norm": 0.6315384419427821,
      "learning_rate": 5.339704014788428e-05,
      "loss": 11.8202,
      "step": 15203
    },
    {
      "epoch": 2.658622950819672,
      "grad_norm": 0.6180682592258216,
      "learning_rate": 5.338451150003008e-05,
      "loss": 12.0463,
      "step": 15204
    },
    {
      "epoch": 2.6587978142076505,
      "grad_norm": 0.6480473554481914,
      "learning_rate": 5.3371983786959545e-05,
      "loss": 12.0006,
      "step": 15205
    },
    {
      "epoch": 2.6589726775956284,
      "grad_norm": 0.5810790266776082,
      "learning_rate": 5.335945700892391e-05,
      "loss": 11.855,
      "step": 15206
    },
    {
      "epoch": 2.6591475409836063,
      "grad_norm": 0.5877183626720373,
      "learning_rate": 5.334693116617435e-05,
      "loss": 11.8696,
      "step": 15207
    },
    {
      "epoch": 2.6593224043715846,
      "grad_norm": 0.6333657695489597,
      "learning_rate": 5.333440625896208e-05,
      "loss": 11.8814,
      "step": 15208
    },
    {
      "epoch": 2.659497267759563,
      "grad_norm": 0.6711816020425619,
      "learning_rate": 5.3321882287538235e-05,
      "loss": 12.1617,
      "step": 15209
    },
    {
      "epoch": 2.659672131147541,
      "grad_norm": 0.6157937322500691,
      "learning_rate": 5.330935925215398e-05,
      "loss": 11.7936,
      "step": 15210
    },
    {
      "epoch": 2.659846994535519,
      "grad_norm": 0.6013866470853205,
      "learning_rate": 5.329683715306043e-05,
      "loss": 11.5913,
      "step": 15211
    },
    {
      "epoch": 2.660021857923497,
      "grad_norm": 0.6481542015833434,
      "learning_rate": 5.3284315990508695e-05,
      "loss": 11.7349,
      "step": 15212
    },
    {
      "epoch": 2.6601967213114754,
      "grad_norm": 0.6555482957666745,
      "learning_rate": 5.3271795764749856e-05,
      "loss": 11.9489,
      "step": 15213
    },
    {
      "epoch": 2.660371584699454,
      "grad_norm": 0.6751829323136356,
      "learning_rate": 5.325927647603498e-05,
      "loss": 11.732,
      "step": 15214
    },
    {
      "epoch": 2.6605464480874317,
      "grad_norm": 0.6167609207179241,
      "learning_rate": 5.3246758124615126e-05,
      "loss": 11.8471,
      "step": 15215
    },
    {
      "epoch": 2.6607213114754096,
      "grad_norm": 0.6095720056733949,
      "learning_rate": 5.3234240710741337e-05,
      "loss": 11.7927,
      "step": 15216
    },
    {
      "epoch": 2.660896174863388,
      "grad_norm": 0.7362992485764166,
      "learning_rate": 5.322172423466461e-05,
      "loss": 11.6952,
      "step": 15217
    },
    {
      "epoch": 2.6610710382513663,
      "grad_norm": 0.6743185088993211,
      "learning_rate": 5.3209208696635926e-05,
      "loss": 12.043,
      "step": 15218
    },
    {
      "epoch": 2.661245901639344,
      "grad_norm": 0.5918845057228287,
      "learning_rate": 5.319669409690628e-05,
      "loss": 11.8392,
      "step": 15219
    },
    {
      "epoch": 2.6614207650273225,
      "grad_norm": 0.6817820016336789,
      "learning_rate": 5.318418043572661e-05,
      "loss": 11.8843,
      "step": 15220
    },
    {
      "epoch": 2.6615956284153004,
      "grad_norm": 0.5961173766390422,
      "learning_rate": 5.3171667713347886e-05,
      "loss": 11.5192,
      "step": 15221
    },
    {
      "epoch": 2.6617704918032787,
      "grad_norm": 0.6747458783544741,
      "learning_rate": 5.3159155930021e-05,
      "loss": 11.8489,
      "step": 15222
    },
    {
      "epoch": 2.661945355191257,
      "grad_norm": 0.7132266308718617,
      "learning_rate": 5.314664508599685e-05,
      "loss": 11.993,
      "step": 15223
    },
    {
      "epoch": 2.662120218579235,
      "grad_norm": 0.5930305018125667,
      "learning_rate": 5.3134135181526336e-05,
      "loss": 11.6503,
      "step": 15224
    },
    {
      "epoch": 2.662295081967213,
      "grad_norm": 0.6573782240328719,
      "learning_rate": 5.31216262168603e-05,
      "loss": 11.8664,
      "step": 15225
    },
    {
      "epoch": 2.662469945355191,
      "grad_norm": 0.6773203248010394,
      "learning_rate": 5.3109118192249596e-05,
      "loss": 11.985,
      "step": 15226
    },
    {
      "epoch": 2.6626448087431696,
      "grad_norm": 0.6104265328703852,
      "learning_rate": 5.309661110794505e-05,
      "loss": 11.7703,
      "step": 15227
    },
    {
      "epoch": 2.6628196721311475,
      "grad_norm": 0.7016700259379303,
      "learning_rate": 5.3084104964197465e-05,
      "loss": 12.0624,
      "step": 15228
    },
    {
      "epoch": 2.662994535519126,
      "grad_norm": 0.641267021180419,
      "learning_rate": 5.307159976125762e-05,
      "loss": 11.8645,
      "step": 15229
    },
    {
      "epoch": 2.6631693989071037,
      "grad_norm": 0.6077527634556784,
      "learning_rate": 5.3059095499376285e-05,
      "loss": 11.9155,
      "step": 15230
    },
    {
      "epoch": 2.663344262295082,
      "grad_norm": 0.6299800211151344,
      "learning_rate": 5.304659217880423e-05,
      "loss": 11.8383,
      "step": 15231
    },
    {
      "epoch": 2.6635191256830604,
      "grad_norm": 0.6245311427196711,
      "learning_rate": 5.3034089799792155e-05,
      "loss": 11.9598,
      "step": 15232
    },
    {
      "epoch": 2.6636939890710383,
      "grad_norm": 0.6275936484731377,
      "learning_rate": 5.302158836259079e-05,
      "loss": 11.86,
      "step": 15233
    },
    {
      "epoch": 2.663868852459016,
      "grad_norm": 0.6390095980100757,
      "learning_rate": 5.300908786745081e-05,
      "loss": 11.8855,
      "step": 15234
    },
    {
      "epoch": 2.6640437158469945,
      "grad_norm": 0.6350152804534355,
      "learning_rate": 5.299658831462291e-05,
      "loss": 12.0516,
      "step": 15235
    },
    {
      "epoch": 2.664218579234973,
      "grad_norm": 0.6140773525570098,
      "learning_rate": 5.298408970435773e-05,
      "loss": 11.9778,
      "step": 15236
    },
    {
      "epoch": 2.6643934426229507,
      "grad_norm": 0.6669902968995871,
      "learning_rate": 5.29715920369059e-05,
      "loss": 11.8401,
      "step": 15237
    },
    {
      "epoch": 2.664568306010929,
      "grad_norm": 0.717002122963436,
      "learning_rate": 5.2959095312518056e-05,
      "loss": 11.8376,
      "step": 15238
    },
    {
      "epoch": 2.664743169398907,
      "grad_norm": 0.6614927337345872,
      "learning_rate": 5.2946599531444784e-05,
      "loss": 12.0629,
      "step": 15239
    },
    {
      "epoch": 2.6649180327868853,
      "grad_norm": 0.6453130019713201,
      "learning_rate": 5.293410469393667e-05,
      "loss": 11.8205,
      "step": 15240
    },
    {
      "epoch": 2.6650928961748632,
      "grad_norm": 0.6861339648169501,
      "learning_rate": 5.2921610800244256e-05,
      "loss": 11.6642,
      "step": 15241
    },
    {
      "epoch": 2.6652677595628416,
      "grad_norm": 0.613857944233135,
      "learning_rate": 5.2909117850618103e-05,
      "loss": 11.7841,
      "step": 15242
    },
    {
      "epoch": 2.6654426229508195,
      "grad_norm": 0.6827652764231236,
      "learning_rate": 5.289662584530872e-05,
      "loss": 11.9282,
      "step": 15243
    },
    {
      "epoch": 2.665617486338798,
      "grad_norm": 0.6293548254639799,
      "learning_rate": 5.2884134784566616e-05,
      "loss": 11.7349,
      "step": 15244
    },
    {
      "epoch": 2.665792349726776,
      "grad_norm": 0.6498150386380043,
      "learning_rate": 5.28716446686423e-05,
      "loss": 11.6763,
      "step": 15245
    },
    {
      "epoch": 2.665967213114754,
      "grad_norm": 0.6837835860150919,
      "learning_rate": 5.285915549778617e-05,
      "loss": 11.6131,
      "step": 15246
    },
    {
      "epoch": 2.6661420765027324,
      "grad_norm": 0.7015333560990196,
      "learning_rate": 5.284666727224878e-05,
      "loss": 11.9426,
      "step": 15247
    },
    {
      "epoch": 2.6663169398907103,
      "grad_norm": 0.6720642084487584,
      "learning_rate": 5.2834179992280394e-05,
      "loss": 11.8797,
      "step": 15248
    },
    {
      "epoch": 2.6664918032786886,
      "grad_norm": 0.606504796431965,
      "learning_rate": 5.282169365813158e-05,
      "loss": 11.5775,
      "step": 15249
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.6859085102853963,
      "learning_rate": 5.280920827005265e-05,
      "loss": 12.0309,
      "step": 15250
    },
    {
      "epoch": 2.666841530054645,
      "grad_norm": 0.6854430186791496,
      "learning_rate": 5.2796723828293995e-05,
      "loss": 11.7313,
      "step": 15251
    },
    {
      "epoch": 2.6670163934426228,
      "grad_norm": 0.6175659962821684,
      "learning_rate": 5.2784240333105984e-05,
      "loss": 11.6611,
      "step": 15252
    },
    {
      "epoch": 2.667191256830601,
      "grad_norm": 0.6716427068880043,
      "learning_rate": 5.277175778473886e-05,
      "loss": 11.817,
      "step": 15253
    },
    {
      "epoch": 2.6673661202185794,
      "grad_norm": 0.7648398127741979,
      "learning_rate": 5.275927618344304e-05,
      "loss": 11.7874,
      "step": 15254
    },
    {
      "epoch": 2.6675409836065573,
      "grad_norm": 0.7059044235985859,
      "learning_rate": 5.2746795529468777e-05,
      "loss": 11.6958,
      "step": 15255
    },
    {
      "epoch": 2.6677158469945357,
      "grad_norm": 0.7158006641718294,
      "learning_rate": 5.2734315823066396e-05,
      "loss": 11.7433,
      "step": 15256
    },
    {
      "epoch": 2.6678907103825136,
      "grad_norm": 0.7325424845290721,
      "learning_rate": 5.272183706448606e-05,
      "loss": 11.5867,
      "step": 15257
    },
    {
      "epoch": 2.668065573770492,
      "grad_norm": 0.6337922619771748,
      "learning_rate": 5.270935925397802e-05,
      "loss": 11.8054,
      "step": 15258
    },
    {
      "epoch": 2.66824043715847,
      "grad_norm": 0.6594076298314215,
      "learning_rate": 5.269688239179258e-05,
      "loss": 11.7171,
      "step": 15259
    },
    {
      "epoch": 2.668415300546448,
      "grad_norm": 0.6703016687335889,
      "learning_rate": 5.268440647817987e-05,
      "loss": 11.7222,
      "step": 15260
    },
    {
      "epoch": 2.668590163934426,
      "grad_norm": 0.6523757455602458,
      "learning_rate": 5.267193151339013e-05,
      "loss": 11.5736,
      "step": 15261
    },
    {
      "epoch": 2.6687650273224044,
      "grad_norm": 0.6192823010969685,
      "learning_rate": 5.2659457497673404e-05,
      "loss": 11.8024,
      "step": 15262
    },
    {
      "epoch": 2.6689398907103827,
      "grad_norm": 0.6404527723476493,
      "learning_rate": 5.2646984431279946e-05,
      "loss": 11.9122,
      "step": 15263
    },
    {
      "epoch": 2.6691147540983606,
      "grad_norm": 0.6603185192070806,
      "learning_rate": 5.263451231445984e-05,
      "loss": 11.7683,
      "step": 15264
    },
    {
      "epoch": 2.6692896174863385,
      "grad_norm": 0.6705380028343438,
      "learning_rate": 5.262204114746324e-05,
      "loss": 11.8996,
      "step": 15265
    },
    {
      "epoch": 2.669464480874317,
      "grad_norm": 0.6522230322032153,
      "learning_rate": 5.260957093054014e-05,
      "loss": 11.8912,
      "step": 15266
    },
    {
      "epoch": 2.669639344262295,
      "grad_norm": 0.6097308908798315,
      "learning_rate": 5.259710166394062e-05,
      "loss": 11.6262,
      "step": 15267
    },
    {
      "epoch": 2.669814207650273,
      "grad_norm": 0.6540542883539449,
      "learning_rate": 5.2584633347914793e-05,
      "loss": 11.6906,
      "step": 15268
    },
    {
      "epoch": 2.6699890710382514,
      "grad_norm": 0.5997273951112635,
      "learning_rate": 5.257216598271264e-05,
      "loss": 11.7763,
      "step": 15269
    },
    {
      "epoch": 2.6701639344262293,
      "grad_norm": 0.6865228306984695,
      "learning_rate": 5.255969956858423e-05,
      "loss": 11.7143,
      "step": 15270
    },
    {
      "epoch": 2.6703387978142077,
      "grad_norm": 0.6630708964852273,
      "learning_rate": 5.2547234105779464e-05,
      "loss": 11.7694,
      "step": 15271
    },
    {
      "epoch": 2.670513661202186,
      "grad_norm": 0.7924063713019158,
      "learning_rate": 5.2534769594548315e-05,
      "loss": 11.78,
      "step": 15272
    },
    {
      "epoch": 2.670688524590164,
      "grad_norm": 0.6444937385657691,
      "learning_rate": 5.252230603514081e-05,
      "loss": 11.8761,
      "step": 15273
    },
    {
      "epoch": 2.670863387978142,
      "grad_norm": 0.6453990533751427,
      "learning_rate": 5.250984342780689e-05,
      "loss": 11.9839,
      "step": 15274
    },
    {
      "epoch": 2.67103825136612,
      "grad_norm": 0.8095499321769988,
      "learning_rate": 5.249738177279638e-05,
      "loss": 11.8739,
      "step": 15275
    },
    {
      "epoch": 2.6712131147540985,
      "grad_norm": 0.7033643236996117,
      "learning_rate": 5.2484921070359226e-05,
      "loss": 11.8744,
      "step": 15276
    },
    {
      "epoch": 2.6713879781420764,
      "grad_norm": 0.7362585129294723,
      "learning_rate": 5.247246132074524e-05,
      "loss": 11.7998,
      "step": 15277
    },
    {
      "epoch": 2.6715628415300547,
      "grad_norm": 0.8213646839099171,
      "learning_rate": 5.246000252420438e-05,
      "loss": 11.9151,
      "step": 15278
    },
    {
      "epoch": 2.6717377049180326,
      "grad_norm": 0.7666460194286508,
      "learning_rate": 5.244754468098648e-05,
      "loss": 11.8348,
      "step": 15279
    },
    {
      "epoch": 2.671912568306011,
      "grad_norm": 0.6995392090636529,
      "learning_rate": 5.2435087791341276e-05,
      "loss": 11.8265,
      "step": 15280
    },
    {
      "epoch": 2.6720874316939893,
      "grad_norm": 0.6688234243978648,
      "learning_rate": 5.242263185551859e-05,
      "loss": 11.9454,
      "step": 15281
    },
    {
      "epoch": 2.672262295081967,
      "grad_norm": 0.762163492968303,
      "learning_rate": 5.241017687376819e-05,
      "loss": 11.8465,
      "step": 15282
    },
    {
      "epoch": 2.672437158469945,
      "grad_norm": 0.6682120827581862,
      "learning_rate": 5.239772284633995e-05,
      "loss": 11.9842,
      "step": 15283
    },
    {
      "epoch": 2.6726120218579235,
      "grad_norm": 0.6677097270767994,
      "learning_rate": 5.238526977348347e-05,
      "loss": 11.9388,
      "step": 15284
    },
    {
      "epoch": 2.672786885245902,
      "grad_norm": 0.7302049210269256,
      "learning_rate": 5.237281765544852e-05,
      "loss": 11.5763,
      "step": 15285
    },
    {
      "epoch": 2.6729617486338797,
      "grad_norm": 0.6708800061358591,
      "learning_rate": 5.236036649248479e-05,
      "loss": 11.9495,
      "step": 15286
    },
    {
      "epoch": 2.673136612021858,
      "grad_norm": 0.632571042464437,
      "learning_rate": 5.234791628484203e-05,
      "loss": 11.8134,
      "step": 15287
    },
    {
      "epoch": 2.673311475409836,
      "grad_norm": 0.6795493691937815,
      "learning_rate": 5.233546703276988e-05,
      "loss": 11.8952,
      "step": 15288
    },
    {
      "epoch": 2.6734863387978143,
      "grad_norm": 0.6605368308720415,
      "learning_rate": 5.232301873651794e-05,
      "loss": 11.9946,
      "step": 15289
    },
    {
      "epoch": 2.6736612021857926,
      "grad_norm": 0.6921734565755487,
      "learning_rate": 5.231057139633586e-05,
      "loss": 12.0402,
      "step": 15290
    },
    {
      "epoch": 2.6738360655737705,
      "grad_norm": 0.6093059679462148,
      "learning_rate": 5.22981250124732e-05,
      "loss": 11.5751,
      "step": 15291
    },
    {
      "epoch": 2.6740109289617484,
      "grad_norm": 0.6974891825403505,
      "learning_rate": 5.228567958517968e-05,
      "loss": 11.8135,
      "step": 15292
    },
    {
      "epoch": 2.6741857923497268,
      "grad_norm": 0.6603055054470801,
      "learning_rate": 5.227323511470474e-05,
      "loss": 11.9107,
      "step": 15293
    },
    {
      "epoch": 2.674360655737705,
      "grad_norm": 0.7644821877563379,
      "learning_rate": 5.2260791601298e-05,
      "loss": 11.8072,
      "step": 15294
    },
    {
      "epoch": 2.674535519125683,
      "grad_norm": 0.7061746502155332,
      "learning_rate": 5.224834904520895e-05,
      "loss": 11.7699,
      "step": 15295
    },
    {
      "epoch": 2.6747103825136613,
      "grad_norm": 0.6883958981408143,
      "learning_rate": 5.2235907446687085e-05,
      "loss": 11.8357,
      "step": 15296
    },
    {
      "epoch": 2.6748852459016392,
      "grad_norm": 0.6516596865666912,
      "learning_rate": 5.2223466805982e-05,
      "loss": 11.833,
      "step": 15297
    },
    {
      "epoch": 2.6750601092896176,
      "grad_norm": 0.6541547775827917,
      "learning_rate": 5.2211027123343074e-05,
      "loss": 11.9691,
      "step": 15298
    },
    {
      "epoch": 2.6752349726775955,
      "grad_norm": 0.6376449238426215,
      "learning_rate": 5.2198588399019765e-05,
      "loss": 11.5731,
      "step": 15299
    },
    {
      "epoch": 2.675409836065574,
      "grad_norm": 0.6128606314876007,
      "learning_rate": 5.2186150633261545e-05,
      "loss": 11.8272,
      "step": 15300
    },
    {
      "epoch": 2.6755846994535517,
      "grad_norm": 0.70790313588712,
      "learning_rate": 5.2173713826317815e-05,
      "loss": 11.6376,
      "step": 15301
    },
    {
      "epoch": 2.67575956284153,
      "grad_norm": 0.6733614528637792,
      "learning_rate": 5.216127797843795e-05,
      "loss": 11.7463,
      "step": 15302
    },
    {
      "epoch": 2.6759344262295084,
      "grad_norm": 0.7846310981702818,
      "learning_rate": 5.214884308987136e-05,
      "loss": 11.8474,
      "step": 15303
    },
    {
      "epoch": 2.6761092896174863,
      "grad_norm": 0.6164897685510912,
      "learning_rate": 5.2136409160867395e-05,
      "loss": 11.7513,
      "step": 15304
    },
    {
      "epoch": 2.6762841530054646,
      "grad_norm": 0.6430281471678293,
      "learning_rate": 5.212397619167537e-05,
      "loss": 11.8855,
      "step": 15305
    },
    {
      "epoch": 2.6764590163934425,
      "grad_norm": 0.6156888684161032,
      "learning_rate": 5.2111544182544624e-05,
      "loss": 11.796,
      "step": 15306
    },
    {
      "epoch": 2.676633879781421,
      "grad_norm": 0.62335000880787,
      "learning_rate": 5.2099113133724465e-05,
      "loss": 11.888,
      "step": 15307
    },
    {
      "epoch": 2.6768087431693988,
      "grad_norm": 0.6803322854367523,
      "learning_rate": 5.208668304546416e-05,
      "loss": 11.7495,
      "step": 15308
    },
    {
      "epoch": 2.676983606557377,
      "grad_norm": 0.6469485197200432,
      "learning_rate": 5.207425391801297e-05,
      "loss": 11.8841,
      "step": 15309
    },
    {
      "epoch": 2.677158469945355,
      "grad_norm": 0.6549879653104034,
      "learning_rate": 5.206182575162013e-05,
      "loss": 11.3968,
      "step": 15310
    },
    {
      "epoch": 2.6773333333333333,
      "grad_norm": 0.6348620604006411,
      "learning_rate": 5.204939854653489e-05,
      "loss": 11.9055,
      "step": 15311
    },
    {
      "epoch": 2.6775081967213117,
      "grad_norm": 0.6017022062619752,
      "learning_rate": 5.2036972303006426e-05,
      "loss": 11.9775,
      "step": 15312
    },
    {
      "epoch": 2.6776830601092896,
      "grad_norm": 0.6594619362454393,
      "learning_rate": 5.2024547021283944e-05,
      "loss": 11.8877,
      "step": 15313
    },
    {
      "epoch": 2.677857923497268,
      "grad_norm": 0.662597190731053,
      "learning_rate": 5.2012122701616584e-05,
      "loss": 11.964,
      "step": 15314
    },
    {
      "epoch": 2.678032786885246,
      "grad_norm": 0.6401564849457114,
      "learning_rate": 5.1999699344253516e-05,
      "loss": 11.5886,
      "step": 15315
    },
    {
      "epoch": 2.678207650273224,
      "grad_norm": 0.6625840632042813,
      "learning_rate": 5.1987276949443856e-05,
      "loss": 11.8979,
      "step": 15316
    },
    {
      "epoch": 2.678382513661202,
      "grad_norm": 0.6254216579081314,
      "learning_rate": 5.1974855517436706e-05,
      "loss": 11.8208,
      "step": 15317
    },
    {
      "epoch": 2.6785573770491804,
      "grad_norm": 0.6345114331152364,
      "learning_rate": 5.196243504848116e-05,
      "loss": 11.8665,
      "step": 15318
    },
    {
      "epoch": 2.6787322404371583,
      "grad_norm": 0.6707930569294295,
      "learning_rate": 5.195001554282629e-05,
      "loss": 11.6966,
      "step": 15319
    },
    {
      "epoch": 2.6789071038251366,
      "grad_norm": 0.6425468538189257,
      "learning_rate": 5.193759700072114e-05,
      "loss": 11.9956,
      "step": 15320
    },
    {
      "epoch": 2.679081967213115,
      "grad_norm": 0.6315506568739222,
      "learning_rate": 5.192517942241474e-05,
      "loss": 11.9987,
      "step": 15321
    },
    {
      "epoch": 2.679256830601093,
      "grad_norm": 0.6645718197647521,
      "learning_rate": 5.19127628081561e-05,
      "loss": 12.0564,
      "step": 15322
    },
    {
      "epoch": 2.6794316939890708,
      "grad_norm": 0.7057042131524127,
      "learning_rate": 5.1900347158194215e-05,
      "loss": 11.8026,
      "step": 15323
    },
    {
      "epoch": 2.679606557377049,
      "grad_norm": 0.7274554760136207,
      "learning_rate": 5.1887932472778056e-05,
      "loss": 11.8544,
      "step": 15324
    },
    {
      "epoch": 2.6797814207650275,
      "grad_norm": 0.6347901267213921,
      "learning_rate": 5.1875518752156574e-05,
      "loss": 11.8872,
      "step": 15325
    },
    {
      "epoch": 2.6799562841530054,
      "grad_norm": 0.6264099203991927,
      "learning_rate": 5.186310599657871e-05,
      "loss": 11.6903,
      "step": 15326
    },
    {
      "epoch": 2.6801311475409837,
      "grad_norm": 0.6816175022704,
      "learning_rate": 5.185069420629336e-05,
      "loss": 11.3827,
      "step": 15327
    },
    {
      "epoch": 2.6803060109289616,
      "grad_norm": 0.6985559349274044,
      "learning_rate": 5.1838283381549435e-05,
      "loss": 11.7618,
      "step": 15328
    },
    {
      "epoch": 2.68048087431694,
      "grad_norm": 0.6379776681743832,
      "learning_rate": 5.18258735225958e-05,
      "loss": 11.6389,
      "step": 15329
    },
    {
      "epoch": 2.6806557377049183,
      "grad_norm": 0.6840180025246133,
      "learning_rate": 5.181346462968131e-05,
      "loss": 11.7376,
      "step": 15330
    },
    {
      "epoch": 2.680830601092896,
      "grad_norm": 0.7108396651280189,
      "learning_rate": 5.1801056703054816e-05,
      "loss": 11.9384,
      "step": 15331
    },
    {
      "epoch": 2.681005464480874,
      "grad_norm": 0.6394754316542183,
      "learning_rate": 5.178864974296511e-05,
      "loss": 11.7879,
      "step": 15332
    },
    {
      "epoch": 2.6811803278688524,
      "grad_norm": 0.7865242719799449,
      "learning_rate": 5.1776243749661026e-05,
      "loss": 11.7958,
      "step": 15333
    },
    {
      "epoch": 2.6813551912568308,
      "grad_norm": 0.6745673952706064,
      "learning_rate": 5.17638387233913e-05,
      "loss": 11.7356,
      "step": 15334
    },
    {
      "epoch": 2.6815300546448086,
      "grad_norm": 0.6811379318659547,
      "learning_rate": 5.175143466440473e-05,
      "loss": 11.9755,
      "step": 15335
    },
    {
      "epoch": 2.681704918032787,
      "grad_norm": 0.5987518350703351,
      "learning_rate": 5.173903157295004e-05,
      "loss": 11.7684,
      "step": 15336
    },
    {
      "epoch": 2.681879781420765,
      "grad_norm": 0.6749779304536481,
      "learning_rate": 5.172662944927593e-05,
      "loss": 11.9263,
      "step": 15337
    },
    {
      "epoch": 2.6820546448087432,
      "grad_norm": 0.6672868041153662,
      "learning_rate": 5.171422829363113e-05,
      "loss": 11.8541,
      "step": 15338
    },
    {
      "epoch": 2.6822295081967216,
      "grad_norm": 0.6542059064874325,
      "learning_rate": 5.1701828106264305e-05,
      "loss": 11.6041,
      "step": 15339
    },
    {
      "epoch": 2.6824043715846995,
      "grad_norm": 0.6814355550954961,
      "learning_rate": 5.1689428887424117e-05,
      "loss": 11.8606,
      "step": 15340
    },
    {
      "epoch": 2.6825792349726774,
      "grad_norm": 0.6664595424552385,
      "learning_rate": 5.167703063735923e-05,
      "loss": 11.7267,
      "step": 15341
    },
    {
      "epoch": 2.6827540983606557,
      "grad_norm": 0.6335008441155449,
      "learning_rate": 5.1664633356318236e-05,
      "loss": 11.859,
      "step": 15342
    },
    {
      "epoch": 2.682928961748634,
      "grad_norm": 0.6486207566850217,
      "learning_rate": 5.165223704454979e-05,
      "loss": 11.988,
      "step": 15343
    },
    {
      "epoch": 2.683103825136612,
      "grad_norm": 0.7790404159043455,
      "learning_rate": 5.163984170230236e-05,
      "loss": 11.7958,
      "step": 15344
    },
    {
      "epoch": 2.6832786885245903,
      "grad_norm": 0.6424979754183475,
      "learning_rate": 5.162744732982463e-05,
      "loss": 11.9743,
      "step": 15345
    },
    {
      "epoch": 2.683453551912568,
      "grad_norm": 0.6050452730426569,
      "learning_rate": 5.161505392736511e-05,
      "loss": 11.661,
      "step": 15346
    },
    {
      "epoch": 2.6836284153005465,
      "grad_norm": 0.7153890902526292,
      "learning_rate": 5.160266149517232e-05,
      "loss": 11.4201,
      "step": 15347
    },
    {
      "epoch": 2.683803278688525,
      "grad_norm": 0.6799332379383579,
      "learning_rate": 5.159027003349479e-05,
      "loss": 11.751,
      "step": 15348
    },
    {
      "epoch": 2.6839781420765028,
      "grad_norm": 0.6909151417229615,
      "learning_rate": 5.157787954258092e-05,
      "loss": 11.9794,
      "step": 15349
    },
    {
      "epoch": 2.6841530054644807,
      "grad_norm": 0.6562644274402524,
      "learning_rate": 5.156549002267927e-05,
      "loss": 11.7782,
      "step": 15350
    },
    {
      "epoch": 2.684327868852459,
      "grad_norm": 0.7184499659241,
      "learning_rate": 5.155310147403827e-05,
      "loss": 11.6499,
      "step": 15351
    },
    {
      "epoch": 2.6845027322404373,
      "grad_norm": 0.6228319516029733,
      "learning_rate": 5.154071389690638e-05,
      "loss": 11.955,
      "step": 15352
    },
    {
      "epoch": 2.6846775956284152,
      "grad_norm": 0.713440612026681,
      "learning_rate": 5.152832729153192e-05,
      "loss": 11.9554,
      "step": 15353
    },
    {
      "epoch": 2.6848524590163936,
      "grad_norm": 0.6931854540200606,
      "learning_rate": 5.1515941658163267e-05,
      "loss": 11.7419,
      "step": 15354
    },
    {
      "epoch": 2.6850273224043715,
      "grad_norm": 0.604418028489953,
      "learning_rate": 5.150355699704892e-05,
      "loss": 11.9497,
      "step": 15355
    },
    {
      "epoch": 2.68520218579235,
      "grad_norm": 0.6752385428181746,
      "learning_rate": 5.149117330843713e-05,
      "loss": 12.0247,
      "step": 15356
    },
    {
      "epoch": 2.6853770491803277,
      "grad_norm": 0.682309562198453,
      "learning_rate": 5.147879059257632e-05,
      "loss": 11.6665,
      "step": 15357
    },
    {
      "epoch": 2.685551912568306,
      "grad_norm": 0.6018922185884066,
      "learning_rate": 5.1466408849714655e-05,
      "loss": 11.9711,
      "step": 15358
    },
    {
      "epoch": 2.685726775956284,
      "grad_norm": 0.6056443365782357,
      "learning_rate": 5.1454028080100534e-05,
      "loss": 11.8877,
      "step": 15359
    },
    {
      "epoch": 2.6859016393442623,
      "grad_norm": 0.6701591942118436,
      "learning_rate": 5.144164828398221e-05,
      "loss": 11.6706,
      "step": 15360
    },
    {
      "epoch": 2.6860765027322406,
      "grad_norm": 0.6219501363620916,
      "learning_rate": 5.142926946160799e-05,
      "loss": 11.826,
      "step": 15361
    },
    {
      "epoch": 2.6862513661202185,
      "grad_norm": 0.6876858267865583,
      "learning_rate": 5.141689161322599e-05,
      "loss": 11.7096,
      "step": 15362
    },
    {
      "epoch": 2.686426229508197,
      "grad_norm": 0.6377665035264009,
      "learning_rate": 5.140451473908445e-05,
      "loss": 11.7828,
      "step": 15363
    },
    {
      "epoch": 2.6866010928961748,
      "grad_norm": 0.6625657581981194,
      "learning_rate": 5.1392138839431636e-05,
      "loss": 11.8372,
      "step": 15364
    },
    {
      "epoch": 2.686775956284153,
      "grad_norm": 0.6650928407073341,
      "learning_rate": 5.137976391451569e-05,
      "loss": 11.7201,
      "step": 15365
    },
    {
      "epoch": 2.686950819672131,
      "grad_norm": 0.682423593215074,
      "learning_rate": 5.13673899645848e-05,
      "loss": 11.6897,
      "step": 15366
    },
    {
      "epoch": 2.6871256830601093,
      "grad_norm": 0.7243849466864453,
      "learning_rate": 5.1355016989887015e-05,
      "loss": 11.7061,
      "step": 15367
    },
    {
      "epoch": 2.6873005464480872,
      "grad_norm": 0.6669947693290569,
      "learning_rate": 5.1342644990670476e-05,
      "loss": 12.002,
      "step": 15368
    },
    {
      "epoch": 2.6874754098360656,
      "grad_norm": 0.6559030036598233,
      "learning_rate": 5.1330273967183315e-05,
      "loss": 11.7697,
      "step": 15369
    },
    {
      "epoch": 2.687650273224044,
      "grad_norm": 0.6659084475742248,
      "learning_rate": 5.131790391967366e-05,
      "loss": 11.7314,
      "step": 15370
    },
    {
      "epoch": 2.687825136612022,
      "grad_norm": 0.6682044406127078,
      "learning_rate": 5.1305534848389445e-05,
      "loss": 12.0318,
      "step": 15371
    },
    {
      "epoch": 2.6879999999999997,
      "grad_norm": 0.6944038511249574,
      "learning_rate": 5.129316675357878e-05,
      "loss": 11.7053,
      "step": 15372
    },
    {
      "epoch": 2.688174863387978,
      "grad_norm": 0.6544102389451707,
      "learning_rate": 5.128079963548964e-05,
      "loss": 12.0264,
      "step": 15373
    },
    {
      "epoch": 2.6883497267759564,
      "grad_norm": 0.7021357436800667,
      "learning_rate": 5.1268433494370084e-05,
      "loss": 11.8183,
      "step": 15374
    },
    {
      "epoch": 2.6885245901639343,
      "grad_norm": 0.6927646039872485,
      "learning_rate": 5.12560683304681e-05,
      "loss": 11.6663,
      "step": 15375
    },
    {
      "epoch": 2.6886994535519126,
      "grad_norm": 0.6105475163516388,
      "learning_rate": 5.124370414403158e-05,
      "loss": 11.7782,
      "step": 15376
    },
    {
      "epoch": 2.6888743169398905,
      "grad_norm": 0.6481044222400446,
      "learning_rate": 5.123134093530848e-05,
      "loss": 11.8418,
      "step": 15377
    },
    {
      "epoch": 2.689049180327869,
      "grad_norm": 0.6501540718393963,
      "learning_rate": 5.12189787045467e-05,
      "loss": 12.0234,
      "step": 15378
    },
    {
      "epoch": 2.6892240437158472,
      "grad_norm": 0.7231134578128233,
      "learning_rate": 5.120661745199427e-05,
      "loss": 11.7359,
      "step": 15379
    },
    {
      "epoch": 2.689398907103825,
      "grad_norm": 0.6730897735146713,
      "learning_rate": 5.119425717789891e-05,
      "loss": 11.8726,
      "step": 15380
    },
    {
      "epoch": 2.689573770491803,
      "grad_norm": 0.6869554145364604,
      "learning_rate": 5.1181897882508554e-05,
      "loss": 11.9609,
      "step": 15381
    },
    {
      "epoch": 2.6897486338797814,
      "grad_norm": 0.6546459241194734,
      "learning_rate": 5.1169539566071e-05,
      "loss": 11.5832,
      "step": 15382
    },
    {
      "epoch": 2.6899234972677597,
      "grad_norm": 0.6236010020966869,
      "learning_rate": 5.1157182228834146e-05,
      "loss": 11.8019,
      "step": 15383
    },
    {
      "epoch": 2.6900983606557376,
      "grad_norm": 0.6536840063093017,
      "learning_rate": 5.1144825871045796e-05,
      "loss": 11.627,
      "step": 15384
    },
    {
      "epoch": 2.690273224043716,
      "grad_norm": 0.5908475009399965,
      "learning_rate": 5.1132470492953656e-05,
      "loss": 11.8942,
      "step": 15385
    },
    {
      "epoch": 2.690448087431694,
      "grad_norm": 0.6406824611098788,
      "learning_rate": 5.1120116094805514e-05,
      "loss": 11.9194,
      "step": 15386
    },
    {
      "epoch": 2.690622950819672,
      "grad_norm": 0.6860321632884131,
      "learning_rate": 5.11077626768491e-05,
      "loss": 11.9441,
      "step": 15387
    },
    {
      "epoch": 2.6907978142076505,
      "grad_norm": 0.6922255798170225,
      "learning_rate": 5.109541023933224e-05,
      "loss": 11.7365,
      "step": 15388
    },
    {
      "epoch": 2.6909726775956284,
      "grad_norm": 0.7136175781766873,
      "learning_rate": 5.108305878250251e-05,
      "loss": 11.6525,
      "step": 15389
    },
    {
      "epoch": 2.6911475409836063,
      "grad_norm": 0.6865615469483418,
      "learning_rate": 5.107070830660765e-05,
      "loss": 11.8408,
      "step": 15390
    },
    {
      "epoch": 2.6913224043715847,
      "grad_norm": 0.6245732402246393,
      "learning_rate": 5.1058358811895335e-05,
      "loss": 11.7883,
      "step": 15391
    },
    {
      "epoch": 2.691497267759563,
      "grad_norm": 0.6666491572744306,
      "learning_rate": 5.1046010298613144e-05,
      "loss": 11.8245,
      "step": 15392
    },
    {
      "epoch": 2.691672131147541,
      "grad_norm": 0.7595554226085478,
      "learning_rate": 5.103366276700884e-05,
      "loss": 11.7406,
      "step": 15393
    },
    {
      "epoch": 2.6918469945355192,
      "grad_norm": 0.6401289491640868,
      "learning_rate": 5.102131621732991e-05,
      "loss": 11.9742,
      "step": 15394
    },
    {
      "epoch": 2.692021857923497,
      "grad_norm": 0.6634584431109907,
      "learning_rate": 5.100897064982395e-05,
      "loss": 11.868,
      "step": 15395
    },
    {
      "epoch": 2.6921967213114755,
      "grad_norm": 0.7383483589341507,
      "learning_rate": 5.0996626064738576e-05,
      "loss": 11.5872,
      "step": 15396
    },
    {
      "epoch": 2.692371584699454,
      "grad_norm": 0.5994855921966429,
      "learning_rate": 5.09842824623213e-05,
      "loss": 11.4826,
      "step": 15397
    },
    {
      "epoch": 2.6925464480874317,
      "grad_norm": 0.6660835014984411,
      "learning_rate": 5.097193984281965e-05,
      "loss": 12.0063,
      "step": 15398
    },
    {
      "epoch": 2.6927213114754096,
      "grad_norm": 0.7123690205525285,
      "learning_rate": 5.0959598206481154e-05,
      "loss": 11.8311,
      "step": 15399
    },
    {
      "epoch": 2.692896174863388,
      "grad_norm": 0.7085506410687978,
      "learning_rate": 5.09472575535533e-05,
      "loss": 12.043,
      "step": 15400
    },
    {
      "epoch": 2.6930710382513663,
      "grad_norm": 0.6134337700539239,
      "learning_rate": 5.093491788428352e-05,
      "loss": 11.8166,
      "step": 15401
    },
    {
      "epoch": 2.693245901639344,
      "grad_norm": 0.6276079035284418,
      "learning_rate": 5.092257919891929e-05,
      "loss": 11.7832,
      "step": 15402
    },
    {
      "epoch": 2.6934207650273225,
      "grad_norm": 0.6464532639408512,
      "learning_rate": 5.091024149770805e-05,
      "loss": 11.8728,
      "step": 15403
    },
    {
      "epoch": 2.6935956284153004,
      "grad_norm": 0.6395756083838361,
      "learning_rate": 5.0897904780897174e-05,
      "loss": 11.8017,
      "step": 15404
    },
    {
      "epoch": 2.6937704918032788,
      "grad_norm": 0.671044291763096,
      "learning_rate": 5.088556904873407e-05,
      "loss": 12.0464,
      "step": 15405
    },
    {
      "epoch": 2.693945355191257,
      "grad_norm": 0.7021278739824582,
      "learning_rate": 5.087323430146612e-05,
      "loss": 11.7561,
      "step": 15406
    },
    {
      "epoch": 2.694120218579235,
      "grad_norm": 0.6743270090304513,
      "learning_rate": 5.086090053934066e-05,
      "loss": 11.8925,
      "step": 15407
    },
    {
      "epoch": 2.694295081967213,
      "grad_norm": 0.6634763271912999,
      "learning_rate": 5.084856776260503e-05,
      "loss": 11.8476,
      "step": 15408
    },
    {
      "epoch": 2.6944699453551912,
      "grad_norm": 0.6873327166250676,
      "learning_rate": 5.083623597150651e-05,
      "loss": 11.8766,
      "step": 15409
    },
    {
      "epoch": 2.6946448087431696,
      "grad_norm": 0.6793173893339227,
      "learning_rate": 5.0823905166292415e-05,
      "loss": 11.8915,
      "step": 15410
    },
    {
      "epoch": 2.6948196721311475,
      "grad_norm": 0.6268339319258005,
      "learning_rate": 5.081157534721002e-05,
      "loss": 11.7789,
      "step": 15411
    },
    {
      "epoch": 2.694994535519126,
      "grad_norm": 0.6487301025259504,
      "learning_rate": 5.079924651450656e-05,
      "loss": 11.6937,
      "step": 15412
    },
    {
      "epoch": 2.6951693989071037,
      "grad_norm": 0.652476896362163,
      "learning_rate": 5.078691866842928e-05,
      "loss": 11.6461,
      "step": 15413
    },
    {
      "epoch": 2.695344262295082,
      "grad_norm": 0.6612527186671775,
      "learning_rate": 5.077459180922539e-05,
      "loss": 11.8828,
      "step": 15414
    },
    {
      "epoch": 2.69551912568306,
      "grad_norm": 0.6272998872828183,
      "learning_rate": 5.0762265937142065e-05,
      "loss": 11.7939,
      "step": 15415
    },
    {
      "epoch": 2.6956939890710383,
      "grad_norm": 0.6240099770340934,
      "learning_rate": 5.07499410524265e-05,
      "loss": 11.8193,
      "step": 15416
    },
    {
      "epoch": 2.695868852459016,
      "grad_norm": 0.7199485821362835,
      "learning_rate": 5.0737617155325825e-05,
      "loss": 11.8805,
      "step": 15417
    },
    {
      "epoch": 2.6960437158469945,
      "grad_norm": 0.6384481620077073,
      "learning_rate": 5.0725294246087185e-05,
      "loss": 11.7657,
      "step": 15418
    },
    {
      "epoch": 2.696218579234973,
      "grad_norm": 0.671824098176975,
      "learning_rate": 5.071297232495769e-05,
      "loss": 11.8514,
      "step": 15419
    },
    {
      "epoch": 2.6963934426229508,
      "grad_norm": 0.700339851469467,
      "learning_rate": 5.070065139218443e-05,
      "loss": 11.809,
      "step": 15420
    },
    {
      "epoch": 2.696568306010929,
      "grad_norm": 0.7027829198204181,
      "learning_rate": 5.068833144801447e-05,
      "loss": 11.9673,
      "step": 15421
    },
    {
      "epoch": 2.696743169398907,
      "grad_norm": 0.6170800180280992,
      "learning_rate": 5.067601249269487e-05,
      "loss": 11.8857,
      "step": 15422
    },
    {
      "epoch": 2.6969180327868854,
      "grad_norm": 0.635429674478823,
      "learning_rate": 5.066369452647268e-05,
      "loss": 12.0547,
      "step": 15423
    },
    {
      "epoch": 2.6970928961748633,
      "grad_norm": 0.6920038977153082,
      "learning_rate": 5.065137754959488e-05,
      "loss": 11.824,
      "step": 15424
    },
    {
      "epoch": 2.6972677595628416,
      "grad_norm": 0.6211319096339598,
      "learning_rate": 5.063906156230853e-05,
      "loss": 11.8306,
      "step": 15425
    },
    {
      "epoch": 2.6974426229508195,
      "grad_norm": 0.6512429671822336,
      "learning_rate": 5.062674656486048e-05,
      "loss": 11.6285,
      "step": 15426
    },
    {
      "epoch": 2.697617486338798,
      "grad_norm": 0.6840798408390717,
      "learning_rate": 5.061443255749778e-05,
      "loss": 11.6745,
      "step": 15427
    },
    {
      "epoch": 2.697792349726776,
      "grad_norm": 0.7000758238823402,
      "learning_rate": 5.060211954046736e-05,
      "loss": 11.6809,
      "step": 15428
    },
    {
      "epoch": 2.697967213114754,
      "grad_norm": 0.6799916363698928,
      "learning_rate": 5.05898075140161e-05,
      "loss": 11.5677,
      "step": 15429
    },
    {
      "epoch": 2.698142076502732,
      "grad_norm": 0.7012548493480772,
      "learning_rate": 5.057749647839092e-05,
      "loss": 11.881,
      "step": 15430
    },
    {
      "epoch": 2.6983169398907103,
      "grad_norm": 0.8051335459101878,
      "learning_rate": 5.056518643383867e-05,
      "loss": 11.5979,
      "step": 15431
    },
    {
      "epoch": 2.6984918032786886,
      "grad_norm": 0.7431238010467243,
      "learning_rate": 5.0552877380606224e-05,
      "loss": 11.8384,
      "step": 15432
    },
    {
      "epoch": 2.6986666666666665,
      "grad_norm": 0.665236653850365,
      "learning_rate": 5.0540569318940414e-05,
      "loss": 11.6979,
      "step": 15433
    },
    {
      "epoch": 2.698841530054645,
      "grad_norm": 0.755662809695075,
      "learning_rate": 5.0528262249088075e-05,
      "loss": 11.9417,
      "step": 15434
    },
    {
      "epoch": 2.699016393442623,
      "grad_norm": 0.7034268002177051,
      "learning_rate": 5.051595617129591e-05,
      "loss": 11.7509,
      "step": 15435
    },
    {
      "epoch": 2.699191256830601,
      "grad_norm": 0.6780806444521266,
      "learning_rate": 5.050365108581079e-05,
      "loss": 11.4762,
      "step": 15436
    },
    {
      "epoch": 2.6993661202185795,
      "grad_norm": 0.7311637658752174,
      "learning_rate": 5.049134699287945e-05,
      "loss": 11.8995,
      "step": 15437
    },
    {
      "epoch": 2.6995409836065574,
      "grad_norm": 0.7666634960742272,
      "learning_rate": 5.04790438927486e-05,
      "loss": 11.7884,
      "step": 15438
    },
    {
      "epoch": 2.6997158469945353,
      "grad_norm": 0.6633341544632221,
      "learning_rate": 5.046674178566502e-05,
      "loss": 11.6829,
      "step": 15439
    },
    {
      "epoch": 2.6998907103825136,
      "grad_norm": 0.6190560734451392,
      "learning_rate": 5.045444067187528e-05,
      "loss": 11.7514,
      "step": 15440
    },
    {
      "epoch": 2.700065573770492,
      "grad_norm": 0.6741899636799512,
      "learning_rate": 5.044214055162615e-05,
      "loss": 11.9659,
      "step": 15441
    },
    {
      "epoch": 2.70024043715847,
      "grad_norm": 0.6828088366237104,
      "learning_rate": 5.042984142516428e-05,
      "loss": 11.771,
      "step": 15442
    },
    {
      "epoch": 2.700415300546448,
      "grad_norm": 0.664723763258667,
      "learning_rate": 5.0417543292736335e-05,
      "loss": 11.7232,
      "step": 15443
    },
    {
      "epoch": 2.700590163934426,
      "grad_norm": 0.614003802825256,
      "learning_rate": 5.0405246154588835e-05,
      "loss": 11.8843,
      "step": 15444
    },
    {
      "epoch": 2.7007650273224044,
      "grad_norm": 0.6245161148564847,
      "learning_rate": 5.039295001096838e-05,
      "loss": 11.8636,
      "step": 15445
    },
    {
      "epoch": 2.7009398907103828,
      "grad_norm": 0.7567326963982413,
      "learning_rate": 5.0380654862121644e-05,
      "loss": 11.9667,
      "step": 15446
    },
    {
      "epoch": 2.7011147540983607,
      "grad_norm": 0.6348185869655365,
      "learning_rate": 5.036836070829512e-05,
      "loss": 11.753,
      "step": 15447
    },
    {
      "epoch": 2.7012896174863386,
      "grad_norm": 0.6774042961562537,
      "learning_rate": 5.035606754973539e-05,
      "loss": 11.8685,
      "step": 15448
    },
    {
      "epoch": 2.701464480874317,
      "grad_norm": 0.6901696586004106,
      "learning_rate": 5.0343775386688906e-05,
      "loss": 11.9169,
      "step": 15449
    },
    {
      "epoch": 2.7016393442622952,
      "grad_norm": 0.6196564915321375,
      "learning_rate": 5.0331484219402134e-05,
      "loss": 11.824,
      "step": 15450
    },
    {
      "epoch": 2.701814207650273,
      "grad_norm": 0.6321423775414277,
      "learning_rate": 5.0319194048121664e-05,
      "loss": 11.8629,
      "step": 15451
    },
    {
      "epoch": 2.7019890710382515,
      "grad_norm": 0.6471035452929789,
      "learning_rate": 5.0306904873093876e-05,
      "loss": 11.7716,
      "step": 15452
    },
    {
      "epoch": 2.7021639344262294,
      "grad_norm": 0.7125807304195346,
      "learning_rate": 5.029461669456528e-05,
      "loss": 11.6702,
      "step": 15453
    },
    {
      "epoch": 2.7023387978142077,
      "grad_norm": 0.6503868403736893,
      "learning_rate": 5.0282329512782136e-05,
      "loss": 11.8343,
      "step": 15454
    },
    {
      "epoch": 2.702513661202186,
      "grad_norm": 0.6231111860873298,
      "learning_rate": 5.027004332799099e-05,
      "loss": 11.7427,
      "step": 15455
    },
    {
      "epoch": 2.702688524590164,
      "grad_norm": 0.707998901005318,
      "learning_rate": 5.025775814043816e-05,
      "loss": 11.636,
      "step": 15456
    },
    {
      "epoch": 2.702863387978142,
      "grad_norm": 0.706484061405206,
      "learning_rate": 5.024547395037005e-05,
      "loss": 11.8831,
      "step": 15457
    },
    {
      "epoch": 2.70303825136612,
      "grad_norm": 0.6423702639420797,
      "learning_rate": 5.0233190758032923e-05,
      "loss": 11.7666,
      "step": 15458
    },
    {
      "epoch": 2.7032131147540985,
      "grad_norm": 0.5954132323032407,
      "learning_rate": 5.022090856367307e-05,
      "loss": 12.0182,
      "step": 15459
    },
    {
      "epoch": 2.7033879781420764,
      "grad_norm": 0.6641620478253045,
      "learning_rate": 5.02086273675369e-05,
      "loss": 11.6577,
      "step": 15460
    },
    {
      "epoch": 2.7035628415300548,
      "grad_norm": 0.6833598675745383,
      "learning_rate": 5.0196347169870616e-05,
      "loss": 11.6637,
      "step": 15461
    },
    {
      "epoch": 2.7037377049180327,
      "grad_norm": 0.6853076380384839,
      "learning_rate": 5.018406797092053e-05,
      "loss": 11.28,
      "step": 15462
    },
    {
      "epoch": 2.703912568306011,
      "grad_norm": 0.6522410624130033,
      "learning_rate": 5.0171789770932806e-05,
      "loss": 11.9782,
      "step": 15463
    },
    {
      "epoch": 2.704087431693989,
      "grad_norm": 0.6656233108235744,
      "learning_rate": 5.0159512570153644e-05,
      "loss": 11.8167,
      "step": 15464
    },
    {
      "epoch": 2.7042622950819672,
      "grad_norm": 0.5889463411222594,
      "learning_rate": 5.014723636882932e-05,
      "loss": 11.8286,
      "step": 15465
    },
    {
      "epoch": 2.704437158469945,
      "grad_norm": 0.656387640293008,
      "learning_rate": 5.013496116720602e-05,
      "loss": 11.6692,
      "step": 15466
    },
    {
      "epoch": 2.7046120218579235,
      "grad_norm": 0.6561702255418286,
      "learning_rate": 5.0122686965529816e-05,
      "loss": 11.6782,
      "step": 15467
    },
    {
      "epoch": 2.704786885245902,
      "grad_norm": 0.6436058334438642,
      "learning_rate": 5.011041376404688e-05,
      "loss": 11.8828,
      "step": 15468
    },
    {
      "epoch": 2.7049617486338797,
      "grad_norm": 0.622465379982044,
      "learning_rate": 5.00981415630033e-05,
      "loss": 11.911,
      "step": 15469
    },
    {
      "epoch": 2.705136612021858,
      "grad_norm": 0.597700369846963,
      "learning_rate": 5.008587036264523e-05,
      "loss": 11.5594,
      "step": 15470
    },
    {
      "epoch": 2.705311475409836,
      "grad_norm": 0.6827396326850945,
      "learning_rate": 5.007360016321876e-05,
      "loss": 11.7221,
      "step": 15471
    },
    {
      "epoch": 2.7054863387978143,
      "grad_norm": 0.5886576763140408,
      "learning_rate": 5.0061330964969854e-05,
      "loss": 11.9974,
      "step": 15472
    },
    {
      "epoch": 2.705661202185792,
      "grad_norm": 0.6214469966183573,
      "learning_rate": 5.0049062768144586e-05,
      "loss": 11.8598,
      "step": 15473
    },
    {
      "epoch": 2.7058360655737705,
      "grad_norm": 0.6717489858331464,
      "learning_rate": 5.003679557298896e-05,
      "loss": 11.7366,
      "step": 15474
    },
    {
      "epoch": 2.7060109289617484,
      "grad_norm": 0.6717056988679008,
      "learning_rate": 5.002452937974906e-05,
      "loss": 12.029,
      "step": 15475
    },
    {
      "epoch": 2.706185792349727,
      "grad_norm": 0.6278529035260247,
      "learning_rate": 5.001226418867075e-05,
      "loss": 11.6474,
      "step": 15476
    },
    {
      "epoch": 2.706360655737705,
      "grad_norm": 0.6424300112932799,
      "learning_rate": 5.000000000000002e-05,
      "loss": 12.0277,
      "step": 15477
    },
    {
      "epoch": 2.706535519125683,
      "grad_norm": 0.6313642619223182,
      "learning_rate": 4.998773681398278e-05,
      "loss": 11.8323,
      "step": 15478
    },
    {
      "epoch": 2.7067103825136614,
      "grad_norm": 0.6271314375363999,
      "learning_rate": 4.9975474630865015e-05,
      "loss": 11.9673,
      "step": 15479
    },
    {
      "epoch": 2.7068852459016393,
      "grad_norm": 0.6406426562989982,
      "learning_rate": 4.996321345089261e-05,
      "loss": 11.905,
      "step": 15480
    },
    {
      "epoch": 2.7070601092896176,
      "grad_norm": 0.6606069914482464,
      "learning_rate": 4.995095327431137e-05,
      "loss": 11.6345,
      "step": 15481
    },
    {
      "epoch": 2.7072349726775955,
      "grad_norm": 0.6595149589691267,
      "learning_rate": 4.993869410136719e-05,
      "loss": 11.6877,
      "step": 15482
    },
    {
      "epoch": 2.707409836065574,
      "grad_norm": 0.6688805326376619,
      "learning_rate": 4.992643593230587e-05,
      "loss": 11.79,
      "step": 15483
    },
    {
      "epoch": 2.7075846994535517,
      "grad_norm": 0.652526097487597,
      "learning_rate": 4.991417876737332e-05,
      "loss": 11.8429,
      "step": 15484
    },
    {
      "epoch": 2.70775956284153,
      "grad_norm": 0.6788738232906992,
      "learning_rate": 4.990192260681524e-05,
      "loss": 11.6898,
      "step": 15485
    },
    {
      "epoch": 2.7079344262295084,
      "grad_norm": 0.6535405410145425,
      "learning_rate": 4.988966745087743e-05,
      "loss": 11.7345,
      "step": 15486
    },
    {
      "epoch": 2.7081092896174863,
      "grad_norm": 0.6474895837076575,
      "learning_rate": 4.987741329980565e-05,
      "loss": 11.5869,
      "step": 15487
    },
    {
      "epoch": 2.708284153005464,
      "grad_norm": 0.681192765270906,
      "learning_rate": 4.9865160153845593e-05,
      "loss": 11.6795,
      "step": 15488
    },
    {
      "epoch": 2.7084590163934426,
      "grad_norm": 0.6564401194170829,
      "learning_rate": 4.985290801324308e-05,
      "loss": 11.866,
      "step": 15489
    },
    {
      "epoch": 2.708633879781421,
      "grad_norm": 0.6758006013515566,
      "learning_rate": 4.98406568782437e-05,
      "loss": 11.8653,
      "step": 15490
    },
    {
      "epoch": 2.708808743169399,
      "grad_norm": 0.6138746068139418,
      "learning_rate": 4.9828406749093145e-05,
      "loss": 11.6069,
      "step": 15491
    },
    {
      "epoch": 2.708983606557377,
      "grad_norm": 0.7640852777909846,
      "learning_rate": 4.98161576260371e-05,
      "loss": 11.9956,
      "step": 15492
    },
    {
      "epoch": 2.709158469945355,
      "grad_norm": 0.6778812325492963,
      "learning_rate": 4.9803909509321166e-05,
      "loss": 11.7991,
      "step": 15493
    },
    {
      "epoch": 2.7093333333333334,
      "grad_norm": 0.6379807307818894,
      "learning_rate": 4.979166239919098e-05,
      "loss": 11.6635,
      "step": 15494
    },
    {
      "epoch": 2.7095081967213117,
      "grad_norm": 0.7306667889678964,
      "learning_rate": 4.977941629589211e-05,
      "loss": 12.0899,
      "step": 15495
    },
    {
      "epoch": 2.7096830601092896,
      "grad_norm": 0.6403290205706589,
      "learning_rate": 4.976717119967015e-05,
      "loss": 11.5881,
      "step": 15496
    },
    {
      "epoch": 2.7098579234972675,
      "grad_norm": 0.668319452568166,
      "learning_rate": 4.97549271107706e-05,
      "loss": 11.9567,
      "step": 15497
    },
    {
      "epoch": 2.710032786885246,
      "grad_norm": 0.6610635478885478,
      "learning_rate": 4.9742684029439115e-05,
      "loss": 11.7031,
      "step": 15498
    },
    {
      "epoch": 2.710207650273224,
      "grad_norm": 0.6486828209070667,
      "learning_rate": 4.973044195592108e-05,
      "loss": 11.6804,
      "step": 15499
    },
    {
      "epoch": 2.710382513661202,
      "grad_norm": 0.6181595031483935,
      "learning_rate": 4.971820089046204e-05,
      "loss": 11.6473,
      "step": 15500
    },
    {
      "epoch": 2.7105573770491804,
      "grad_norm": 0.6381717160237255,
      "learning_rate": 4.9705960833307455e-05,
      "loss": 11.8215,
      "step": 15501
    },
    {
      "epoch": 2.7107322404371583,
      "grad_norm": 0.7326002307824934,
      "learning_rate": 4.969372178470277e-05,
      "loss": 11.9818,
      "step": 15502
    },
    {
      "epoch": 2.7109071038251367,
      "grad_norm": 0.634708344627088,
      "learning_rate": 4.968148374489344e-05,
      "loss": 12.0302,
      "step": 15503
    },
    {
      "epoch": 2.711081967213115,
      "grad_norm": 0.6099873776036354,
      "learning_rate": 4.966924671412485e-05,
      "loss": 11.8835,
      "step": 15504
    },
    {
      "epoch": 2.711256830601093,
      "grad_norm": 0.6076009761556068,
      "learning_rate": 4.965701069264239e-05,
      "loss": 11.9382,
      "step": 15505
    },
    {
      "epoch": 2.711431693989071,
      "grad_norm": 0.6142029649378036,
      "learning_rate": 4.964477568069146e-05,
      "loss": 11.7122,
      "step": 15506
    },
    {
      "epoch": 2.711606557377049,
      "grad_norm": 0.6680602082408585,
      "learning_rate": 4.963254167851738e-05,
      "loss": 11.9256,
      "step": 15507
    },
    {
      "epoch": 2.7117814207650275,
      "grad_norm": 0.6429979138578618,
      "learning_rate": 4.962030868636548e-05,
      "loss": 11.9054,
      "step": 15508
    },
    {
      "epoch": 2.7119562841530054,
      "grad_norm": 0.6857977222920985,
      "learning_rate": 4.9608076704481075e-05,
      "loss": 12.0166,
      "step": 15509
    },
    {
      "epoch": 2.7121311475409837,
      "grad_norm": 0.598698950375332,
      "learning_rate": 4.9595845733109455e-05,
      "loss": 11.7922,
      "step": 15510
    },
    {
      "epoch": 2.7123060109289616,
      "grad_norm": 0.6411685878462037,
      "learning_rate": 4.95836157724959e-05,
      "loss": 12.0192,
      "step": 15511
    },
    {
      "epoch": 2.71248087431694,
      "grad_norm": 0.6257832731423782,
      "learning_rate": 4.957138682288564e-05,
      "loss": 11.7043,
      "step": 15512
    },
    {
      "epoch": 2.7126557377049183,
      "grad_norm": 0.654080129331742,
      "learning_rate": 4.9559158884523916e-05,
      "loss": 11.7647,
      "step": 15513
    },
    {
      "epoch": 2.712830601092896,
      "grad_norm": 0.5979066245383735,
      "learning_rate": 4.954693195765593e-05,
      "loss": 11.7578,
      "step": 15514
    },
    {
      "epoch": 2.713005464480874,
      "grad_norm": 0.5915982125010617,
      "learning_rate": 4.953470604252687e-05,
      "loss": 11.6002,
      "step": 15515
    },
    {
      "epoch": 2.7131803278688524,
      "grad_norm": 0.673014010096836,
      "learning_rate": 4.9522481139381894e-05,
      "loss": 11.8669,
      "step": 15516
    },
    {
      "epoch": 2.713355191256831,
      "grad_norm": 0.6758175218399309,
      "learning_rate": 4.9510257248466166e-05,
      "loss": 11.8627,
      "step": 15517
    },
    {
      "epoch": 2.7135300546448087,
      "grad_norm": 0.6916679621703267,
      "learning_rate": 4.9498034370024804e-05,
      "loss": 11.7015,
      "step": 15518
    },
    {
      "epoch": 2.713704918032787,
      "grad_norm": 0.6651192696251927,
      "learning_rate": 4.948581250430291e-05,
      "loss": 12.0793,
      "step": 15519
    },
    {
      "epoch": 2.713879781420765,
      "grad_norm": 0.6855895835857284,
      "learning_rate": 4.947359165154557e-05,
      "loss": 11.7447,
      "step": 15520
    },
    {
      "epoch": 2.7140546448087433,
      "grad_norm": 0.6541795703976625,
      "learning_rate": 4.9461371811997856e-05,
      "loss": 12.0179,
      "step": 15521
    },
    {
      "epoch": 2.714229508196721,
      "grad_norm": 0.642797382539765,
      "learning_rate": 4.944915298590482e-05,
      "loss": 11.749,
      "step": 15522
    },
    {
      "epoch": 2.7144043715846995,
      "grad_norm": 0.6597939470370637,
      "learning_rate": 4.943693517351147e-05,
      "loss": 11.8921,
      "step": 15523
    },
    {
      "epoch": 2.7145792349726774,
      "grad_norm": 0.6363431450254992,
      "learning_rate": 4.9424718375062825e-05,
      "loss": 11.9568,
      "step": 15524
    },
    {
      "epoch": 2.7147540983606557,
      "grad_norm": 0.6173560187051551,
      "learning_rate": 4.9412502590803855e-05,
      "loss": 11.8369,
      "step": 15525
    },
    {
      "epoch": 2.714928961748634,
      "grad_norm": 0.6069061782276376,
      "learning_rate": 4.940028782097954e-05,
      "loss": 11.9676,
      "step": 15526
    },
    {
      "epoch": 2.715103825136612,
      "grad_norm": 0.6135513981646109,
      "learning_rate": 4.9388074065834814e-05,
      "loss": 11.5893,
      "step": 15527
    },
    {
      "epoch": 2.7152786885245903,
      "grad_norm": 0.7062821906466596,
      "learning_rate": 4.9375861325614606e-05,
      "loss": 11.8338,
      "step": 15528
    },
    {
      "epoch": 2.715453551912568,
      "grad_norm": 0.6508906742540802,
      "learning_rate": 4.93636496005638e-05,
      "loss": 11.8418,
      "step": 15529
    },
    {
      "epoch": 2.7156284153005465,
      "grad_norm": 0.6595308737818862,
      "learning_rate": 4.935143889092734e-05,
      "loss": 11.9221,
      "step": 15530
    },
    {
      "epoch": 2.7158032786885244,
      "grad_norm": 0.6155604355278546,
      "learning_rate": 4.933922919694996e-05,
      "loss": 11.7285,
      "step": 15531
    },
    {
      "epoch": 2.715978142076503,
      "grad_norm": 0.672320915339658,
      "learning_rate": 4.9327020518876634e-05,
      "loss": 11.6789,
      "step": 15532
    },
    {
      "epoch": 2.7161530054644807,
      "grad_norm": 0.6812544160102656,
      "learning_rate": 4.9314812856952134e-05,
      "loss": 11.871,
      "step": 15533
    },
    {
      "epoch": 2.716327868852459,
      "grad_norm": 0.6724375041479734,
      "learning_rate": 4.9302606211421254e-05,
      "loss": 11.9,
      "step": 15534
    },
    {
      "epoch": 2.7165027322404374,
      "grad_norm": 0.7679821280759612,
      "learning_rate": 4.9290400582528815e-05,
      "loss": 11.7377,
      "step": 15535
    },
    {
      "epoch": 2.7166775956284153,
      "grad_norm": 0.6646015958095789,
      "learning_rate": 4.9278195970519467e-05,
      "loss": 12.0392,
      "step": 15536
    },
    {
      "epoch": 2.7168524590163936,
      "grad_norm": 0.6737080324742319,
      "learning_rate": 4.926599237563807e-05,
      "loss": 11.7382,
      "step": 15537
    },
    {
      "epoch": 2.7170273224043715,
      "grad_norm": 0.6663595748663713,
      "learning_rate": 4.92537897981293e-05,
      "loss": 11.8746,
      "step": 15538
    },
    {
      "epoch": 2.71720218579235,
      "grad_norm": 0.6982588847598828,
      "learning_rate": 4.924158823823789e-05,
      "loss": 11.7803,
      "step": 15539
    },
    {
      "epoch": 2.7173770491803277,
      "grad_norm": 0.6322054297437056,
      "learning_rate": 4.922938769620845e-05,
      "loss": 11.9149,
      "step": 15540
    },
    {
      "epoch": 2.717551912568306,
      "grad_norm": 0.692242229225119,
      "learning_rate": 4.9217188172285635e-05,
      "loss": 12.0233,
      "step": 15541
    },
    {
      "epoch": 2.717726775956284,
      "grad_norm": 0.6470545348955202,
      "learning_rate": 4.920498966671415e-05,
      "loss": 11.4551,
      "step": 15542
    },
    {
      "epoch": 2.7179016393442623,
      "grad_norm": 0.6651998631500752,
      "learning_rate": 4.919279217973859e-05,
      "loss": 11.9105,
      "step": 15543
    },
    {
      "epoch": 2.7180765027322407,
      "grad_norm": 0.660875637397739,
      "learning_rate": 4.918059571160358e-05,
      "loss": 12.0145,
      "step": 15544
    },
    {
      "epoch": 2.7182513661202186,
      "grad_norm": 0.6444023358187517,
      "learning_rate": 4.916840026255359e-05,
      "loss": 11.8802,
      "step": 15545
    },
    {
      "epoch": 2.7184262295081965,
      "grad_norm": 0.6208218545825203,
      "learning_rate": 4.915620583283329e-05,
      "loss": 11.8607,
      "step": 15546
    },
    {
      "epoch": 2.718601092896175,
      "grad_norm": 0.6598819846110021,
      "learning_rate": 4.914401242268717e-05,
      "loss": 11.9437,
      "step": 15547
    },
    {
      "epoch": 2.718775956284153,
      "grad_norm": 0.7003426026355316,
      "learning_rate": 4.913182003235979e-05,
      "loss": 11.8704,
      "step": 15548
    },
    {
      "epoch": 2.718950819672131,
      "grad_norm": 0.6643290854176714,
      "learning_rate": 4.911962866209556e-05,
      "loss": 11.7229,
      "step": 15549
    },
    {
      "epoch": 2.7191256830601094,
      "grad_norm": 0.6469584322507023,
      "learning_rate": 4.910743831213897e-05,
      "loss": 11.9048,
      "step": 15550
    },
    {
      "epoch": 2.7193005464480873,
      "grad_norm": 0.647651566249084,
      "learning_rate": 4.909524898273453e-05,
      "loss": 11.8452,
      "step": 15551
    },
    {
      "epoch": 2.7194754098360656,
      "grad_norm": 0.5903888825951961,
      "learning_rate": 4.908306067412665e-05,
      "loss": 11.3778,
      "step": 15552
    },
    {
      "epoch": 2.719650273224044,
      "grad_norm": 0.6100455125653483,
      "learning_rate": 4.907087338655978e-05,
      "loss": 11.8353,
      "step": 15553
    },
    {
      "epoch": 2.719825136612022,
      "grad_norm": 0.6074493815857415,
      "learning_rate": 4.905868712027822e-05,
      "loss": 11.8875,
      "step": 15554
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.6913728641603644,
      "learning_rate": 4.904650187552637e-05,
      "loss": 11.7849,
      "step": 15555
    },
    {
      "epoch": 2.720174863387978,
      "grad_norm": 0.7711108866515943,
      "learning_rate": 4.9034317652548645e-05,
      "loss": 11.5213,
      "step": 15556
    },
    {
      "epoch": 2.7203497267759564,
      "grad_norm": 0.613507536769338,
      "learning_rate": 4.902213445158936e-05,
      "loss": 11.7816,
      "step": 15557
    },
    {
      "epoch": 2.7205245901639343,
      "grad_norm": 0.6098845063346321,
      "learning_rate": 4.900995227289278e-05,
      "loss": 11.7428,
      "step": 15558
    },
    {
      "epoch": 2.7206994535519127,
      "grad_norm": 0.6237900155494641,
      "learning_rate": 4.8997771116703204e-05,
      "loss": 11.8948,
      "step": 15559
    },
    {
      "epoch": 2.7208743169398906,
      "grad_norm": 0.6300648393356607,
      "learning_rate": 4.898559098326488e-05,
      "loss": 11.8298,
      "step": 15560
    },
    {
      "epoch": 2.721049180327869,
      "grad_norm": 0.6712262821325483,
      "learning_rate": 4.897341187282214e-05,
      "loss": 11.8262,
      "step": 15561
    },
    {
      "epoch": 2.7212240437158473,
      "grad_norm": 0.5932780082469558,
      "learning_rate": 4.896123378561919e-05,
      "loss": 11.6893,
      "step": 15562
    },
    {
      "epoch": 2.721398907103825,
      "grad_norm": 0.5999501846238926,
      "learning_rate": 4.8949056721900175e-05,
      "loss": 11.9581,
      "step": 15563
    },
    {
      "epoch": 2.721573770491803,
      "grad_norm": 0.6064051899980188,
      "learning_rate": 4.893688068190932e-05,
      "loss": 11.7081,
      "step": 15564
    },
    {
      "epoch": 2.7217486338797814,
      "grad_norm": 0.6877442286873116,
      "learning_rate": 4.892470566589077e-05,
      "loss": 11.7168,
      "step": 15565
    },
    {
      "epoch": 2.7219234972677597,
      "grad_norm": 0.6424303576579279,
      "learning_rate": 4.891253167408875e-05,
      "loss": 11.5365,
      "step": 15566
    },
    {
      "epoch": 2.7220983606557376,
      "grad_norm": 0.6211308609577927,
      "learning_rate": 4.89003587067473e-05,
      "loss": 11.958,
      "step": 15567
    },
    {
      "epoch": 2.722273224043716,
      "grad_norm": 0.592622789926363,
      "learning_rate": 4.888818676411054e-05,
      "loss": 11.8354,
      "step": 15568
    },
    {
      "epoch": 2.722448087431694,
      "grad_norm": 0.6758653080689682,
      "learning_rate": 4.8876015846422554e-05,
      "loss": 11.698,
      "step": 15569
    },
    {
      "epoch": 2.722622950819672,
      "grad_norm": 0.6681750335975218,
      "learning_rate": 4.886384595392744e-05,
      "loss": 11.9369,
      "step": 15570
    },
    {
      "epoch": 2.7227978142076505,
      "grad_norm": 0.6731815168737323,
      "learning_rate": 4.8851677086869275e-05,
      "loss": 11.8949,
      "step": 15571
    },
    {
      "epoch": 2.7229726775956284,
      "grad_norm": 0.6585188696712853,
      "learning_rate": 4.8839509245491985e-05,
      "loss": 11.8464,
      "step": 15572
    },
    {
      "epoch": 2.7231475409836063,
      "grad_norm": 0.6359419704528417,
      "learning_rate": 4.8827342430039624e-05,
      "loss": 11.9422,
      "step": 15573
    },
    {
      "epoch": 2.7233224043715847,
      "grad_norm": 0.7093449888352815,
      "learning_rate": 4.881517664075612e-05,
      "loss": 11.9229,
      "step": 15574
    },
    {
      "epoch": 2.723497267759563,
      "grad_norm": 0.6997118254377498,
      "learning_rate": 4.8803011877885564e-05,
      "loss": 11.7009,
      "step": 15575
    },
    {
      "epoch": 2.723672131147541,
      "grad_norm": 0.6908992519446114,
      "learning_rate": 4.8790848141671775e-05,
      "loss": 11.7743,
      "step": 15576
    },
    {
      "epoch": 2.7238469945355193,
      "grad_norm": 0.636122178060083,
      "learning_rate": 4.877868543235872e-05,
      "loss": 11.761,
      "step": 15577
    },
    {
      "epoch": 2.724021857923497,
      "grad_norm": 0.6200335171316237,
      "learning_rate": 4.8766523750190286e-05,
      "loss": 11.9263,
      "step": 15578
    },
    {
      "epoch": 2.7241967213114755,
      "grad_norm": 0.6302937159499264,
      "learning_rate": 4.875436309541033e-05,
      "loss": 11.6731,
      "step": 15579
    },
    {
      "epoch": 2.7243715846994534,
      "grad_norm": 0.6525937636961646,
      "learning_rate": 4.87422034682628e-05,
      "loss": 11.7121,
      "step": 15580
    },
    {
      "epoch": 2.7245464480874317,
      "grad_norm": 0.7064632797037447,
      "learning_rate": 4.873004486899144e-05,
      "loss": 11.8796,
      "step": 15581
    },
    {
      "epoch": 2.7247213114754096,
      "grad_norm": 0.6758301731626666,
      "learning_rate": 4.87178872978401e-05,
      "loss": 11.7407,
      "step": 15582
    },
    {
      "epoch": 2.724896174863388,
      "grad_norm": 0.7073474774633485,
      "learning_rate": 4.870573075505258e-05,
      "loss": 11.8928,
      "step": 15583
    },
    {
      "epoch": 2.7250710382513663,
      "grad_norm": 0.6640108579109198,
      "learning_rate": 4.869357524087266e-05,
      "loss": 11.8444,
      "step": 15584
    },
    {
      "epoch": 2.725245901639344,
      "grad_norm": 0.6644098834122334,
      "learning_rate": 4.868142075554407e-05,
      "loss": 11.9837,
      "step": 15585
    },
    {
      "epoch": 2.7254207650273226,
      "grad_norm": 0.6153344158524381,
      "learning_rate": 4.866926729931057e-05,
      "loss": 11.6599,
      "step": 15586
    },
    {
      "epoch": 2.7255956284153005,
      "grad_norm": 0.6409133676547326,
      "learning_rate": 4.8657114872415865e-05,
      "loss": 11.9615,
      "step": 15587
    },
    {
      "epoch": 2.725770491803279,
      "grad_norm": 0.6975394435780765,
      "learning_rate": 4.864496347510366e-05,
      "loss": 11.89,
      "step": 15588
    },
    {
      "epoch": 2.7259453551912567,
      "grad_norm": 0.6298534599902863,
      "learning_rate": 4.8632813107617605e-05,
      "loss": 11.9728,
      "step": 15589
    },
    {
      "epoch": 2.726120218579235,
      "grad_norm": 0.688217311507654,
      "learning_rate": 4.8620663770201366e-05,
      "loss": 11.7429,
      "step": 15590
    },
    {
      "epoch": 2.726295081967213,
      "grad_norm": 0.6385643129682345,
      "learning_rate": 4.860851546309858e-05,
      "loss": 11.8471,
      "step": 15591
    },
    {
      "epoch": 2.7264699453551913,
      "grad_norm": 0.5847282084338531,
      "learning_rate": 4.8596368186552854e-05,
      "loss": 11.9475,
      "step": 15592
    },
    {
      "epoch": 2.7266448087431696,
      "grad_norm": 0.6481850794108059,
      "learning_rate": 4.8584221940807774e-05,
      "loss": 11.7884,
      "step": 15593
    },
    {
      "epoch": 2.7268196721311475,
      "grad_norm": 0.5820805509618525,
      "learning_rate": 4.857207672610691e-05,
      "loss": 11.5777,
      "step": 15594
    },
    {
      "epoch": 2.7269945355191254,
      "grad_norm": 0.6741063486916774,
      "learning_rate": 4.855993254269381e-05,
      "loss": 11.5201,
      "step": 15595
    },
    {
      "epoch": 2.7271693989071037,
      "grad_norm": 0.6060318614452811,
      "learning_rate": 4.8547789390812e-05,
      "loss": 11.8706,
      "step": 15596
    },
    {
      "epoch": 2.727344262295082,
      "grad_norm": 0.6685245486792447,
      "learning_rate": 4.8535647270705e-05,
      "loss": 12.008,
      "step": 15597
    },
    {
      "epoch": 2.72751912568306,
      "grad_norm": 0.6198355131704194,
      "learning_rate": 4.8523506182616285e-05,
      "loss": 11.8922,
      "step": 15598
    },
    {
      "epoch": 2.7276939890710383,
      "grad_norm": 0.613770951923661,
      "learning_rate": 4.8511366126789324e-05,
      "loss": 11.9758,
      "step": 15599
    },
    {
      "epoch": 2.7278688524590162,
      "grad_norm": 0.5997493671109946,
      "learning_rate": 4.8499227103467574e-05,
      "loss": 11.8974,
      "step": 15600
    },
    {
      "epoch": 2.7280437158469946,
      "grad_norm": 0.5974807685038627,
      "learning_rate": 4.848708911289444e-05,
      "loss": 11.9475,
      "step": 15601
    },
    {
      "epoch": 2.728218579234973,
      "grad_norm": 0.6182644123669085,
      "learning_rate": 4.8474952155313325e-05,
      "loss": 11.9832,
      "step": 15602
    },
    {
      "epoch": 2.728393442622951,
      "grad_norm": 0.6362304699034562,
      "learning_rate": 4.846281623096763e-05,
      "loss": 11.6501,
      "step": 15603
    },
    {
      "epoch": 2.7285683060109287,
      "grad_norm": 0.744723533917843,
      "learning_rate": 4.845068134010071e-05,
      "loss": 11.7922,
      "step": 15604
    },
    {
      "epoch": 2.728743169398907,
      "grad_norm": 0.6322394096763347,
      "learning_rate": 4.843854748295591e-05,
      "loss": 11.7962,
      "step": 15605
    },
    {
      "epoch": 2.7289180327868854,
      "grad_norm": 0.7062730031230808,
      "learning_rate": 4.842641465977654e-05,
      "loss": 11.7915,
      "step": 15606
    },
    {
      "epoch": 2.7290928961748633,
      "grad_norm": 0.5864566153719263,
      "learning_rate": 4.8414282870805915e-05,
      "loss": 11.3619,
      "step": 15607
    },
    {
      "epoch": 2.7292677595628416,
      "grad_norm": 0.600876935858748,
      "learning_rate": 4.84021521162873e-05,
      "loss": 11.8571,
      "step": 15608
    },
    {
      "epoch": 2.7294426229508195,
      "grad_norm": 0.6197011048727795,
      "learning_rate": 4.8390022396463965e-05,
      "loss": 11.6268,
      "step": 15609
    },
    {
      "epoch": 2.729617486338798,
      "grad_norm": 0.6845018194975725,
      "learning_rate": 4.837789371157915e-05,
      "loss": 11.7455,
      "step": 15610
    },
    {
      "epoch": 2.729792349726776,
      "grad_norm": 0.6738249067834581,
      "learning_rate": 4.836576606187606e-05,
      "loss": 11.7841,
      "step": 15611
    },
    {
      "epoch": 2.729967213114754,
      "grad_norm": 0.6867126072684469,
      "learning_rate": 4.8353639447597895e-05,
      "loss": 11.8287,
      "step": 15612
    },
    {
      "epoch": 2.730142076502732,
      "grad_norm": 0.7621911719397577,
      "learning_rate": 4.8341513868987845e-05,
      "loss": 11.94,
      "step": 15613
    },
    {
      "epoch": 2.7303169398907103,
      "grad_norm": 0.6308124592334162,
      "learning_rate": 4.832938932628904e-05,
      "loss": 11.6882,
      "step": 15614
    },
    {
      "epoch": 2.7304918032786887,
      "grad_norm": 0.6579164409505212,
      "learning_rate": 4.831726581974464e-05,
      "loss": 11.7906,
      "step": 15615
    },
    {
      "epoch": 2.7306666666666666,
      "grad_norm": 0.6589552320344052,
      "learning_rate": 4.830514334959775e-05,
      "loss": 11.8247,
      "step": 15616
    },
    {
      "epoch": 2.730841530054645,
      "grad_norm": 0.7463190538926914,
      "learning_rate": 4.829302191609145e-05,
      "loss": 11.8701,
      "step": 15617
    },
    {
      "epoch": 2.731016393442623,
      "grad_norm": 0.6595610796127427,
      "learning_rate": 4.828090151946882e-05,
      "loss": 12.0061,
      "step": 15618
    },
    {
      "epoch": 2.731191256830601,
      "grad_norm": 0.646282344159238,
      "learning_rate": 4.8268782159972924e-05,
      "loss": 11.7172,
      "step": 15619
    },
    {
      "epoch": 2.7313661202185795,
      "grad_norm": 0.7031222606977385,
      "learning_rate": 4.8256663837846774e-05,
      "loss": 11.9096,
      "step": 15620
    },
    {
      "epoch": 2.7315409836065574,
      "grad_norm": 0.6936640075850736,
      "learning_rate": 4.8244546553333395e-05,
      "loss": 11.8492,
      "step": 15621
    },
    {
      "epoch": 2.7317158469945353,
      "grad_norm": 0.6633152818910946,
      "learning_rate": 4.823243030667576e-05,
      "loss": 11.692,
      "step": 15622
    },
    {
      "epoch": 2.7318907103825136,
      "grad_norm": 0.6872264446460155,
      "learning_rate": 4.8220315098116845e-05,
      "loss": 12.0486,
      "step": 15623
    },
    {
      "epoch": 2.732065573770492,
      "grad_norm": 0.7035690760901654,
      "learning_rate": 4.820820092789959e-05,
      "loss": 11.9669,
      "step": 15624
    },
    {
      "epoch": 2.73224043715847,
      "grad_norm": 0.6461628636093555,
      "learning_rate": 4.8196087796266944e-05,
      "loss": 11.8072,
      "step": 15625
    },
    {
      "epoch": 2.732415300546448,
      "grad_norm": 0.7145124217704797,
      "learning_rate": 4.818397570346183e-05,
      "loss": 11.8409,
      "step": 15626
    },
    {
      "epoch": 2.732590163934426,
      "grad_norm": 0.763994965261242,
      "learning_rate": 4.817186464972702e-05,
      "loss": 11.7034,
      "step": 15627
    },
    {
      "epoch": 2.7327650273224044,
      "grad_norm": 0.6537377479548045,
      "learning_rate": 4.815975463530549e-05,
      "loss": 11.6318,
      "step": 15628
    },
    {
      "epoch": 2.732939890710383,
      "grad_norm": 0.7064580764890348,
      "learning_rate": 4.8147645660440055e-05,
      "loss": 11.8644,
      "step": 15629
    },
    {
      "epoch": 2.7331147540983607,
      "grad_norm": 0.6694621792072438,
      "learning_rate": 4.813553772537354e-05,
      "loss": 11.8501,
      "step": 15630
    },
    {
      "epoch": 2.7332896174863386,
      "grad_norm": 0.7326055608981884,
      "learning_rate": 4.812343083034876e-05,
      "loss": 12.0314,
      "step": 15631
    },
    {
      "epoch": 2.733464480874317,
      "grad_norm": 0.6606864438428337,
      "learning_rate": 4.811132497560841e-05,
      "loss": 11.8429,
      "step": 15632
    },
    {
      "epoch": 2.7336393442622953,
      "grad_norm": 0.6354187209040731,
      "learning_rate": 4.8099220161395355e-05,
      "loss": 11.5439,
      "step": 15633
    },
    {
      "epoch": 2.733814207650273,
      "grad_norm": 0.6053975705135142,
      "learning_rate": 4.808711638795228e-05,
      "loss": 11.73,
      "step": 15634
    },
    {
      "epoch": 2.7339890710382515,
      "grad_norm": 0.7124205149981098,
      "learning_rate": 4.807501365552196e-05,
      "loss": 11.6061,
      "step": 15635
    },
    {
      "epoch": 2.7341639344262294,
      "grad_norm": 0.7222337776021666,
      "learning_rate": 4.8062911964347004e-05,
      "loss": 11.8097,
      "step": 15636
    },
    {
      "epoch": 2.7343387978142077,
      "grad_norm": 0.5899512824297327,
      "learning_rate": 4.80508113146701e-05,
      "loss": 11.4849,
      "step": 15637
    },
    {
      "epoch": 2.7345136612021856,
      "grad_norm": 0.6774842667214453,
      "learning_rate": 4.803871170673396e-05,
      "loss": 11.7855,
      "step": 15638
    },
    {
      "epoch": 2.734688524590164,
      "grad_norm": 0.6189495552558599,
      "learning_rate": 4.802661314078121e-05,
      "loss": 11.7714,
      "step": 15639
    },
    {
      "epoch": 2.734863387978142,
      "grad_norm": 0.5931974470503638,
      "learning_rate": 4.801451561705447e-05,
      "loss": 11.8499,
      "step": 15640
    },
    {
      "epoch": 2.73503825136612,
      "grad_norm": 0.6644962557053192,
      "learning_rate": 4.800241913579624e-05,
      "loss": 11.7796,
      "step": 15641
    },
    {
      "epoch": 2.7352131147540986,
      "grad_norm": 0.6466533250485066,
      "learning_rate": 4.7990323697249194e-05,
      "loss": 11.9507,
      "step": 15642
    },
    {
      "epoch": 2.7353879781420765,
      "grad_norm": 0.683041047146604,
      "learning_rate": 4.797822930165584e-05,
      "loss": 11.9869,
      "step": 15643
    },
    {
      "epoch": 2.735562841530055,
      "grad_norm": 0.7492567301528095,
      "learning_rate": 4.7966135949258776e-05,
      "loss": 11.8646,
      "step": 15644
    },
    {
      "epoch": 2.7357377049180327,
      "grad_norm": 0.6044348409841231,
      "learning_rate": 4.7954043640300394e-05,
      "loss": 11.6292,
      "step": 15645
    },
    {
      "epoch": 2.735912568306011,
      "grad_norm": 0.6483473881878588,
      "learning_rate": 4.7941952375023205e-05,
      "loss": 11.9302,
      "step": 15646
    },
    {
      "epoch": 2.736087431693989,
      "grad_norm": 0.6699328132530702,
      "learning_rate": 4.792986215366976e-05,
      "loss": 11.7827,
      "step": 15647
    },
    {
      "epoch": 2.7362622950819673,
      "grad_norm": 0.6396742586211563,
      "learning_rate": 4.791777297648243e-05,
      "loss": 11.63,
      "step": 15648
    },
    {
      "epoch": 2.736437158469945,
      "grad_norm": 0.6263245061976273,
      "learning_rate": 4.790568484370372e-05,
      "loss": 11.6394,
      "step": 15649
    },
    {
      "epoch": 2.7366120218579235,
      "grad_norm": 0.6014700847899922,
      "learning_rate": 4.789359775557595e-05,
      "loss": 11.9312,
      "step": 15650
    },
    {
      "epoch": 2.736786885245902,
      "grad_norm": 0.6341904002301796,
      "learning_rate": 4.7881511712341484e-05,
      "loss": 11.8673,
      "step": 15651
    },
    {
      "epoch": 2.7369617486338798,
      "grad_norm": 0.6423704629029998,
      "learning_rate": 4.786942671424277e-05,
      "loss": 11.6086,
      "step": 15652
    },
    {
      "epoch": 2.7371366120218577,
      "grad_norm": 0.6352286256674475,
      "learning_rate": 4.785734276152215e-05,
      "loss": 11.8651,
      "step": 15653
    },
    {
      "epoch": 2.737311475409836,
      "grad_norm": 0.6348395327865766,
      "learning_rate": 4.7845259854421875e-05,
      "loss": 11.9594,
      "step": 15654
    },
    {
      "epoch": 2.7374863387978143,
      "grad_norm": 0.7358406696943259,
      "learning_rate": 4.783317799318427e-05,
      "loss": 11.793,
      "step": 15655
    },
    {
      "epoch": 2.7376612021857922,
      "grad_norm": 0.675977651473555,
      "learning_rate": 4.782109717805159e-05,
      "loss": 12.0195,
      "step": 15656
    },
    {
      "epoch": 2.7378360655737706,
      "grad_norm": 0.6385344161374613,
      "learning_rate": 4.780901740926614e-05,
      "loss": 11.9178,
      "step": 15657
    },
    {
      "epoch": 2.7380109289617485,
      "grad_norm": 0.633085119274037,
      "learning_rate": 4.779693868707019e-05,
      "loss": 11.8983,
      "step": 15658
    },
    {
      "epoch": 2.738185792349727,
      "grad_norm": 0.6759003255430811,
      "learning_rate": 4.778486101170587e-05,
      "loss": 12.0391,
      "step": 15659
    },
    {
      "epoch": 2.738360655737705,
      "grad_norm": 0.6510402585752402,
      "learning_rate": 4.77727843834154e-05,
      "loss": 11.9337,
      "step": 15660
    },
    {
      "epoch": 2.738535519125683,
      "grad_norm": 0.5869057986433612,
      "learning_rate": 4.776070880244093e-05,
      "loss": 11.5811,
      "step": 15661
    },
    {
      "epoch": 2.738710382513661,
      "grad_norm": 0.6213731083254908,
      "learning_rate": 4.774863426902472e-05,
      "loss": 11.9792,
      "step": 15662
    },
    {
      "epoch": 2.7388852459016393,
      "grad_norm": 0.6030613045446408,
      "learning_rate": 4.773656078340879e-05,
      "loss": 11.9034,
      "step": 15663
    },
    {
      "epoch": 2.7390601092896176,
      "grad_norm": 0.6303919317635542,
      "learning_rate": 4.772448834583527e-05,
      "loss": 11.79,
      "step": 15664
    },
    {
      "epoch": 2.7392349726775955,
      "grad_norm": 0.643086359402343,
      "learning_rate": 4.771241695654625e-05,
      "loss": 11.6298,
      "step": 15665
    },
    {
      "epoch": 2.739409836065574,
      "grad_norm": 0.6373399609257314,
      "learning_rate": 4.770034661578383e-05,
      "loss": 11.9499,
      "step": 15666
    },
    {
      "epoch": 2.7395846994535518,
      "grad_norm": 0.6389579937467336,
      "learning_rate": 4.76882773237901e-05,
      "loss": 12.0231,
      "step": 15667
    },
    {
      "epoch": 2.73975956284153,
      "grad_norm": 0.6656674550695586,
      "learning_rate": 4.767620908080698e-05,
      "loss": 11.7556,
      "step": 15668
    },
    {
      "epoch": 2.7399344262295084,
      "grad_norm": 0.6497513901920124,
      "learning_rate": 4.766414188707652e-05,
      "loss": 11.9263,
      "step": 15669
    },
    {
      "epoch": 2.7401092896174863,
      "grad_norm": 0.5972872772119833,
      "learning_rate": 4.7652075742840674e-05,
      "loss": 11.9717,
      "step": 15670
    },
    {
      "epoch": 2.7402841530054642,
      "grad_norm": 0.6847693935963028,
      "learning_rate": 4.7640010648341514e-05,
      "loss": 12.0181,
      "step": 15671
    },
    {
      "epoch": 2.7404590163934426,
      "grad_norm": 0.6285490828811645,
      "learning_rate": 4.762794660382086e-05,
      "loss": 11.8349,
      "step": 15672
    },
    {
      "epoch": 2.740633879781421,
      "grad_norm": 0.6600375550359421,
      "learning_rate": 4.7615883609520693e-05,
      "loss": 11.5403,
      "step": 15673
    },
    {
      "epoch": 2.740808743169399,
      "grad_norm": 0.6006696325948322,
      "learning_rate": 4.760382166568289e-05,
      "loss": 11.918,
      "step": 15674
    },
    {
      "epoch": 2.740983606557377,
      "grad_norm": 0.7038629897569286,
      "learning_rate": 4.759176077254931e-05,
      "loss": 11.7211,
      "step": 15675
    },
    {
      "epoch": 2.741158469945355,
      "grad_norm": 0.7062832732273837,
      "learning_rate": 4.7579700930361915e-05,
      "loss": 11.8655,
      "step": 15676
    },
    {
      "epoch": 2.7413333333333334,
      "grad_norm": 0.7218221108304069,
      "learning_rate": 4.756764213936242e-05,
      "loss": 12.1389,
      "step": 15677
    },
    {
      "epoch": 2.7415081967213117,
      "grad_norm": 0.6790080985638289,
      "learning_rate": 4.7555584399792706e-05,
      "loss": 11.8931,
      "step": 15678
    },
    {
      "epoch": 2.7416830601092896,
      "grad_norm": 0.6406352020842322,
      "learning_rate": 4.754352771189454e-05,
      "loss": 11.9577,
      "step": 15679
    },
    {
      "epoch": 2.7418579234972675,
      "grad_norm": 0.72882249560402,
      "learning_rate": 4.753147207590971e-05,
      "loss": 11.7874,
      "step": 15680
    },
    {
      "epoch": 2.742032786885246,
      "grad_norm": 0.7706256331521248,
      "learning_rate": 4.751941749207995e-05,
      "loss": 11.6708,
      "step": 15681
    },
    {
      "epoch": 2.742207650273224,
      "grad_norm": 0.6920621185882717,
      "learning_rate": 4.7507363960647026e-05,
      "loss": 11.764,
      "step": 15682
    },
    {
      "epoch": 2.742382513661202,
      "grad_norm": 0.6653787147367539,
      "learning_rate": 4.7495311481852625e-05,
      "loss": 11.9851,
      "step": 15683
    },
    {
      "epoch": 2.7425573770491805,
      "grad_norm": 0.7057901389512843,
      "learning_rate": 4.7483260055938426e-05,
      "loss": 11.8434,
      "step": 15684
    },
    {
      "epoch": 2.7427322404371584,
      "grad_norm": 0.7243925082519103,
      "learning_rate": 4.7471209683146124e-05,
      "loss": 11.9694,
      "step": 15685
    },
    {
      "epoch": 2.7429071038251367,
      "grad_norm": 0.6898525278467954,
      "learning_rate": 4.745916036371735e-05,
      "loss": 11.942,
      "step": 15686
    },
    {
      "epoch": 2.7430819672131146,
      "grad_norm": 0.6117127170577626,
      "learning_rate": 4.744711209789372e-05,
      "loss": 11.9297,
      "step": 15687
    },
    {
      "epoch": 2.743256830601093,
      "grad_norm": 0.6923133607559432,
      "learning_rate": 4.743506488591686e-05,
      "loss": 12.0261,
      "step": 15688
    },
    {
      "epoch": 2.743431693989071,
      "grad_norm": 0.6292432724340862,
      "learning_rate": 4.742301872802835e-05,
      "loss": 11.7727,
      "step": 15689
    },
    {
      "epoch": 2.743606557377049,
      "grad_norm": 0.6605277785885043,
      "learning_rate": 4.741097362446973e-05,
      "loss": 11.6849,
      "step": 15690
    },
    {
      "epoch": 2.7437814207650275,
      "grad_norm": 0.6569813934123953,
      "learning_rate": 4.739892957548258e-05,
      "loss": 11.7335,
      "step": 15691
    },
    {
      "epoch": 2.7439562841530054,
      "grad_norm": 0.6228664771203468,
      "learning_rate": 4.7386886581308385e-05,
      "loss": 11.8281,
      "step": 15692
    },
    {
      "epoch": 2.7441311475409838,
      "grad_norm": 0.7465813471109697,
      "learning_rate": 4.737484464218867e-05,
      "loss": 11.6713,
      "step": 15693
    },
    {
      "epoch": 2.7443060109289616,
      "grad_norm": 0.699553994569845,
      "learning_rate": 4.736280375836488e-05,
      "loss": 11.4729,
      "step": 15694
    },
    {
      "epoch": 2.74448087431694,
      "grad_norm": 0.6197558116466396,
      "learning_rate": 4.735076393007851e-05,
      "loss": 11.8654,
      "step": 15695
    },
    {
      "epoch": 2.744655737704918,
      "grad_norm": 0.6323497345022175,
      "learning_rate": 4.7338725157570974e-05,
      "loss": 11.8814,
      "step": 15696
    },
    {
      "epoch": 2.7448306010928962,
      "grad_norm": 0.6470076231098978,
      "learning_rate": 4.7326687441083686e-05,
      "loss": 11.8329,
      "step": 15697
    },
    {
      "epoch": 2.745005464480874,
      "grad_norm": 0.65118862894982,
      "learning_rate": 4.731465078085804e-05,
      "loss": 11.8989,
      "step": 15698
    },
    {
      "epoch": 2.7451803278688525,
      "grad_norm": 0.6970201120935304,
      "learning_rate": 4.730261517713541e-05,
      "loss": 11.8321,
      "step": 15699
    },
    {
      "epoch": 2.745355191256831,
      "grad_norm": 0.6562779595623885,
      "learning_rate": 4.729058063015716e-05,
      "loss": 11.8061,
      "step": 15700
    },
    {
      "epoch": 2.7455300546448087,
      "grad_norm": 0.6105142643038655,
      "learning_rate": 4.72785471401646e-05,
      "loss": 11.8867,
      "step": 15701
    },
    {
      "epoch": 2.745704918032787,
      "grad_norm": 0.6498431738833017,
      "learning_rate": 4.726651470739905e-05,
      "loss": 11.8366,
      "step": 15702
    },
    {
      "epoch": 2.745879781420765,
      "grad_norm": 0.6558909861244869,
      "learning_rate": 4.7254483332101784e-05,
      "loss": 11.7209,
      "step": 15703
    },
    {
      "epoch": 2.7460546448087433,
      "grad_norm": 0.6319318665897498,
      "learning_rate": 4.7242453014514084e-05,
      "loss": 12.0487,
      "step": 15704
    },
    {
      "epoch": 2.746229508196721,
      "grad_norm": 0.6293484472234834,
      "learning_rate": 4.723042375487718e-05,
      "loss": 11.7414,
      "step": 15705
    },
    {
      "epoch": 2.7464043715846995,
      "grad_norm": 0.6276019125784281,
      "learning_rate": 4.7218395553432315e-05,
      "loss": 11.4583,
      "step": 15706
    },
    {
      "epoch": 2.7465792349726774,
      "grad_norm": 0.6469127389240645,
      "learning_rate": 4.7206368410420675e-05,
      "loss": 11.8473,
      "step": 15707
    },
    {
      "epoch": 2.7467540983606558,
      "grad_norm": 0.7120475813385422,
      "learning_rate": 4.719434232608345e-05,
      "loss": 11.631,
      "step": 15708
    },
    {
      "epoch": 2.746928961748634,
      "grad_norm": 0.6301305226771063,
      "learning_rate": 4.7182317300661796e-05,
      "loss": 11.586,
      "step": 15709
    },
    {
      "epoch": 2.747103825136612,
      "grad_norm": 0.623104585484498,
      "learning_rate": 4.717029333439684e-05,
      "loss": 11.3937,
      "step": 15710
    },
    {
      "epoch": 2.74727868852459,
      "grad_norm": 0.5793617812072561,
      "learning_rate": 4.7158270427529726e-05,
      "loss": 11.6553,
      "step": 15711
    },
    {
      "epoch": 2.7474535519125682,
      "grad_norm": 0.6332434212754648,
      "learning_rate": 4.714624858030153e-05,
      "loss": 11.6973,
      "step": 15712
    },
    {
      "epoch": 2.7476284153005466,
      "grad_norm": 0.6439227176177854,
      "learning_rate": 4.7134227792953337e-05,
      "loss": 11.9048,
      "step": 15713
    },
    {
      "epoch": 2.7478032786885245,
      "grad_norm": 0.660505083461518,
      "learning_rate": 4.712220806572618e-05,
      "loss": 12.0202,
      "step": 15714
    },
    {
      "epoch": 2.747978142076503,
      "grad_norm": 0.6559304853518556,
      "learning_rate": 4.711018939886112e-05,
      "loss": 11.6959,
      "step": 15715
    },
    {
      "epoch": 2.7481530054644807,
      "grad_norm": 0.5898262079040222,
      "learning_rate": 4.709817179259916e-05,
      "loss": 11.7253,
      "step": 15716
    },
    {
      "epoch": 2.748327868852459,
      "grad_norm": 0.7024757591337556,
      "learning_rate": 4.708615524718128e-05,
      "loss": 11.8892,
      "step": 15717
    },
    {
      "epoch": 2.7485027322404374,
      "grad_norm": 0.7108997453338173,
      "learning_rate": 4.707413976284845e-05,
      "loss": 11.5399,
      "step": 15718
    },
    {
      "epoch": 2.7486775956284153,
      "grad_norm": 0.753382251315731,
      "learning_rate": 4.7062125339841626e-05,
      "loss": 11.9035,
      "step": 15719
    },
    {
      "epoch": 2.748852459016393,
      "grad_norm": 0.5888930016138209,
      "learning_rate": 4.705011197840172e-05,
      "loss": 11.5951,
      "step": 15720
    },
    {
      "epoch": 2.7490273224043715,
      "grad_norm": 0.6394074330443686,
      "learning_rate": 4.703809967876965e-05,
      "loss": 11.7391,
      "step": 15721
    },
    {
      "epoch": 2.74920218579235,
      "grad_norm": 0.6973243893273143,
      "learning_rate": 4.702608844118635e-05,
      "loss": 11.6717,
      "step": 15722
    },
    {
      "epoch": 2.7493770491803278,
      "grad_norm": 0.648522877829995,
      "learning_rate": 4.7014078265892533e-05,
      "loss": 11.6933,
      "step": 15723
    },
    {
      "epoch": 2.749551912568306,
      "grad_norm": 0.622140160091988,
      "learning_rate": 4.700206915312918e-05,
      "loss": 11.539,
      "step": 15724
    },
    {
      "epoch": 2.749726775956284,
      "grad_norm": 0.6671111939262085,
      "learning_rate": 4.699006110313706e-05,
      "loss": 11.8239,
      "step": 15725
    },
    {
      "epoch": 2.7499016393442623,
      "grad_norm": 0.7383860202485154,
      "learning_rate": 4.6978054116156987e-05,
      "loss": 11.6161,
      "step": 15726
    },
    {
      "epoch": 2.7500765027322407,
      "grad_norm": 0.6889697659054885,
      "learning_rate": 4.696604819242977e-05,
      "loss": 11.6773,
      "step": 15727
    },
    {
      "epoch": 2.7502513661202186,
      "grad_norm": 0.6877301095592364,
      "learning_rate": 4.695404333219603e-05,
      "loss": 11.8384,
      "step": 15728
    },
    {
      "epoch": 2.7504262295081965,
      "grad_norm": 0.6920412849184038,
      "learning_rate": 4.6942039535696635e-05,
      "loss": 12.0215,
      "step": 15729
    },
    {
      "epoch": 2.750601092896175,
      "grad_norm": 0.640186181704944,
      "learning_rate": 4.6930036803172265e-05,
      "loss": 11.8036,
      "step": 15730
    },
    {
      "epoch": 2.750775956284153,
      "grad_norm": 0.6516524817665823,
      "learning_rate": 4.691803513486363e-05,
      "loss": 11.9437,
      "step": 15731
    },
    {
      "epoch": 2.750950819672131,
      "grad_norm": 0.6746339997185623,
      "learning_rate": 4.6906034531011346e-05,
      "loss": 11.7245,
      "step": 15732
    },
    {
      "epoch": 2.7511256830601094,
      "grad_norm": 0.5948772907071604,
      "learning_rate": 4.6894034991856054e-05,
      "loss": 11.8841,
      "step": 15733
    },
    {
      "epoch": 2.7513005464480873,
      "grad_norm": 0.6078638324446444,
      "learning_rate": 4.688203651763845e-05,
      "loss": 11.791,
      "step": 15734
    },
    {
      "epoch": 2.7514754098360656,
      "grad_norm": 0.6133589675091251,
      "learning_rate": 4.687003910859911e-05,
      "loss": 11.9082,
      "step": 15735
    },
    {
      "epoch": 2.751650273224044,
      "grad_norm": 0.6050939464768189,
      "learning_rate": 4.685804276497865e-05,
      "loss": 11.7036,
      "step": 15736
    },
    {
      "epoch": 2.751825136612022,
      "grad_norm": 0.585581612336681,
      "learning_rate": 4.684604748701752e-05,
      "loss": 11.5865,
      "step": 15737
    },
    {
      "epoch": 2.752,
      "grad_norm": 0.6484167811374819,
      "learning_rate": 4.683405327495638e-05,
      "loss": 12.0281,
      "step": 15738
    },
    {
      "epoch": 2.752174863387978,
      "grad_norm": 0.6681739491198521,
      "learning_rate": 4.6822060129035716e-05,
      "loss": 11.6736,
      "step": 15739
    },
    {
      "epoch": 2.7523497267759565,
      "grad_norm": 0.6587445918709384,
      "learning_rate": 4.681006804949607e-05,
      "loss": 11.777,
      "step": 15740
    },
    {
      "epoch": 2.7525245901639344,
      "grad_norm": 0.6227995354124625,
      "learning_rate": 4.679807703657782e-05,
      "loss": 11.8028,
      "step": 15741
    },
    {
      "epoch": 2.7526994535519127,
      "grad_norm": 0.6704167222157164,
      "learning_rate": 4.6786087090521456e-05,
      "loss": 11.7929,
      "step": 15742
    },
    {
      "epoch": 2.7528743169398906,
      "grad_norm": 0.651490752628738,
      "learning_rate": 4.677409821156746e-05,
      "loss": 12.0475,
      "step": 15743
    },
    {
      "epoch": 2.753049180327869,
      "grad_norm": 0.5584732154869185,
      "learning_rate": 4.676211039995623e-05,
      "loss": 11.5188,
      "step": 15744
    },
    {
      "epoch": 2.753224043715847,
      "grad_norm": 0.7619759944914951,
      "learning_rate": 4.675012365592818e-05,
      "loss": 11.5933,
      "step": 15745
    },
    {
      "epoch": 2.753398907103825,
      "grad_norm": 0.694061410359469,
      "learning_rate": 4.673813797972362e-05,
      "loss": 11.8231,
      "step": 15746
    },
    {
      "epoch": 2.753573770491803,
      "grad_norm": 0.6698802188316909,
      "learning_rate": 4.672615337158287e-05,
      "loss": 11.9035,
      "step": 15747
    },
    {
      "epoch": 2.7537486338797814,
      "grad_norm": 0.6637484377525635,
      "learning_rate": 4.671416983174637e-05,
      "loss": 11.2106,
      "step": 15748
    },
    {
      "epoch": 2.7539234972677598,
      "grad_norm": 0.6919858144512413,
      "learning_rate": 4.670218736045441e-05,
      "loss": 11.7473,
      "step": 15749
    },
    {
      "epoch": 2.7540983606557377,
      "grad_norm": 0.7217794865099403,
      "learning_rate": 4.669020595794721e-05,
      "loss": 11.7399,
      "step": 15750
    },
    {
      "epoch": 2.754273224043716,
      "grad_norm": 0.6327597135555578,
      "learning_rate": 4.6678225624465043e-05,
      "loss": 11.7483,
      "step": 15751
    },
    {
      "epoch": 2.754448087431694,
      "grad_norm": 0.6566383928122695,
      "learning_rate": 4.6666246360248144e-05,
      "loss": 11.9149,
      "step": 15752
    },
    {
      "epoch": 2.7546229508196722,
      "grad_norm": 0.6521832983286447,
      "learning_rate": 4.6654268165536805e-05,
      "loss": 11.8745,
      "step": 15753
    },
    {
      "epoch": 2.75479781420765,
      "grad_norm": 0.6919894975541258,
      "learning_rate": 4.664229104057122e-05,
      "loss": 11.8776,
      "step": 15754
    },
    {
      "epoch": 2.7549726775956285,
      "grad_norm": 0.6590126629542437,
      "learning_rate": 4.663031498559148e-05,
      "loss": 11.9009,
      "step": 15755
    },
    {
      "epoch": 2.7551475409836064,
      "grad_norm": 0.6096170895183661,
      "learning_rate": 4.66183400008378e-05,
      "loss": 11.7758,
      "step": 15756
    },
    {
      "epoch": 2.7553224043715847,
      "grad_norm": 0.6759583123953578,
      "learning_rate": 4.660636608655027e-05,
      "loss": 11.996,
      "step": 15757
    },
    {
      "epoch": 2.755497267759563,
      "grad_norm": 0.656362694330923,
      "learning_rate": 4.65943932429691e-05,
      "loss": 11.9408,
      "step": 15758
    },
    {
      "epoch": 2.755672131147541,
      "grad_norm": 0.7029712525349515,
      "learning_rate": 4.65824214703343e-05,
      "loss": 11.7431,
      "step": 15759
    },
    {
      "epoch": 2.7558469945355193,
      "grad_norm": 0.7077179364659959,
      "learning_rate": 4.6570450768885966e-05,
      "loss": 11.9124,
      "step": 15760
    },
    {
      "epoch": 2.756021857923497,
      "grad_norm": 0.5989738834774018,
      "learning_rate": 4.65584811388641e-05,
      "loss": 11.7543,
      "step": 15761
    },
    {
      "epoch": 2.7561967213114755,
      "grad_norm": 0.6297288141894696,
      "learning_rate": 4.6546512580508804e-05,
      "loss": 11.7293,
      "step": 15762
    },
    {
      "epoch": 2.7563715846994534,
      "grad_norm": 0.6516009772520966,
      "learning_rate": 4.65345450940601e-05,
      "loss": 11.941,
      "step": 15763
    },
    {
      "epoch": 2.7565464480874318,
      "grad_norm": 0.7051437639537159,
      "learning_rate": 4.65225786797579e-05,
      "loss": 11.61,
      "step": 15764
    },
    {
      "epoch": 2.7567213114754097,
      "grad_norm": 0.6996759973251963,
      "learning_rate": 4.651061333784218e-05,
      "loss": 11.7417,
      "step": 15765
    },
    {
      "epoch": 2.756896174863388,
      "grad_norm": 0.6577960398608742,
      "learning_rate": 4.649864906855286e-05,
      "loss": 11.7186,
      "step": 15766
    },
    {
      "epoch": 2.7570710382513663,
      "grad_norm": 0.6321766172337534,
      "learning_rate": 4.648668587212997e-05,
      "loss": 11.9027,
      "step": 15767
    },
    {
      "epoch": 2.7572459016393442,
      "grad_norm": 0.7078177891215941,
      "learning_rate": 4.647472374881331e-05,
      "loss": 11.9209,
      "step": 15768
    },
    {
      "epoch": 2.757420765027322,
      "grad_norm": 0.6731196014361626,
      "learning_rate": 4.646276269884278e-05,
      "loss": 11.8218,
      "step": 15769
    },
    {
      "epoch": 2.7575956284153005,
      "grad_norm": 0.6422907373180746,
      "learning_rate": 4.645080272245823e-05,
      "loss": 11.7954,
      "step": 15770
    },
    {
      "epoch": 2.757770491803279,
      "grad_norm": 0.8093540942868682,
      "learning_rate": 4.643884381989947e-05,
      "loss": 11.8229,
      "step": 15771
    },
    {
      "epoch": 2.7579453551912567,
      "grad_norm": 0.730231654577006,
      "learning_rate": 4.642688599140642e-05,
      "loss": 11.6098,
      "step": 15772
    },
    {
      "epoch": 2.758120218579235,
      "grad_norm": 0.6012112141670769,
      "learning_rate": 4.641492923721876e-05,
      "loss": 11.9653,
      "step": 15773
    },
    {
      "epoch": 2.758295081967213,
      "grad_norm": 0.7464179436611452,
      "learning_rate": 4.64029735575763e-05,
      "loss": 11.9293,
      "step": 15774
    },
    {
      "epoch": 2.7584699453551913,
      "grad_norm": 0.7424845643650816,
      "learning_rate": 4.639101895271879e-05,
      "loss": 11.7668,
      "step": 15775
    },
    {
      "epoch": 2.7586448087431696,
      "grad_norm": 0.7409911036551843,
      "learning_rate": 4.637906542288594e-05,
      "loss": 11.8382,
      "step": 15776
    },
    {
      "epoch": 2.7588196721311475,
      "grad_norm": 0.6065092091211224,
      "learning_rate": 4.636711296831747e-05,
      "loss": 11.9242,
      "step": 15777
    },
    {
      "epoch": 2.7589945355191254,
      "grad_norm": 0.6318735263966624,
      "learning_rate": 4.635516158925306e-05,
      "loss": 11.7271,
      "step": 15778
    },
    {
      "epoch": 2.7591693989071038,
      "grad_norm": 0.6667444552553057,
      "learning_rate": 4.634321128593237e-05,
      "loss": 11.973,
      "step": 15779
    },
    {
      "epoch": 2.759344262295082,
      "grad_norm": 0.6576452496890551,
      "learning_rate": 4.633126205859504e-05,
      "loss": 11.5565,
      "step": 15780
    },
    {
      "epoch": 2.75951912568306,
      "grad_norm": 0.6274596853128369,
      "learning_rate": 4.63193139074807e-05,
      "loss": 11.6179,
      "step": 15781
    },
    {
      "epoch": 2.7596939890710384,
      "grad_norm": 0.6596202720174538,
      "learning_rate": 4.6307366832828927e-05,
      "loss": 11.8812,
      "step": 15782
    },
    {
      "epoch": 2.7598688524590163,
      "grad_norm": 0.658302657775301,
      "learning_rate": 4.629542083487931e-05,
      "loss": 11.6565,
      "step": 15783
    },
    {
      "epoch": 2.7600437158469946,
      "grad_norm": 0.6129169604862454,
      "learning_rate": 4.62834759138714e-05,
      "loss": 11.8162,
      "step": 15784
    },
    {
      "epoch": 2.760218579234973,
      "grad_norm": 0.6476022119232122,
      "learning_rate": 4.627153207004473e-05,
      "loss": 11.6321,
      "step": 15785
    },
    {
      "epoch": 2.760393442622951,
      "grad_norm": 0.6296173187681336,
      "learning_rate": 4.6259589303638804e-05,
      "loss": 11.8216,
      "step": 15786
    },
    {
      "epoch": 2.7605683060109287,
      "grad_norm": 0.5818468617646887,
      "learning_rate": 4.624764761489313e-05,
      "loss": 11.727,
      "step": 15787
    },
    {
      "epoch": 2.760743169398907,
      "grad_norm": 0.6871754272819177,
      "learning_rate": 4.6235707004047145e-05,
      "loss": 11.7151,
      "step": 15788
    },
    {
      "epoch": 2.7609180327868854,
      "grad_norm": 0.7339299509537089,
      "learning_rate": 4.6223767471340326e-05,
      "loss": 11.7676,
      "step": 15789
    },
    {
      "epoch": 2.7610928961748633,
      "grad_norm": 0.5766596955664118,
      "learning_rate": 4.6211829017012084e-05,
      "loss": 11.6675,
      "step": 15790
    },
    {
      "epoch": 2.7612677595628417,
      "grad_norm": 0.6790596570715502,
      "learning_rate": 4.619989164130182e-05,
      "loss": 11.7852,
      "step": 15791
    },
    {
      "epoch": 2.7614426229508195,
      "grad_norm": 0.6051368917661691,
      "learning_rate": 4.618795534444892e-05,
      "loss": 11.9759,
      "step": 15792
    },
    {
      "epoch": 2.761617486338798,
      "grad_norm": 0.6172670860648827,
      "learning_rate": 4.617602012669273e-05,
      "loss": 11.9683,
      "step": 15793
    },
    {
      "epoch": 2.7617923497267762,
      "grad_norm": 0.6926339010393954,
      "learning_rate": 4.61640859882726e-05,
      "loss": 11.731,
      "step": 15794
    },
    {
      "epoch": 2.761967213114754,
      "grad_norm": 0.6027134613616143,
      "learning_rate": 4.615215292942785e-05,
      "loss": 11.7654,
      "step": 15795
    },
    {
      "epoch": 2.762142076502732,
      "grad_norm": 0.6020472750932436,
      "learning_rate": 4.6140220950397764e-05,
      "loss": 11.8713,
      "step": 15796
    },
    {
      "epoch": 2.7623169398907104,
      "grad_norm": 0.6201465857633305,
      "learning_rate": 4.612829005142162e-05,
      "loss": 11.9153,
      "step": 15797
    },
    {
      "epoch": 2.7624918032786887,
      "grad_norm": 0.5787793332699268,
      "learning_rate": 4.6116360232738675e-05,
      "loss": 11.7421,
      "step": 15798
    },
    {
      "epoch": 2.7626666666666666,
      "grad_norm": 0.6038538906305577,
      "learning_rate": 4.610443149458814e-05,
      "loss": 11.4097,
      "step": 15799
    },
    {
      "epoch": 2.762841530054645,
      "grad_norm": 0.7160629888205697,
      "learning_rate": 4.609250383720924e-05,
      "loss": 11.8287,
      "step": 15800
    },
    {
      "epoch": 2.763016393442623,
      "grad_norm": 0.6400194666455149,
      "learning_rate": 4.6080577260841165e-05,
      "loss": 11.8945,
      "step": 15801
    },
    {
      "epoch": 2.763191256830601,
      "grad_norm": 0.6877461230606786,
      "learning_rate": 4.606865176572306e-05,
      "loss": 11.9053,
      "step": 15802
    },
    {
      "epoch": 2.763366120218579,
      "grad_norm": 0.6477595470167432,
      "learning_rate": 4.6056727352094076e-05,
      "loss": 11.5611,
      "step": 15803
    },
    {
      "epoch": 2.7635409836065574,
      "grad_norm": 0.6241075911083837,
      "learning_rate": 4.6044804020193334e-05,
      "loss": 11.9569,
      "step": 15804
    },
    {
      "epoch": 2.7637158469945353,
      "grad_norm": 0.7125796718768008,
      "learning_rate": 4.6032881770259953e-05,
      "loss": 11.7554,
      "step": 15805
    },
    {
      "epoch": 2.7638907103825137,
      "grad_norm": 0.6316394347333589,
      "learning_rate": 4.602096060253297e-05,
      "loss": 11.7864,
      "step": 15806
    },
    {
      "epoch": 2.764065573770492,
      "grad_norm": 0.6897931139838472,
      "learning_rate": 4.600904051725148e-05,
      "loss": 11.7166,
      "step": 15807
    },
    {
      "epoch": 2.76424043715847,
      "grad_norm": 0.6314245741012217,
      "learning_rate": 4.599712151465449e-05,
      "loss": 11.6313,
      "step": 15808
    },
    {
      "epoch": 2.7644153005464482,
      "grad_norm": 0.6721392297731407,
      "learning_rate": 4.5985203594981044e-05,
      "loss": 11.7058,
      "step": 15809
    },
    {
      "epoch": 2.764590163934426,
      "grad_norm": 0.5873700841298013,
      "learning_rate": 4.59732867584701e-05,
      "loss": 11.7594,
      "step": 15810
    },
    {
      "epoch": 2.7647650273224045,
      "grad_norm": 0.6711009122389615,
      "learning_rate": 4.5961371005360645e-05,
      "loss": 11.8294,
      "step": 15811
    },
    {
      "epoch": 2.7649398907103824,
      "grad_norm": 0.5829974896078218,
      "learning_rate": 4.594945633589162e-05,
      "loss": 11.7509,
      "step": 15812
    },
    {
      "epoch": 2.7651147540983607,
      "grad_norm": 0.6049019160816796,
      "learning_rate": 4.593754275030194e-05,
      "loss": 11.8829,
      "step": 15813
    },
    {
      "epoch": 2.7652896174863386,
      "grad_norm": 0.6597000021299438,
      "learning_rate": 4.592563024883053e-05,
      "loss": 11.9322,
      "step": 15814
    },
    {
      "epoch": 2.765464480874317,
      "grad_norm": 0.6644656770363,
      "learning_rate": 4.591371883171627e-05,
      "loss": 11.8646,
      "step": 15815
    },
    {
      "epoch": 2.7656393442622953,
      "grad_norm": 0.6327195090493014,
      "learning_rate": 4.5901808499198004e-05,
      "loss": 11.7258,
      "step": 15816
    },
    {
      "epoch": 2.765814207650273,
      "grad_norm": 0.6515026925072962,
      "learning_rate": 4.588989925151459e-05,
      "loss": 12.0696,
      "step": 15817
    },
    {
      "epoch": 2.765989071038251,
      "grad_norm": 0.6664030493892041,
      "learning_rate": 4.587799108890487e-05,
      "loss": 11.7318,
      "step": 15818
    },
    {
      "epoch": 2.7661639344262294,
      "grad_norm": 0.6220297508991677,
      "learning_rate": 4.586608401160754e-05,
      "loss": 11.7809,
      "step": 15819
    },
    {
      "epoch": 2.7663387978142078,
      "grad_norm": 0.69883904832504,
      "learning_rate": 4.585417801986146e-05,
      "loss": 11.9103,
      "step": 15820
    },
    {
      "epoch": 2.7665136612021857,
      "grad_norm": 0.7204557225746714,
      "learning_rate": 4.5842273113905365e-05,
      "loss": 12.0122,
      "step": 15821
    },
    {
      "epoch": 2.766688524590164,
      "grad_norm": 0.6296942241255109,
      "learning_rate": 4.583036929397798e-05,
      "loss": 11.6585,
      "step": 15822
    },
    {
      "epoch": 2.766863387978142,
      "grad_norm": 0.6566941090285977,
      "learning_rate": 4.581846656031805e-05,
      "loss": 11.7272,
      "step": 15823
    },
    {
      "epoch": 2.7670382513661202,
      "grad_norm": 0.6266905402197993,
      "learning_rate": 4.5806564913164155e-05,
      "loss": 11.6162,
      "step": 15824
    },
    {
      "epoch": 2.7672131147540986,
      "grad_norm": 0.7143894309016563,
      "learning_rate": 4.5794664352755055e-05,
      "loss": 11.5296,
      "step": 15825
    },
    {
      "epoch": 2.7673879781420765,
      "grad_norm": 0.6456723392217996,
      "learning_rate": 4.578276487932938e-05,
      "loss": 11.7312,
      "step": 15826
    },
    {
      "epoch": 2.7675628415300544,
      "grad_norm": 0.7118927755915057,
      "learning_rate": 4.5770866493125776e-05,
      "loss": 11.7277,
      "step": 15827
    },
    {
      "epoch": 2.7677377049180327,
      "grad_norm": 0.6276511291507385,
      "learning_rate": 4.575896919438276e-05,
      "loss": 11.8533,
      "step": 15828
    },
    {
      "epoch": 2.767912568306011,
      "grad_norm": 0.6261279956540571,
      "learning_rate": 4.574707298333891e-05,
      "loss": 11.7158,
      "step": 15829
    },
    {
      "epoch": 2.768087431693989,
      "grad_norm": 0.6873672040253309,
      "learning_rate": 4.573517786023287e-05,
      "loss": 11.7503,
      "step": 15830
    },
    {
      "epoch": 2.7682622950819673,
      "grad_norm": 0.6792656894535789,
      "learning_rate": 4.5723283825303133e-05,
      "loss": 11.9176,
      "step": 15831
    },
    {
      "epoch": 2.768437158469945,
      "grad_norm": 0.700523090382574,
      "learning_rate": 4.571139087878824e-05,
      "loss": 11.8129,
      "step": 15832
    },
    {
      "epoch": 2.7686120218579235,
      "grad_norm": 0.6851682891616878,
      "learning_rate": 4.5699499020926575e-05,
      "loss": 11.7196,
      "step": 15833
    },
    {
      "epoch": 2.768786885245902,
      "grad_norm": 0.6445853646946468,
      "learning_rate": 4.5687608251956714e-05,
      "loss": 11.9493,
      "step": 15834
    },
    {
      "epoch": 2.76896174863388,
      "grad_norm": 0.5994880658198979,
      "learning_rate": 4.5675718572117064e-05,
      "loss": 11.9518,
      "step": 15835
    },
    {
      "epoch": 2.7691366120218577,
      "grad_norm": 0.6413705073581344,
      "learning_rate": 4.56638299816461e-05,
      "loss": 11.5542,
      "step": 15836
    },
    {
      "epoch": 2.769311475409836,
      "grad_norm": 0.6485481084856187,
      "learning_rate": 4.565194248078214e-05,
      "loss": 11.9248,
      "step": 15837
    },
    {
      "epoch": 2.7694863387978144,
      "grad_norm": 0.6522371062389023,
      "learning_rate": 4.564005606976356e-05,
      "loss": 11.9384,
      "step": 15838
    },
    {
      "epoch": 2.7696612021857923,
      "grad_norm": 0.6969324858758136,
      "learning_rate": 4.5628170748828804e-05,
      "loss": 11.7605,
      "step": 15839
    },
    {
      "epoch": 2.7698360655737706,
      "grad_norm": 0.6900004178907535,
      "learning_rate": 4.561628651821617e-05,
      "loss": 11.8655,
      "step": 15840
    },
    {
      "epoch": 2.7700109289617485,
      "grad_norm": 0.6436633144728988,
      "learning_rate": 4.560440337816401e-05,
      "loss": 11.6358,
      "step": 15841
    },
    {
      "epoch": 2.770185792349727,
      "grad_norm": 0.6694649159593555,
      "learning_rate": 4.559252132891053e-05,
      "loss": 11.5822,
      "step": 15842
    },
    {
      "epoch": 2.770360655737705,
      "grad_norm": 0.6907991401594012,
      "learning_rate": 4.5580640370694027e-05,
      "loss": 11.8351,
      "step": 15843
    },
    {
      "epoch": 2.770535519125683,
      "grad_norm": 0.7105803242008947,
      "learning_rate": 4.556876050375281e-05,
      "loss": 11.666,
      "step": 15844
    },
    {
      "epoch": 2.770710382513661,
      "grad_norm": 0.7031384324179392,
      "learning_rate": 4.5556881728325094e-05,
      "loss": 11.6954,
      "step": 15845
    },
    {
      "epoch": 2.7708852459016393,
      "grad_norm": 0.6967815353730195,
      "learning_rate": 4.5545004044649034e-05,
      "loss": 12.0165,
      "step": 15846
    },
    {
      "epoch": 2.7710601092896177,
      "grad_norm": 0.6273197245319079,
      "learning_rate": 4.553312745296284e-05,
      "loss": 11.8412,
      "step": 15847
    },
    {
      "epoch": 2.7712349726775956,
      "grad_norm": 0.6917241260315997,
      "learning_rate": 4.552125195350464e-05,
      "loss": 11.8641,
      "step": 15848
    },
    {
      "epoch": 2.771409836065574,
      "grad_norm": 0.6108854147183291,
      "learning_rate": 4.5509377546512656e-05,
      "loss": 11.8734,
      "step": 15849
    },
    {
      "epoch": 2.771584699453552,
      "grad_norm": 0.6609962022409975,
      "learning_rate": 4.5497504232224986e-05,
      "loss": 11.7632,
      "step": 15850
    },
    {
      "epoch": 2.77175956284153,
      "grad_norm": 0.6561329814908782,
      "learning_rate": 4.548563201087966e-05,
      "loss": 11.5381,
      "step": 15851
    },
    {
      "epoch": 2.771934426229508,
      "grad_norm": 0.6062330920089517,
      "learning_rate": 4.54737608827148e-05,
      "loss": 11.9416,
      "step": 15852
    },
    {
      "epoch": 2.7721092896174864,
      "grad_norm": 0.6533976301204246,
      "learning_rate": 4.5461890847968416e-05,
      "loss": 11.8023,
      "step": 15853
    },
    {
      "epoch": 2.7722841530054643,
      "grad_norm": 0.6171045747263519,
      "learning_rate": 4.545002190687865e-05,
      "loss": 11.8032,
      "step": 15854
    },
    {
      "epoch": 2.7724590163934426,
      "grad_norm": 0.680560954180562,
      "learning_rate": 4.5438154059683394e-05,
      "loss": 11.7083,
      "step": 15855
    },
    {
      "epoch": 2.772633879781421,
      "grad_norm": 0.6304224730627858,
      "learning_rate": 4.5426287306620675e-05,
      "loss": 11.8552,
      "step": 15856
    },
    {
      "epoch": 2.772808743169399,
      "grad_norm": 0.6949615439074387,
      "learning_rate": 4.541442164792843e-05,
      "loss": 11.8297,
      "step": 15857
    },
    {
      "epoch": 2.772983606557377,
      "grad_norm": 0.6454915986713221,
      "learning_rate": 4.5402557083844654e-05,
      "loss": 11.8304,
      "step": 15858
    },
    {
      "epoch": 2.773158469945355,
      "grad_norm": 0.6566387044032168,
      "learning_rate": 4.5390693614607296e-05,
      "loss": 11.7573,
      "step": 15859
    },
    {
      "epoch": 2.7733333333333334,
      "grad_norm": 0.704148056672308,
      "learning_rate": 4.537883124045416e-05,
      "loss": 11.9171,
      "step": 15860
    },
    {
      "epoch": 2.7735081967213113,
      "grad_norm": 0.6565148265367327,
      "learning_rate": 4.5366969961623166e-05,
      "loss": 11.6481,
      "step": 15861
    },
    {
      "epoch": 2.7736830601092897,
      "grad_norm": 0.6839696322272422,
      "learning_rate": 4.5355109778352135e-05,
      "loss": 11.4494,
      "step": 15862
    },
    {
      "epoch": 2.7738579234972676,
      "grad_norm": 0.6971237886288469,
      "learning_rate": 4.5343250690879e-05,
      "loss": 11.8948,
      "step": 15863
    },
    {
      "epoch": 2.774032786885246,
      "grad_norm": 0.6054073049280182,
      "learning_rate": 4.533139269944148e-05,
      "loss": 11.7038,
      "step": 15864
    },
    {
      "epoch": 2.7742076502732242,
      "grad_norm": 0.6089240352565222,
      "learning_rate": 4.531953580427739e-05,
      "loss": 11.8668,
      "step": 15865
    },
    {
      "epoch": 2.774382513661202,
      "grad_norm": 0.6611374732180073,
      "learning_rate": 4.53076800056245e-05,
      "loss": 11.8751,
      "step": 15866
    },
    {
      "epoch": 2.7745573770491805,
      "grad_norm": 0.6470589284174472,
      "learning_rate": 4.5295825303720505e-05,
      "loss": 11.9554,
      "step": 15867
    },
    {
      "epoch": 2.7747322404371584,
      "grad_norm": 0.6477877163842998,
      "learning_rate": 4.528397169880326e-05,
      "loss": 11.5824,
      "step": 15868
    },
    {
      "epoch": 2.7749071038251367,
      "grad_norm": 0.6473692223466235,
      "learning_rate": 4.527211919111034e-05,
      "loss": 11.7966,
      "step": 15869
    },
    {
      "epoch": 2.7750819672131146,
      "grad_norm": 0.6697892609712215,
      "learning_rate": 4.526026778087947e-05,
      "loss": 11.8985,
      "step": 15870
    },
    {
      "epoch": 2.775256830601093,
      "grad_norm": 0.6941588650506728,
      "learning_rate": 4.524841746834831e-05,
      "loss": 11.9444,
      "step": 15871
    },
    {
      "epoch": 2.775431693989071,
      "grad_norm": 0.6284669199169821,
      "learning_rate": 4.523656825375449e-05,
      "loss": 11.7458,
      "step": 15872
    },
    {
      "epoch": 2.775606557377049,
      "grad_norm": 0.6681678394105061,
      "learning_rate": 4.522472013733561e-05,
      "loss": 11.9929,
      "step": 15873
    },
    {
      "epoch": 2.7757814207650275,
      "grad_norm": 0.6122258080198764,
      "learning_rate": 4.521287311932928e-05,
      "loss": 11.8769,
      "step": 15874
    },
    {
      "epoch": 2.7759562841530054,
      "grad_norm": 0.7135499437873686,
      "learning_rate": 4.520102719997306e-05,
      "loss": 11.8264,
      "step": 15875
    },
    {
      "epoch": 2.7761311475409833,
      "grad_norm": 0.6214085902009655,
      "learning_rate": 4.51891823795045e-05,
      "loss": 11.7366,
      "step": 15876
    },
    {
      "epoch": 2.7763060109289617,
      "grad_norm": 0.6059915314969605,
      "learning_rate": 4.517733865816113e-05,
      "loss": 11.9279,
      "step": 15877
    },
    {
      "epoch": 2.77648087431694,
      "grad_norm": 0.642955438172694,
      "learning_rate": 4.5165496036180445e-05,
      "loss": 11.8805,
      "step": 15878
    },
    {
      "epoch": 2.776655737704918,
      "grad_norm": 0.6598206477487796,
      "learning_rate": 4.515365451379993e-05,
      "loss": 11.7327,
      "step": 15879
    },
    {
      "epoch": 2.7768306010928963,
      "grad_norm": 0.6369949805923146,
      "learning_rate": 4.514181409125704e-05,
      "loss": 11.6128,
      "step": 15880
    },
    {
      "epoch": 2.777005464480874,
      "grad_norm": 0.676903296619598,
      "learning_rate": 4.5129974768789206e-05,
      "loss": 11.5578,
      "step": 15881
    },
    {
      "epoch": 2.7771803278688525,
      "grad_norm": 0.6802889481220534,
      "learning_rate": 4.511813654663386e-05,
      "loss": 11.8953,
      "step": 15882
    },
    {
      "epoch": 2.777355191256831,
      "grad_norm": 0.6520942788803918,
      "learning_rate": 4.510629942502839e-05,
      "loss": 11.8778,
      "step": 15883
    },
    {
      "epoch": 2.7775300546448087,
      "grad_norm": 0.6899775808339511,
      "learning_rate": 4.509446340421015e-05,
      "loss": 11.9029,
      "step": 15884
    },
    {
      "epoch": 2.7777049180327866,
      "grad_norm": 0.6289827364380083,
      "learning_rate": 4.5082628484416514e-05,
      "loss": 11.6451,
      "step": 15885
    },
    {
      "epoch": 2.777879781420765,
      "grad_norm": 0.6348420980805791,
      "learning_rate": 4.5070794665884786e-05,
      "loss": 11.7919,
      "step": 15886
    },
    {
      "epoch": 2.7780546448087433,
      "grad_norm": 0.6952417750025395,
      "learning_rate": 4.505896194885229e-05,
      "loss": 11.9491,
      "step": 15887
    },
    {
      "epoch": 2.778229508196721,
      "grad_norm": 0.5947965978555044,
      "learning_rate": 4.504713033355629e-05,
      "loss": 11.6443,
      "step": 15888
    },
    {
      "epoch": 2.7784043715846996,
      "grad_norm": 0.6668209174170739,
      "learning_rate": 4.5035299820234054e-05,
      "loss": 11.6222,
      "step": 15889
    },
    {
      "epoch": 2.7785792349726774,
      "grad_norm": 0.6837771484103807,
      "learning_rate": 4.502347040912283e-05,
      "loss": 11.7973,
      "step": 15890
    },
    {
      "epoch": 2.778754098360656,
      "grad_norm": 0.6456236724827438,
      "learning_rate": 4.5011642100459804e-05,
      "loss": 11.8849,
      "step": 15891
    },
    {
      "epoch": 2.778928961748634,
      "grad_norm": 0.6711914703389997,
      "learning_rate": 4.499981489448221e-05,
      "loss": 11.8959,
      "step": 15892
    },
    {
      "epoch": 2.779103825136612,
      "grad_norm": 0.6371222973675349,
      "learning_rate": 4.49879887914272e-05,
      "loss": 11.7362,
      "step": 15893
    },
    {
      "epoch": 2.77927868852459,
      "grad_norm": 0.8141797416038973,
      "learning_rate": 4.49761637915319e-05,
      "loss": 11.632,
      "step": 15894
    },
    {
      "epoch": 2.7794535519125683,
      "grad_norm": 0.6447911101879532,
      "learning_rate": 4.4964339895033484e-05,
      "loss": 11.2264,
      "step": 15895
    },
    {
      "epoch": 2.7796284153005466,
      "grad_norm": 0.6987621494863604,
      "learning_rate": 4.4952517102169026e-05,
      "loss": 11.7623,
      "step": 15896
    },
    {
      "epoch": 2.7798032786885245,
      "grad_norm": 0.6065316253368839,
      "learning_rate": 4.4940695413175626e-05,
      "loss": 11.7371,
      "step": 15897
    },
    {
      "epoch": 2.779978142076503,
      "grad_norm": 0.6979702681896727,
      "learning_rate": 4.492887482829032e-05,
      "loss": 11.8737,
      "step": 15898
    },
    {
      "epoch": 2.7801530054644807,
      "grad_norm": 0.5904740867720243,
      "learning_rate": 4.4917055347750184e-05,
      "loss": 11.5955,
      "step": 15899
    },
    {
      "epoch": 2.780327868852459,
      "grad_norm": 0.650401650547548,
      "learning_rate": 4.4905236971792244e-05,
      "loss": 11.7049,
      "step": 15900
    },
    {
      "epoch": 2.7805027322404374,
      "grad_norm": 0.6117365013130999,
      "learning_rate": 4.4893419700653394e-05,
      "loss": 11.7869,
      "step": 15901
    },
    {
      "epoch": 2.7806775956284153,
      "grad_norm": 0.6171869023542428,
      "learning_rate": 4.4881603534570715e-05,
      "loss": 11.5987,
      "step": 15902
    },
    {
      "epoch": 2.780852459016393,
      "grad_norm": 0.6161161256393191,
      "learning_rate": 4.486978847378112e-05,
      "loss": 11.8726,
      "step": 15903
    },
    {
      "epoch": 2.7810273224043716,
      "grad_norm": 0.7558418782105872,
      "learning_rate": 4.485797451852154e-05,
      "loss": 11.8144,
      "step": 15904
    },
    {
      "epoch": 2.78120218579235,
      "grad_norm": 0.6113577086548748,
      "learning_rate": 4.4846161669028876e-05,
      "loss": 11.7757,
      "step": 15905
    },
    {
      "epoch": 2.781377049180328,
      "grad_norm": 0.6135884459714882,
      "learning_rate": 4.483434992554001e-05,
      "loss": 11.844,
      "step": 15906
    },
    {
      "epoch": 2.781551912568306,
      "grad_norm": 0.5973152955755602,
      "learning_rate": 4.482253928829182e-05,
      "loss": 11.9182,
      "step": 15907
    },
    {
      "epoch": 2.781726775956284,
      "grad_norm": 0.6781120961738631,
      "learning_rate": 4.481072975752113e-05,
      "loss": 11.6895,
      "step": 15908
    },
    {
      "epoch": 2.7819016393442624,
      "grad_norm": 0.5850164571089848,
      "learning_rate": 4.4798921333464796e-05,
      "loss": 11.7406,
      "step": 15909
    },
    {
      "epoch": 2.7820765027322403,
      "grad_norm": 0.6273583233513235,
      "learning_rate": 4.47871140163595e-05,
      "loss": 11.8786,
      "step": 15910
    },
    {
      "epoch": 2.7822513661202186,
      "grad_norm": 0.6508517948454213,
      "learning_rate": 4.4775307806442136e-05,
      "loss": 11.7663,
      "step": 15911
    },
    {
      "epoch": 2.7824262295081965,
      "grad_norm": 0.641862231874394,
      "learning_rate": 4.476350270394942e-05,
      "loss": 11.7318,
      "step": 15912
    },
    {
      "epoch": 2.782601092896175,
      "grad_norm": 0.6243238175439009,
      "learning_rate": 4.475169870911806e-05,
      "loss": 11.5908,
      "step": 15913
    },
    {
      "epoch": 2.782775956284153,
      "grad_norm": 0.6554383302302792,
      "learning_rate": 4.473989582218483e-05,
      "loss": 11.9255,
      "step": 15914
    },
    {
      "epoch": 2.782950819672131,
      "grad_norm": 0.6859175650499639,
      "learning_rate": 4.472809404338627e-05,
      "loss": 11.6137,
      "step": 15915
    },
    {
      "epoch": 2.7831256830601094,
      "grad_norm": 0.640438896270139,
      "learning_rate": 4.4716293372959176e-05,
      "loss": 11.9757,
      "step": 15916
    },
    {
      "epoch": 2.7833005464480873,
      "grad_norm": 0.7778864815341926,
      "learning_rate": 4.4704493811140136e-05,
      "loss": 11.8857,
      "step": 15917
    },
    {
      "epoch": 2.7834754098360657,
      "grad_norm": 0.6597893412910386,
      "learning_rate": 4.4692695358165815e-05,
      "loss": 11.8466,
      "step": 15918
    },
    {
      "epoch": 2.7836502732240436,
      "grad_norm": 0.630554080419985,
      "learning_rate": 4.468089801427273e-05,
      "loss": 11.8585,
      "step": 15919
    },
    {
      "epoch": 2.783825136612022,
      "grad_norm": 0.7043307807332212,
      "learning_rate": 4.4669101779697465e-05,
      "loss": 11.9222,
      "step": 15920
    },
    {
      "epoch": 2.784,
      "grad_norm": 0.6530984777480121,
      "learning_rate": 4.4657306654676623e-05,
      "loss": 11.7295,
      "step": 15921
    },
    {
      "epoch": 2.784174863387978,
      "grad_norm": 0.7152053186204405,
      "learning_rate": 4.4645512639446704e-05,
      "loss": 11.8367,
      "step": 15922
    },
    {
      "epoch": 2.7843497267759565,
      "grad_norm": 0.5842958593538154,
      "learning_rate": 4.463371973424426e-05,
      "loss": 11.7354,
      "step": 15923
    },
    {
      "epoch": 2.7845245901639344,
      "grad_norm": 0.6488581240023901,
      "learning_rate": 4.4621927939305695e-05,
      "loss": 11.5558,
      "step": 15924
    },
    {
      "epoch": 2.7846994535519127,
      "grad_norm": 0.6081135420457996,
      "learning_rate": 4.461013725486748e-05,
      "loss": 11.6133,
      "step": 15925
    },
    {
      "epoch": 2.7848743169398906,
      "grad_norm": 0.6298226858956413,
      "learning_rate": 4.4598347681166107e-05,
      "loss": 11.8941,
      "step": 15926
    },
    {
      "epoch": 2.785049180327869,
      "grad_norm": 0.6523699855488126,
      "learning_rate": 4.4586559218438005e-05,
      "loss": 11.8453,
      "step": 15927
    },
    {
      "epoch": 2.785224043715847,
      "grad_norm": 0.7777960486581371,
      "learning_rate": 4.4574771866919505e-05,
      "loss": 11.9437,
      "step": 15928
    },
    {
      "epoch": 2.785398907103825,
      "grad_norm": 0.5735235339293157,
      "learning_rate": 4.456298562684696e-05,
      "loss": 11.7036,
      "step": 15929
    },
    {
      "epoch": 2.785573770491803,
      "grad_norm": 0.630833262884415,
      "learning_rate": 4.4551200498456814e-05,
      "loss": 11.6316,
      "step": 15930
    },
    {
      "epoch": 2.7857486338797814,
      "grad_norm": 0.7009903711923254,
      "learning_rate": 4.453941648198534e-05,
      "loss": 12.0248,
      "step": 15931
    },
    {
      "epoch": 2.78592349726776,
      "grad_norm": 0.6448833110850284,
      "learning_rate": 4.4527633577668895e-05,
      "loss": 11.89,
      "step": 15932
    },
    {
      "epoch": 2.7860983606557377,
      "grad_norm": 0.629278113063941,
      "learning_rate": 4.451585178574368e-05,
      "loss": 11.7161,
      "step": 15933
    },
    {
      "epoch": 2.7862732240437156,
      "grad_norm": 0.6300719078264593,
      "learning_rate": 4.450407110644595e-05,
      "loss": 11.7184,
      "step": 15934
    },
    {
      "epoch": 2.786448087431694,
      "grad_norm": 0.6371333529126576,
      "learning_rate": 4.4492291540012035e-05,
      "loss": 11.8035,
      "step": 15935
    },
    {
      "epoch": 2.7866229508196723,
      "grad_norm": 0.7460152041942549,
      "learning_rate": 4.44805130866781e-05,
      "loss": 11.8999,
      "step": 15936
    },
    {
      "epoch": 2.78679781420765,
      "grad_norm": 0.56840405481692,
      "learning_rate": 4.446873574668039e-05,
      "loss": 11.8282,
      "step": 15937
    },
    {
      "epoch": 2.7869726775956285,
      "grad_norm": 0.7487211878233039,
      "learning_rate": 4.4456959520254994e-05,
      "loss": 11.9063,
      "step": 15938
    },
    {
      "epoch": 2.7871475409836064,
      "grad_norm": 0.6053694032775482,
      "learning_rate": 4.444518440763805e-05,
      "loss": 11.5798,
      "step": 15939
    },
    {
      "epoch": 2.7873224043715847,
      "grad_norm": 0.6608338930825978,
      "learning_rate": 4.443341040906578e-05,
      "loss": 11.9372,
      "step": 15940
    },
    {
      "epoch": 2.787497267759563,
      "grad_norm": 0.6930373216733917,
      "learning_rate": 4.4421637524774285e-05,
      "loss": 11.7025,
      "step": 15941
    },
    {
      "epoch": 2.787672131147541,
      "grad_norm": 0.6281919893461317,
      "learning_rate": 4.440986575499956e-05,
      "loss": 11.8504,
      "step": 15942
    },
    {
      "epoch": 2.787846994535519,
      "grad_norm": 0.6010158743310446,
      "learning_rate": 4.439809509997772e-05,
      "loss": 11.921,
      "step": 15943
    },
    {
      "epoch": 2.788021857923497,
      "grad_norm": 0.6180983254588627,
      "learning_rate": 4.438632555994475e-05,
      "loss": 11.7155,
      "step": 15944
    },
    {
      "epoch": 2.7881967213114756,
      "grad_norm": 0.6207605151155372,
      "learning_rate": 4.4374557135136744e-05,
      "loss": 11.6628,
      "step": 15945
    },
    {
      "epoch": 2.7883715846994535,
      "grad_norm": 0.6131544062822187,
      "learning_rate": 4.43627898257897e-05,
      "loss": 11.8261,
      "step": 15946
    },
    {
      "epoch": 2.788546448087432,
      "grad_norm": 0.6542772333252113,
      "learning_rate": 4.435102363213951e-05,
      "loss": 11.798,
      "step": 15947
    },
    {
      "epoch": 2.7887213114754097,
      "grad_norm": 0.6204102058159647,
      "learning_rate": 4.433925855442216e-05,
      "loss": 11.8076,
      "step": 15948
    },
    {
      "epoch": 2.788896174863388,
      "grad_norm": 0.6663540386663366,
      "learning_rate": 4.4327494592873544e-05,
      "loss": 11.8001,
      "step": 15949
    },
    {
      "epoch": 2.7890710382513664,
      "grad_norm": 0.7039976753140098,
      "learning_rate": 4.431573174772967e-05,
      "loss": 11.7477,
      "step": 15950
    },
    {
      "epoch": 2.7892459016393443,
      "grad_norm": 0.649970555206536,
      "learning_rate": 4.430397001922631e-05,
      "loss": 11.8949,
      "step": 15951
    },
    {
      "epoch": 2.789420765027322,
      "grad_norm": 0.6251814025813617,
      "learning_rate": 4.429220940759936e-05,
      "loss": 11.7419,
      "step": 15952
    },
    {
      "epoch": 2.7895956284153005,
      "grad_norm": 0.6470451148940864,
      "learning_rate": 4.428044991308463e-05,
      "loss": 11.8808,
      "step": 15953
    },
    {
      "epoch": 2.789770491803279,
      "grad_norm": 0.6268385150595952,
      "learning_rate": 4.4268691535917983e-05,
      "loss": 11.8487,
      "step": 15954
    },
    {
      "epoch": 2.7899453551912567,
      "grad_norm": 0.618673606271153,
      "learning_rate": 4.4256934276335235e-05,
      "loss": 12.0069,
      "step": 15955
    },
    {
      "epoch": 2.790120218579235,
      "grad_norm": 0.6304797377595375,
      "learning_rate": 4.424517813457206e-05,
      "loss": 11.8458,
      "step": 15956
    },
    {
      "epoch": 2.790295081967213,
      "grad_norm": 0.6732794269855548,
      "learning_rate": 4.423342311086427e-05,
      "loss": 11.7189,
      "step": 15957
    },
    {
      "epoch": 2.7904699453551913,
      "grad_norm": 0.6564770063327343,
      "learning_rate": 4.422166920544752e-05,
      "loss": 11.9152,
      "step": 15958
    },
    {
      "epoch": 2.7906448087431697,
      "grad_norm": 0.6150788299109458,
      "learning_rate": 4.4209916418557654e-05,
      "loss": 11.7839,
      "step": 15959
    },
    {
      "epoch": 2.7908196721311476,
      "grad_norm": 0.6910759063466415,
      "learning_rate": 4.4198164750430217e-05,
      "loss": 11.6588,
      "step": 15960
    },
    {
      "epoch": 2.7909945355191255,
      "grad_norm": 0.6262300096021131,
      "learning_rate": 4.418641420130091e-05,
      "loss": 11.7826,
      "step": 15961
    },
    {
      "epoch": 2.791169398907104,
      "grad_norm": 0.6533496482430211,
      "learning_rate": 4.4174664771405364e-05,
      "loss": 11.7549,
      "step": 15962
    },
    {
      "epoch": 2.791344262295082,
      "grad_norm": 0.5913007718624441,
      "learning_rate": 4.416291646097916e-05,
      "loss": 11.8197,
      "step": 15963
    },
    {
      "epoch": 2.79151912568306,
      "grad_norm": 0.6634446649110142,
      "learning_rate": 4.4151169270258005e-05,
      "loss": 11.8254,
      "step": 15964
    },
    {
      "epoch": 2.7916939890710384,
      "grad_norm": 0.6395968379071583,
      "learning_rate": 4.413942319947734e-05,
      "loss": 12.0815,
      "step": 15965
    },
    {
      "epoch": 2.7918688524590163,
      "grad_norm": 0.6350378388150925,
      "learning_rate": 4.412767824887276e-05,
      "loss": 11.9249,
      "step": 15966
    },
    {
      "epoch": 2.7920437158469946,
      "grad_norm": 0.6501083608231779,
      "learning_rate": 4.411593441867978e-05,
      "loss": 11.9297,
      "step": 15967
    },
    {
      "epoch": 2.7922185792349725,
      "grad_norm": 0.6313642488754738,
      "learning_rate": 4.41041917091339e-05,
      "loss": 11.8099,
      "step": 15968
    },
    {
      "epoch": 2.792393442622951,
      "grad_norm": 0.632449571173157,
      "learning_rate": 4.40924501204706e-05,
      "loss": 11.6305,
      "step": 15969
    },
    {
      "epoch": 2.7925683060109288,
      "grad_norm": 0.6261638467066494,
      "learning_rate": 4.4080709652925336e-05,
      "loss": 11.6238,
      "step": 15970
    },
    {
      "epoch": 2.792743169398907,
      "grad_norm": 0.7568083542456199,
      "learning_rate": 4.4068970306733526e-05,
      "loss": 11.929,
      "step": 15971
    },
    {
      "epoch": 2.7929180327868854,
      "grad_norm": 0.6181519632827338,
      "learning_rate": 4.405723208213061e-05,
      "loss": 11.7607,
      "step": 15972
    },
    {
      "epoch": 2.7930928961748633,
      "grad_norm": 0.6706161199416012,
      "learning_rate": 4.404549497935194e-05,
      "loss": 11.4815,
      "step": 15973
    },
    {
      "epoch": 2.7932677595628417,
      "grad_norm": 0.7094043861800668,
      "learning_rate": 4.4033758998632916e-05,
      "loss": 11.6598,
      "step": 15974
    },
    {
      "epoch": 2.7934426229508196,
      "grad_norm": 0.6443666106260809,
      "learning_rate": 4.402202414020885e-05,
      "loss": 11.6345,
      "step": 15975
    },
    {
      "epoch": 2.793617486338798,
      "grad_norm": 0.6393517328704638,
      "learning_rate": 4.401029040431508e-05,
      "loss": 11.6853,
      "step": 15976
    },
    {
      "epoch": 2.793792349726776,
      "grad_norm": 0.6163457236735871,
      "learning_rate": 4.39985577911869e-05,
      "loss": 11.7134,
      "step": 15977
    },
    {
      "epoch": 2.793967213114754,
      "grad_norm": 0.6157671794917724,
      "learning_rate": 4.398682630105958e-05,
      "loss": 11.9524,
      "step": 15978
    },
    {
      "epoch": 2.794142076502732,
      "grad_norm": 0.6436330402110557,
      "learning_rate": 4.397509593416838e-05,
      "loss": 11.4504,
      "step": 15979
    },
    {
      "epoch": 2.7943169398907104,
      "grad_norm": 0.6167230013526258,
      "learning_rate": 4.396336669074852e-05,
      "loss": 11.7012,
      "step": 15980
    },
    {
      "epoch": 2.7944918032786887,
      "grad_norm": 0.6325453459941227,
      "learning_rate": 4.3951638571035215e-05,
      "loss": 11.9875,
      "step": 15981
    },
    {
      "epoch": 2.7946666666666666,
      "grad_norm": 0.6317752952541623,
      "learning_rate": 4.3939911575263645e-05,
      "loss": 11.5631,
      "step": 15982
    },
    {
      "epoch": 2.794841530054645,
      "grad_norm": 0.6645388949607652,
      "learning_rate": 4.392818570366898e-05,
      "loss": 11.7503,
      "step": 15983
    },
    {
      "epoch": 2.795016393442623,
      "grad_norm": 0.725492146443513,
      "learning_rate": 4.391646095648635e-05,
      "loss": 11.5798,
      "step": 15984
    },
    {
      "epoch": 2.795191256830601,
      "grad_norm": 0.643603310587616,
      "learning_rate": 4.3904737333950886e-05,
      "loss": 11.9046,
      "step": 15985
    },
    {
      "epoch": 2.795366120218579,
      "grad_norm": 0.6541168052285967,
      "learning_rate": 4.3893014836297665e-05,
      "loss": 11.6945,
      "step": 15986
    },
    {
      "epoch": 2.7955409836065575,
      "grad_norm": 0.6095213048029741,
      "learning_rate": 4.388129346376178e-05,
      "loss": 11.8222,
      "step": 15987
    },
    {
      "epoch": 2.7957158469945353,
      "grad_norm": 0.646535144677048,
      "learning_rate": 4.386957321657825e-05,
      "loss": 11.8869,
      "step": 15988
    },
    {
      "epoch": 2.7958907103825137,
      "grad_norm": 0.6232055091158265,
      "learning_rate": 4.385785409498212e-05,
      "loss": 11.6531,
      "step": 15989
    },
    {
      "epoch": 2.796065573770492,
      "grad_norm": 0.6659627697923409,
      "learning_rate": 4.3846136099208414e-05,
      "loss": 11.949,
      "step": 15990
    },
    {
      "epoch": 2.79624043715847,
      "grad_norm": 0.629200032698947,
      "learning_rate": 4.383441922949207e-05,
      "loss": 11.8542,
      "step": 15991
    },
    {
      "epoch": 2.796415300546448,
      "grad_norm": 0.7095026013323232,
      "learning_rate": 4.382270348606809e-05,
      "loss": 11.661,
      "step": 15992
    },
    {
      "epoch": 2.796590163934426,
      "grad_norm": 0.7899434684802105,
      "learning_rate": 4.3810988869171385e-05,
      "loss": 11.6141,
      "step": 15993
    },
    {
      "epoch": 2.7967650273224045,
      "grad_norm": 0.7284869502446165,
      "learning_rate": 4.3799275379036874e-05,
      "loss": 11.8044,
      "step": 15994
    },
    {
      "epoch": 2.7969398907103824,
      "grad_norm": 0.6904255635389446,
      "learning_rate": 4.378756301589946e-05,
      "loss": 11.9378,
      "step": 15995
    },
    {
      "epoch": 2.7971147540983607,
      "grad_norm": 0.779264994633778,
      "learning_rate": 4.377585177999404e-05,
      "loss": 11.8439,
      "step": 15996
    },
    {
      "epoch": 2.7972896174863386,
      "grad_norm": 0.6403337604654589,
      "learning_rate": 4.3764141671555336e-05,
      "loss": 11.5043,
      "step": 15997
    },
    {
      "epoch": 2.797464480874317,
      "grad_norm": 0.6889514812264398,
      "learning_rate": 4.37524326908183e-05,
      "loss": 11.6027,
      "step": 15998
    },
    {
      "epoch": 2.7976393442622953,
      "grad_norm": 0.636751704657427,
      "learning_rate": 4.374072483801769e-05,
      "loss": 11.9034,
      "step": 15999
    },
    {
      "epoch": 2.797814207650273,
      "grad_norm": 0.6354535334577662,
      "learning_rate": 4.3729018113388286e-05,
      "loss": 11.7402,
      "step": 16000
    },
    {
      "epoch": 2.797989071038251,
      "grad_norm": 0.6949605209679823,
      "learning_rate": 4.371731251716485e-05,
      "loss": 11.7454,
      "step": 16001
    },
    {
      "epoch": 2.7981639344262295,
      "grad_norm": 0.6419131954313634,
      "learning_rate": 4.37056080495821e-05,
      "loss": 11.785,
      "step": 16002
    },
    {
      "epoch": 2.798338797814208,
      "grad_norm": 0.7429303295331616,
      "learning_rate": 4.369390471087477e-05,
      "loss": 11.5607,
      "step": 16003
    },
    {
      "epoch": 2.7985136612021857,
      "grad_norm": 0.7247989504542676,
      "learning_rate": 4.368220250127751e-05,
      "loss": 11.9614,
      "step": 16004
    },
    {
      "epoch": 2.798688524590164,
      "grad_norm": 0.6532546669944731,
      "learning_rate": 4.367050142102507e-05,
      "loss": 11.759,
      "step": 16005
    },
    {
      "epoch": 2.798863387978142,
      "grad_norm": 0.6982065180456601,
      "learning_rate": 4.365880147035194e-05,
      "loss": 11.5352,
      "step": 16006
    },
    {
      "epoch": 2.7990382513661203,
      "grad_norm": 0.7186518682371309,
      "learning_rate": 4.3647102649492866e-05,
      "loss": 11.9309,
      "step": 16007
    },
    {
      "epoch": 2.7992131147540986,
      "grad_norm": 0.6566698203649237,
      "learning_rate": 4.363540495868241e-05,
      "loss": 11.7724,
      "step": 16008
    },
    {
      "epoch": 2.7993879781420765,
      "grad_norm": 0.5887223661228925,
      "learning_rate": 4.362370839815516e-05,
      "loss": 11.8132,
      "step": 16009
    },
    {
      "epoch": 2.7995628415300544,
      "grad_norm": 0.6637779164466764,
      "learning_rate": 4.361201296814569e-05,
      "loss": 12.0187,
      "step": 16010
    },
    {
      "epoch": 2.7997377049180328,
      "grad_norm": 0.6005829907427734,
      "learning_rate": 4.3600318668888406e-05,
      "loss": 11.9164,
      "step": 16011
    },
    {
      "epoch": 2.799912568306011,
      "grad_norm": 0.6466164851069435,
      "learning_rate": 4.358862550061795e-05,
      "loss": 11.7473,
      "step": 16012
    },
    {
      "epoch": 2.800087431693989,
      "grad_norm": 0.6264543291792074,
      "learning_rate": 4.357693346356876e-05,
      "loss": 11.7346,
      "step": 16013
    },
    {
      "epoch": 2.8002622950819673,
      "grad_norm": 0.6992967977603455,
      "learning_rate": 4.3565242557975326e-05,
      "loss": 11.8627,
      "step": 16014
    },
    {
      "epoch": 2.8004371584699452,
      "grad_norm": 0.6634720794189165,
      "learning_rate": 4.355355278407202e-05,
      "loss": 11.935,
      "step": 16015
    },
    {
      "epoch": 2.8006120218579236,
      "grad_norm": 0.6197187744207255,
      "learning_rate": 4.354186414209326e-05,
      "loss": 11.6183,
      "step": 16016
    },
    {
      "epoch": 2.800786885245902,
      "grad_norm": 0.7040664034748295,
      "learning_rate": 4.353017663227351e-05,
      "loss": 11.7548,
      "step": 16017
    },
    {
      "epoch": 2.80096174863388,
      "grad_norm": 0.6668740552976508,
      "learning_rate": 4.35184902548471e-05,
      "loss": 11.5353,
      "step": 16018
    },
    {
      "epoch": 2.8011366120218577,
      "grad_norm": 0.657491666254844,
      "learning_rate": 4.350680501004842e-05,
      "loss": 11.7371,
      "step": 16019
    },
    {
      "epoch": 2.801311475409836,
      "grad_norm": 0.6954136990622561,
      "learning_rate": 4.349512089811172e-05,
      "loss": 11.717,
      "step": 16020
    },
    {
      "epoch": 2.8014863387978144,
      "grad_norm": 0.7082996869138527,
      "learning_rate": 4.34834379192713e-05,
      "loss": 11.7414,
      "step": 16021
    },
    {
      "epoch": 2.8016612021857923,
      "grad_norm": 0.6343531062255241,
      "learning_rate": 4.347175607376152e-05,
      "loss": 11.7584,
      "step": 16022
    },
    {
      "epoch": 2.8018360655737706,
      "grad_norm": 0.6427019855647239,
      "learning_rate": 4.3460075361816635e-05,
      "loss": 11.4097,
      "step": 16023
    },
    {
      "epoch": 2.8020109289617485,
      "grad_norm": 0.6022529906965622,
      "learning_rate": 4.344839578367079e-05,
      "loss": 11.6244,
      "step": 16024
    },
    {
      "epoch": 2.802185792349727,
      "grad_norm": 0.596804569980975,
      "learning_rate": 4.343671733955822e-05,
      "loss": 11.8029,
      "step": 16025
    },
    {
      "epoch": 2.8023606557377048,
      "grad_norm": 0.635738708943527,
      "learning_rate": 4.342504002971317e-05,
      "loss": 11.6668,
      "step": 16026
    },
    {
      "epoch": 2.802535519125683,
      "grad_norm": 0.6047684626458961,
      "learning_rate": 4.341336385436977e-05,
      "loss": 11.4901,
      "step": 16027
    },
    {
      "epoch": 2.802710382513661,
      "grad_norm": 0.6232194979128418,
      "learning_rate": 4.340168881376222e-05,
      "loss": 11.8884,
      "step": 16028
    },
    {
      "epoch": 2.8028852459016393,
      "grad_norm": 0.616738524983898,
      "learning_rate": 4.339001490812454e-05,
      "loss": 11.8821,
      "step": 16029
    },
    {
      "epoch": 2.8030601092896177,
      "grad_norm": 0.6169246598851448,
      "learning_rate": 4.337834213769085e-05,
      "loss": 11.7434,
      "step": 16030
    },
    {
      "epoch": 2.8032349726775956,
      "grad_norm": 0.6753296572713794,
      "learning_rate": 4.336667050269528e-05,
      "loss": 11.7998,
      "step": 16031
    },
    {
      "epoch": 2.803409836065574,
      "grad_norm": 0.7189779124055562,
      "learning_rate": 4.335500000337189e-05,
      "loss": 11.8596,
      "step": 16032
    },
    {
      "epoch": 2.803584699453552,
      "grad_norm": 0.6914764649129501,
      "learning_rate": 4.334333063995464e-05,
      "loss": 11.5278,
      "step": 16033
    },
    {
      "epoch": 2.80375956284153,
      "grad_norm": 0.7056010384804133,
      "learning_rate": 4.3331662412677576e-05,
      "loss": 11.8256,
      "step": 16034
    },
    {
      "epoch": 2.803934426229508,
      "grad_norm": 0.668179366279812,
      "learning_rate": 4.331999532177463e-05,
      "loss": 11.7581,
      "step": 16035
    },
    {
      "epoch": 2.8041092896174864,
      "grad_norm": 0.6894764918446925,
      "learning_rate": 4.330832936747985e-05,
      "loss": 11.7381,
      "step": 16036
    },
    {
      "epoch": 2.8042841530054643,
      "grad_norm": 0.6026919179799658,
      "learning_rate": 4.3296664550027156e-05,
      "loss": 11.9905,
      "step": 16037
    },
    {
      "epoch": 2.8044590163934426,
      "grad_norm": 0.5607856040888998,
      "learning_rate": 4.328500086965042e-05,
      "loss": 11.5049,
      "step": 16038
    },
    {
      "epoch": 2.804633879781421,
      "grad_norm": 0.6645898546091408,
      "learning_rate": 4.327333832658354e-05,
      "loss": 11.6967,
      "step": 16039
    },
    {
      "epoch": 2.804808743169399,
      "grad_norm": 0.6786270271827773,
      "learning_rate": 4.326167692106038e-05,
      "loss": 11.7055,
      "step": 16040
    },
    {
      "epoch": 2.8049836065573768,
      "grad_norm": 0.6298115070804662,
      "learning_rate": 4.3250016653314864e-05,
      "loss": 11.8031,
      "step": 16041
    },
    {
      "epoch": 2.805158469945355,
      "grad_norm": 0.5888590276955561,
      "learning_rate": 4.323835752358073e-05,
      "loss": 11.8443,
      "step": 16042
    },
    {
      "epoch": 2.8053333333333335,
      "grad_norm": 0.6881244792693637,
      "learning_rate": 4.322669953209181e-05,
      "loss": 11.6153,
      "step": 16043
    },
    {
      "epoch": 2.8055081967213114,
      "grad_norm": 0.6870038175072758,
      "learning_rate": 4.3215042679081886e-05,
      "loss": 11.7436,
      "step": 16044
    },
    {
      "epoch": 2.8056830601092897,
      "grad_norm": 0.6362003306555953,
      "learning_rate": 4.3203386964784664e-05,
      "loss": 11.6897,
      "step": 16045
    },
    {
      "epoch": 2.8058579234972676,
      "grad_norm": 0.6528098383328088,
      "learning_rate": 4.319173238943399e-05,
      "loss": 11.7863,
      "step": 16046
    },
    {
      "epoch": 2.806032786885246,
      "grad_norm": 0.637492709318133,
      "learning_rate": 4.318007895326348e-05,
      "loss": 11.8184,
      "step": 16047
    },
    {
      "epoch": 2.8062076502732243,
      "grad_norm": 0.6883213544344918,
      "learning_rate": 4.316842665650684e-05,
      "loss": 11.633,
      "step": 16048
    },
    {
      "epoch": 2.806382513661202,
      "grad_norm": 0.6581385765688886,
      "learning_rate": 4.3156775499397715e-05,
      "loss": 11.8873,
      "step": 16049
    },
    {
      "epoch": 2.80655737704918,
      "grad_norm": 0.623346469227896,
      "learning_rate": 4.314512548216985e-05,
      "loss": 11.6649,
      "step": 16050
    },
    {
      "epoch": 2.8067322404371584,
      "grad_norm": 0.6823071224156502,
      "learning_rate": 4.313347660505673e-05,
      "loss": 11.8502,
      "step": 16051
    },
    {
      "epoch": 2.8069071038251368,
      "grad_norm": 0.6635639816023938,
      "learning_rate": 4.312182886829201e-05,
      "loss": 11.8653,
      "step": 16052
    },
    {
      "epoch": 2.8070819672131146,
      "grad_norm": 0.6314570402283398,
      "learning_rate": 4.311018227210928e-05,
      "loss": 11.78,
      "step": 16053
    },
    {
      "epoch": 2.807256830601093,
      "grad_norm": 0.6674030205464685,
      "learning_rate": 4.309853681674202e-05,
      "loss": 11.7114,
      "step": 16054
    },
    {
      "epoch": 2.807431693989071,
      "grad_norm": 0.7208387313475546,
      "learning_rate": 4.308689250242388e-05,
      "loss": 11.7509,
      "step": 16055
    },
    {
      "epoch": 2.8076065573770492,
      "grad_norm": 0.6280162329646489,
      "learning_rate": 4.3075249329388255e-05,
      "loss": 11.6852,
      "step": 16056
    },
    {
      "epoch": 2.8077814207650276,
      "grad_norm": 0.6094535964305592,
      "learning_rate": 4.306360729786867e-05,
      "loss": 11.6016,
      "step": 16057
    },
    {
      "epoch": 2.8079562841530055,
      "grad_norm": 0.6649468499779553,
      "learning_rate": 4.3051966408098564e-05,
      "loss": 11.7628,
      "step": 16058
    },
    {
      "epoch": 2.8081311475409834,
      "grad_norm": 0.7045805915694695,
      "learning_rate": 4.304032666031139e-05,
      "loss": 12.0319,
      "step": 16059
    },
    {
      "epoch": 2.8083060109289617,
      "grad_norm": 0.6212491158798357,
      "learning_rate": 4.3028688054740565e-05,
      "loss": 11.7953,
      "step": 16060
    },
    {
      "epoch": 2.80848087431694,
      "grad_norm": 0.6415445571139846,
      "learning_rate": 4.3017050591619454e-05,
      "loss": 11.6353,
      "step": 16061
    },
    {
      "epoch": 2.808655737704918,
      "grad_norm": 0.6455089705556682,
      "learning_rate": 4.3005414271181445e-05,
      "loss": 11.7026,
      "step": 16062
    },
    {
      "epoch": 2.8088306010928963,
      "grad_norm": 0.6423912796882381,
      "learning_rate": 4.299377909365988e-05,
      "loss": 11.7665,
      "step": 16063
    },
    {
      "epoch": 2.809005464480874,
      "grad_norm": 0.62263923050478,
      "learning_rate": 4.298214505928807e-05,
      "loss": 11.8303,
      "step": 16064
    },
    {
      "epoch": 2.8091803278688525,
      "grad_norm": 0.6673588848829014,
      "learning_rate": 4.2970512168299325e-05,
      "loss": 11.8837,
      "step": 16065
    },
    {
      "epoch": 2.809355191256831,
      "grad_norm": 0.604767463998557,
      "learning_rate": 4.2958880420926916e-05,
      "loss": 11.7143,
      "step": 16066
    },
    {
      "epoch": 2.8095300546448088,
      "grad_norm": 0.6458977387824754,
      "learning_rate": 4.294724981740409e-05,
      "loss": 11.5151,
      "step": 16067
    },
    {
      "epoch": 2.8097049180327867,
      "grad_norm": 0.6802669957328507,
      "learning_rate": 4.2935620357964076e-05,
      "loss": 11.8407,
      "step": 16068
    },
    {
      "epoch": 2.809879781420765,
      "grad_norm": 0.6357532733860453,
      "learning_rate": 4.29239920428401e-05,
      "loss": 11.8908,
      "step": 16069
    },
    {
      "epoch": 2.8100546448087433,
      "grad_norm": 0.706586256136723,
      "learning_rate": 4.291236487226531e-05,
      "loss": 12.0039,
      "step": 16070
    },
    {
      "epoch": 2.8102295081967212,
      "grad_norm": 0.6394864614920546,
      "learning_rate": 4.29007388464729e-05,
      "loss": 11.639,
      "step": 16071
    },
    {
      "epoch": 2.8104043715846996,
      "grad_norm": 0.6992174972693205,
      "learning_rate": 4.288911396569599e-05,
      "loss": 11.6932,
      "step": 16072
    },
    {
      "epoch": 2.8105792349726775,
      "grad_norm": 0.7519250751908895,
      "learning_rate": 4.2877490230167695e-05,
      "loss": 11.6408,
      "step": 16073
    },
    {
      "epoch": 2.810754098360656,
      "grad_norm": 0.6141489493404069,
      "learning_rate": 4.2865867640121117e-05,
      "loss": 11.6128,
      "step": 16074
    },
    {
      "epoch": 2.8109289617486337,
      "grad_norm": 0.6864047102901644,
      "learning_rate": 4.2854246195789314e-05,
      "loss": 11.7614,
      "step": 16075
    },
    {
      "epoch": 2.811103825136612,
      "grad_norm": 0.6512389845504778,
      "learning_rate": 4.284262589740533e-05,
      "loss": 11.8822,
      "step": 16076
    },
    {
      "epoch": 2.81127868852459,
      "grad_norm": 0.7301458701289026,
      "learning_rate": 4.283100674520219e-05,
      "loss": 11.7942,
      "step": 16077
    },
    {
      "epoch": 2.8114535519125683,
      "grad_norm": 0.646661914611906,
      "learning_rate": 4.281938873941291e-05,
      "loss": 11.9271,
      "step": 16078
    },
    {
      "epoch": 2.8116284153005466,
      "grad_norm": 0.7502372609107288,
      "learning_rate": 4.280777188027045e-05,
      "loss": 11.7322,
      "step": 16079
    },
    {
      "epoch": 2.8118032786885245,
      "grad_norm": 0.6890686540245602,
      "learning_rate": 4.2796156168007774e-05,
      "loss": 11.9887,
      "step": 16080
    },
    {
      "epoch": 2.811978142076503,
      "grad_norm": 0.6194126898364666,
      "learning_rate": 4.2784541602857786e-05,
      "loss": 11.6231,
      "step": 16081
    },
    {
      "epoch": 2.8121530054644808,
      "grad_norm": 0.63244232136307,
      "learning_rate": 4.277292818505343e-05,
      "loss": 11.5727,
      "step": 16082
    },
    {
      "epoch": 2.812327868852459,
      "grad_norm": 0.6414048016257985,
      "learning_rate": 4.276131591482757e-05,
      "loss": 11.6952,
      "step": 16083
    },
    {
      "epoch": 2.812502732240437,
      "grad_norm": 0.6441919215656429,
      "learning_rate": 4.274970479241307e-05,
      "loss": 11.9665,
      "step": 16084
    },
    {
      "epoch": 2.8126775956284154,
      "grad_norm": 0.6572790100244915,
      "learning_rate": 4.2738094818042764e-05,
      "loss": 11.8591,
      "step": 16085
    },
    {
      "epoch": 2.8128524590163932,
      "grad_norm": 0.6460907453316204,
      "learning_rate": 4.272648599194948e-05,
      "loss": 11.8236,
      "step": 16086
    },
    {
      "epoch": 2.8130273224043716,
      "grad_norm": 0.6655767874905911,
      "learning_rate": 4.271487831436601e-05,
      "loss": 11.9945,
      "step": 16087
    },
    {
      "epoch": 2.81320218579235,
      "grad_norm": 0.6221695290926712,
      "learning_rate": 4.270327178552511e-05,
      "loss": 11.6783,
      "step": 16088
    },
    {
      "epoch": 2.813377049180328,
      "grad_norm": 0.6189128489330801,
      "learning_rate": 4.269166640565954e-05,
      "loss": 11.8687,
      "step": 16089
    },
    {
      "epoch": 2.813551912568306,
      "grad_norm": 0.6580476659345169,
      "learning_rate": 4.268006217500201e-05,
      "loss": 11.7232,
      "step": 16090
    },
    {
      "epoch": 2.813726775956284,
      "grad_norm": 0.6531418396896571,
      "learning_rate": 4.266845909378524e-05,
      "loss": 11.8423,
      "step": 16091
    },
    {
      "epoch": 2.8139016393442624,
      "grad_norm": 0.6583302641187578,
      "learning_rate": 4.265685716224193e-05,
      "loss": 11.5264,
      "step": 16092
    },
    {
      "epoch": 2.8140765027322403,
      "grad_norm": 0.6414570399844999,
      "learning_rate": 4.264525638060463e-05,
      "loss": 11.7328,
      "step": 16093
    },
    {
      "epoch": 2.8142513661202186,
      "grad_norm": 0.6045339366053031,
      "learning_rate": 4.263365674910606e-05,
      "loss": 11.7837,
      "step": 16094
    },
    {
      "epoch": 2.8144262295081965,
      "grad_norm": 0.6315618248759649,
      "learning_rate": 4.262205826797883e-05,
      "loss": 11.616,
      "step": 16095
    },
    {
      "epoch": 2.814601092896175,
      "grad_norm": 0.6558538005678631,
      "learning_rate": 4.26104609374555e-05,
      "loss": 11.4175,
      "step": 16096
    },
    {
      "epoch": 2.8147759562841532,
      "grad_norm": 0.630541076453954,
      "learning_rate": 4.259886475776863e-05,
      "loss": 11.7224,
      "step": 16097
    },
    {
      "epoch": 2.814950819672131,
      "grad_norm": 0.611595617720985,
      "learning_rate": 4.258726972915077e-05,
      "loss": 11.6612,
      "step": 16098
    },
    {
      "epoch": 2.815125683060109,
      "grad_norm": 0.6229642884601446,
      "learning_rate": 4.257567585183444e-05,
      "loss": 12.1485,
      "step": 16099
    },
    {
      "epoch": 2.8153005464480874,
      "grad_norm": 0.6587089378052978,
      "learning_rate": 4.256408312605211e-05,
      "loss": 11.7917,
      "step": 16100
    },
    {
      "epoch": 2.8154754098360657,
      "grad_norm": 0.7604855059727491,
      "learning_rate": 4.255249155203632e-05,
      "loss": 11.6521,
      "step": 16101
    },
    {
      "epoch": 2.8156502732240436,
      "grad_norm": 0.6088875373161255,
      "learning_rate": 4.2540901130019375e-05,
      "loss": 11.6742,
      "step": 16102
    },
    {
      "epoch": 2.815825136612022,
      "grad_norm": 0.6252650563008783,
      "learning_rate": 4.252931186023382e-05,
      "loss": 11.6026,
      "step": 16103
    },
    {
      "epoch": 2.816,
      "grad_norm": 0.6409548383577703,
      "learning_rate": 4.251772374291203e-05,
      "loss": 11.8217,
      "step": 16104
    },
    {
      "epoch": 2.816174863387978,
      "grad_norm": 0.6746586867152456,
      "learning_rate": 4.2506136778286365e-05,
      "loss": 11.5247,
      "step": 16105
    },
    {
      "epoch": 2.8163497267759565,
      "grad_norm": 0.6354536112375337,
      "learning_rate": 4.249455096658923e-05,
      "loss": 11.7435,
      "step": 16106
    },
    {
      "epoch": 2.8165245901639344,
      "grad_norm": 0.6492988195108684,
      "learning_rate": 4.248296630805284e-05,
      "loss": 11.6468,
      "step": 16107
    },
    {
      "epoch": 2.8166994535519123,
      "grad_norm": 0.6513518175762475,
      "learning_rate": 4.247138280290961e-05,
      "loss": 11.8338,
      "step": 16108
    },
    {
      "epoch": 2.8168743169398907,
      "grad_norm": 0.6101772190331989,
      "learning_rate": 4.245980045139178e-05,
      "loss": 11.7071,
      "step": 16109
    },
    {
      "epoch": 2.817049180327869,
      "grad_norm": 0.7204746350949558,
      "learning_rate": 4.244821925373168e-05,
      "loss": 11.9649,
      "step": 16110
    },
    {
      "epoch": 2.817224043715847,
      "grad_norm": 0.6545856400451351,
      "learning_rate": 4.2436639210161444e-05,
      "loss": 11.7094,
      "step": 16111
    },
    {
      "epoch": 2.8173989071038252,
      "grad_norm": 0.695160417418621,
      "learning_rate": 4.242506032091329e-05,
      "loss": 11.4724,
      "step": 16112
    },
    {
      "epoch": 2.817573770491803,
      "grad_norm": 0.6666155259313798,
      "learning_rate": 4.24134825862195e-05,
      "loss": 11.7107,
      "step": 16113
    },
    {
      "epoch": 2.8177486338797815,
      "grad_norm": 0.6919890586593532,
      "learning_rate": 4.24019060063122e-05,
      "loss": 11.902,
      "step": 16114
    },
    {
      "epoch": 2.81792349726776,
      "grad_norm": 0.6316853700707264,
      "learning_rate": 4.239033058142356e-05,
      "loss": 11.7969,
      "step": 16115
    },
    {
      "epoch": 2.8180983606557377,
      "grad_norm": 0.7051462888987224,
      "learning_rate": 4.237875631178565e-05,
      "loss": 11.8558,
      "step": 16116
    },
    {
      "epoch": 2.8182732240437156,
      "grad_norm": 0.7170479744946988,
      "learning_rate": 4.2367183197630564e-05,
      "loss": 11.5897,
      "step": 16117
    },
    {
      "epoch": 2.818448087431694,
      "grad_norm": 0.6465911920056951,
      "learning_rate": 4.235561123919044e-05,
      "loss": 11.8241,
      "step": 16118
    },
    {
      "epoch": 2.8186229508196723,
      "grad_norm": 0.6337373429764815,
      "learning_rate": 4.234404043669734e-05,
      "loss": 11.8045,
      "step": 16119
    },
    {
      "epoch": 2.81879781420765,
      "grad_norm": 0.6397004782632072,
      "learning_rate": 4.2332470790383214e-05,
      "loss": 11.9425,
      "step": 16120
    },
    {
      "epoch": 2.8189726775956285,
      "grad_norm": 0.7159196898367639,
      "learning_rate": 4.232090230048008e-05,
      "loss": 11.9198,
      "step": 16121
    },
    {
      "epoch": 2.8191475409836064,
      "grad_norm": 0.6537209518553404,
      "learning_rate": 4.2309334967219995e-05,
      "loss": 11.3162,
      "step": 16122
    },
    {
      "epoch": 2.8193224043715848,
      "grad_norm": 0.6567896356844686,
      "learning_rate": 4.229776879083487e-05,
      "loss": 11.6939,
      "step": 16123
    },
    {
      "epoch": 2.819497267759563,
      "grad_norm": 0.6586667666918701,
      "learning_rate": 4.2286203771556684e-05,
      "loss": 11.5187,
      "step": 16124
    },
    {
      "epoch": 2.819672131147541,
      "grad_norm": 0.7232985646721698,
      "learning_rate": 4.2274639909617295e-05,
      "loss": 11.7399,
      "step": 16125
    },
    {
      "epoch": 2.819846994535519,
      "grad_norm": 0.6270771671182567,
      "learning_rate": 4.226307720524857e-05,
      "loss": 11.5724,
      "step": 16126
    },
    {
      "epoch": 2.8200218579234972,
      "grad_norm": 0.6504998899366887,
      "learning_rate": 4.225151565868246e-05,
      "loss": 11.7676,
      "step": 16127
    },
    {
      "epoch": 2.8201967213114756,
      "grad_norm": 0.6645243761730307,
      "learning_rate": 4.22399552701508e-05,
      "loss": 11.6409,
      "step": 16128
    },
    {
      "epoch": 2.8203715846994535,
      "grad_norm": 0.6857943427288404,
      "learning_rate": 4.222839603988535e-05,
      "loss": 11.762,
      "step": 16129
    },
    {
      "epoch": 2.820546448087432,
      "grad_norm": 0.646539672037311,
      "learning_rate": 4.2216837968117954e-05,
      "loss": 11.6924,
      "step": 16130
    },
    {
      "epoch": 2.8207213114754097,
      "grad_norm": 0.6886557462131574,
      "learning_rate": 4.2205281055080325e-05,
      "loss": 11.6373,
      "step": 16131
    },
    {
      "epoch": 2.820896174863388,
      "grad_norm": 0.6871502152640191,
      "learning_rate": 4.219372530100431e-05,
      "loss": 11.8721,
      "step": 16132
    },
    {
      "epoch": 2.821071038251366,
      "grad_norm": 0.6186258376542925,
      "learning_rate": 4.2182170706121626e-05,
      "loss": 11.7655,
      "step": 16133
    },
    {
      "epoch": 2.8212459016393443,
      "grad_norm": 0.706586607982707,
      "learning_rate": 4.21706172706639e-05,
      "loss": 11.7705,
      "step": 16134
    },
    {
      "epoch": 2.821420765027322,
      "grad_norm": 0.7736371377398912,
      "learning_rate": 4.2159064994862866e-05,
      "loss": 11.703,
      "step": 16135
    },
    {
      "epoch": 2.8215956284153005,
      "grad_norm": 0.6285935212877098,
      "learning_rate": 4.214751387895013e-05,
      "loss": 11.7272,
      "step": 16136
    },
    {
      "epoch": 2.821770491803279,
      "grad_norm": 0.6488243723081649,
      "learning_rate": 4.213596392315744e-05,
      "loss": 11.5017,
      "step": 16137
    },
    {
      "epoch": 2.8219453551912568,
      "grad_norm": 0.6292073751914417,
      "learning_rate": 4.212441512771631e-05,
      "loss": 11.6376,
      "step": 16138
    },
    {
      "epoch": 2.822120218579235,
      "grad_norm": 0.6178655606341074,
      "learning_rate": 4.211286749285836e-05,
      "loss": 11.5237,
      "step": 16139
    },
    {
      "epoch": 2.822295081967213,
      "grad_norm": 0.7100272663905559,
      "learning_rate": 4.210132101881516e-05,
      "loss": 11.6229,
      "step": 16140
    },
    {
      "epoch": 2.8224699453551914,
      "grad_norm": 0.5803086361229093,
      "learning_rate": 4.208977570581819e-05,
      "loss": 11.9478,
      "step": 16141
    },
    {
      "epoch": 2.8226448087431693,
      "grad_norm": 0.5853773401263045,
      "learning_rate": 4.207823155409911e-05,
      "loss": 11.838,
      "step": 16142
    },
    {
      "epoch": 2.8228196721311476,
      "grad_norm": 0.6235765532603834,
      "learning_rate": 4.206668856388929e-05,
      "loss": 11.9134,
      "step": 16143
    },
    {
      "epoch": 2.8229945355191255,
      "grad_norm": 0.6513413798182554,
      "learning_rate": 4.2055146735420245e-05,
      "loss": 12.021,
      "step": 16144
    },
    {
      "epoch": 2.823169398907104,
      "grad_norm": 0.6280316481192588,
      "learning_rate": 4.204360606892339e-05,
      "loss": 11.8189,
      "step": 16145
    },
    {
      "epoch": 2.823344262295082,
      "grad_norm": 0.6587964652087018,
      "learning_rate": 4.203206656463026e-05,
      "loss": 11.8246,
      "step": 16146
    },
    {
      "epoch": 2.82351912568306,
      "grad_norm": 0.6594529754244287,
      "learning_rate": 4.2020528222772123e-05,
      "loss": 11.802,
      "step": 16147
    },
    {
      "epoch": 2.8236939890710384,
      "grad_norm": 0.9811792640403898,
      "learning_rate": 4.2008991043580434e-05,
      "loss": 12.0606,
      "step": 16148
    },
    {
      "epoch": 2.8238688524590163,
      "grad_norm": 0.6280899759852382,
      "learning_rate": 4.1997455027286525e-05,
      "loss": 11.8738,
      "step": 16149
    },
    {
      "epoch": 2.8240437158469947,
      "grad_norm": 0.6480640040168751,
      "learning_rate": 4.19859201741217e-05,
      "loss": 11.7618,
      "step": 16150
    },
    {
      "epoch": 2.8242185792349725,
      "grad_norm": 0.6482696295621495,
      "learning_rate": 4.1974386484317363e-05,
      "loss": 11.7338,
      "step": 16151
    },
    {
      "epoch": 2.824393442622951,
      "grad_norm": 0.6265988610240475,
      "learning_rate": 4.196285395810472e-05,
      "loss": 11.7576,
      "step": 16152
    },
    {
      "epoch": 2.824568306010929,
      "grad_norm": 0.6939207413273584,
      "learning_rate": 4.195132259571504e-05,
      "loss": 11.7763,
      "step": 16153
    },
    {
      "epoch": 2.824743169398907,
      "grad_norm": 0.7165499061508829,
      "learning_rate": 4.193979239737957e-05,
      "loss": 11.8807,
      "step": 16154
    },
    {
      "epoch": 2.8249180327868855,
      "grad_norm": 0.7071655566653058,
      "learning_rate": 4.1928263363329547e-05,
      "loss": 11.5826,
      "step": 16155
    },
    {
      "epoch": 2.8250928961748634,
      "grad_norm": 0.6715422797811447,
      "learning_rate": 4.1916735493796146e-05,
      "loss": 11.8203,
      "step": 16156
    },
    {
      "epoch": 2.8252677595628413,
      "grad_norm": 0.6575108135347161,
      "learning_rate": 4.190520878901052e-05,
      "loss": 11.783,
      "step": 16157
    },
    {
      "epoch": 2.8254426229508196,
      "grad_norm": 0.7099362190456362,
      "learning_rate": 4.189368324920385e-05,
      "loss": 11.9673,
      "step": 16158
    },
    {
      "epoch": 2.825617486338798,
      "grad_norm": 0.6264379017143978,
      "learning_rate": 4.188215887460723e-05,
      "loss": 11.5227,
      "step": 16159
    },
    {
      "epoch": 2.825792349726776,
      "grad_norm": 0.5891998421713939,
      "learning_rate": 4.187063566545177e-05,
      "loss": 11.8374,
      "step": 16160
    },
    {
      "epoch": 2.825967213114754,
      "grad_norm": 0.6824289903663184,
      "learning_rate": 4.185911362196855e-05,
      "loss": 11.9146,
      "step": 16161
    },
    {
      "epoch": 2.826142076502732,
      "grad_norm": 0.6900123109354289,
      "learning_rate": 4.184759274438862e-05,
      "loss": 11.7553,
      "step": 16162
    },
    {
      "epoch": 2.8263169398907104,
      "grad_norm": 0.6666156051246214,
      "learning_rate": 4.1836073032943e-05,
      "loss": 11.8372,
      "step": 16163
    },
    {
      "epoch": 2.8264918032786888,
      "grad_norm": 0.6445596029373358,
      "learning_rate": 4.182455448786271e-05,
      "loss": 11.8747,
      "step": 16164
    },
    {
      "epoch": 2.8266666666666667,
      "grad_norm": 0.6084453940358961,
      "learning_rate": 4.1813037109378715e-05,
      "loss": 11.9194,
      "step": 16165
    },
    {
      "epoch": 2.8268415300546446,
      "grad_norm": 0.6475570985383483,
      "learning_rate": 4.180152089772199e-05,
      "loss": 11.6916,
      "step": 16166
    },
    {
      "epoch": 2.827016393442623,
      "grad_norm": 0.6360171330719513,
      "learning_rate": 4.179000585312347e-05,
      "loss": 11.9238,
      "step": 16167
    },
    {
      "epoch": 2.8271912568306012,
      "grad_norm": 0.659210457397447,
      "learning_rate": 4.177849197581405e-05,
      "loss": 12.0501,
      "step": 16168
    },
    {
      "epoch": 2.827366120218579,
      "grad_norm": 0.6023130081871548,
      "learning_rate": 4.176697926602464e-05,
      "loss": 11.6723,
      "step": 16169
    },
    {
      "epoch": 2.8275409836065575,
      "grad_norm": 0.6682207601181344,
      "learning_rate": 4.1755467723986084e-05,
      "loss": 11.8153,
      "step": 16170
    },
    {
      "epoch": 2.8277158469945354,
      "grad_norm": 0.6008922753537823,
      "learning_rate": 4.174395734992924e-05,
      "loss": 11.6948,
      "step": 16171
    },
    {
      "epoch": 2.8278907103825137,
      "grad_norm": 0.6415980108969143,
      "learning_rate": 4.173244814408491e-05,
      "loss": 11.7522,
      "step": 16172
    },
    {
      "epoch": 2.828065573770492,
      "grad_norm": 0.6006247263326878,
      "learning_rate": 4.172094010668391e-05,
      "loss": 12.0164,
      "step": 16173
    },
    {
      "epoch": 2.82824043715847,
      "grad_norm": 0.6112562874811973,
      "learning_rate": 4.170943323795701e-05,
      "loss": 11.9678,
      "step": 16174
    },
    {
      "epoch": 2.828415300546448,
      "grad_norm": 0.6961333944538841,
      "learning_rate": 4.1697927538134925e-05,
      "loss": 11.9008,
      "step": 16175
    },
    {
      "epoch": 2.828590163934426,
      "grad_norm": 0.6058192342750152,
      "learning_rate": 4.1686423007448426e-05,
      "loss": 11.6543,
      "step": 16176
    },
    {
      "epoch": 2.8287650273224045,
      "grad_norm": 0.6634235690135871,
      "learning_rate": 4.1674919646128186e-05,
      "loss": 11.853,
      "step": 16177
    },
    {
      "epoch": 2.8289398907103824,
      "grad_norm": 0.7041812204394554,
      "learning_rate": 4.166341745440488e-05,
      "loss": 11.8025,
      "step": 16178
    },
    {
      "epoch": 2.8291147540983608,
      "grad_norm": 0.6195113543555141,
      "learning_rate": 4.1651916432509176e-05,
      "loss": 11.7338,
      "step": 16179
    },
    {
      "epoch": 2.8292896174863387,
      "grad_norm": 0.6399162172395472,
      "learning_rate": 4.1640416580671706e-05,
      "loss": 11.8976,
      "step": 16180
    },
    {
      "epoch": 2.829464480874317,
      "grad_norm": 0.6716053351031621,
      "learning_rate": 4.1628917899123065e-05,
      "loss": 11.9037,
      "step": 16181
    },
    {
      "epoch": 2.8296393442622954,
      "grad_norm": 0.6585264844635735,
      "learning_rate": 4.161742038809384e-05,
      "loss": 11.3936,
      "step": 16182
    },
    {
      "epoch": 2.8298142076502733,
      "grad_norm": 0.6199860519260577,
      "learning_rate": 4.16059240478146e-05,
      "loss": 11.6663,
      "step": 16183
    },
    {
      "epoch": 2.829989071038251,
      "grad_norm": 0.5979561726905301,
      "learning_rate": 4.159442887851588e-05,
      "loss": 11.565,
      "step": 16184
    },
    {
      "epoch": 2.8301639344262295,
      "grad_norm": 0.6071747135694165,
      "learning_rate": 4.158293488042818e-05,
      "loss": 11.864,
      "step": 16185
    },
    {
      "epoch": 2.830338797814208,
      "grad_norm": 0.6862568713638284,
      "learning_rate": 4.157144205378199e-05,
      "loss": 11.8255,
      "step": 16186
    },
    {
      "epoch": 2.8305136612021857,
      "grad_norm": 0.6291363773282029,
      "learning_rate": 4.155995039880781e-05,
      "loss": 11.7293,
      "step": 16187
    },
    {
      "epoch": 2.830688524590164,
      "grad_norm": 0.6213179521323378,
      "learning_rate": 4.154845991573608e-05,
      "loss": 12.0744,
      "step": 16188
    },
    {
      "epoch": 2.830863387978142,
      "grad_norm": 0.6270577475466254,
      "learning_rate": 4.1536970604797134e-05,
      "loss": 11.7601,
      "step": 16189
    },
    {
      "epoch": 2.8310382513661203,
      "grad_norm": 0.6424071827610851,
      "learning_rate": 4.152548246622147e-05,
      "loss": 11.8301,
      "step": 16190
    },
    {
      "epoch": 2.831213114754098,
      "grad_norm": 0.6531636132187761,
      "learning_rate": 4.151399550023941e-05,
      "loss": 11.888,
      "step": 16191
    },
    {
      "epoch": 2.8313879781420765,
      "grad_norm": 0.6382854082817906,
      "learning_rate": 4.150250970708133e-05,
      "loss": 11.8539,
      "step": 16192
    },
    {
      "epoch": 2.8315628415300544,
      "grad_norm": 0.6596085012820064,
      "learning_rate": 4.149102508697753e-05,
      "loss": 11.7563,
      "step": 16193
    },
    {
      "epoch": 2.831737704918033,
      "grad_norm": 0.6683193997481073,
      "learning_rate": 4.147954164015832e-05,
      "loss": 11.4539,
      "step": 16194
    },
    {
      "epoch": 2.831912568306011,
      "grad_norm": 0.6981962739765225,
      "learning_rate": 4.146805936685399e-05,
      "loss": 11.6578,
      "step": 16195
    },
    {
      "epoch": 2.832087431693989,
      "grad_norm": 0.6311993663021123,
      "learning_rate": 4.14565782672948e-05,
      "loss": 11.6891,
      "step": 16196
    },
    {
      "epoch": 2.8322622950819674,
      "grad_norm": 0.6938115567533243,
      "learning_rate": 4.144509834171099e-05,
      "loss": 11.8191,
      "step": 16197
    },
    {
      "epoch": 2.8324371584699453,
      "grad_norm": 0.6878898247708514,
      "learning_rate": 4.1433619590332675e-05,
      "loss": 11.902,
      "step": 16198
    },
    {
      "epoch": 2.8326120218579236,
      "grad_norm": 0.6126327087392652,
      "learning_rate": 4.142214201339014e-05,
      "loss": 11.6564,
      "step": 16199
    },
    {
      "epoch": 2.8327868852459015,
      "grad_norm": 0.598710758151943,
      "learning_rate": 4.141066561111352e-05,
      "loss": 11.6531,
      "step": 16200
    },
    {
      "epoch": 2.83296174863388,
      "grad_norm": 0.6001187481227942,
      "learning_rate": 4.139919038373294e-05,
      "loss": 11.882,
      "step": 16201
    },
    {
      "epoch": 2.8331366120218577,
      "grad_norm": 0.7186799420014564,
      "learning_rate": 4.1387716331478565e-05,
      "loss": 11.8837,
      "step": 16202
    },
    {
      "epoch": 2.833311475409836,
      "grad_norm": 0.6408794896932323,
      "learning_rate": 4.1376243454580366e-05,
      "loss": 11.7397,
      "step": 16203
    },
    {
      "epoch": 2.8334863387978144,
      "grad_norm": 0.6492172355713747,
      "learning_rate": 4.136477175326853e-05,
      "loss": 11.8738,
      "step": 16204
    },
    {
      "epoch": 2.8336612021857923,
      "grad_norm": 0.6849843111311997,
      "learning_rate": 4.135330122777305e-05,
      "loss": 11.8792,
      "step": 16205
    },
    {
      "epoch": 2.8338360655737707,
      "grad_norm": 0.6585509809316055,
      "learning_rate": 4.1341831878323967e-05,
      "loss": 11.6408,
      "step": 16206
    },
    {
      "epoch": 2.8340109289617486,
      "grad_norm": 0.6372076154102072,
      "learning_rate": 4.133036370515124e-05,
      "loss": 11.8008,
      "step": 16207
    },
    {
      "epoch": 2.834185792349727,
      "grad_norm": 0.6665641634856537,
      "learning_rate": 4.131889670848482e-05,
      "loss": 11.8594,
      "step": 16208
    },
    {
      "epoch": 2.834360655737705,
      "grad_norm": 0.6539176484798684,
      "learning_rate": 4.130743088855473e-05,
      "loss": 11.7737,
      "step": 16209
    },
    {
      "epoch": 2.834535519125683,
      "grad_norm": 0.6120374774535495,
      "learning_rate": 4.1295966245590867e-05,
      "loss": 11.7444,
      "step": 16210
    },
    {
      "epoch": 2.834710382513661,
      "grad_norm": 0.6641178142860016,
      "learning_rate": 4.1284502779823166e-05,
      "loss": 11.6465,
      "step": 16211
    },
    {
      "epoch": 2.8348852459016394,
      "grad_norm": 0.6435148896539443,
      "learning_rate": 4.127304049148142e-05,
      "loss": 11.8732,
      "step": 16212
    },
    {
      "epoch": 2.8350601092896177,
      "grad_norm": 0.6371713233669976,
      "learning_rate": 4.126157938079549e-05,
      "loss": 11.9474,
      "step": 16213
    },
    {
      "epoch": 2.8352349726775956,
      "grad_norm": 0.6903723442479754,
      "learning_rate": 4.1250119447995284e-05,
      "loss": 11.7008,
      "step": 16214
    },
    {
      "epoch": 2.8354098360655735,
      "grad_norm": 0.6263567920764647,
      "learning_rate": 4.12386606933106e-05,
      "loss": 11.6772,
      "step": 16215
    },
    {
      "epoch": 2.835584699453552,
      "grad_norm": 0.7950311362027216,
      "learning_rate": 4.1227203116971156e-05,
      "loss": 12.012,
      "step": 16216
    },
    {
      "epoch": 2.83575956284153,
      "grad_norm": 0.7276250512470692,
      "learning_rate": 4.1215746719206716e-05,
      "loss": 11.5863,
      "step": 16217
    },
    {
      "epoch": 2.835934426229508,
      "grad_norm": 0.7158803271704693,
      "learning_rate": 4.120429150024707e-05,
      "loss": 11.6514,
      "step": 16218
    },
    {
      "epoch": 2.8361092896174864,
      "grad_norm": 0.6105438264692323,
      "learning_rate": 4.119283746032192e-05,
      "loss": 11.7324,
      "step": 16219
    },
    {
      "epoch": 2.8362841530054643,
      "grad_norm": 0.6940392091918448,
      "learning_rate": 4.118138459966097e-05,
      "loss": 11.6069,
      "step": 16220
    },
    {
      "epoch": 2.8364590163934427,
      "grad_norm": 0.656492603079563,
      "learning_rate": 4.116993291849381e-05,
      "loss": 11.8655,
      "step": 16221
    },
    {
      "epoch": 2.836633879781421,
      "grad_norm": 0.6474295246500527,
      "learning_rate": 4.11584824170501e-05,
      "loss": 11.9142,
      "step": 16222
    },
    {
      "epoch": 2.836808743169399,
      "grad_norm": 0.6511794915244612,
      "learning_rate": 4.114703309555953e-05,
      "loss": 11.7455,
      "step": 16223
    },
    {
      "epoch": 2.836983606557377,
      "grad_norm": 0.6830935235372066,
      "learning_rate": 4.1135584954251675e-05,
      "loss": 11.7595,
      "step": 16224
    },
    {
      "epoch": 2.837158469945355,
      "grad_norm": 0.6451584795240932,
      "learning_rate": 4.1124137993356035e-05,
      "loss": 11.8458,
      "step": 16225
    },
    {
      "epoch": 2.8373333333333335,
      "grad_norm": 0.6855451478735718,
      "learning_rate": 4.111269221310221e-05,
      "loss": 11.6938,
      "step": 16226
    },
    {
      "epoch": 2.8375081967213114,
      "grad_norm": 0.7404742342304588,
      "learning_rate": 4.110124761371969e-05,
      "loss": 11.8497,
      "step": 16227
    },
    {
      "epoch": 2.8376830601092897,
      "grad_norm": 0.7177599874606986,
      "learning_rate": 4.108980419543801e-05,
      "loss": 11.696,
      "step": 16228
    },
    {
      "epoch": 2.8378579234972676,
      "grad_norm": 0.6412197311618326,
      "learning_rate": 4.107836195848669e-05,
      "loss": 11.461,
      "step": 16229
    },
    {
      "epoch": 2.838032786885246,
      "grad_norm": 0.672888304958149,
      "learning_rate": 4.1066920903095076e-05,
      "loss": 11.7381,
      "step": 16230
    },
    {
      "epoch": 2.8382076502732243,
      "grad_norm": 0.6389571435935335,
      "learning_rate": 4.1055481029492645e-05,
      "loss": 11.9453,
      "step": 16231
    },
    {
      "epoch": 2.838382513661202,
      "grad_norm": 0.62275499385071,
      "learning_rate": 4.104404233790878e-05,
      "loss": 11.9335,
      "step": 16232
    },
    {
      "epoch": 2.83855737704918,
      "grad_norm": 0.6186498046471517,
      "learning_rate": 4.103260482857295e-05,
      "loss": 11.8297,
      "step": 16233
    },
    {
      "epoch": 2.8387322404371584,
      "grad_norm": 0.6176459097975866,
      "learning_rate": 4.102116850171441e-05,
      "loss": 11.7761,
      "step": 16234
    },
    {
      "epoch": 2.838907103825137,
      "grad_norm": 0.6789864150052585,
      "learning_rate": 4.100973335756253e-05,
      "loss": 11.4461,
      "step": 16235
    },
    {
      "epoch": 2.8390819672131147,
      "grad_norm": 0.5611477286444294,
      "learning_rate": 4.099829939634663e-05,
      "loss": 11.8741,
      "step": 16236
    },
    {
      "epoch": 2.839256830601093,
      "grad_norm": 0.6383132422688235,
      "learning_rate": 4.098686661829594e-05,
      "loss": 11.8687,
      "step": 16237
    },
    {
      "epoch": 2.839431693989071,
      "grad_norm": 0.646708956905714,
      "learning_rate": 4.097543502363984e-05,
      "loss": 11.6383,
      "step": 16238
    },
    {
      "epoch": 2.8396065573770493,
      "grad_norm": 0.6743401252879155,
      "learning_rate": 4.0964004612607465e-05,
      "loss": 11.9492,
      "step": 16239
    },
    {
      "epoch": 2.8397814207650276,
      "grad_norm": 0.7268575072395493,
      "learning_rate": 4.095257538542806e-05,
      "loss": 11.838,
      "step": 16240
    },
    {
      "epoch": 2.8399562841530055,
      "grad_norm": 0.5944527450541471,
      "learning_rate": 4.094114734233078e-05,
      "loss": 11.8125,
      "step": 16241
    },
    {
      "epoch": 2.8401311475409834,
      "grad_norm": 0.6528218364531396,
      "learning_rate": 4.092972048354491e-05,
      "loss": 11.8844,
      "step": 16242
    },
    {
      "epoch": 2.8403060109289617,
      "grad_norm": 0.6594420734980307,
      "learning_rate": 4.091829480929946e-05,
      "loss": 11.924,
      "step": 16243
    },
    {
      "epoch": 2.84048087431694,
      "grad_norm": 0.6729067929443469,
      "learning_rate": 4.090687031982362e-05,
      "loss": 11.5831,
      "step": 16244
    },
    {
      "epoch": 2.840655737704918,
      "grad_norm": 0.6855666991113594,
      "learning_rate": 4.0895447015346465e-05,
      "loss": 11.6599,
      "step": 16245
    },
    {
      "epoch": 2.8408306010928963,
      "grad_norm": 0.5906130249381241,
      "learning_rate": 4.088402489609704e-05,
      "loss": 11.3741,
      "step": 16246
    },
    {
      "epoch": 2.841005464480874,
      "grad_norm": 0.6884912454833726,
      "learning_rate": 4.087260396230449e-05,
      "loss": 11.8307,
      "step": 16247
    },
    {
      "epoch": 2.8411803278688526,
      "grad_norm": 0.6516002735262902,
      "learning_rate": 4.086118421419774e-05,
      "loss": 11.5634,
      "step": 16248
    },
    {
      "epoch": 2.8413551912568304,
      "grad_norm": 0.6105830951126631,
      "learning_rate": 4.0849765652005834e-05,
      "loss": 11.8468,
      "step": 16249
    },
    {
      "epoch": 2.841530054644809,
      "grad_norm": 0.7271066330351419,
      "learning_rate": 4.083834827595774e-05,
      "loss": 11.5466,
      "step": 16250
    },
    {
      "epoch": 2.8417049180327867,
      "grad_norm": 0.6557490612641753,
      "learning_rate": 4.082693208628241e-05,
      "loss": 11.7447,
      "step": 16251
    },
    {
      "epoch": 2.841879781420765,
      "grad_norm": 0.6143941523998756,
      "learning_rate": 4.0815517083208785e-05,
      "loss": 11.8949,
      "step": 16252
    },
    {
      "epoch": 2.8420546448087434,
      "grad_norm": 0.7395347197570223,
      "learning_rate": 4.080410326696576e-05,
      "loss": 11.6172,
      "step": 16253
    },
    {
      "epoch": 2.8422295081967213,
      "grad_norm": 0.6296108892705887,
      "learning_rate": 4.0792690637782205e-05,
      "loss": 11.5876,
      "step": 16254
    },
    {
      "epoch": 2.8424043715846996,
      "grad_norm": 0.7357035389790474,
      "learning_rate": 4.078127919588701e-05,
      "loss": 11.8591,
      "step": 16255
    },
    {
      "epoch": 2.8425792349726775,
      "grad_norm": 0.6311361370159456,
      "learning_rate": 4.076986894150898e-05,
      "loss": 11.9811,
      "step": 16256
    },
    {
      "epoch": 2.842754098360656,
      "grad_norm": 0.6681361421663855,
      "learning_rate": 4.0758459874876954e-05,
      "loss": 11.8304,
      "step": 16257
    },
    {
      "epoch": 2.8429289617486337,
      "grad_norm": 0.7099823201974699,
      "learning_rate": 4.074705199621969e-05,
      "loss": 11.7349,
      "step": 16258
    },
    {
      "epoch": 2.843103825136612,
      "grad_norm": 0.7157977627019411,
      "learning_rate": 4.073564530576597e-05,
      "loss": 11.8964,
      "step": 16259
    },
    {
      "epoch": 2.84327868852459,
      "grad_norm": 0.6039039778903663,
      "learning_rate": 4.072423980374452e-05,
      "loss": 11.8895,
      "step": 16260
    },
    {
      "epoch": 2.8434535519125683,
      "grad_norm": 0.6295211004325462,
      "learning_rate": 4.071283549038407e-05,
      "loss": 11.7265,
      "step": 16261
    },
    {
      "epoch": 2.8436284153005467,
      "grad_norm": 0.774995478583581,
      "learning_rate": 4.07014323659133e-05,
      "loss": 11.8357,
      "step": 16262
    },
    {
      "epoch": 2.8438032786885246,
      "grad_norm": 0.6311579660438313,
      "learning_rate": 4.0690030430560874e-05,
      "loss": 11.6346,
      "step": 16263
    },
    {
      "epoch": 2.8439781420765025,
      "grad_norm": 0.6540654693661826,
      "learning_rate": 4.067862968455546e-05,
      "loss": 11.8318,
      "step": 16264
    },
    {
      "epoch": 2.844153005464481,
      "grad_norm": 0.6638999514539191,
      "learning_rate": 4.066723012812566e-05,
      "loss": 11.7864,
      "step": 16265
    },
    {
      "epoch": 2.844327868852459,
      "grad_norm": 0.7076015498150278,
      "learning_rate": 4.065583176150005e-05,
      "loss": 11.9842,
      "step": 16266
    },
    {
      "epoch": 2.844502732240437,
      "grad_norm": 0.6906909479065237,
      "learning_rate": 4.0644434584907244e-05,
      "loss": 11.9118,
      "step": 16267
    },
    {
      "epoch": 2.8446775956284154,
      "grad_norm": 0.6941822927577007,
      "learning_rate": 4.063303859857577e-05,
      "loss": 11.912,
      "step": 16268
    },
    {
      "epoch": 2.8448524590163933,
      "grad_norm": 0.6478145176258906,
      "learning_rate": 4.062164380273414e-05,
      "loss": 11.8578,
      "step": 16269
    },
    {
      "epoch": 2.8450273224043716,
      "grad_norm": 0.607311945527539,
      "learning_rate": 4.061025019761088e-05,
      "loss": 11.9987,
      "step": 16270
    },
    {
      "epoch": 2.84520218579235,
      "grad_norm": 0.6250007672204781,
      "learning_rate": 4.059885778343444e-05,
      "loss": 11.9386,
      "step": 16271
    },
    {
      "epoch": 2.845377049180328,
      "grad_norm": 0.6843344963393575,
      "learning_rate": 4.0587466560433296e-05,
      "loss": 11.8959,
      "step": 16272
    },
    {
      "epoch": 2.8455519125683058,
      "grad_norm": 0.6770175163694354,
      "learning_rate": 4.057607652883587e-05,
      "loss": 11.9352,
      "step": 16273
    },
    {
      "epoch": 2.845726775956284,
      "grad_norm": 0.6581840221717248,
      "learning_rate": 4.0564687688870563e-05,
      "loss": 11.7997,
      "step": 16274
    },
    {
      "epoch": 2.8459016393442624,
      "grad_norm": 0.6790000786603582,
      "learning_rate": 4.0553300040765755e-05,
      "loss": 11.4523,
      "step": 16275
    },
    {
      "epoch": 2.8460765027322403,
      "grad_norm": 0.708936663685117,
      "learning_rate": 4.054191358474981e-05,
      "loss": 11.7719,
      "step": 16276
    },
    {
      "epoch": 2.8462513661202187,
      "grad_norm": 0.7227365684160434,
      "learning_rate": 4.053052832105105e-05,
      "loss": 11.9139,
      "step": 16277
    },
    {
      "epoch": 2.8464262295081966,
      "grad_norm": 0.6477560681875654,
      "learning_rate": 4.05191442498978e-05,
      "loss": 11.8651,
      "step": 16278
    },
    {
      "epoch": 2.846601092896175,
      "grad_norm": 0.653367060820596,
      "learning_rate": 4.050776137151834e-05,
      "loss": 11.6667,
      "step": 16279
    },
    {
      "epoch": 2.8467759562841533,
      "grad_norm": 0.622769090224846,
      "learning_rate": 4.0496379686140926e-05,
      "loss": 11.816,
      "step": 16280
    },
    {
      "epoch": 2.846950819672131,
      "grad_norm": 0.681641689496583,
      "learning_rate": 4.04849991939938e-05,
      "loss": 11.8791,
      "step": 16281
    },
    {
      "epoch": 2.847125683060109,
      "grad_norm": 0.6805696831471224,
      "learning_rate": 4.047361989530518e-05,
      "loss": 11.8742,
      "step": 16282
    },
    {
      "epoch": 2.8473005464480874,
      "grad_norm": 0.6477408162041861,
      "learning_rate": 4.046224179030325e-05,
      "loss": 11.5951,
      "step": 16283
    },
    {
      "epoch": 2.8474754098360657,
      "grad_norm": 0.6911947753526122,
      "learning_rate": 4.04508648792162e-05,
      "loss": 11.8413,
      "step": 16284
    },
    {
      "epoch": 2.8476502732240436,
      "grad_norm": 0.7133984756359603,
      "learning_rate": 4.043948916227209e-05,
      "loss": 11.6063,
      "step": 16285
    },
    {
      "epoch": 2.847825136612022,
      "grad_norm": 0.6102903126565293,
      "learning_rate": 4.0428114639699135e-05,
      "loss": 11.735,
      "step": 16286
    },
    {
      "epoch": 2.848,
      "grad_norm": 0.6927144988127392,
      "learning_rate": 4.041674131172538e-05,
      "loss": 11.8287,
      "step": 16287
    },
    {
      "epoch": 2.848174863387978,
      "grad_norm": 0.6318035969760539,
      "learning_rate": 4.040536917857891e-05,
      "loss": 11.7447,
      "step": 16288
    },
    {
      "epoch": 2.8483497267759565,
      "grad_norm": 0.6733470133609061,
      "learning_rate": 4.039399824048777e-05,
      "loss": 11.8584,
      "step": 16289
    },
    {
      "epoch": 2.8485245901639344,
      "grad_norm": 0.6569840924415568,
      "learning_rate": 4.038262849767999e-05,
      "loss": 11.6095,
      "step": 16290
    },
    {
      "epoch": 2.8486994535519123,
      "grad_norm": 0.6175697996109759,
      "learning_rate": 4.037125995038354e-05,
      "loss": 11.4153,
      "step": 16291
    },
    {
      "epoch": 2.8488743169398907,
      "grad_norm": 0.6055266808462842,
      "learning_rate": 4.035989259882642e-05,
      "loss": 11.5387,
      "step": 16292
    },
    {
      "epoch": 2.849049180327869,
      "grad_norm": 0.6708654285238054,
      "learning_rate": 4.034852644323661e-05,
      "loss": 11.5898,
      "step": 16293
    },
    {
      "epoch": 2.849224043715847,
      "grad_norm": 0.6544077214653626,
      "learning_rate": 4.033716148384192e-05,
      "loss": 11.6686,
      "step": 16294
    },
    {
      "epoch": 2.8493989071038253,
      "grad_norm": 0.6349979227283445,
      "learning_rate": 4.0325797720870376e-05,
      "loss": 11.8981,
      "step": 16295
    },
    {
      "epoch": 2.849573770491803,
      "grad_norm": 0.634331598195688,
      "learning_rate": 4.031443515454981e-05,
      "loss": 12.0253,
      "step": 16296
    },
    {
      "epoch": 2.8497486338797815,
      "grad_norm": 0.6392571327176138,
      "learning_rate": 4.030307378510807e-05,
      "loss": 11.8445,
      "step": 16297
    },
    {
      "epoch": 2.8499234972677594,
      "grad_norm": 0.6544560168720897,
      "learning_rate": 4.029171361277304e-05,
      "loss": 12.1643,
      "step": 16298
    },
    {
      "epoch": 2.8500983606557377,
      "grad_norm": 0.6243794496322531,
      "learning_rate": 4.02803546377724e-05,
      "loss": 11.5898,
      "step": 16299
    },
    {
      "epoch": 2.8502732240437156,
      "grad_norm": 0.6407186038882009,
      "learning_rate": 4.026899686033406e-05,
      "loss": 11.7485,
      "step": 16300
    },
    {
      "epoch": 2.850448087431694,
      "grad_norm": 0.5998030208512759,
      "learning_rate": 4.025764028068573e-05,
      "loss": 11.8464,
      "step": 16301
    },
    {
      "epoch": 2.8506229508196723,
      "grad_norm": 0.6531275934508195,
      "learning_rate": 4.024628489905517e-05,
      "loss": 11.4455,
      "step": 16302
    },
    {
      "epoch": 2.85079781420765,
      "grad_norm": 0.6515888864893755,
      "learning_rate": 4.023493071567004e-05,
      "loss": 11.606,
      "step": 16303
    },
    {
      "epoch": 2.8509726775956286,
      "grad_norm": 0.6971842615244045,
      "learning_rate": 4.022357773075801e-05,
      "loss": 11.6229,
      "step": 16304
    },
    {
      "epoch": 2.8511475409836065,
      "grad_norm": 0.6885461198963636,
      "learning_rate": 4.0212225944546825e-05,
      "loss": 11.7985,
      "step": 16305
    },
    {
      "epoch": 2.851322404371585,
      "grad_norm": 0.6869989957591783,
      "learning_rate": 4.020087535726408e-05,
      "loss": 11.6062,
      "step": 16306
    },
    {
      "epoch": 2.8514972677595627,
      "grad_norm": 0.6595764425708797,
      "learning_rate": 4.018952596913742e-05,
      "loss": 11.8382,
      "step": 16307
    },
    {
      "epoch": 2.851672131147541,
      "grad_norm": 0.6519430668778394,
      "learning_rate": 4.0178177780394376e-05,
      "loss": 11.7764,
      "step": 16308
    },
    {
      "epoch": 2.851846994535519,
      "grad_norm": 0.6406756028361158,
      "learning_rate": 4.016683079126251e-05,
      "loss": 11.6135,
      "step": 16309
    },
    {
      "epoch": 2.8520218579234973,
      "grad_norm": 0.6644090821968175,
      "learning_rate": 4.015548500196944e-05,
      "loss": 11.9183,
      "step": 16310
    },
    {
      "epoch": 2.8521967213114756,
      "grad_norm": 0.5784221595880709,
      "learning_rate": 4.014414041274267e-05,
      "loss": 11.8915,
      "step": 16311
    },
    {
      "epoch": 2.8523715846994535,
      "grad_norm": 0.6305958302172892,
      "learning_rate": 4.013279702380963e-05,
      "loss": 11.825,
      "step": 16312
    },
    {
      "epoch": 2.852546448087432,
      "grad_norm": 0.7091696066044442,
      "learning_rate": 4.01214548353978e-05,
      "loss": 11.4784,
      "step": 16313
    },
    {
      "epoch": 2.8527213114754097,
      "grad_norm": 0.7050355407928159,
      "learning_rate": 4.011011384773467e-05,
      "loss": 11.9271,
      "step": 16314
    },
    {
      "epoch": 2.852896174863388,
      "grad_norm": 0.6627537041131114,
      "learning_rate": 4.009877406104766e-05,
      "loss": 11.8035,
      "step": 16315
    },
    {
      "epoch": 2.853071038251366,
      "grad_norm": 0.6632514189995216,
      "learning_rate": 4.008743547556418e-05,
      "loss": 11.8646,
      "step": 16316
    },
    {
      "epoch": 2.8532459016393443,
      "grad_norm": 0.585731483961139,
      "learning_rate": 4.007609809151155e-05,
      "loss": 11.8185,
      "step": 16317
    },
    {
      "epoch": 2.8534207650273222,
      "grad_norm": 0.6151683195168207,
      "learning_rate": 4.00647619091171e-05,
      "loss": 11.9488,
      "step": 16318
    },
    {
      "epoch": 2.8535956284153006,
      "grad_norm": 0.6558693071264707,
      "learning_rate": 4.005342692860823e-05,
      "loss": 11.8896,
      "step": 16319
    },
    {
      "epoch": 2.853770491803279,
      "grad_norm": 0.6605922964052664,
      "learning_rate": 4.004209315021225e-05,
      "loss": 11.7969,
      "step": 16320
    },
    {
      "epoch": 2.853945355191257,
      "grad_norm": 0.620252409700845,
      "learning_rate": 4.003076057415637e-05,
      "loss": 11.9518,
      "step": 16321
    },
    {
      "epoch": 2.8541202185792347,
      "grad_norm": 0.6288905274622963,
      "learning_rate": 4.0019429200667856e-05,
      "loss": 11.6203,
      "step": 16322
    },
    {
      "epoch": 2.854295081967213,
      "grad_norm": 0.667217321868065,
      "learning_rate": 4.000809902997392e-05,
      "loss": 11.8797,
      "step": 16323
    },
    {
      "epoch": 2.8544699453551914,
      "grad_norm": 0.6095418218273356,
      "learning_rate": 3.9996770062301836e-05,
      "loss": 11.7867,
      "step": 16324
    },
    {
      "epoch": 2.8546448087431693,
      "grad_norm": 0.6091724713881271,
      "learning_rate": 3.9985442297878773e-05,
      "loss": 11.7133,
      "step": 16325
    },
    {
      "epoch": 2.8548196721311476,
      "grad_norm": 0.7389516103485692,
      "learning_rate": 3.997411573693184e-05,
      "loss": 11.8402,
      "step": 16326
    },
    {
      "epoch": 2.8549945355191255,
      "grad_norm": 0.6521623252417994,
      "learning_rate": 3.996279037968817e-05,
      "loss": 11.6326,
      "step": 16327
    },
    {
      "epoch": 2.855169398907104,
      "grad_norm": 0.6321768604373432,
      "learning_rate": 3.995146622637488e-05,
      "loss": 11.576,
      "step": 16328
    },
    {
      "epoch": 2.855344262295082,
      "grad_norm": 0.6150075890881659,
      "learning_rate": 3.994014327721912e-05,
      "loss": 11.6889,
      "step": 16329
    },
    {
      "epoch": 2.85551912568306,
      "grad_norm": 0.6674648785680196,
      "learning_rate": 3.9928821532447866e-05,
      "loss": 11.789,
      "step": 16330
    },
    {
      "epoch": 2.855693989071038,
      "grad_norm": 0.600157615655983,
      "learning_rate": 3.991750099228818e-05,
      "loss": 11.5587,
      "step": 16331
    },
    {
      "epoch": 2.8558688524590163,
      "grad_norm": 0.6302948866440788,
      "learning_rate": 3.990618165696708e-05,
      "loss": 11.7062,
      "step": 16332
    },
    {
      "epoch": 2.8560437158469947,
      "grad_norm": 0.6146705702601785,
      "learning_rate": 3.989486352671151e-05,
      "loss": 11.7181,
      "step": 16333
    },
    {
      "epoch": 2.8562185792349726,
      "grad_norm": 0.6038185083681019,
      "learning_rate": 3.988354660174855e-05,
      "loss": 11.9176,
      "step": 16334
    },
    {
      "epoch": 2.856393442622951,
      "grad_norm": 0.5965910469785168,
      "learning_rate": 3.987223088230503e-05,
      "loss": 11.8599,
      "step": 16335
    },
    {
      "epoch": 2.856568306010929,
      "grad_norm": 0.587412353771166,
      "learning_rate": 3.986091636860788e-05,
      "loss": 11.7888,
      "step": 16336
    },
    {
      "epoch": 2.856743169398907,
      "grad_norm": 0.6207416792734302,
      "learning_rate": 3.984960306088399e-05,
      "loss": 11.5793,
      "step": 16337
    },
    {
      "epoch": 2.8569180327868855,
      "grad_norm": 0.6529177016406864,
      "learning_rate": 3.9838290959360304e-05,
      "loss": 11.814,
      "step": 16338
    },
    {
      "epoch": 2.8570928961748634,
      "grad_norm": 0.6270472098242691,
      "learning_rate": 3.982698006426357e-05,
      "loss": 11.6538,
      "step": 16339
    },
    {
      "epoch": 2.8572677595628413,
      "grad_norm": 0.7146790004545074,
      "learning_rate": 3.9815670375820655e-05,
      "loss": 11.9929,
      "step": 16340
    },
    {
      "epoch": 2.8574426229508196,
      "grad_norm": 0.7206605334684218,
      "learning_rate": 3.980436189425832e-05,
      "loss": 11.5589,
      "step": 16341
    },
    {
      "epoch": 2.857617486338798,
      "grad_norm": 0.628346551843646,
      "learning_rate": 3.979305461980334e-05,
      "loss": 11.2983,
      "step": 16342
    },
    {
      "epoch": 2.857792349726776,
      "grad_norm": 0.7238266151260263,
      "learning_rate": 3.978174855268253e-05,
      "loss": 11.477,
      "step": 16343
    },
    {
      "epoch": 2.857967213114754,
      "grad_norm": 0.6475156449309049,
      "learning_rate": 3.9770443693122526e-05,
      "loss": 11.7566,
      "step": 16344
    },
    {
      "epoch": 2.858142076502732,
      "grad_norm": 0.5931854789497004,
      "learning_rate": 3.9759140041350054e-05,
      "loss": 11.9456,
      "step": 16345
    },
    {
      "epoch": 2.8583169398907105,
      "grad_norm": 0.6333084347391651,
      "learning_rate": 3.974783759759179e-05,
      "loss": 11.6145,
      "step": 16346
    },
    {
      "epoch": 2.858491803278689,
      "grad_norm": 0.6361226706253171,
      "learning_rate": 3.973653636207437e-05,
      "loss": 11.8135,
      "step": 16347
    },
    {
      "epoch": 2.8586666666666667,
      "grad_norm": 0.5731714728963988,
      "learning_rate": 3.9725236335024443e-05,
      "loss": 11.4748,
      "step": 16348
    },
    {
      "epoch": 2.8588415300546446,
      "grad_norm": 0.6259675733715255,
      "learning_rate": 3.9713937516668584e-05,
      "loss": 11.7864,
      "step": 16349
    },
    {
      "epoch": 2.859016393442623,
      "grad_norm": 0.6280715513070706,
      "learning_rate": 3.970263990723338e-05,
      "loss": 11.8256,
      "step": 16350
    },
    {
      "epoch": 2.8591912568306013,
      "grad_norm": 0.760075573442802,
      "learning_rate": 3.96913435069454e-05,
      "loss": 11.6822,
      "step": 16351
    },
    {
      "epoch": 2.859366120218579,
      "grad_norm": 0.6212491514703982,
      "learning_rate": 3.9680048316031134e-05,
      "loss": 11.6329,
      "step": 16352
    },
    {
      "epoch": 2.8595409836065575,
      "grad_norm": 0.6571784643042767,
      "learning_rate": 3.966875433471712e-05,
      "loss": 11.7316,
      "step": 16353
    },
    {
      "epoch": 2.8597158469945354,
      "grad_norm": 0.6577306598638575,
      "learning_rate": 3.965746156322981e-05,
      "loss": 11.647,
      "step": 16354
    },
    {
      "epoch": 2.8598907103825137,
      "grad_norm": 0.8061572533872945,
      "learning_rate": 3.964617000179569e-05,
      "loss": 11.5941,
      "step": 16355
    },
    {
      "epoch": 2.8600655737704916,
      "grad_norm": 0.6248145697345238,
      "learning_rate": 3.9634879650641153e-05,
      "loss": 11.8373,
      "step": 16356
    },
    {
      "epoch": 2.86024043715847,
      "grad_norm": 0.6743277707149415,
      "learning_rate": 3.962359050999264e-05,
      "loss": 11.5552,
      "step": 16357
    },
    {
      "epoch": 2.860415300546448,
      "grad_norm": 0.7081992296631214,
      "learning_rate": 3.961230258007651e-05,
      "loss": 11.6826,
      "step": 16358
    },
    {
      "epoch": 2.8605901639344262,
      "grad_norm": 0.7372572171489572,
      "learning_rate": 3.960101586111913e-05,
      "loss": 11.6392,
      "step": 16359
    },
    {
      "epoch": 2.8607650273224046,
      "grad_norm": 0.69684162483018,
      "learning_rate": 3.958973035334682e-05,
      "loss": 11.8539,
      "step": 16360
    },
    {
      "epoch": 2.8609398907103825,
      "grad_norm": 0.6602822023538828,
      "learning_rate": 3.9578446056985916e-05,
      "loss": 11.6991,
      "step": 16361
    },
    {
      "epoch": 2.861114754098361,
      "grad_norm": 0.7747167105389579,
      "learning_rate": 3.956716297226268e-05,
      "loss": 11.8326,
      "step": 16362
    },
    {
      "epoch": 2.8612896174863387,
      "grad_norm": 0.6651293538435364,
      "learning_rate": 3.955588109940338e-05,
      "loss": 11.777,
      "step": 16363
    },
    {
      "epoch": 2.861464480874317,
      "grad_norm": 0.6046720593671081,
      "learning_rate": 3.954460043863425e-05,
      "loss": 11.4558,
      "step": 16364
    },
    {
      "epoch": 2.861639344262295,
      "grad_norm": 0.6776063350810732,
      "learning_rate": 3.953332099018151e-05,
      "loss": 11.8027,
      "step": 16365
    },
    {
      "epoch": 2.8618142076502733,
      "grad_norm": 0.6576592924793998,
      "learning_rate": 3.952204275427134e-05,
      "loss": 11.6583,
      "step": 16366
    },
    {
      "epoch": 2.861989071038251,
      "grad_norm": 0.6558694824339109,
      "learning_rate": 3.95107657311299e-05,
      "loss": 11.7381,
      "step": 16367
    },
    {
      "epoch": 2.8621639344262295,
      "grad_norm": 0.6732057851980008,
      "learning_rate": 3.9499489920983335e-05,
      "loss": 11.6426,
      "step": 16368
    },
    {
      "epoch": 2.862338797814208,
      "grad_norm": 0.671268834348523,
      "learning_rate": 3.948821532405776e-05,
      "loss": 11.9233,
      "step": 16369
    },
    {
      "epoch": 2.8625136612021858,
      "grad_norm": 0.6384621757659488,
      "learning_rate": 3.9476941940579246e-05,
      "loss": 11.9843,
      "step": 16370
    },
    {
      "epoch": 2.862688524590164,
      "grad_norm": 0.6261226141175161,
      "learning_rate": 3.946566977077389e-05,
      "loss": 11.754,
      "step": 16371
    },
    {
      "epoch": 2.862863387978142,
      "grad_norm": 0.611270719236316,
      "learning_rate": 3.945439881486773e-05,
      "loss": 11.8035,
      "step": 16372
    },
    {
      "epoch": 2.8630382513661203,
      "grad_norm": 0.6751014603583773,
      "learning_rate": 3.944312907308676e-05,
      "loss": 11.9065,
      "step": 16373
    },
    {
      "epoch": 2.8632131147540982,
      "grad_norm": 0.6234895282583324,
      "learning_rate": 3.943186054565699e-05,
      "loss": 11.5622,
      "step": 16374
    },
    {
      "epoch": 2.8633879781420766,
      "grad_norm": 0.6603193133293515,
      "learning_rate": 3.9420593232804425e-05,
      "loss": 11.8391,
      "step": 16375
    },
    {
      "epoch": 2.8635628415300545,
      "grad_norm": 0.6428069850586001,
      "learning_rate": 3.9409327134754895e-05,
      "loss": 11.6791,
      "step": 16376
    },
    {
      "epoch": 2.863737704918033,
      "grad_norm": 0.6911836248458134,
      "learning_rate": 3.9398062251734425e-05,
      "loss": 12.0107,
      "step": 16377
    },
    {
      "epoch": 2.863912568306011,
      "grad_norm": 0.6403485728483872,
      "learning_rate": 3.938679858396889e-05,
      "loss": 11.6832,
      "step": 16378
    },
    {
      "epoch": 2.864087431693989,
      "grad_norm": 0.6733717478562804,
      "learning_rate": 3.9375536131684134e-05,
      "loss": 11.7954,
      "step": 16379
    },
    {
      "epoch": 2.864262295081967,
      "grad_norm": 0.6409355449778892,
      "learning_rate": 3.9364274895106066e-05,
      "loss": 11.8667,
      "step": 16380
    },
    {
      "epoch": 2.8644371584699453,
      "grad_norm": 0.6418231009611319,
      "learning_rate": 3.9353014874460395e-05,
      "loss": 11.5712,
      "step": 16381
    },
    {
      "epoch": 2.8646120218579236,
      "grad_norm": 0.6368160805120258,
      "learning_rate": 3.9341756069973015e-05,
      "loss": 11.8033,
      "step": 16382
    },
    {
      "epoch": 2.8647868852459015,
      "grad_norm": 0.6042600059305111,
      "learning_rate": 3.933049848186967e-05,
      "loss": 11.8298,
      "step": 16383
    },
    {
      "epoch": 2.86496174863388,
      "grad_norm": 0.6590099052697025,
      "learning_rate": 3.931924211037614e-05,
      "loss": 11.9229,
      "step": 16384
    },
    {
      "epoch": 2.8651366120218578,
      "grad_norm": 0.7030816460305606,
      "learning_rate": 3.930798695571807e-05,
      "loss": 11.6197,
      "step": 16385
    },
    {
      "epoch": 2.865311475409836,
      "grad_norm": 0.6734774848850613,
      "learning_rate": 3.929673301812122e-05,
      "loss": 11.8898,
      "step": 16386
    },
    {
      "epoch": 2.8654863387978144,
      "grad_norm": 0.5938352436794472,
      "learning_rate": 3.9285480297811276e-05,
      "loss": 11.5252,
      "step": 16387
    },
    {
      "epoch": 2.8656612021857923,
      "grad_norm": 0.6547251571984001,
      "learning_rate": 3.927422879501387e-05,
      "loss": 11.8318,
      "step": 16388
    },
    {
      "epoch": 2.8658360655737702,
      "grad_norm": 0.6607585832117815,
      "learning_rate": 3.9262978509954654e-05,
      "loss": 11.7532,
      "step": 16389
    },
    {
      "epoch": 2.8660109289617486,
      "grad_norm": 0.7080137141665569,
      "learning_rate": 3.9251729442859143e-05,
      "loss": 12.0221,
      "step": 16390
    },
    {
      "epoch": 2.866185792349727,
      "grad_norm": 0.6950539100882774,
      "learning_rate": 3.924048159395301e-05,
      "loss": 12.0461,
      "step": 16391
    },
    {
      "epoch": 2.866360655737705,
      "grad_norm": 0.6152815156218225,
      "learning_rate": 3.9229234963461766e-05,
      "loss": 11.8645,
      "step": 16392
    },
    {
      "epoch": 2.866535519125683,
      "grad_norm": 0.8072467980016201,
      "learning_rate": 3.921798955161101e-05,
      "loss": 11.8928,
      "step": 16393
    },
    {
      "epoch": 2.866710382513661,
      "grad_norm": 0.6637935711384337,
      "learning_rate": 3.920674535862614e-05,
      "loss": 11.7718,
      "step": 16394
    },
    {
      "epoch": 2.8668852459016394,
      "grad_norm": 0.6387232532621915,
      "learning_rate": 3.919550238473265e-05,
      "loss": 11.7943,
      "step": 16395
    },
    {
      "epoch": 2.8670601092896177,
      "grad_norm": 0.6871728992206266,
      "learning_rate": 3.9184260630156064e-05,
      "loss": 11.7924,
      "step": 16396
    },
    {
      "epoch": 2.8672349726775956,
      "grad_norm": 0.6327775919367962,
      "learning_rate": 3.9173020095121794e-05,
      "loss": 11.8184,
      "step": 16397
    },
    {
      "epoch": 2.8674098360655735,
      "grad_norm": 0.6435229386104923,
      "learning_rate": 3.916178077985526e-05,
      "loss": 11.988,
      "step": 16398
    },
    {
      "epoch": 2.867584699453552,
      "grad_norm": 0.6143477103008167,
      "learning_rate": 3.915054268458179e-05,
      "loss": 11.854,
      "step": 16399
    },
    {
      "epoch": 2.86775956284153,
      "grad_norm": 0.59466906730206,
      "learning_rate": 3.913930580952673e-05,
      "loss": 11.9344,
      "step": 16400
    },
    {
      "epoch": 2.867934426229508,
      "grad_norm": 0.5795095204530967,
      "learning_rate": 3.9128070154915496e-05,
      "loss": 11.6608,
      "step": 16401
    },
    {
      "epoch": 2.8681092896174865,
      "grad_norm": 0.6546957249941795,
      "learning_rate": 3.911683572097339e-05,
      "loss": 11.7658,
      "step": 16402
    },
    {
      "epoch": 2.8682841530054644,
      "grad_norm": 0.6176633253230235,
      "learning_rate": 3.9105602507925624e-05,
      "loss": 11.693,
      "step": 16403
    },
    {
      "epoch": 2.8684590163934427,
      "grad_norm": 0.66614790796391,
      "learning_rate": 3.9094370515997504e-05,
      "loss": 11.6192,
      "step": 16404
    },
    {
      "epoch": 2.868633879781421,
      "grad_norm": 0.6511119435861019,
      "learning_rate": 3.908313974541422e-05,
      "loss": 11.81,
      "step": 16405
    },
    {
      "epoch": 2.868808743169399,
      "grad_norm": 0.6097770053696489,
      "learning_rate": 3.9071910196401064e-05,
      "loss": 11.5984,
      "step": 16406
    },
    {
      "epoch": 2.868983606557377,
      "grad_norm": 0.7391514312521028,
      "learning_rate": 3.9060681869183216e-05,
      "loss": 11.767,
      "step": 16407
    },
    {
      "epoch": 2.869158469945355,
      "grad_norm": 0.6889636209887056,
      "learning_rate": 3.904945476398576e-05,
      "loss": 11.5942,
      "step": 16408
    },
    {
      "epoch": 2.8693333333333335,
      "grad_norm": 0.6468231107602351,
      "learning_rate": 3.903822888103386e-05,
      "loss": 11.8569,
      "step": 16409
    },
    {
      "epoch": 2.8695081967213114,
      "grad_norm": 0.6824714991173892,
      "learning_rate": 3.902700422055266e-05,
      "loss": 11.837,
      "step": 16410
    },
    {
      "epoch": 2.8696830601092898,
      "grad_norm": 0.6697531331499741,
      "learning_rate": 3.901578078276726e-05,
      "loss": 11.9225,
      "step": 16411
    },
    {
      "epoch": 2.8698579234972676,
      "grad_norm": 0.6709429772180803,
      "learning_rate": 3.900455856790273e-05,
      "loss": 11.7119,
      "step": 16412
    },
    {
      "epoch": 2.870032786885246,
      "grad_norm": 0.6026540669550877,
      "learning_rate": 3.899333757618404e-05,
      "loss": 11.7919,
      "step": 16413
    },
    {
      "epoch": 2.870207650273224,
      "grad_norm": 0.6491319857626451,
      "learning_rate": 3.8982117807836214e-05,
      "loss": 11.9093,
      "step": 16414
    },
    {
      "epoch": 2.8703825136612022,
      "grad_norm": 0.6224819061972552,
      "learning_rate": 3.8970899263084314e-05,
      "loss": 12.0134,
      "step": 16415
    },
    {
      "epoch": 2.87055737704918,
      "grad_norm": 0.6216672615772111,
      "learning_rate": 3.8959681942153304e-05,
      "loss": 11.7632,
      "step": 16416
    },
    {
      "epoch": 2.8707322404371585,
      "grad_norm": 0.6308282026722358,
      "learning_rate": 3.8948465845268055e-05,
      "loss": 11.8235,
      "step": 16417
    },
    {
      "epoch": 2.870907103825137,
      "grad_norm": 0.646617821317203,
      "learning_rate": 3.893725097265352e-05,
      "loss": 11.9818,
      "step": 16418
    },
    {
      "epoch": 2.8710819672131147,
      "grad_norm": 0.6165765194444711,
      "learning_rate": 3.892603732453455e-05,
      "loss": 11.7331,
      "step": 16419
    },
    {
      "epoch": 2.871256830601093,
      "grad_norm": 0.7159854883167185,
      "learning_rate": 3.891482490113608e-05,
      "loss": 11.7939,
      "step": 16420
    },
    {
      "epoch": 2.871431693989071,
      "grad_norm": 0.5975077019891544,
      "learning_rate": 3.8903613702682975e-05,
      "loss": 11.7571,
      "step": 16421
    },
    {
      "epoch": 2.8716065573770493,
      "grad_norm": 0.6829120125520437,
      "learning_rate": 3.8892403729399975e-05,
      "loss": 11.958,
      "step": 16422
    },
    {
      "epoch": 2.871781420765027,
      "grad_norm": 0.6637245195450222,
      "learning_rate": 3.8881194981511896e-05,
      "loss": 11.8436,
      "step": 16423
    },
    {
      "epoch": 2.8719562841530055,
      "grad_norm": 0.7312346642786689,
      "learning_rate": 3.8869987459243486e-05,
      "loss": 11.7457,
      "step": 16424
    },
    {
      "epoch": 2.8721311475409834,
      "grad_norm": 0.6339588615453781,
      "learning_rate": 3.885878116281958e-05,
      "loss": 11.4499,
      "step": 16425
    },
    {
      "epoch": 2.8723060109289618,
      "grad_norm": 0.6359503620692217,
      "learning_rate": 3.884757609246481e-05,
      "loss": 11.8417,
      "step": 16426
    },
    {
      "epoch": 2.87248087431694,
      "grad_norm": 0.6887582792050564,
      "learning_rate": 3.883637224840391e-05,
      "loss": 11.5957,
      "step": 16427
    },
    {
      "epoch": 2.872655737704918,
      "grad_norm": 0.5966203960677343,
      "learning_rate": 3.882516963086154e-05,
      "loss": 11.9353,
      "step": 16428
    },
    {
      "epoch": 2.8728306010928963,
      "grad_norm": 0.6794527090725492,
      "learning_rate": 3.8813968240062314e-05,
      "loss": 11.8395,
      "step": 16429
    },
    {
      "epoch": 2.8730054644808742,
      "grad_norm": 0.5949359298427886,
      "learning_rate": 3.880276807623095e-05,
      "loss": 11.7019,
      "step": 16430
    },
    {
      "epoch": 2.8731803278688526,
      "grad_norm": 0.6706587296243346,
      "learning_rate": 3.879156913959198e-05,
      "loss": 11.7596,
      "step": 16431
    },
    {
      "epoch": 2.8733551912568305,
      "grad_norm": 0.6471536670213925,
      "learning_rate": 3.878037143036995e-05,
      "loss": 11.729,
      "step": 16432
    },
    {
      "epoch": 2.873530054644809,
      "grad_norm": 0.6679787459396235,
      "learning_rate": 3.876917494878943e-05,
      "loss": 11.6732,
      "step": 16433
    },
    {
      "epoch": 2.8737049180327867,
      "grad_norm": 0.6842865078885573,
      "learning_rate": 3.875797969507502e-05,
      "loss": 11.8388,
      "step": 16434
    },
    {
      "epoch": 2.873879781420765,
      "grad_norm": 0.6494985767750643,
      "learning_rate": 3.874678566945111e-05,
      "loss": 11.7708,
      "step": 16435
    },
    {
      "epoch": 2.8740546448087434,
      "grad_norm": 0.6685006198462332,
      "learning_rate": 3.8735592872142236e-05,
      "loss": 11.6327,
      "step": 16436
    },
    {
      "epoch": 2.8742295081967213,
      "grad_norm": 0.6497076438662912,
      "learning_rate": 3.872440130337281e-05,
      "loss": 11.7175,
      "step": 16437
    },
    {
      "epoch": 2.874404371584699,
      "grad_norm": 0.6510521256977346,
      "learning_rate": 3.8713210963367266e-05,
      "loss": 11.7518,
      "step": 16438
    },
    {
      "epoch": 2.8745792349726775,
      "grad_norm": 0.6587634926874711,
      "learning_rate": 3.870202185235008e-05,
      "loss": 11.6795,
      "step": 16439
    },
    {
      "epoch": 2.874754098360656,
      "grad_norm": 0.6341339464618468,
      "learning_rate": 3.8690833970545526e-05,
      "loss": 11.7368,
      "step": 16440
    },
    {
      "epoch": 2.8749289617486338,
      "grad_norm": 0.6723694798613337,
      "learning_rate": 3.867964731817799e-05,
      "loss": 11.6242,
      "step": 16441
    },
    {
      "epoch": 2.875103825136612,
      "grad_norm": 0.6375724936060058,
      "learning_rate": 3.866846189547181e-05,
      "loss": 11.6816,
      "step": 16442
    },
    {
      "epoch": 2.87527868852459,
      "grad_norm": 0.6371972791131957,
      "learning_rate": 3.8657277702651286e-05,
      "loss": 11.6571,
      "step": 16443
    },
    {
      "epoch": 2.8754535519125684,
      "grad_norm": 0.7070497639781256,
      "learning_rate": 3.864609473994069e-05,
      "loss": 11.809,
      "step": 16444
    },
    {
      "epoch": 2.8756284153005467,
      "grad_norm": 0.6574396036239337,
      "learning_rate": 3.863491300756427e-05,
      "loss": 11.9112,
      "step": 16445
    },
    {
      "epoch": 2.8758032786885246,
      "grad_norm": 0.6575060360975936,
      "learning_rate": 3.862373250574626e-05,
      "loss": 11.8386,
      "step": 16446
    },
    {
      "epoch": 2.8759781420765025,
      "grad_norm": 0.6282679021653019,
      "learning_rate": 3.861255323471086e-05,
      "loss": 11.7754,
      "step": 16447
    },
    {
      "epoch": 2.876153005464481,
      "grad_norm": 0.6375666374556943,
      "learning_rate": 3.8601375194682254e-05,
      "loss": 11.9426,
      "step": 16448
    },
    {
      "epoch": 2.876327868852459,
      "grad_norm": 0.6611498431324946,
      "learning_rate": 3.85901983858846e-05,
      "loss": 11.6392,
      "step": 16449
    },
    {
      "epoch": 2.876502732240437,
      "grad_norm": 0.6387746102375967,
      "learning_rate": 3.857902280854202e-05,
      "loss": 11.4354,
      "step": 16450
    },
    {
      "epoch": 2.8766775956284154,
      "grad_norm": 0.6886238405352936,
      "learning_rate": 3.856784846287862e-05,
      "loss": 11.8167,
      "step": 16451
    },
    {
      "epoch": 2.8768524590163933,
      "grad_norm": 0.6760792839717732,
      "learning_rate": 3.8556675349118474e-05,
      "loss": 11.8316,
      "step": 16452
    },
    {
      "epoch": 2.8770273224043716,
      "grad_norm": 0.6128854265886222,
      "learning_rate": 3.854550346748564e-05,
      "loss": 11.796,
      "step": 16453
    },
    {
      "epoch": 2.87720218579235,
      "grad_norm": 0.6171382917250644,
      "learning_rate": 3.853433281820417e-05,
      "loss": 11.7634,
      "step": 16454
    },
    {
      "epoch": 2.877377049180328,
      "grad_norm": 0.7011487988790303,
      "learning_rate": 3.852316340149803e-05,
      "loss": 11.7941,
      "step": 16455
    },
    {
      "epoch": 2.877551912568306,
      "grad_norm": 0.6696548019225125,
      "learning_rate": 3.851199521759124e-05,
      "loss": 11.6943,
      "step": 16456
    },
    {
      "epoch": 2.877726775956284,
      "grad_norm": 0.6691045188260729,
      "learning_rate": 3.850082826670772e-05,
      "loss": 11.7435,
      "step": 16457
    },
    {
      "epoch": 2.8779016393442625,
      "grad_norm": 0.6698058088531954,
      "learning_rate": 3.848966254907144e-05,
      "loss": 11.9859,
      "step": 16458
    },
    {
      "epoch": 2.8780765027322404,
      "grad_norm": 0.6497531902525145,
      "learning_rate": 3.8478498064906273e-05,
      "loss": 11.6465,
      "step": 16459
    },
    {
      "epoch": 2.8782513661202187,
      "grad_norm": 0.6635308390822,
      "learning_rate": 3.846733481443613e-05,
      "loss": 12.0363,
      "step": 16460
    },
    {
      "epoch": 2.8784262295081966,
      "grad_norm": 0.6298592901389264,
      "learning_rate": 3.845617279788484e-05,
      "loss": 11.5003,
      "step": 16461
    },
    {
      "epoch": 2.878601092896175,
      "grad_norm": 0.667638780980032,
      "learning_rate": 3.8445012015476256e-05,
      "loss": 11.8358,
      "step": 16462
    },
    {
      "epoch": 2.8787759562841533,
      "grad_norm": 0.6140748215288014,
      "learning_rate": 3.843385246743417e-05,
      "loss": 11.9062,
      "step": 16463
    },
    {
      "epoch": 2.878950819672131,
      "grad_norm": 0.731747526539925,
      "learning_rate": 3.842269415398239e-05,
      "loss": 11.8591,
      "step": 16464
    },
    {
      "epoch": 2.879125683060109,
      "grad_norm": 0.639307625413618,
      "learning_rate": 3.841153707534465e-05,
      "loss": 11.5306,
      "step": 16465
    },
    {
      "epoch": 2.8793005464480874,
      "grad_norm": 0.6687886180413396,
      "learning_rate": 3.84003812317447e-05,
      "loss": 11.8182,
      "step": 16466
    },
    {
      "epoch": 2.8794754098360658,
      "grad_norm": 0.6791879911999837,
      "learning_rate": 3.838922662340624e-05,
      "loss": 11.7612,
      "step": 16467
    },
    {
      "epoch": 2.8796502732240437,
      "grad_norm": 0.6338575965479747,
      "learning_rate": 3.8378073250552946e-05,
      "loss": 11.6866,
      "step": 16468
    },
    {
      "epoch": 2.879825136612022,
      "grad_norm": 0.6058216699711478,
      "learning_rate": 3.836692111340851e-05,
      "loss": 11.6392,
      "step": 16469
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.6339857942883113,
      "learning_rate": 3.835577021219652e-05,
      "loss": 11.6659,
      "step": 16470
    },
    {
      "epoch": 2.8801748633879782,
      "grad_norm": 0.7219227025200481,
      "learning_rate": 3.8344620547140655e-05,
      "loss": 11.8431,
      "step": 16471
    },
    {
      "epoch": 2.880349726775956,
      "grad_norm": 0.667745437596426,
      "learning_rate": 3.8333472118464386e-05,
      "loss": 11.6478,
      "step": 16472
    },
    {
      "epoch": 2.8805245901639345,
      "grad_norm": 0.6798600458178106,
      "learning_rate": 3.832232492639137e-05,
      "loss": 11.8412,
      "step": 16473
    },
    {
      "epoch": 2.8806994535519124,
      "grad_norm": 0.6261486185325588,
      "learning_rate": 3.831117897114512e-05,
      "loss": 11.7529,
      "step": 16474
    },
    {
      "epoch": 2.8808743169398907,
      "grad_norm": 0.6118692820001775,
      "learning_rate": 3.8300034252949146e-05,
      "loss": 11.8925,
      "step": 16475
    },
    {
      "epoch": 2.881049180327869,
      "grad_norm": 0.6083544481428393,
      "learning_rate": 3.8288890772026965e-05,
      "loss": 11.4883,
      "step": 16476
    },
    {
      "epoch": 2.881224043715847,
      "grad_norm": 0.6387198480356623,
      "learning_rate": 3.827774852860192e-05,
      "loss": 11.6955,
      "step": 16477
    },
    {
      "epoch": 2.8813989071038253,
      "grad_norm": 0.630063839716951,
      "learning_rate": 3.826660752289758e-05,
      "loss": 11.745,
      "step": 16478
    },
    {
      "epoch": 2.881573770491803,
      "grad_norm": 0.6131174429940532,
      "learning_rate": 3.825546775513729e-05,
      "loss": 11.5439,
      "step": 16479
    },
    {
      "epoch": 2.8817486338797815,
      "grad_norm": 0.6932820334002587,
      "learning_rate": 3.824432922554449e-05,
      "loss": 11.865,
      "step": 16480
    },
    {
      "epoch": 2.8819234972677594,
      "grad_norm": 0.6914625351143436,
      "learning_rate": 3.823319193434244e-05,
      "loss": 11.5898,
      "step": 16481
    },
    {
      "epoch": 2.8820983606557378,
      "grad_norm": 0.6267839444093377,
      "learning_rate": 3.822205588175457e-05,
      "loss": 11.7747,
      "step": 16482
    },
    {
      "epoch": 2.8822732240437157,
      "grad_norm": 0.692456115236965,
      "learning_rate": 3.821092106800417e-05,
      "loss": 11.8181,
      "step": 16483
    },
    {
      "epoch": 2.882448087431694,
      "grad_norm": 0.6658340864427662,
      "learning_rate": 3.819978749331452e-05,
      "loss": 11.8145,
      "step": 16484
    },
    {
      "epoch": 2.8826229508196723,
      "grad_norm": 0.636917538908193,
      "learning_rate": 3.8188655157908916e-05,
      "loss": 11.7032,
      "step": 16485
    },
    {
      "epoch": 2.8827978142076502,
      "grad_norm": 0.6696209064754861,
      "learning_rate": 3.8177524062010486e-05,
      "loss": 11.5968,
      "step": 16486
    },
    {
      "epoch": 2.882972677595628,
      "grad_norm": 0.6202672204195405,
      "learning_rate": 3.816639420584255e-05,
      "loss": 11.903,
      "step": 16487
    },
    {
      "epoch": 2.8831475409836065,
      "grad_norm": 0.6457297595681177,
      "learning_rate": 3.8155265589628276e-05,
      "loss": 11.6704,
      "step": 16488
    },
    {
      "epoch": 2.883322404371585,
      "grad_norm": 0.6223080358798458,
      "learning_rate": 3.814413821359084e-05,
      "loss": 11.5226,
      "step": 16489
    },
    {
      "epoch": 2.8834972677595627,
      "grad_norm": 0.6488211121158226,
      "learning_rate": 3.8133012077953325e-05,
      "loss": 11.9347,
      "step": 16490
    },
    {
      "epoch": 2.883672131147541,
      "grad_norm": 0.6268598734480963,
      "learning_rate": 3.8121887182938845e-05,
      "loss": 11.8467,
      "step": 16491
    },
    {
      "epoch": 2.883846994535519,
      "grad_norm": 0.6236917159165886,
      "learning_rate": 3.811076352877054e-05,
      "loss": 11.6645,
      "step": 16492
    },
    {
      "epoch": 2.8840218579234973,
      "grad_norm": 0.6655421384166053,
      "learning_rate": 3.809964111567146e-05,
      "loss": 11.8364,
      "step": 16493
    },
    {
      "epoch": 2.8841967213114756,
      "grad_norm": 0.6686715731951028,
      "learning_rate": 3.8088519943864675e-05,
      "loss": 11.8771,
      "step": 16494
    },
    {
      "epoch": 2.8843715846994535,
      "grad_norm": 0.6712463725659027,
      "learning_rate": 3.8077400013573116e-05,
      "loss": 11.8451,
      "step": 16495
    },
    {
      "epoch": 2.8845464480874314,
      "grad_norm": 0.6197215464650396,
      "learning_rate": 3.806628132501978e-05,
      "loss": 11.7981,
      "step": 16496
    },
    {
      "epoch": 2.8847213114754098,
      "grad_norm": 0.607124214580636,
      "learning_rate": 3.80551638784277e-05,
      "loss": 11.7577,
      "step": 16497
    },
    {
      "epoch": 2.884896174863388,
      "grad_norm": 0.6739132838153995,
      "learning_rate": 3.804404767401981e-05,
      "loss": 11.6803,
      "step": 16498
    },
    {
      "epoch": 2.885071038251366,
      "grad_norm": 0.7663417872880158,
      "learning_rate": 3.8032932712018966e-05,
      "loss": 11.2349,
      "step": 16499
    },
    {
      "epoch": 2.8852459016393444,
      "grad_norm": 0.7022957139213221,
      "learning_rate": 3.802181899264809e-05,
      "loss": 11.6385,
      "step": 16500
    },
    {
      "epoch": 2.8854207650273223,
      "grad_norm": 0.6085696746059532,
      "learning_rate": 3.801070651612999e-05,
      "loss": 11.3667,
      "step": 16501
    },
    {
      "epoch": 2.8855956284153006,
      "grad_norm": 0.6841947599704238,
      "learning_rate": 3.79995952826876e-05,
      "loss": 11.7921,
      "step": 16502
    },
    {
      "epoch": 2.885770491803279,
      "grad_norm": 0.7161996703243511,
      "learning_rate": 3.798848529254373e-05,
      "loss": 11.822,
      "step": 16503
    },
    {
      "epoch": 2.885945355191257,
      "grad_norm": 0.6393338210269662,
      "learning_rate": 3.7977376545921084e-05,
      "loss": 11.6905,
      "step": 16504
    },
    {
      "epoch": 2.8861202185792347,
      "grad_norm": 0.6314707598794542,
      "learning_rate": 3.7966269043042445e-05,
      "loss": 11.6735,
      "step": 16505
    },
    {
      "epoch": 2.886295081967213,
      "grad_norm": 0.6205729829621293,
      "learning_rate": 3.7955162784130615e-05,
      "loss": 11.6269,
      "step": 16506
    },
    {
      "epoch": 2.8864699453551914,
      "grad_norm": 0.6277291141790149,
      "learning_rate": 3.794405776940832e-05,
      "loss": 11.8638,
      "step": 16507
    },
    {
      "epoch": 2.8866448087431693,
      "grad_norm": 0.6271352101591423,
      "learning_rate": 3.793295399909815e-05,
      "loss": 11.9437,
      "step": 16508
    },
    {
      "epoch": 2.8868196721311477,
      "grad_norm": 0.6033290432660332,
      "learning_rate": 3.7921851473422834e-05,
      "loss": 11.6995,
      "step": 16509
    },
    {
      "epoch": 2.8869945355191255,
      "grad_norm": 0.6151898261539716,
      "learning_rate": 3.791075019260496e-05,
      "loss": 11.7482,
      "step": 16510
    },
    {
      "epoch": 2.887169398907104,
      "grad_norm": 0.6557178568855094,
      "learning_rate": 3.789965015686722e-05,
      "loss": 11.9481,
      "step": 16511
    },
    {
      "epoch": 2.8873442622950822,
      "grad_norm": 0.6550664356731385,
      "learning_rate": 3.78885513664322e-05,
      "loss": 11.6356,
      "step": 16512
    },
    {
      "epoch": 2.88751912568306,
      "grad_norm": 0.5996253979880866,
      "learning_rate": 3.7877453821522405e-05,
      "loss": 11.9707,
      "step": 16513
    },
    {
      "epoch": 2.887693989071038,
      "grad_norm": 0.6983930682219812,
      "learning_rate": 3.78663575223604e-05,
      "loss": 11.8658,
      "step": 16514
    },
    {
      "epoch": 2.8878688524590164,
      "grad_norm": 0.6626099498557195,
      "learning_rate": 3.785526246916866e-05,
      "loss": 11.8003,
      "step": 16515
    },
    {
      "epoch": 2.8880437158469947,
      "grad_norm": 0.6029713471500587,
      "learning_rate": 3.78441686621698e-05,
      "loss": 11.8472,
      "step": 16516
    },
    {
      "epoch": 2.8882185792349726,
      "grad_norm": 0.6469225835540413,
      "learning_rate": 3.783307610158615e-05,
      "loss": 11.8976,
      "step": 16517
    },
    {
      "epoch": 2.888393442622951,
      "grad_norm": 0.6134012195364534,
      "learning_rate": 3.782198478764021e-05,
      "loss": 11.5709,
      "step": 16518
    },
    {
      "epoch": 2.888568306010929,
      "grad_norm": 0.6066263148995567,
      "learning_rate": 3.781089472055439e-05,
      "loss": 11.5532,
      "step": 16519
    },
    {
      "epoch": 2.888743169398907,
      "grad_norm": 0.5887938753786859,
      "learning_rate": 3.779980590055105e-05,
      "loss": 11.8877,
      "step": 16520
    },
    {
      "epoch": 2.888918032786885,
      "grad_norm": 0.6448566800430462,
      "learning_rate": 3.778871832785262e-05,
      "loss": 11.8109,
      "step": 16521
    },
    {
      "epoch": 2.8890928961748634,
      "grad_norm": 0.6491487100407667,
      "learning_rate": 3.77776320026814e-05,
      "loss": 11.8826,
      "step": 16522
    },
    {
      "epoch": 2.8892677595628413,
      "grad_norm": 0.6858466937297788,
      "learning_rate": 3.776654692525968e-05,
      "loss": 11.8195,
      "step": 16523
    },
    {
      "epoch": 2.8894426229508197,
      "grad_norm": 0.5969822425334969,
      "learning_rate": 3.7755463095809794e-05,
      "loss": 11.8296,
      "step": 16524
    },
    {
      "epoch": 2.889617486338798,
      "grad_norm": 0.6702825958923923,
      "learning_rate": 3.774438051455398e-05,
      "loss": 11.9097,
      "step": 16525
    },
    {
      "epoch": 2.889792349726776,
      "grad_norm": 0.6816260452467925,
      "learning_rate": 3.77332991817145e-05,
      "loss": 11.3053,
      "step": 16526
    },
    {
      "epoch": 2.8899672131147542,
      "grad_norm": 0.6363529948020398,
      "learning_rate": 3.772221909751353e-05,
      "loss": 11.8395,
      "step": 16527
    },
    {
      "epoch": 2.890142076502732,
      "grad_norm": 0.6529044685712618,
      "learning_rate": 3.771114026217331e-05,
      "loss": 11.7255,
      "step": 16528
    },
    {
      "epoch": 2.8903169398907105,
      "grad_norm": 0.7061138566500297,
      "learning_rate": 3.770006267591594e-05,
      "loss": 11.8509,
      "step": 16529
    },
    {
      "epoch": 2.8904918032786884,
      "grad_norm": 0.6597632675814461,
      "learning_rate": 3.768898633896366e-05,
      "loss": 11.7101,
      "step": 16530
    },
    {
      "epoch": 2.8906666666666667,
      "grad_norm": 0.6562950928771916,
      "learning_rate": 3.7677911251538486e-05,
      "loss": 11.8801,
      "step": 16531
    },
    {
      "epoch": 2.8908415300546446,
      "grad_norm": 0.6904809819211949,
      "learning_rate": 3.766683741386255e-05,
      "loss": 12.0361,
      "step": 16532
    },
    {
      "epoch": 2.891016393442623,
      "grad_norm": 0.6478375002319189,
      "learning_rate": 3.765576482615792e-05,
      "loss": 11.9914,
      "step": 16533
    },
    {
      "epoch": 2.8911912568306013,
      "grad_norm": 0.6628412185352697,
      "learning_rate": 3.764469348864662e-05,
      "loss": 11.9993,
      "step": 16534
    },
    {
      "epoch": 2.891366120218579,
      "grad_norm": 0.6108202153186975,
      "learning_rate": 3.7633623401550674e-05,
      "loss": 11.6638,
      "step": 16535
    },
    {
      "epoch": 2.8915409836065575,
      "grad_norm": 0.6740162315084814,
      "learning_rate": 3.762255456509206e-05,
      "loss": 11.4762,
      "step": 16536
    },
    {
      "epoch": 2.8917158469945354,
      "grad_norm": 0.6329553879876801,
      "learning_rate": 3.761148697949276e-05,
      "loss": 11.625,
      "step": 16537
    },
    {
      "epoch": 2.8918907103825138,
      "grad_norm": 0.5960180729827146,
      "learning_rate": 3.760042064497471e-05,
      "loss": 11.8506,
      "step": 16538
    },
    {
      "epoch": 2.8920655737704917,
      "grad_norm": 0.6206819914940619,
      "learning_rate": 3.7589355561759806e-05,
      "loss": 11.7925,
      "step": 16539
    },
    {
      "epoch": 2.89224043715847,
      "grad_norm": 0.5981389569932523,
      "learning_rate": 3.757829173006995e-05,
      "loss": 11.7822,
      "step": 16540
    },
    {
      "epoch": 2.892415300546448,
      "grad_norm": 0.5972965550748379,
      "learning_rate": 3.7567229150127016e-05,
      "loss": 11.7056,
      "step": 16541
    },
    {
      "epoch": 2.8925901639344263,
      "grad_norm": 0.6222395307340183,
      "learning_rate": 3.755616782215282e-05,
      "loss": 11.7322,
      "step": 16542
    },
    {
      "epoch": 2.8927650273224046,
      "grad_norm": 0.6610127143139948,
      "learning_rate": 3.7545107746369194e-05,
      "loss": 12.0111,
      "step": 16543
    },
    {
      "epoch": 2.8929398907103825,
      "grad_norm": 0.6751441379254888,
      "learning_rate": 3.753404892299791e-05,
      "loss": 11.8582,
      "step": 16544
    },
    {
      "epoch": 2.8931147540983604,
      "grad_norm": 0.6604205954224114,
      "learning_rate": 3.752299135226074e-05,
      "loss": 11.895,
      "step": 16545
    },
    {
      "epoch": 2.8932896174863387,
      "grad_norm": 0.6489053389090298,
      "learning_rate": 3.751193503437943e-05,
      "loss": 11.7434,
      "step": 16546
    },
    {
      "epoch": 2.893464480874317,
      "grad_norm": 0.6857462083986603,
      "learning_rate": 3.7500879969575685e-05,
      "loss": 12.0254,
      "step": 16547
    },
    {
      "epoch": 2.893639344262295,
      "grad_norm": 0.7102513742160756,
      "learning_rate": 3.7489826158071194e-05,
      "loss": 11.8,
      "step": 16548
    },
    {
      "epoch": 2.8938142076502733,
      "grad_norm": 0.5709423824925679,
      "learning_rate": 3.7478773600087624e-05,
      "loss": 11.8044,
      "step": 16549
    },
    {
      "epoch": 2.893989071038251,
      "grad_norm": 0.6013470739427017,
      "learning_rate": 3.74677222958466e-05,
      "loss": 11.5422,
      "step": 16550
    },
    {
      "epoch": 2.8941639344262295,
      "grad_norm": 0.6717165237002913,
      "learning_rate": 3.7456672245569746e-05,
      "loss": 11.7649,
      "step": 16551
    },
    {
      "epoch": 2.894338797814208,
      "grad_norm": 0.6692510030360281,
      "learning_rate": 3.744562344947865e-05,
      "loss": 11.8635,
      "step": 16552
    },
    {
      "epoch": 2.894513661202186,
      "grad_norm": 0.6208742045361457,
      "learning_rate": 3.743457590779488e-05,
      "loss": 11.8975,
      "step": 16553
    },
    {
      "epoch": 2.8946885245901637,
      "grad_norm": 0.6712692697677655,
      "learning_rate": 3.742352962073995e-05,
      "loss": 11.5656,
      "step": 16554
    },
    {
      "epoch": 2.894863387978142,
      "grad_norm": 0.5914741019270915,
      "learning_rate": 3.741248458853539e-05,
      "loss": 11.607,
      "step": 16555
    },
    {
      "epoch": 2.8950382513661204,
      "grad_norm": 0.6162288074176112,
      "learning_rate": 3.7401440811402676e-05,
      "loss": 11.6882,
      "step": 16556
    },
    {
      "epoch": 2.8952131147540983,
      "grad_norm": 0.6187807061573612,
      "learning_rate": 3.739039828956329e-05,
      "loss": 11.829,
      "step": 16557
    },
    {
      "epoch": 2.8953879781420766,
      "grad_norm": 0.6408086130710223,
      "learning_rate": 3.7379357023238646e-05,
      "loss": 12.0341,
      "step": 16558
    },
    {
      "epoch": 2.8955628415300545,
      "grad_norm": 0.732953306845618,
      "learning_rate": 3.7368317012650165e-05,
      "loss": 11.8816,
      "step": 16559
    },
    {
      "epoch": 2.895737704918033,
      "grad_norm": 0.6432856642955298,
      "learning_rate": 3.7357278258019235e-05,
      "loss": 11.6038,
      "step": 16560
    },
    {
      "epoch": 2.895912568306011,
      "grad_norm": 0.6592895077370466,
      "learning_rate": 3.734624075956721e-05,
      "loss": 11.7648,
      "step": 16561
    },
    {
      "epoch": 2.896087431693989,
      "grad_norm": 0.5952407275752691,
      "learning_rate": 3.733520451751543e-05,
      "loss": 11.8476,
      "step": 16562
    },
    {
      "epoch": 2.896262295081967,
      "grad_norm": 0.6878697401773683,
      "learning_rate": 3.732416953208522e-05,
      "loss": 11.7687,
      "step": 16563
    },
    {
      "epoch": 2.8964371584699453,
      "grad_norm": 0.6763249555415717,
      "learning_rate": 3.7313135803497835e-05,
      "loss": 11.7696,
      "step": 16564
    },
    {
      "epoch": 2.8966120218579237,
      "grad_norm": 0.6392100290552716,
      "learning_rate": 3.730210333197456e-05,
      "loss": 11.6932,
      "step": 16565
    },
    {
      "epoch": 2.8967868852459016,
      "grad_norm": 0.6479070217995684,
      "learning_rate": 3.7291072117736614e-05,
      "loss": 11.7466,
      "step": 16566
    },
    {
      "epoch": 2.89696174863388,
      "grad_norm": 0.672483623757288,
      "learning_rate": 3.7280042161005246e-05,
      "loss": 11.6827,
      "step": 16567
    },
    {
      "epoch": 2.897136612021858,
      "grad_norm": 0.6482573680121191,
      "learning_rate": 3.7269013462001545e-05,
      "loss": 11.7593,
      "step": 16568
    },
    {
      "epoch": 2.897311475409836,
      "grad_norm": 0.6074149575651948,
      "learning_rate": 3.7257986020946776e-05,
      "loss": 11.6864,
      "step": 16569
    },
    {
      "epoch": 2.8974863387978145,
      "grad_norm": 0.6397494405368715,
      "learning_rate": 3.724695983806202e-05,
      "loss": 11.6608,
      "step": 16570
    },
    {
      "epoch": 2.8976612021857924,
      "grad_norm": 0.6285714143179241,
      "learning_rate": 3.7235934913568395e-05,
      "loss": 11.7651,
      "step": 16571
    },
    {
      "epoch": 2.8978360655737703,
      "grad_norm": 0.6754727019876354,
      "learning_rate": 3.722491124768702e-05,
      "loss": 11.9506,
      "step": 16572
    },
    {
      "epoch": 2.8980109289617486,
      "grad_norm": 0.6864807717947492,
      "learning_rate": 3.7213888840638865e-05,
      "loss": 11.7578,
      "step": 16573
    },
    {
      "epoch": 2.898185792349727,
      "grad_norm": 0.6520385714526805,
      "learning_rate": 3.720286769264504e-05,
      "loss": 11.7074,
      "step": 16574
    },
    {
      "epoch": 2.898360655737705,
      "grad_norm": 0.6759330208280974,
      "learning_rate": 3.7191847803926526e-05,
      "loss": 11.7828,
      "step": 16575
    },
    {
      "epoch": 2.898535519125683,
      "grad_norm": 0.6499937323041345,
      "learning_rate": 3.7180829174704354e-05,
      "loss": 11.483,
      "step": 16576
    },
    {
      "epoch": 2.898710382513661,
      "grad_norm": 0.6553697578825388,
      "learning_rate": 3.716981180519936e-05,
      "loss": 11.643,
      "step": 16577
    },
    {
      "epoch": 2.8988852459016394,
      "grad_norm": 0.6582834889742066,
      "learning_rate": 3.71587956956326e-05,
      "loss": 11.6508,
      "step": 16578
    },
    {
      "epoch": 2.8990601092896173,
      "grad_norm": 0.6406400499757466,
      "learning_rate": 3.714778084622492e-05,
      "loss": 11.6696,
      "step": 16579
    },
    {
      "epoch": 2.8992349726775957,
      "grad_norm": 0.6710225577408018,
      "learning_rate": 3.713676725719723e-05,
      "loss": 11.6627,
      "step": 16580
    },
    {
      "epoch": 2.8994098360655736,
      "grad_norm": 0.6143293652704077,
      "learning_rate": 3.71257549287704e-05,
      "loss": 11.822,
      "step": 16581
    },
    {
      "epoch": 2.899584699453552,
      "grad_norm": 0.7485282339946832,
      "learning_rate": 3.711474386116516e-05,
      "loss": 11.4952,
      "step": 16582
    },
    {
      "epoch": 2.8997595628415302,
      "grad_norm": 0.6468365948446791,
      "learning_rate": 3.710373405460242e-05,
      "loss": 11.5777,
      "step": 16583
    },
    {
      "epoch": 2.899934426229508,
      "grad_norm": 0.6696508354563563,
      "learning_rate": 3.709272550930294e-05,
      "loss": 11.795,
      "step": 16584
    },
    {
      "epoch": 2.9001092896174865,
      "grad_norm": 0.6928978729341282,
      "learning_rate": 3.7081718225487486e-05,
      "loss": 11.8545,
      "step": 16585
    },
    {
      "epoch": 2.9002841530054644,
      "grad_norm": 0.6110075822979114,
      "learning_rate": 3.707071220337675e-05,
      "loss": 11.8777,
      "step": 16586
    },
    {
      "epoch": 2.9004590163934427,
      "grad_norm": 0.6477765126308551,
      "learning_rate": 3.7059707443191395e-05,
      "loss": 11.5621,
      "step": 16587
    },
    {
      "epoch": 2.9006338797814206,
      "grad_norm": 0.6000091460129912,
      "learning_rate": 3.7048703945152205e-05,
      "loss": 11.4564,
      "step": 16588
    },
    {
      "epoch": 2.900808743169399,
      "grad_norm": 0.6063007547670246,
      "learning_rate": 3.7037701709479786e-05,
      "loss": 11.7007,
      "step": 16589
    },
    {
      "epoch": 2.900983606557377,
      "grad_norm": 0.6432943739024995,
      "learning_rate": 3.70267007363948e-05,
      "loss": 11.7288,
      "step": 16590
    },
    {
      "epoch": 2.901158469945355,
      "grad_norm": 0.6185322558968994,
      "learning_rate": 3.701570102611779e-05,
      "loss": 11.5986,
      "step": 16591
    },
    {
      "epoch": 2.9013333333333335,
      "grad_norm": 0.6232941015751713,
      "learning_rate": 3.700470257886932e-05,
      "loss": 11.8294,
      "step": 16592
    },
    {
      "epoch": 2.9015081967213114,
      "grad_norm": 0.6854291313117448,
      "learning_rate": 3.699370539487003e-05,
      "loss": 11.685,
      "step": 16593
    },
    {
      "epoch": 2.90168306010929,
      "grad_norm": 0.6225110648031978,
      "learning_rate": 3.6982709474340414e-05,
      "loss": 11.8503,
      "step": 16594
    },
    {
      "epoch": 2.9018579234972677,
      "grad_norm": 0.6386109717176901,
      "learning_rate": 3.697171481750095e-05,
      "loss": 11.6632,
      "step": 16595
    },
    {
      "epoch": 2.902032786885246,
      "grad_norm": 0.6794939996354987,
      "learning_rate": 3.696072142457211e-05,
      "loss": 11.9361,
      "step": 16596
    },
    {
      "epoch": 2.902207650273224,
      "grad_norm": 0.6699754135871314,
      "learning_rate": 3.694972929577434e-05,
      "loss": 11.8281,
      "step": 16597
    },
    {
      "epoch": 2.9023825136612023,
      "grad_norm": 0.6793231523378607,
      "learning_rate": 3.693873843132811e-05,
      "loss": 12.0989,
      "step": 16598
    },
    {
      "epoch": 2.90255737704918,
      "grad_norm": 0.6270887567836728,
      "learning_rate": 3.6927748831453836e-05,
      "loss": 11.6912,
      "step": 16599
    },
    {
      "epoch": 2.9027322404371585,
      "grad_norm": 0.7222801131991018,
      "learning_rate": 3.6916760496371826e-05,
      "loss": 11.8218,
      "step": 16600
    },
    {
      "epoch": 2.902907103825137,
      "grad_norm": 0.6170570313275462,
      "learning_rate": 3.690577342630243e-05,
      "loss": 11.8228,
      "step": 16601
    },
    {
      "epoch": 2.9030819672131147,
      "grad_norm": 0.6532568310994963,
      "learning_rate": 3.689478762146602e-05,
      "loss": 11.8904,
      "step": 16602
    },
    {
      "epoch": 2.9032568306010926,
      "grad_norm": 0.7114607239711598,
      "learning_rate": 3.6883803082082916e-05,
      "loss": 11.7548,
      "step": 16603
    },
    {
      "epoch": 2.903431693989071,
      "grad_norm": 0.6555467407070281,
      "learning_rate": 3.6872819808373316e-05,
      "loss": 11.6388,
      "step": 16604
    },
    {
      "epoch": 2.9036065573770493,
      "grad_norm": 0.5750432581744043,
      "learning_rate": 3.686183780055751e-05,
      "loss": 11.666,
      "step": 16605
    },
    {
      "epoch": 2.903781420765027,
      "grad_norm": 0.598482162149765,
      "learning_rate": 3.6850857058855683e-05,
      "loss": 11.8764,
      "step": 16606
    },
    {
      "epoch": 2.9039562841530056,
      "grad_norm": 0.6693816501337425,
      "learning_rate": 3.683987758348809e-05,
      "loss": 11.9356,
      "step": 16607
    },
    {
      "epoch": 2.9041311475409834,
      "grad_norm": 0.6989486772530029,
      "learning_rate": 3.682889937467493e-05,
      "loss": 11.8947,
      "step": 16608
    },
    {
      "epoch": 2.904306010928962,
      "grad_norm": 0.5757766470711944,
      "learning_rate": 3.681792243263625e-05,
      "loss": 11.9245,
      "step": 16609
    },
    {
      "epoch": 2.90448087431694,
      "grad_norm": 0.6142143933600543,
      "learning_rate": 3.680694675759222e-05,
      "loss": 11.509,
      "step": 16610
    },
    {
      "epoch": 2.904655737704918,
      "grad_norm": 0.6490976008043778,
      "learning_rate": 3.679597234976291e-05,
      "loss": 11.7024,
      "step": 16611
    },
    {
      "epoch": 2.904830601092896,
      "grad_norm": 0.6507146441798483,
      "learning_rate": 3.678499920936848e-05,
      "loss": 11.8236,
      "step": 16612
    },
    {
      "epoch": 2.9050054644808743,
      "grad_norm": 0.6444464351933207,
      "learning_rate": 3.677402733662887e-05,
      "loss": 11.9238,
      "step": 16613
    },
    {
      "epoch": 2.9051803278688526,
      "grad_norm": 0.6765450197865785,
      "learning_rate": 3.6763056731764143e-05,
      "loss": 12.006,
      "step": 16614
    },
    {
      "epoch": 2.9053551912568305,
      "grad_norm": 0.636985961241045,
      "learning_rate": 3.675208739499431e-05,
      "loss": 11.8854,
      "step": 16615
    },
    {
      "epoch": 2.905530054644809,
      "grad_norm": 0.7017060137740224,
      "learning_rate": 3.674111932653926e-05,
      "loss": 11.8144,
      "step": 16616
    },
    {
      "epoch": 2.9057049180327867,
      "grad_norm": 0.6624330823028514,
      "learning_rate": 3.673015252661909e-05,
      "loss": 11.7957,
      "step": 16617
    },
    {
      "epoch": 2.905879781420765,
      "grad_norm": 0.5930955629018277,
      "learning_rate": 3.671918699545358e-05,
      "loss": 11.4855,
      "step": 16618
    },
    {
      "epoch": 2.9060546448087434,
      "grad_norm": 0.68239999896168,
      "learning_rate": 3.670822273326266e-05,
      "loss": 11.5977,
      "step": 16619
    },
    {
      "epoch": 2.9062295081967213,
      "grad_norm": 0.6047894078424593,
      "learning_rate": 3.669725974026622e-05,
      "loss": 11.5987,
      "step": 16620
    },
    {
      "epoch": 2.906404371584699,
      "grad_norm": 0.6293137341661772,
      "learning_rate": 3.6686298016684085e-05,
      "loss": 11.7148,
      "step": 16621
    },
    {
      "epoch": 2.9065792349726776,
      "grad_norm": 0.6556572869867815,
      "learning_rate": 3.667533756273607e-05,
      "loss": 11.8406,
      "step": 16622
    },
    {
      "epoch": 2.906754098360656,
      "grad_norm": 0.5817815864547723,
      "learning_rate": 3.666437837864196e-05,
      "loss": 11.9739,
      "step": 16623
    },
    {
      "epoch": 2.906928961748634,
      "grad_norm": 0.60952750983337,
      "learning_rate": 3.6653420464621544e-05,
      "loss": 11.7378,
      "step": 16624
    },
    {
      "epoch": 2.907103825136612,
      "grad_norm": 0.6067478356008158,
      "learning_rate": 3.664246382089451e-05,
      "loss": 11.7195,
      "step": 16625
    },
    {
      "epoch": 2.90727868852459,
      "grad_norm": 0.6246231237711627,
      "learning_rate": 3.6631508447680675e-05,
      "loss": 11.8088,
      "step": 16626
    },
    {
      "epoch": 2.9074535519125684,
      "grad_norm": 0.6935075011064974,
      "learning_rate": 3.662055434519963e-05,
      "loss": 11.7429,
      "step": 16627
    },
    {
      "epoch": 2.9076284153005467,
      "grad_norm": 0.6367280257918405,
      "learning_rate": 3.660960151367107e-05,
      "loss": 11.9227,
      "step": 16628
    },
    {
      "epoch": 2.9078032786885246,
      "grad_norm": 0.6409730990159455,
      "learning_rate": 3.6598649953314636e-05,
      "loss": 11.8723,
      "step": 16629
    },
    {
      "epoch": 2.9079781420765025,
      "grad_norm": 0.6081234276492402,
      "learning_rate": 3.658769966434993e-05,
      "loss": 11.7852,
      "step": 16630
    },
    {
      "epoch": 2.908153005464481,
      "grad_norm": 0.5978180709201472,
      "learning_rate": 3.657675064699655e-05,
      "loss": 11.688,
      "step": 16631
    },
    {
      "epoch": 2.908327868852459,
      "grad_norm": 0.6303801825157112,
      "learning_rate": 3.6565802901474066e-05,
      "loss": 11.9813,
      "step": 16632
    },
    {
      "epoch": 2.908502732240437,
      "grad_norm": 0.6130994615889577,
      "learning_rate": 3.655485642800199e-05,
      "loss": 11.8753,
      "step": 16633
    },
    {
      "epoch": 2.9086775956284154,
      "grad_norm": 0.6073842915540797,
      "learning_rate": 3.6543911226799845e-05,
      "loss": 11.603,
      "step": 16634
    },
    {
      "epoch": 2.9088524590163933,
      "grad_norm": 0.5830972370175849,
      "learning_rate": 3.653296729808712e-05,
      "loss": 11.6925,
      "step": 16635
    },
    {
      "epoch": 2.9090273224043717,
      "grad_norm": 0.6308238705295771,
      "learning_rate": 3.6522024642083276e-05,
      "loss": 11.5576,
      "step": 16636
    },
    {
      "epoch": 2.9092021857923496,
      "grad_norm": 0.5839671448279972,
      "learning_rate": 3.651108325900773e-05,
      "loss": 11.8569,
      "step": 16637
    },
    {
      "epoch": 2.909377049180328,
      "grad_norm": 0.5988971520068752,
      "learning_rate": 3.6500143149079903e-05,
      "loss": 11.9234,
      "step": 16638
    },
    {
      "epoch": 2.909551912568306,
      "grad_norm": 0.6714891576544337,
      "learning_rate": 3.6489204312519176e-05,
      "loss": 11.61,
      "step": 16639
    },
    {
      "epoch": 2.909726775956284,
      "grad_norm": 0.6359511350100105,
      "learning_rate": 3.647826674954491e-05,
      "loss": 11.6276,
      "step": 16640
    },
    {
      "epoch": 2.9099016393442625,
      "grad_norm": 0.6240423546079208,
      "learning_rate": 3.646733046037644e-05,
      "loss": 11.8924,
      "step": 16641
    },
    {
      "epoch": 2.9100765027322404,
      "grad_norm": 0.6472683726366886,
      "learning_rate": 3.645639544523305e-05,
      "loss": 11.764,
      "step": 16642
    },
    {
      "epoch": 2.9102513661202187,
      "grad_norm": 0.6942355775685155,
      "learning_rate": 3.644546170433404e-05,
      "loss": 11.728,
      "step": 16643
    },
    {
      "epoch": 2.9104262295081966,
      "grad_norm": 0.6374916792775365,
      "learning_rate": 3.643452923789866e-05,
      "loss": 11.7461,
      "step": 16644
    },
    {
      "epoch": 2.910601092896175,
      "grad_norm": 0.7003742673499647,
      "learning_rate": 3.6423598046146154e-05,
      "loss": 11.7068,
      "step": 16645
    },
    {
      "epoch": 2.910775956284153,
      "grad_norm": 0.6681840595851455,
      "learning_rate": 3.64126681292957e-05,
      "loss": 11.6969,
      "step": 16646
    },
    {
      "epoch": 2.910950819672131,
      "grad_norm": 0.6885114454062836,
      "learning_rate": 3.640173948756649e-05,
      "loss": 11.605,
      "step": 16647
    },
    {
      "epoch": 2.911125683060109,
      "grad_norm": 0.6573680904081546,
      "learning_rate": 3.6390812121177675e-05,
      "loss": 11.8562,
      "step": 16648
    },
    {
      "epoch": 2.9113005464480874,
      "grad_norm": 0.682429677739206,
      "learning_rate": 3.6379886030348386e-05,
      "loss": 11.7527,
      "step": 16649
    },
    {
      "epoch": 2.911475409836066,
      "grad_norm": 0.6729733702834652,
      "learning_rate": 3.636896121529773e-05,
      "loss": 11.742,
      "step": 16650
    },
    {
      "epoch": 2.9116502732240437,
      "grad_norm": 0.6638115690226946,
      "learning_rate": 3.635803767624476e-05,
      "loss": 11.8278,
      "step": 16651
    },
    {
      "epoch": 2.9118251366120216,
      "grad_norm": 0.6444078157306365,
      "learning_rate": 3.634711541340854e-05,
      "loss": 11.641,
      "step": 16652
    },
    {
      "epoch": 2.912,
      "grad_norm": 0.6812793772414266,
      "learning_rate": 3.633619442700811e-05,
      "loss": 11.8061,
      "step": 16653
    },
    {
      "epoch": 2.9121748633879783,
      "grad_norm": 0.6606855925565066,
      "learning_rate": 3.6325274717262436e-05,
      "loss": 11.831,
      "step": 16654
    },
    {
      "epoch": 2.912349726775956,
      "grad_norm": 0.6295334053757292,
      "learning_rate": 3.6314356284390525e-05,
      "loss": 11.6358,
      "step": 16655
    },
    {
      "epoch": 2.9125245901639345,
      "grad_norm": 0.7034401731131738,
      "learning_rate": 3.63034391286113e-05,
      "loss": 11.8519,
      "step": 16656
    },
    {
      "epoch": 2.9126994535519124,
      "grad_norm": 0.6673106078315868,
      "learning_rate": 3.6292523250143705e-05,
      "loss": 11.833,
      "step": 16657
    },
    {
      "epoch": 2.9128743169398907,
      "grad_norm": 0.5909561516227357,
      "learning_rate": 3.628160864920663e-05,
      "loss": 11.8526,
      "step": 16658
    },
    {
      "epoch": 2.913049180327869,
      "grad_norm": 0.6556202525092902,
      "learning_rate": 3.6270695326018934e-05,
      "loss": 11.7875,
      "step": 16659
    },
    {
      "epoch": 2.913224043715847,
      "grad_norm": 0.6573276273815282,
      "learning_rate": 3.6259783280799474e-05,
      "loss": 11.7342,
      "step": 16660
    },
    {
      "epoch": 2.913398907103825,
      "grad_norm": 0.6723851635955839,
      "learning_rate": 3.6248872513767076e-05,
      "loss": 11.8297,
      "step": 16661
    },
    {
      "epoch": 2.913573770491803,
      "grad_norm": 0.6645683509017782,
      "learning_rate": 3.623796302514051e-05,
      "loss": 11.8126,
      "step": 16662
    },
    {
      "epoch": 2.9137486338797816,
      "grad_norm": 0.6160933073063162,
      "learning_rate": 3.6227054815138604e-05,
      "loss": 11.8261,
      "step": 16663
    },
    {
      "epoch": 2.9139234972677595,
      "grad_norm": 0.6359189660183334,
      "learning_rate": 3.621614788397999e-05,
      "loss": 11.7142,
      "step": 16664
    },
    {
      "epoch": 2.914098360655738,
      "grad_norm": 0.6494064641312277,
      "learning_rate": 3.620524223188349e-05,
      "loss": 11.7789,
      "step": 16665
    },
    {
      "epoch": 2.9142732240437157,
      "grad_norm": 0.6142870515601846,
      "learning_rate": 3.619433785906775e-05,
      "loss": 11.4964,
      "step": 16666
    },
    {
      "epoch": 2.914448087431694,
      "grad_norm": 0.6014393861680095,
      "learning_rate": 3.6183434765751445e-05,
      "loss": 11.6041,
      "step": 16667
    },
    {
      "epoch": 2.9146229508196724,
      "grad_norm": 0.657372563426173,
      "learning_rate": 3.617253295215325e-05,
      "loss": 11.9378,
      "step": 16668
    },
    {
      "epoch": 2.9147978142076503,
      "grad_norm": 0.6676537338578916,
      "learning_rate": 3.616163241849169e-05,
      "loss": 11.5698,
      "step": 16669
    },
    {
      "epoch": 2.914972677595628,
      "grad_norm": 0.6341128930938014,
      "learning_rate": 3.6150733164985426e-05,
      "loss": 11.902,
      "step": 16670
    },
    {
      "epoch": 2.9151475409836065,
      "grad_norm": 0.6104626949502014,
      "learning_rate": 3.613983519185301e-05,
      "loss": 11.8414,
      "step": 16671
    },
    {
      "epoch": 2.915322404371585,
      "grad_norm": 0.6207345132221563,
      "learning_rate": 3.6128938499313017e-05,
      "loss": 11.5011,
      "step": 16672
    },
    {
      "epoch": 2.9154972677595627,
      "grad_norm": 0.6736179484288681,
      "learning_rate": 3.6118043087583843e-05,
      "loss": 11.8132,
      "step": 16673
    },
    {
      "epoch": 2.915672131147541,
      "grad_norm": 0.634750431262872,
      "learning_rate": 3.6107148956884084e-05,
      "loss": 11.8114,
      "step": 16674
    },
    {
      "epoch": 2.915846994535519,
      "grad_norm": 0.6575766015373465,
      "learning_rate": 3.609625610743217e-05,
      "loss": 11.8907,
      "step": 16675
    },
    {
      "epoch": 2.9160218579234973,
      "grad_norm": 0.6184229412724295,
      "learning_rate": 3.608536453944652e-05,
      "loss": 11.9553,
      "step": 16676
    },
    {
      "epoch": 2.9161967213114757,
      "grad_norm": 0.6292386172946727,
      "learning_rate": 3.60744742531456e-05,
      "loss": 11.8791,
      "step": 16677
    },
    {
      "epoch": 2.9163715846994536,
      "grad_norm": 0.6759907982022336,
      "learning_rate": 3.606358524874767e-05,
      "loss": 12.0312,
      "step": 16678
    },
    {
      "epoch": 2.9165464480874315,
      "grad_norm": 0.6745619542590819,
      "learning_rate": 3.60526975264712e-05,
      "loss": 11.5767,
      "step": 16679
    },
    {
      "epoch": 2.91672131147541,
      "grad_norm": 0.6509959272417943,
      "learning_rate": 3.604181108653449e-05,
      "loss": 11.6218,
      "step": 16680
    },
    {
      "epoch": 2.916896174863388,
      "grad_norm": 0.6220467894436196,
      "learning_rate": 3.603092592915589e-05,
      "loss": 11.4691,
      "step": 16681
    },
    {
      "epoch": 2.917071038251366,
      "grad_norm": 0.6663066810905054,
      "learning_rate": 3.6020042054553596e-05,
      "loss": 11.8881,
      "step": 16682
    },
    {
      "epoch": 2.9172459016393444,
      "grad_norm": 0.6502679874603741,
      "learning_rate": 3.600915946294587e-05,
      "loss": 11.6947,
      "step": 16683
    },
    {
      "epoch": 2.9174207650273223,
      "grad_norm": 0.6277702592909432,
      "learning_rate": 3.5998278154551004e-05,
      "loss": 11.6122,
      "step": 16684
    },
    {
      "epoch": 2.9175956284153006,
      "grad_norm": 0.6440594201551179,
      "learning_rate": 3.598739812958718e-05,
      "loss": 11.8797,
      "step": 16685
    },
    {
      "epoch": 2.917770491803279,
      "grad_norm": 0.6799478753734127,
      "learning_rate": 3.597651938827259e-05,
      "loss": 11.5331,
      "step": 16686
    },
    {
      "epoch": 2.917945355191257,
      "grad_norm": 0.6561374571006607,
      "learning_rate": 3.596564193082534e-05,
      "loss": 11.8043,
      "step": 16687
    },
    {
      "epoch": 2.9181202185792348,
      "grad_norm": 0.6070307347398926,
      "learning_rate": 3.595476575746355e-05,
      "loss": 11.9148,
      "step": 16688
    },
    {
      "epoch": 2.918295081967213,
      "grad_norm": 0.6186208118313893,
      "learning_rate": 3.594389086840537e-05,
      "loss": 11.8319,
      "step": 16689
    },
    {
      "epoch": 2.9184699453551914,
      "grad_norm": 0.6245130482092477,
      "learning_rate": 3.593301726386889e-05,
      "loss": 11.4225,
      "step": 16690
    },
    {
      "epoch": 2.9186448087431693,
      "grad_norm": 0.6451772352197263,
      "learning_rate": 3.592214494407211e-05,
      "loss": 11.7237,
      "step": 16691
    },
    {
      "epoch": 2.9188196721311477,
      "grad_norm": 0.5944417035009981,
      "learning_rate": 3.591127390923306e-05,
      "loss": 11.6038,
      "step": 16692
    },
    {
      "epoch": 2.9189945355191256,
      "grad_norm": 0.6501546367323447,
      "learning_rate": 3.5900404159569715e-05,
      "loss": 11.6477,
      "step": 16693
    },
    {
      "epoch": 2.919169398907104,
      "grad_norm": 0.6304635547631154,
      "learning_rate": 3.5889535695300105e-05,
      "loss": 11.6698,
      "step": 16694
    },
    {
      "epoch": 2.919344262295082,
      "grad_norm": 0.6017416964894944,
      "learning_rate": 3.587866851664219e-05,
      "loss": 11.7853,
      "step": 16695
    },
    {
      "epoch": 2.91951912568306,
      "grad_norm": 0.6289667474279806,
      "learning_rate": 3.586780262381381e-05,
      "loss": 11.4183,
      "step": 16696
    },
    {
      "epoch": 2.919693989071038,
      "grad_norm": 0.6831949823929384,
      "learning_rate": 3.585693801703287e-05,
      "loss": 11.748,
      "step": 16697
    },
    {
      "epoch": 2.9198688524590164,
      "grad_norm": 0.6604231098680261,
      "learning_rate": 3.58460746965173e-05,
      "loss": 11.8328,
      "step": 16698
    },
    {
      "epoch": 2.9200437158469947,
      "grad_norm": 0.652396170637515,
      "learning_rate": 3.583521266248494e-05,
      "loss": 11.5154,
      "step": 16699
    },
    {
      "epoch": 2.9202185792349726,
      "grad_norm": 0.7761803031645593,
      "learning_rate": 3.582435191515354e-05,
      "loss": 11.7649,
      "step": 16700
    },
    {
      "epoch": 2.920393442622951,
      "grad_norm": 0.6361296641360981,
      "learning_rate": 3.581349245474095e-05,
      "loss": 11.8572,
      "step": 16701
    },
    {
      "epoch": 2.920568306010929,
      "grad_norm": 0.6170933125958549,
      "learning_rate": 3.580263428146486e-05,
      "loss": 11.5672,
      "step": 16702
    },
    {
      "epoch": 2.920743169398907,
      "grad_norm": 0.70233859452884,
      "learning_rate": 3.579177739554311e-05,
      "loss": 11.7113,
      "step": 16703
    },
    {
      "epoch": 2.920918032786885,
      "grad_norm": 0.7009015469011226,
      "learning_rate": 3.57809217971934e-05,
      "loss": 11.72,
      "step": 16704
    },
    {
      "epoch": 2.9210928961748635,
      "grad_norm": 0.6371229818696286,
      "learning_rate": 3.577006748663335e-05,
      "loss": 11.8758,
      "step": 16705
    },
    {
      "epoch": 2.9212677595628413,
      "grad_norm": 0.607284454551943,
      "learning_rate": 3.575921446408066e-05,
      "loss": 11.4388,
      "step": 16706
    },
    {
      "epoch": 2.9214426229508197,
      "grad_norm": 0.6600441310333053,
      "learning_rate": 3.574836272975293e-05,
      "loss": 11.8377,
      "step": 16707
    },
    {
      "epoch": 2.921617486338798,
      "grad_norm": 0.6598805284704213,
      "learning_rate": 3.573751228386787e-05,
      "loss": 11.8283,
      "step": 16708
    },
    {
      "epoch": 2.921792349726776,
      "grad_norm": 0.6444993266482827,
      "learning_rate": 3.5726663126642976e-05,
      "loss": 11.7891,
      "step": 16709
    },
    {
      "epoch": 2.921967213114754,
      "grad_norm": 0.617351849809399,
      "learning_rate": 3.571581525829583e-05,
      "loss": 11.8222,
      "step": 16710
    },
    {
      "epoch": 2.922142076502732,
      "grad_norm": 0.6378054242226192,
      "learning_rate": 3.570496867904396e-05,
      "loss": 11.9047,
      "step": 16711
    },
    {
      "epoch": 2.9223169398907105,
      "grad_norm": 0.6751280503879816,
      "learning_rate": 3.569412338910485e-05,
      "loss": 11.8084,
      "step": 16712
    },
    {
      "epoch": 2.9224918032786884,
      "grad_norm": 0.6812223809012501,
      "learning_rate": 3.568327938869608e-05,
      "loss": 11.759,
      "step": 16713
    },
    {
      "epoch": 2.9226666666666667,
      "grad_norm": 0.6936488010262933,
      "learning_rate": 3.567243667803499e-05,
      "loss": 11.7036,
      "step": 16714
    },
    {
      "epoch": 2.9228415300546446,
      "grad_norm": 0.6267049482328406,
      "learning_rate": 3.5661595257339076e-05,
      "loss": 11.7792,
      "step": 16715
    },
    {
      "epoch": 2.923016393442623,
      "grad_norm": 0.6605283477517617,
      "learning_rate": 3.5650755126825706e-05,
      "loss": 11.9602,
      "step": 16716
    },
    {
      "epoch": 2.9231912568306013,
      "grad_norm": 0.6489436696000278,
      "learning_rate": 3.5639916286712285e-05,
      "loss": 11.7864,
      "step": 16717
    },
    {
      "epoch": 2.9233661202185792,
      "grad_norm": 0.7517929701498162,
      "learning_rate": 3.562907873721615e-05,
      "loss": 11.844,
      "step": 16718
    },
    {
      "epoch": 2.923540983606557,
      "grad_norm": 0.6994934042268816,
      "learning_rate": 3.561824247855462e-05,
      "loss": 11.6877,
      "step": 16719
    },
    {
      "epoch": 2.9237158469945355,
      "grad_norm": 0.6590717342627201,
      "learning_rate": 3.5607407510945025e-05,
      "loss": 11.5935,
      "step": 16720
    },
    {
      "epoch": 2.923890710382514,
      "grad_norm": 0.7179144158068484,
      "learning_rate": 3.559657383460457e-05,
      "loss": 11.9244,
      "step": 16721
    },
    {
      "epoch": 2.9240655737704917,
      "grad_norm": 0.618456646881728,
      "learning_rate": 3.558574144975063e-05,
      "loss": 11.5362,
      "step": 16722
    },
    {
      "epoch": 2.92424043715847,
      "grad_norm": 0.7522850827902013,
      "learning_rate": 3.557491035660032e-05,
      "loss": 11.7886,
      "step": 16723
    },
    {
      "epoch": 2.924415300546448,
      "grad_norm": 0.5962732656395031,
      "learning_rate": 3.556408055537087e-05,
      "loss": 11.6267,
      "step": 16724
    },
    {
      "epoch": 2.9245901639344263,
      "grad_norm": 0.6307238646084875,
      "learning_rate": 3.555325204627944e-05,
      "loss": 11.6155,
      "step": 16725
    },
    {
      "epoch": 2.9247650273224046,
      "grad_norm": 0.6703738231829847,
      "learning_rate": 3.55424248295432e-05,
      "loss": 11.8725,
      "step": 16726
    },
    {
      "epoch": 2.9249398907103825,
      "grad_norm": 0.6584886846522823,
      "learning_rate": 3.553159890537924e-05,
      "loss": 11.8832,
      "step": 16727
    },
    {
      "epoch": 2.9251147540983604,
      "grad_norm": 0.6620184688567746,
      "learning_rate": 3.5520774274004665e-05,
      "loss": 11.9044,
      "step": 16728
    },
    {
      "epoch": 2.9252896174863388,
      "grad_norm": 0.5733843401339791,
      "learning_rate": 3.550995093563654e-05,
      "loss": 11.7218,
      "step": 16729
    },
    {
      "epoch": 2.925464480874317,
      "grad_norm": 0.6747636095699677,
      "learning_rate": 3.5499128890491914e-05,
      "loss": 11.814,
      "step": 16730
    },
    {
      "epoch": 2.925639344262295,
      "grad_norm": 0.6362939555230557,
      "learning_rate": 3.5488308138787804e-05,
      "loss": 11.9001,
      "step": 16731
    },
    {
      "epoch": 2.9258142076502733,
      "grad_norm": 0.6379215280242699,
      "learning_rate": 3.547748868074119e-05,
      "loss": 11.8634,
      "step": 16732
    },
    {
      "epoch": 2.9259890710382512,
      "grad_norm": 0.6544551627011121,
      "learning_rate": 3.5466670516569035e-05,
      "loss": 11.9074,
      "step": 16733
    },
    {
      "epoch": 2.9261639344262296,
      "grad_norm": 0.6708065506890244,
      "learning_rate": 3.545585364648828e-05,
      "loss": 11.643,
      "step": 16734
    },
    {
      "epoch": 2.926338797814208,
      "grad_norm": 0.7290651665132747,
      "learning_rate": 3.544503807071583e-05,
      "loss": 11.8363,
      "step": 16735
    },
    {
      "epoch": 2.926513661202186,
      "grad_norm": 0.6482791467924007,
      "learning_rate": 3.543422378946858e-05,
      "loss": 11.8518,
      "step": 16736
    },
    {
      "epoch": 2.9266885245901637,
      "grad_norm": 0.7673429568516512,
      "learning_rate": 3.54234108029634e-05,
      "loss": 11.8893,
      "step": 16737
    },
    {
      "epoch": 2.926863387978142,
      "grad_norm": 0.6323214774993021,
      "learning_rate": 3.5412599111417085e-05,
      "loss": 11.7867,
      "step": 16738
    },
    {
      "epoch": 2.9270382513661204,
      "grad_norm": 0.641927491956161,
      "learning_rate": 3.540178871504648e-05,
      "loss": 11.6398,
      "step": 16739
    },
    {
      "epoch": 2.9272131147540983,
      "grad_norm": 0.6251079535739597,
      "learning_rate": 3.5390979614068356e-05,
      "loss": 11.7877,
      "step": 16740
    },
    {
      "epoch": 2.9273879781420766,
      "grad_norm": 0.6141413048039382,
      "learning_rate": 3.5380171808699466e-05,
      "loss": 11.6195,
      "step": 16741
    },
    {
      "epoch": 2.9275628415300545,
      "grad_norm": 0.6778635818646029,
      "learning_rate": 3.5369365299156545e-05,
      "loss": 11.6238,
      "step": 16742
    },
    {
      "epoch": 2.927737704918033,
      "grad_norm": 0.6168890521721568,
      "learning_rate": 3.5358560085656276e-05,
      "loss": 11.9671,
      "step": 16743
    },
    {
      "epoch": 2.9279125683060108,
      "grad_norm": 0.6241089717663799,
      "learning_rate": 3.534775616841537e-05,
      "loss": 11.669,
      "step": 16744
    },
    {
      "epoch": 2.928087431693989,
      "grad_norm": 0.617508480178979,
      "learning_rate": 3.5336953547650455e-05,
      "loss": 11.8299,
      "step": 16745
    },
    {
      "epoch": 2.928262295081967,
      "grad_norm": 0.6470309184229641,
      "learning_rate": 3.532615222357816e-05,
      "loss": 11.7597,
      "step": 16746
    },
    {
      "epoch": 2.9284371584699453,
      "grad_norm": 0.7174510400494155,
      "learning_rate": 3.531535219641509e-05,
      "loss": 11.7361,
      "step": 16747
    },
    {
      "epoch": 2.9286120218579237,
      "grad_norm": 0.6530090178753168,
      "learning_rate": 3.530455346637781e-05,
      "loss": 11.4574,
      "step": 16748
    },
    {
      "epoch": 2.9287868852459016,
      "grad_norm": 0.6022499541357899,
      "learning_rate": 3.529375603368288e-05,
      "loss": 11.683,
      "step": 16749
    },
    {
      "epoch": 2.92896174863388,
      "grad_norm": 0.6643871351144579,
      "learning_rate": 3.5282959898546817e-05,
      "loss": 11.7072,
      "step": 16750
    },
    {
      "epoch": 2.929136612021858,
      "grad_norm": 0.7457427500723857,
      "learning_rate": 3.527216506118611e-05,
      "loss": 11.7469,
      "step": 16751
    },
    {
      "epoch": 2.929311475409836,
      "grad_norm": 0.6644032457541119,
      "learning_rate": 3.5261371521817244e-05,
      "loss": 11.8212,
      "step": 16752
    },
    {
      "epoch": 2.929486338797814,
      "grad_norm": 0.6290942908261333,
      "learning_rate": 3.5250579280656636e-05,
      "loss": 11.693,
      "step": 16753
    },
    {
      "epoch": 2.9296612021857924,
      "grad_norm": 0.649029026365476,
      "learning_rate": 3.523978833792073e-05,
      "loss": 11.5504,
      "step": 16754
    },
    {
      "epoch": 2.9298360655737703,
      "grad_norm": 0.6720080125123525,
      "learning_rate": 3.522899869382591e-05,
      "loss": 11.6821,
      "step": 16755
    },
    {
      "epoch": 2.9300109289617486,
      "grad_norm": 0.6874834650554981,
      "learning_rate": 3.5218210348588544e-05,
      "loss": 11.8381,
      "step": 16756
    },
    {
      "epoch": 2.930185792349727,
      "grad_norm": 0.6471202994886467,
      "learning_rate": 3.5207423302424956e-05,
      "loss": 11.7611,
      "step": 16757
    },
    {
      "epoch": 2.930360655737705,
      "grad_norm": 0.6659017059692849,
      "learning_rate": 3.5196637555551484e-05,
      "loss": 11.5802,
      "step": 16758
    },
    {
      "epoch": 2.930535519125683,
      "grad_norm": 0.621918221499862,
      "learning_rate": 3.518585310818442e-05,
      "loss": 11.8454,
      "step": 16759
    },
    {
      "epoch": 2.930710382513661,
      "grad_norm": 0.6609461564009449,
      "learning_rate": 3.517506996053994e-05,
      "loss": 11.9548,
      "step": 16760
    },
    {
      "epoch": 2.9308852459016395,
      "grad_norm": 0.6608197796689458,
      "learning_rate": 3.516428811283439e-05,
      "loss": 11.8658,
      "step": 16761
    },
    {
      "epoch": 2.9310601092896174,
      "grad_norm": 0.7541073965403698,
      "learning_rate": 3.5153507565283925e-05,
      "loss": 11.6423,
      "step": 16762
    },
    {
      "epoch": 2.9312349726775957,
      "grad_norm": 0.655391983906339,
      "learning_rate": 3.514272831810474e-05,
      "loss": 11.8831,
      "step": 16763
    },
    {
      "epoch": 2.9314098360655736,
      "grad_norm": 0.655278841875019,
      "learning_rate": 3.513195037151303e-05,
      "loss": 11.908,
      "step": 16764
    },
    {
      "epoch": 2.931584699453552,
      "grad_norm": 0.6046665929809043,
      "learning_rate": 3.5121173725724823e-05,
      "loss": 11.3912,
      "step": 16765
    },
    {
      "epoch": 2.9317595628415303,
      "grad_norm": 0.6566989893563369,
      "learning_rate": 3.511039838095631e-05,
      "loss": 11.9282,
      "step": 16766
    },
    {
      "epoch": 2.931934426229508,
      "grad_norm": 0.6206943660724551,
      "learning_rate": 3.5099624337423554e-05,
      "loss": 11.7525,
      "step": 16767
    },
    {
      "epoch": 2.932109289617486,
      "grad_norm": 0.7556406869462464,
      "learning_rate": 3.508885159534263e-05,
      "loss": 12.0128,
      "step": 16768
    },
    {
      "epoch": 2.9322841530054644,
      "grad_norm": 0.6537907446406347,
      "learning_rate": 3.507808015492946e-05,
      "loss": 11.5067,
      "step": 16769
    },
    {
      "epoch": 2.9324590163934428,
      "grad_norm": 0.6023887257750412,
      "learning_rate": 3.506731001640017e-05,
      "loss": 11.5247,
      "step": 16770
    },
    {
      "epoch": 2.9326338797814206,
      "grad_norm": 0.6346774345148185,
      "learning_rate": 3.505654117997067e-05,
      "loss": 11.574,
      "step": 16771
    },
    {
      "epoch": 2.932808743169399,
      "grad_norm": 0.6492181847737207,
      "learning_rate": 3.5045773645856925e-05,
      "loss": 11.965,
      "step": 16772
    },
    {
      "epoch": 2.932983606557377,
      "grad_norm": 0.6370330994224993,
      "learning_rate": 3.503500741427489e-05,
      "loss": 11.8364,
      "step": 16773
    },
    {
      "epoch": 2.9331584699453552,
      "grad_norm": 0.5968607964114935,
      "learning_rate": 3.502424248544036e-05,
      "loss": 11.6147,
      "step": 16774
    },
    {
      "epoch": 2.9333333333333336,
      "grad_norm": 0.6454463067384958,
      "learning_rate": 3.501347885956929e-05,
      "loss": 11.6279,
      "step": 16775
    },
    {
      "epoch": 2.9335081967213115,
      "grad_norm": 0.6615686056980666,
      "learning_rate": 3.500271653687752e-05,
      "loss": 11.9259,
      "step": 16776
    },
    {
      "epoch": 2.9336830601092894,
      "grad_norm": 0.6961198782277371,
      "learning_rate": 3.4991955517580875e-05,
      "loss": 11.3634,
      "step": 16777
    },
    {
      "epoch": 2.9338579234972677,
      "grad_norm": 0.622645434183216,
      "learning_rate": 3.498119580189509e-05,
      "loss": 11.4577,
      "step": 16778
    },
    {
      "epoch": 2.934032786885246,
      "grad_norm": 0.6352309457621048,
      "learning_rate": 3.497043739003594e-05,
      "loss": 11.5998,
      "step": 16779
    },
    {
      "epoch": 2.934207650273224,
      "grad_norm": 0.6446871048445036,
      "learning_rate": 3.4959680282219223e-05,
      "loss": 11.6017,
      "step": 16780
    },
    {
      "epoch": 2.9343825136612023,
      "grad_norm": 0.6258435075340046,
      "learning_rate": 3.4948924478660594e-05,
      "loss": 11.7265,
      "step": 16781
    },
    {
      "epoch": 2.93455737704918,
      "grad_norm": 0.5865172232067838,
      "learning_rate": 3.493816997957582e-05,
      "loss": 11.8637,
      "step": 16782
    },
    {
      "epoch": 2.9347322404371585,
      "grad_norm": 0.6884531550488077,
      "learning_rate": 3.492741678518046e-05,
      "loss": 11.6243,
      "step": 16783
    },
    {
      "epoch": 2.934907103825137,
      "grad_norm": 0.630541480272886,
      "learning_rate": 3.491666489569016e-05,
      "loss": 11.6499,
      "step": 16784
    },
    {
      "epoch": 2.9350819672131148,
      "grad_norm": 0.5992081101538426,
      "learning_rate": 3.490591431132059e-05,
      "loss": 11.8382,
      "step": 16785
    },
    {
      "epoch": 2.9352568306010927,
      "grad_norm": 0.693837957734071,
      "learning_rate": 3.489516503228733e-05,
      "loss": 11.7719,
      "step": 16786
    },
    {
      "epoch": 2.935431693989071,
      "grad_norm": 0.7077727966516814,
      "learning_rate": 3.4884417058805887e-05,
      "loss": 11.7496,
      "step": 16787
    },
    {
      "epoch": 2.9356065573770493,
      "grad_norm": 0.5854828137373724,
      "learning_rate": 3.487367039109182e-05,
      "loss": 11.4058,
      "step": 16788
    },
    {
      "epoch": 2.9357814207650272,
      "grad_norm": 0.6285178719051453,
      "learning_rate": 3.486292502936057e-05,
      "loss": 11.6701,
      "step": 16789
    },
    {
      "epoch": 2.9359562841530056,
      "grad_norm": 0.6318819463592321,
      "learning_rate": 3.485218097382772e-05,
      "loss": 11.7086,
      "step": 16790
    },
    {
      "epoch": 2.9361311475409835,
      "grad_norm": 0.638352652490743,
      "learning_rate": 3.484143822470871e-05,
      "loss": 11.8123,
      "step": 16791
    },
    {
      "epoch": 2.936306010928962,
      "grad_norm": 0.6139340279422922,
      "learning_rate": 3.483069678221889e-05,
      "loss": 11.7528,
      "step": 16792
    },
    {
      "epoch": 2.93648087431694,
      "grad_norm": 0.6557095051218632,
      "learning_rate": 3.481995664657365e-05,
      "loss": 11.8616,
      "step": 16793
    },
    {
      "epoch": 2.936655737704918,
      "grad_norm": 0.6166383093259873,
      "learning_rate": 3.480921781798845e-05,
      "loss": 11.873,
      "step": 16794
    },
    {
      "epoch": 2.936830601092896,
      "grad_norm": 0.6684286544020996,
      "learning_rate": 3.4798480296678624e-05,
      "loss": 11.7462,
      "step": 16795
    },
    {
      "epoch": 2.9370054644808743,
      "grad_norm": 0.6166356707869461,
      "learning_rate": 3.478774408285943e-05,
      "loss": 11.7761,
      "step": 16796
    },
    {
      "epoch": 2.9371803278688526,
      "grad_norm": 0.6755834226963209,
      "learning_rate": 3.47770091767462e-05,
      "loss": 11.698,
      "step": 16797
    },
    {
      "epoch": 2.9373551912568305,
      "grad_norm": 0.6529932598791405,
      "learning_rate": 3.476627557855416e-05,
      "loss": 12.01,
      "step": 16798
    },
    {
      "epoch": 2.937530054644809,
      "grad_norm": 0.7014669933280463,
      "learning_rate": 3.475554328849863e-05,
      "loss": 11.9375,
      "step": 16799
    },
    {
      "epoch": 2.9377049180327868,
      "grad_norm": 0.5804777991672592,
      "learning_rate": 3.4744812306794804e-05,
      "loss": 11.6891,
      "step": 16800
    },
    {
      "epoch": 2.937879781420765,
      "grad_norm": 0.6090887071188572,
      "learning_rate": 3.473408263365783e-05,
      "loss": 11.785,
      "step": 16801
    },
    {
      "epoch": 2.938054644808743,
      "grad_norm": 0.6730941064379454,
      "learning_rate": 3.4723354269302885e-05,
      "loss": 11.9773,
      "step": 16802
    },
    {
      "epoch": 2.9382295081967214,
      "grad_norm": 0.6816528973927077,
      "learning_rate": 3.4712627213945084e-05,
      "loss": 11.6337,
      "step": 16803
    },
    {
      "epoch": 2.9384043715846992,
      "grad_norm": 0.5661419928440922,
      "learning_rate": 3.4701901467799624e-05,
      "loss": 11.3239,
      "step": 16804
    },
    {
      "epoch": 2.9385792349726776,
      "grad_norm": 0.6403586220889816,
      "learning_rate": 3.46911770310815e-05,
      "loss": 11.9496,
      "step": 16805
    },
    {
      "epoch": 2.938754098360656,
      "grad_norm": 0.6165240179032732,
      "learning_rate": 3.4680453904005805e-05,
      "loss": 11.8143,
      "step": 16806
    },
    {
      "epoch": 2.938928961748634,
      "grad_norm": 0.6348025980340223,
      "learning_rate": 3.4669732086787564e-05,
      "loss": 11.4168,
      "step": 16807
    },
    {
      "epoch": 2.939103825136612,
      "grad_norm": 0.5971662405458763,
      "learning_rate": 3.4659011579641754e-05,
      "loss": 11.8452,
      "step": 16808
    },
    {
      "epoch": 2.93927868852459,
      "grad_norm": 0.8940082480984838,
      "learning_rate": 3.4648292382783455e-05,
      "loss": 11.7235,
      "step": 16809
    },
    {
      "epoch": 2.9394535519125684,
      "grad_norm": 0.6397142519747883,
      "learning_rate": 3.46375744964275e-05,
      "loss": 11.3157,
      "step": 16810
    },
    {
      "epoch": 2.9396284153005463,
      "grad_norm": 0.6282869114639112,
      "learning_rate": 3.462685792078888e-05,
      "loss": 11.6751,
      "step": 16811
    },
    {
      "epoch": 2.9398032786885246,
      "grad_norm": 0.6443965354718075,
      "learning_rate": 3.4616142656082485e-05,
      "loss": 11.9698,
      "step": 16812
    },
    {
      "epoch": 2.9399781420765025,
      "grad_norm": 0.7009534094999369,
      "learning_rate": 3.460542870252318e-05,
      "loss": 11.7469,
      "step": 16813
    },
    {
      "epoch": 2.940153005464481,
      "grad_norm": 0.6034092235794968,
      "learning_rate": 3.45947160603258e-05,
      "loss": 11.796,
      "step": 16814
    },
    {
      "epoch": 2.9403278688524592,
      "grad_norm": 0.843120469331885,
      "learning_rate": 3.4584004729705213e-05,
      "loss": 11.6335,
      "step": 16815
    },
    {
      "epoch": 2.940502732240437,
      "grad_norm": 0.7901309448937488,
      "learning_rate": 3.457329471087617e-05,
      "loss": 11.9255,
      "step": 16816
    },
    {
      "epoch": 2.9406775956284155,
      "grad_norm": 0.6385988502864001,
      "learning_rate": 3.456258600405342e-05,
      "loss": 11.5297,
      "step": 16817
    },
    {
      "epoch": 2.9408524590163934,
      "grad_norm": 0.6458025435766372,
      "learning_rate": 3.45518786094518e-05,
      "loss": 11.5256,
      "step": 16818
    },
    {
      "epoch": 2.9410273224043717,
      "grad_norm": 0.6462756653672201,
      "learning_rate": 3.454117252728595e-05,
      "loss": 11.3681,
      "step": 16819
    },
    {
      "epoch": 2.9412021857923496,
      "grad_norm": 0.6268054621744934,
      "learning_rate": 3.4530467757770556e-05,
      "loss": 11.6265,
      "step": 16820
    },
    {
      "epoch": 2.941377049180328,
      "grad_norm": 0.6367624635254732,
      "learning_rate": 3.451976430112032e-05,
      "loss": 11.655,
      "step": 16821
    },
    {
      "epoch": 2.941551912568306,
      "grad_norm": 0.6851153763864434,
      "learning_rate": 3.450906215754984e-05,
      "loss": 12.0148,
      "step": 16822
    },
    {
      "epoch": 2.941726775956284,
      "grad_norm": 0.6896069630758107,
      "learning_rate": 3.4498361327273766e-05,
      "loss": 11.9848,
      "step": 16823
    },
    {
      "epoch": 2.9419016393442625,
      "grad_norm": 0.6152220051819531,
      "learning_rate": 3.4487661810506656e-05,
      "loss": 11.9131,
      "step": 16824
    },
    {
      "epoch": 2.9420765027322404,
      "grad_norm": 0.631887139530762,
      "learning_rate": 3.4476963607463076e-05,
      "loss": 11.8951,
      "step": 16825
    },
    {
      "epoch": 2.9422513661202183,
      "grad_norm": 0.6151808938210453,
      "learning_rate": 3.446626671835756e-05,
      "loss": 11.926,
      "step": 16826
    },
    {
      "epoch": 2.9424262295081967,
      "grad_norm": 0.6582592084550424,
      "learning_rate": 3.4455571143404606e-05,
      "loss": 11.8336,
      "step": 16827
    },
    {
      "epoch": 2.942601092896175,
      "grad_norm": 0.653785409905621,
      "learning_rate": 3.444487688281871e-05,
      "loss": 11.8469,
      "step": 16828
    },
    {
      "epoch": 2.942775956284153,
      "grad_norm": 0.6688325036550833,
      "learning_rate": 3.44341839368143e-05,
      "loss": 11.7008,
      "step": 16829
    },
    {
      "epoch": 2.9429508196721312,
      "grad_norm": 0.6514307957289802,
      "learning_rate": 3.442349230560582e-05,
      "loss": 11.8713,
      "step": 16830
    },
    {
      "epoch": 2.943125683060109,
      "grad_norm": 0.6629673019751641,
      "learning_rate": 3.441280198940767e-05,
      "loss": 11.6513,
      "step": 16831
    },
    {
      "epoch": 2.9433005464480875,
      "grad_norm": 0.7112951708439416,
      "learning_rate": 3.440211298843422e-05,
      "loss": 11.499,
      "step": 16832
    },
    {
      "epoch": 2.943475409836066,
      "grad_norm": 0.5854099152000808,
      "learning_rate": 3.439142530289981e-05,
      "loss": 11.6206,
      "step": 16833
    },
    {
      "epoch": 2.9436502732240437,
      "grad_norm": 0.6804845537195511,
      "learning_rate": 3.438073893301876e-05,
      "loss": 11.7083,
      "step": 16834
    },
    {
      "epoch": 2.9438251366120216,
      "grad_norm": 0.6304015199019748,
      "learning_rate": 3.4370053879005395e-05,
      "loss": 11.6431,
      "step": 16835
    },
    {
      "epoch": 2.944,
      "grad_norm": 0.6316947475569692,
      "learning_rate": 3.435937014107394e-05,
      "loss": 11.8309,
      "step": 16836
    },
    {
      "epoch": 2.9441748633879783,
      "grad_norm": 0.6285189268988075,
      "learning_rate": 3.4348687719438665e-05,
      "loss": 11.8034,
      "step": 16837
    },
    {
      "epoch": 2.944349726775956,
      "grad_norm": 0.6609268870817192,
      "learning_rate": 3.433800661431378e-05,
      "loss": 11.7129,
      "step": 16838
    },
    {
      "epoch": 2.9445245901639345,
      "grad_norm": 0.6708512919318975,
      "learning_rate": 3.4327326825913465e-05,
      "loss": 11.6543,
      "step": 16839
    },
    {
      "epoch": 2.9446994535519124,
      "grad_norm": 0.6008364219935246,
      "learning_rate": 3.4316648354451895e-05,
      "loss": 11.8691,
      "step": 16840
    },
    {
      "epoch": 2.9448743169398908,
      "grad_norm": 0.5978248484030598,
      "learning_rate": 3.430597120014319e-05,
      "loss": 11.1626,
      "step": 16841
    },
    {
      "epoch": 2.945049180327869,
      "grad_norm": 0.6265096304103479,
      "learning_rate": 3.4295295363201476e-05,
      "loss": 11.6192,
      "step": 16842
    },
    {
      "epoch": 2.945224043715847,
      "grad_norm": 0.6880234944483363,
      "learning_rate": 3.4284620843840834e-05,
      "loss": 11.6743,
      "step": 16843
    },
    {
      "epoch": 2.945398907103825,
      "grad_norm": 0.6251751162675543,
      "learning_rate": 3.427394764227531e-05,
      "loss": 11.6679,
      "step": 16844
    },
    {
      "epoch": 2.9455737704918032,
      "grad_norm": 0.6258848493022995,
      "learning_rate": 3.426327575871895e-05,
      "loss": 11.2838,
      "step": 16845
    },
    {
      "epoch": 2.9457486338797816,
      "grad_norm": 0.6691920775394796,
      "learning_rate": 3.425260519338573e-05,
      "loss": 11.9181,
      "step": 16846
    },
    {
      "epoch": 2.9459234972677595,
      "grad_norm": 0.6844421307165772,
      "learning_rate": 3.424193594648967e-05,
      "loss": 11.695,
      "step": 16847
    },
    {
      "epoch": 2.946098360655738,
      "grad_norm": 0.6230783554707627,
      "learning_rate": 3.423126801824469e-05,
      "loss": 11.7385,
      "step": 16848
    },
    {
      "epoch": 2.9462732240437157,
      "grad_norm": 0.6720925431660784,
      "learning_rate": 3.422060140886472e-05,
      "loss": 11.8459,
      "step": 16849
    },
    {
      "epoch": 2.946448087431694,
      "grad_norm": 0.6164001714350537,
      "learning_rate": 3.420993611856369e-05,
      "loss": 11.8305,
      "step": 16850
    },
    {
      "epoch": 2.9466229508196724,
      "grad_norm": 0.717897893204414,
      "learning_rate": 3.419927214755538e-05,
      "loss": 11.785,
      "step": 16851
    },
    {
      "epoch": 2.9467978142076503,
      "grad_norm": 0.7118828500195914,
      "learning_rate": 3.418860949605374e-05,
      "loss": 11.8016,
      "step": 16852
    },
    {
      "epoch": 2.946972677595628,
      "grad_norm": 0.6506129373707803,
      "learning_rate": 3.4177948164272535e-05,
      "loss": 11.8394,
      "step": 16853
    },
    {
      "epoch": 2.9471475409836065,
      "grad_norm": 0.6384825457979523,
      "learning_rate": 3.416728815242557e-05,
      "loss": 11.7387,
      "step": 16854
    },
    {
      "epoch": 2.947322404371585,
      "grad_norm": 0.6390474923178433,
      "learning_rate": 3.4156629460726655e-05,
      "loss": 11.8304,
      "step": 16855
    },
    {
      "epoch": 2.9474972677595628,
      "grad_norm": 0.7270116628877857,
      "learning_rate": 3.414597208938941e-05,
      "loss": 12.0773,
      "step": 16856
    },
    {
      "epoch": 2.947672131147541,
      "grad_norm": 0.6886164692731084,
      "learning_rate": 3.413531603862765e-05,
      "loss": 11.6563,
      "step": 16857
    },
    {
      "epoch": 2.947846994535519,
      "grad_norm": 0.6249848808946411,
      "learning_rate": 3.412466130865505e-05,
      "loss": 11.808,
      "step": 16858
    },
    {
      "epoch": 2.9480218579234974,
      "grad_norm": 0.6045740604699329,
      "learning_rate": 3.411400789968527e-05,
      "loss": 11.4423,
      "step": 16859
    },
    {
      "epoch": 2.9481967213114753,
      "grad_norm": 0.7162306069582537,
      "learning_rate": 3.4103355811931915e-05,
      "loss": 11.9202,
      "step": 16860
    },
    {
      "epoch": 2.9483715846994536,
      "grad_norm": 0.6021640132812616,
      "learning_rate": 3.4092705045608564e-05,
      "loss": 11.7241,
      "step": 16861
    },
    {
      "epoch": 2.9485464480874315,
      "grad_norm": 0.6424942163229761,
      "learning_rate": 3.4082055600928867e-05,
      "loss": 11.9088,
      "step": 16862
    },
    {
      "epoch": 2.94872131147541,
      "grad_norm": 0.5984909957969382,
      "learning_rate": 3.407140747810636e-05,
      "loss": 11.539,
      "step": 16863
    },
    {
      "epoch": 2.948896174863388,
      "grad_norm": 0.7364235748028,
      "learning_rate": 3.406076067735458e-05,
      "loss": 11.6555,
      "step": 16864
    },
    {
      "epoch": 2.949071038251366,
      "grad_norm": 0.608680359867599,
      "learning_rate": 3.405011519888694e-05,
      "loss": 11.8903,
      "step": 16865
    },
    {
      "epoch": 2.9492459016393444,
      "grad_norm": 0.642515399319775,
      "learning_rate": 3.403947104291704e-05,
      "loss": 11.6639,
      "step": 16866
    },
    {
      "epoch": 2.9494207650273223,
      "grad_norm": 0.6229455190523504,
      "learning_rate": 3.402882820965825e-05,
      "loss": 11.6426,
      "step": 16867
    },
    {
      "epoch": 2.9495956284153007,
      "grad_norm": 0.6223266995488965,
      "learning_rate": 3.401818669932405e-05,
      "loss": 11.6146,
      "step": 16868
    },
    {
      "epoch": 2.9497704918032785,
      "grad_norm": 0.6659088853530407,
      "learning_rate": 3.400754651212776e-05,
      "loss": 11.698,
      "step": 16869
    },
    {
      "epoch": 2.949945355191257,
      "grad_norm": 0.8608525607762716,
      "learning_rate": 3.399690764828277e-05,
      "loss": 11.8804,
      "step": 16870
    },
    {
      "epoch": 2.950120218579235,
      "grad_norm": 0.6477763035428029,
      "learning_rate": 3.398627010800246e-05,
      "loss": 11.925,
      "step": 16871
    },
    {
      "epoch": 2.950295081967213,
      "grad_norm": 0.6730791186508477,
      "learning_rate": 3.3975633891500114e-05,
      "loss": 11.8731,
      "step": 16872
    },
    {
      "epoch": 2.9504699453551915,
      "grad_norm": 0.6528136046868052,
      "learning_rate": 3.396499899898907e-05,
      "loss": 11.6443,
      "step": 16873
    },
    {
      "epoch": 2.9506448087431694,
      "grad_norm": 0.6337996523019734,
      "learning_rate": 3.395436543068252e-05,
      "loss": 11.6884,
      "step": 16874
    },
    {
      "epoch": 2.9508196721311473,
      "grad_norm": 0.6436350606544795,
      "learning_rate": 3.3943733186793704e-05,
      "loss": 11.8113,
      "step": 16875
    },
    {
      "epoch": 2.9509945355191256,
      "grad_norm": 0.7048850941163454,
      "learning_rate": 3.393310226753588e-05,
      "loss": 11.8079,
      "step": 16876
    },
    {
      "epoch": 2.951169398907104,
      "grad_norm": 0.6503973223754038,
      "learning_rate": 3.392247267312224e-05,
      "loss": 11.7613,
      "step": 16877
    },
    {
      "epoch": 2.951344262295082,
      "grad_norm": 0.5996808665566095,
      "learning_rate": 3.391184440376588e-05,
      "loss": 11.4525,
      "step": 16878
    },
    {
      "epoch": 2.95151912568306,
      "grad_norm": 0.649244422394427,
      "learning_rate": 3.3901217459679956e-05,
      "loss": 11.5518,
      "step": 16879
    },
    {
      "epoch": 2.951693989071038,
      "grad_norm": 0.605492804201371,
      "learning_rate": 3.389059184107753e-05,
      "loss": 11.6149,
      "step": 16880
    },
    {
      "epoch": 2.9518688524590164,
      "grad_norm": 0.618415798856161,
      "learning_rate": 3.387996754817176e-05,
      "loss": 11.7996,
      "step": 16881
    },
    {
      "epoch": 2.9520437158469948,
      "grad_norm": 0.6644914614314389,
      "learning_rate": 3.386934458117569e-05,
      "loss": 11.7127,
      "step": 16882
    },
    {
      "epoch": 2.9522185792349727,
      "grad_norm": 0.6763967468213937,
      "learning_rate": 3.385872294030228e-05,
      "loss": 11.9546,
      "step": 16883
    },
    {
      "epoch": 2.9523934426229506,
      "grad_norm": 0.6553068190822522,
      "learning_rate": 3.384810262576455e-05,
      "loss": 11.6201,
      "step": 16884
    },
    {
      "epoch": 2.952568306010929,
      "grad_norm": 0.6910961108173714,
      "learning_rate": 3.383748363777543e-05,
      "loss": 11.338,
      "step": 16885
    },
    {
      "epoch": 2.9527431693989072,
      "grad_norm": 0.6001244839901495,
      "learning_rate": 3.382686597654796e-05,
      "loss": 11.7181,
      "step": 16886
    },
    {
      "epoch": 2.952918032786885,
      "grad_norm": 0.6032016329404565,
      "learning_rate": 3.381624964229504e-05,
      "loss": 11.82,
      "step": 16887
    },
    {
      "epoch": 2.9530928961748635,
      "grad_norm": 0.6364970857257469,
      "learning_rate": 3.380563463522948e-05,
      "loss": 11.8251,
      "step": 16888
    },
    {
      "epoch": 2.9532677595628414,
      "grad_norm": 0.6410446139865517,
      "learning_rate": 3.379502095556416e-05,
      "loss": 11.6841,
      "step": 16889
    },
    {
      "epoch": 2.9534426229508197,
      "grad_norm": 0.6129941232835511,
      "learning_rate": 3.3784408603511974e-05,
      "loss": 11.8933,
      "step": 16890
    },
    {
      "epoch": 2.953617486338798,
      "grad_norm": 0.6259545059335363,
      "learning_rate": 3.377379757928575e-05,
      "loss": 11.8858,
      "step": 16891
    },
    {
      "epoch": 2.953792349726776,
      "grad_norm": 0.6852789146995886,
      "learning_rate": 3.376318788309818e-05,
      "loss": 11.7698,
      "step": 16892
    },
    {
      "epoch": 2.953967213114754,
      "grad_norm": 0.6529753220949072,
      "learning_rate": 3.375257951516207e-05,
      "loss": 11.582,
      "step": 16893
    },
    {
      "epoch": 2.954142076502732,
      "grad_norm": 0.743464297808948,
      "learning_rate": 3.374197247569012e-05,
      "loss": 11.7551,
      "step": 16894
    },
    {
      "epoch": 2.9543169398907105,
      "grad_norm": 0.6306816431632365,
      "learning_rate": 3.373136676489509e-05,
      "loss": 11.8313,
      "step": 16895
    },
    {
      "epoch": 2.9544918032786884,
      "grad_norm": 0.6223170330023035,
      "learning_rate": 3.3720762382989654e-05,
      "loss": 11.9573,
      "step": 16896
    },
    {
      "epoch": 2.9546666666666668,
      "grad_norm": 0.6431494923536073,
      "learning_rate": 3.371015933018642e-05,
      "loss": 11.7792,
      "step": 16897
    },
    {
      "epoch": 2.9548415300546447,
      "grad_norm": 0.7009336236120316,
      "learning_rate": 3.3699557606698015e-05,
      "loss": 11.7827,
      "step": 16898
    },
    {
      "epoch": 2.955016393442623,
      "grad_norm": 0.6781088847732372,
      "learning_rate": 3.368895721273702e-05,
      "loss": 11.4231,
      "step": 16899
    },
    {
      "epoch": 2.9551912568306014,
      "grad_norm": 0.6814797900236282,
      "learning_rate": 3.36783581485161e-05,
      "loss": 11.6454,
      "step": 16900
    },
    {
      "epoch": 2.9553661202185793,
      "grad_norm": 0.6434281269778638,
      "learning_rate": 3.36677604142477e-05,
      "loss": 11.8087,
      "step": 16901
    },
    {
      "epoch": 2.955540983606557,
      "grad_norm": 0.618291310620995,
      "learning_rate": 3.365716401014437e-05,
      "loss": 11.6024,
      "step": 16902
    },
    {
      "epoch": 2.9557158469945355,
      "grad_norm": 0.6519779269639366,
      "learning_rate": 3.3646568936418596e-05,
      "loss": 11.7848,
      "step": 16903
    },
    {
      "epoch": 2.955890710382514,
      "grad_norm": 0.7264601236692649,
      "learning_rate": 3.363597519328282e-05,
      "loss": 11.1771,
      "step": 16904
    },
    {
      "epoch": 2.9560655737704917,
      "grad_norm": 0.6706348091751576,
      "learning_rate": 3.3625382780949574e-05,
      "loss": 11.8047,
      "step": 16905
    },
    {
      "epoch": 2.95624043715847,
      "grad_norm": 0.6392694417654426,
      "learning_rate": 3.361479169963117e-05,
      "loss": 11.6386,
      "step": 16906
    },
    {
      "epoch": 2.956415300546448,
      "grad_norm": 0.657583781581,
      "learning_rate": 3.360420194954e-05,
      "loss": 11.8991,
      "step": 16907
    },
    {
      "epoch": 2.9565901639344263,
      "grad_norm": 0.6395896150323487,
      "learning_rate": 3.359361353088847e-05,
      "loss": 11.807,
      "step": 16908
    },
    {
      "epoch": 2.9567650273224046,
      "grad_norm": 0.6604087086900172,
      "learning_rate": 3.358302644388887e-05,
      "loss": 11.6723,
      "step": 16909
    },
    {
      "epoch": 2.9569398907103825,
      "grad_norm": 0.6260656219591942,
      "learning_rate": 3.357244068875352e-05,
      "loss": 11.6372,
      "step": 16910
    },
    {
      "epoch": 2.9571147540983604,
      "grad_norm": 0.6908377779958402,
      "learning_rate": 3.356185626569468e-05,
      "loss": 11.8572,
      "step": 16911
    },
    {
      "epoch": 2.957289617486339,
      "grad_norm": 0.6101787791023362,
      "learning_rate": 3.355127317492463e-05,
      "loss": 11.7668,
      "step": 16912
    },
    {
      "epoch": 2.957464480874317,
      "grad_norm": 0.6623246084245186,
      "learning_rate": 3.354069141665553e-05,
      "loss": 11.7648,
      "step": 16913
    },
    {
      "epoch": 2.957639344262295,
      "grad_norm": 0.6385552320506688,
      "learning_rate": 3.3530110991099706e-05,
      "loss": 11.8502,
      "step": 16914
    },
    {
      "epoch": 2.9578142076502734,
      "grad_norm": 0.5933959593490723,
      "learning_rate": 3.35195318984692e-05,
      "loss": 11.6904,
      "step": 16915
    },
    {
      "epoch": 2.9579890710382513,
      "grad_norm": 0.6358607695150789,
      "learning_rate": 3.3508954138976214e-05,
      "loss": 11.7693,
      "step": 16916
    },
    {
      "epoch": 2.9581639344262296,
      "grad_norm": 0.6623188311695224,
      "learning_rate": 3.349837771283285e-05,
      "loss": 11.5426,
      "step": 16917
    },
    {
      "epoch": 2.9583387978142075,
      "grad_norm": 0.6245738352755763,
      "learning_rate": 3.34878026202512e-05,
      "loss": 11.7561,
      "step": 16918
    },
    {
      "epoch": 2.958513661202186,
      "grad_norm": 0.6765415656766655,
      "learning_rate": 3.347722886144331e-05,
      "loss": 11.4891,
      "step": 16919
    },
    {
      "epoch": 2.9586885245901637,
      "grad_norm": 0.648331697703546,
      "learning_rate": 3.346665643662125e-05,
      "loss": 11.7227,
      "step": 16920
    },
    {
      "epoch": 2.958863387978142,
      "grad_norm": 0.6591548777159772,
      "learning_rate": 3.3456085345997e-05,
      "loss": 11.5957,
      "step": 16921
    },
    {
      "epoch": 2.9590382513661204,
      "grad_norm": 0.5975565539747628,
      "learning_rate": 3.3445515589782574e-05,
      "loss": 11.8649,
      "step": 16922
    },
    {
      "epoch": 2.9592131147540983,
      "grad_norm": 0.6258061142408667,
      "learning_rate": 3.343494716818989e-05,
      "loss": 11.5262,
      "step": 16923
    },
    {
      "epoch": 2.9593879781420767,
      "grad_norm": 0.6318379001289239,
      "learning_rate": 3.3424380081430905e-05,
      "loss": 11.5814,
      "step": 16924
    },
    {
      "epoch": 2.9595628415300546,
      "grad_norm": 0.6134631925730928,
      "learning_rate": 3.3413814329717516e-05,
      "loss": 11.8105,
      "step": 16925
    },
    {
      "epoch": 2.959737704918033,
      "grad_norm": 0.6191059419164362,
      "learning_rate": 3.340324991326159e-05,
      "loss": 11.6345,
      "step": 16926
    },
    {
      "epoch": 2.959912568306011,
      "grad_norm": 0.6963982191305452,
      "learning_rate": 3.339268683227499e-05,
      "loss": 11.7839,
      "step": 16927
    },
    {
      "epoch": 2.960087431693989,
      "grad_norm": 0.645725167142561,
      "learning_rate": 3.338212508696953e-05,
      "loss": 11.7367,
      "step": 16928
    },
    {
      "epoch": 2.960262295081967,
      "grad_norm": 0.6769513724776927,
      "learning_rate": 3.3371564677557e-05,
      "loss": 11.4986,
      "step": 16929
    },
    {
      "epoch": 2.9604371584699454,
      "grad_norm": 0.6526516876823174,
      "learning_rate": 3.3361005604249176e-05,
      "loss": 11.8021,
      "step": 16930
    },
    {
      "epoch": 2.9606120218579237,
      "grad_norm": 0.6509031350236267,
      "learning_rate": 3.33504478672578e-05,
      "loss": 11.5555,
      "step": 16931
    },
    {
      "epoch": 2.9607868852459016,
      "grad_norm": 0.6774519694656955,
      "learning_rate": 3.333989146679458e-05,
      "loss": 11.5987,
      "step": 16932
    },
    {
      "epoch": 2.9609617486338795,
      "grad_norm": 0.6361300837346665,
      "learning_rate": 3.332933640307121e-05,
      "loss": 11.4672,
      "step": 16933
    },
    {
      "epoch": 2.961136612021858,
      "grad_norm": 0.610902548143187,
      "learning_rate": 3.3318782676299356e-05,
      "loss": 11.8305,
      "step": 16934
    },
    {
      "epoch": 2.961311475409836,
      "grad_norm": 0.6543171567314381,
      "learning_rate": 3.330823028669064e-05,
      "loss": 12.0299,
      "step": 16935
    },
    {
      "epoch": 2.961486338797814,
      "grad_norm": 0.6640278535503208,
      "learning_rate": 3.3297679234456694e-05,
      "loss": 11.7915,
      "step": 16936
    },
    {
      "epoch": 2.9616612021857924,
      "grad_norm": 0.6909257629105586,
      "learning_rate": 3.328712951980907e-05,
      "loss": 11.8038,
      "step": 16937
    },
    {
      "epoch": 2.9618360655737703,
      "grad_norm": 0.6348492924572458,
      "learning_rate": 3.327658114295934e-05,
      "loss": 11.8377,
      "step": 16938
    },
    {
      "epoch": 2.9620109289617487,
      "grad_norm": 0.8464932626818017,
      "learning_rate": 3.3266034104119014e-05,
      "loss": 11.6332,
      "step": 16939
    },
    {
      "epoch": 2.962185792349727,
      "grad_norm": 0.6590949225837319,
      "learning_rate": 3.3255488403499615e-05,
      "loss": 11.5655,
      "step": 16940
    },
    {
      "epoch": 2.962360655737705,
      "grad_norm": 0.6329669679958981,
      "learning_rate": 3.324494404131261e-05,
      "loss": 11.5635,
      "step": 16941
    },
    {
      "epoch": 2.962535519125683,
      "grad_norm": 0.6214249287098225,
      "learning_rate": 3.3234401017769434e-05,
      "loss": 11.6929,
      "step": 16942
    },
    {
      "epoch": 2.962710382513661,
      "grad_norm": 0.6853210613513322,
      "learning_rate": 3.322385933308151e-05,
      "loss": 11.8366,
      "step": 16943
    },
    {
      "epoch": 2.9628852459016395,
      "grad_norm": 0.6890461208084597,
      "learning_rate": 3.321331898746024e-05,
      "loss": 11.8153,
      "step": 16944
    },
    {
      "epoch": 2.9630601092896174,
      "grad_norm": 0.6919072159602552,
      "learning_rate": 3.320277998111699e-05,
      "loss": 11.8249,
      "step": 16945
    },
    {
      "epoch": 2.9632349726775957,
      "grad_norm": 0.5914667808881668,
      "learning_rate": 3.3192242314263134e-05,
      "loss": 11.6405,
      "step": 16946
    },
    {
      "epoch": 2.9634098360655736,
      "grad_norm": 0.6602778583658452,
      "learning_rate": 3.318170598710987e-05,
      "loss": 11.7102,
      "step": 16947
    },
    {
      "epoch": 2.963584699453552,
      "grad_norm": 0.629839229792346,
      "learning_rate": 3.3171170999868594e-05,
      "loss": 11.7293,
      "step": 16948
    },
    {
      "epoch": 2.9637595628415303,
      "grad_norm": 0.645880721308206,
      "learning_rate": 3.3160637352750534e-05,
      "loss": 11.4963,
      "step": 16949
    },
    {
      "epoch": 2.963934426229508,
      "grad_norm": 0.6450160610582065,
      "learning_rate": 3.315010504596692e-05,
      "loss": 11.6714,
      "step": 16950
    },
    {
      "epoch": 2.964109289617486,
      "grad_norm": 0.6237301431770036,
      "learning_rate": 3.313957407972899e-05,
      "loss": 11.8052,
      "step": 16951
    },
    {
      "epoch": 2.9642841530054644,
      "grad_norm": 0.6456891760124803,
      "learning_rate": 3.312904445424781e-05,
      "loss": 11.6242,
      "step": 16952
    },
    {
      "epoch": 2.964459016393443,
      "grad_norm": 0.7566501191411523,
      "learning_rate": 3.311851616973466e-05,
      "loss": 11.5781,
      "step": 16953
    },
    {
      "epoch": 2.9646338797814207,
      "grad_norm": 0.6574658897456708,
      "learning_rate": 3.31079892264006e-05,
      "loss": 11.7221,
      "step": 16954
    },
    {
      "epoch": 2.964808743169399,
      "grad_norm": 0.6486971247263791,
      "learning_rate": 3.309746362445679e-05,
      "loss": 11.5736,
      "step": 16955
    },
    {
      "epoch": 2.964983606557377,
      "grad_norm": 0.7450775067582789,
      "learning_rate": 3.308693936411421e-05,
      "loss": 11.8028,
      "step": 16956
    },
    {
      "epoch": 2.9651584699453553,
      "grad_norm": 0.6732840901590829,
      "learning_rate": 3.3076416445583916e-05,
      "loss": 11.6983,
      "step": 16957
    },
    {
      "epoch": 2.9653333333333336,
      "grad_norm": 0.7425375507173216,
      "learning_rate": 3.3065894869076975e-05,
      "loss": 11.8522,
      "step": 16958
    },
    {
      "epoch": 2.9655081967213115,
      "grad_norm": 0.6417850477525676,
      "learning_rate": 3.305537463480437e-05,
      "loss": 11.8737,
      "step": 16959
    },
    {
      "epoch": 2.9656830601092894,
      "grad_norm": 0.6917229045319987,
      "learning_rate": 3.3044855742977074e-05,
      "loss": 11.5819,
      "step": 16960
    },
    {
      "epoch": 2.9658579234972677,
      "grad_norm": 0.6517681158608509,
      "learning_rate": 3.303433819380595e-05,
      "loss": 11.9581,
      "step": 16961
    },
    {
      "epoch": 2.966032786885246,
      "grad_norm": 0.6181330701127906,
      "learning_rate": 3.302382198750198e-05,
      "loss": 11.6514,
      "step": 16962
    },
    {
      "epoch": 2.966207650273224,
      "grad_norm": 0.6489151846229941,
      "learning_rate": 3.3013307124276014e-05,
      "loss": 11.7698,
      "step": 16963
    },
    {
      "epoch": 2.9663825136612023,
      "grad_norm": 0.6862869543222445,
      "learning_rate": 3.3002793604338956e-05,
      "loss": 11.7356,
      "step": 16964
    },
    {
      "epoch": 2.96655737704918,
      "grad_norm": 0.6368463646774389,
      "learning_rate": 3.2992281427901564e-05,
      "loss": 11.8915,
      "step": 16965
    },
    {
      "epoch": 2.9667322404371586,
      "grad_norm": 0.6566889650036779,
      "learning_rate": 3.2981770595174634e-05,
      "loss": 11.9903,
      "step": 16966
    },
    {
      "epoch": 2.9669071038251364,
      "grad_norm": 0.6426219658446957,
      "learning_rate": 3.297126110636901e-05,
      "loss": 11.6934,
      "step": 16967
    },
    {
      "epoch": 2.967081967213115,
      "grad_norm": 0.6292864171357638,
      "learning_rate": 3.296075296169542e-05,
      "loss": 11.5025,
      "step": 16968
    },
    {
      "epoch": 2.9672568306010927,
      "grad_norm": 0.6834447544236287,
      "learning_rate": 3.295024616136458e-05,
      "loss": 11.7716,
      "step": 16969
    },
    {
      "epoch": 2.967431693989071,
      "grad_norm": 0.6710613216047477,
      "learning_rate": 3.2939740705587154e-05,
      "loss": 11.7707,
      "step": 16970
    },
    {
      "epoch": 2.9676065573770494,
      "grad_norm": 0.7601836045025074,
      "learning_rate": 3.2929236594573796e-05,
      "loss": 11.699,
      "step": 16971
    },
    {
      "epoch": 2.9677814207650273,
      "grad_norm": 0.6201057245130531,
      "learning_rate": 3.291873382853521e-05,
      "loss": 11.9878,
      "step": 16972
    },
    {
      "epoch": 2.9679562841530056,
      "grad_norm": 0.6344418461638064,
      "learning_rate": 3.290823240768202e-05,
      "loss": 11.7023,
      "step": 16973
    },
    {
      "epoch": 2.9681311475409835,
      "grad_norm": 0.6845913800389078,
      "learning_rate": 3.2897732332224715e-05,
      "loss": 11.7613,
      "step": 16974
    },
    {
      "epoch": 2.968306010928962,
      "grad_norm": 0.6836849576825326,
      "learning_rate": 3.288723360237392e-05,
      "loss": 11.756,
      "step": 16975
    },
    {
      "epoch": 2.9684808743169397,
      "grad_norm": 0.6604054305078986,
      "learning_rate": 3.287673621834012e-05,
      "loss": 11.8678,
      "step": 16976
    },
    {
      "epoch": 2.968655737704918,
      "grad_norm": 0.6351183522008932,
      "learning_rate": 3.286624018033389e-05,
      "loss": 11.7648,
      "step": 16977
    },
    {
      "epoch": 2.968830601092896,
      "grad_norm": 0.608955881279012,
      "learning_rate": 3.2855745488565705e-05,
      "loss": 11.5926,
      "step": 16978
    },
    {
      "epoch": 2.9690054644808743,
      "grad_norm": 0.7386216060874119,
      "learning_rate": 3.284525214324594e-05,
      "loss": 11.7516,
      "step": 16979
    },
    {
      "epoch": 2.9691803278688527,
      "grad_norm": 0.6134192705705398,
      "learning_rate": 3.283476014458507e-05,
      "loss": 11.5509,
      "step": 16980
    },
    {
      "epoch": 2.9693551912568306,
      "grad_norm": 0.7143105922373455,
      "learning_rate": 3.282426949279346e-05,
      "loss": 11.7302,
      "step": 16981
    },
    {
      "epoch": 2.969530054644809,
      "grad_norm": 0.6998541467016034,
      "learning_rate": 3.281378018808156e-05,
      "loss": 11.7379,
      "step": 16982
    },
    {
      "epoch": 2.969704918032787,
      "grad_norm": 0.6451855895721142,
      "learning_rate": 3.2803292230659624e-05,
      "loss": 11.694,
      "step": 16983
    },
    {
      "epoch": 2.969879781420765,
      "grad_norm": 0.6702797687137368,
      "learning_rate": 3.279280562073799e-05,
      "loss": 11.8981,
      "step": 16984
    },
    {
      "epoch": 2.970054644808743,
      "grad_norm": 0.6804248101130161,
      "learning_rate": 3.278232035852693e-05,
      "loss": 11.8408,
      "step": 16985
    },
    {
      "epoch": 2.9702295081967214,
      "grad_norm": 0.6228390380627505,
      "learning_rate": 3.277183644423677e-05,
      "loss": 11.573,
      "step": 16986
    },
    {
      "epoch": 2.9704043715846993,
      "grad_norm": 0.6279083699371352,
      "learning_rate": 3.276135387807774e-05,
      "loss": 11.5944,
      "step": 16987
    },
    {
      "epoch": 2.9705792349726776,
      "grad_norm": 0.6415339596652095,
      "learning_rate": 3.275087266025998e-05,
      "loss": 11.8288,
      "step": 16988
    },
    {
      "epoch": 2.970754098360656,
      "grad_norm": 0.7363238583990104,
      "learning_rate": 3.27403927909937e-05,
      "loss": 11.5236,
      "step": 16989
    },
    {
      "epoch": 2.970928961748634,
      "grad_norm": 0.6870870769806204,
      "learning_rate": 3.2729914270489026e-05,
      "loss": 11.7474,
      "step": 16990
    },
    {
      "epoch": 2.9711038251366118,
      "grad_norm": 0.6040073658496583,
      "learning_rate": 3.271943709895619e-05,
      "loss": 11.635,
      "step": 16991
    },
    {
      "epoch": 2.97127868852459,
      "grad_norm": 0.6558335484125383,
      "learning_rate": 3.2708961276605177e-05,
      "loss": 11.861,
      "step": 16992
    },
    {
      "epoch": 2.9714535519125684,
      "grad_norm": 0.6503728074994878,
      "learning_rate": 3.269848680364611e-05,
      "loss": 11.6635,
      "step": 16993
    },
    {
      "epoch": 2.9716284153005463,
      "grad_norm": 0.566321235827364,
      "learning_rate": 3.268801368028902e-05,
      "loss": 11.6392,
      "step": 16994
    },
    {
      "epoch": 2.9718032786885247,
      "grad_norm": 0.6255399131710024,
      "learning_rate": 3.267754190674389e-05,
      "loss": 11.9114,
      "step": 16995
    },
    {
      "epoch": 2.9719781420765026,
      "grad_norm": 0.5992370194839012,
      "learning_rate": 3.2667071483220836e-05,
      "loss": 11.942,
      "step": 16996
    },
    {
      "epoch": 2.972153005464481,
      "grad_norm": 0.6218486148223547,
      "learning_rate": 3.265660240992969e-05,
      "loss": 11.565,
      "step": 16997
    },
    {
      "epoch": 2.9723278688524593,
      "grad_norm": 0.6422071693167187,
      "learning_rate": 3.2646134687080444e-05,
      "loss": 11.442,
      "step": 16998
    },
    {
      "epoch": 2.972502732240437,
      "grad_norm": 0.6688697687355196,
      "learning_rate": 3.2635668314883e-05,
      "loss": 11.748,
      "step": 16999
    },
    {
      "epoch": 2.972677595628415,
      "grad_norm": 0.6787193938855305,
      "learning_rate": 3.2625203293547246e-05,
      "loss": 11.5928,
      "step": 17000
    },
    {
      "epoch": 2.9728524590163934,
      "grad_norm": 0.6908590505627741,
      "learning_rate": 3.261473962328303e-05,
      "loss": 11.7377,
      "step": 17001
    },
    {
      "epoch": 2.9730273224043717,
      "grad_norm": 0.6087539281864249,
      "learning_rate": 3.260427730430018e-05,
      "loss": 11.8877,
      "step": 17002
    },
    {
      "epoch": 2.9732021857923496,
      "grad_norm": 0.6479218662338908,
      "learning_rate": 3.25938163368085e-05,
      "loss": 11.7935,
      "step": 17003
    },
    {
      "epoch": 2.973377049180328,
      "grad_norm": 0.7165176721988049,
      "learning_rate": 3.258335672101778e-05,
      "loss": 11.7965,
      "step": 17004
    },
    {
      "epoch": 2.973551912568306,
      "grad_norm": 0.5980777762483418,
      "learning_rate": 3.257289845713775e-05,
      "loss": 11.6583,
      "step": 17005
    },
    {
      "epoch": 2.973726775956284,
      "grad_norm": 0.6879638271684135,
      "learning_rate": 3.256244154537813e-05,
      "loss": 11.6791,
      "step": 17006
    },
    {
      "epoch": 2.9739016393442625,
      "grad_norm": 0.6589546847902028,
      "learning_rate": 3.2551985985948616e-05,
      "loss": 11.6858,
      "step": 17007
    },
    {
      "epoch": 2.9740765027322404,
      "grad_norm": 0.6091135783771228,
      "learning_rate": 3.254153177905888e-05,
      "loss": 11.6808,
      "step": 17008
    },
    {
      "epoch": 2.9742513661202183,
      "grad_norm": 0.6619534582494614,
      "learning_rate": 3.2531078924918546e-05,
      "loss": 11.8331,
      "step": 17009
    },
    {
      "epoch": 2.9744262295081967,
      "grad_norm": 0.6173419049251846,
      "learning_rate": 3.2520627423737246e-05,
      "loss": 11.5842,
      "step": 17010
    },
    {
      "epoch": 2.974601092896175,
      "grad_norm": 0.6121392307943513,
      "learning_rate": 3.251017727572456e-05,
      "loss": 11.8476,
      "step": 17011
    },
    {
      "epoch": 2.974775956284153,
      "grad_norm": 0.6640274461828761,
      "learning_rate": 3.249972848109003e-05,
      "loss": 11.8979,
      "step": 17012
    },
    {
      "epoch": 2.9749508196721313,
      "grad_norm": 0.6687098512484929,
      "learning_rate": 3.248928104004321e-05,
      "loss": 11.7657,
      "step": 17013
    },
    {
      "epoch": 2.975125683060109,
      "grad_norm": 0.6163237564401779,
      "learning_rate": 3.247883495279358e-05,
      "loss": 11.3669,
      "step": 17014
    },
    {
      "epoch": 2.9753005464480875,
      "grad_norm": 0.7266625076630259,
      "learning_rate": 3.246839021955062e-05,
      "loss": 11.8333,
      "step": 17015
    },
    {
      "epoch": 2.975475409836066,
      "grad_norm": 0.6638872731914416,
      "learning_rate": 3.24579468405238e-05,
      "loss": 12.0063,
      "step": 17016
    },
    {
      "epoch": 2.9756502732240437,
      "grad_norm": 0.6334011269798433,
      "learning_rate": 3.244750481592251e-05,
      "loss": 11.8848,
      "step": 17017
    },
    {
      "epoch": 2.9758251366120216,
      "grad_norm": 0.6111970525220575,
      "learning_rate": 3.2437064145956174e-05,
      "loss": 11.4751,
      "step": 17018
    },
    {
      "epoch": 2.976,
      "grad_norm": 0.6630882778928938,
      "learning_rate": 3.242662483083415e-05,
      "loss": 11.5491,
      "step": 17019
    },
    {
      "epoch": 2.9761748633879783,
      "grad_norm": 0.6134480667908264,
      "learning_rate": 3.241618687076577e-05,
      "loss": 12.0128,
      "step": 17020
    },
    {
      "epoch": 2.976349726775956,
      "grad_norm": 0.6389042613956308,
      "learning_rate": 3.2405750265960353e-05,
      "loss": 11.8067,
      "step": 17021
    },
    {
      "epoch": 2.9765245901639346,
      "grad_norm": 0.5973513644029049,
      "learning_rate": 3.2395315016627195e-05,
      "loss": 11.9009,
      "step": 17022
    },
    {
      "epoch": 2.9766994535519125,
      "grad_norm": 0.6388008274062672,
      "learning_rate": 3.238488112297553e-05,
      "loss": 11.6152,
      "step": 17023
    },
    {
      "epoch": 2.976874316939891,
      "grad_norm": 0.5943299404161171,
      "learning_rate": 3.2374448585214614e-05,
      "loss": 11.6036,
      "step": 17024
    },
    {
      "epoch": 2.9770491803278687,
      "grad_norm": 0.6608624981676086,
      "learning_rate": 3.236401740355364e-05,
      "loss": 11.7688,
      "step": 17025
    },
    {
      "epoch": 2.977224043715847,
      "grad_norm": 0.6040318320303998,
      "learning_rate": 3.2353587578201785e-05,
      "loss": 11.7437,
      "step": 17026
    },
    {
      "epoch": 2.977398907103825,
      "grad_norm": 0.6406986915167232,
      "learning_rate": 3.2343159109368196e-05,
      "loss": 11.8062,
      "step": 17027
    },
    {
      "epoch": 2.9775737704918033,
      "grad_norm": 0.6353361472354978,
      "learning_rate": 3.2332731997262e-05,
      "loss": 11.7691,
      "step": 17028
    },
    {
      "epoch": 2.9777486338797816,
      "grad_norm": 0.6539560914602619,
      "learning_rate": 3.2322306242092306e-05,
      "loss": 11.8409,
      "step": 17029
    },
    {
      "epoch": 2.9779234972677595,
      "grad_norm": 0.650291897387501,
      "learning_rate": 3.231188184406816e-05,
      "loss": 11.8103,
      "step": 17030
    },
    {
      "epoch": 2.978098360655738,
      "grad_norm": 0.7014152076010002,
      "learning_rate": 3.230145880339861e-05,
      "loss": 11.8152,
      "step": 17031
    },
    {
      "epoch": 2.9782732240437158,
      "grad_norm": 0.6521070930907498,
      "learning_rate": 3.229103712029268e-05,
      "loss": 11.7663,
      "step": 17032
    },
    {
      "epoch": 2.978448087431694,
      "grad_norm": 0.6124456146970386,
      "learning_rate": 3.228061679495935e-05,
      "loss": 11.6269,
      "step": 17033
    },
    {
      "epoch": 2.978622950819672,
      "grad_norm": 0.6309990206017463,
      "learning_rate": 3.2270197827607575e-05,
      "loss": 11.6976,
      "step": 17034
    },
    {
      "epoch": 2.9787978142076503,
      "grad_norm": 0.6240498569219376,
      "learning_rate": 3.225978021844629e-05,
      "loss": 11.7539,
      "step": 17035
    },
    {
      "epoch": 2.9789726775956282,
      "grad_norm": 0.62062312916419,
      "learning_rate": 3.22493639676844e-05,
      "loss": 11.6357,
      "step": 17036
    },
    {
      "epoch": 2.9791475409836066,
      "grad_norm": 0.7334231187304959,
      "learning_rate": 3.223894907553079e-05,
      "loss": 11.8426,
      "step": 17037
    },
    {
      "epoch": 2.979322404371585,
      "grad_norm": 0.691902358101553,
      "learning_rate": 3.222853554219431e-05,
      "loss": 11.8141,
      "step": 17038
    },
    {
      "epoch": 2.979497267759563,
      "grad_norm": 0.6608709553768382,
      "learning_rate": 3.221812336788377e-05,
      "loss": 11.6727,
      "step": 17039
    },
    {
      "epoch": 2.979672131147541,
      "grad_norm": 0.600662375093881,
      "learning_rate": 3.220771255280797e-05,
      "loss": 11.8883,
      "step": 17040
    },
    {
      "epoch": 2.979846994535519,
      "grad_norm": 0.6461096382680658,
      "learning_rate": 3.219730309717569e-05,
      "loss": 11.5419,
      "step": 17041
    },
    {
      "epoch": 2.9800218579234974,
      "grad_norm": 0.5810576278592705,
      "learning_rate": 3.2186895001195693e-05,
      "loss": 11.5988,
      "step": 17042
    },
    {
      "epoch": 2.9801967213114753,
      "grad_norm": 0.572544845624717,
      "learning_rate": 3.2176488265076596e-05,
      "loss": 11.4428,
      "step": 17043
    },
    {
      "epoch": 2.9803715846994536,
      "grad_norm": 0.6321451755549957,
      "learning_rate": 3.21660828890272e-05,
      "loss": 11.6379,
      "step": 17044
    },
    {
      "epoch": 2.9805464480874315,
      "grad_norm": 0.675460240207978,
      "learning_rate": 3.215567887325611e-05,
      "loss": 12.0682,
      "step": 17045
    },
    {
      "epoch": 2.98072131147541,
      "grad_norm": 0.6441810632358042,
      "learning_rate": 3.2145276217971975e-05,
      "loss": 11.683,
      "step": 17046
    },
    {
      "epoch": 2.980896174863388,
      "grad_norm": 0.5861592030324075,
      "learning_rate": 3.213487492338342e-05,
      "loss": 11.604,
      "step": 17047
    },
    {
      "epoch": 2.981071038251366,
      "grad_norm": 0.6122659600033432,
      "learning_rate": 3.2124474989698926e-05,
      "loss": 11.6697,
      "step": 17048
    },
    {
      "epoch": 2.981245901639344,
      "grad_norm": 0.6293418800479746,
      "learning_rate": 3.211407641712716e-05,
      "loss": 11.8562,
      "step": 17049
    },
    {
      "epoch": 2.9814207650273223,
      "grad_norm": 0.6478048730475789,
      "learning_rate": 3.2103679205876594e-05,
      "loss": 11.6129,
      "step": 17050
    },
    {
      "epoch": 2.9815956284153007,
      "grad_norm": 0.6164084176612576,
      "learning_rate": 3.209328335615576e-05,
      "loss": 11.7297,
      "step": 17051
    },
    {
      "epoch": 2.9817704918032786,
      "grad_norm": 0.697545812090867,
      "learning_rate": 3.208288886817306e-05,
      "loss": 11.5713,
      "step": 17052
    },
    {
      "epoch": 2.981945355191257,
      "grad_norm": 0.6408975478085785,
      "learning_rate": 3.2072495742136944e-05,
      "loss": 11.6309,
      "step": 17053
    },
    {
      "epoch": 2.982120218579235,
      "grad_norm": 0.7627374008867814,
      "learning_rate": 3.2062103978255875e-05,
      "loss": 11.7458,
      "step": 17054
    },
    {
      "epoch": 2.982295081967213,
      "grad_norm": 0.6432517264226951,
      "learning_rate": 3.205171357673823e-05,
      "loss": 11.5532,
      "step": 17055
    },
    {
      "epoch": 2.9824699453551915,
      "grad_norm": 0.6641032194922568,
      "learning_rate": 3.2041324537792385e-05,
      "loss": 11.7898,
      "step": 17056
    },
    {
      "epoch": 2.9826448087431694,
      "grad_norm": 0.652822010795314,
      "learning_rate": 3.203093686162658e-05,
      "loss": 11.6392,
      "step": 17057
    },
    {
      "epoch": 2.9828196721311473,
      "grad_norm": 0.6208458359870196,
      "learning_rate": 3.202055054844921e-05,
      "loss": 11.9613,
      "step": 17058
    },
    {
      "epoch": 2.9829945355191256,
      "grad_norm": 0.6161255527034092,
      "learning_rate": 3.201016559846853e-05,
      "loss": 11.6312,
      "step": 17059
    },
    {
      "epoch": 2.983169398907104,
      "grad_norm": 0.6439745990443294,
      "learning_rate": 3.199978201189282e-05,
      "loss": 11.9724,
      "step": 17060
    },
    {
      "epoch": 2.983344262295082,
      "grad_norm": 0.7015596160471163,
      "learning_rate": 3.198939978893023e-05,
      "loss": 11.986,
      "step": 17061
    },
    {
      "epoch": 2.98351912568306,
      "grad_norm": 0.5654587499557334,
      "learning_rate": 3.197901892978897e-05,
      "loss": 11.8468,
      "step": 17062
    },
    {
      "epoch": 2.983693989071038,
      "grad_norm": 0.7065170299705662,
      "learning_rate": 3.196863943467726e-05,
      "loss": 11.6743,
      "step": 17063
    },
    {
      "epoch": 2.9838688524590165,
      "grad_norm": 0.6541042914026158,
      "learning_rate": 3.1958261303803215e-05,
      "loss": 11.9496,
      "step": 17064
    },
    {
      "epoch": 2.984043715846995,
      "grad_norm": 0.7090136047004729,
      "learning_rate": 3.194788453737497e-05,
      "loss": 11.6823,
      "step": 17065
    },
    {
      "epoch": 2.9842185792349727,
      "grad_norm": 0.6704299917476345,
      "learning_rate": 3.193750913560057e-05,
      "loss": 11.9107,
      "step": 17066
    },
    {
      "epoch": 2.9843934426229506,
      "grad_norm": 0.5995298526026321,
      "learning_rate": 3.1927135098688056e-05,
      "loss": 11.4628,
      "step": 17067
    },
    {
      "epoch": 2.984568306010929,
      "grad_norm": 0.6517852694515588,
      "learning_rate": 3.191676242684551e-05,
      "loss": 11.4656,
      "step": 17068
    },
    {
      "epoch": 2.9847431693989073,
      "grad_norm": 0.6652951744938828,
      "learning_rate": 3.1906391120280974e-05,
      "loss": 11.9601,
      "step": 17069
    },
    {
      "epoch": 2.984918032786885,
      "grad_norm": 0.6433956113390842,
      "learning_rate": 3.189602117920233e-05,
      "loss": 11.557,
      "step": 17070
    },
    {
      "epoch": 2.9850928961748635,
      "grad_norm": 0.6384205464718241,
      "learning_rate": 3.1885652603817574e-05,
      "loss": 11.6734,
      "step": 17071
    },
    {
      "epoch": 2.9852677595628414,
      "grad_norm": 0.6497628057798643,
      "learning_rate": 3.187528539433458e-05,
      "loss": 11.7402,
      "step": 17072
    },
    {
      "epoch": 2.9854426229508197,
      "grad_norm": 0.6776085274511194,
      "learning_rate": 3.1864919550961314e-05,
      "loss": 11.6166,
      "step": 17073
    },
    {
      "epoch": 2.985617486338798,
      "grad_norm": 0.7011673756440775,
      "learning_rate": 3.185455507390566e-05,
      "loss": 11.5072,
      "step": 17074
    },
    {
      "epoch": 2.985792349726776,
      "grad_norm": 0.6708432725134323,
      "learning_rate": 3.184419196337537e-05,
      "loss": 11.834,
      "step": 17075
    },
    {
      "epoch": 2.985967213114754,
      "grad_norm": 0.646841158416971,
      "learning_rate": 3.1833830219578284e-05,
      "loss": 11.8684,
      "step": 17076
    },
    {
      "epoch": 2.9861420765027322,
      "grad_norm": 0.6782341745389319,
      "learning_rate": 3.1823469842722186e-05,
      "loss": 11.9451,
      "step": 17077
    },
    {
      "epoch": 2.9863169398907106,
      "grad_norm": 0.6474094229806217,
      "learning_rate": 3.18131108330149e-05,
      "loss": 11.5231,
      "step": 17078
    },
    {
      "epoch": 2.9864918032786885,
      "grad_norm": 0.6376662959389618,
      "learning_rate": 3.180275319066408e-05,
      "loss": 11.7111,
      "step": 17079
    },
    {
      "epoch": 2.986666666666667,
      "grad_norm": 0.6245676625326945,
      "learning_rate": 3.1792396915877454e-05,
      "loss": 11.4244,
      "step": 17080
    },
    {
      "epoch": 2.9868415300546447,
      "grad_norm": 0.6463022226141325,
      "learning_rate": 3.178204200886266e-05,
      "loss": 11.6407,
      "step": 17081
    },
    {
      "epoch": 2.987016393442623,
      "grad_norm": 0.6602516715695309,
      "learning_rate": 3.17716884698274e-05,
      "loss": 11.2595,
      "step": 17082
    },
    {
      "epoch": 2.987191256830601,
      "grad_norm": 0.6012125941137345,
      "learning_rate": 3.1761336298979315e-05,
      "loss": 11.7087,
      "step": 17083
    },
    {
      "epoch": 2.9873661202185793,
      "grad_norm": 0.6170370624663336,
      "learning_rate": 3.175098549652593e-05,
      "loss": 11.847,
      "step": 17084
    },
    {
      "epoch": 2.987540983606557,
      "grad_norm": 0.6249765417662915,
      "learning_rate": 3.174063606267483e-05,
      "loss": 11.5714,
      "step": 17085
    },
    {
      "epoch": 2.9877158469945355,
      "grad_norm": 0.5870063272670568,
      "learning_rate": 3.1730287997633536e-05,
      "loss": 11.8005,
      "step": 17086
    },
    {
      "epoch": 2.987890710382514,
      "grad_norm": 0.652089169483592,
      "learning_rate": 3.171994130160965e-05,
      "loss": 11.9514,
      "step": 17087
    },
    {
      "epoch": 2.9880655737704918,
      "grad_norm": 0.6480790410244548,
      "learning_rate": 3.170959597481055e-05,
      "loss": 11.8697,
      "step": 17088
    },
    {
      "epoch": 2.98824043715847,
      "grad_norm": 0.5786900622275978,
      "learning_rate": 3.169925201744373e-05,
      "loss": 11.7509,
      "step": 17089
    },
    {
      "epoch": 2.988415300546448,
      "grad_norm": 0.6959403297365221,
      "learning_rate": 3.168890942971662e-05,
      "loss": 11.6881,
      "step": 17090
    },
    {
      "epoch": 2.9885901639344263,
      "grad_norm": 0.6568279691232337,
      "learning_rate": 3.167856821183658e-05,
      "loss": 11.8196,
      "step": 17091
    },
    {
      "epoch": 2.9887650273224042,
      "grad_norm": 0.6384519460277078,
      "learning_rate": 3.166822836401109e-05,
      "loss": 11.8709,
      "step": 17092
    },
    {
      "epoch": 2.9889398907103826,
      "grad_norm": 0.6692898909256522,
      "learning_rate": 3.1657889886447386e-05,
      "loss": 11.9743,
      "step": 17093
    },
    {
      "epoch": 2.9891147540983605,
      "grad_norm": 0.6702563819670662,
      "learning_rate": 3.164755277935284e-05,
      "loss": 11.7422,
      "step": 17094
    },
    {
      "epoch": 2.989289617486339,
      "grad_norm": 0.6605822887145797,
      "learning_rate": 3.163721704293473e-05,
      "loss": 11.8481,
      "step": 17095
    },
    {
      "epoch": 2.989464480874317,
      "grad_norm": 0.6465081908427284,
      "learning_rate": 3.162688267740031e-05,
      "loss": 11.6908,
      "step": 17096
    },
    {
      "epoch": 2.989639344262295,
      "grad_norm": 0.6816296907184525,
      "learning_rate": 3.161654968295684e-05,
      "loss": 11.9333,
      "step": 17097
    },
    {
      "epoch": 2.989814207650273,
      "grad_norm": 0.5908385996310923,
      "learning_rate": 3.160621805981151e-05,
      "loss": 11.8782,
      "step": 17098
    },
    {
      "epoch": 2.9899890710382513,
      "grad_norm": 0.6794062974110624,
      "learning_rate": 3.15958878081715e-05,
      "loss": 11.5335,
      "step": 17099
    },
    {
      "epoch": 2.9901639344262296,
      "grad_norm": 0.8028804357175631,
      "learning_rate": 3.158555892824396e-05,
      "loss": 11.7882,
      "step": 17100
    },
    {
      "epoch": 2.9903387978142075,
      "grad_norm": 0.6474173979008615,
      "learning_rate": 3.157523142023604e-05,
      "loss": 11.7855,
      "step": 17101
    },
    {
      "epoch": 2.990513661202186,
      "grad_norm": 0.637865834582248,
      "learning_rate": 3.156490528435481e-05,
      "loss": 11.6657,
      "step": 17102
    },
    {
      "epoch": 2.9906885245901638,
      "grad_norm": 0.6651854225988452,
      "learning_rate": 3.155458052080735e-05,
      "loss": 11.802,
      "step": 17103
    },
    {
      "epoch": 2.990863387978142,
      "grad_norm": 0.6575262724844527,
      "learning_rate": 3.1544257129800716e-05,
      "loss": 11.7793,
      "step": 17104
    },
    {
      "epoch": 2.9910382513661204,
      "grad_norm": 0.6910447476626946,
      "learning_rate": 3.1533935111541915e-05,
      "loss": 11.6676,
      "step": 17105
    },
    {
      "epoch": 2.9912131147540983,
      "grad_norm": 0.653982195407327,
      "learning_rate": 3.152361446623792e-05,
      "loss": 11.9862,
      "step": 17106
    },
    {
      "epoch": 2.9913879781420762,
      "grad_norm": 0.6441357973321188,
      "learning_rate": 3.151329519409572e-05,
      "loss": 11.7706,
      "step": 17107
    },
    {
      "epoch": 2.9915628415300546,
      "grad_norm": 0.7467259705709393,
      "learning_rate": 3.150297729532222e-05,
      "loss": 11.6507,
      "step": 17108
    },
    {
      "epoch": 2.991737704918033,
      "grad_norm": 0.7437858745520524,
      "learning_rate": 3.1492660770124336e-05,
      "loss": 11.8095,
      "step": 17109
    },
    {
      "epoch": 2.991912568306011,
      "grad_norm": 0.64214687350436,
      "learning_rate": 3.148234561870897e-05,
      "loss": 11.7777,
      "step": 17110
    },
    {
      "epoch": 2.992087431693989,
      "grad_norm": 0.6277804626281576,
      "learning_rate": 3.147203184128293e-05,
      "loss": 11.8865,
      "step": 17111
    },
    {
      "epoch": 2.992262295081967,
      "grad_norm": 0.6633442944284542,
      "learning_rate": 3.146171943805307e-05,
      "loss": 11.9661,
      "step": 17112
    },
    {
      "epoch": 2.9924371584699454,
      "grad_norm": 0.6223392534247152,
      "learning_rate": 3.145140840922617e-05,
      "loss": 11.3452,
      "step": 17113
    },
    {
      "epoch": 2.9926120218579237,
      "grad_norm": 0.6543586729232949,
      "learning_rate": 3.1441098755009005e-05,
      "loss": 11.4829,
      "step": 17114
    },
    {
      "epoch": 2.9927868852459016,
      "grad_norm": 0.6736386826600124,
      "learning_rate": 3.143079047560832e-05,
      "loss": 11.6549,
      "step": 17115
    },
    {
      "epoch": 2.9929617486338795,
      "grad_norm": 0.6258125442374418,
      "learning_rate": 3.1420483571230816e-05,
      "loss": 11.7816,
      "step": 17116
    },
    {
      "epoch": 2.993136612021858,
      "grad_norm": 0.6562102910698495,
      "learning_rate": 3.1410178042083185e-05,
      "loss": 11.8323,
      "step": 17117
    },
    {
      "epoch": 2.993311475409836,
      "grad_norm": 0.6992602383184408,
      "learning_rate": 3.1399873888372075e-05,
      "loss": 11.7698,
      "step": 17118
    },
    {
      "epoch": 2.993486338797814,
      "grad_norm": 0.6735134880634379,
      "learning_rate": 3.138957111030413e-05,
      "loss": 11.3153,
      "step": 17119
    },
    {
      "epoch": 2.9936612021857925,
      "grad_norm": 0.6397333387376547,
      "learning_rate": 3.137926970808595e-05,
      "loss": 11.656,
      "step": 17120
    },
    {
      "epoch": 2.9938360655737704,
      "grad_norm": 0.7101536470649796,
      "learning_rate": 3.13689696819241e-05,
      "loss": 11.9927,
      "step": 17121
    },
    {
      "epoch": 2.9940109289617487,
      "grad_norm": 0.5861198155003247,
      "learning_rate": 3.135867103202513e-05,
      "loss": 11.7122,
      "step": 17122
    },
    {
      "epoch": 2.994185792349727,
      "grad_norm": 0.6622453911164156,
      "learning_rate": 3.1348373758595564e-05,
      "loss": 11.8649,
      "step": 17123
    },
    {
      "epoch": 2.994360655737705,
      "grad_norm": 0.6815959499543925,
      "learning_rate": 3.1338077861841896e-05,
      "loss": 11.7675,
      "step": 17124
    },
    {
      "epoch": 2.994535519125683,
      "grad_norm": 0.8019922720678597,
      "learning_rate": 3.132778334197059e-05,
      "loss": 11.6803,
      "step": 17125
    },
    {
      "epoch": 2.994710382513661,
      "grad_norm": 0.6375392490288699,
      "learning_rate": 3.131749019918807e-05,
      "loss": 11.6431,
      "step": 17126
    },
    {
      "epoch": 2.9948852459016395,
      "grad_norm": 0.6345020094916412,
      "learning_rate": 3.1307198433700756e-05,
      "loss": 11.4492,
      "step": 17127
    },
    {
      "epoch": 2.9950601092896174,
      "grad_norm": 0.7066556437648074,
      "learning_rate": 3.1296908045715036e-05,
      "loss": 11.755,
      "step": 17128
    },
    {
      "epoch": 2.9952349726775958,
      "grad_norm": 0.6628295623330936,
      "learning_rate": 3.128661903543725e-05,
      "loss": 11.7597,
      "step": 17129
    },
    {
      "epoch": 2.9954098360655736,
      "grad_norm": 0.6065886823029708,
      "learning_rate": 3.1276331403073735e-05,
      "loss": 11.9542,
      "step": 17130
    },
    {
      "epoch": 2.995584699453552,
      "grad_norm": 0.6136504762719257,
      "learning_rate": 3.126604514883078e-05,
      "loss": 11.9623,
      "step": 17131
    },
    {
      "epoch": 2.99575956284153,
      "grad_norm": 0.6374500140239989,
      "learning_rate": 3.1255760272914656e-05,
      "loss": 11.9056,
      "step": 17132
    },
    {
      "epoch": 2.9959344262295082,
      "grad_norm": 0.6064602276770421,
      "learning_rate": 3.124547677553162e-05,
      "loss": 11.9467,
      "step": 17133
    },
    {
      "epoch": 2.996109289617486,
      "grad_norm": 0.6436119693999092,
      "learning_rate": 3.123519465688788e-05,
      "loss": 11.9271,
      "step": 17134
    },
    {
      "epoch": 2.9962841530054645,
      "grad_norm": 0.6566780702437138,
      "learning_rate": 3.122491391718961e-05,
      "loss": 11.8403,
      "step": 17135
    },
    {
      "epoch": 2.996459016393443,
      "grad_norm": 0.6390681125883632,
      "learning_rate": 3.1214634556643e-05,
      "loss": 11.6619,
      "step": 17136
    },
    {
      "epoch": 2.9966338797814207,
      "grad_norm": 0.6617441601655701,
      "learning_rate": 3.120435657545415e-05,
      "loss": 11.8208,
      "step": 17137
    },
    {
      "epoch": 2.996808743169399,
      "grad_norm": 0.6497118797885548,
      "learning_rate": 3.1194079973829215e-05,
      "loss": 11.3999,
      "step": 17138
    },
    {
      "epoch": 2.996983606557377,
      "grad_norm": 0.6353605562434406,
      "learning_rate": 3.118380475197419e-05,
      "loss": 11.6147,
      "step": 17139
    },
    {
      "epoch": 2.9971584699453553,
      "grad_norm": 0.6465108824239489,
      "learning_rate": 3.117353091009519e-05,
      "loss": 11.5802,
      "step": 17140
    },
    {
      "epoch": 2.997333333333333,
      "grad_norm": 0.6344739159181587,
      "learning_rate": 3.116325844839823e-05,
      "loss": 11.6328,
      "step": 17141
    },
    {
      "epoch": 2.9975081967213115,
      "grad_norm": 0.6648763285087471,
      "learning_rate": 3.115298736708928e-05,
      "loss": 11.758,
      "step": 17142
    },
    {
      "epoch": 2.9976830601092894,
      "grad_norm": 0.7063831286321698,
      "learning_rate": 3.114271766637437e-05,
      "loss": 11.903,
      "step": 17143
    },
    {
      "epoch": 2.9978579234972678,
      "grad_norm": 0.770939959176482,
      "learning_rate": 3.113244934645931e-05,
      "loss": 11.7263,
      "step": 17144
    },
    {
      "epoch": 2.998032786885246,
      "grad_norm": 0.6917283904332149,
      "learning_rate": 3.1122182407550135e-05,
      "loss": 11.7128,
      "step": 17145
    },
    {
      "epoch": 2.998207650273224,
      "grad_norm": 0.6174944636697124,
      "learning_rate": 3.111191684985269e-05,
      "loss": 11.8597,
      "step": 17146
    },
    {
      "epoch": 2.9983825136612023,
      "grad_norm": 0.6137431690511203,
      "learning_rate": 3.1101652673572854e-05,
      "loss": 11.9231,
      "step": 17147
    },
    {
      "epoch": 2.9985573770491802,
      "grad_norm": 0.6748801428320145,
      "learning_rate": 3.109138987891639e-05,
      "loss": 11.6554,
      "step": 17148
    },
    {
      "epoch": 2.9987322404371586,
      "grad_norm": 0.6431176033987472,
      "learning_rate": 3.1081128466089115e-05,
      "loss": 11.7188,
      "step": 17149
    },
    {
      "epoch": 2.9989071038251365,
      "grad_norm": 0.6515953889410575,
      "learning_rate": 3.107086843529684e-05,
      "loss": 11.4649,
      "step": 17150
    },
    {
      "epoch": 2.999081967213115,
      "grad_norm": 0.6548740334553543,
      "learning_rate": 3.1060609786745306e-05,
      "loss": 11.72,
      "step": 17151
    },
    {
      "epoch": 2.9992568306010927,
      "grad_norm": 0.6719945908045701,
      "learning_rate": 3.1050352520640245e-05,
      "loss": 11.6611,
      "step": 17152
    },
    {
      "epoch": 2.999431693989071,
      "grad_norm": 0.6949934050434943,
      "learning_rate": 3.1040096637187253e-05,
      "loss": 11.5761,
      "step": 17153
    },
    {
      "epoch": 2.9996065573770494,
      "grad_norm": 0.566572090395668,
      "learning_rate": 3.102984213659209e-05,
      "loss": 11.52,
      "step": 17154
    },
    {
      "epoch": 2.9997814207650273,
      "grad_norm": 0.5739180799178876,
      "learning_rate": 3.1019589019060356e-05,
      "loss": 11.784,
      "step": 17155
    },
    {
      "epoch": 2.999956284153005,
      "grad_norm": 0.6303411091530349,
      "learning_rate": 3.10093372847977e-05,
      "loss": 12.0894,
      "step": 17156
    },
    {
      "epoch": 3.0001311475409835,
      "grad_norm": 0.7106729183763246,
      "learning_rate": 3.0999086934009625e-05,
      "loss": 10.998,
      "step": 17157
    },
    {
      "epoch": 3.000306010928962,
      "grad_norm": 0.6612174715362318,
      "learning_rate": 3.098883796690169e-05,
      "loss": 11.5388,
      "step": 17158
    },
    {
      "epoch": 3.0004808743169398,
      "grad_norm": 0.6923534254032744,
      "learning_rate": 3.097859038367947e-05,
      "loss": 11.4099,
      "step": 17159
    },
    {
      "epoch": 3.000655737704918,
      "grad_norm": 0.6107014229546639,
      "learning_rate": 3.0968344184548435e-05,
      "loss": 11.4658,
      "step": 17160
    },
    {
      "epoch": 3.000830601092896,
      "grad_norm": 0.6210924985777937,
      "learning_rate": 3.09580993697141e-05,
      "loss": 11.4067,
      "step": 17161
    },
    {
      "epoch": 3.0010054644808744,
      "grad_norm": 0.6377529099363936,
      "learning_rate": 3.094785593938181e-05,
      "loss": 11.4991,
      "step": 17162
    },
    {
      "epoch": 3.0011803278688522,
      "grad_norm": 0.7200879595939041,
      "learning_rate": 3.0937613893757e-05,
      "loss": 11.5115,
      "step": 17163
    },
    {
      "epoch": 3.0013551912568306,
      "grad_norm": 0.6575175027306085,
      "learning_rate": 3.092737323304512e-05,
      "loss": 11.378,
      "step": 17164
    },
    {
      "epoch": 3.001530054644809,
      "grad_norm": 0.7102866578194814,
      "learning_rate": 3.091713395745152e-05,
      "loss": 11.6147,
      "step": 17165
    },
    {
      "epoch": 3.001704918032787,
      "grad_norm": 0.6882773938126625,
      "learning_rate": 3.090689606718146e-05,
      "loss": 11.2692,
      "step": 17166
    },
    {
      "epoch": 3.001879781420765,
      "grad_norm": 0.7015433969865326,
      "learning_rate": 3.0896659562440265e-05,
      "loss": 11.5012,
      "step": 17167
    },
    {
      "epoch": 3.002054644808743,
      "grad_norm": 0.7630715621824006,
      "learning_rate": 3.08864244434332e-05,
      "loss": 11.366,
      "step": 17168
    },
    {
      "epoch": 3.0022295081967214,
      "grad_norm": 0.5635850575876371,
      "learning_rate": 3.087619071036557e-05,
      "loss": 11.227,
      "step": 17169
    },
    {
      "epoch": 3.0024043715846993,
      "grad_norm": 0.6964796513802004,
      "learning_rate": 3.0865958363442584e-05,
      "loss": 11.655,
      "step": 17170
    },
    {
      "epoch": 3.0025792349726776,
      "grad_norm": 0.6545421326539286,
      "learning_rate": 3.085572740286936e-05,
      "loss": 11.6692,
      "step": 17171
    },
    {
      "epoch": 3.0027540983606555,
      "grad_norm": 0.6629894023095236,
      "learning_rate": 3.0845497828851114e-05,
      "loss": 11.3859,
      "step": 17172
    },
    {
      "epoch": 3.002928961748634,
      "grad_norm": 0.6334398692636856,
      "learning_rate": 3.083526964159292e-05,
      "loss": 11.4353,
      "step": 17173
    },
    {
      "epoch": 3.0031038251366122,
      "grad_norm": 0.6815264209958048,
      "learning_rate": 3.082504284130001e-05,
      "loss": 11.6317,
      "step": 17174
    },
    {
      "epoch": 3.00327868852459,
      "grad_norm": 0.6382980318351462,
      "learning_rate": 3.081481742817736e-05,
      "loss": 11.5452,
      "step": 17175
    },
    {
      "epoch": 3.0034535519125685,
      "grad_norm": 0.6471264019110968,
      "learning_rate": 3.0804593402430036e-05,
      "loss": 11.0173,
      "step": 17176
    },
    {
      "epoch": 3.0036284153005464,
      "grad_norm": 0.5865929292427982,
      "learning_rate": 3.0794370764263045e-05,
      "loss": 11.2284,
      "step": 17177
    },
    {
      "epoch": 3.0038032786885247,
      "grad_norm": 0.7510026059497202,
      "learning_rate": 3.078414951388144e-05,
      "loss": 11.4519,
      "step": 17178
    },
    {
      "epoch": 3.0039781420765026,
      "grad_norm": 0.6529006962115255,
      "learning_rate": 3.077392965149019e-05,
      "loss": 11.5596,
      "step": 17179
    },
    {
      "epoch": 3.004153005464481,
      "grad_norm": 0.6602939535723896,
      "learning_rate": 3.0763711177294175e-05,
      "loss": 11.3863,
      "step": 17180
    },
    {
      "epoch": 3.004327868852459,
      "grad_norm": 0.5771060292380805,
      "learning_rate": 3.075349409149832e-05,
      "loss": 11.0913,
      "step": 17181
    },
    {
      "epoch": 3.004502732240437,
      "grad_norm": 0.6747980097584909,
      "learning_rate": 3.0743278394307505e-05,
      "loss": 11.2414,
      "step": 17182
    },
    {
      "epoch": 3.0046775956284155,
      "grad_norm": 0.6123965414367977,
      "learning_rate": 3.073306408592667e-05,
      "loss": 11.3402,
      "step": 17183
    },
    {
      "epoch": 3.0048524590163934,
      "grad_norm": 0.6732391470568523,
      "learning_rate": 3.072285116656053e-05,
      "loss": 11.5882,
      "step": 17184
    },
    {
      "epoch": 3.0050273224043718,
      "grad_norm": 0.6869306688832055,
      "learning_rate": 3.0712639636413944e-05,
      "loss": 11.569,
      "step": 17185
    },
    {
      "epoch": 3.0052021857923497,
      "grad_norm": 0.643685097395783,
      "learning_rate": 3.0702429495691675e-05,
      "loss": 11.3489,
      "step": 17186
    },
    {
      "epoch": 3.005377049180328,
      "grad_norm": 0.6101174308957664,
      "learning_rate": 3.069222074459843e-05,
      "loss": 11.4909,
      "step": 17187
    },
    {
      "epoch": 3.005551912568306,
      "grad_norm": 0.6555890849404662,
      "learning_rate": 3.068201338333903e-05,
      "loss": 11.3331,
      "step": 17188
    },
    {
      "epoch": 3.0057267759562842,
      "grad_norm": 0.6849291350422702,
      "learning_rate": 3.067180741211806e-05,
      "loss": 11.4481,
      "step": 17189
    },
    {
      "epoch": 3.005901639344262,
      "grad_norm": 0.5612259560870072,
      "learning_rate": 3.0661602831140214e-05,
      "loss": 11.5059,
      "step": 17190
    },
    {
      "epoch": 3.0060765027322405,
      "grad_norm": 0.6433563425936997,
      "learning_rate": 3.065139964061014e-05,
      "loss": 11.0511,
      "step": 17191
    },
    {
      "epoch": 3.0062513661202184,
      "grad_norm": 0.6408790809008809,
      "learning_rate": 3.064119784073242e-05,
      "loss": 11.4444,
      "step": 17192
    },
    {
      "epoch": 3.0064262295081967,
      "grad_norm": 0.6862788136612532,
      "learning_rate": 3.0630997431711636e-05,
      "loss": 11.3225,
      "step": 17193
    },
    {
      "epoch": 3.006601092896175,
      "grad_norm": 0.658488881456411,
      "learning_rate": 3.0620798413752363e-05,
      "loss": 11.6292,
      "step": 17194
    },
    {
      "epoch": 3.006775956284153,
      "grad_norm": 0.6094199731507608,
      "learning_rate": 3.0610600787059086e-05,
      "loss": 11.3158,
      "step": 17195
    },
    {
      "epoch": 3.0069508196721313,
      "grad_norm": 0.6851419775527819,
      "learning_rate": 3.0600404551836324e-05,
      "loss": 11.4813,
      "step": 17196
    },
    {
      "epoch": 3.007125683060109,
      "grad_norm": 0.7116682579651568,
      "learning_rate": 3.059020970828854e-05,
      "loss": 11.4671,
      "step": 17197
    },
    {
      "epoch": 3.0073005464480875,
      "grad_norm": 0.6257141011372284,
      "learning_rate": 3.058001625662017e-05,
      "loss": 11.4083,
      "step": 17198
    },
    {
      "epoch": 3.0074754098360654,
      "grad_norm": 0.604478235189052,
      "learning_rate": 3.056982419703561e-05,
      "loss": 11.4535,
      "step": 17199
    },
    {
      "epoch": 3.0076502732240438,
      "grad_norm": 0.6367100912014487,
      "learning_rate": 3.0559633529739265e-05,
      "loss": 11.2729,
      "step": 17200
    },
    {
      "epoch": 3.0078251366120217,
      "grad_norm": 0.6199018594741647,
      "learning_rate": 3.0549444254935466e-05,
      "loss": 11.3788,
      "step": 17201
    },
    {
      "epoch": 3.008,
      "grad_norm": 0.6499607246459562,
      "learning_rate": 3.053925637282856e-05,
      "loss": 11.5155,
      "step": 17202
    },
    {
      "epoch": 3.0081748633879783,
      "grad_norm": 0.6109598957929057,
      "learning_rate": 3.0529069883622826e-05,
      "loss": 11.5802,
      "step": 17203
    },
    {
      "epoch": 3.0083497267759562,
      "grad_norm": 0.6015992118915167,
      "learning_rate": 3.051888478752256e-05,
      "loss": 11.4819,
      "step": 17204
    },
    {
      "epoch": 3.0085245901639346,
      "grad_norm": 0.648280058537349,
      "learning_rate": 3.0508701084731983e-05,
      "loss": 11.5551,
      "step": 17205
    },
    {
      "epoch": 3.0086994535519125,
      "grad_norm": 0.6281911039369529,
      "learning_rate": 3.0498518775455308e-05,
      "loss": 11.5265,
      "step": 17206
    },
    {
      "epoch": 3.008874316939891,
      "grad_norm": 0.7522710010127232,
      "learning_rate": 3.048833785989673e-05,
      "loss": 11.2534,
      "step": 17207
    },
    {
      "epoch": 3.0090491803278687,
      "grad_norm": 0.6139801354786427,
      "learning_rate": 3.047815833826042e-05,
      "loss": 11.7118,
      "step": 17208
    },
    {
      "epoch": 3.009224043715847,
      "grad_norm": 0.6600366393697782,
      "learning_rate": 3.046798021075048e-05,
      "loss": 11.5678,
      "step": 17209
    },
    {
      "epoch": 3.009398907103825,
      "grad_norm": 0.6322094014833759,
      "learning_rate": 3.0457803477571033e-05,
      "loss": 11.3045,
      "step": 17210
    },
    {
      "epoch": 3.0095737704918033,
      "grad_norm": 0.6726963104639465,
      "learning_rate": 3.0447628138926156e-05,
      "loss": 11.3559,
      "step": 17211
    },
    {
      "epoch": 3.0097486338797816,
      "grad_norm": 0.6386722549025003,
      "learning_rate": 3.043745419501989e-05,
      "loss": 11.4825,
      "step": 17212
    },
    {
      "epoch": 3.0099234972677595,
      "grad_norm": 0.7695788744984764,
      "learning_rate": 3.0427281646056238e-05,
      "loss": 11.5285,
      "step": 17213
    },
    {
      "epoch": 3.010098360655738,
      "grad_norm": 0.637245682605003,
      "learning_rate": 3.041711049223921e-05,
      "loss": 11.2712,
      "step": 17214
    },
    {
      "epoch": 3.010273224043716,
      "grad_norm": 0.7573395090901136,
      "learning_rate": 3.040694073377277e-05,
      "loss": 11.4612,
      "step": 17215
    },
    {
      "epoch": 3.010448087431694,
      "grad_norm": 0.6422140786794462,
      "learning_rate": 3.039677237086085e-05,
      "loss": 11.0615,
      "step": 17216
    },
    {
      "epoch": 3.010622950819672,
      "grad_norm": 0.6959725146476327,
      "learning_rate": 3.0386605403707346e-05,
      "loss": 11.163,
      "step": 17217
    },
    {
      "epoch": 3.0107978142076504,
      "grad_norm": 0.7142077809101265,
      "learning_rate": 3.0376439832516147e-05,
      "loss": 11.4184,
      "step": 17218
    },
    {
      "epoch": 3.0109726775956283,
      "grad_norm": 0.6534204994776671,
      "learning_rate": 3.0366275657491093e-05,
      "loss": 11.6098,
      "step": 17219
    },
    {
      "epoch": 3.0111475409836066,
      "grad_norm": 0.6055661303775585,
      "learning_rate": 3.035611287883603e-05,
      "loss": 11.5988,
      "step": 17220
    },
    {
      "epoch": 3.0113224043715845,
      "grad_norm": 0.6632600830790151,
      "learning_rate": 3.0345951496754733e-05,
      "loss": 11.5944,
      "step": 17221
    },
    {
      "epoch": 3.011497267759563,
      "grad_norm": 0.6281246650388685,
      "learning_rate": 3.0335791511450973e-05,
      "loss": 11.1828,
      "step": 17222
    },
    {
      "epoch": 3.011672131147541,
      "grad_norm": 0.6241210704127826,
      "learning_rate": 3.03256329231285e-05,
      "loss": 11.6824,
      "step": 17223
    },
    {
      "epoch": 3.011846994535519,
      "grad_norm": 0.6518015500415135,
      "learning_rate": 3.0315475731991006e-05,
      "loss": 11.4167,
      "step": 17224
    },
    {
      "epoch": 3.0120218579234974,
      "grad_norm": 0.6195641095290535,
      "learning_rate": 3.030531993824218e-05,
      "loss": 11.6043,
      "step": 17225
    },
    {
      "epoch": 3.0121967213114753,
      "grad_norm": 0.6581784771983948,
      "learning_rate": 3.029516554208568e-05,
      "loss": 11.7622,
      "step": 17226
    },
    {
      "epoch": 3.0123715846994537,
      "grad_norm": 0.6601193313089326,
      "learning_rate": 3.0285012543725145e-05,
      "loss": 11.287,
      "step": 17227
    },
    {
      "epoch": 3.0125464480874315,
      "grad_norm": 0.6936725143654606,
      "learning_rate": 3.027486094336415e-05,
      "loss": 11.1081,
      "step": 17228
    },
    {
      "epoch": 3.01272131147541,
      "grad_norm": 0.6616697153058494,
      "learning_rate": 3.0264710741206283e-05,
      "loss": 11.4754,
      "step": 17229
    },
    {
      "epoch": 3.012896174863388,
      "grad_norm": 0.6452028259786221,
      "learning_rate": 3.0254561937455083e-05,
      "loss": 11.5678,
      "step": 17230
    },
    {
      "epoch": 3.013071038251366,
      "grad_norm": 0.6373033914609509,
      "learning_rate": 3.0244414532314057e-05,
      "loss": 11.4924,
      "step": 17231
    },
    {
      "epoch": 3.0132459016393445,
      "grad_norm": 0.7327778918826156,
      "learning_rate": 3.0234268525986696e-05,
      "loss": 11.694,
      "step": 17232
    },
    {
      "epoch": 3.0134207650273224,
      "grad_norm": 0.71391495660604,
      "learning_rate": 3.022412391867646e-05,
      "loss": 11.5428,
      "step": 17233
    },
    {
      "epoch": 3.0135956284153007,
      "grad_norm": 0.6479120273910801,
      "learning_rate": 3.021398071058682e-05,
      "loss": 11.4771,
      "step": 17234
    },
    {
      "epoch": 3.0137704918032786,
      "grad_norm": 0.6443042832153236,
      "learning_rate": 3.0203838901921066e-05,
      "loss": 11.4491,
      "step": 17235
    },
    {
      "epoch": 3.013945355191257,
      "grad_norm": 0.6291809630357383,
      "learning_rate": 3.019369849288268e-05,
      "loss": 11.2715,
      "step": 17236
    },
    {
      "epoch": 3.014120218579235,
      "grad_norm": 0.6673773947675214,
      "learning_rate": 3.0183559483674974e-05,
      "loss": 11.4587,
      "step": 17237
    },
    {
      "epoch": 3.014295081967213,
      "grad_norm": 0.7215570178027082,
      "learning_rate": 3.0173421874501262e-05,
      "loss": 11.4765,
      "step": 17238
    },
    {
      "epoch": 3.014469945355191,
      "grad_norm": 0.6302715701990761,
      "learning_rate": 3.0163285665564868e-05,
      "loss": 11.6668,
      "step": 17239
    },
    {
      "epoch": 3.0146448087431694,
      "grad_norm": 0.6602982574002401,
      "learning_rate": 3.0153150857068967e-05,
      "loss": 11.5138,
      "step": 17240
    },
    {
      "epoch": 3.0148196721311473,
      "grad_norm": 0.6867345648762934,
      "learning_rate": 3.0143017449216872e-05,
      "loss": 11.4583,
      "step": 17241
    },
    {
      "epoch": 3.0149945355191257,
      "grad_norm": 0.6353841671017716,
      "learning_rate": 3.0132885442211768e-05,
      "loss": 11.5311,
      "step": 17242
    },
    {
      "epoch": 3.015169398907104,
      "grad_norm": 0.6462431779429438,
      "learning_rate": 3.0122754836256872e-05,
      "loss": 11.6836,
      "step": 17243
    },
    {
      "epoch": 3.015344262295082,
      "grad_norm": 0.6642656999250864,
      "learning_rate": 3.011262563155525e-05,
      "loss": 11.7053,
      "step": 17244
    },
    {
      "epoch": 3.0155191256830602,
      "grad_norm": 0.6581935333777106,
      "learning_rate": 3.010249782831004e-05,
      "loss": 11.2662,
      "step": 17245
    },
    {
      "epoch": 3.015693989071038,
      "grad_norm": 0.647133407924541,
      "learning_rate": 3.0092371426724398e-05,
      "loss": 11.6496,
      "step": 17246
    },
    {
      "epoch": 3.0158688524590165,
      "grad_norm": 0.6987448166677713,
      "learning_rate": 3.0082246427001347e-05,
      "loss": 11.4872,
      "step": 17247
    },
    {
      "epoch": 3.0160437158469944,
      "grad_norm": 0.6214100329897504,
      "learning_rate": 3.007212282934396e-05,
      "loss": 11.3407,
      "step": 17248
    },
    {
      "epoch": 3.0162185792349727,
      "grad_norm": 0.5878918918404336,
      "learning_rate": 3.0062000633955156e-05,
      "loss": 11.4539,
      "step": 17249
    },
    {
      "epoch": 3.0163934426229506,
      "grad_norm": 0.6310179852356012,
      "learning_rate": 3.0051879841038013e-05,
      "loss": 11.5489,
      "step": 17250
    },
    {
      "epoch": 3.016568306010929,
      "grad_norm": 0.6488899081309264,
      "learning_rate": 3.0041760450795442e-05,
      "loss": 11.5067,
      "step": 17251
    },
    {
      "epoch": 3.0167431693989073,
      "grad_norm": 0.681870780165754,
      "learning_rate": 3.0031642463430408e-05,
      "loss": 11.3618,
      "step": 17252
    },
    {
      "epoch": 3.016918032786885,
      "grad_norm": 0.6721027320339323,
      "learning_rate": 3.002152587914574e-05,
      "loss": 11.3532,
      "step": 17253
    },
    {
      "epoch": 3.0170928961748635,
      "grad_norm": 0.6256927804766931,
      "learning_rate": 3.0011410698144314e-05,
      "loss": 11.0195,
      "step": 17254
    },
    {
      "epoch": 3.0172677595628414,
      "grad_norm": 0.6504669525695018,
      "learning_rate": 3.000129692062903e-05,
      "loss": 11.288,
      "step": 17255
    },
    {
      "epoch": 3.0174426229508198,
      "grad_norm": 0.7363807393458456,
      "learning_rate": 2.9991184546802663e-05,
      "loss": 11.2539,
      "step": 17256
    },
    {
      "epoch": 3.0176174863387977,
      "grad_norm": 0.6769460939349603,
      "learning_rate": 2.9981073576868034e-05,
      "loss": 11.6614,
      "step": 17257
    },
    {
      "epoch": 3.017792349726776,
      "grad_norm": 0.5870176031544768,
      "learning_rate": 2.9970964011027837e-05,
      "loss": 11.4091,
      "step": 17258
    },
    {
      "epoch": 3.017967213114754,
      "grad_norm": 0.6157245907978505,
      "learning_rate": 2.9960855849484782e-05,
      "loss": 11.2134,
      "step": 17259
    },
    {
      "epoch": 3.0181420765027323,
      "grad_norm": 0.681000457379947,
      "learning_rate": 2.9950749092441664e-05,
      "loss": 11.2229,
      "step": 17260
    },
    {
      "epoch": 3.0183169398907106,
      "grad_norm": 0.6628503438948707,
      "learning_rate": 2.9940643740101137e-05,
      "loss": 11.5287,
      "step": 17261
    },
    {
      "epoch": 3.0184918032786885,
      "grad_norm": 0.6218949339140377,
      "learning_rate": 2.993053979266577e-05,
      "loss": 11.159,
      "step": 17262
    },
    {
      "epoch": 3.018666666666667,
      "grad_norm": 0.6274795635435088,
      "learning_rate": 2.992043725033823e-05,
      "loss": 11.4517,
      "step": 17263
    },
    {
      "epoch": 3.0188415300546447,
      "grad_norm": 0.6320693296462251,
      "learning_rate": 2.9910336113321068e-05,
      "loss": 11.4974,
      "step": 17264
    },
    {
      "epoch": 3.019016393442623,
      "grad_norm": 0.6824629805135057,
      "learning_rate": 2.9900236381816893e-05,
      "loss": 11.3468,
      "step": 17265
    },
    {
      "epoch": 3.019191256830601,
      "grad_norm": 0.7558028780548134,
      "learning_rate": 2.989013805602825e-05,
      "loss": 11.3228,
      "step": 17266
    },
    {
      "epoch": 3.0193661202185793,
      "grad_norm": 0.6238632278243811,
      "learning_rate": 2.9880041136157565e-05,
      "loss": 11.4034,
      "step": 17267
    },
    {
      "epoch": 3.019540983606557,
      "grad_norm": 0.6606947654905787,
      "learning_rate": 2.986994562240737e-05,
      "loss": 11.5944,
      "step": 17268
    },
    {
      "epoch": 3.0197158469945355,
      "grad_norm": 0.57395450547448,
      "learning_rate": 2.985985151498004e-05,
      "loss": 11.7376,
      "step": 17269
    },
    {
      "epoch": 3.0198907103825134,
      "grad_norm": 0.6126463069725029,
      "learning_rate": 2.9849758814078123e-05,
      "loss": 11.5939,
      "step": 17270
    },
    {
      "epoch": 3.020065573770492,
      "grad_norm": 0.665133692914614,
      "learning_rate": 2.983966751990389e-05,
      "loss": 11.6978,
      "step": 17271
    },
    {
      "epoch": 3.02024043715847,
      "grad_norm": 0.6942710414338887,
      "learning_rate": 2.9829577632659754e-05,
      "loss": 11.3206,
      "step": 17272
    },
    {
      "epoch": 3.020415300546448,
      "grad_norm": 0.6512869709496407,
      "learning_rate": 2.9819489152547996e-05,
      "loss": 11.4941,
      "step": 17273
    },
    {
      "epoch": 3.0205901639344264,
      "grad_norm": 0.712250269759999,
      "learning_rate": 2.980940207977101e-05,
      "loss": 11.4405,
      "step": 17274
    },
    {
      "epoch": 3.0207650273224043,
      "grad_norm": 0.7013882095409326,
      "learning_rate": 2.979931641453104e-05,
      "loss": 11.3666,
      "step": 17275
    },
    {
      "epoch": 3.0209398907103826,
      "grad_norm": 0.7063381427122939,
      "learning_rate": 2.97892321570303e-05,
      "loss": 11.4236,
      "step": 17276
    },
    {
      "epoch": 3.0211147540983605,
      "grad_norm": 0.6479144483677118,
      "learning_rate": 2.977914930747102e-05,
      "loss": 11.6581,
      "step": 17277
    },
    {
      "epoch": 3.021289617486339,
      "grad_norm": 0.7366524087676724,
      "learning_rate": 2.976906786605538e-05,
      "loss": 11.3073,
      "step": 17278
    },
    {
      "epoch": 3.0214644808743167,
      "grad_norm": 0.6401400937376863,
      "learning_rate": 2.975898783298563e-05,
      "loss": 11.3478,
      "step": 17279
    },
    {
      "epoch": 3.021639344262295,
      "grad_norm": 0.6886276499162153,
      "learning_rate": 2.9748909208463804e-05,
      "loss": 11.4407,
      "step": 17280
    },
    {
      "epoch": 3.0218142076502734,
      "grad_norm": 0.7099264481029613,
      "learning_rate": 2.9738831992692052e-05,
      "loss": 10.9963,
      "step": 17281
    },
    {
      "epoch": 3.0219890710382513,
      "grad_norm": 0.7287325706010119,
      "learning_rate": 2.972875618587245e-05,
      "loss": 11.5795,
      "step": 17282
    },
    {
      "epoch": 3.0221639344262297,
      "grad_norm": 0.6746792304506799,
      "learning_rate": 2.9718681788207016e-05,
      "loss": 11.4748,
      "step": 17283
    },
    {
      "epoch": 3.0223387978142076,
      "grad_norm": 0.6519658455107099,
      "learning_rate": 2.9708608799897865e-05,
      "loss": 11.5083,
      "step": 17284
    },
    {
      "epoch": 3.022513661202186,
      "grad_norm": 0.7353474713689754,
      "learning_rate": 2.9698537221146904e-05,
      "loss": 11.194,
      "step": 17285
    },
    {
      "epoch": 3.022688524590164,
      "grad_norm": 0.6479922891264004,
      "learning_rate": 2.968846705215612e-05,
      "loss": 11.3348,
      "step": 17286
    },
    {
      "epoch": 3.022863387978142,
      "grad_norm": 0.6711306288657707,
      "learning_rate": 2.9678398293127452e-05,
      "loss": 11.5722,
      "step": 17287
    },
    {
      "epoch": 3.02303825136612,
      "grad_norm": 0.656545862628213,
      "learning_rate": 2.9668330944262823e-05,
      "loss": 11.4887,
      "step": 17288
    },
    {
      "epoch": 3.0232131147540984,
      "grad_norm": 0.6167686980457318,
      "learning_rate": 2.9658265005764098e-05,
      "loss": 11.5365,
      "step": 17289
    },
    {
      "epoch": 3.0233879781420767,
      "grad_norm": 0.6666440135220345,
      "learning_rate": 2.9648200477833144e-05,
      "loss": 11.4272,
      "step": 17290
    },
    {
      "epoch": 3.0235628415300546,
      "grad_norm": 0.6938398162232194,
      "learning_rate": 2.963813736067178e-05,
      "loss": 11.4488,
      "step": 17291
    },
    {
      "epoch": 3.023737704918033,
      "grad_norm": 0.6602166223075356,
      "learning_rate": 2.962807565448179e-05,
      "loss": 11.445,
      "step": 17292
    },
    {
      "epoch": 3.023912568306011,
      "grad_norm": 0.636103263408412,
      "learning_rate": 2.9618015359464958e-05,
      "loss": 11.4914,
      "step": 17293
    },
    {
      "epoch": 3.024087431693989,
      "grad_norm": 0.7005211696646306,
      "learning_rate": 2.960795647582303e-05,
      "loss": 11.5328,
      "step": 17294
    },
    {
      "epoch": 3.024262295081967,
      "grad_norm": 0.6431742509058773,
      "learning_rate": 2.959789900375769e-05,
      "loss": 11.1724,
      "step": 17295
    },
    {
      "epoch": 3.0244371584699454,
      "grad_norm": 0.6077189864978149,
      "learning_rate": 2.9587842943470647e-05,
      "loss": 11.6656,
      "step": 17296
    },
    {
      "epoch": 3.0246120218579233,
      "grad_norm": 0.6675529174061714,
      "learning_rate": 2.9577788295163545e-05,
      "loss": 11.6799,
      "step": 17297
    },
    {
      "epoch": 3.0247868852459017,
      "grad_norm": 0.6863878902146023,
      "learning_rate": 2.9567735059038014e-05,
      "loss": 11.5688,
      "step": 17298
    },
    {
      "epoch": 3.0249617486338796,
      "grad_norm": 0.6582020274160804,
      "learning_rate": 2.9557683235295653e-05,
      "loss": 11.5559,
      "step": 17299
    },
    {
      "epoch": 3.025136612021858,
      "grad_norm": 0.623621123492818,
      "learning_rate": 2.9547632824138015e-05,
      "loss": 11.617,
      "step": 17300
    },
    {
      "epoch": 3.0253114754098362,
      "grad_norm": 0.6339578296005635,
      "learning_rate": 2.9537583825766667e-05,
      "loss": 11.5655,
      "step": 17301
    },
    {
      "epoch": 3.025486338797814,
      "grad_norm": 0.6472090650296058,
      "learning_rate": 2.9527536240383112e-05,
      "loss": 11.4586,
      "step": 17302
    },
    {
      "epoch": 3.0256612021857925,
      "grad_norm": 0.6107000855711868,
      "learning_rate": 2.9517490068188835e-05,
      "loss": 11.6622,
      "step": 17303
    },
    {
      "epoch": 3.0258360655737704,
      "grad_norm": 0.6318976450714048,
      "learning_rate": 2.9507445309385294e-05,
      "loss": 11.5948,
      "step": 17304
    },
    {
      "epoch": 3.0260109289617487,
      "grad_norm": 0.6929643861959167,
      "learning_rate": 2.949740196417391e-05,
      "loss": 11.39,
      "step": 17305
    },
    {
      "epoch": 3.0261857923497266,
      "grad_norm": 0.7057913633917287,
      "learning_rate": 2.9487360032756084e-05,
      "loss": 11.4619,
      "step": 17306
    },
    {
      "epoch": 3.026360655737705,
      "grad_norm": 0.6771295765261194,
      "learning_rate": 2.94773195153332e-05,
      "loss": 11.4947,
      "step": 17307
    },
    {
      "epoch": 3.026535519125683,
      "grad_norm": 0.7009496676045981,
      "learning_rate": 2.946728041210659e-05,
      "loss": 11.4087,
      "step": 17308
    },
    {
      "epoch": 3.026710382513661,
      "grad_norm": 0.6398370278009038,
      "learning_rate": 2.9457242723277566e-05,
      "loss": 11.2125,
      "step": 17309
    },
    {
      "epoch": 3.0268852459016395,
      "grad_norm": 0.7048403724891568,
      "learning_rate": 2.9447206449047427e-05,
      "loss": 11.4784,
      "step": 17310
    },
    {
      "epoch": 3.0270601092896174,
      "grad_norm": 0.7143501759462603,
      "learning_rate": 2.943717158961743e-05,
      "loss": 11.6473,
      "step": 17311
    },
    {
      "epoch": 3.027234972677596,
      "grad_norm": 0.624181781446546,
      "learning_rate": 2.942713814518878e-05,
      "loss": 11.2489,
      "step": 17312
    },
    {
      "epoch": 3.0274098360655737,
      "grad_norm": 0.6398678199138074,
      "learning_rate": 2.941710611596271e-05,
      "loss": 11.3928,
      "step": 17313
    },
    {
      "epoch": 3.027584699453552,
      "grad_norm": 0.6752141865776123,
      "learning_rate": 2.940707550214038e-05,
      "loss": 11.0202,
      "step": 17314
    },
    {
      "epoch": 3.02775956284153,
      "grad_norm": 0.8055205560907458,
      "learning_rate": 2.939704630392294e-05,
      "loss": 11.5891,
      "step": 17315
    },
    {
      "epoch": 3.0279344262295083,
      "grad_norm": 0.6101779960741733,
      "learning_rate": 2.9387018521511522e-05,
      "loss": 11.3963,
      "step": 17316
    },
    {
      "epoch": 3.028109289617486,
      "grad_norm": 0.7045656941266153,
      "learning_rate": 2.937699215510714e-05,
      "loss": 11.2317,
      "step": 17317
    },
    {
      "epoch": 3.0282841530054645,
      "grad_norm": 0.665976454978327,
      "learning_rate": 2.9366967204910933e-05,
      "loss": 11.1937,
      "step": 17318
    },
    {
      "epoch": 3.028459016393443,
      "grad_norm": 0.7119342757150204,
      "learning_rate": 2.9356943671123904e-05,
      "loss": 11.4066,
      "step": 17319
    },
    {
      "epoch": 3.0286338797814207,
      "grad_norm": 0.6325013345582488,
      "learning_rate": 2.934692155394706e-05,
      "loss": 11.4777,
      "step": 17320
    },
    {
      "epoch": 3.028808743169399,
      "grad_norm": 0.7044741821099663,
      "learning_rate": 2.9336900853581363e-05,
      "loss": 11.3602,
      "step": 17321
    },
    {
      "epoch": 3.028983606557377,
      "grad_norm": 0.6969255745438825,
      "learning_rate": 2.932688157022778e-05,
      "loss": 11.6067,
      "step": 17322
    },
    {
      "epoch": 3.0291584699453553,
      "grad_norm": 0.6250368540426618,
      "learning_rate": 2.9316863704087206e-05,
      "loss": 11.2654,
      "step": 17323
    },
    {
      "epoch": 3.029333333333333,
      "grad_norm": 0.6741929137633442,
      "learning_rate": 2.930684725536055e-05,
      "loss": 11.6179,
      "step": 17324
    },
    {
      "epoch": 3.0295081967213116,
      "grad_norm": 0.6478424128556904,
      "learning_rate": 2.9296832224248694e-05,
      "loss": 11.2166,
      "step": 17325
    },
    {
      "epoch": 3.0296830601092894,
      "grad_norm": 0.6767452356033884,
      "learning_rate": 2.928681861095237e-05,
      "loss": 11.4061,
      "step": 17326
    },
    {
      "epoch": 3.029857923497268,
      "grad_norm": 0.6126844353382728,
      "learning_rate": 2.9276806415672485e-05,
      "loss": 11.6481,
      "step": 17327
    },
    {
      "epoch": 3.0300327868852457,
      "grad_norm": 0.6867711624747325,
      "learning_rate": 2.926679563860978e-05,
      "loss": 11.2734,
      "step": 17328
    },
    {
      "epoch": 3.030207650273224,
      "grad_norm": 0.6295886674729836,
      "learning_rate": 2.9256786279964998e-05,
      "loss": 11.5462,
      "step": 17329
    },
    {
      "epoch": 3.0303825136612024,
      "grad_norm": 0.6633299786009451,
      "learning_rate": 2.9246778339938895e-05,
      "loss": 11.5511,
      "step": 17330
    },
    {
      "epoch": 3.0305573770491803,
      "grad_norm": 0.604795611164258,
      "learning_rate": 2.9236771818732077e-05,
      "loss": 11.6378,
      "step": 17331
    },
    {
      "epoch": 3.0307322404371586,
      "grad_norm": 0.6322902560948023,
      "learning_rate": 2.9226766716545272e-05,
      "loss": 11.6844,
      "step": 17332
    },
    {
      "epoch": 3.0309071038251365,
      "grad_norm": 0.6598047909897181,
      "learning_rate": 2.9216763033579097e-05,
      "loss": 11.3645,
      "step": 17333
    },
    {
      "epoch": 3.031081967213115,
      "grad_norm": 0.6295505521097583,
      "learning_rate": 2.9206760770034193e-05,
      "loss": 11.3971,
      "step": 17334
    },
    {
      "epoch": 3.0312568306010927,
      "grad_norm": 0.6387005910322521,
      "learning_rate": 2.919675992611107e-05,
      "loss": 11.2067,
      "step": 17335
    },
    {
      "epoch": 3.031431693989071,
      "grad_norm": 0.6027212181428194,
      "learning_rate": 2.918676050201027e-05,
      "loss": 11.3865,
      "step": 17336
    },
    {
      "epoch": 3.031606557377049,
      "grad_norm": 0.6152024169342173,
      "learning_rate": 2.9176762497932375e-05,
      "loss": 11.6027,
      "step": 17337
    },
    {
      "epoch": 3.0317814207650273,
      "grad_norm": 0.6230626889708976,
      "learning_rate": 2.9166765914077852e-05,
      "loss": 11.5347,
      "step": 17338
    },
    {
      "epoch": 3.0319562841530057,
      "grad_norm": 0.6839958737352435,
      "learning_rate": 2.9156770750647187e-05,
      "loss": 11.3238,
      "step": 17339
    },
    {
      "epoch": 3.0321311475409836,
      "grad_norm": 0.6985632995790343,
      "learning_rate": 2.9146777007840764e-05,
      "loss": 11.6165,
      "step": 17340
    },
    {
      "epoch": 3.032306010928962,
      "grad_norm": 0.6540695165170436,
      "learning_rate": 2.9136784685858974e-05,
      "loss": 11.6367,
      "step": 17341
    },
    {
      "epoch": 3.03248087431694,
      "grad_norm": 0.7000641021115769,
      "learning_rate": 2.9126793784902262e-05,
      "loss": 11.3704,
      "step": 17342
    },
    {
      "epoch": 3.032655737704918,
      "grad_norm": 0.6851539213976633,
      "learning_rate": 2.9116804305170965e-05,
      "loss": 11.5271,
      "step": 17343
    },
    {
      "epoch": 3.032830601092896,
      "grad_norm": 0.7096357047387206,
      "learning_rate": 2.9106816246865365e-05,
      "loss": 11.2921,
      "step": 17344
    },
    {
      "epoch": 3.0330054644808744,
      "grad_norm": 0.7002480291900172,
      "learning_rate": 2.909682961018573e-05,
      "loss": 11.4901,
      "step": 17345
    },
    {
      "epoch": 3.0331803278688523,
      "grad_norm": 0.6568654045678397,
      "learning_rate": 2.9086844395332392e-05,
      "loss": 11.4797,
      "step": 17346
    },
    {
      "epoch": 3.0333551912568306,
      "grad_norm": 0.6402052512854057,
      "learning_rate": 2.9076860602505564e-05,
      "loss": 11.4138,
      "step": 17347
    },
    {
      "epoch": 3.033530054644809,
      "grad_norm": 0.6447066759889009,
      "learning_rate": 2.906687823190547e-05,
      "loss": 11.3306,
      "step": 17348
    },
    {
      "epoch": 3.033704918032787,
      "grad_norm": 0.6533609237773872,
      "learning_rate": 2.905689728373222e-05,
      "loss": 11.3231,
      "step": 17349
    },
    {
      "epoch": 3.033879781420765,
      "grad_norm": 0.7338632433983548,
      "learning_rate": 2.904691775818599e-05,
      "loss": 11.3166,
      "step": 17350
    },
    {
      "epoch": 3.034054644808743,
      "grad_norm": 0.6214726300530828,
      "learning_rate": 2.9036939655466932e-05,
      "loss": 11.3223,
      "step": 17351
    },
    {
      "epoch": 3.0342295081967214,
      "grad_norm": 0.6497525892280369,
      "learning_rate": 2.9026962975775153e-05,
      "loss": 11.4038,
      "step": 17352
    },
    {
      "epoch": 3.0344043715846993,
      "grad_norm": 0.5928574152992038,
      "learning_rate": 2.901698771931065e-05,
      "loss": 11.4976,
      "step": 17353
    },
    {
      "epoch": 3.0345792349726777,
      "grad_norm": 0.6496036493202734,
      "learning_rate": 2.9007013886273494e-05,
      "loss": 11.1859,
      "step": 17354
    },
    {
      "epoch": 3.0347540983606556,
      "grad_norm": 0.7264139496690502,
      "learning_rate": 2.899704147686366e-05,
      "loss": 11.4725,
      "step": 17355
    },
    {
      "epoch": 3.034928961748634,
      "grad_norm": 0.6395104036905169,
      "learning_rate": 2.8987070491281176e-05,
      "loss": 11.2821,
      "step": 17356
    },
    {
      "epoch": 3.035103825136612,
      "grad_norm": 0.8535414250702786,
      "learning_rate": 2.8977100929725996e-05,
      "loss": 11.5724,
      "step": 17357
    },
    {
      "epoch": 3.03527868852459,
      "grad_norm": 0.6110407813566751,
      "learning_rate": 2.8967132792397987e-05,
      "loss": 11.3632,
      "step": 17358
    },
    {
      "epoch": 3.0354535519125685,
      "grad_norm": 0.644305604550459,
      "learning_rate": 2.8957166079497056e-05,
      "loss": 11.41,
      "step": 17359
    },
    {
      "epoch": 3.0356284153005464,
      "grad_norm": 0.6709154676911282,
      "learning_rate": 2.894720079122305e-05,
      "loss": 11.4351,
      "step": 17360
    },
    {
      "epoch": 3.0358032786885247,
      "grad_norm": 0.6586610518132348,
      "learning_rate": 2.8937236927775892e-05,
      "loss": 11.5375,
      "step": 17361
    },
    {
      "epoch": 3.0359781420765026,
      "grad_norm": 0.6559595640515115,
      "learning_rate": 2.8927274489355293e-05,
      "loss": 11.5233,
      "step": 17362
    },
    {
      "epoch": 3.036153005464481,
      "grad_norm": 0.6254202832001331,
      "learning_rate": 2.8917313476161066e-05,
      "loss": 11.1176,
      "step": 17363
    },
    {
      "epoch": 3.036327868852459,
      "grad_norm": 0.6290688448088049,
      "learning_rate": 2.890735388839295e-05,
      "loss": 11.567,
      "step": 17364
    },
    {
      "epoch": 3.036502732240437,
      "grad_norm": 0.6587980048487955,
      "learning_rate": 2.8897395726250642e-05,
      "loss": 11.4708,
      "step": 17365
    },
    {
      "epoch": 3.036677595628415,
      "grad_norm": 0.6334041911492547,
      "learning_rate": 2.8887438989933922e-05,
      "loss": 11.4983,
      "step": 17366
    },
    {
      "epoch": 3.0368524590163934,
      "grad_norm": 0.6813061720451196,
      "learning_rate": 2.887748367964237e-05,
      "loss": 11.373,
      "step": 17367
    },
    {
      "epoch": 3.037027322404372,
      "grad_norm": 0.6340473185551543,
      "learning_rate": 2.8867529795575642e-05,
      "loss": 11.172,
      "step": 17368
    },
    {
      "epoch": 3.0372021857923497,
      "grad_norm": 0.670872729335981,
      "learning_rate": 2.885757733793332e-05,
      "loss": 11.3918,
      "step": 17369
    },
    {
      "epoch": 3.037377049180328,
      "grad_norm": 0.6070822657143333,
      "learning_rate": 2.8847626306915032e-05,
      "loss": 11.5574,
      "step": 17370
    },
    {
      "epoch": 3.037551912568306,
      "grad_norm": 0.6303567838755748,
      "learning_rate": 2.883767670272034e-05,
      "loss": 11.5448,
      "step": 17371
    },
    {
      "epoch": 3.0377267759562843,
      "grad_norm": 0.6533674541771343,
      "learning_rate": 2.8827728525548692e-05,
      "loss": 11.3658,
      "step": 17372
    },
    {
      "epoch": 3.037901639344262,
      "grad_norm": 0.6253472338955497,
      "learning_rate": 2.8817781775599618e-05,
      "loss": 11.5863,
      "step": 17373
    },
    {
      "epoch": 3.0380765027322405,
      "grad_norm": 0.6688154909748536,
      "learning_rate": 2.8807836453072544e-05,
      "loss": 11.5836,
      "step": 17374
    },
    {
      "epoch": 3.0382513661202184,
      "grad_norm": 0.7249778373333065,
      "learning_rate": 2.8797892558167006e-05,
      "loss": 11.6146,
      "step": 17375
    },
    {
      "epoch": 3.0384262295081967,
      "grad_norm": 0.6291365946009938,
      "learning_rate": 2.8787950091082306e-05,
      "loss": 11.2957,
      "step": 17376
    },
    {
      "epoch": 3.038601092896175,
      "grad_norm": 0.6498834891093563,
      "learning_rate": 2.877800905201785e-05,
      "loss": 10.995,
      "step": 17377
    },
    {
      "epoch": 3.038775956284153,
      "grad_norm": 0.6575780369633822,
      "learning_rate": 2.8768069441173006e-05,
      "loss": 11.2889,
      "step": 17378
    },
    {
      "epoch": 3.0389508196721313,
      "grad_norm": 0.662405106550659,
      "learning_rate": 2.8758131258747034e-05,
      "loss": 11.4915,
      "step": 17379
    },
    {
      "epoch": 3.039125683060109,
      "grad_norm": 0.6838222571583271,
      "learning_rate": 2.8748194504939353e-05,
      "loss": 11.5472,
      "step": 17380
    },
    {
      "epoch": 3.0393005464480876,
      "grad_norm": 0.692627132600887,
      "learning_rate": 2.87382591799491e-05,
      "loss": 11.2066,
      "step": 17381
    },
    {
      "epoch": 3.0394754098360655,
      "grad_norm": 0.6684421376761056,
      "learning_rate": 2.8728325283975553e-05,
      "loss": 11.4791,
      "step": 17382
    },
    {
      "epoch": 3.039650273224044,
      "grad_norm": 0.6768310998276669,
      "learning_rate": 2.871839281721791e-05,
      "loss": 11.4186,
      "step": 17383
    },
    {
      "epoch": 3.0398251366120217,
      "grad_norm": 0.6539052292649007,
      "learning_rate": 2.8708461779875362e-05,
      "loss": 11.5399,
      "step": 17384
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.739458314771431,
      "learning_rate": 2.8698532172147053e-05,
      "loss": 11.5043,
      "step": 17385
    },
    {
      "epoch": 3.040174863387978,
      "grad_norm": 0.6224605860839031,
      "learning_rate": 2.8688603994232098e-05,
      "loss": 11.5241,
      "step": 17386
    },
    {
      "epoch": 3.0403497267759563,
      "grad_norm": 0.7094094474816718,
      "learning_rate": 2.867867724632959e-05,
      "loss": 11.437,
      "step": 17387
    },
    {
      "epoch": 3.0405245901639346,
      "grad_norm": 0.7192805919671967,
      "learning_rate": 2.866875192863857e-05,
      "loss": 11.3022,
      "step": 17388
    },
    {
      "epoch": 3.0406994535519125,
      "grad_norm": 0.6627559253330647,
      "learning_rate": 2.865882804135811e-05,
      "loss": 11.2937,
      "step": 17389
    },
    {
      "epoch": 3.040874316939891,
      "grad_norm": 0.6293470222602253,
      "learning_rate": 2.864890558468718e-05,
      "loss": 11.5837,
      "step": 17390
    },
    {
      "epoch": 3.0410491803278688,
      "grad_norm": 0.6275250782212715,
      "learning_rate": 2.8638984558824777e-05,
      "loss": 11.4844,
      "step": 17391
    },
    {
      "epoch": 3.041224043715847,
      "grad_norm": 0.6910851449414166,
      "learning_rate": 2.8629064963969842e-05,
      "loss": 11.3204,
      "step": 17392
    },
    {
      "epoch": 3.041398907103825,
      "grad_norm": 0.6422137827658062,
      "learning_rate": 2.8619146800321294e-05,
      "loss": 11.106,
      "step": 17393
    },
    {
      "epoch": 3.0415737704918033,
      "grad_norm": 0.6583184666205663,
      "learning_rate": 2.860923006807801e-05,
      "loss": 10.8993,
      "step": 17394
    },
    {
      "epoch": 3.0417486338797812,
      "grad_norm": 0.6767407759846221,
      "learning_rate": 2.8599314767438868e-05,
      "loss": 11.0104,
      "step": 17395
    },
    {
      "epoch": 3.0419234972677596,
      "grad_norm": 0.6188079536101325,
      "learning_rate": 2.8589400898602693e-05,
      "loss": 11.4457,
      "step": 17396
    },
    {
      "epoch": 3.042098360655738,
      "grad_norm": 0.6741598448777888,
      "learning_rate": 2.857948846176829e-05,
      "loss": 11.5159,
      "step": 17397
    },
    {
      "epoch": 3.042273224043716,
      "grad_norm": 0.6718834526431532,
      "learning_rate": 2.8569577457134434e-05,
      "loss": 11.377,
      "step": 17398
    },
    {
      "epoch": 3.042448087431694,
      "grad_norm": 0.756247902672072,
      "learning_rate": 2.8559667884899877e-05,
      "loss": 11.2586,
      "step": 17399
    },
    {
      "epoch": 3.042622950819672,
      "grad_norm": 0.6341574497036808,
      "learning_rate": 2.8549759745263314e-05,
      "loss": 11.3347,
      "step": 17400
    },
    {
      "epoch": 3.0427978142076504,
      "grad_norm": 0.6416694425704045,
      "learning_rate": 2.853985303842346e-05,
      "loss": 11.341,
      "step": 17401
    },
    {
      "epoch": 3.0429726775956283,
      "grad_norm": 0.6996711917400208,
      "learning_rate": 2.8529947764578957e-05,
      "loss": 11.494,
      "step": 17402
    },
    {
      "epoch": 3.0431475409836066,
      "grad_norm": 0.6649100278632103,
      "learning_rate": 2.8520043923928454e-05,
      "loss": 11.2823,
      "step": 17403
    },
    {
      "epoch": 3.0433224043715845,
      "grad_norm": 0.6964861437460124,
      "learning_rate": 2.851014151667053e-05,
      "loss": 11.4911,
      "step": 17404
    },
    {
      "epoch": 3.043497267759563,
      "grad_norm": 0.6539451306330778,
      "learning_rate": 2.8500240543003774e-05,
      "loss": 11.6153,
      "step": 17405
    },
    {
      "epoch": 3.0436721311475408,
      "grad_norm": 0.6421524867654523,
      "learning_rate": 2.849034100312674e-05,
      "loss": 11.5142,
      "step": 17406
    },
    {
      "epoch": 3.043846994535519,
      "grad_norm": 0.6374496483080904,
      "learning_rate": 2.8480442897237926e-05,
      "loss": 11.6789,
      "step": 17407
    },
    {
      "epoch": 3.0440218579234974,
      "grad_norm": 0.6457763156915559,
      "learning_rate": 2.847054622553582e-05,
      "loss": 11.5375,
      "step": 17408
    },
    {
      "epoch": 3.0441967213114753,
      "grad_norm": 0.6512039912873079,
      "learning_rate": 2.8460650988218886e-05,
      "loss": 11.2277,
      "step": 17409
    },
    {
      "epoch": 3.0443715846994537,
      "grad_norm": 0.7598491236865765,
      "learning_rate": 2.8450757185485566e-05,
      "loss": 11.5055,
      "step": 17410
    },
    {
      "epoch": 3.0445464480874316,
      "grad_norm": 0.6491574845893132,
      "learning_rate": 2.844086481753424e-05,
      "loss": 11.5748,
      "step": 17411
    },
    {
      "epoch": 3.04472131147541,
      "grad_norm": 0.6333578173777102,
      "learning_rate": 2.8430973884563316e-05,
      "loss": 11.3166,
      "step": 17412
    },
    {
      "epoch": 3.044896174863388,
      "grad_norm": 0.6957466991573162,
      "learning_rate": 2.8421084386771057e-05,
      "loss": 11.2954,
      "step": 17413
    },
    {
      "epoch": 3.045071038251366,
      "grad_norm": 0.7047644421647723,
      "learning_rate": 2.841119632435586e-05,
      "loss": 11.2035,
      "step": 17414
    },
    {
      "epoch": 3.045245901639344,
      "grad_norm": 0.6737008332246057,
      "learning_rate": 2.840130969751599e-05,
      "loss": 11.5883,
      "step": 17415
    },
    {
      "epoch": 3.0454207650273224,
      "grad_norm": 0.6922060868166454,
      "learning_rate": 2.8391424506449693e-05,
      "loss": 11.4669,
      "step": 17416
    },
    {
      "epoch": 3.0455956284153007,
      "grad_norm": 0.7189861646131193,
      "learning_rate": 2.8381540751355208e-05,
      "loss": 11.6032,
      "step": 17417
    },
    {
      "epoch": 3.0457704918032786,
      "grad_norm": 0.6502738588126694,
      "learning_rate": 2.8371658432430716e-05,
      "loss": 11.6484,
      "step": 17418
    },
    {
      "epoch": 3.045945355191257,
      "grad_norm": 0.6711323357416769,
      "learning_rate": 2.8361777549874403e-05,
      "loss": 11.2578,
      "step": 17419
    },
    {
      "epoch": 3.046120218579235,
      "grad_norm": 0.700253370633462,
      "learning_rate": 2.835189810388441e-05,
      "loss": 11.084,
      "step": 17420
    },
    {
      "epoch": 3.046295081967213,
      "grad_norm": 0.7210102596759016,
      "learning_rate": 2.834202009465888e-05,
      "loss": 11.4013,
      "step": 17421
    },
    {
      "epoch": 3.046469945355191,
      "grad_norm": 0.7020541998902211,
      "learning_rate": 2.8332143522395814e-05,
      "loss": 11.3447,
      "step": 17422
    },
    {
      "epoch": 3.0466448087431695,
      "grad_norm": 0.6426706308799902,
      "learning_rate": 2.8322268387293337e-05,
      "loss": 11.3956,
      "step": 17423
    },
    {
      "epoch": 3.0468196721311473,
      "grad_norm": 0.6524867656217708,
      "learning_rate": 2.8312394689549472e-05,
      "loss": 11.4042,
      "step": 17424
    },
    {
      "epoch": 3.0469945355191257,
      "grad_norm": 0.7265766433228373,
      "learning_rate": 2.830252242936219e-05,
      "loss": 11.4216,
      "step": 17425
    },
    {
      "epoch": 3.047169398907104,
      "grad_norm": 0.6556207551549674,
      "learning_rate": 2.829265160692951e-05,
      "loss": 11.1608,
      "step": 17426
    },
    {
      "epoch": 3.047344262295082,
      "grad_norm": 0.608499333664148,
      "learning_rate": 2.8282782222449267e-05,
      "loss": 11.512,
      "step": 17427
    },
    {
      "epoch": 3.0475191256830603,
      "grad_norm": 0.6685395746688686,
      "learning_rate": 2.8272914276119476e-05,
      "loss": 11.3902,
      "step": 17428
    },
    {
      "epoch": 3.047693989071038,
      "grad_norm": 0.6365981411591272,
      "learning_rate": 2.8263047768137997e-05,
      "loss": 11.6618,
      "step": 17429
    },
    {
      "epoch": 3.0478688524590165,
      "grad_norm": 0.6169512712854789,
      "learning_rate": 2.8253182698702686e-05,
      "loss": 11.4662,
      "step": 17430
    },
    {
      "epoch": 3.0480437158469944,
      "grad_norm": 0.6614809746036329,
      "learning_rate": 2.824331906801133e-05,
      "loss": 11.6109,
      "step": 17431
    },
    {
      "epoch": 3.0482185792349727,
      "grad_norm": 0.6431730532213421,
      "learning_rate": 2.8233456876261723e-05,
      "loss": 11.3902,
      "step": 17432
    },
    {
      "epoch": 3.0483934426229506,
      "grad_norm": 0.6670659987490534,
      "learning_rate": 2.8223596123651673e-05,
      "loss": 11.4686,
      "step": 17433
    },
    {
      "epoch": 3.048568306010929,
      "grad_norm": 0.657841183146564,
      "learning_rate": 2.821373681037892e-05,
      "loss": 11.2715,
      "step": 17434
    },
    {
      "epoch": 3.0487431693989073,
      "grad_norm": 0.6370866274778371,
      "learning_rate": 2.8203878936641182e-05,
      "loss": 11.5479,
      "step": 17435
    },
    {
      "epoch": 3.0489180327868852,
      "grad_norm": 0.7126473112921786,
      "learning_rate": 2.8194022502636075e-05,
      "loss": 11.6332,
      "step": 17436
    },
    {
      "epoch": 3.0490928961748636,
      "grad_norm": 0.6466036827056312,
      "learning_rate": 2.8184167508561265e-05,
      "loss": 11.5634,
      "step": 17437
    },
    {
      "epoch": 3.0492677595628415,
      "grad_norm": 0.666895840442808,
      "learning_rate": 2.817431395461444e-05,
      "loss": 11.5217,
      "step": 17438
    },
    {
      "epoch": 3.04944262295082,
      "grad_norm": 0.6602569315639457,
      "learning_rate": 2.8164461840993173e-05,
      "loss": 11.1575,
      "step": 17439
    },
    {
      "epoch": 3.0496174863387977,
      "grad_norm": 0.6728355681279753,
      "learning_rate": 2.8154611167894984e-05,
      "loss": 11.4232,
      "step": 17440
    },
    {
      "epoch": 3.049792349726776,
      "grad_norm": 0.686152376562545,
      "learning_rate": 2.8144761935517406e-05,
      "loss": 11.567,
      "step": 17441
    },
    {
      "epoch": 3.049967213114754,
      "grad_norm": 0.8564620480270854,
      "learning_rate": 2.8134914144058e-05,
      "loss": 11.3954,
      "step": 17442
    },
    {
      "epoch": 3.0501420765027323,
      "grad_norm": 0.6986707385445128,
      "learning_rate": 2.812506779371422e-05,
      "loss": 11.3556,
      "step": 17443
    },
    {
      "epoch": 3.05031693989071,
      "grad_norm": 0.6820280886809919,
      "learning_rate": 2.8115222884683557e-05,
      "loss": 11.4573,
      "step": 17444
    },
    {
      "epoch": 3.0504918032786885,
      "grad_norm": 0.647864104839038,
      "learning_rate": 2.8105379417163357e-05,
      "loss": 11.5101,
      "step": 17445
    },
    {
      "epoch": 3.050666666666667,
      "grad_norm": 0.6933960136702262,
      "learning_rate": 2.809553739135101e-05,
      "loss": 11.5282,
      "step": 17446
    },
    {
      "epoch": 3.0508415300546448,
      "grad_norm": 0.6329528572618175,
      "learning_rate": 2.808569680744394e-05,
      "loss": 11.4934,
      "step": 17447
    },
    {
      "epoch": 3.051016393442623,
      "grad_norm": 0.7565606590609005,
      "learning_rate": 2.8075857665639492e-05,
      "loss": 11.3754,
      "step": 17448
    },
    {
      "epoch": 3.051191256830601,
      "grad_norm": 0.6721724032286784,
      "learning_rate": 2.8066019966134904e-05,
      "loss": 11.4129,
      "step": 17449
    },
    {
      "epoch": 3.0513661202185793,
      "grad_norm": 0.6852973769917997,
      "learning_rate": 2.805618370912748e-05,
      "loss": 11.2066,
      "step": 17450
    },
    {
      "epoch": 3.0515409836065572,
      "grad_norm": 0.6326156330582591,
      "learning_rate": 2.8046348894814446e-05,
      "loss": 11.4925,
      "step": 17451
    },
    {
      "epoch": 3.0517158469945356,
      "grad_norm": 0.7495817865567795,
      "learning_rate": 2.803651552339307e-05,
      "loss": 11.0471,
      "step": 17452
    },
    {
      "epoch": 3.0518907103825135,
      "grad_norm": 0.6475508902982505,
      "learning_rate": 2.8026683595060545e-05,
      "loss": 11.5913,
      "step": 17453
    },
    {
      "epoch": 3.052065573770492,
      "grad_norm": 0.6957053536577743,
      "learning_rate": 2.801685311001396e-05,
      "loss": 11.3232,
      "step": 17454
    },
    {
      "epoch": 3.05224043715847,
      "grad_norm": 0.6722419843943374,
      "learning_rate": 2.8007024068450495e-05,
      "loss": 11.5249,
      "step": 17455
    },
    {
      "epoch": 3.052415300546448,
      "grad_norm": 0.6724867600423172,
      "learning_rate": 2.7997196470567212e-05,
      "loss": 11.346,
      "step": 17456
    },
    {
      "epoch": 3.0525901639344264,
      "grad_norm": 0.6119337927162668,
      "learning_rate": 2.7987370316561288e-05,
      "loss": 11.6634,
      "step": 17457
    },
    {
      "epoch": 3.0527650273224043,
      "grad_norm": 0.6810595790827787,
      "learning_rate": 2.7977545606629663e-05,
      "loss": 11.5385,
      "step": 17458
    },
    {
      "epoch": 3.0529398907103826,
      "grad_norm": 0.7488298943155685,
      "learning_rate": 2.7967722340969383e-05,
      "loss": 11.566,
      "step": 17459
    },
    {
      "epoch": 3.0531147540983605,
      "grad_norm": 0.6212963449022316,
      "learning_rate": 2.7957900519777437e-05,
      "loss": 11.4597,
      "step": 17460
    },
    {
      "epoch": 3.053289617486339,
      "grad_norm": 0.6466153251696771,
      "learning_rate": 2.7948080143250754e-05,
      "loss": 11.5449,
      "step": 17461
    },
    {
      "epoch": 3.0534644808743168,
      "grad_norm": 0.7629957825071629,
      "learning_rate": 2.7938261211586347e-05,
      "loss": 11.6238,
      "step": 17462
    },
    {
      "epoch": 3.053639344262295,
      "grad_norm": 0.7097966099031807,
      "learning_rate": 2.7928443724981045e-05,
      "loss": 11.5243,
      "step": 17463
    },
    {
      "epoch": 3.053814207650273,
      "grad_norm": 0.7445254406350322,
      "learning_rate": 2.7918627683631728e-05,
      "loss": 11.2565,
      "step": 17464
    },
    {
      "epoch": 3.0539890710382513,
      "grad_norm": 0.6823691380449804,
      "learning_rate": 2.7908813087735207e-05,
      "loss": 11.1733,
      "step": 17465
    },
    {
      "epoch": 3.0541639344262297,
      "grad_norm": 0.6323902409124593,
      "learning_rate": 2.789899993748841e-05,
      "loss": 11.1536,
      "step": 17466
    },
    {
      "epoch": 3.0543387978142076,
      "grad_norm": 0.6378197554514126,
      "learning_rate": 2.7889188233088005e-05,
      "loss": 11.502,
      "step": 17467
    },
    {
      "epoch": 3.054513661202186,
      "grad_norm": 0.7134504675299114,
      "learning_rate": 2.787937797473078e-05,
      "loss": 11.4157,
      "step": 17468
    },
    {
      "epoch": 3.054688524590164,
      "grad_norm": 0.634925370224679,
      "learning_rate": 2.7869569162613473e-05,
      "loss": 11.469,
      "step": 17469
    },
    {
      "epoch": 3.054863387978142,
      "grad_norm": 0.648474496702814,
      "learning_rate": 2.7859761796932738e-05,
      "loss": 11.6348,
      "step": 17470
    },
    {
      "epoch": 3.05503825136612,
      "grad_norm": 0.6609589286190102,
      "learning_rate": 2.7849955877885346e-05,
      "loss": 11.4748,
      "step": 17471
    },
    {
      "epoch": 3.0552131147540984,
      "grad_norm": 0.6920308164869118,
      "learning_rate": 2.7840151405667837e-05,
      "loss": 11.508,
      "step": 17472
    },
    {
      "epoch": 3.0553879781420763,
      "grad_norm": 0.6627360329063502,
      "learning_rate": 2.7830348380476844e-05,
      "loss": 11.56,
      "step": 17473
    },
    {
      "epoch": 3.0555628415300546,
      "grad_norm": 0.6902965615864243,
      "learning_rate": 2.7820546802508963e-05,
      "loss": 11.5769,
      "step": 17474
    },
    {
      "epoch": 3.055737704918033,
      "grad_norm": 0.6895304206001681,
      "learning_rate": 2.781074667196073e-05,
      "loss": 11.7258,
      "step": 17475
    },
    {
      "epoch": 3.055912568306011,
      "grad_norm": 0.6603111564848577,
      "learning_rate": 2.7800947989028692e-05,
      "loss": 11.5375,
      "step": 17476
    },
    {
      "epoch": 3.056087431693989,
      "grad_norm": 0.7662848584040595,
      "learning_rate": 2.779115075390931e-05,
      "loss": 11.3574,
      "step": 17477
    },
    {
      "epoch": 3.056262295081967,
      "grad_norm": 0.683018536429244,
      "learning_rate": 2.7781354966799078e-05,
      "loss": 11.3572,
      "step": 17478
    },
    {
      "epoch": 3.0564371584699455,
      "grad_norm": 0.6387125649281484,
      "learning_rate": 2.777156062789441e-05,
      "loss": 11.4197,
      "step": 17479
    },
    {
      "epoch": 3.0566120218579234,
      "grad_norm": 0.8047427602807213,
      "learning_rate": 2.776176773739173e-05,
      "loss": 11.5212,
      "step": 17480
    },
    {
      "epoch": 3.0567868852459017,
      "grad_norm": 0.6904771305419554,
      "learning_rate": 2.7751976295487402e-05,
      "loss": 11.2373,
      "step": 17481
    },
    {
      "epoch": 3.0569617486338796,
      "grad_norm": 0.684970664618615,
      "learning_rate": 2.7742186302377794e-05,
      "loss": 11.5199,
      "step": 17482
    },
    {
      "epoch": 3.057136612021858,
      "grad_norm": 0.7129592923156375,
      "learning_rate": 2.773239775825919e-05,
      "loss": 11.5807,
      "step": 17483
    },
    {
      "epoch": 3.0573114754098363,
      "grad_norm": 0.6445505708309938,
      "learning_rate": 2.772261066332792e-05,
      "loss": 11.48,
      "step": 17484
    },
    {
      "epoch": 3.057486338797814,
      "grad_norm": 0.6776664895047234,
      "learning_rate": 2.771282501778021e-05,
      "loss": 11.4947,
      "step": 17485
    },
    {
      "epoch": 3.0576612021857925,
      "grad_norm": 0.6801241521456506,
      "learning_rate": 2.770304082181232e-05,
      "loss": 11.4073,
      "step": 17486
    },
    {
      "epoch": 3.0578360655737704,
      "grad_norm": 0.7087494053984046,
      "learning_rate": 2.769325807562044e-05,
      "loss": 11.5766,
      "step": 17487
    },
    {
      "epoch": 3.0580109289617488,
      "grad_norm": 0.64415936269471,
      "learning_rate": 2.7683476779400742e-05,
      "loss": 11.2488,
      "step": 17488
    },
    {
      "epoch": 3.0581857923497267,
      "grad_norm": 0.645961173619213,
      "learning_rate": 2.767369693334938e-05,
      "loss": 11.3537,
      "step": 17489
    },
    {
      "epoch": 3.058360655737705,
      "grad_norm": 0.628877202988786,
      "learning_rate": 2.766391853766247e-05,
      "loss": 11.3553,
      "step": 17490
    },
    {
      "epoch": 3.058535519125683,
      "grad_norm": 0.6799315016341773,
      "learning_rate": 2.7654141592536086e-05,
      "loss": 11.3478,
      "step": 17491
    },
    {
      "epoch": 3.0587103825136612,
      "grad_norm": 0.6809655000567727,
      "learning_rate": 2.7644366098166297e-05,
      "loss": 11.3627,
      "step": 17492
    },
    {
      "epoch": 3.058885245901639,
      "grad_norm": 0.6777402868671558,
      "learning_rate": 2.7634592054749133e-05,
      "loss": 11.1968,
      "step": 17493
    },
    {
      "epoch": 3.0590601092896175,
      "grad_norm": 0.6393807435428993,
      "learning_rate": 2.7624819462480578e-05,
      "loss": 11.5722,
      "step": 17494
    },
    {
      "epoch": 3.059234972677596,
      "grad_norm": 0.6951905365247502,
      "learning_rate": 2.7615048321556624e-05,
      "loss": 11.5234,
      "step": 17495
    },
    {
      "epoch": 3.0594098360655737,
      "grad_norm": 0.6315726544213752,
      "learning_rate": 2.7605278632173202e-05,
      "loss": 11.5101,
      "step": 17496
    },
    {
      "epoch": 3.059584699453552,
      "grad_norm": 0.6950675538588635,
      "learning_rate": 2.7595510394526213e-05,
      "loss": 11.3674,
      "step": 17497
    },
    {
      "epoch": 3.05975956284153,
      "grad_norm": 0.5930313548647428,
      "learning_rate": 2.7585743608811565e-05,
      "loss": 11.4599,
      "step": 17498
    },
    {
      "epoch": 3.0599344262295083,
      "grad_norm": 0.6263896098779995,
      "learning_rate": 2.757597827522509e-05,
      "loss": 11.4967,
      "step": 17499
    },
    {
      "epoch": 3.060109289617486,
      "grad_norm": 0.6284635004877213,
      "learning_rate": 2.756621439396263e-05,
      "loss": 11.418,
      "step": 17500
    },
    {
      "epoch": 3.0602841530054645,
      "grad_norm": 0.6492511023464015,
      "learning_rate": 2.7556451965219965e-05,
      "loss": 11.535,
      "step": 17501
    },
    {
      "epoch": 3.0604590163934424,
      "grad_norm": 0.6679273562452894,
      "learning_rate": 2.754669098919287e-05,
      "loss": 11.3899,
      "step": 17502
    },
    {
      "epoch": 3.0606338797814208,
      "grad_norm": 0.6357239312737402,
      "learning_rate": 2.7536931466077086e-05,
      "loss": 11.4458,
      "step": 17503
    },
    {
      "epoch": 3.060808743169399,
      "grad_norm": 0.6415829198359004,
      "learning_rate": 2.752717339606832e-05,
      "loss": 11.4979,
      "step": 17504
    },
    {
      "epoch": 3.060983606557377,
      "grad_norm": 0.6569284356936073,
      "learning_rate": 2.7517416779362237e-05,
      "loss": 11.1198,
      "step": 17505
    },
    {
      "epoch": 3.0611584699453553,
      "grad_norm": 0.6796177806134034,
      "learning_rate": 2.75076616161545e-05,
      "loss": 11.5681,
      "step": 17506
    },
    {
      "epoch": 3.0613333333333332,
      "grad_norm": 0.6431522637771648,
      "learning_rate": 2.7497907906640742e-05,
      "loss": 11.429,
      "step": 17507
    },
    {
      "epoch": 3.0615081967213116,
      "grad_norm": 0.7439630713148989,
      "learning_rate": 2.7488155651016556e-05,
      "loss": 11.4648,
      "step": 17508
    },
    {
      "epoch": 3.0616830601092895,
      "grad_norm": 0.7512399512272165,
      "learning_rate": 2.7478404849477436e-05,
      "loss": 11.3583,
      "step": 17509
    },
    {
      "epoch": 3.061857923497268,
      "grad_norm": 0.6292282459598704,
      "learning_rate": 2.7468655502218998e-05,
      "loss": 11.5691,
      "step": 17510
    },
    {
      "epoch": 3.0620327868852457,
      "grad_norm": 0.7337405459692549,
      "learning_rate": 2.7458907609436714e-05,
      "loss": 11.3381,
      "step": 17511
    },
    {
      "epoch": 3.062207650273224,
      "grad_norm": 0.8001617935980276,
      "learning_rate": 2.7449161171326066e-05,
      "loss": 11.3062,
      "step": 17512
    },
    {
      "epoch": 3.0623825136612024,
      "grad_norm": 0.6691480798223567,
      "learning_rate": 2.74394161880825e-05,
      "loss": 11.5109,
      "step": 17513
    },
    {
      "epoch": 3.0625573770491803,
      "grad_norm": 0.6918311712890826,
      "learning_rate": 2.7429672659901427e-05,
      "loss": 11.2628,
      "step": 17514
    },
    {
      "epoch": 3.0627322404371586,
      "grad_norm": 0.6220556889988251,
      "learning_rate": 2.7419930586978237e-05,
      "loss": 11.5871,
      "step": 17515
    },
    {
      "epoch": 3.0629071038251365,
      "grad_norm": 0.7102559328411749,
      "learning_rate": 2.7410189969508294e-05,
      "loss": 11.5563,
      "step": 17516
    },
    {
      "epoch": 3.063081967213115,
      "grad_norm": 0.617900094237443,
      "learning_rate": 2.7400450807686938e-05,
      "loss": 11.4817,
      "step": 17517
    },
    {
      "epoch": 3.0632568306010928,
      "grad_norm": 0.6500208874337822,
      "learning_rate": 2.739071310170941e-05,
      "loss": 11.4203,
      "step": 17518
    },
    {
      "epoch": 3.063431693989071,
      "grad_norm": 0.725144756963461,
      "learning_rate": 2.7380976851771045e-05,
      "loss": 11.0734,
      "step": 17519
    },
    {
      "epoch": 3.063606557377049,
      "grad_norm": 0.6305736072753233,
      "learning_rate": 2.737124205806707e-05,
      "loss": 11.4374,
      "step": 17520
    },
    {
      "epoch": 3.0637814207650274,
      "grad_norm": 0.7403128408873962,
      "learning_rate": 2.7361508720792682e-05,
      "loss": 11.4765,
      "step": 17521
    },
    {
      "epoch": 3.0639562841530052,
      "grad_norm": 0.6949054325212565,
      "learning_rate": 2.7351776840143106e-05,
      "loss": 11.5265,
      "step": 17522
    },
    {
      "epoch": 3.0641311475409836,
      "grad_norm": 0.6215980414678305,
      "learning_rate": 2.7342046416313406e-05,
      "loss": 11.6938,
      "step": 17523
    },
    {
      "epoch": 3.064306010928962,
      "grad_norm": 0.6698769031010618,
      "learning_rate": 2.7332317449498802e-05,
      "loss": 11.7015,
      "step": 17524
    },
    {
      "epoch": 3.06448087431694,
      "grad_norm": 0.6466680232056248,
      "learning_rate": 2.732258993989434e-05,
      "loss": 11.5923,
      "step": 17525
    },
    {
      "epoch": 3.064655737704918,
      "grad_norm": 0.6881921902170304,
      "learning_rate": 2.731286388769514e-05,
      "loss": 11.4113,
      "step": 17526
    },
    {
      "epoch": 3.064830601092896,
      "grad_norm": 0.6373868844283169,
      "learning_rate": 2.7303139293096157e-05,
      "loss": 11.2269,
      "step": 17527
    },
    {
      "epoch": 3.0650054644808744,
      "grad_norm": 0.6817032219491679,
      "learning_rate": 2.7293416156292418e-05,
      "loss": 11.5865,
      "step": 17528
    },
    {
      "epoch": 3.0651803278688523,
      "grad_norm": 0.6835649307683507,
      "learning_rate": 2.728369447747895e-05,
      "loss": 11.2212,
      "step": 17529
    },
    {
      "epoch": 3.0653551912568306,
      "grad_norm": 0.6954780761950828,
      "learning_rate": 2.7273974256850675e-05,
      "loss": 11.5128,
      "step": 17530
    },
    {
      "epoch": 3.0655300546448085,
      "grad_norm": 0.676815550200656,
      "learning_rate": 2.726425549460254e-05,
      "loss": 11.3656,
      "step": 17531
    },
    {
      "epoch": 3.065704918032787,
      "grad_norm": 0.6264001025909264,
      "learning_rate": 2.725453819092939e-05,
      "loss": 11.5403,
      "step": 17532
    },
    {
      "epoch": 3.0658797814207652,
      "grad_norm": 0.6503522242278931,
      "learning_rate": 2.7244822346026065e-05,
      "loss": 11.3832,
      "step": 17533
    },
    {
      "epoch": 3.066054644808743,
      "grad_norm": 0.6693324149375154,
      "learning_rate": 2.723510796008748e-05,
      "loss": 11.434,
      "step": 17534
    },
    {
      "epoch": 3.0662295081967215,
      "grad_norm": 0.6762233029981305,
      "learning_rate": 2.722539503330843e-05,
      "loss": 11.4761,
      "step": 17535
    },
    {
      "epoch": 3.0664043715846994,
      "grad_norm": 0.6220370270991701,
      "learning_rate": 2.721568356588362e-05,
      "loss": 11.3368,
      "step": 17536
    },
    {
      "epoch": 3.0665792349726777,
      "grad_norm": 0.6736860278241547,
      "learning_rate": 2.7205973558007802e-05,
      "loss": 11.1195,
      "step": 17537
    },
    {
      "epoch": 3.0667540983606556,
      "grad_norm": 0.6705112691407463,
      "learning_rate": 2.7196265009875754e-05,
      "loss": 11.2282,
      "step": 17538
    },
    {
      "epoch": 3.066928961748634,
      "grad_norm": 0.6922487881991978,
      "learning_rate": 2.718655792168213e-05,
      "loss": 11.341,
      "step": 17539
    },
    {
      "epoch": 3.067103825136612,
      "grad_norm": 0.7193974462178214,
      "learning_rate": 2.7176852293621612e-05,
      "loss": 11.5304,
      "step": 17540
    },
    {
      "epoch": 3.06727868852459,
      "grad_norm": 0.6826607387860276,
      "learning_rate": 2.7167148125888775e-05,
      "loss": 11.755,
      "step": 17541
    },
    {
      "epoch": 3.0674535519125685,
      "grad_norm": 0.6984922638144089,
      "learning_rate": 2.7157445418678216e-05,
      "loss": 11.3997,
      "step": 17542
    },
    {
      "epoch": 3.0676284153005464,
      "grad_norm": 0.7161265545195848,
      "learning_rate": 2.7147744172184564e-05,
      "loss": 11.5508,
      "step": 17543
    },
    {
      "epoch": 3.0678032786885248,
      "grad_norm": 0.7234461105068725,
      "learning_rate": 2.7138044386602358e-05,
      "loss": 11.5697,
      "step": 17544
    },
    {
      "epoch": 3.0679781420765027,
      "grad_norm": 0.7129310238924459,
      "learning_rate": 2.7128346062126043e-05,
      "loss": 11.3648,
      "step": 17545
    },
    {
      "epoch": 3.068153005464481,
      "grad_norm": 0.6570847449051258,
      "learning_rate": 2.711864919895013e-05,
      "loss": 11.6526,
      "step": 17546
    },
    {
      "epoch": 3.068327868852459,
      "grad_norm": 0.6856318185659895,
      "learning_rate": 2.7108953797269056e-05,
      "loss": 11.4529,
      "step": 17547
    },
    {
      "epoch": 3.0685027322404372,
      "grad_norm": 0.7533299359231573,
      "learning_rate": 2.709925985727728e-05,
      "loss": 11.3233,
      "step": 17548
    },
    {
      "epoch": 3.068677595628415,
      "grad_norm": 0.8333680243043937,
      "learning_rate": 2.7089567379169222e-05,
      "loss": 11.5784,
      "step": 17549
    },
    {
      "epoch": 3.0688524590163935,
      "grad_norm": 0.6461961208685523,
      "learning_rate": 2.7079876363139168e-05,
      "loss": 11.2242,
      "step": 17550
    },
    {
      "epoch": 3.0690273224043714,
      "grad_norm": 0.6457879486967132,
      "learning_rate": 2.707018680938148e-05,
      "loss": 11.5615,
      "step": 17551
    },
    {
      "epoch": 3.0692021857923497,
      "grad_norm": 0.6405889272495477,
      "learning_rate": 2.7060498718090454e-05,
      "loss": 11.616,
      "step": 17552
    },
    {
      "epoch": 3.069377049180328,
      "grad_norm": 0.6972767237869465,
      "learning_rate": 2.705081208946043e-05,
      "loss": 11.43,
      "step": 17553
    },
    {
      "epoch": 3.069551912568306,
      "grad_norm": 0.653852376977053,
      "learning_rate": 2.704112692368559e-05,
      "loss": 11.6102,
      "step": 17554
    },
    {
      "epoch": 3.0697267759562843,
      "grad_norm": 0.683092621638205,
      "learning_rate": 2.7031443220960163e-05,
      "loss": 11.5623,
      "step": 17555
    },
    {
      "epoch": 3.069901639344262,
      "grad_norm": 0.6079197718250015,
      "learning_rate": 2.702176098147834e-05,
      "loss": 11.3585,
      "step": 17556
    },
    {
      "epoch": 3.0700765027322405,
      "grad_norm": 0.6734791131173555,
      "learning_rate": 2.7012080205434242e-05,
      "loss": 11.2715,
      "step": 17557
    },
    {
      "epoch": 3.0702513661202184,
      "grad_norm": 0.6761463717520633,
      "learning_rate": 2.7002400893022117e-05,
      "loss": 11.4853,
      "step": 17558
    },
    {
      "epoch": 3.0704262295081968,
      "grad_norm": 0.6562924584159032,
      "learning_rate": 2.699272304443594e-05,
      "loss": 11.5705,
      "step": 17559
    },
    {
      "epoch": 3.0706010928961747,
      "grad_norm": 0.6846584397196246,
      "learning_rate": 2.6983046659869838e-05,
      "loss": 11.2628,
      "step": 17560
    },
    {
      "epoch": 3.070775956284153,
      "grad_norm": 0.7042680930167302,
      "learning_rate": 2.6973371739517804e-05,
      "loss": 11.6436,
      "step": 17561
    },
    {
      "epoch": 3.0709508196721313,
      "grad_norm": 0.6798048484523869,
      "learning_rate": 2.6963698283573958e-05,
      "loss": 11.3727,
      "step": 17562
    },
    {
      "epoch": 3.0711256830601092,
      "grad_norm": 0.7756747647057688,
      "learning_rate": 2.6954026292232182e-05,
      "loss": 11.2367,
      "step": 17563
    },
    {
      "epoch": 3.0713005464480876,
      "grad_norm": 0.633040003380015,
      "learning_rate": 2.6944355765686447e-05,
      "loss": 11.3167,
      "step": 17564
    },
    {
      "epoch": 3.0714754098360655,
      "grad_norm": 0.659277291942463,
      "learning_rate": 2.6934686704130696e-05,
      "loss": 11.5303,
      "step": 17565
    },
    {
      "epoch": 3.071650273224044,
      "grad_norm": 0.6700739693209256,
      "learning_rate": 2.6925019107758785e-05,
      "loss": 11.6125,
      "step": 17566
    },
    {
      "epoch": 3.0718251366120217,
      "grad_norm": 0.6675695474502777,
      "learning_rate": 2.6915352976764675e-05,
      "loss": 11.5208,
      "step": 17567
    },
    {
      "epoch": 3.072,
      "grad_norm": 0.6713118143422722,
      "learning_rate": 2.6905688311342104e-05,
      "loss": 11.4695,
      "step": 17568
    },
    {
      "epoch": 3.072174863387978,
      "grad_norm": 0.6155229284088688,
      "learning_rate": 2.689602511168492e-05,
      "loss": 11.2712,
      "step": 17569
    },
    {
      "epoch": 3.0723497267759563,
      "grad_norm": 0.6931348384909793,
      "learning_rate": 2.688636337798688e-05,
      "loss": 11.5644,
      "step": 17570
    },
    {
      "epoch": 3.072524590163934,
      "grad_norm": 0.6559306898349573,
      "learning_rate": 2.6876703110441747e-05,
      "loss": 11.5915,
      "step": 17571
    },
    {
      "epoch": 3.0726994535519125,
      "grad_norm": 0.6906409312480685,
      "learning_rate": 2.6867044309243237e-05,
      "loss": 11.3285,
      "step": 17572
    },
    {
      "epoch": 3.072874316939891,
      "grad_norm": 0.6765257816567116,
      "learning_rate": 2.685738697458503e-05,
      "loss": 11.5915,
      "step": 17573
    },
    {
      "epoch": 3.073049180327869,
      "grad_norm": 0.620983367048087,
      "learning_rate": 2.684773110666079e-05,
      "loss": 11.666,
      "step": 17574
    },
    {
      "epoch": 3.073224043715847,
      "grad_norm": 0.69399828586327,
      "learning_rate": 2.6838076705664162e-05,
      "loss": 11.4615,
      "step": 17575
    },
    {
      "epoch": 3.073398907103825,
      "grad_norm": 0.6696537136604138,
      "learning_rate": 2.6828423771788725e-05,
      "loss": 10.9261,
      "step": 17576
    },
    {
      "epoch": 3.0735737704918034,
      "grad_norm": 0.7350650548250047,
      "learning_rate": 2.6818772305228046e-05,
      "loss": 11.4086,
      "step": 17577
    },
    {
      "epoch": 3.0737486338797813,
      "grad_norm": 0.6632462756768457,
      "learning_rate": 2.6809122306175692e-05,
      "loss": 11.479,
      "step": 17578
    },
    {
      "epoch": 3.0739234972677596,
      "grad_norm": 0.6245226880100085,
      "learning_rate": 2.6799473774825156e-05,
      "loss": 11.4496,
      "step": 17579
    },
    {
      "epoch": 3.0740983606557375,
      "grad_norm": 0.5975000895099919,
      "learning_rate": 2.6789826711369924e-05,
      "loss": 11.4888,
      "step": 17580
    },
    {
      "epoch": 3.074273224043716,
      "grad_norm": 0.6124458812927911,
      "learning_rate": 2.6780181116003455e-05,
      "loss": 11.3222,
      "step": 17581
    },
    {
      "epoch": 3.074448087431694,
      "grad_norm": 0.6891904453917825,
      "learning_rate": 2.677053698891917e-05,
      "loss": 11.4099,
      "step": 17582
    },
    {
      "epoch": 3.074622950819672,
      "grad_norm": 0.657876363325627,
      "learning_rate": 2.676089433031046e-05,
      "loss": 11.3296,
      "step": 17583
    },
    {
      "epoch": 3.0747978142076504,
      "grad_norm": 0.6492635232807932,
      "learning_rate": 2.6751253140370704e-05,
      "loss": 11.2759,
      "step": 17584
    },
    {
      "epoch": 3.0749726775956283,
      "grad_norm": 0.7169898691422316,
      "learning_rate": 2.674161341929321e-05,
      "loss": 11.644,
      "step": 17585
    },
    {
      "epoch": 3.0751475409836067,
      "grad_norm": 0.7465326971067203,
      "learning_rate": 2.6731975167271307e-05,
      "loss": 11.3487,
      "step": 17586
    },
    {
      "epoch": 3.0753224043715846,
      "grad_norm": 0.6645402436316078,
      "learning_rate": 2.6722338384498257e-05,
      "loss": 11.6165,
      "step": 17587
    },
    {
      "epoch": 3.075497267759563,
      "grad_norm": 0.7414266725344828,
      "learning_rate": 2.6712703071167323e-05,
      "loss": 11.6233,
      "step": 17588
    },
    {
      "epoch": 3.075672131147541,
      "grad_norm": 0.6768664025965236,
      "learning_rate": 2.670306922747171e-05,
      "loss": 11.4454,
      "step": 17589
    },
    {
      "epoch": 3.075846994535519,
      "grad_norm": 0.6708519203990948,
      "learning_rate": 2.6693436853604615e-05,
      "loss": 11.5485,
      "step": 17590
    },
    {
      "epoch": 3.0760218579234975,
      "grad_norm": 0.6351474602563095,
      "learning_rate": 2.6683805949759176e-05,
      "loss": 11.2075,
      "step": 17591
    },
    {
      "epoch": 3.0761967213114754,
      "grad_norm": 0.7523782612368746,
      "learning_rate": 2.6674176516128556e-05,
      "loss": 11.5839,
      "step": 17592
    },
    {
      "epoch": 3.0763715846994537,
      "grad_norm": 0.6938552188542505,
      "learning_rate": 2.666454855290583e-05,
      "loss": 11.1526,
      "step": 17593
    },
    {
      "epoch": 3.0765464480874316,
      "grad_norm": 0.7366737305148257,
      "learning_rate": 2.665492206028407e-05,
      "loss": 11.4307,
      "step": 17594
    },
    {
      "epoch": 3.07672131147541,
      "grad_norm": 0.6477914211100306,
      "learning_rate": 2.6645297038456328e-05,
      "loss": 11.3855,
      "step": 17595
    },
    {
      "epoch": 3.076896174863388,
      "grad_norm": 0.6453941837307218,
      "learning_rate": 2.663567348761561e-05,
      "loss": 11.5699,
      "step": 17596
    },
    {
      "epoch": 3.077071038251366,
      "grad_norm": 0.6357064646196424,
      "learning_rate": 2.6626051407954898e-05,
      "loss": 11.1718,
      "step": 17597
    },
    {
      "epoch": 3.077245901639344,
      "grad_norm": 0.6529765690359238,
      "learning_rate": 2.6616430799667136e-05,
      "loss": 11.4765,
      "step": 17598
    },
    {
      "epoch": 3.0774207650273224,
      "grad_norm": 0.6590371379539891,
      "learning_rate": 2.660681166294525e-05,
      "loss": 11.4572,
      "step": 17599
    },
    {
      "epoch": 3.0775956284153008,
      "grad_norm": 0.690669247705963,
      "learning_rate": 2.659719399798215e-05,
      "loss": 11.4073,
      "step": 17600
    },
    {
      "epoch": 3.0777704918032787,
      "grad_norm": 0.6810760754687986,
      "learning_rate": 2.6587577804970677e-05,
      "loss": 11.5454,
      "step": 17601
    },
    {
      "epoch": 3.077945355191257,
      "grad_norm": 0.7146174793658822,
      "learning_rate": 2.6577963084103684e-05,
      "loss": 11.3519,
      "step": 17602
    },
    {
      "epoch": 3.078120218579235,
      "grad_norm": 0.6619318660003114,
      "learning_rate": 2.656834983557396e-05,
      "loss": 11.1531,
      "step": 17603
    },
    {
      "epoch": 3.0782950819672132,
      "grad_norm": 0.6995354441413765,
      "learning_rate": 2.655873805957432e-05,
      "loss": 11.3875,
      "step": 17604
    },
    {
      "epoch": 3.078469945355191,
      "grad_norm": 0.670733132577643,
      "learning_rate": 2.654912775629741e-05,
      "loss": 11.4325,
      "step": 17605
    },
    {
      "epoch": 3.0786448087431695,
      "grad_norm": 0.6620198566247459,
      "learning_rate": 2.6539518925936046e-05,
      "loss": 11.4684,
      "step": 17606
    },
    {
      "epoch": 3.0788196721311474,
      "grad_norm": 0.6234449482236697,
      "learning_rate": 2.6529911568682876e-05,
      "loss": 11.57,
      "step": 17607
    },
    {
      "epoch": 3.0789945355191257,
      "grad_norm": 0.6596257918238594,
      "learning_rate": 2.6520305684730563e-05,
      "loss": 11.6475,
      "step": 17608
    },
    {
      "epoch": 3.0791693989071036,
      "grad_norm": 0.6783992632999952,
      "learning_rate": 2.6510701274271745e-05,
      "loss": 11.5993,
      "step": 17609
    },
    {
      "epoch": 3.079344262295082,
      "grad_norm": 0.6300001935913905,
      "learning_rate": 2.6501098337498998e-05,
      "loss": 11.4622,
      "step": 17610
    },
    {
      "epoch": 3.0795191256830603,
      "grad_norm": 0.6778626989932139,
      "learning_rate": 2.6491496874604894e-05,
      "loss": 11.2936,
      "step": 17611
    },
    {
      "epoch": 3.079693989071038,
      "grad_norm": 0.760698269257253,
      "learning_rate": 2.6481896885781988e-05,
      "loss": 11.5324,
      "step": 17612
    },
    {
      "epoch": 3.0798688524590165,
      "grad_norm": 0.7237041050017462,
      "learning_rate": 2.6472298371222816e-05,
      "loss": 11.3288,
      "step": 17613
    },
    {
      "epoch": 3.0800437158469944,
      "grad_norm": 0.7267734951114517,
      "learning_rate": 2.6462701331119744e-05,
      "loss": 11.2533,
      "step": 17614
    },
    {
      "epoch": 3.0802185792349728,
      "grad_norm": 0.682290531700884,
      "learning_rate": 2.6453105765665342e-05,
      "loss": 11.3935,
      "step": 17615
    },
    {
      "epoch": 3.0803934426229507,
      "grad_norm": 0.6410858637826181,
      "learning_rate": 2.644351167505199e-05,
      "loss": 11.4823,
      "step": 17616
    },
    {
      "epoch": 3.080568306010929,
      "grad_norm": 0.6650871823019938,
      "learning_rate": 2.6433919059472067e-05,
      "loss": 11.1208,
      "step": 17617
    },
    {
      "epoch": 3.080743169398907,
      "grad_norm": 0.6755152820230225,
      "learning_rate": 2.6424327919117986e-05,
      "loss": 11.3213,
      "step": 17618
    },
    {
      "epoch": 3.0809180327868853,
      "grad_norm": 0.7187645115525684,
      "learning_rate": 2.641473825418197e-05,
      "loss": 11.2353,
      "step": 17619
    },
    {
      "epoch": 3.0810928961748636,
      "grad_norm": 0.7408470745287353,
      "learning_rate": 2.640515006485641e-05,
      "loss": 11.5289,
      "step": 17620
    },
    {
      "epoch": 3.0812677595628415,
      "grad_norm": 0.6509871154927782,
      "learning_rate": 2.6395563351333573e-05,
      "loss": 11.3795,
      "step": 17621
    },
    {
      "epoch": 3.08144262295082,
      "grad_norm": 0.7120589344223495,
      "learning_rate": 2.6385978113805708e-05,
      "loss": 11.5142,
      "step": 17622
    },
    {
      "epoch": 3.0816174863387977,
      "grad_norm": 0.7931508950001598,
      "learning_rate": 2.6376394352464972e-05,
      "loss": 11.3844,
      "step": 17623
    },
    {
      "epoch": 3.081792349726776,
      "grad_norm": 0.6364669795054808,
      "learning_rate": 2.6366812067503565e-05,
      "loss": 11.5959,
      "step": 17624
    },
    {
      "epoch": 3.081967213114754,
      "grad_norm": 0.6337188848654085,
      "learning_rate": 2.635723125911368e-05,
      "loss": 11.5973,
      "step": 17625
    },
    {
      "epoch": 3.0821420765027323,
      "grad_norm": 0.7755928729966978,
      "learning_rate": 2.6347651927487428e-05,
      "loss": 11.1021,
      "step": 17626
    },
    {
      "epoch": 3.08231693989071,
      "grad_norm": 0.7094841013118157,
      "learning_rate": 2.6338074072816922e-05,
      "loss": 11.1804,
      "step": 17627
    },
    {
      "epoch": 3.0824918032786885,
      "grad_norm": 0.6564937836918885,
      "learning_rate": 2.6328497695294173e-05,
      "loss": 11.5799,
      "step": 17628
    },
    {
      "epoch": 3.0826666666666664,
      "grad_norm": 0.6929135492151886,
      "learning_rate": 2.6318922795111223e-05,
      "loss": 11.5456,
      "step": 17629
    },
    {
      "epoch": 3.082841530054645,
      "grad_norm": 0.6497468213091026,
      "learning_rate": 2.6309349372460122e-05,
      "loss": 11.6168,
      "step": 17630
    },
    {
      "epoch": 3.083016393442623,
      "grad_norm": 0.6777763308091838,
      "learning_rate": 2.629977742753287e-05,
      "loss": 11.5257,
      "step": 17631
    },
    {
      "epoch": 3.083191256830601,
      "grad_norm": 0.7230911130450616,
      "learning_rate": 2.6290206960521323e-05,
      "loss": 11.5014,
      "step": 17632
    },
    {
      "epoch": 3.0833661202185794,
      "grad_norm": 0.6555107377269204,
      "learning_rate": 2.6280637971617428e-05,
      "loss": 11.4534,
      "step": 17633
    },
    {
      "epoch": 3.0835409836065573,
      "grad_norm": 0.6572665615269073,
      "learning_rate": 2.6271070461013116e-05,
      "loss": 11.536,
      "step": 17634
    },
    {
      "epoch": 3.0837158469945356,
      "grad_norm": 0.6464715075347709,
      "learning_rate": 2.626150442890022e-05,
      "loss": 11.0203,
      "step": 17635
    },
    {
      "epoch": 3.0838907103825135,
      "grad_norm": 0.6905682371520849,
      "learning_rate": 2.625193987547059e-05,
      "loss": 11.5506,
      "step": 17636
    },
    {
      "epoch": 3.084065573770492,
      "grad_norm": 0.6742945880992308,
      "learning_rate": 2.624237680091598e-05,
      "loss": 11.4969,
      "step": 17637
    },
    {
      "epoch": 3.0842404371584697,
      "grad_norm": 0.768680160584072,
      "learning_rate": 2.623281520542814e-05,
      "loss": 11.6126,
      "step": 17638
    },
    {
      "epoch": 3.084415300546448,
      "grad_norm": 0.6630820965560027,
      "learning_rate": 2.6223255089198895e-05,
      "loss": 11.4305,
      "step": 17639
    },
    {
      "epoch": 3.0845901639344264,
      "grad_norm": 0.6773246572088458,
      "learning_rate": 2.6213696452419933e-05,
      "loss": 11.3981,
      "step": 17640
    },
    {
      "epoch": 3.0847650273224043,
      "grad_norm": 0.6667186867390895,
      "learning_rate": 2.620413929528288e-05,
      "loss": 11.4962,
      "step": 17641
    },
    {
      "epoch": 3.0849398907103827,
      "grad_norm": 0.6648337757628288,
      "learning_rate": 2.6194583617979417e-05,
      "loss": 11.4584,
      "step": 17642
    },
    {
      "epoch": 3.0851147540983606,
      "grad_norm": 0.6486835242239579,
      "learning_rate": 2.6185029420701136e-05,
      "loss": 11.4537,
      "step": 17643
    },
    {
      "epoch": 3.085289617486339,
      "grad_norm": 0.6832281718890713,
      "learning_rate": 2.617547670363969e-05,
      "loss": 11.5716,
      "step": 17644
    },
    {
      "epoch": 3.085464480874317,
      "grad_norm": 0.6094309643604922,
      "learning_rate": 2.616592546698663e-05,
      "loss": 11.328,
      "step": 17645
    },
    {
      "epoch": 3.085639344262295,
      "grad_norm": 0.669902369997977,
      "learning_rate": 2.6156375710933436e-05,
      "loss": 11.2906,
      "step": 17646
    },
    {
      "epoch": 3.085814207650273,
      "grad_norm": 0.65265377543733,
      "learning_rate": 2.614682743567164e-05,
      "loss": 11.5138,
      "step": 17647
    },
    {
      "epoch": 3.0859890710382514,
      "grad_norm": 0.6675183691774362,
      "learning_rate": 2.613728064139268e-05,
      "loss": 11.5373,
      "step": 17648
    },
    {
      "epoch": 3.0861639344262297,
      "grad_norm": 0.7596835147113506,
      "learning_rate": 2.6127735328288106e-05,
      "loss": 11.3676,
      "step": 17649
    },
    {
      "epoch": 3.0863387978142076,
      "grad_norm": 0.6586658013144729,
      "learning_rate": 2.6118191496549205e-05,
      "loss": 11.2545,
      "step": 17650
    },
    {
      "epoch": 3.086513661202186,
      "grad_norm": 0.7014160076525671,
      "learning_rate": 2.6108649146367425e-05,
      "loss": 11.4544,
      "step": 17651
    },
    {
      "epoch": 3.086688524590164,
      "grad_norm": 0.6485464094340491,
      "learning_rate": 2.6099108277934103e-05,
      "loss": 11.3862,
      "step": 17652
    },
    {
      "epoch": 3.086863387978142,
      "grad_norm": 0.6671717507076209,
      "learning_rate": 2.608956889144054e-05,
      "loss": 11.206,
      "step": 17653
    },
    {
      "epoch": 3.08703825136612,
      "grad_norm": 0.6853916235584585,
      "learning_rate": 2.608003098707811e-05,
      "loss": 11.1995,
      "step": 17654
    },
    {
      "epoch": 3.0872131147540984,
      "grad_norm": 0.6403919127434282,
      "learning_rate": 2.6070494565037994e-05,
      "loss": 11.4907,
      "step": 17655
    },
    {
      "epoch": 3.0873879781420763,
      "grad_norm": 0.6624766318406469,
      "learning_rate": 2.606095962551145e-05,
      "loss": 11.4477,
      "step": 17656
    },
    {
      "epoch": 3.0875628415300547,
      "grad_norm": 0.7029785191824817,
      "learning_rate": 2.6051426168689665e-05,
      "loss": 11.4171,
      "step": 17657
    },
    {
      "epoch": 3.087737704918033,
      "grad_norm": 0.6176531751398472,
      "learning_rate": 2.6041894194763895e-05,
      "loss": 11.4612,
      "step": 17658
    },
    {
      "epoch": 3.087912568306011,
      "grad_norm": 0.6583014467436706,
      "learning_rate": 2.60323637039252e-05,
      "loss": 11.4093,
      "step": 17659
    },
    {
      "epoch": 3.0880874316939892,
      "grad_norm": 0.8283775214905842,
      "learning_rate": 2.6022834696364716e-05,
      "loss": 11.3298,
      "step": 17660
    },
    {
      "epoch": 3.088262295081967,
      "grad_norm": 0.6686439230561125,
      "learning_rate": 2.6013307172273548e-05,
      "loss": 11.5854,
      "step": 17661
    },
    {
      "epoch": 3.0884371584699455,
      "grad_norm": 0.7445915236943307,
      "learning_rate": 2.600378113184271e-05,
      "loss": 11.5457,
      "step": 17662
    },
    {
      "epoch": 3.0886120218579234,
      "grad_norm": 0.7114413137910922,
      "learning_rate": 2.599425657526331e-05,
      "loss": 11.7206,
      "step": 17663
    },
    {
      "epoch": 3.0887868852459017,
      "grad_norm": 0.7087864453819113,
      "learning_rate": 2.5984733502726267e-05,
      "loss": 11.1635,
      "step": 17664
    },
    {
      "epoch": 3.0889617486338796,
      "grad_norm": 0.6893054106425166,
      "learning_rate": 2.5975211914422572e-05,
      "loss": 11.4787,
      "step": 17665
    },
    {
      "epoch": 3.089136612021858,
      "grad_norm": 0.6725612045445679,
      "learning_rate": 2.5965691810543157e-05,
      "loss": 11.4578,
      "step": 17666
    },
    {
      "epoch": 3.089311475409836,
      "grad_norm": 0.7557585848005427,
      "learning_rate": 2.5956173191278942e-05,
      "loss": 11.4131,
      "step": 17667
    },
    {
      "epoch": 3.089486338797814,
      "grad_norm": 0.7096587973289236,
      "learning_rate": 2.5946656056820806e-05,
      "loss": 11.5807,
      "step": 17668
    },
    {
      "epoch": 3.0896612021857925,
      "grad_norm": 0.6213661561219151,
      "learning_rate": 2.5937140407359583e-05,
      "loss": 10.8646,
      "step": 17669
    },
    {
      "epoch": 3.0898360655737704,
      "grad_norm": 0.8087200627477938,
      "learning_rate": 2.59276262430861e-05,
      "loss": 11.1061,
      "step": 17670
    },
    {
      "epoch": 3.090010928961749,
      "grad_norm": 0.6531782609069479,
      "learning_rate": 2.5918113564191137e-05,
      "loss": 11.4948,
      "step": 17671
    },
    {
      "epoch": 3.0901857923497267,
      "grad_norm": 0.6391450934403712,
      "learning_rate": 2.5908602370865465e-05,
      "loss": 11.6007,
      "step": 17672
    },
    {
      "epoch": 3.090360655737705,
      "grad_norm": 0.6826924558254599,
      "learning_rate": 2.5899092663299807e-05,
      "loss": 11.5787,
      "step": 17673
    },
    {
      "epoch": 3.090535519125683,
      "grad_norm": 0.657691748296156,
      "learning_rate": 2.5889584441684856e-05,
      "loss": 11.2666,
      "step": 17674
    },
    {
      "epoch": 3.0907103825136613,
      "grad_norm": 0.6880179586468922,
      "learning_rate": 2.588007770621129e-05,
      "loss": 11.5611,
      "step": 17675
    },
    {
      "epoch": 3.090885245901639,
      "grad_norm": 0.6550657423552514,
      "learning_rate": 2.5870572457069742e-05,
      "loss": 11.8269,
      "step": 17676
    },
    {
      "epoch": 3.0910601092896175,
      "grad_norm": 0.6752603199432989,
      "learning_rate": 2.586106869445082e-05,
      "loss": 11.6289,
      "step": 17677
    },
    {
      "epoch": 3.091234972677596,
      "grad_norm": 0.7032347408869959,
      "learning_rate": 2.5851566418545114e-05,
      "loss": 11.3885,
      "step": 17678
    },
    {
      "epoch": 3.0914098360655737,
      "grad_norm": 0.7449952239225253,
      "learning_rate": 2.5842065629543166e-05,
      "loss": 11.4499,
      "step": 17679
    },
    {
      "epoch": 3.091584699453552,
      "grad_norm": 0.6354713037133931,
      "learning_rate": 2.583256632763549e-05,
      "loss": 11.6446,
      "step": 17680
    },
    {
      "epoch": 3.09175956284153,
      "grad_norm": 0.6460533143731362,
      "learning_rate": 2.5823068513012595e-05,
      "loss": 11.3951,
      "step": 17681
    },
    {
      "epoch": 3.0919344262295083,
      "grad_norm": 0.6605437909099312,
      "learning_rate": 2.581357218586492e-05,
      "loss": 11.7069,
      "step": 17682
    },
    {
      "epoch": 3.092109289617486,
      "grad_norm": 0.7347312944228158,
      "learning_rate": 2.5804077346382905e-05,
      "loss": 11.3207,
      "step": 17683
    },
    {
      "epoch": 3.0922841530054646,
      "grad_norm": 0.7257398946037107,
      "learning_rate": 2.579458399475695e-05,
      "loss": 11.47,
      "step": 17684
    },
    {
      "epoch": 3.0924590163934425,
      "grad_norm": 0.6554233756284487,
      "learning_rate": 2.5785092131177424e-05,
      "loss": 11.3579,
      "step": 17685
    },
    {
      "epoch": 3.092633879781421,
      "grad_norm": 0.6904539061304034,
      "learning_rate": 2.577560175583468e-05,
      "loss": 11.4705,
      "step": 17686
    },
    {
      "epoch": 3.0928087431693987,
      "grad_norm": 0.721806542938868,
      "learning_rate": 2.576611286891901e-05,
      "loss": 11.5384,
      "step": 17687
    },
    {
      "epoch": 3.092983606557377,
      "grad_norm": 0.7411918809598131,
      "learning_rate": 2.575662547062071e-05,
      "loss": 11.4514,
      "step": 17688
    },
    {
      "epoch": 3.0931584699453554,
      "grad_norm": 0.6730547280365721,
      "learning_rate": 2.574713956113003e-05,
      "loss": 11.7237,
      "step": 17689
    },
    {
      "epoch": 3.0933333333333333,
      "grad_norm": 0.7074144546145512,
      "learning_rate": 2.5737655140637186e-05,
      "loss": 11.4868,
      "step": 17690
    },
    {
      "epoch": 3.0935081967213116,
      "grad_norm": 0.668131511991168,
      "learning_rate": 2.5728172209332367e-05,
      "loss": 11.5331,
      "step": 17691
    },
    {
      "epoch": 3.0936830601092895,
      "grad_norm": 0.7811385777342142,
      "learning_rate": 2.571869076740574e-05,
      "loss": 11.2113,
      "step": 17692
    },
    {
      "epoch": 3.093857923497268,
      "grad_norm": 0.736713244050532,
      "learning_rate": 2.5709210815047447e-05,
      "loss": 11.6394,
      "step": 17693
    },
    {
      "epoch": 3.0940327868852457,
      "grad_norm": 0.7168844889687129,
      "learning_rate": 2.5699732352447568e-05,
      "loss": 11.2572,
      "step": 17694
    },
    {
      "epoch": 3.094207650273224,
      "grad_norm": 0.6959640552905947,
      "learning_rate": 2.569025537979619e-05,
      "loss": 11.4943,
      "step": 17695
    },
    {
      "epoch": 3.094382513661202,
      "grad_norm": 0.6509533422644824,
      "learning_rate": 2.568077989728336e-05,
      "loss": 11.4966,
      "step": 17696
    },
    {
      "epoch": 3.0945573770491803,
      "grad_norm": 0.7387327810352815,
      "learning_rate": 2.5671305905099075e-05,
      "loss": 11.396,
      "step": 17697
    },
    {
      "epoch": 3.0947322404371587,
      "grad_norm": 0.6837756027654529,
      "learning_rate": 2.566183340343332e-05,
      "loss": 11.1981,
      "step": 17698
    },
    {
      "epoch": 3.0949071038251366,
      "grad_norm": 0.7706121669097543,
      "learning_rate": 2.565236239247607e-05,
      "loss": 11.4242,
      "step": 17699
    },
    {
      "epoch": 3.095081967213115,
      "grad_norm": 0.627602875852065,
      "learning_rate": 2.5642892872417247e-05,
      "loss": 11.6552,
      "step": 17700
    },
    {
      "epoch": 3.095256830601093,
      "grad_norm": 0.6560770674194705,
      "learning_rate": 2.5633424843446674e-05,
      "loss": 11.4968,
      "step": 17701
    },
    {
      "epoch": 3.095431693989071,
      "grad_norm": 0.7125070745287648,
      "learning_rate": 2.5623958305754292e-05,
      "loss": 11.2289,
      "step": 17702
    },
    {
      "epoch": 3.095606557377049,
      "grad_norm": 0.7958257755008223,
      "learning_rate": 2.5614493259529927e-05,
      "loss": 11.5807,
      "step": 17703
    },
    {
      "epoch": 3.0957814207650274,
      "grad_norm": 0.7894094398376692,
      "learning_rate": 2.5605029704963346e-05,
      "loss": 11.5844,
      "step": 17704
    },
    {
      "epoch": 3.0959562841530053,
      "grad_norm": 0.8514615234488089,
      "learning_rate": 2.5595567642244354e-05,
      "loss": 11.5257,
      "step": 17705
    },
    {
      "epoch": 3.0961311475409836,
      "grad_norm": 0.6971802616064691,
      "learning_rate": 2.558610707156268e-05,
      "loss": 11.2541,
      "step": 17706
    },
    {
      "epoch": 3.096306010928962,
      "grad_norm": 0.7118031421554616,
      "learning_rate": 2.5576647993108038e-05,
      "loss": 10.9028,
      "step": 17707
    },
    {
      "epoch": 3.09648087431694,
      "grad_norm": 0.735669341148519,
      "learning_rate": 2.556719040707012e-05,
      "loss": 11.4313,
      "step": 17708
    },
    {
      "epoch": 3.096655737704918,
      "grad_norm": 0.6619229495484744,
      "learning_rate": 2.555773431363859e-05,
      "loss": 11.2998,
      "step": 17709
    },
    {
      "epoch": 3.096830601092896,
      "grad_norm": 0.6854665062911172,
      "learning_rate": 2.5548279713002997e-05,
      "loss": 11.5985,
      "step": 17710
    },
    {
      "epoch": 3.0970054644808744,
      "grad_norm": 0.7175808965955636,
      "learning_rate": 2.553882660535304e-05,
      "loss": 11.4011,
      "step": 17711
    },
    {
      "epoch": 3.0971803278688523,
      "grad_norm": 0.5971114555479986,
      "learning_rate": 2.552937499087822e-05,
      "loss": 11.3281,
      "step": 17712
    },
    {
      "epoch": 3.0973551912568307,
      "grad_norm": 0.6639033365906486,
      "learning_rate": 2.5519924869768086e-05,
      "loss": 11.3921,
      "step": 17713
    },
    {
      "epoch": 3.0975300546448086,
      "grad_norm": 0.6830522838015932,
      "learning_rate": 2.551047624221218e-05,
      "loss": 11.5118,
      "step": 17714
    },
    {
      "epoch": 3.097704918032787,
      "grad_norm": 0.6507225018294692,
      "learning_rate": 2.5501029108399866e-05,
      "loss": 11.6033,
      "step": 17715
    },
    {
      "epoch": 3.0978797814207653,
      "grad_norm": 0.7106207819606536,
      "learning_rate": 2.5491583468520708e-05,
      "loss": 11.3816,
      "step": 17716
    },
    {
      "epoch": 3.098054644808743,
      "grad_norm": 0.6343845465263755,
      "learning_rate": 2.5482139322764054e-05,
      "loss": 11.3577,
      "step": 17717
    },
    {
      "epoch": 3.0982295081967215,
      "grad_norm": 0.6545146121830873,
      "learning_rate": 2.547269667131935e-05,
      "loss": 11.6212,
      "step": 17718
    },
    {
      "epoch": 3.0984043715846994,
      "grad_norm": 0.7678551071390466,
      "learning_rate": 2.5463255514375873e-05,
      "loss": 11.333,
      "step": 17719
    },
    {
      "epoch": 3.0985792349726777,
      "grad_norm": 0.7070599392831847,
      "learning_rate": 2.5453815852122942e-05,
      "loss": 11.4898,
      "step": 17720
    },
    {
      "epoch": 3.0987540983606556,
      "grad_norm": 0.7360810840354101,
      "learning_rate": 2.5444377684749932e-05,
      "loss": 11.2889,
      "step": 17721
    },
    {
      "epoch": 3.098928961748634,
      "grad_norm": 0.603890597977591,
      "learning_rate": 2.5434941012446056e-05,
      "loss": 11.6353,
      "step": 17722
    },
    {
      "epoch": 3.099103825136612,
      "grad_norm": 0.6617554899458721,
      "learning_rate": 2.54255058354006e-05,
      "loss": 11.197,
      "step": 17723
    },
    {
      "epoch": 3.09927868852459,
      "grad_norm": 0.6369120382612925,
      "learning_rate": 2.5416072153802683e-05,
      "loss": 11.4374,
      "step": 17724
    },
    {
      "epoch": 3.099453551912568,
      "grad_norm": 0.6776545759108101,
      "learning_rate": 2.5406639967841494e-05,
      "loss": 11.3776,
      "step": 17725
    },
    {
      "epoch": 3.0996284153005464,
      "grad_norm": 0.7816211415402772,
      "learning_rate": 2.5397209277706246e-05,
      "loss": 11.2207,
      "step": 17726
    },
    {
      "epoch": 3.099803278688525,
      "grad_norm": 0.7103787455194353,
      "learning_rate": 2.538778008358603e-05,
      "loss": 11.4946,
      "step": 17727
    },
    {
      "epoch": 3.0999781420765027,
      "grad_norm": 0.6518664448353361,
      "learning_rate": 2.537835238566989e-05,
      "loss": 11.3842,
      "step": 17728
    },
    {
      "epoch": 3.100153005464481,
      "grad_norm": 0.7072789833631121,
      "learning_rate": 2.536892618414687e-05,
      "loss": 11.6418,
      "step": 17729
    },
    {
      "epoch": 3.100327868852459,
      "grad_norm": 0.6680974295179235,
      "learning_rate": 2.535950147920606e-05,
      "loss": 11.5603,
      "step": 17730
    },
    {
      "epoch": 3.1005027322404373,
      "grad_norm": 0.6816207400530236,
      "learning_rate": 2.5350078271036414e-05,
      "loss": 11.5553,
      "step": 17731
    },
    {
      "epoch": 3.100677595628415,
      "grad_norm": 0.6480071156832831,
      "learning_rate": 2.5340656559826947e-05,
      "loss": 11.4577,
      "step": 17732
    },
    {
      "epoch": 3.1008524590163935,
      "grad_norm": 0.7874681589938544,
      "learning_rate": 2.5331236345766517e-05,
      "loss": 11.4167,
      "step": 17733
    },
    {
      "epoch": 3.1010273224043714,
      "grad_norm": 0.7374713560558684,
      "learning_rate": 2.532181762904402e-05,
      "loss": 11.3694,
      "step": 17734
    },
    {
      "epoch": 3.1012021857923497,
      "grad_norm": 0.7192942409260564,
      "learning_rate": 2.5312400409848413e-05,
      "loss": 11.3966,
      "step": 17735
    },
    {
      "epoch": 3.101377049180328,
      "grad_norm": 0.7066083147760547,
      "learning_rate": 2.5302984688368526e-05,
      "loss": 11.5685,
      "step": 17736
    },
    {
      "epoch": 3.101551912568306,
      "grad_norm": 0.6866728251143424,
      "learning_rate": 2.5293570464793113e-05,
      "loss": 11.3535,
      "step": 17737
    },
    {
      "epoch": 3.1017267759562843,
      "grad_norm": 0.6426150317735863,
      "learning_rate": 2.5284157739310997e-05,
      "loss": 11.2827,
      "step": 17738
    },
    {
      "epoch": 3.101901639344262,
      "grad_norm": 0.6846701391346468,
      "learning_rate": 2.527474651211089e-05,
      "loss": 11.35,
      "step": 17739
    },
    {
      "epoch": 3.1020765027322406,
      "grad_norm": 0.6590678383205423,
      "learning_rate": 2.5265336783381588e-05,
      "loss": 11.5005,
      "step": 17740
    },
    {
      "epoch": 3.1022513661202185,
      "grad_norm": 0.6665200684709799,
      "learning_rate": 2.525592855331177e-05,
      "loss": 11.4972,
      "step": 17741
    },
    {
      "epoch": 3.102426229508197,
      "grad_norm": 0.615805104731051,
      "learning_rate": 2.5246521822090064e-05,
      "loss": 11.6044,
      "step": 17742
    },
    {
      "epoch": 3.1026010928961747,
      "grad_norm": 0.6093810323082867,
      "learning_rate": 2.5237116589905108e-05,
      "loss": 11.4029,
      "step": 17743
    },
    {
      "epoch": 3.102775956284153,
      "grad_norm": 0.6942689999085137,
      "learning_rate": 2.5227712856945484e-05,
      "loss": 11.6574,
      "step": 17744
    },
    {
      "epoch": 3.102950819672131,
      "grad_norm": 0.6658135980424115,
      "learning_rate": 2.5218310623399867e-05,
      "loss": 11.5209,
      "step": 17745
    },
    {
      "epoch": 3.1031256830601093,
      "grad_norm": 0.7100101575256239,
      "learning_rate": 2.52089098894567e-05,
      "loss": 11.4853,
      "step": 17746
    },
    {
      "epoch": 3.1033005464480876,
      "grad_norm": 0.6936714379583291,
      "learning_rate": 2.5199510655304525e-05,
      "loss": 11.2946,
      "step": 17747
    },
    {
      "epoch": 3.1034754098360655,
      "grad_norm": 0.6039674541313171,
      "learning_rate": 2.5190112921131824e-05,
      "loss": 11.3556,
      "step": 17748
    },
    {
      "epoch": 3.103650273224044,
      "grad_norm": 0.6020833803515504,
      "learning_rate": 2.5180716687127037e-05,
      "loss": 11.5038,
      "step": 17749
    },
    {
      "epoch": 3.1038251366120218,
      "grad_norm": 0.6164045054256334,
      "learning_rate": 2.5171321953478654e-05,
      "loss": 11.4411,
      "step": 17750
    },
    {
      "epoch": 3.104,
      "grad_norm": 0.6189017724333525,
      "learning_rate": 2.5161928720374993e-05,
      "loss": 11.4762,
      "step": 17751
    },
    {
      "epoch": 3.104174863387978,
      "grad_norm": 0.6595569150298002,
      "learning_rate": 2.515253698800445e-05,
      "loss": 11.4298,
      "step": 17752
    },
    {
      "epoch": 3.1043497267759563,
      "grad_norm": 0.6384458956071338,
      "learning_rate": 2.5143146756555326e-05,
      "loss": 11.4919,
      "step": 17753
    },
    {
      "epoch": 3.1045245901639342,
      "grad_norm": 0.6881226646142349,
      "learning_rate": 2.5133758026216003e-05,
      "loss": 11.4975,
      "step": 17754
    },
    {
      "epoch": 3.1046994535519126,
      "grad_norm": 0.6532799850215478,
      "learning_rate": 2.5124370797174668e-05,
      "loss": 11.6911,
      "step": 17755
    },
    {
      "epoch": 3.104874316939891,
      "grad_norm": 0.6053005442175524,
      "learning_rate": 2.51149850696196e-05,
      "loss": 11.6225,
      "step": 17756
    },
    {
      "epoch": 3.105049180327869,
      "grad_norm": 0.6644945686642303,
      "learning_rate": 2.5105600843739008e-05,
      "loss": 11.5266,
      "step": 17757
    },
    {
      "epoch": 3.105224043715847,
      "grad_norm": 0.7240589155546312,
      "learning_rate": 2.5096218119721048e-05,
      "loss": 11.5825,
      "step": 17758
    },
    {
      "epoch": 3.105398907103825,
      "grad_norm": 0.6707882603213974,
      "learning_rate": 2.508683689775396e-05,
      "loss": 11.714,
      "step": 17759
    },
    {
      "epoch": 3.1055737704918034,
      "grad_norm": 0.6651593880271957,
      "learning_rate": 2.5077457178025777e-05,
      "loss": 11.4541,
      "step": 17760
    },
    {
      "epoch": 3.1057486338797813,
      "grad_norm": 0.6514715256810399,
      "learning_rate": 2.5068078960724618e-05,
      "loss": 11.5113,
      "step": 17761
    },
    {
      "epoch": 3.1059234972677596,
      "grad_norm": 0.7040280510940551,
      "learning_rate": 2.5058702246038544e-05,
      "loss": 11.6793,
      "step": 17762
    },
    {
      "epoch": 3.1060983606557375,
      "grad_norm": 0.7479784455269234,
      "learning_rate": 2.504932703415559e-05,
      "loss": 11.554,
      "step": 17763
    },
    {
      "epoch": 3.106273224043716,
      "grad_norm": 0.6946779798690105,
      "learning_rate": 2.5039953325263764e-05,
      "loss": 11.3599,
      "step": 17764
    },
    {
      "epoch": 3.106448087431694,
      "grad_norm": 0.6522678252669271,
      "learning_rate": 2.503058111955102e-05,
      "loss": 11.1812,
      "step": 17765
    },
    {
      "epoch": 3.106622950819672,
      "grad_norm": 0.684386784240768,
      "learning_rate": 2.502121041720532e-05,
      "loss": 11.4469,
      "step": 17766
    },
    {
      "epoch": 3.1067978142076504,
      "grad_norm": 0.6332317333742525,
      "learning_rate": 2.5011841218414566e-05,
      "loss": 11.6252,
      "step": 17767
    },
    {
      "epoch": 3.1069726775956283,
      "grad_norm": 0.7320481433173093,
      "learning_rate": 2.500247352336664e-05,
      "loss": 11.5033,
      "step": 17768
    },
    {
      "epoch": 3.1071475409836067,
      "grad_norm": 0.6797468418406737,
      "learning_rate": 2.4993107332249387e-05,
      "loss": 11.5052,
      "step": 17769
    },
    {
      "epoch": 3.1073224043715846,
      "grad_norm": 0.6787007383899657,
      "learning_rate": 2.4983742645250642e-05,
      "loss": 11.5197,
      "step": 17770
    },
    {
      "epoch": 3.107497267759563,
      "grad_norm": 0.669076588994991,
      "learning_rate": 2.4974379462558182e-05,
      "loss": 11.5297,
      "step": 17771
    },
    {
      "epoch": 3.107672131147541,
      "grad_norm": 0.6505893427512448,
      "learning_rate": 2.496501778435977e-05,
      "loss": 11.2888,
      "step": 17772
    },
    {
      "epoch": 3.107846994535519,
      "grad_norm": 0.6832469755786699,
      "learning_rate": 2.495565761084314e-05,
      "loss": 11.5424,
      "step": 17773
    },
    {
      "epoch": 3.108021857923497,
      "grad_norm": 0.6575817004197724,
      "learning_rate": 2.494629894219599e-05,
      "loss": 11.5776,
      "step": 17774
    },
    {
      "epoch": 3.1081967213114754,
      "grad_norm": 0.6489647160383701,
      "learning_rate": 2.4936941778605994e-05,
      "loss": 11.3781,
      "step": 17775
    },
    {
      "epoch": 3.1083715846994537,
      "grad_norm": 0.6929423604680267,
      "learning_rate": 2.4927586120260793e-05,
      "loss": 11.3617,
      "step": 17776
    },
    {
      "epoch": 3.1085464480874316,
      "grad_norm": 0.6561367238750487,
      "learning_rate": 2.4918231967347993e-05,
      "loss": 11.4272,
      "step": 17777
    },
    {
      "epoch": 3.10872131147541,
      "grad_norm": 0.660216784871195,
      "learning_rate": 2.4908879320055167e-05,
      "loss": 11.1394,
      "step": 17778
    },
    {
      "epoch": 3.108896174863388,
      "grad_norm": 0.7164908585762308,
      "learning_rate": 2.4899528178569874e-05,
      "loss": 11.4992,
      "step": 17779
    },
    {
      "epoch": 3.109071038251366,
      "grad_norm": 0.6433082387954189,
      "learning_rate": 2.4890178543079624e-05,
      "loss": 11.3706,
      "step": 17780
    },
    {
      "epoch": 3.109245901639344,
      "grad_norm": 0.6908882954253645,
      "learning_rate": 2.4880830413771915e-05,
      "loss": 11.4897,
      "step": 17781
    },
    {
      "epoch": 3.1094207650273225,
      "grad_norm": 0.6523695567760622,
      "learning_rate": 2.4871483790834216e-05,
      "loss": 11.6065,
      "step": 17782
    },
    {
      "epoch": 3.1095956284153004,
      "grad_norm": 0.6325212979640329,
      "learning_rate": 2.486213867445393e-05,
      "loss": 11.4606,
      "step": 17783
    },
    {
      "epoch": 3.1097704918032787,
      "grad_norm": 0.6507996694138763,
      "learning_rate": 2.4852795064818458e-05,
      "loss": 11.4667,
      "step": 17784
    },
    {
      "epoch": 3.109945355191257,
      "grad_norm": 0.6223338368758017,
      "learning_rate": 2.484345296211519e-05,
      "loss": 11.4023,
      "step": 17785
    },
    {
      "epoch": 3.110120218579235,
      "grad_norm": 0.6561236542847522,
      "learning_rate": 2.4834112366531447e-05,
      "loss": 11.5488,
      "step": 17786
    },
    {
      "epoch": 3.1102950819672133,
      "grad_norm": 0.641780488227816,
      "learning_rate": 2.4824773278254544e-05,
      "loss": 11.5301,
      "step": 17787
    },
    {
      "epoch": 3.110469945355191,
      "grad_norm": 0.6655028530654629,
      "learning_rate": 2.4815435697471746e-05,
      "loss": 11.3907,
      "step": 17788
    },
    {
      "epoch": 3.1106448087431695,
      "grad_norm": 0.6787856526818996,
      "learning_rate": 2.4806099624370317e-05,
      "loss": 11.4037,
      "step": 17789
    },
    {
      "epoch": 3.1108196721311474,
      "grad_norm": 0.6841395409208503,
      "learning_rate": 2.4796765059137462e-05,
      "loss": 11.2596,
      "step": 17790
    },
    {
      "epoch": 3.1109945355191257,
      "grad_norm": 0.652247404759437,
      "learning_rate": 2.478743200196041e-05,
      "loss": 11.3096,
      "step": 17791
    },
    {
      "epoch": 3.1111693989071036,
      "grad_norm": 0.6619972036770084,
      "learning_rate": 2.477810045302621e-05,
      "loss": 11.5534,
      "step": 17792
    },
    {
      "epoch": 3.111344262295082,
      "grad_norm": 0.6897046467618451,
      "learning_rate": 2.476877041252209e-05,
      "loss": 11.4045,
      "step": 17793
    },
    {
      "epoch": 3.11151912568306,
      "grad_norm": 0.6434636449496023,
      "learning_rate": 2.4759441880635115e-05,
      "loss": 11.0098,
      "step": 17794
    },
    {
      "epoch": 3.1116939890710382,
      "grad_norm": 0.657735178499973,
      "learning_rate": 2.4750114857552355e-05,
      "loss": 11.5649,
      "step": 17795
    },
    {
      "epoch": 3.1118688524590166,
      "grad_norm": 0.6613034196116991,
      "learning_rate": 2.4740789343460857e-05,
      "loss": 11.3223,
      "step": 17796
    },
    {
      "epoch": 3.1120437158469945,
      "grad_norm": 0.7300777936555097,
      "learning_rate": 2.4731465338547556e-05,
      "loss": 11.5924,
      "step": 17797
    },
    {
      "epoch": 3.112218579234973,
      "grad_norm": 0.7004841540818783,
      "learning_rate": 2.472214284299951e-05,
      "loss": 11.4408,
      "step": 17798
    },
    {
      "epoch": 3.1123934426229507,
      "grad_norm": 0.6095410930016262,
      "learning_rate": 2.4712821857003632e-05,
      "loss": 11.5097,
      "step": 17799
    },
    {
      "epoch": 3.112568306010929,
      "grad_norm": 0.6930412140148309,
      "learning_rate": 2.4703502380746867e-05,
      "loss": 11.4625,
      "step": 17800
    },
    {
      "epoch": 3.112743169398907,
      "grad_norm": 0.703273202789803,
      "learning_rate": 2.469418441441601e-05,
      "loss": 11.3263,
      "step": 17801
    },
    {
      "epoch": 3.1129180327868853,
      "grad_norm": 0.6708230184581019,
      "learning_rate": 2.4684867958198e-05,
      "loss": 11.3255,
      "step": 17802
    },
    {
      "epoch": 3.113092896174863,
      "grad_norm": 0.7015588581111187,
      "learning_rate": 2.467555301227964e-05,
      "loss": 11.3624,
      "step": 17803
    },
    {
      "epoch": 3.1132677595628415,
      "grad_norm": 0.5904437222348037,
      "learning_rate": 2.4666239576847727e-05,
      "loss": 11.4976,
      "step": 17804
    },
    {
      "epoch": 3.11344262295082,
      "grad_norm": 0.6610459111103978,
      "learning_rate": 2.4656927652089034e-05,
      "loss": 11.5086,
      "step": 17805
    },
    {
      "epoch": 3.1136174863387978,
      "grad_norm": 0.8454158282618192,
      "learning_rate": 2.4647617238190223e-05,
      "loss": 11.4903,
      "step": 17806
    },
    {
      "epoch": 3.113792349726776,
      "grad_norm": 0.7034897876034787,
      "learning_rate": 2.4638308335338066e-05,
      "loss": 11.2881,
      "step": 17807
    },
    {
      "epoch": 3.113967213114754,
      "grad_norm": 0.6109966468960091,
      "learning_rate": 2.4629000943719228e-05,
      "loss": 11.3656,
      "step": 17808
    },
    {
      "epoch": 3.1141420765027323,
      "grad_norm": 0.6562643078005519,
      "learning_rate": 2.4619695063520367e-05,
      "loss": 11.5048,
      "step": 17809
    },
    {
      "epoch": 3.1143169398907102,
      "grad_norm": 0.6383239745536157,
      "learning_rate": 2.4610390694928043e-05,
      "loss": 11.4004,
      "step": 17810
    },
    {
      "epoch": 3.1144918032786886,
      "grad_norm": 0.6477940250864438,
      "learning_rate": 2.4601087838128824e-05,
      "loss": 11.5201,
      "step": 17811
    },
    {
      "epoch": 3.1146666666666665,
      "grad_norm": 0.6824766064163362,
      "learning_rate": 2.459178649330933e-05,
      "loss": 11.2058,
      "step": 17812
    },
    {
      "epoch": 3.114841530054645,
      "grad_norm": 0.6268783765116611,
      "learning_rate": 2.4582486660656055e-05,
      "loss": 11.368,
      "step": 17813
    },
    {
      "epoch": 3.115016393442623,
      "grad_norm": 0.6912770552157561,
      "learning_rate": 2.457318834035551e-05,
      "loss": 11.4302,
      "step": 17814
    },
    {
      "epoch": 3.115191256830601,
      "grad_norm": 0.7237895933052179,
      "learning_rate": 2.4563891532594106e-05,
      "loss": 11.4535,
      "step": 17815
    },
    {
      "epoch": 3.1153661202185794,
      "grad_norm": 0.74194699814845,
      "learning_rate": 2.455459623755827e-05,
      "loss": 11.263,
      "step": 17816
    },
    {
      "epoch": 3.1155409836065573,
      "grad_norm": 0.7754685696884475,
      "learning_rate": 2.454530245543445e-05,
      "loss": 11.4927,
      "step": 17817
    },
    {
      "epoch": 3.1157158469945356,
      "grad_norm": 0.7382301318475445,
      "learning_rate": 2.453601018640902e-05,
      "loss": 11.4719,
      "step": 17818
    },
    {
      "epoch": 3.1158907103825135,
      "grad_norm": 0.7291709971573453,
      "learning_rate": 2.4526719430668266e-05,
      "loss": 11.4388,
      "step": 17819
    },
    {
      "epoch": 3.116065573770492,
      "grad_norm": 0.7805387807106245,
      "learning_rate": 2.4517430188398517e-05,
      "loss": 11.4669,
      "step": 17820
    },
    {
      "epoch": 3.1162404371584698,
      "grad_norm": 0.6649965795777633,
      "learning_rate": 2.450814245978602e-05,
      "loss": 11.3843,
      "step": 17821
    },
    {
      "epoch": 3.116415300546448,
      "grad_norm": 0.6109007676412962,
      "learning_rate": 2.44988562450171e-05,
      "loss": 11.3121,
      "step": 17822
    },
    {
      "epoch": 3.1165901639344264,
      "grad_norm": 0.7259778052953714,
      "learning_rate": 2.4489571544277945e-05,
      "loss": 11.5813,
      "step": 17823
    },
    {
      "epoch": 3.1167650273224043,
      "grad_norm": 0.6566138855923176,
      "learning_rate": 2.448028835775471e-05,
      "loss": 11.4735,
      "step": 17824
    },
    {
      "epoch": 3.1169398907103827,
      "grad_norm": 0.6994720829238836,
      "learning_rate": 2.4471006685633536e-05,
      "loss": 11.3881,
      "step": 17825
    },
    {
      "epoch": 3.1171147540983606,
      "grad_norm": 0.6246700996108956,
      "learning_rate": 2.4461726528100615e-05,
      "loss": 11.2822,
      "step": 17826
    },
    {
      "epoch": 3.117289617486339,
      "grad_norm": 0.6474087403273289,
      "learning_rate": 2.4452447885342033e-05,
      "loss": 11.2579,
      "step": 17827
    },
    {
      "epoch": 3.117464480874317,
      "grad_norm": 0.6455481199517434,
      "learning_rate": 2.444317075754381e-05,
      "loss": 11.4502,
      "step": 17828
    },
    {
      "epoch": 3.117639344262295,
      "grad_norm": 0.7440975935093463,
      "learning_rate": 2.4433895144892005e-05,
      "loss": 11.2829,
      "step": 17829
    },
    {
      "epoch": 3.117814207650273,
      "grad_norm": 0.6589505820820564,
      "learning_rate": 2.4424621047572583e-05,
      "loss": 11.3406,
      "step": 17830
    },
    {
      "epoch": 3.1179890710382514,
      "grad_norm": 0.5938169566327913,
      "learning_rate": 2.441534846577159e-05,
      "loss": 11.5856,
      "step": 17831
    },
    {
      "epoch": 3.1181639344262293,
      "grad_norm": 0.7981880157967043,
      "learning_rate": 2.4406077399674963e-05,
      "loss": 11.6179,
      "step": 17832
    },
    {
      "epoch": 3.1183387978142076,
      "grad_norm": 0.655516818533025,
      "learning_rate": 2.4396807849468562e-05,
      "loss": 11.2106,
      "step": 17833
    },
    {
      "epoch": 3.118513661202186,
      "grad_norm": 0.6747285763625293,
      "learning_rate": 2.438753981533829e-05,
      "loss": 11.4263,
      "step": 17834
    },
    {
      "epoch": 3.118688524590164,
      "grad_norm": 0.6870355500321268,
      "learning_rate": 2.4378273297469978e-05,
      "loss": 11.2958,
      "step": 17835
    },
    {
      "epoch": 3.118863387978142,
      "grad_norm": 0.7273688004645998,
      "learning_rate": 2.4369008296049533e-05,
      "loss": 11.3195,
      "step": 17836
    },
    {
      "epoch": 3.11903825136612,
      "grad_norm": 0.6449144760434472,
      "learning_rate": 2.4359744811262665e-05,
      "loss": 11.482,
      "step": 17837
    },
    {
      "epoch": 3.1192131147540985,
      "grad_norm": 0.5657714037311841,
      "learning_rate": 2.4350482843295164e-05,
      "loss": 11.8309,
      "step": 17838
    },
    {
      "epoch": 3.1193879781420764,
      "grad_norm": 0.7084208891328143,
      "learning_rate": 2.4341222392332742e-05,
      "loss": 11.4728,
      "step": 17839
    },
    {
      "epoch": 3.1195628415300547,
      "grad_norm": 0.6872415089610366,
      "learning_rate": 2.4331963458561102e-05,
      "loss": 11.5266,
      "step": 17840
    },
    {
      "epoch": 3.1197377049180326,
      "grad_norm": 0.6767555806590271,
      "learning_rate": 2.4322706042165967e-05,
      "loss": 11.469,
      "step": 17841
    },
    {
      "epoch": 3.119912568306011,
      "grad_norm": 0.6272097553763123,
      "learning_rate": 2.4313450143332916e-05,
      "loss": 11.3426,
      "step": 17842
    },
    {
      "epoch": 3.1200874316939893,
      "grad_norm": 0.7138097628537662,
      "learning_rate": 2.4304195762247585e-05,
      "loss": 11.5347,
      "step": 17843
    },
    {
      "epoch": 3.120262295081967,
      "grad_norm": 0.686682397639243,
      "learning_rate": 2.429494289909553e-05,
      "loss": 11.2428,
      "step": 17844
    },
    {
      "epoch": 3.1204371584699455,
      "grad_norm": 0.670362189545172,
      "learning_rate": 2.4285691554062295e-05,
      "loss": 11.19,
      "step": 17845
    },
    {
      "epoch": 3.1206120218579234,
      "grad_norm": 0.6326323996160323,
      "learning_rate": 2.4276441727333476e-05,
      "loss": 11.4782,
      "step": 17846
    },
    {
      "epoch": 3.1207868852459018,
      "grad_norm": 0.6162762274306984,
      "learning_rate": 2.4267193419094468e-05,
      "loss": 11.6027,
      "step": 17847
    },
    {
      "epoch": 3.1209617486338797,
      "grad_norm": 0.632306874619392,
      "learning_rate": 2.425794662953076e-05,
      "loss": 11.7366,
      "step": 17848
    },
    {
      "epoch": 3.121136612021858,
      "grad_norm": 0.6563690534905006,
      "learning_rate": 2.4248701358827764e-05,
      "loss": 11.1134,
      "step": 17849
    },
    {
      "epoch": 3.121311475409836,
      "grad_norm": 0.6340789958141454,
      "learning_rate": 2.4239457607170946e-05,
      "loss": 11.5683,
      "step": 17850
    },
    {
      "epoch": 3.1214863387978142,
      "grad_norm": 0.6821749297337492,
      "learning_rate": 2.423021537474558e-05,
      "loss": 11.3423,
      "step": 17851
    },
    {
      "epoch": 3.121661202185792,
      "grad_norm": 0.6702514293576471,
      "learning_rate": 2.422097466173705e-05,
      "loss": 11.4368,
      "step": 17852
    },
    {
      "epoch": 3.1218360655737705,
      "grad_norm": 0.6811799243549325,
      "learning_rate": 2.4211735468330642e-05,
      "loss": 11.3691,
      "step": 17853
    },
    {
      "epoch": 3.122010928961749,
      "grad_norm": 0.6582621657697622,
      "learning_rate": 2.420249779471162e-05,
      "loss": 11.3529,
      "step": 17854
    },
    {
      "epoch": 3.1221857923497267,
      "grad_norm": 0.6875853661459597,
      "learning_rate": 2.41932616410653e-05,
      "loss": 11.4029,
      "step": 17855
    },
    {
      "epoch": 3.122360655737705,
      "grad_norm": 0.6351549738477642,
      "learning_rate": 2.4184027007576803e-05,
      "loss": 11.5741,
      "step": 17856
    },
    {
      "epoch": 3.122535519125683,
      "grad_norm": 0.705809029788488,
      "learning_rate": 2.4174793894431358e-05,
      "loss": 11.4362,
      "step": 17857
    },
    {
      "epoch": 3.1227103825136613,
      "grad_norm": 0.6341559177330363,
      "learning_rate": 2.4165562301814116e-05,
      "loss": 11.3555,
      "step": 17858
    },
    {
      "epoch": 3.122885245901639,
      "grad_norm": 0.6832805121471619,
      "learning_rate": 2.4156332229910182e-05,
      "loss": 11.4964,
      "step": 17859
    },
    {
      "epoch": 3.1230601092896175,
      "grad_norm": 0.7552862745602854,
      "learning_rate": 2.414710367890466e-05,
      "loss": 11.3745,
      "step": 17860
    },
    {
      "epoch": 3.1232349726775954,
      "grad_norm": 0.6856960178259563,
      "learning_rate": 2.4137876648982606e-05,
      "loss": 11.6207,
      "step": 17861
    },
    {
      "epoch": 3.1234098360655738,
      "grad_norm": 0.7615748009396127,
      "learning_rate": 2.412865114032906e-05,
      "loss": 11.4591,
      "step": 17862
    },
    {
      "epoch": 3.123584699453552,
      "grad_norm": 0.6786283974578292,
      "learning_rate": 2.4119427153129004e-05,
      "loss": 11.6133,
      "step": 17863
    },
    {
      "epoch": 3.12375956284153,
      "grad_norm": 0.6501357330823689,
      "learning_rate": 2.4110204687567427e-05,
      "loss": 11.636,
      "step": 17864
    },
    {
      "epoch": 3.1239344262295083,
      "grad_norm": 0.6464853005862302,
      "learning_rate": 2.4100983743829252e-05,
      "loss": 11.5267,
      "step": 17865
    },
    {
      "epoch": 3.1241092896174862,
      "grad_norm": 0.6482565875307905,
      "learning_rate": 2.4091764322099396e-05,
      "loss": 11.2198,
      "step": 17866
    },
    {
      "epoch": 3.1242841530054646,
      "grad_norm": 0.6721809921052476,
      "learning_rate": 2.4082546422562736e-05,
      "loss": 11.4417,
      "step": 17867
    },
    {
      "epoch": 3.1244590163934425,
      "grad_norm": 0.673409566033141,
      "learning_rate": 2.4073330045404118e-05,
      "loss": 11.444,
      "step": 17868
    },
    {
      "epoch": 3.124633879781421,
      "grad_norm": 0.6463560948297495,
      "learning_rate": 2.4064115190808357e-05,
      "loss": 11.5811,
      "step": 17869
    },
    {
      "epoch": 3.1248087431693987,
      "grad_norm": 0.6584641839101553,
      "learning_rate": 2.4054901858960244e-05,
      "loss": 11.2359,
      "step": 17870
    },
    {
      "epoch": 3.124983606557377,
      "grad_norm": 0.5991857752057093,
      "learning_rate": 2.404569005004452e-05,
      "loss": 11.4459,
      "step": 17871
    },
    {
      "epoch": 3.1251584699453554,
      "grad_norm": 0.6741097803319797,
      "learning_rate": 2.4036479764245924e-05,
      "loss": 11.4481,
      "step": 17872
    },
    {
      "epoch": 3.1253333333333333,
      "grad_norm": 0.6932662419587077,
      "learning_rate": 2.4027271001749154e-05,
      "loss": 11.6288,
      "step": 17873
    },
    {
      "epoch": 3.1255081967213116,
      "grad_norm": 0.719992880049048,
      "learning_rate": 2.4018063762738863e-05,
      "loss": 11.3032,
      "step": 17874
    },
    {
      "epoch": 3.1256830601092895,
      "grad_norm": 0.7028600092633194,
      "learning_rate": 2.4008858047399684e-05,
      "loss": 11.3446,
      "step": 17875
    },
    {
      "epoch": 3.125857923497268,
      "grad_norm": 0.6673191760291669,
      "learning_rate": 2.3999653855916237e-05,
      "loss": 11.5447,
      "step": 17876
    },
    {
      "epoch": 3.1260327868852458,
      "grad_norm": 0.7690699855634909,
      "learning_rate": 2.3990451188473073e-05,
      "loss": 11.3637,
      "step": 17877
    },
    {
      "epoch": 3.126207650273224,
      "grad_norm": 0.8075338373138372,
      "learning_rate": 2.3981250045254745e-05,
      "loss": 11.4334,
      "step": 17878
    },
    {
      "epoch": 3.126382513661202,
      "grad_norm": 0.7194256314839592,
      "learning_rate": 2.3972050426445758e-05,
      "loss": 11.513,
      "step": 17879
    },
    {
      "epoch": 3.1265573770491804,
      "grad_norm": 0.7206058144142891,
      "learning_rate": 2.39628523322306e-05,
      "loss": 11.2486,
      "step": 17880
    },
    {
      "epoch": 3.1267322404371587,
      "grad_norm": 0.7119600811309159,
      "learning_rate": 2.3953655762793716e-05,
      "loss": 11.2979,
      "step": 17881
    },
    {
      "epoch": 3.1269071038251366,
      "grad_norm": 0.6063415925591296,
      "learning_rate": 2.3944460718319526e-05,
      "loss": 11.6243,
      "step": 17882
    },
    {
      "epoch": 3.127081967213115,
      "grad_norm": 0.6973832805740637,
      "learning_rate": 2.3935267198992428e-05,
      "loss": 11.282,
      "step": 17883
    },
    {
      "epoch": 3.127256830601093,
      "grad_norm": 0.7318936931850779,
      "learning_rate": 2.392607520499677e-05,
      "loss": 11.4486,
      "step": 17884
    },
    {
      "epoch": 3.127431693989071,
      "grad_norm": 0.6095277410795148,
      "learning_rate": 2.3916884736516877e-05,
      "loss": 11.4648,
      "step": 17885
    },
    {
      "epoch": 3.127606557377049,
      "grad_norm": 0.648014629078848,
      "learning_rate": 2.390769579373705e-05,
      "loss": 11.6809,
      "step": 17886
    },
    {
      "epoch": 3.1277814207650274,
      "grad_norm": 0.6540427434054303,
      "learning_rate": 2.3898508376841577e-05,
      "loss": 11.6069,
      "step": 17887
    },
    {
      "epoch": 3.1279562841530053,
      "grad_norm": 0.6314704907432223,
      "learning_rate": 2.3889322486014632e-05,
      "loss": 11.3981,
      "step": 17888
    },
    {
      "epoch": 3.1281311475409836,
      "grad_norm": 0.6318168094483515,
      "learning_rate": 2.388013812144049e-05,
      "loss": 11.5379,
      "step": 17889
    },
    {
      "epoch": 3.1283060109289615,
      "grad_norm": 0.6459330127822469,
      "learning_rate": 2.3870955283303287e-05,
      "loss": 11.3804,
      "step": 17890
    },
    {
      "epoch": 3.12848087431694,
      "grad_norm": 0.6790527007643107,
      "learning_rate": 2.3861773971787184e-05,
      "loss": 11.4094,
      "step": 17891
    },
    {
      "epoch": 3.1286557377049182,
      "grad_norm": 0.6534577145695757,
      "learning_rate": 2.3852594187076306e-05,
      "loss": 11.3362,
      "step": 17892
    },
    {
      "epoch": 3.128830601092896,
      "grad_norm": 0.6672487094356159,
      "learning_rate": 2.3843415929354674e-05,
      "loss": 11.2991,
      "step": 17893
    },
    {
      "epoch": 3.1290054644808745,
      "grad_norm": 0.7198733001989932,
      "learning_rate": 2.3834239198806408e-05,
      "loss": 11.2265,
      "step": 17894
    },
    {
      "epoch": 3.1291803278688524,
      "grad_norm": 0.6694225089065863,
      "learning_rate": 2.3825063995615505e-05,
      "loss": 11.3954,
      "step": 17895
    },
    {
      "epoch": 3.1293551912568307,
      "grad_norm": 0.7695589933185834,
      "learning_rate": 2.3815890319965984e-05,
      "loss": 11.7351,
      "step": 17896
    },
    {
      "epoch": 3.1295300546448086,
      "grad_norm": 0.6880951075652881,
      "learning_rate": 2.380671817204172e-05,
      "loss": 11.6449,
      "step": 17897
    },
    {
      "epoch": 3.129704918032787,
      "grad_norm": 0.6738055606749463,
      "learning_rate": 2.3797547552026732e-05,
      "loss": 11.3303,
      "step": 17898
    },
    {
      "epoch": 3.129879781420765,
      "grad_norm": 0.7164096434567412,
      "learning_rate": 2.378837846010489e-05,
      "loss": 11.5076,
      "step": 17899
    },
    {
      "epoch": 3.130054644808743,
      "grad_norm": 0.6569163443112508,
      "learning_rate": 2.3779210896460045e-05,
      "loss": 11.4678,
      "step": 17900
    },
    {
      "epoch": 3.130229508196721,
      "grad_norm": 0.6977717793502706,
      "learning_rate": 2.3770044861276096e-05,
      "loss": 11.3788,
      "step": 17901
    },
    {
      "epoch": 3.1304043715846994,
      "grad_norm": 0.7215037979997087,
      "learning_rate": 2.3760880354736748e-05,
      "loss": 11.6507,
      "step": 17902
    },
    {
      "epoch": 3.1305792349726778,
      "grad_norm": 0.6142356940335998,
      "learning_rate": 2.3751717377025852e-05,
      "loss": 11.6095,
      "step": 17903
    },
    {
      "epoch": 3.1307540983606557,
      "grad_norm": 0.6809208497754432,
      "learning_rate": 2.3742555928327137e-05,
      "loss": 11.5483,
      "step": 17904
    },
    {
      "epoch": 3.130928961748634,
      "grad_norm": 0.6682334560385572,
      "learning_rate": 2.373339600882435e-05,
      "loss": 11.6475,
      "step": 17905
    },
    {
      "epoch": 3.131103825136612,
      "grad_norm": 0.6716972242439038,
      "learning_rate": 2.372423761870112e-05,
      "loss": 11.1862,
      "step": 17906
    },
    {
      "epoch": 3.1312786885245902,
      "grad_norm": 0.6837602368272655,
      "learning_rate": 2.3715080758141083e-05,
      "loss": 11.434,
      "step": 17907
    },
    {
      "epoch": 3.131453551912568,
      "grad_norm": 0.7110683149183264,
      "learning_rate": 2.3705925427327937e-05,
      "loss": 11.4278,
      "step": 17908
    },
    {
      "epoch": 3.1316284153005465,
      "grad_norm": 0.7213612239226779,
      "learning_rate": 2.3696771626445236e-05,
      "loss": 11.4778,
      "step": 17909
    },
    {
      "epoch": 3.1318032786885244,
      "grad_norm": 0.6662876907795434,
      "learning_rate": 2.368761935567657e-05,
      "loss": 11.6783,
      "step": 17910
    },
    {
      "epoch": 3.1319781420765027,
      "grad_norm": 0.6260751388345966,
      "learning_rate": 2.367846861520543e-05,
      "loss": 11.7075,
      "step": 17911
    },
    {
      "epoch": 3.132153005464481,
      "grad_norm": 0.7310740217341444,
      "learning_rate": 2.3669319405215296e-05,
      "loss": 11.3744,
      "step": 17912
    },
    {
      "epoch": 3.132327868852459,
      "grad_norm": 0.6793972179902891,
      "learning_rate": 2.36601717258897e-05,
      "loss": 11.4427,
      "step": 17913
    },
    {
      "epoch": 3.1325027322404373,
      "grad_norm": 0.6686698187769331,
      "learning_rate": 2.3651025577412077e-05,
      "loss": 11.6932,
      "step": 17914
    },
    {
      "epoch": 3.132677595628415,
      "grad_norm": 0.6204821287690302,
      "learning_rate": 2.36418809599658e-05,
      "loss": 11.3477,
      "step": 17915
    },
    {
      "epoch": 3.1328524590163935,
      "grad_norm": 0.6968133090392692,
      "learning_rate": 2.3632737873734258e-05,
      "loss": 11.7111,
      "step": 17916
    },
    {
      "epoch": 3.1330273224043714,
      "grad_norm": 0.7009635086289269,
      "learning_rate": 2.3623596318900763e-05,
      "loss": 11.3885,
      "step": 17917
    },
    {
      "epoch": 3.1332021857923498,
      "grad_norm": 0.6500132835363601,
      "learning_rate": 2.36144562956487e-05,
      "loss": 11.5669,
      "step": 17918
    },
    {
      "epoch": 3.1333770491803277,
      "grad_norm": 0.7961996259674593,
      "learning_rate": 2.3605317804161353e-05,
      "loss": 11.2788,
      "step": 17919
    },
    {
      "epoch": 3.133551912568306,
      "grad_norm": 0.6513484844416154,
      "learning_rate": 2.3596180844621917e-05,
      "loss": 11.6925,
      "step": 17920
    },
    {
      "epoch": 3.1337267759562843,
      "grad_norm": 0.660342226430966,
      "learning_rate": 2.3587045417213616e-05,
      "loss": 11.3819,
      "step": 17921
    },
    {
      "epoch": 3.1339016393442622,
      "grad_norm": 0.6260796842154408,
      "learning_rate": 2.35779115221197e-05,
      "loss": 10.9349,
      "step": 17922
    },
    {
      "epoch": 3.1340765027322406,
      "grad_norm": 0.6941923813097215,
      "learning_rate": 2.3568779159523334e-05,
      "loss": 11.3029,
      "step": 17923
    },
    {
      "epoch": 3.1342513661202185,
      "grad_norm": 0.6846153816284546,
      "learning_rate": 2.3559648329607596e-05,
      "loss": 11.5608,
      "step": 17924
    },
    {
      "epoch": 3.134426229508197,
      "grad_norm": 0.7124032741767836,
      "learning_rate": 2.355051903255562e-05,
      "loss": 11.2474,
      "step": 17925
    },
    {
      "epoch": 3.1346010928961747,
      "grad_norm": 0.6397556381423543,
      "learning_rate": 2.3541391268550417e-05,
      "loss": 11.398,
      "step": 17926
    },
    {
      "epoch": 3.134775956284153,
      "grad_norm": 0.7765327003753256,
      "learning_rate": 2.3532265037775124e-05,
      "loss": 11.4084,
      "step": 17927
    },
    {
      "epoch": 3.134950819672131,
      "grad_norm": 0.6989001077703713,
      "learning_rate": 2.352314034041273e-05,
      "loss": 11.5143,
      "step": 17928
    },
    {
      "epoch": 3.1351256830601093,
      "grad_norm": 0.7010373501877935,
      "learning_rate": 2.351401717664615e-05,
      "loss": 11.1428,
      "step": 17929
    },
    {
      "epoch": 3.1353005464480876,
      "grad_norm": 0.7124856780889423,
      "learning_rate": 2.350489554665838e-05,
      "loss": 11.3264,
      "step": 17930
    },
    {
      "epoch": 3.1354754098360655,
      "grad_norm": 0.6329104786850933,
      "learning_rate": 2.3495775450632283e-05,
      "loss": 11.322,
      "step": 17931
    },
    {
      "epoch": 3.135650273224044,
      "grad_norm": 0.6662580143319091,
      "learning_rate": 2.3486656888750858e-05,
      "loss": 11.5241,
      "step": 17932
    },
    {
      "epoch": 3.135825136612022,
      "grad_norm": 0.6685838563391103,
      "learning_rate": 2.3477539861196853e-05,
      "loss": 11.5,
      "step": 17933
    },
    {
      "epoch": 3.136,
      "grad_norm": 0.6495851742786984,
      "learning_rate": 2.346842436815313e-05,
      "loss": 11.276,
      "step": 17934
    },
    {
      "epoch": 3.136174863387978,
      "grad_norm": 0.6536871692995705,
      "learning_rate": 2.345931040980247e-05,
      "loss": 11.6715,
      "step": 17935
    },
    {
      "epoch": 3.1363497267759564,
      "grad_norm": 0.6563741885773919,
      "learning_rate": 2.3450197986327637e-05,
      "loss": 11.3221,
      "step": 17936
    },
    {
      "epoch": 3.1365245901639343,
      "grad_norm": 0.7233218519443616,
      "learning_rate": 2.3441087097911418e-05,
      "loss": 11.5246,
      "step": 17937
    },
    {
      "epoch": 3.1366994535519126,
      "grad_norm": 0.6987575912352167,
      "learning_rate": 2.3431977744736445e-05,
      "loss": 11.7118,
      "step": 17938
    },
    {
      "epoch": 3.136874316939891,
      "grad_norm": 0.7216717312838933,
      "learning_rate": 2.3422869926985415e-05,
      "loss": 11.1852,
      "step": 17939
    },
    {
      "epoch": 3.137049180327869,
      "grad_norm": 0.6578614934378297,
      "learning_rate": 2.341376364484097e-05,
      "loss": 11.2807,
      "step": 17940
    },
    {
      "epoch": 3.137224043715847,
      "grad_norm": 0.6866222496925669,
      "learning_rate": 2.3404658898485697e-05,
      "loss": 11.5886,
      "step": 17941
    },
    {
      "epoch": 3.137398907103825,
      "grad_norm": 0.6839605638415112,
      "learning_rate": 2.339555568810221e-05,
      "loss": 11.3439,
      "step": 17942
    },
    {
      "epoch": 3.1375737704918034,
      "grad_norm": 0.6481163600072225,
      "learning_rate": 2.3386454013873025e-05,
      "loss": 11.4301,
      "step": 17943
    },
    {
      "epoch": 3.1377486338797813,
      "grad_norm": 0.6073143484066507,
      "learning_rate": 2.337735387598068e-05,
      "loss": 11.4202,
      "step": 17944
    },
    {
      "epoch": 3.1379234972677597,
      "grad_norm": 0.6872171625401449,
      "learning_rate": 2.336825527460762e-05,
      "loss": 11.3277,
      "step": 17945
    },
    {
      "epoch": 3.1380983606557376,
      "grad_norm": 0.7039544249417994,
      "learning_rate": 2.3359158209936394e-05,
      "loss": 11.4457,
      "step": 17946
    },
    {
      "epoch": 3.138273224043716,
      "grad_norm": 0.6542673659095142,
      "learning_rate": 2.3350062682149332e-05,
      "loss": 11.6221,
      "step": 17947
    },
    {
      "epoch": 3.138448087431694,
      "grad_norm": 0.6252380706356863,
      "learning_rate": 2.3340968691428866e-05,
      "loss": 11.6206,
      "step": 17948
    },
    {
      "epoch": 3.138622950819672,
      "grad_norm": 0.6461120157343089,
      "learning_rate": 2.333187623795734e-05,
      "loss": 11.135,
      "step": 17949
    },
    {
      "epoch": 3.1387978142076505,
      "grad_norm": 0.6681842172537469,
      "learning_rate": 2.3322785321917106e-05,
      "loss": 11.3692,
      "step": 17950
    },
    {
      "epoch": 3.1389726775956284,
      "grad_norm": 0.6332226751441238,
      "learning_rate": 2.3313695943490454e-05,
      "loss": 11.5986,
      "step": 17951
    },
    {
      "epoch": 3.1391475409836067,
      "grad_norm": 0.6847463304415564,
      "learning_rate": 2.3304608102859648e-05,
      "loss": 11.4864,
      "step": 17952
    },
    {
      "epoch": 3.1393224043715846,
      "grad_norm": 0.6634219974748183,
      "learning_rate": 2.329552180020694e-05,
      "loss": 11.6183,
      "step": 17953
    },
    {
      "epoch": 3.139497267759563,
      "grad_norm": 0.7140842088676175,
      "learning_rate": 2.3286437035714547e-05,
      "loss": 11.4844,
      "step": 17954
    },
    {
      "epoch": 3.139672131147541,
      "grad_norm": 0.647818335894153,
      "learning_rate": 2.3277353809564617e-05,
      "loss": 11.3872,
      "step": 17955
    },
    {
      "epoch": 3.139846994535519,
      "grad_norm": 0.6767899236189339,
      "learning_rate": 2.3268272121939316e-05,
      "loss": 11.205,
      "step": 17956
    },
    {
      "epoch": 3.140021857923497,
      "grad_norm": 0.6653421790775463,
      "learning_rate": 2.3259191973020768e-05,
      "loss": 11.3204,
      "step": 17957
    },
    {
      "epoch": 3.1401967213114754,
      "grad_norm": 0.669849403551724,
      "learning_rate": 2.325011336299103e-05,
      "loss": 11.4068,
      "step": 17958
    },
    {
      "epoch": 3.1403715846994533,
      "grad_norm": 0.6641667159404185,
      "learning_rate": 2.324103629203218e-05,
      "loss": 11.5521,
      "step": 17959
    },
    {
      "epoch": 3.1405464480874317,
      "grad_norm": 0.6116074197989104,
      "learning_rate": 2.323196076032623e-05,
      "loss": 11.4163,
      "step": 17960
    },
    {
      "epoch": 3.14072131147541,
      "grad_norm": 0.702846104642491,
      "learning_rate": 2.3222886768055174e-05,
      "loss": 11.4121,
      "step": 17961
    },
    {
      "epoch": 3.140896174863388,
      "grad_norm": 0.6863466618876185,
      "learning_rate": 2.3213814315400984e-05,
      "loss": 10.9467,
      "step": 17962
    },
    {
      "epoch": 3.1410710382513662,
      "grad_norm": 0.648569806559998,
      "learning_rate": 2.320474340254557e-05,
      "loss": 11.4828,
      "step": 17963
    },
    {
      "epoch": 3.141245901639344,
      "grad_norm": 0.6551836996114673,
      "learning_rate": 2.3195674029670854e-05,
      "loss": 11.2415,
      "step": 17964
    },
    {
      "epoch": 3.1414207650273225,
      "grad_norm": 0.674317363491089,
      "learning_rate": 2.3186606196958693e-05,
      "loss": 11.3876,
      "step": 17965
    },
    {
      "epoch": 3.1415956284153004,
      "grad_norm": 0.6576848422324713,
      "learning_rate": 2.3177539904590916e-05,
      "loss": 11.6456,
      "step": 17966
    },
    {
      "epoch": 3.1417704918032787,
      "grad_norm": 0.6265061817853108,
      "learning_rate": 2.3168475152749346e-05,
      "loss": 11.5318,
      "step": 17967
    },
    {
      "epoch": 3.1419453551912566,
      "grad_norm": 0.6569760764457445,
      "learning_rate": 2.315941194161575e-05,
      "loss": 11.3991,
      "step": 17968
    },
    {
      "epoch": 3.142120218579235,
      "grad_norm": 0.6869017803382444,
      "learning_rate": 2.3150350271371878e-05,
      "loss": 11.2785,
      "step": 17969
    },
    {
      "epoch": 3.1422950819672133,
      "grad_norm": 0.6287989320566898,
      "learning_rate": 2.3141290142199445e-05,
      "loss": 11.4576,
      "step": 17970
    },
    {
      "epoch": 3.142469945355191,
      "grad_norm": 0.6738254166528975,
      "learning_rate": 2.3132231554280136e-05,
      "loss": 11.6151,
      "step": 17971
    },
    {
      "epoch": 3.1426448087431695,
      "grad_norm": 0.694639757969725,
      "learning_rate": 2.3123174507795586e-05,
      "loss": 11.46,
      "step": 17972
    },
    {
      "epoch": 3.1428196721311474,
      "grad_norm": 0.6471437828090819,
      "learning_rate": 2.311411900292745e-05,
      "loss": 11.4211,
      "step": 17973
    },
    {
      "epoch": 3.1429945355191258,
      "grad_norm": 0.6236390646346741,
      "learning_rate": 2.3105065039857275e-05,
      "loss": 11.5366,
      "step": 17974
    },
    {
      "epoch": 3.1431693989071037,
      "grad_norm": 0.6892956283666096,
      "learning_rate": 2.309601261876666e-05,
      "loss": 11.5109,
      "step": 17975
    },
    {
      "epoch": 3.143344262295082,
      "grad_norm": 0.6717334953907335,
      "learning_rate": 2.308696173983711e-05,
      "loss": 11.4015,
      "step": 17976
    },
    {
      "epoch": 3.14351912568306,
      "grad_norm": 0.6721998902667159,
      "learning_rate": 2.3077912403250134e-05,
      "loss": 11.5333,
      "step": 17977
    },
    {
      "epoch": 3.1436939890710383,
      "grad_norm": 0.7066189229282542,
      "learning_rate": 2.30688646091872e-05,
      "loss": 11.4661,
      "step": 17978
    },
    {
      "epoch": 3.1438688524590166,
      "grad_norm": 0.6530314834451173,
      "learning_rate": 2.3059818357829744e-05,
      "loss": 11.2352,
      "step": 17979
    },
    {
      "epoch": 3.1440437158469945,
      "grad_norm": 0.6314376878609402,
      "learning_rate": 2.305077364935916e-05,
      "loss": 11.5015,
      "step": 17980
    },
    {
      "epoch": 3.144218579234973,
      "grad_norm": 0.6655709616358988,
      "learning_rate": 2.3041730483956823e-05,
      "loss": 11.6467,
      "step": 17981
    },
    {
      "epoch": 3.1443934426229507,
      "grad_norm": 0.6189121478429852,
      "learning_rate": 2.30326888618041e-05,
      "loss": 11.4635,
      "step": 17982
    },
    {
      "epoch": 3.144568306010929,
      "grad_norm": 0.7182231386903047,
      "learning_rate": 2.3023648783082297e-05,
      "loss": 11.4171,
      "step": 17983
    },
    {
      "epoch": 3.144743169398907,
      "grad_norm": 0.6795653857735997,
      "learning_rate": 2.3014610247972635e-05,
      "loss": 11.385,
      "step": 17984
    },
    {
      "epoch": 3.1449180327868853,
      "grad_norm": 0.6320837243475772,
      "learning_rate": 2.3005573256656443e-05,
      "loss": 11.3407,
      "step": 17985
    },
    {
      "epoch": 3.145092896174863,
      "grad_norm": 0.729113851532604,
      "learning_rate": 2.299653780931491e-05,
      "loss": 11.0098,
      "step": 17986
    },
    {
      "epoch": 3.1452677595628415,
      "grad_norm": 0.7229715326218444,
      "learning_rate": 2.2987503906129228e-05,
      "loss": 11.4337,
      "step": 17987
    },
    {
      "epoch": 3.14544262295082,
      "grad_norm": 0.6721422061066118,
      "learning_rate": 2.297847154728058e-05,
      "loss": 11.3859,
      "step": 17988
    },
    {
      "epoch": 3.145617486338798,
      "grad_norm": 0.9488275134156025,
      "learning_rate": 2.296944073295e-05,
      "loss": 11.2791,
      "step": 17989
    },
    {
      "epoch": 3.145792349726776,
      "grad_norm": 0.6642076780393581,
      "learning_rate": 2.296041146331869e-05,
      "loss": 11.546,
      "step": 17990
    },
    {
      "epoch": 3.145967213114754,
      "grad_norm": 0.6405616395130619,
      "learning_rate": 2.2951383738567656e-05,
      "loss": 11.8106,
      "step": 17991
    },
    {
      "epoch": 3.1461420765027324,
      "grad_norm": 0.6409450869009966,
      "learning_rate": 2.294235755887799e-05,
      "loss": 11.3751,
      "step": 17992
    },
    {
      "epoch": 3.1463169398907103,
      "grad_norm": 0.8689425067769061,
      "learning_rate": 2.2933332924430593e-05,
      "loss": 11.4247,
      "step": 17993
    },
    {
      "epoch": 3.1464918032786886,
      "grad_norm": 0.684228003080014,
      "learning_rate": 2.292430983540652e-05,
      "loss": 11.6334,
      "step": 17994
    },
    {
      "epoch": 3.1466666666666665,
      "grad_norm": 0.6870994590610169,
      "learning_rate": 2.291528829198669e-05,
      "loss": 11.555,
      "step": 17995
    },
    {
      "epoch": 3.146841530054645,
      "grad_norm": 0.633531206803064,
      "learning_rate": 2.2906268294352017e-05,
      "loss": 11.3512,
      "step": 17996
    },
    {
      "epoch": 3.147016393442623,
      "grad_norm": 0.636408885916802,
      "learning_rate": 2.2897249842683398e-05,
      "loss": 11.2811,
      "step": 17997
    },
    {
      "epoch": 3.147191256830601,
      "grad_norm": 0.6419988065201472,
      "learning_rate": 2.2888232937161603e-05,
      "loss": 11.6479,
      "step": 17998
    },
    {
      "epoch": 3.1473661202185794,
      "grad_norm": 0.683077066117231,
      "learning_rate": 2.2879217577967527e-05,
      "loss": 11.0952,
      "step": 17999
    },
    {
      "epoch": 3.1475409836065573,
      "grad_norm": 0.5972440188571562,
      "learning_rate": 2.2870203765281926e-05,
      "loss": 11.1544,
      "step": 18000
    },
    {
      "epoch": 3.1477158469945357,
      "grad_norm": 0.7267945056354755,
      "learning_rate": 2.2861191499285596e-05,
      "loss": 11.4868,
      "step": 18001
    },
    {
      "epoch": 3.1478907103825136,
      "grad_norm": 0.7758794723952211,
      "learning_rate": 2.2852180780159194e-05,
      "loss": 11.2846,
      "step": 18002
    },
    {
      "epoch": 3.148065573770492,
      "grad_norm": 0.6107374005638595,
      "learning_rate": 2.2843171608083414e-05,
      "loss": 10.8029,
      "step": 18003
    },
    {
      "epoch": 3.14824043715847,
      "grad_norm": 0.8079182627855357,
      "learning_rate": 2.2834163983238967e-05,
      "loss": 11.5835,
      "step": 18004
    },
    {
      "epoch": 3.148415300546448,
      "grad_norm": 0.6395777600043151,
      "learning_rate": 2.282515790580647e-05,
      "loss": 11.4685,
      "step": 18005
    },
    {
      "epoch": 3.148590163934426,
      "grad_norm": 0.7193947987288941,
      "learning_rate": 2.2816153375966543e-05,
      "loss": 11.4746,
      "step": 18006
    },
    {
      "epoch": 3.1487650273224044,
      "grad_norm": 0.7001429751004999,
      "learning_rate": 2.280715039389969e-05,
      "loss": 11.4062,
      "step": 18007
    },
    {
      "epoch": 3.1489398907103827,
      "grad_norm": 0.6770510350918177,
      "learning_rate": 2.279814895978646e-05,
      "loss": 11.4952,
      "step": 18008
    },
    {
      "epoch": 3.1491147540983606,
      "grad_norm": 0.6102459987801984,
      "learning_rate": 2.2789149073807414e-05,
      "loss": 11.5785,
      "step": 18009
    },
    {
      "epoch": 3.149289617486339,
      "grad_norm": 0.7148533755254189,
      "learning_rate": 2.2780150736143025e-05,
      "loss": 11.1185,
      "step": 18010
    },
    {
      "epoch": 3.149464480874317,
      "grad_norm": 0.8488770058077093,
      "learning_rate": 2.2771153946973688e-05,
      "loss": 11.3007,
      "step": 18011
    },
    {
      "epoch": 3.149639344262295,
      "grad_norm": 0.6452807423960288,
      "learning_rate": 2.276215870647983e-05,
      "loss": 11.4634,
      "step": 18012
    },
    {
      "epoch": 3.149814207650273,
      "grad_norm": 0.6356577692899844,
      "learning_rate": 2.2753165014841815e-05,
      "loss": 11.3102,
      "step": 18013
    },
    {
      "epoch": 3.1499890710382514,
      "grad_norm": 0.7037346889254374,
      "learning_rate": 2.2744172872240043e-05,
      "loss": 11.3653,
      "step": 18014
    },
    {
      "epoch": 3.1501639344262293,
      "grad_norm": 0.6773095552831566,
      "learning_rate": 2.273518227885485e-05,
      "loss": 11.4723,
      "step": 18015
    },
    {
      "epoch": 3.1503387978142077,
      "grad_norm": 0.7331344432889222,
      "learning_rate": 2.2726193234866455e-05,
      "loss": 11.3278,
      "step": 18016
    },
    {
      "epoch": 3.1505136612021856,
      "grad_norm": 0.6709568087846648,
      "learning_rate": 2.2717205740455115e-05,
      "loss": 11.3921,
      "step": 18017
    },
    {
      "epoch": 3.150688524590164,
      "grad_norm": 0.6474459614262956,
      "learning_rate": 2.270821979580112e-05,
      "loss": 11.163,
      "step": 18018
    },
    {
      "epoch": 3.1508633879781422,
      "grad_norm": 0.6774444672443417,
      "learning_rate": 2.2699235401084673e-05,
      "loss": 11.3459,
      "step": 18019
    },
    {
      "epoch": 3.15103825136612,
      "grad_norm": 0.7179064375341507,
      "learning_rate": 2.2690252556485858e-05,
      "loss": 11.5138,
      "step": 18020
    },
    {
      "epoch": 3.1512131147540985,
      "grad_norm": 0.6437014674190992,
      "learning_rate": 2.2681271262184856e-05,
      "loss": 11.4403,
      "step": 18021
    },
    {
      "epoch": 3.1513879781420764,
      "grad_norm": 0.7013173292660759,
      "learning_rate": 2.2672291518361743e-05,
      "loss": 11.433,
      "step": 18022
    },
    {
      "epoch": 3.1515628415300547,
      "grad_norm": 0.6687168876385599,
      "learning_rate": 2.2663313325196632e-05,
      "loss": 11.5444,
      "step": 18023
    },
    {
      "epoch": 3.1517377049180326,
      "grad_norm": 0.659323077680958,
      "learning_rate": 2.265433668286957e-05,
      "loss": 11.6536,
      "step": 18024
    },
    {
      "epoch": 3.151912568306011,
      "grad_norm": 0.6827085042869405,
      "learning_rate": 2.264536159156052e-05,
      "loss": 11.5894,
      "step": 18025
    },
    {
      "epoch": 3.152087431693989,
      "grad_norm": 0.6566483850927225,
      "learning_rate": 2.2636388051449475e-05,
      "loss": 11.3847,
      "step": 18026
    },
    {
      "epoch": 3.152262295081967,
      "grad_norm": 0.6734164862078851,
      "learning_rate": 2.2627416062716366e-05,
      "loss": 11.4162,
      "step": 18027
    },
    {
      "epoch": 3.1524371584699455,
      "grad_norm": 0.6906481500385788,
      "learning_rate": 2.2618445625541174e-05,
      "loss": 11.4688,
      "step": 18028
    },
    {
      "epoch": 3.1526120218579234,
      "grad_norm": 0.6378518460935249,
      "learning_rate": 2.260947674010372e-05,
      "loss": 11.5199,
      "step": 18029
    },
    {
      "epoch": 3.152786885245902,
      "grad_norm": 0.7163964073017274,
      "learning_rate": 2.260050940658388e-05,
      "loss": 11.472,
      "step": 18030
    },
    {
      "epoch": 3.1529617486338797,
      "grad_norm": 0.662682056631731,
      "learning_rate": 2.2591543625161472e-05,
      "loss": 11.3805,
      "step": 18031
    },
    {
      "epoch": 3.153136612021858,
      "grad_norm": 0.6721860488929109,
      "learning_rate": 2.2582579396016257e-05,
      "loss": 11.3863,
      "step": 18032
    },
    {
      "epoch": 3.153311475409836,
      "grad_norm": 0.7008230211567471,
      "learning_rate": 2.2573616719328084e-05,
      "loss": 11.334,
      "step": 18033
    },
    {
      "epoch": 3.1534863387978143,
      "grad_norm": 0.753136736148162,
      "learning_rate": 2.2564655595276596e-05,
      "loss": 11.1012,
      "step": 18034
    },
    {
      "epoch": 3.153661202185792,
      "grad_norm": 0.694840764566596,
      "learning_rate": 2.255569602404153e-05,
      "loss": 11.6034,
      "step": 18035
    },
    {
      "epoch": 3.1538360655737705,
      "grad_norm": 0.6899806728514597,
      "learning_rate": 2.2546738005802538e-05,
      "loss": 11.3988,
      "step": 18036
    },
    {
      "epoch": 3.154010928961749,
      "grad_norm": 0.6548150551743478,
      "learning_rate": 2.2537781540739255e-05,
      "loss": 11.3912,
      "step": 18037
    },
    {
      "epoch": 3.1541857923497267,
      "grad_norm": 0.6748797903782855,
      "learning_rate": 2.2528826629031306e-05,
      "loss": 11.228,
      "step": 18038
    },
    {
      "epoch": 3.154360655737705,
      "grad_norm": 0.6835921861310061,
      "learning_rate": 2.251987327085825e-05,
      "loss": 11.3366,
      "step": 18039
    },
    {
      "epoch": 3.154535519125683,
      "grad_norm": 0.7432451671078707,
      "learning_rate": 2.2510921466399616e-05,
      "loss": 11.3273,
      "step": 18040
    },
    {
      "epoch": 3.1547103825136613,
      "grad_norm": 0.6459258378836207,
      "learning_rate": 2.2501971215834915e-05,
      "loss": 11.5803,
      "step": 18041
    },
    {
      "epoch": 3.154885245901639,
      "grad_norm": 0.6042910124359865,
      "learning_rate": 2.2493022519343686e-05,
      "loss": 11.5632,
      "step": 18042
    },
    {
      "epoch": 3.1550601092896176,
      "grad_norm": 0.6046369964380213,
      "learning_rate": 2.2484075377105318e-05,
      "loss": 11.3702,
      "step": 18043
    },
    {
      "epoch": 3.1552349726775955,
      "grad_norm": 0.6212855923963887,
      "learning_rate": 2.2475129789299243e-05,
      "loss": 11.017,
      "step": 18044
    },
    {
      "epoch": 3.155409836065574,
      "grad_norm": 0.7694229220369835,
      "learning_rate": 2.2466185756104842e-05,
      "loss": 11.4542,
      "step": 18045
    },
    {
      "epoch": 3.155584699453552,
      "grad_norm": 0.63507881245537,
      "learning_rate": 2.245724327770148e-05,
      "loss": 11.5539,
      "step": 18046
    },
    {
      "epoch": 3.15575956284153,
      "grad_norm": 0.6213522412651515,
      "learning_rate": 2.244830235426847e-05,
      "loss": 11.5549,
      "step": 18047
    },
    {
      "epoch": 3.1559344262295084,
      "grad_norm": 0.640839352287823,
      "learning_rate": 2.2439362985985124e-05,
      "loss": 11.3,
      "step": 18048
    },
    {
      "epoch": 3.1561092896174863,
      "grad_norm": 0.6795202150390146,
      "learning_rate": 2.2430425173030677e-05,
      "loss": 11.246,
      "step": 18049
    },
    {
      "epoch": 3.1562841530054646,
      "grad_norm": 0.6610363929614039,
      "learning_rate": 2.2421488915584387e-05,
      "loss": 11.4007,
      "step": 18050
    },
    {
      "epoch": 3.1564590163934425,
      "grad_norm": 0.7667746797322046,
      "learning_rate": 2.241255421382544e-05,
      "loss": 11.5173,
      "step": 18051
    },
    {
      "epoch": 3.156633879781421,
      "grad_norm": 0.6770112242753762,
      "learning_rate": 2.2403621067933012e-05,
      "loss": 11.4755,
      "step": 18052
    },
    {
      "epoch": 3.1568087431693987,
      "grad_norm": 0.6780744140821667,
      "learning_rate": 2.239468947808623e-05,
      "loss": 11.5166,
      "step": 18053
    },
    {
      "epoch": 3.156983606557377,
      "grad_norm": 0.6136165485878035,
      "learning_rate": 2.2385759444464194e-05,
      "loss": 11.476,
      "step": 18054
    },
    {
      "epoch": 3.157158469945355,
      "grad_norm": 0.6769767210418589,
      "learning_rate": 2.2376830967246e-05,
      "loss": 11.3057,
      "step": 18055
    },
    {
      "epoch": 3.1573333333333333,
      "grad_norm": 0.6783990612390935,
      "learning_rate": 2.2367904046610665e-05,
      "loss": 11.3447,
      "step": 18056
    },
    {
      "epoch": 3.1575081967213117,
      "grad_norm": 0.6541671423578863,
      "learning_rate": 2.235897868273723e-05,
      "loss": 11.351,
      "step": 18057
    },
    {
      "epoch": 3.1576830601092896,
      "grad_norm": 0.6991672109461947,
      "learning_rate": 2.235005487580466e-05,
      "loss": 11.5904,
      "step": 18058
    },
    {
      "epoch": 3.157857923497268,
      "grad_norm": 0.6330635679550914,
      "learning_rate": 2.23411326259919e-05,
      "loss": 11.3589,
      "step": 18059
    },
    {
      "epoch": 3.158032786885246,
      "grad_norm": 0.664753021485409,
      "learning_rate": 2.2332211933477877e-05,
      "loss": 11.4494,
      "step": 18060
    },
    {
      "epoch": 3.158207650273224,
      "grad_norm": 0.7130038208314815,
      "learning_rate": 2.2323292798441474e-05,
      "loss": 11.5458,
      "step": 18061
    },
    {
      "epoch": 3.158382513661202,
      "grad_norm": 0.6221312967845415,
      "learning_rate": 2.2314375221061558e-05,
      "loss": 11.5881,
      "step": 18062
    },
    {
      "epoch": 3.1585573770491804,
      "grad_norm": 0.6724531539028833,
      "learning_rate": 2.2305459201516944e-05,
      "loss": 11.0068,
      "step": 18063
    },
    {
      "epoch": 3.1587322404371583,
      "grad_norm": 0.6080603391829913,
      "learning_rate": 2.2296544739986414e-05,
      "loss": 11.5598,
      "step": 18064
    },
    {
      "epoch": 3.1589071038251366,
      "grad_norm": 0.6815238385256703,
      "learning_rate": 2.2287631836648748e-05,
      "loss": 11.4876,
      "step": 18065
    },
    {
      "epoch": 3.159081967213115,
      "grad_norm": 0.650145932527557,
      "learning_rate": 2.2278720491682682e-05,
      "loss": 11.6451,
      "step": 18066
    },
    {
      "epoch": 3.159256830601093,
      "grad_norm": 0.7100790740146846,
      "learning_rate": 2.226981070526689e-05,
      "loss": 11.4871,
      "step": 18067
    },
    {
      "epoch": 3.159431693989071,
      "grad_norm": 0.6687743694106633,
      "learning_rate": 2.226090247758007e-05,
      "loss": 11.5676,
      "step": 18068
    },
    {
      "epoch": 3.159606557377049,
      "grad_norm": 0.7020461202468903,
      "learning_rate": 2.225199580880084e-05,
      "loss": 11.3503,
      "step": 18069
    },
    {
      "epoch": 3.1597814207650274,
      "grad_norm": 0.6542628399564627,
      "learning_rate": 2.2243090699107815e-05,
      "loss": 11.388,
      "step": 18070
    },
    {
      "epoch": 3.1599562841530053,
      "grad_norm": 0.6970921531548635,
      "learning_rate": 2.2234187148679564e-05,
      "loss": 11.695,
      "step": 18071
    },
    {
      "epoch": 3.1601311475409837,
      "grad_norm": 0.6695978728657568,
      "learning_rate": 2.2225285157694632e-05,
      "loss": 11.3055,
      "step": 18072
    },
    {
      "epoch": 3.1603060109289616,
      "grad_norm": 0.68885872544599,
      "learning_rate": 2.221638472633153e-05,
      "loss": 11.4989,
      "step": 18073
    },
    {
      "epoch": 3.16048087431694,
      "grad_norm": 0.6473209186907906,
      "learning_rate": 2.220748585476875e-05,
      "loss": 11.6142,
      "step": 18074
    },
    {
      "epoch": 3.160655737704918,
      "grad_norm": 0.6466144860127483,
      "learning_rate": 2.2198588543184728e-05,
      "loss": 11.3346,
      "step": 18075
    },
    {
      "epoch": 3.160830601092896,
      "grad_norm": 0.6354052027381779,
      "learning_rate": 2.2189692791757888e-05,
      "loss": 11.7998,
      "step": 18076
    },
    {
      "epoch": 3.1610054644808745,
      "grad_norm": 0.7183331308745943,
      "learning_rate": 2.218079860066662e-05,
      "loss": 11.4212,
      "step": 18077
    },
    {
      "epoch": 3.1611803278688524,
      "grad_norm": 0.7013762743049481,
      "learning_rate": 2.2171905970089268e-05,
      "loss": 11.1392,
      "step": 18078
    },
    {
      "epoch": 3.1613551912568307,
      "grad_norm": 0.7619379519045063,
      "learning_rate": 2.2163014900204193e-05,
      "loss": 11.509,
      "step": 18079
    },
    {
      "epoch": 3.1615300546448086,
      "grad_norm": 0.641027039446464,
      "learning_rate": 2.2154125391189607e-05,
      "loss": 11.239,
      "step": 18080
    },
    {
      "epoch": 3.161704918032787,
      "grad_norm": 0.6523404497873249,
      "learning_rate": 2.2145237443223854e-05,
      "loss": 11.1083,
      "step": 18081
    },
    {
      "epoch": 3.161879781420765,
      "grad_norm": 0.7156487581670797,
      "learning_rate": 2.2136351056485137e-05,
      "loss": 11.4748,
      "step": 18082
    },
    {
      "epoch": 3.162054644808743,
      "grad_norm": 0.6601706090670469,
      "learning_rate": 2.212746623115166e-05,
      "loss": 11.4988,
      "step": 18083
    },
    {
      "epoch": 3.162229508196721,
      "grad_norm": 0.6909381477512377,
      "learning_rate": 2.2118582967401604e-05,
      "loss": 11.3212,
      "step": 18084
    },
    {
      "epoch": 3.1624043715846994,
      "grad_norm": 0.6342269543070053,
      "learning_rate": 2.210970126541303e-05,
      "loss": 11.2719,
      "step": 18085
    },
    {
      "epoch": 3.162579234972678,
      "grad_norm": 0.6397377774789963,
      "learning_rate": 2.2100821125364137e-05,
      "loss": 11.5982,
      "step": 18086
    },
    {
      "epoch": 3.1627540983606557,
      "grad_norm": 0.6459072866563161,
      "learning_rate": 2.2091942547432955e-05,
      "loss": 11.3265,
      "step": 18087
    },
    {
      "epoch": 3.162928961748634,
      "grad_norm": 0.6899207438635012,
      "learning_rate": 2.2083065531797564e-05,
      "loss": 11.4525,
      "step": 18088
    },
    {
      "epoch": 3.163103825136612,
      "grad_norm": 0.6708580993727393,
      "learning_rate": 2.207419007863589e-05,
      "loss": 11.3249,
      "step": 18089
    },
    {
      "epoch": 3.1632786885245903,
      "grad_norm": 0.6531657654655497,
      "learning_rate": 2.2065316188125995e-05,
      "loss": 11.37,
      "step": 18090
    },
    {
      "epoch": 3.163453551912568,
      "grad_norm": 0.6961869928006198,
      "learning_rate": 2.2056443860445808e-05,
      "loss": 11.2352,
      "step": 18091
    },
    {
      "epoch": 3.1636284153005465,
      "grad_norm": 0.68414775254903,
      "learning_rate": 2.2047573095773234e-05,
      "loss": 11.4481,
      "step": 18092
    },
    {
      "epoch": 3.1638032786885244,
      "grad_norm": 0.7042116424074218,
      "learning_rate": 2.2038703894286182e-05,
      "loss": 11.3335,
      "step": 18093
    },
    {
      "epoch": 3.1639781420765027,
      "grad_norm": 0.7718712094158643,
      "learning_rate": 2.202983625616245e-05,
      "loss": 11.5144,
      "step": 18094
    },
    {
      "epoch": 3.164153005464481,
      "grad_norm": 0.7319098810424771,
      "learning_rate": 2.2020970181579926e-05,
      "loss": 11.5255,
      "step": 18095
    },
    {
      "epoch": 3.164327868852459,
      "grad_norm": 0.6659822655703389,
      "learning_rate": 2.2012105670716377e-05,
      "loss": 11.39,
      "step": 18096
    },
    {
      "epoch": 3.1645027322404373,
      "grad_norm": 0.6290239155891207,
      "learning_rate": 2.2003242723749583e-05,
      "loss": 11.353,
      "step": 18097
    },
    {
      "epoch": 3.164677595628415,
      "grad_norm": 0.6154604674739467,
      "learning_rate": 2.1994381340857227e-05,
      "loss": 11.3838,
      "step": 18098
    },
    {
      "epoch": 3.1648524590163936,
      "grad_norm": 0.6302420211225704,
      "learning_rate": 2.1985521522217024e-05,
      "loss": 11.3861,
      "step": 18099
    },
    {
      "epoch": 3.1650273224043715,
      "grad_norm": 0.7006551181877438,
      "learning_rate": 2.1976663268006657e-05,
      "loss": 11.1286,
      "step": 18100
    },
    {
      "epoch": 3.16520218579235,
      "grad_norm": 0.6575483058667067,
      "learning_rate": 2.196780657840376e-05,
      "loss": 11.148,
      "step": 18101
    },
    {
      "epoch": 3.1653770491803277,
      "grad_norm": 0.6592428262962323,
      "learning_rate": 2.1958951453585964e-05,
      "loss": 11.4909,
      "step": 18102
    },
    {
      "epoch": 3.165551912568306,
      "grad_norm": 0.6971872898631726,
      "learning_rate": 2.1950097893730782e-05,
      "loss": 11.3369,
      "step": 18103
    },
    {
      "epoch": 3.1657267759562844,
      "grad_norm": 0.6637525886863924,
      "learning_rate": 2.194124589901575e-05,
      "loss": 11.2383,
      "step": 18104
    },
    {
      "epoch": 3.1659016393442623,
      "grad_norm": 0.7199212681884032,
      "learning_rate": 2.1932395469618437e-05,
      "loss": 11.403,
      "step": 18105
    },
    {
      "epoch": 3.1660765027322406,
      "grad_norm": 0.6737788925114313,
      "learning_rate": 2.1923546605716327e-05,
      "loss": 11.3597,
      "step": 18106
    },
    {
      "epoch": 3.1662513661202185,
      "grad_norm": 0.6856178746747065,
      "learning_rate": 2.1914699307486796e-05,
      "loss": 11.6343,
      "step": 18107
    },
    {
      "epoch": 3.166426229508197,
      "grad_norm": 0.6299379245625095,
      "learning_rate": 2.1905853575107305e-05,
      "loss": 11.5923,
      "step": 18108
    },
    {
      "epoch": 3.1666010928961748,
      "grad_norm": 0.657549974574543,
      "learning_rate": 2.18970094087552e-05,
      "loss": 11.6157,
      "step": 18109
    },
    {
      "epoch": 3.166775956284153,
      "grad_norm": 0.6917688566926984,
      "learning_rate": 2.188816680860789e-05,
      "loss": 11.5128,
      "step": 18110
    },
    {
      "epoch": 3.166950819672131,
      "grad_norm": 0.6471350801565797,
      "learning_rate": 2.187932577484271e-05,
      "loss": 11.72,
      "step": 18111
    },
    {
      "epoch": 3.1671256830601093,
      "grad_norm": 0.6681170967127443,
      "learning_rate": 2.1870486307636873e-05,
      "loss": 11.5157,
      "step": 18112
    },
    {
      "epoch": 3.1673005464480872,
      "grad_norm": 0.6840029782178702,
      "learning_rate": 2.1861648407167646e-05,
      "loss": 11.5816,
      "step": 18113
    },
    {
      "epoch": 3.1674754098360656,
      "grad_norm": 0.7331840757435784,
      "learning_rate": 2.1852812073612304e-05,
      "loss": 11.1895,
      "step": 18114
    },
    {
      "epoch": 3.167650273224044,
      "grad_norm": 0.6862884523430232,
      "learning_rate": 2.184397730714807e-05,
      "loss": 11.5247,
      "step": 18115
    },
    {
      "epoch": 3.167825136612022,
      "grad_norm": 0.6447048083646134,
      "learning_rate": 2.1835144107952022e-05,
      "loss": 11.3435,
      "step": 18116
    },
    {
      "epoch": 3.168,
      "grad_norm": 0.6786448494358235,
      "learning_rate": 2.1826312476201338e-05,
      "loss": 11.4858,
      "step": 18117
    },
    {
      "epoch": 3.168174863387978,
      "grad_norm": 0.6707803371060841,
      "learning_rate": 2.1817482412073075e-05,
      "loss": 11.3,
      "step": 18118
    },
    {
      "epoch": 3.1683497267759564,
      "grad_norm": 0.6596078359693284,
      "learning_rate": 2.180865391574438e-05,
      "loss": 11.6602,
      "step": 18119
    },
    {
      "epoch": 3.1685245901639343,
      "grad_norm": 0.7465107926595479,
      "learning_rate": 2.179982698739228e-05,
      "loss": 11.5211,
      "step": 18120
    },
    {
      "epoch": 3.1686994535519126,
      "grad_norm": 0.696441668872285,
      "learning_rate": 2.179100162719372e-05,
      "loss": 11.2414,
      "step": 18121
    },
    {
      "epoch": 3.1688743169398905,
      "grad_norm": 0.6850002304268675,
      "learning_rate": 2.1782177835325712e-05,
      "loss": 11.448,
      "step": 18122
    },
    {
      "epoch": 3.169049180327869,
      "grad_norm": 0.7594954566004679,
      "learning_rate": 2.1773355611965174e-05,
      "loss": 11.5634,
      "step": 18123
    },
    {
      "epoch": 3.1692240437158468,
      "grad_norm": 0.7053094629429575,
      "learning_rate": 2.1764534957289106e-05,
      "loss": 11.4469,
      "step": 18124
    },
    {
      "epoch": 3.169398907103825,
      "grad_norm": 0.6981103811171131,
      "learning_rate": 2.1755715871474293e-05,
      "loss": 11.2537,
      "step": 18125
    },
    {
      "epoch": 3.1695737704918034,
      "grad_norm": 0.747923902538509,
      "learning_rate": 2.174689835469761e-05,
      "loss": 11.4641,
      "step": 18126
    },
    {
      "epoch": 3.1697486338797813,
      "grad_norm": 0.6579243759932308,
      "learning_rate": 2.1738082407135895e-05,
      "loss": 11.4903,
      "step": 18127
    },
    {
      "epoch": 3.1699234972677597,
      "grad_norm": 0.6426959091535691,
      "learning_rate": 2.172926802896589e-05,
      "loss": 11.4224,
      "step": 18128
    },
    {
      "epoch": 3.1700983606557376,
      "grad_norm": 0.7053495563863406,
      "learning_rate": 2.1720455220364444e-05,
      "loss": 11.2704,
      "step": 18129
    },
    {
      "epoch": 3.170273224043716,
      "grad_norm": 0.7443727390575497,
      "learning_rate": 2.171164398150819e-05,
      "loss": 11.4782,
      "step": 18130
    },
    {
      "epoch": 3.170448087431694,
      "grad_norm": 0.6321805492788728,
      "learning_rate": 2.170283431257385e-05,
      "loss": 11.5846,
      "step": 18131
    },
    {
      "epoch": 3.170622950819672,
      "grad_norm": 0.6436955755948588,
      "learning_rate": 2.1694026213738082e-05,
      "loss": 11.6898,
      "step": 18132
    },
    {
      "epoch": 3.17079781420765,
      "grad_norm": 0.661705888715084,
      "learning_rate": 2.1685219685177527e-05,
      "loss": 11.5933,
      "step": 18133
    },
    {
      "epoch": 3.1709726775956284,
      "grad_norm": 0.6889942053113968,
      "learning_rate": 2.1676414727068772e-05,
      "loss": 11.6612,
      "step": 18134
    },
    {
      "epoch": 3.1711475409836067,
      "grad_norm": 0.7182210221704887,
      "learning_rate": 2.166761133958838e-05,
      "loss": 11.4353,
      "step": 18135
    },
    {
      "epoch": 3.1713224043715846,
      "grad_norm": 0.6132328363046741,
      "learning_rate": 2.1658809522912905e-05,
      "loss": 11.5717,
      "step": 18136
    },
    {
      "epoch": 3.171497267759563,
      "grad_norm": 0.6730477609705878,
      "learning_rate": 2.1650009277218795e-05,
      "loss": 11.2487,
      "step": 18137
    },
    {
      "epoch": 3.171672131147541,
      "grad_norm": 0.6136470643271219,
      "learning_rate": 2.1641210602682637e-05,
      "loss": 11.5825,
      "step": 18138
    },
    {
      "epoch": 3.171846994535519,
      "grad_norm": 0.6560670721453897,
      "learning_rate": 2.1632413499480762e-05,
      "loss": 11.6073,
      "step": 18139
    },
    {
      "epoch": 3.172021857923497,
      "grad_norm": 0.7220664818008988,
      "learning_rate": 2.162361796778962e-05,
      "loss": 11.4169,
      "step": 18140
    },
    {
      "epoch": 3.1721967213114755,
      "grad_norm": 0.6685534688061163,
      "learning_rate": 2.1614824007785582e-05,
      "loss": 11.5135,
      "step": 18141
    },
    {
      "epoch": 3.1723715846994534,
      "grad_norm": 0.6899664011262499,
      "learning_rate": 2.1606031619645005e-05,
      "loss": 11.6085,
      "step": 18142
    },
    {
      "epoch": 3.1725464480874317,
      "grad_norm": 0.7190409131677512,
      "learning_rate": 2.1597240803544182e-05,
      "loss": 11.1576,
      "step": 18143
    },
    {
      "epoch": 3.17272131147541,
      "grad_norm": 0.6443193812863954,
      "learning_rate": 2.1588451559659416e-05,
      "loss": 11.3683,
      "step": 18144
    },
    {
      "epoch": 3.172896174863388,
      "grad_norm": 0.7104054633525488,
      "learning_rate": 2.1579663888166956e-05,
      "loss": 10.9787,
      "step": 18145
    },
    {
      "epoch": 3.1730710382513663,
      "grad_norm": 0.6432956187698434,
      "learning_rate": 2.1570877789243004e-05,
      "loss": 11.2039,
      "step": 18146
    },
    {
      "epoch": 3.173245901639344,
      "grad_norm": 0.6264301333783929,
      "learning_rate": 2.1562093263063777e-05,
      "loss": 11.5017,
      "step": 18147
    },
    {
      "epoch": 3.1734207650273225,
      "grad_norm": 0.684416282156868,
      "learning_rate": 2.1553310309805407e-05,
      "loss": 11.41,
      "step": 18148
    },
    {
      "epoch": 3.1735956284153004,
      "grad_norm": 0.6745441722896802,
      "learning_rate": 2.1544528929644027e-05,
      "loss": 11.3213,
      "step": 18149
    },
    {
      "epoch": 3.1737704918032787,
      "grad_norm": 0.6634259405121882,
      "learning_rate": 2.1535749122755733e-05,
      "loss": 11.5722,
      "step": 18150
    },
    {
      "epoch": 3.1739453551912566,
      "grad_norm": 0.7259372703573795,
      "learning_rate": 2.152697088931659e-05,
      "loss": 11.3018,
      "step": 18151
    },
    {
      "epoch": 3.174120218579235,
      "grad_norm": 0.6939036637899475,
      "learning_rate": 2.1518194229502632e-05,
      "loss": 11.5477,
      "step": 18152
    },
    {
      "epoch": 3.1742950819672133,
      "grad_norm": 0.7429485484874888,
      "learning_rate": 2.150941914348984e-05,
      "loss": 11.3881,
      "step": 18153
    },
    {
      "epoch": 3.1744699453551912,
      "grad_norm": 0.7293220346156175,
      "learning_rate": 2.1500645631454187e-05,
      "loss": 11.3829,
      "step": 18154
    },
    {
      "epoch": 3.1746448087431696,
      "grad_norm": 0.7049046213033453,
      "learning_rate": 2.1491873693571617e-05,
      "loss": 11.4808,
      "step": 18155
    },
    {
      "epoch": 3.1748196721311475,
      "grad_norm": 0.7550688076853438,
      "learning_rate": 2.148310333001804e-05,
      "loss": 11.4778,
      "step": 18156
    },
    {
      "epoch": 3.174994535519126,
      "grad_norm": 0.7219619464872122,
      "learning_rate": 2.147433454096931e-05,
      "loss": 11.4921,
      "step": 18157
    },
    {
      "epoch": 3.1751693989071037,
      "grad_norm": 0.6339415909146194,
      "learning_rate": 2.1465567326601278e-05,
      "loss": 11.3912,
      "step": 18158
    },
    {
      "epoch": 3.175344262295082,
      "grad_norm": 0.6105550533124507,
      "learning_rate": 2.1456801687089756e-05,
      "loss": 11.4733,
      "step": 18159
    },
    {
      "epoch": 3.17551912568306,
      "grad_norm": 0.6717754319197031,
      "learning_rate": 2.1448037622610528e-05,
      "loss": 11.3425,
      "step": 18160
    },
    {
      "epoch": 3.1756939890710383,
      "grad_norm": 0.6999809379723553,
      "learning_rate": 2.143927513333932e-05,
      "loss": 11.5458,
      "step": 18161
    },
    {
      "epoch": 3.1758688524590166,
      "grad_norm": 0.7098631790548346,
      "learning_rate": 2.1430514219451857e-05,
      "loss": 11.4276,
      "step": 18162
    },
    {
      "epoch": 3.1760437158469945,
      "grad_norm": 0.7473598019726366,
      "learning_rate": 2.1421754881123834e-05,
      "loss": 11.5671,
      "step": 18163
    },
    {
      "epoch": 3.176218579234973,
      "grad_norm": 0.6711169793162745,
      "learning_rate": 2.1412997118530885e-05,
      "loss": 11.4064,
      "step": 18164
    },
    {
      "epoch": 3.1763934426229508,
      "grad_norm": 0.6345235979506899,
      "learning_rate": 2.140424093184864e-05,
      "loss": 11.695,
      "step": 18165
    },
    {
      "epoch": 3.176568306010929,
      "grad_norm": 0.710166777885323,
      "learning_rate": 2.139548632125269e-05,
      "loss": 11.2866,
      "step": 18166
    },
    {
      "epoch": 3.176743169398907,
      "grad_norm": 0.6930131917927722,
      "learning_rate": 2.138673328691858e-05,
      "loss": 11.3069,
      "step": 18167
    },
    {
      "epoch": 3.1769180327868853,
      "grad_norm": 0.7757211829107267,
      "learning_rate": 2.137798182902184e-05,
      "loss": 11.4694,
      "step": 18168
    },
    {
      "epoch": 3.1770928961748632,
      "grad_norm": 0.6889474355473028,
      "learning_rate": 2.136923194773798e-05,
      "loss": 11.3762,
      "step": 18169
    },
    {
      "epoch": 3.1772677595628416,
      "grad_norm": 0.71590217014163,
      "learning_rate": 2.1360483643242434e-05,
      "loss": 11.4777,
      "step": 18170
    },
    {
      "epoch": 3.1774426229508195,
      "grad_norm": 0.6704371071791196,
      "learning_rate": 2.1351736915710652e-05,
      "loss": 11.3139,
      "step": 18171
    },
    {
      "epoch": 3.177617486338798,
      "grad_norm": 0.7760272866093332,
      "learning_rate": 2.134299176531802e-05,
      "loss": 11.6368,
      "step": 18172
    },
    {
      "epoch": 3.177792349726776,
      "grad_norm": 0.7180675720932815,
      "learning_rate": 2.1334248192239926e-05,
      "loss": 11.2615,
      "step": 18173
    },
    {
      "epoch": 3.177967213114754,
      "grad_norm": 0.7320174146440367,
      "learning_rate": 2.132550619665168e-05,
      "loss": 11.4892,
      "step": 18174
    },
    {
      "epoch": 3.1781420765027324,
      "grad_norm": 0.6380695180426978,
      "learning_rate": 2.131676577872862e-05,
      "loss": 11.5555,
      "step": 18175
    },
    {
      "epoch": 3.1783169398907103,
      "grad_norm": 0.6803399675657292,
      "learning_rate": 2.130802693864595e-05,
      "loss": 11.4706,
      "step": 18176
    },
    {
      "epoch": 3.1784918032786886,
      "grad_norm": 0.6243449096808481,
      "learning_rate": 2.1299289676578982e-05,
      "loss": 11.3925,
      "step": 18177
    },
    {
      "epoch": 3.1786666666666665,
      "grad_norm": 0.6729412552544098,
      "learning_rate": 2.12905539927029e-05,
      "loss": 11.5483,
      "step": 18178
    },
    {
      "epoch": 3.178841530054645,
      "grad_norm": 0.6489897217068977,
      "learning_rate": 2.1281819887192865e-05,
      "loss": 11.3701,
      "step": 18179
    },
    {
      "epoch": 3.1790163934426228,
      "grad_norm": 0.6247721907432926,
      "learning_rate": 2.1273087360224087e-05,
      "loss": 11.5211,
      "step": 18180
    },
    {
      "epoch": 3.179191256830601,
      "grad_norm": 0.6320512097205742,
      "learning_rate": 2.1264356411971566e-05,
      "loss": 11.6926,
      "step": 18181
    },
    {
      "epoch": 3.179366120218579,
      "grad_norm": 0.6925826844041824,
      "learning_rate": 2.125562704261047e-05,
      "loss": 11.3923,
      "step": 18182
    },
    {
      "epoch": 3.1795409836065573,
      "grad_norm": 0.5952593479188064,
      "learning_rate": 2.1246899252315843e-05,
      "loss": 11.6403,
      "step": 18183
    },
    {
      "epoch": 3.1797158469945357,
      "grad_norm": 0.6891599808289944,
      "learning_rate": 2.123817304126271e-05,
      "loss": 11.3385,
      "step": 18184
    },
    {
      "epoch": 3.1798907103825136,
      "grad_norm": 0.601844464972584,
      "learning_rate": 2.1229448409625985e-05,
      "loss": 11.2684,
      "step": 18185
    },
    {
      "epoch": 3.180065573770492,
      "grad_norm": 0.725705167463798,
      "learning_rate": 2.1220725357580695e-05,
      "loss": 11.345,
      "step": 18186
    },
    {
      "epoch": 3.18024043715847,
      "grad_norm": 0.6608612453647997,
      "learning_rate": 2.1212003885301758e-05,
      "loss": 11.4838,
      "step": 18187
    },
    {
      "epoch": 3.180415300546448,
      "grad_norm": 0.7014651692084978,
      "learning_rate": 2.120328399296405e-05,
      "loss": 11.5446,
      "step": 18188
    },
    {
      "epoch": 3.180590163934426,
      "grad_norm": 0.6635379231589792,
      "learning_rate": 2.1194565680742463e-05,
      "loss": 11.2346,
      "step": 18189
    },
    {
      "epoch": 3.1807650273224044,
      "grad_norm": 0.6363249895126373,
      "learning_rate": 2.1185848948811736e-05,
      "loss": 11.5472,
      "step": 18190
    },
    {
      "epoch": 3.1809398907103823,
      "grad_norm": 0.6855427336988099,
      "learning_rate": 2.117713379734677e-05,
      "loss": 11.5926,
      "step": 18191
    },
    {
      "epoch": 3.1811147540983606,
      "grad_norm": 0.6935130135780725,
      "learning_rate": 2.116842022652228e-05,
      "loss": 11.5031,
      "step": 18192
    },
    {
      "epoch": 3.181289617486339,
      "grad_norm": 0.7044824051715761,
      "learning_rate": 2.1159708236513044e-05,
      "loss": 11.4289,
      "step": 18193
    },
    {
      "epoch": 3.181464480874317,
      "grad_norm": 0.658805893224882,
      "learning_rate": 2.115099782749369e-05,
      "loss": 11.7676,
      "step": 18194
    },
    {
      "epoch": 3.181639344262295,
      "grad_norm": 0.6813641542673045,
      "learning_rate": 2.1142288999638906e-05,
      "loss": 11.4849,
      "step": 18195
    },
    {
      "epoch": 3.181814207650273,
      "grad_norm": 0.6926144222778653,
      "learning_rate": 2.113358175312339e-05,
      "loss": 11.5336,
      "step": 18196
    },
    {
      "epoch": 3.1819890710382515,
      "grad_norm": 0.6630719201860058,
      "learning_rate": 2.1124876088121692e-05,
      "loss": 11.2442,
      "step": 18197
    },
    {
      "epoch": 3.1821639344262294,
      "grad_norm": 0.7690855362819217,
      "learning_rate": 2.1116172004808453e-05,
      "loss": 11.5025,
      "step": 18198
    },
    {
      "epoch": 3.1823387978142077,
      "grad_norm": 0.7160202745494539,
      "learning_rate": 2.1107469503358134e-05,
      "loss": 11.3252,
      "step": 18199
    },
    {
      "epoch": 3.1825136612021856,
      "grad_norm": 0.6295795592494602,
      "learning_rate": 2.109876858394526e-05,
      "loss": 11.4404,
      "step": 18200
    },
    {
      "epoch": 3.182688524590164,
      "grad_norm": 0.7046272868442574,
      "learning_rate": 2.109006924674436e-05,
      "loss": 11.3957,
      "step": 18201
    },
    {
      "epoch": 3.1828633879781423,
      "grad_norm": 0.6994232925339223,
      "learning_rate": 2.108137149192988e-05,
      "loss": 11.6165,
      "step": 18202
    },
    {
      "epoch": 3.18303825136612,
      "grad_norm": 0.6873592959347792,
      "learning_rate": 2.107267531967618e-05,
      "loss": 11.3028,
      "step": 18203
    },
    {
      "epoch": 3.1832131147540985,
      "grad_norm": 0.694545454529193,
      "learning_rate": 2.1063980730157697e-05,
      "loss": 11.6091,
      "step": 18204
    },
    {
      "epoch": 3.1833879781420764,
      "grad_norm": 0.7456481462571151,
      "learning_rate": 2.1055287723548733e-05,
      "loss": 11.1807,
      "step": 18205
    },
    {
      "epoch": 3.1835628415300548,
      "grad_norm": 0.6789409093798355,
      "learning_rate": 2.104659630002366e-05,
      "loss": 11.5628,
      "step": 18206
    },
    {
      "epoch": 3.1837377049180327,
      "grad_norm": 0.6859786291582751,
      "learning_rate": 2.1037906459756785e-05,
      "loss": 11.265,
      "step": 18207
    },
    {
      "epoch": 3.183912568306011,
      "grad_norm": 0.6455532055065748,
      "learning_rate": 2.102921820292231e-05,
      "loss": 11.5232,
      "step": 18208
    },
    {
      "epoch": 3.184087431693989,
      "grad_norm": 0.648886088392338,
      "learning_rate": 2.1020531529694443e-05,
      "loss": 11.3189,
      "step": 18209
    },
    {
      "epoch": 3.1842622950819672,
      "grad_norm": 0.6920651391762288,
      "learning_rate": 2.101184644024745e-05,
      "loss": 11.3172,
      "step": 18210
    },
    {
      "epoch": 3.1844371584699456,
      "grad_norm": 0.7117224509629179,
      "learning_rate": 2.1003162934755483e-05,
      "loss": 11.0236,
      "step": 18211
    },
    {
      "epoch": 3.1846120218579235,
      "grad_norm": 0.664334048784574,
      "learning_rate": 2.099448101339263e-05,
      "loss": 11.3553,
      "step": 18212
    },
    {
      "epoch": 3.184786885245902,
      "grad_norm": 0.6610246237944165,
      "learning_rate": 2.098580067633301e-05,
      "loss": 11.3224,
      "step": 18213
    },
    {
      "epoch": 3.1849617486338797,
      "grad_norm": 0.6122765179509316,
      "learning_rate": 2.0977121923750663e-05,
      "loss": 11.4784,
      "step": 18214
    },
    {
      "epoch": 3.185136612021858,
      "grad_norm": 0.669065539904725,
      "learning_rate": 2.0968444755819684e-05,
      "loss": 11.325,
      "step": 18215
    },
    {
      "epoch": 3.185311475409836,
      "grad_norm": 0.7452467082428548,
      "learning_rate": 2.095976917271407e-05,
      "loss": 11.2213,
      "step": 18216
    },
    {
      "epoch": 3.1854863387978143,
      "grad_norm": 0.6558981702243228,
      "learning_rate": 2.0951095174607748e-05,
      "loss": 11.5798,
      "step": 18217
    },
    {
      "epoch": 3.185661202185792,
      "grad_norm": 0.662344947798294,
      "learning_rate": 2.0942422761674674e-05,
      "loss": 11.4535,
      "step": 18218
    },
    {
      "epoch": 3.1858360655737705,
      "grad_norm": 0.6363223584525945,
      "learning_rate": 2.0933751934088743e-05,
      "loss": 11.5049,
      "step": 18219
    },
    {
      "epoch": 3.186010928961749,
      "grad_norm": 0.6853782557648975,
      "learning_rate": 2.092508269202389e-05,
      "loss": 11.411,
      "step": 18220
    },
    {
      "epoch": 3.1861857923497268,
      "grad_norm": 0.6589358488748334,
      "learning_rate": 2.0916415035653903e-05,
      "loss": 11.6085,
      "step": 18221
    },
    {
      "epoch": 3.186360655737705,
      "grad_norm": 0.6698310704578031,
      "learning_rate": 2.0907748965152618e-05,
      "loss": 11.5658,
      "step": 18222
    },
    {
      "epoch": 3.186535519125683,
      "grad_norm": 0.6303686440404671,
      "learning_rate": 2.089908448069381e-05,
      "loss": 11.4455,
      "step": 18223
    },
    {
      "epoch": 3.1867103825136613,
      "grad_norm": 0.6506085576974853,
      "learning_rate": 2.089042158245119e-05,
      "loss": 11.2423,
      "step": 18224
    },
    {
      "epoch": 3.1868852459016392,
      "grad_norm": 0.6804206227495226,
      "learning_rate": 2.088176027059858e-05,
      "loss": 11.4732,
      "step": 18225
    },
    {
      "epoch": 3.1870601092896176,
      "grad_norm": 0.7071574357717624,
      "learning_rate": 2.0873100545309578e-05,
      "loss": 11.478,
      "step": 18226
    },
    {
      "epoch": 3.1872349726775955,
      "grad_norm": 0.7474538522449854,
      "learning_rate": 2.0864442406757857e-05,
      "loss": 11.5103,
      "step": 18227
    },
    {
      "epoch": 3.187409836065574,
      "grad_norm": 0.7490814917625056,
      "learning_rate": 2.085578585511705e-05,
      "loss": 11.4206,
      "step": 18228
    },
    {
      "epoch": 3.1875846994535517,
      "grad_norm": 0.6715600604000077,
      "learning_rate": 2.0847130890560728e-05,
      "loss": 11.6291,
      "step": 18229
    },
    {
      "epoch": 3.18775956284153,
      "grad_norm": 0.6859261425454632,
      "learning_rate": 2.0838477513262468e-05,
      "loss": 11.3814,
      "step": 18230
    },
    {
      "epoch": 3.1879344262295084,
      "grad_norm": 0.7176731796015771,
      "learning_rate": 2.08298257233958e-05,
      "loss": 11.24,
      "step": 18231
    },
    {
      "epoch": 3.1881092896174863,
      "grad_norm": 0.717676405625713,
      "learning_rate": 2.0821175521134207e-05,
      "loss": 11.3859,
      "step": 18232
    },
    {
      "epoch": 3.1882841530054646,
      "grad_norm": 0.680145594726162,
      "learning_rate": 2.081252690665112e-05,
      "loss": 11.482,
      "step": 18233
    },
    {
      "epoch": 3.1884590163934425,
      "grad_norm": 0.6717083709748886,
      "learning_rate": 2.080387988012007e-05,
      "loss": 11.3628,
      "step": 18234
    },
    {
      "epoch": 3.188633879781421,
      "grad_norm": 0.6445418358136688,
      "learning_rate": 2.0795234441714363e-05,
      "loss": 11.4653,
      "step": 18235
    },
    {
      "epoch": 3.1888087431693988,
      "grad_norm": 0.6645875924751535,
      "learning_rate": 2.078659059160739e-05,
      "loss": 11.4503,
      "step": 18236
    },
    {
      "epoch": 3.188983606557377,
      "grad_norm": 0.6486205079430732,
      "learning_rate": 2.0777948329972497e-05,
      "loss": 11.3903,
      "step": 18237
    },
    {
      "epoch": 3.189158469945355,
      "grad_norm": 0.7398688235141484,
      "learning_rate": 2.0769307656982983e-05,
      "loss": 11.316,
      "step": 18238
    },
    {
      "epoch": 3.1893333333333334,
      "grad_norm": 0.7823296369278077,
      "learning_rate": 2.0760668572812124e-05,
      "loss": 11.2046,
      "step": 18239
    },
    {
      "epoch": 3.1895081967213113,
      "grad_norm": 0.6778399841313993,
      "learning_rate": 2.075203107763315e-05,
      "loss": 11.387,
      "step": 18240
    },
    {
      "epoch": 3.1896830601092896,
      "grad_norm": 0.6775920475784073,
      "learning_rate": 2.0743395171619285e-05,
      "loss": 11.2607,
      "step": 18241
    },
    {
      "epoch": 3.189857923497268,
      "grad_norm": 0.7744259547390406,
      "learning_rate": 2.073476085494369e-05,
      "loss": 11.2571,
      "step": 18242
    },
    {
      "epoch": 3.190032786885246,
      "grad_norm": 0.6618739160265458,
      "learning_rate": 2.072612812777952e-05,
      "loss": 11.5148,
      "step": 18243
    },
    {
      "epoch": 3.190207650273224,
      "grad_norm": 0.6343800386985294,
      "learning_rate": 2.0717496990299878e-05,
      "loss": 11.4329,
      "step": 18244
    },
    {
      "epoch": 3.190382513661202,
      "grad_norm": 0.7089248934872793,
      "learning_rate": 2.070886744267786e-05,
      "loss": 11.6799,
      "step": 18245
    },
    {
      "epoch": 3.1905573770491804,
      "grad_norm": 0.6414465131377314,
      "learning_rate": 2.0700239485086505e-05,
      "loss": 11.5243,
      "step": 18246
    },
    {
      "epoch": 3.1907322404371583,
      "grad_norm": 0.6529288573741442,
      "learning_rate": 2.0691613117698815e-05,
      "loss": 11.7788,
      "step": 18247
    },
    {
      "epoch": 3.1909071038251366,
      "grad_norm": 0.7038106576193283,
      "learning_rate": 2.0682988340687804e-05,
      "loss": 11.3488,
      "step": 18248
    },
    {
      "epoch": 3.1910819672131145,
      "grad_norm": 0.6249877065882762,
      "learning_rate": 2.0674365154226415e-05,
      "loss": 11.5793,
      "step": 18249
    },
    {
      "epoch": 3.191256830601093,
      "grad_norm": 0.7341665123218976,
      "learning_rate": 2.066574355848756e-05,
      "loss": 11.1412,
      "step": 18250
    },
    {
      "epoch": 3.1914316939890712,
      "grad_norm": 0.7487797751265086,
      "learning_rate": 2.065712355364413e-05,
      "loss": 11.5773,
      "step": 18251
    },
    {
      "epoch": 3.191606557377049,
      "grad_norm": 0.6904515653048156,
      "learning_rate": 2.064850513986899e-05,
      "loss": 11.4187,
      "step": 18252
    },
    {
      "epoch": 3.1917814207650275,
      "grad_norm": 0.7040924223660795,
      "learning_rate": 2.0639888317334967e-05,
      "loss": 11.4395,
      "step": 18253
    },
    {
      "epoch": 3.1919562841530054,
      "grad_norm": 0.6930872006228541,
      "learning_rate": 2.0631273086214852e-05,
      "loss": 11.6487,
      "step": 18254
    },
    {
      "epoch": 3.1921311475409837,
      "grad_norm": 0.7228283154631945,
      "learning_rate": 2.06226594466814e-05,
      "loss": 11.3228,
      "step": 18255
    },
    {
      "epoch": 3.1923060109289616,
      "grad_norm": 0.7350756244413215,
      "learning_rate": 2.0614047398907356e-05,
      "loss": 11.3804,
      "step": 18256
    },
    {
      "epoch": 3.19248087431694,
      "grad_norm": 0.7291356271236681,
      "learning_rate": 2.0605436943065405e-05,
      "loss": 11.2609,
      "step": 18257
    },
    {
      "epoch": 3.192655737704918,
      "grad_norm": 0.6926784630336738,
      "learning_rate": 2.0596828079328212e-05,
      "loss": 11.4325,
      "step": 18258
    },
    {
      "epoch": 3.192830601092896,
      "grad_norm": 0.6702545619292848,
      "learning_rate": 2.058822080786842e-05,
      "loss": 11.39,
      "step": 18259
    },
    {
      "epoch": 3.1930054644808745,
      "grad_norm": 0.6334882571312911,
      "learning_rate": 2.057961512885862e-05,
      "loss": 11.4259,
      "step": 18260
    },
    {
      "epoch": 3.1931803278688524,
      "grad_norm": 0.6340429325702185,
      "learning_rate": 2.05710110424714e-05,
      "loss": 11.5009,
      "step": 18261
    },
    {
      "epoch": 3.1933551912568308,
      "grad_norm": 0.6640361188368239,
      "learning_rate": 2.0562408548879274e-05,
      "loss": 11.4458,
      "step": 18262
    },
    {
      "epoch": 3.1935300546448087,
      "grad_norm": 0.685641388633725,
      "learning_rate": 2.0553807648254774e-05,
      "loss": 11.3821,
      "step": 18263
    },
    {
      "epoch": 3.193704918032787,
      "grad_norm": 0.6441969546091613,
      "learning_rate": 2.054520834077036e-05,
      "loss": 11.4347,
      "step": 18264
    },
    {
      "epoch": 3.193879781420765,
      "grad_norm": 0.7563791758147347,
      "learning_rate": 2.0536610626598474e-05,
      "loss": 11.5081,
      "step": 18265
    },
    {
      "epoch": 3.1940546448087432,
      "grad_norm": 0.6679623241803742,
      "learning_rate": 2.052801450591155e-05,
      "loss": 11.5627,
      "step": 18266
    },
    {
      "epoch": 3.194229508196721,
      "grad_norm": 0.74182730456218,
      "learning_rate": 2.0519419978881903e-05,
      "loss": 11.3294,
      "step": 18267
    },
    {
      "epoch": 3.1944043715846995,
      "grad_norm": 0.7404681195400176,
      "learning_rate": 2.0510827045681948e-05,
      "loss": 11.3652,
      "step": 18268
    },
    {
      "epoch": 3.194579234972678,
      "grad_norm": 0.618409821111594,
      "learning_rate": 2.0502235706483975e-05,
      "loss": 11.6287,
      "step": 18269
    },
    {
      "epoch": 3.1947540983606557,
      "grad_norm": 0.7396330701472609,
      "learning_rate": 2.049364596146026e-05,
      "loss": 11.5341,
      "step": 18270
    },
    {
      "epoch": 3.194928961748634,
      "grad_norm": 0.765578679472474,
      "learning_rate": 2.0485057810783092e-05,
      "loss": 11.5504,
      "step": 18271
    },
    {
      "epoch": 3.195103825136612,
      "grad_norm": 0.7198889922804678,
      "learning_rate": 2.0476471254624606e-05,
      "loss": 11.4522,
      "step": 18272
    },
    {
      "epoch": 3.1952786885245903,
      "grad_norm": 0.6765121738911818,
      "learning_rate": 2.046788629315707e-05,
      "loss": 11.5248,
      "step": 18273
    },
    {
      "epoch": 3.195453551912568,
      "grad_norm": 0.6580438598996471,
      "learning_rate": 2.0459302926552605e-05,
      "loss": 11.6373,
      "step": 18274
    },
    {
      "epoch": 3.1956284153005465,
      "grad_norm": 0.6729538561136429,
      "learning_rate": 2.0450721154983365e-05,
      "loss": 11.5071,
      "step": 18275
    },
    {
      "epoch": 3.1958032786885244,
      "grad_norm": 0.6615452110222759,
      "learning_rate": 2.0442140978621384e-05,
      "loss": 11.453,
      "step": 18276
    },
    {
      "epoch": 3.1959781420765028,
      "grad_norm": 0.7278229847588904,
      "learning_rate": 2.043356239763873e-05,
      "loss": 11.4921,
      "step": 18277
    },
    {
      "epoch": 3.1961530054644807,
      "grad_norm": 0.7479519331478228,
      "learning_rate": 2.0424985412207477e-05,
      "loss": 11.4351,
      "step": 18278
    },
    {
      "epoch": 3.196327868852459,
      "grad_norm": 0.6883769725427912,
      "learning_rate": 2.0416410022499588e-05,
      "loss": 11.6528,
      "step": 18279
    },
    {
      "epoch": 3.1965027322404373,
      "grad_norm": 0.7503114767977432,
      "learning_rate": 2.040783622868706e-05,
      "loss": 11.3001,
      "step": 18280
    },
    {
      "epoch": 3.1966775956284152,
      "grad_norm": 0.6802974771247225,
      "learning_rate": 2.0399264030941745e-05,
      "loss": 11.4535,
      "step": 18281
    },
    {
      "epoch": 3.1968524590163936,
      "grad_norm": 0.7371986722654842,
      "learning_rate": 2.0390693429435627e-05,
      "loss": 11.6044,
      "step": 18282
    },
    {
      "epoch": 3.1970273224043715,
      "grad_norm": 0.7755629979101271,
      "learning_rate": 2.038212442434052e-05,
      "loss": 11.5196,
      "step": 18283
    },
    {
      "epoch": 3.19720218579235,
      "grad_norm": 0.6931927013303639,
      "learning_rate": 2.0373557015828315e-05,
      "loss": 11.5097,
      "step": 18284
    },
    {
      "epoch": 3.1973770491803277,
      "grad_norm": 0.688668418798795,
      "learning_rate": 2.036499120407076e-05,
      "loss": 11.4034,
      "step": 18285
    },
    {
      "epoch": 3.197551912568306,
      "grad_norm": 0.6601912321113166,
      "learning_rate": 2.0356426989239596e-05,
      "loss": 11.4641,
      "step": 18286
    },
    {
      "epoch": 3.197726775956284,
      "grad_norm": 0.6994921758194907,
      "learning_rate": 2.0347864371506652e-05,
      "loss": 11.4852,
      "step": 18287
    },
    {
      "epoch": 3.1979016393442623,
      "grad_norm": 0.7595871293760619,
      "learning_rate": 2.033930335104358e-05,
      "loss": 11.3308,
      "step": 18288
    },
    {
      "epoch": 3.1980765027322406,
      "grad_norm": 0.6283626542642583,
      "learning_rate": 2.03307439280221e-05,
      "loss": 11.1642,
      "step": 18289
    },
    {
      "epoch": 3.1982513661202185,
      "grad_norm": 0.659934118812899,
      "learning_rate": 2.0322186102613795e-05,
      "loss": 11.6107,
      "step": 18290
    },
    {
      "epoch": 3.198426229508197,
      "grad_norm": 0.6522415800094735,
      "learning_rate": 2.031362987499027e-05,
      "loss": 11.6332,
      "step": 18291
    },
    {
      "epoch": 3.198601092896175,
      "grad_norm": 0.671284952208065,
      "learning_rate": 2.030507524532317e-05,
      "loss": 11.4413,
      "step": 18292
    },
    {
      "epoch": 3.198775956284153,
      "grad_norm": 0.6630114240780226,
      "learning_rate": 2.029652221378403e-05,
      "loss": 11.4914,
      "step": 18293
    },
    {
      "epoch": 3.198950819672131,
      "grad_norm": 0.7212092203783906,
      "learning_rate": 2.0287970780544307e-05,
      "loss": 11.6356,
      "step": 18294
    },
    {
      "epoch": 3.1991256830601094,
      "grad_norm": 0.6653838308110488,
      "learning_rate": 2.0279420945775528e-05,
      "loss": 11.4936,
      "step": 18295
    },
    {
      "epoch": 3.1993005464480873,
      "grad_norm": 0.6889580382871173,
      "learning_rate": 2.0270872709649102e-05,
      "loss": 11.466,
      "step": 18296
    },
    {
      "epoch": 3.1994754098360656,
      "grad_norm": 0.6447408663461632,
      "learning_rate": 2.02623260723365e-05,
      "loss": 11.2836,
      "step": 18297
    },
    {
      "epoch": 3.1996502732240435,
      "grad_norm": 0.6635651547907081,
      "learning_rate": 2.0253781034009122e-05,
      "loss": 11.461,
      "step": 18298
    },
    {
      "epoch": 3.199825136612022,
      "grad_norm": 0.6660906182266747,
      "learning_rate": 2.024523759483826e-05,
      "loss": 11.3074,
      "step": 18299
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6494830274079937,
      "learning_rate": 2.023669575499526e-05,
      "loss": 11.5131,
      "step": 18300
    },
    {
      "epoch": 3.200174863387978,
      "grad_norm": 0.7219003767240785,
      "learning_rate": 2.0228155514651392e-05,
      "loss": 11.446,
      "step": 18301
    },
    {
      "epoch": 3.2003497267759564,
      "grad_norm": 0.673627201892341,
      "learning_rate": 2.0219616873977986e-05,
      "loss": 11.3103,
      "step": 18302
    },
    {
      "epoch": 3.2005245901639343,
      "grad_norm": 0.6581443890105413,
      "learning_rate": 2.0211079833146197e-05,
      "loss": 11.6182,
      "step": 18303
    },
    {
      "epoch": 3.2006994535519127,
      "grad_norm": 0.6617194256764601,
      "learning_rate": 2.020254439232724e-05,
      "loss": 11.1852,
      "step": 18304
    },
    {
      "epoch": 3.2008743169398906,
      "grad_norm": 0.652071498474875,
      "learning_rate": 2.0194010551692255e-05,
      "loss": 11.4959,
      "step": 18305
    },
    {
      "epoch": 3.201049180327869,
      "grad_norm": 0.6811301597037007,
      "learning_rate": 2.018547831141242e-05,
      "loss": 11.2534,
      "step": 18306
    },
    {
      "epoch": 3.201224043715847,
      "grad_norm": 0.7474447956023816,
      "learning_rate": 2.017694767165883e-05,
      "loss": 11.4054,
      "step": 18307
    },
    {
      "epoch": 3.201398907103825,
      "grad_norm": 0.6952885899650433,
      "learning_rate": 2.0168418632602505e-05,
      "loss": 11.4181,
      "step": 18308
    },
    {
      "epoch": 3.2015737704918035,
      "grad_norm": 0.720862953105837,
      "learning_rate": 2.0159891194414504e-05,
      "loss": 11.3801,
      "step": 18309
    },
    {
      "epoch": 3.2017486338797814,
      "grad_norm": 0.6492009506395213,
      "learning_rate": 2.0151365357265794e-05,
      "loss": 11.5797,
      "step": 18310
    },
    {
      "epoch": 3.2019234972677597,
      "grad_norm": 0.7483264880958819,
      "learning_rate": 2.014284112132744e-05,
      "loss": 11.18,
      "step": 18311
    },
    {
      "epoch": 3.2020983606557376,
      "grad_norm": 0.6485135504803072,
      "learning_rate": 2.0134318486770276e-05,
      "loss": 11.6824,
      "step": 18312
    },
    {
      "epoch": 3.202273224043716,
      "grad_norm": 0.6945674472980536,
      "learning_rate": 2.0125797453765248e-05,
      "loss": 11.6672,
      "step": 18313
    },
    {
      "epoch": 3.202448087431694,
      "grad_norm": 0.6840269993021902,
      "learning_rate": 2.0117278022483234e-05,
      "loss": 11.4363,
      "step": 18314
    },
    {
      "epoch": 3.202622950819672,
      "grad_norm": 0.6995845093525501,
      "learning_rate": 2.010876019309503e-05,
      "loss": 11.6122,
      "step": 18315
    },
    {
      "epoch": 3.20279781420765,
      "grad_norm": 0.6863492021891705,
      "learning_rate": 2.010024396577155e-05,
      "loss": 11.3313,
      "step": 18316
    },
    {
      "epoch": 3.2029726775956284,
      "grad_norm": 0.6774564994764621,
      "learning_rate": 2.0091729340683474e-05,
      "loss": 11.5001,
      "step": 18317
    },
    {
      "epoch": 3.2031475409836068,
      "grad_norm": 0.7183273580653087,
      "learning_rate": 2.0083216318001564e-05,
      "loss": 11.6005,
      "step": 18318
    },
    {
      "epoch": 3.2033224043715847,
      "grad_norm": 0.6794667079025047,
      "learning_rate": 2.0074704897896558e-05,
      "loss": 11.2518,
      "step": 18319
    },
    {
      "epoch": 3.203497267759563,
      "grad_norm": 0.6820068525573743,
      "learning_rate": 2.0066195080539085e-05,
      "loss": 11.3307,
      "step": 18320
    },
    {
      "epoch": 3.203672131147541,
      "grad_norm": 0.6350980213525775,
      "learning_rate": 2.0057686866099877e-05,
      "loss": 11.3801,
      "step": 18321
    },
    {
      "epoch": 3.2038469945355192,
      "grad_norm": 0.7017316743529055,
      "learning_rate": 2.004918025474949e-05,
      "loss": 11.1173,
      "step": 18322
    },
    {
      "epoch": 3.204021857923497,
      "grad_norm": 0.7055541457335262,
      "learning_rate": 2.0040675246658513e-05,
      "loss": 11.3252,
      "step": 18323
    },
    {
      "epoch": 3.2041967213114755,
      "grad_norm": 0.6994272484672909,
      "learning_rate": 2.0032171841997504e-05,
      "loss": 11.2673,
      "step": 18324
    },
    {
      "epoch": 3.2043715846994534,
      "grad_norm": 0.7929456272719027,
      "learning_rate": 2.002367004093699e-05,
      "loss": 11.5389,
      "step": 18325
    },
    {
      "epoch": 3.2045464480874317,
      "grad_norm": 0.6722589769830476,
      "learning_rate": 2.001516984364744e-05,
      "loss": 11.4995,
      "step": 18326
    },
    {
      "epoch": 3.20472131147541,
      "grad_norm": 0.6755822419788172,
      "learning_rate": 2.0006671250299337e-05,
      "loss": 11.4393,
      "step": 18327
    },
    {
      "epoch": 3.204896174863388,
      "grad_norm": 0.7059252426067968,
      "learning_rate": 1.9998174261063073e-05,
      "loss": 11.6132,
      "step": 18328
    },
    {
      "epoch": 3.2050710382513663,
      "grad_norm": 0.6806376050893825,
      "learning_rate": 1.998967887610903e-05,
      "loss": 11.1276,
      "step": 18329
    },
    {
      "epoch": 3.205245901639344,
      "grad_norm": 0.735841307683102,
      "learning_rate": 1.9981185095607648e-05,
      "loss": 11.3197,
      "step": 18330
    },
    {
      "epoch": 3.2054207650273225,
      "grad_norm": 0.6308185273779091,
      "learning_rate": 1.997269291972915e-05,
      "loss": 11.3661,
      "step": 18331
    },
    {
      "epoch": 3.2055956284153004,
      "grad_norm": 0.65161870714719,
      "learning_rate": 1.996420234864389e-05,
      "loss": 11.0224,
      "step": 18332
    },
    {
      "epoch": 3.2057704918032788,
      "grad_norm": 0.7165635961555885,
      "learning_rate": 1.9955713382522113e-05,
      "loss": 11.1385,
      "step": 18333
    },
    {
      "epoch": 3.2059453551912567,
      "grad_norm": 0.6455117946731245,
      "learning_rate": 1.994722602153406e-05,
      "loss": 11.3603,
      "step": 18334
    },
    {
      "epoch": 3.206120218579235,
      "grad_norm": 0.6705415196341068,
      "learning_rate": 1.993874026584991e-05,
      "loss": 11.5171,
      "step": 18335
    },
    {
      "epoch": 3.206295081967213,
      "grad_norm": 0.7182733592578017,
      "learning_rate": 1.9930256115639832e-05,
      "loss": 11.4762,
      "step": 18336
    },
    {
      "epoch": 3.2064699453551913,
      "grad_norm": 0.6361892472122983,
      "learning_rate": 1.9921773571073976e-05,
      "loss": 11.3574,
      "step": 18337
    },
    {
      "epoch": 3.2066448087431696,
      "grad_norm": 0.7011266849327081,
      "learning_rate": 1.9913292632322424e-05,
      "loss": 11.4699,
      "step": 18338
    },
    {
      "epoch": 3.2068196721311475,
      "grad_norm": 0.7125482878419774,
      "learning_rate": 1.9904813299555257e-05,
      "loss": 11.2687,
      "step": 18339
    },
    {
      "epoch": 3.206994535519126,
      "grad_norm": 0.7519723335259242,
      "learning_rate": 1.9896335572942515e-05,
      "loss": 11.2681,
      "step": 18340
    },
    {
      "epoch": 3.2071693989071037,
      "grad_norm": 0.7202463699753421,
      "learning_rate": 1.988785945265419e-05,
      "loss": 11.3799,
      "step": 18341
    },
    {
      "epoch": 3.207344262295082,
      "grad_norm": 0.6504460824157422,
      "learning_rate": 1.9879384938860258e-05,
      "loss": 11.6075,
      "step": 18342
    },
    {
      "epoch": 3.20751912568306,
      "grad_norm": 0.6725707963924042,
      "learning_rate": 1.987091203173066e-05,
      "loss": 11.4953,
      "step": 18343
    },
    {
      "epoch": 3.2076939890710383,
      "grad_norm": 0.6959535693593452,
      "learning_rate": 1.9862440731435306e-05,
      "loss": 11.6554,
      "step": 18344
    },
    {
      "epoch": 3.207868852459016,
      "grad_norm": 0.7041695158865768,
      "learning_rate": 1.985397103814407e-05,
      "loss": 11.1355,
      "step": 18345
    },
    {
      "epoch": 3.2080437158469945,
      "grad_norm": 0.6768461842055341,
      "learning_rate": 1.9845502952026795e-05,
      "loss": 11.0122,
      "step": 18346
    },
    {
      "epoch": 3.2082185792349724,
      "grad_norm": 0.6368060822130559,
      "learning_rate": 1.983703647325329e-05,
      "loss": 11.2913,
      "step": 18347
    },
    {
      "epoch": 3.208393442622951,
      "grad_norm": 0.7198752181926907,
      "learning_rate": 1.982857160199334e-05,
      "loss": 11.2703,
      "step": 18348
    },
    {
      "epoch": 3.208568306010929,
      "grad_norm": 0.6982462390314793,
      "learning_rate": 1.9820108338416675e-05,
      "loss": 11.4687,
      "step": 18349
    },
    {
      "epoch": 3.208743169398907,
      "grad_norm": 0.7002331249129182,
      "learning_rate": 1.9811646682693042e-05,
      "loss": 11.4657,
      "step": 18350
    },
    {
      "epoch": 3.2089180327868854,
      "grad_norm": 0.6851669641871753,
      "learning_rate": 1.98031866349921e-05,
      "loss": 11.2433,
      "step": 18351
    },
    {
      "epoch": 3.2090928961748633,
      "grad_norm": 0.718243296643392,
      "learning_rate": 1.979472819548349e-05,
      "loss": 11.4473,
      "step": 18352
    },
    {
      "epoch": 3.2092677595628416,
      "grad_norm": 0.7101459480467677,
      "learning_rate": 1.978627136433685e-05,
      "loss": 11.437,
      "step": 18353
    },
    {
      "epoch": 3.2094426229508195,
      "grad_norm": 0.7029656540623618,
      "learning_rate": 1.977781614172176e-05,
      "loss": 11.2243,
      "step": 18354
    },
    {
      "epoch": 3.209617486338798,
      "grad_norm": 0.6534964483685428,
      "learning_rate": 1.9769362527807768e-05,
      "loss": 11.3249,
      "step": 18355
    },
    {
      "epoch": 3.2097923497267757,
      "grad_norm": 0.6921483697460823,
      "learning_rate": 1.9760910522764398e-05,
      "loss": 11.2517,
      "step": 18356
    },
    {
      "epoch": 3.209967213114754,
      "grad_norm": 0.6474817861310455,
      "learning_rate": 1.9752460126761142e-05,
      "loss": 11.332,
      "step": 18357
    },
    {
      "epoch": 3.2101420765027324,
      "grad_norm": 0.677031331705404,
      "learning_rate": 1.974401133996745e-05,
      "loss": 11.5558,
      "step": 18358
    },
    {
      "epoch": 3.2103169398907103,
      "grad_norm": 0.5825085575540362,
      "learning_rate": 1.9735564162552744e-05,
      "loss": 11.6147,
      "step": 18359
    },
    {
      "epoch": 3.2104918032786887,
      "grad_norm": 0.6683076791161594,
      "learning_rate": 1.9727118594686422e-05,
      "loss": 11.5675,
      "step": 18360
    },
    {
      "epoch": 3.2106666666666666,
      "grad_norm": 0.7123497325350725,
      "learning_rate": 1.9718674636537847e-05,
      "loss": 11.4404,
      "step": 18361
    },
    {
      "epoch": 3.210841530054645,
      "grad_norm": 0.7190511151535381,
      "learning_rate": 1.9710232288276365e-05,
      "loss": 11.2358,
      "step": 18362
    },
    {
      "epoch": 3.211016393442623,
      "grad_norm": 0.6416657553205853,
      "learning_rate": 1.9701791550071202e-05,
      "loss": 11.5031,
      "step": 18363
    },
    {
      "epoch": 3.211191256830601,
      "grad_norm": 0.6055270799908684,
      "learning_rate": 1.9693352422091682e-05,
      "loss": 11.5801,
      "step": 18364
    },
    {
      "epoch": 3.211366120218579,
      "grad_norm": 0.71114223591245,
      "learning_rate": 1.9684914904507023e-05,
      "loss": 11.2961,
      "step": 18365
    },
    {
      "epoch": 3.2115409836065574,
      "grad_norm": 0.7095669512103554,
      "learning_rate": 1.9676478997486426e-05,
      "loss": 11.3846,
      "step": 18366
    },
    {
      "epoch": 3.2117158469945357,
      "grad_norm": 0.6728616788446472,
      "learning_rate": 1.9668044701199074e-05,
      "loss": 11.382,
      "step": 18367
    },
    {
      "epoch": 3.2118907103825136,
      "grad_norm": 0.6362309306282719,
      "learning_rate": 1.965961201581402e-05,
      "loss": 11.4165,
      "step": 18368
    },
    {
      "epoch": 3.212065573770492,
      "grad_norm": 0.7219958964390456,
      "learning_rate": 1.9651180941500446e-05,
      "loss": 11.0478,
      "step": 18369
    },
    {
      "epoch": 3.21224043715847,
      "grad_norm": 0.6451154075128022,
      "learning_rate": 1.964275147842741e-05,
      "loss": 11.2093,
      "step": 18370
    },
    {
      "epoch": 3.212415300546448,
      "grad_norm": 0.7544538584381881,
      "learning_rate": 1.963432362676395e-05,
      "loss": 11.5401,
      "step": 18371
    },
    {
      "epoch": 3.212590163934426,
      "grad_norm": 0.6778655503395384,
      "learning_rate": 1.9625897386679038e-05,
      "loss": 11.5587,
      "step": 18372
    },
    {
      "epoch": 3.2127650273224044,
      "grad_norm": 0.6608679939787696,
      "learning_rate": 1.961747275834164e-05,
      "loss": 11.6311,
      "step": 18373
    },
    {
      "epoch": 3.2129398907103823,
      "grad_norm": 0.7089126948970556,
      "learning_rate": 1.9609049741920738e-05,
      "loss": 11.4299,
      "step": 18374
    },
    {
      "epoch": 3.2131147540983607,
      "grad_norm": 0.7364260852374688,
      "learning_rate": 1.9600628337585225e-05,
      "loss": 11.6061,
      "step": 18375
    },
    {
      "epoch": 3.213289617486339,
      "grad_norm": 0.7084963641505564,
      "learning_rate": 1.9592208545503998e-05,
      "loss": 11.5698,
      "step": 18376
    },
    {
      "epoch": 3.213464480874317,
      "grad_norm": 0.7024433227580477,
      "learning_rate": 1.9583790365845822e-05,
      "loss": 11.4863,
      "step": 18377
    },
    {
      "epoch": 3.2136393442622952,
      "grad_norm": 0.699191102230123,
      "learning_rate": 1.9575373798779583e-05,
      "loss": 11.4805,
      "step": 18378
    },
    {
      "epoch": 3.213814207650273,
      "grad_norm": 0.6750560935855697,
      "learning_rate": 1.9566958844474036e-05,
      "loss": 11.1456,
      "step": 18379
    },
    {
      "epoch": 3.2139890710382515,
      "grad_norm": 0.6567433760221906,
      "learning_rate": 1.955854550309796e-05,
      "loss": 11.2676,
      "step": 18380
    },
    {
      "epoch": 3.2141639344262294,
      "grad_norm": 0.6954803613269535,
      "learning_rate": 1.9550133774820002e-05,
      "loss": 11.5523,
      "step": 18381
    },
    {
      "epoch": 3.2143387978142077,
      "grad_norm": 0.6688172077819977,
      "learning_rate": 1.9541723659808853e-05,
      "loss": 11.4682,
      "step": 18382
    },
    {
      "epoch": 3.2145136612021856,
      "grad_norm": 0.7511212806695066,
      "learning_rate": 1.95333151582332e-05,
      "loss": 11.4771,
      "step": 18383
    },
    {
      "epoch": 3.214688524590164,
      "grad_norm": 0.6966952392593564,
      "learning_rate": 1.9524908270261656e-05,
      "loss": 11.3724,
      "step": 18384
    },
    {
      "epoch": 3.2148633879781423,
      "grad_norm": 0.6739340228510853,
      "learning_rate": 1.9516502996062815e-05,
      "loss": 11.4418,
      "step": 18385
    },
    {
      "epoch": 3.21503825136612,
      "grad_norm": 0.6894785234112111,
      "learning_rate": 1.9508099335805173e-05,
      "loss": 11.4556,
      "step": 18386
    },
    {
      "epoch": 3.2152131147540985,
      "grad_norm": 0.6859909711513713,
      "learning_rate": 1.949969728965726e-05,
      "loss": 11.609,
      "step": 18387
    },
    {
      "epoch": 3.2153879781420764,
      "grad_norm": 0.5988494914907035,
      "learning_rate": 1.949129685778761e-05,
      "loss": 11.198,
      "step": 18388
    },
    {
      "epoch": 3.215562841530055,
      "grad_norm": 0.6753622891928835,
      "learning_rate": 1.9482898040364685e-05,
      "loss": 11.6253,
      "step": 18389
    },
    {
      "epoch": 3.2157377049180327,
      "grad_norm": 0.6791063105785187,
      "learning_rate": 1.9474500837556842e-05,
      "loss": 11.4343,
      "step": 18390
    },
    {
      "epoch": 3.215912568306011,
      "grad_norm": 0.892205815605738,
      "learning_rate": 1.9466105249532495e-05,
      "loss": 11.2779,
      "step": 18391
    },
    {
      "epoch": 3.216087431693989,
      "grad_norm": 0.6664646570439309,
      "learning_rate": 1.945771127645999e-05,
      "loss": 11.6054,
      "step": 18392
    },
    {
      "epoch": 3.2162622950819673,
      "grad_norm": 0.7545923012043835,
      "learning_rate": 1.9449318918507698e-05,
      "loss": 11.2874,
      "step": 18393
    },
    {
      "epoch": 3.216437158469945,
      "grad_norm": 0.6601412474225611,
      "learning_rate": 1.944092817584392e-05,
      "loss": 11.1904,
      "step": 18394
    },
    {
      "epoch": 3.2166120218579235,
      "grad_norm": 0.7053215424863761,
      "learning_rate": 1.9432539048636843e-05,
      "loss": 11.3897,
      "step": 18395
    },
    {
      "epoch": 3.216786885245902,
      "grad_norm": 0.6508170287772858,
      "learning_rate": 1.9424151537054737e-05,
      "loss": 11.3991,
      "step": 18396
    },
    {
      "epoch": 3.2169617486338797,
      "grad_norm": 0.6903948075143236,
      "learning_rate": 1.941576564126576e-05,
      "loss": 11.4312,
      "step": 18397
    },
    {
      "epoch": 3.217136612021858,
      "grad_norm": 0.6706216103234355,
      "learning_rate": 1.9407381361438172e-05,
      "loss": 11.46,
      "step": 18398
    },
    {
      "epoch": 3.217311475409836,
      "grad_norm": 0.7309102743454087,
      "learning_rate": 1.9398998697740002e-05,
      "loss": 11.5416,
      "step": 18399
    },
    {
      "epoch": 3.2174863387978143,
      "grad_norm": 0.6367872234199208,
      "learning_rate": 1.9390617650339395e-05,
      "loss": 11.3756,
      "step": 18400
    },
    {
      "epoch": 3.217661202185792,
      "grad_norm": 0.678549645287535,
      "learning_rate": 1.938223821940438e-05,
      "loss": 11.4392,
      "step": 18401
    },
    {
      "epoch": 3.2178360655737706,
      "grad_norm": 0.6215755084257641,
      "learning_rate": 1.937386040510304e-05,
      "loss": 11.4571,
      "step": 18402
    },
    {
      "epoch": 3.2180109289617485,
      "grad_norm": 0.7226900808156191,
      "learning_rate": 1.936548420760338e-05,
      "loss": 11.5393,
      "step": 18403
    },
    {
      "epoch": 3.218185792349727,
      "grad_norm": 0.7850078606838085,
      "learning_rate": 1.935710962707331e-05,
      "loss": 11.3659,
      "step": 18404
    },
    {
      "epoch": 3.2183606557377047,
      "grad_norm": 0.5966118480852399,
      "learning_rate": 1.9348736663680813e-05,
      "loss": 11.2995,
      "step": 18405
    },
    {
      "epoch": 3.218535519125683,
      "grad_norm": 0.7167188021290691,
      "learning_rate": 1.9340365317593746e-05,
      "loss": 11.2607,
      "step": 18406
    },
    {
      "epoch": 3.2187103825136614,
      "grad_norm": 0.7441211262513702,
      "learning_rate": 1.9331995588980057e-05,
      "loss": 11.4199,
      "step": 18407
    },
    {
      "epoch": 3.2188852459016393,
      "grad_norm": 0.6293766379497033,
      "learning_rate": 1.9323627478007522e-05,
      "loss": 11.4877,
      "step": 18408
    },
    {
      "epoch": 3.2190601092896176,
      "grad_norm": 0.6496178067563868,
      "learning_rate": 1.9315260984843963e-05,
      "loss": 11.7385,
      "step": 18409
    },
    {
      "epoch": 3.2192349726775955,
      "grad_norm": 0.6742544417966659,
      "learning_rate": 1.9306896109657145e-05,
      "loss": 11.5142,
      "step": 18410
    },
    {
      "epoch": 3.219409836065574,
      "grad_norm": 0.6464181198876839,
      "learning_rate": 1.9298532852614816e-05,
      "loss": 11.5753,
      "step": 18411
    },
    {
      "epoch": 3.2195846994535517,
      "grad_norm": 0.6519887795603411,
      "learning_rate": 1.929017121388472e-05,
      "loss": 11.5889,
      "step": 18412
    },
    {
      "epoch": 3.21975956284153,
      "grad_norm": 0.6509220077078972,
      "learning_rate": 1.9281811193634492e-05,
      "loss": 11.5667,
      "step": 18413
    },
    {
      "epoch": 3.219934426229508,
      "grad_norm": 0.6236598010370634,
      "learning_rate": 1.9273452792031777e-05,
      "loss": 11.3175,
      "step": 18414
    },
    {
      "epoch": 3.2201092896174863,
      "grad_norm": 0.6705334913944644,
      "learning_rate": 1.9265096009244198e-05,
      "loss": 11.2654,
      "step": 18415
    },
    {
      "epoch": 3.2202841530054647,
      "grad_norm": 0.7440970917490084,
      "learning_rate": 1.9256740845439337e-05,
      "loss": 11.5536,
      "step": 18416
    },
    {
      "epoch": 3.2204590163934426,
      "grad_norm": 0.7038356998619381,
      "learning_rate": 1.924838730078474e-05,
      "loss": 11.4253,
      "step": 18417
    },
    {
      "epoch": 3.220633879781421,
      "grad_norm": 0.7074296180406301,
      "learning_rate": 1.9240035375447907e-05,
      "loss": 11.5234,
      "step": 18418
    },
    {
      "epoch": 3.220808743169399,
      "grad_norm": 0.6594199415441624,
      "learning_rate": 1.923168506959634e-05,
      "loss": 11.5193,
      "step": 18419
    },
    {
      "epoch": 3.220983606557377,
      "grad_norm": 0.7810249505190269,
      "learning_rate": 1.922333638339747e-05,
      "loss": 11.4557,
      "step": 18420
    },
    {
      "epoch": 3.221158469945355,
      "grad_norm": 0.6215784433946172,
      "learning_rate": 1.9214989317018728e-05,
      "loss": 11.5292,
      "step": 18421
    },
    {
      "epoch": 3.2213333333333334,
      "grad_norm": 0.7147114020254892,
      "learning_rate": 1.9206643870627495e-05,
      "loss": 11.0629,
      "step": 18422
    },
    {
      "epoch": 3.2215081967213113,
      "grad_norm": 0.6566031737640534,
      "learning_rate": 1.9198300044391115e-05,
      "loss": 11.4604,
      "step": 18423
    },
    {
      "epoch": 3.2216830601092896,
      "grad_norm": 0.7049979070005843,
      "learning_rate": 1.9189957838476925e-05,
      "loss": 11.3573,
      "step": 18424
    },
    {
      "epoch": 3.221857923497268,
      "grad_norm": 0.7226485976979249,
      "learning_rate": 1.9181617253052187e-05,
      "loss": 11.535,
      "step": 18425
    },
    {
      "epoch": 3.222032786885246,
      "grad_norm": 0.6848635258405862,
      "learning_rate": 1.917327828828417e-05,
      "loss": 11.5316,
      "step": 18426
    },
    {
      "epoch": 3.222207650273224,
      "grad_norm": 0.6921279006620388,
      "learning_rate": 1.9164940944340092e-05,
      "loss": 11.4014,
      "step": 18427
    },
    {
      "epoch": 3.222382513661202,
      "grad_norm": 0.7261373072674511,
      "learning_rate": 1.9156605221387147e-05,
      "loss": 11.4109,
      "step": 18428
    },
    {
      "epoch": 3.2225573770491804,
      "grad_norm": 0.6840829743078435,
      "learning_rate": 1.9148271119592485e-05,
      "loss": 11.4473,
      "step": 18429
    },
    {
      "epoch": 3.2227322404371583,
      "grad_norm": 0.7053309318314023,
      "learning_rate": 1.9139938639123245e-05,
      "loss": 11.6659,
      "step": 18430
    },
    {
      "epoch": 3.2229071038251367,
      "grad_norm": 0.6809824171879749,
      "learning_rate": 1.91316077801465e-05,
      "loss": 11.3279,
      "step": 18431
    },
    {
      "epoch": 3.2230819672131146,
      "grad_norm": 0.8231204747154512,
      "learning_rate": 1.9123278542829325e-05,
      "loss": 11.4265,
      "step": 18432
    },
    {
      "epoch": 3.223256830601093,
      "grad_norm": 0.6738225469487797,
      "learning_rate": 1.9114950927338737e-05,
      "loss": 11.5031,
      "step": 18433
    },
    {
      "epoch": 3.2234316939890713,
      "grad_norm": 0.7279443141560475,
      "learning_rate": 1.9106624933841722e-05,
      "loss": 11.2672,
      "step": 18434
    },
    {
      "epoch": 3.223606557377049,
      "grad_norm": 0.6847333671594915,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 11.4468,
      "step": 18435
    },
    {
      "epoch": 3.2237814207650275,
      "grad_norm": 0.7164888378392905,
      "learning_rate": 1.9089977813496272e-05,
      "loss": 11.5681,
      "step": 18436
    },
    {
      "epoch": 3.2239562841530054,
      "grad_norm": 0.6649796523266691,
      "learning_rate": 1.9081656686981664e-05,
      "loss": 11.3136,
      "step": 18437
    },
    {
      "epoch": 3.2241311475409837,
      "grad_norm": 0.6404618091269391,
      "learning_rate": 1.9073337183128283e-05,
      "loss": 11.2066,
      "step": 18438
    },
    {
      "epoch": 3.2243060109289616,
      "grad_norm": 0.6143506734709266,
      "learning_rate": 1.906501930210297e-05,
      "loss": 11.6347,
      "step": 18439
    },
    {
      "epoch": 3.22448087431694,
      "grad_norm": 0.6640999137291675,
      "learning_rate": 1.9056703044072522e-05,
      "loss": 11.3234,
      "step": 18440
    },
    {
      "epoch": 3.224655737704918,
      "grad_norm": 0.6901267092192701,
      "learning_rate": 1.90483884092037e-05,
      "loss": 11.3146,
      "step": 18441
    },
    {
      "epoch": 3.224830601092896,
      "grad_norm": 0.6703014243705924,
      "learning_rate": 1.904007539766326e-05,
      "loss": 11.4854,
      "step": 18442
    },
    {
      "epoch": 3.2250054644808746,
      "grad_norm": 0.6685060996112674,
      "learning_rate": 1.903176400961788e-05,
      "loss": 11.3263,
      "step": 18443
    },
    {
      "epoch": 3.2251803278688524,
      "grad_norm": 0.6950517109527913,
      "learning_rate": 1.902345424523423e-05,
      "loss": 11.6727,
      "step": 18444
    },
    {
      "epoch": 3.225355191256831,
      "grad_norm": 0.6555787313763444,
      "learning_rate": 1.901514610467896e-05,
      "loss": 11.7127,
      "step": 18445
    },
    {
      "epoch": 3.2255300546448087,
      "grad_norm": 0.6605990332253149,
      "learning_rate": 1.9006839588118664e-05,
      "loss": 11.7032,
      "step": 18446
    },
    {
      "epoch": 3.225704918032787,
      "grad_norm": 0.6870106552893871,
      "learning_rate": 1.8998534695719917e-05,
      "loss": 11.7697,
      "step": 18447
    },
    {
      "epoch": 3.225879781420765,
      "grad_norm": 0.6324234203118286,
      "learning_rate": 1.8990231427649264e-05,
      "loss": 11.5245,
      "step": 18448
    },
    {
      "epoch": 3.2260546448087433,
      "grad_norm": 0.6566515885350009,
      "learning_rate": 1.8981929784073193e-05,
      "loss": 11.4058,
      "step": 18449
    },
    {
      "epoch": 3.226229508196721,
      "grad_norm": 0.7113955099650304,
      "learning_rate": 1.8973629765158197e-05,
      "loss": 11.2908,
      "step": 18450
    },
    {
      "epoch": 3.2264043715846995,
      "grad_norm": 0.6222205988423071,
      "learning_rate": 1.8965331371070695e-05,
      "loss": 11.422,
      "step": 18451
    },
    {
      "epoch": 3.2265792349726774,
      "grad_norm": 0.681362110638737,
      "learning_rate": 1.8957034601977118e-05,
      "loss": 11.4291,
      "step": 18452
    },
    {
      "epoch": 3.2267540983606557,
      "grad_norm": 0.6581730708736877,
      "learning_rate": 1.894873945804383e-05,
      "loss": 11.5093,
      "step": 18453
    },
    {
      "epoch": 3.226928961748634,
      "grad_norm": 0.667128498377112,
      "learning_rate": 1.894044593943718e-05,
      "loss": 11.2957,
      "step": 18454
    },
    {
      "epoch": 3.227103825136612,
      "grad_norm": 0.7181697785628012,
      "learning_rate": 1.893215404632347e-05,
      "loss": 11.0257,
      "step": 18455
    },
    {
      "epoch": 3.2272786885245903,
      "grad_norm": 0.6800015078389297,
      "learning_rate": 1.8923863778868976e-05,
      "loss": 11.746,
      "step": 18456
    },
    {
      "epoch": 3.227453551912568,
      "grad_norm": 0.6840864694594101,
      "learning_rate": 1.891557513723996e-05,
      "loss": 11.0905,
      "step": 18457
    },
    {
      "epoch": 3.2276284153005466,
      "grad_norm": 0.7653923932504966,
      "learning_rate": 1.890728812160264e-05,
      "loss": 11.0739,
      "step": 18458
    },
    {
      "epoch": 3.2278032786885245,
      "grad_norm": 0.6572668143550124,
      "learning_rate": 1.8899002732123138e-05,
      "loss": 11.435,
      "step": 18459
    },
    {
      "epoch": 3.227978142076503,
      "grad_norm": 0.7450923795759821,
      "learning_rate": 1.8890718968967668e-05,
      "loss": 11.5131,
      "step": 18460
    },
    {
      "epoch": 3.2281530054644807,
      "grad_norm": 0.6300448870414798,
      "learning_rate": 1.888243683230232e-05,
      "loss": 11.6311,
      "step": 18461
    },
    {
      "epoch": 3.228327868852459,
      "grad_norm": 0.7053311282720381,
      "learning_rate": 1.887415632229318e-05,
      "loss": 11.3033,
      "step": 18462
    },
    {
      "epoch": 3.228502732240437,
      "grad_norm": 0.6676630811747691,
      "learning_rate": 1.8865877439106317e-05,
      "loss": 11.4873,
      "step": 18463
    },
    {
      "epoch": 3.2286775956284153,
      "grad_norm": 0.6396617169879879,
      "learning_rate": 1.8857600182907675e-05,
      "loss": 11.3844,
      "step": 18464
    },
    {
      "epoch": 3.2288524590163936,
      "grad_norm": 0.6533982599981537,
      "learning_rate": 1.8849324553863313e-05,
      "loss": 11.1818,
      "step": 18465
    },
    {
      "epoch": 3.2290273224043715,
      "grad_norm": 0.6788175898692115,
      "learning_rate": 1.884105055213916e-05,
      "loss": 11.3923,
      "step": 18466
    },
    {
      "epoch": 3.22920218579235,
      "grad_norm": 0.6896884364328048,
      "learning_rate": 1.8832778177901154e-05,
      "loss": 11.4529,
      "step": 18467
    },
    {
      "epoch": 3.2293770491803278,
      "grad_norm": 0.7062200450421549,
      "learning_rate": 1.882450743131515e-05,
      "loss": 11.3063,
      "step": 18468
    },
    {
      "epoch": 3.229551912568306,
      "grad_norm": 0.6255041746247957,
      "learning_rate": 1.8816238312546974e-05,
      "loss": 11.4832,
      "step": 18469
    },
    {
      "epoch": 3.229726775956284,
      "grad_norm": 0.6749913790849772,
      "learning_rate": 1.880797082176252e-05,
      "loss": 11.4932,
      "step": 18470
    },
    {
      "epoch": 3.2299016393442623,
      "grad_norm": 0.6939513253248578,
      "learning_rate": 1.879970495912755e-05,
      "loss": 11.4855,
      "step": 18471
    },
    {
      "epoch": 3.2300765027322402,
      "grad_norm": 0.6289192795766596,
      "learning_rate": 1.8791440724807828e-05,
      "loss": 11.3861,
      "step": 18472
    },
    {
      "epoch": 3.2302513661202186,
      "grad_norm": 0.752803107020367,
      "learning_rate": 1.8783178118969013e-05,
      "loss": 11.4051,
      "step": 18473
    },
    {
      "epoch": 3.230426229508197,
      "grad_norm": 0.6908673537549626,
      "learning_rate": 1.877491714177687e-05,
      "loss": 11.5231,
      "step": 18474
    },
    {
      "epoch": 3.230601092896175,
      "grad_norm": 0.6873172112657511,
      "learning_rate": 1.8766657793397036e-05,
      "loss": 11.6628,
      "step": 18475
    },
    {
      "epoch": 3.230775956284153,
      "grad_norm": 0.7512135051302395,
      "learning_rate": 1.875840007399515e-05,
      "loss": 11.2679,
      "step": 18476
    },
    {
      "epoch": 3.230950819672131,
      "grad_norm": 0.6712736670625588,
      "learning_rate": 1.8750143983736766e-05,
      "loss": 11.2024,
      "step": 18477
    },
    {
      "epoch": 3.2311256830601094,
      "grad_norm": 0.701524397522116,
      "learning_rate": 1.874188952278745e-05,
      "loss": 11.3023,
      "step": 18478
    },
    {
      "epoch": 3.2313005464480873,
      "grad_norm": 0.7002772832313817,
      "learning_rate": 1.8733636691312762e-05,
      "loss": 11.5618,
      "step": 18479
    },
    {
      "epoch": 3.2314754098360656,
      "grad_norm": 0.8226860028313066,
      "learning_rate": 1.8725385489478176e-05,
      "loss": 11.4818,
      "step": 18480
    },
    {
      "epoch": 3.2316502732240435,
      "grad_norm": 0.6643435757077353,
      "learning_rate": 1.871713591744918e-05,
      "loss": 11.5647,
      "step": 18481
    },
    {
      "epoch": 3.231825136612022,
      "grad_norm": 0.7986837617668949,
      "learning_rate": 1.8708887975391155e-05,
      "loss": 11.3459,
      "step": 18482
    },
    {
      "epoch": 3.232,
      "grad_norm": 0.6371109148798757,
      "learning_rate": 1.8700641663469497e-05,
      "loss": 11.3089,
      "step": 18483
    },
    {
      "epoch": 3.232174863387978,
      "grad_norm": 0.6159510188920078,
      "learning_rate": 1.8692396981849635e-05,
      "loss": 11.5706,
      "step": 18484
    },
    {
      "epoch": 3.2323497267759564,
      "grad_norm": 0.6821504488621466,
      "learning_rate": 1.868415393069687e-05,
      "loss": 11.3472,
      "step": 18485
    },
    {
      "epoch": 3.2325245901639343,
      "grad_norm": 0.6837119688942741,
      "learning_rate": 1.8675912510176473e-05,
      "loss": 11.5671,
      "step": 18486
    },
    {
      "epoch": 3.2326994535519127,
      "grad_norm": 0.6787596536999281,
      "learning_rate": 1.866767272045372e-05,
      "loss": 11.27,
      "step": 18487
    },
    {
      "epoch": 3.2328743169398906,
      "grad_norm": 0.7136375875201489,
      "learning_rate": 1.8659434561693835e-05,
      "loss": 11.3055,
      "step": 18488
    },
    {
      "epoch": 3.233049180327869,
      "grad_norm": 0.6536485006741727,
      "learning_rate": 1.8651198034062058e-05,
      "loss": 11.5393,
      "step": 18489
    },
    {
      "epoch": 3.233224043715847,
      "grad_norm": 0.7474047074568677,
      "learning_rate": 1.8642963137723556e-05,
      "loss": 11.4966,
      "step": 18490
    },
    {
      "epoch": 3.233398907103825,
      "grad_norm": 0.7210622656344604,
      "learning_rate": 1.8634729872843426e-05,
      "loss": 11.1729,
      "step": 18491
    },
    {
      "epoch": 3.2335737704918035,
      "grad_norm": 0.6276304692552378,
      "learning_rate": 1.8626498239586776e-05,
      "loss": 11.3496,
      "step": 18492
    },
    {
      "epoch": 3.2337486338797814,
      "grad_norm": 0.6678746260435521,
      "learning_rate": 1.8618268238118675e-05,
      "loss": 11.4956,
      "step": 18493
    },
    {
      "epoch": 3.2339234972677597,
      "grad_norm": 0.650046871501226,
      "learning_rate": 1.86100398686042e-05,
      "loss": 11.4339,
      "step": 18494
    },
    {
      "epoch": 3.2340983606557376,
      "grad_norm": 0.6076962230999963,
      "learning_rate": 1.860181313120831e-05,
      "loss": 11.5831,
      "step": 18495
    },
    {
      "epoch": 3.234273224043716,
      "grad_norm": 0.7605971277016744,
      "learning_rate": 1.8593588026096e-05,
      "loss": 11.5496,
      "step": 18496
    },
    {
      "epoch": 3.234448087431694,
      "grad_norm": 0.6265796936775716,
      "learning_rate": 1.8585364553432173e-05,
      "loss": 11.4971,
      "step": 18497
    },
    {
      "epoch": 3.234622950819672,
      "grad_norm": 0.6676658678310456,
      "learning_rate": 1.857714271338178e-05,
      "loss": 11.4046,
      "step": 18498
    },
    {
      "epoch": 3.23479781420765,
      "grad_norm": 0.6956962247655567,
      "learning_rate": 1.856892250610971e-05,
      "loss": 11.5228,
      "step": 18499
    },
    {
      "epoch": 3.2349726775956285,
      "grad_norm": 0.8180819844607056,
      "learning_rate": 1.8560703931780733e-05,
      "loss": 11.4123,
      "step": 18500
    },
    {
      "epoch": 3.2351475409836064,
      "grad_norm": 0.6722302781241417,
      "learning_rate": 1.8552486990559703e-05,
      "loss": 11.4735,
      "step": 18501
    },
    {
      "epoch": 3.2353224043715847,
      "grad_norm": 0.6517022210345573,
      "learning_rate": 1.854427168261136e-05,
      "loss": 11.5283,
      "step": 18502
    },
    {
      "epoch": 3.235497267759563,
      "grad_norm": 0.6314425861959088,
      "learning_rate": 1.853605800810052e-05,
      "loss": 11.3769,
      "step": 18503
    },
    {
      "epoch": 3.235672131147541,
      "grad_norm": 0.6982326606027754,
      "learning_rate": 1.8527845967191813e-05,
      "loss": 11.5381,
      "step": 18504
    },
    {
      "epoch": 3.2358469945355193,
      "grad_norm": 0.7105835892417648,
      "learning_rate": 1.8519635560049952e-05,
      "loss": 11.3026,
      "step": 18505
    },
    {
      "epoch": 3.236021857923497,
      "grad_norm": 0.6578842398864154,
      "learning_rate": 1.8511426786839582e-05,
      "loss": 11.3938,
      "step": 18506
    },
    {
      "epoch": 3.2361967213114755,
      "grad_norm": 0.7026089065226306,
      "learning_rate": 1.850321964772528e-05,
      "loss": 11.036,
      "step": 18507
    },
    {
      "epoch": 3.2363715846994534,
      "grad_norm": 0.6697042211679032,
      "learning_rate": 1.8495014142871693e-05,
      "loss": 10.9976,
      "step": 18508
    },
    {
      "epoch": 3.2365464480874317,
      "grad_norm": 0.6038637737200893,
      "learning_rate": 1.848681027244331e-05,
      "loss": 11.3612,
      "step": 18509
    },
    {
      "epoch": 3.2367213114754096,
      "grad_norm": 0.65374789127881,
      "learning_rate": 1.847860803660465e-05,
      "loss": 11.5455,
      "step": 18510
    },
    {
      "epoch": 3.236896174863388,
      "grad_norm": 0.7133100131929345,
      "learning_rate": 1.847040743552021e-05,
      "loss": 11.4139,
      "step": 18511
    },
    {
      "epoch": 3.2370710382513663,
      "grad_norm": 0.6612523606043222,
      "learning_rate": 1.8462208469354437e-05,
      "loss": 11.2522,
      "step": 18512
    },
    {
      "epoch": 3.2372459016393442,
      "grad_norm": 0.7373015554420087,
      "learning_rate": 1.845401113827172e-05,
      "loss": 11.4953,
      "step": 18513
    },
    {
      "epoch": 3.2374207650273226,
      "grad_norm": 0.6970671605107842,
      "learning_rate": 1.8445815442436476e-05,
      "loss": 11.2709,
      "step": 18514
    },
    {
      "epoch": 3.2375956284153005,
      "grad_norm": 0.6854978289270255,
      "learning_rate": 1.8437621382013027e-05,
      "loss": 11.4479,
      "step": 18515
    },
    {
      "epoch": 3.237770491803279,
      "grad_norm": 0.7745986375974354,
      "learning_rate": 1.8429428957165696e-05,
      "loss": 11.5372,
      "step": 18516
    },
    {
      "epoch": 3.2379453551912567,
      "grad_norm": 0.6916223666981814,
      "learning_rate": 1.8421238168058773e-05,
      "loss": 11.455,
      "step": 18517
    },
    {
      "epoch": 3.238120218579235,
      "grad_norm": 0.7307117524859663,
      "learning_rate": 1.841304901485651e-05,
      "loss": 11.4713,
      "step": 18518
    },
    {
      "epoch": 3.238295081967213,
      "grad_norm": 0.679296627157282,
      "learning_rate": 1.8404861497723113e-05,
      "loss": 11.4504,
      "step": 18519
    },
    {
      "epoch": 3.2384699453551913,
      "grad_norm": 0.6721560846214575,
      "learning_rate": 1.8396675616822766e-05,
      "loss": 11.633,
      "step": 18520
    },
    {
      "epoch": 3.238644808743169,
      "grad_norm": 0.7318721938433894,
      "learning_rate": 1.8388491372319638e-05,
      "loss": 11.6913,
      "step": 18521
    },
    {
      "epoch": 3.2388196721311475,
      "grad_norm": 0.6485687883140246,
      "learning_rate": 1.8380308764377842e-05,
      "loss": 11.4749,
      "step": 18522
    },
    {
      "epoch": 3.238994535519126,
      "grad_norm": 0.6695375190041666,
      "learning_rate": 1.8372127793161452e-05,
      "loss": 11.4489,
      "step": 18523
    },
    {
      "epoch": 3.2391693989071038,
      "grad_norm": 0.6609548089331546,
      "learning_rate": 1.8363948458834536e-05,
      "loss": 11.2937,
      "step": 18524
    },
    {
      "epoch": 3.239344262295082,
      "grad_norm": 0.6093946855135155,
      "learning_rate": 1.8355770761561098e-05,
      "loss": 11.2996,
      "step": 18525
    },
    {
      "epoch": 3.23951912568306,
      "grad_norm": 0.6883563779898382,
      "learning_rate": 1.8347594701505146e-05,
      "loss": 11.4138,
      "step": 18526
    },
    {
      "epoch": 3.2396939890710383,
      "grad_norm": 0.7041606347601088,
      "learning_rate": 1.8339420278830633e-05,
      "loss": 11.4888,
      "step": 18527
    },
    {
      "epoch": 3.2398688524590162,
      "grad_norm": 0.689771075818869,
      "learning_rate": 1.8331247493701465e-05,
      "loss": 11.5726,
      "step": 18528
    },
    {
      "epoch": 3.2400437158469946,
      "grad_norm": 0.6739186724073092,
      "learning_rate": 1.832307634628154e-05,
      "loss": 11.4141,
      "step": 18529
    },
    {
      "epoch": 3.2402185792349725,
      "grad_norm": 0.631239348049243,
      "learning_rate": 1.8314906836734713e-05,
      "loss": 11.554,
      "step": 18530
    },
    {
      "epoch": 3.240393442622951,
      "grad_norm": 0.6667697640148296,
      "learning_rate": 1.8306738965224814e-05,
      "loss": 11.3757,
      "step": 18531
    },
    {
      "epoch": 3.240568306010929,
      "grad_norm": 0.6603165430145909,
      "learning_rate": 1.829857273191563e-05,
      "loss": 11.4856,
      "step": 18532
    },
    {
      "epoch": 3.240743169398907,
      "grad_norm": 0.6246999942394662,
      "learning_rate": 1.8290408136970925e-05,
      "loss": 11.5344,
      "step": 18533
    },
    {
      "epoch": 3.2409180327868854,
      "grad_norm": 0.7060345479642798,
      "learning_rate": 1.8282245180554413e-05,
      "loss": 11.2522,
      "step": 18534
    },
    {
      "epoch": 3.2410928961748633,
      "grad_norm": 0.7003244053750838,
      "learning_rate": 1.8274083862829795e-05,
      "loss": 11.4645,
      "step": 18535
    },
    {
      "epoch": 3.2412677595628416,
      "grad_norm": 0.6714310719123375,
      "learning_rate": 1.8265924183960716e-05,
      "loss": 11.1707,
      "step": 18536
    },
    {
      "epoch": 3.2414426229508195,
      "grad_norm": 0.7873918176503278,
      "learning_rate": 1.825776614411082e-05,
      "loss": 11.346,
      "step": 18537
    },
    {
      "epoch": 3.241617486338798,
      "grad_norm": 0.7180240338047326,
      "learning_rate": 1.8249609743443698e-05,
      "loss": 11.3032,
      "step": 18538
    },
    {
      "epoch": 3.2417923497267758,
      "grad_norm": 0.6699900663975563,
      "learning_rate": 1.8241454982122907e-05,
      "loss": 11.4703,
      "step": 18539
    },
    {
      "epoch": 3.241967213114754,
      "grad_norm": 0.7377615538157283,
      "learning_rate": 1.8233301860311967e-05,
      "loss": 11.4841,
      "step": 18540
    },
    {
      "epoch": 3.2421420765027325,
      "grad_norm": 0.6915658000511091,
      "learning_rate": 1.8225150378174393e-05,
      "loss": 11.5335,
      "step": 18541
    },
    {
      "epoch": 3.2423169398907103,
      "grad_norm": 0.6809400271734299,
      "learning_rate": 1.821700053587363e-05,
      "loss": 11.1668,
      "step": 18542
    },
    {
      "epoch": 3.2424918032786887,
      "grad_norm": 0.706906918802161,
      "learning_rate": 1.820885233357311e-05,
      "loss": 11.5904,
      "step": 18543
    },
    {
      "epoch": 3.2426666666666666,
      "grad_norm": 0.7429468979950111,
      "learning_rate": 1.8200705771436233e-05,
      "loss": 11.4376,
      "step": 18544
    },
    {
      "epoch": 3.242841530054645,
      "grad_norm": 0.6661838700710521,
      "learning_rate": 1.8192560849626372e-05,
      "loss": 11.2194,
      "step": 18545
    },
    {
      "epoch": 3.243016393442623,
      "grad_norm": 0.661757595404119,
      "learning_rate": 1.8184417568306834e-05,
      "loss": 11.3462,
      "step": 18546
    },
    {
      "epoch": 3.243191256830601,
      "grad_norm": 0.6879963492477605,
      "learning_rate": 1.817627592764094e-05,
      "loss": 11.4939,
      "step": 18547
    },
    {
      "epoch": 3.243366120218579,
      "grad_norm": 0.7088854860535629,
      "learning_rate": 1.816813592779194e-05,
      "loss": 11.229,
      "step": 18548
    },
    {
      "epoch": 3.2435409836065574,
      "grad_norm": 0.6876374938086774,
      "learning_rate": 1.8159997568923083e-05,
      "loss": 11.1733,
      "step": 18549
    },
    {
      "epoch": 3.2437158469945357,
      "grad_norm": 0.6601787959050552,
      "learning_rate": 1.8151860851197543e-05,
      "loss": 11.4943,
      "step": 18550
    },
    {
      "epoch": 3.2438907103825136,
      "grad_norm": 0.6915870405367227,
      "learning_rate": 1.8143725774778508e-05,
      "loss": 11.1699,
      "step": 18551
    },
    {
      "epoch": 3.244065573770492,
      "grad_norm": 0.6061138288295475,
      "learning_rate": 1.8135592339829098e-05,
      "loss": 11.7814,
      "step": 18552
    },
    {
      "epoch": 3.24424043715847,
      "grad_norm": 0.7046664810253639,
      "learning_rate": 1.812746054651242e-05,
      "loss": 11.3913,
      "step": 18553
    },
    {
      "epoch": 3.244415300546448,
      "grad_norm": 0.6943411060222069,
      "learning_rate": 1.811933039499156e-05,
      "loss": 11.6089,
      "step": 18554
    },
    {
      "epoch": 3.244590163934426,
      "grad_norm": 0.6820589218290458,
      "learning_rate": 1.8111201885429486e-05,
      "loss": 11.0772,
      "step": 18555
    },
    {
      "epoch": 3.2447650273224045,
      "grad_norm": 0.6650669601190057,
      "learning_rate": 1.810307501798927e-05,
      "loss": 11.534,
      "step": 18556
    },
    {
      "epoch": 3.2449398907103824,
      "grad_norm": 0.6971242712779762,
      "learning_rate": 1.809494979283386e-05,
      "loss": 11.2644,
      "step": 18557
    },
    {
      "epoch": 3.2451147540983607,
      "grad_norm": 0.7047259294201085,
      "learning_rate": 1.8086826210126183e-05,
      "loss": 11.5998,
      "step": 18558
    },
    {
      "epoch": 3.2452896174863386,
      "grad_norm": 0.6601773045932833,
      "learning_rate": 1.807870427002918e-05,
      "loss": 11.4669,
      "step": 18559
    },
    {
      "epoch": 3.245464480874317,
      "grad_norm": 0.6741677555286999,
      "learning_rate": 1.8070583972705636e-05,
      "loss": 11.7511,
      "step": 18560
    },
    {
      "epoch": 3.2456393442622953,
      "grad_norm": 0.6668897876133552,
      "learning_rate": 1.8062465318318454e-05,
      "loss": 11.5585,
      "step": 18561
    },
    {
      "epoch": 3.245814207650273,
      "grad_norm": 0.7033541732537911,
      "learning_rate": 1.805434830703043e-05,
      "loss": 11.3854,
      "step": 18562
    },
    {
      "epoch": 3.2459890710382515,
      "grad_norm": 0.6833106501637588,
      "learning_rate": 1.8046232939004358e-05,
      "loss": 11.6578,
      "step": 18563
    },
    {
      "epoch": 3.2461639344262294,
      "grad_norm": 0.6830261526558686,
      "learning_rate": 1.8038119214402926e-05,
      "loss": 11.4645,
      "step": 18564
    },
    {
      "epoch": 3.2463387978142078,
      "grad_norm": 0.6566923001208098,
      "learning_rate": 1.8030007133388825e-05,
      "loss": 11.4206,
      "step": 18565
    },
    {
      "epoch": 3.2465136612021857,
      "grad_norm": 0.6784250044434237,
      "learning_rate": 1.8021896696124785e-05,
      "loss": 11.4106,
      "step": 18566
    },
    {
      "epoch": 3.246688524590164,
      "grad_norm": 0.6994180537369754,
      "learning_rate": 1.8013787902773426e-05,
      "loss": 11.6718,
      "step": 18567
    },
    {
      "epoch": 3.246863387978142,
      "grad_norm": 0.6672782391457021,
      "learning_rate": 1.8005680753497377e-05,
      "loss": 11.5074,
      "step": 18568
    },
    {
      "epoch": 3.2470382513661202,
      "grad_norm": 0.7044978772364643,
      "learning_rate": 1.799757524845913e-05,
      "loss": 11.2719,
      "step": 18569
    },
    {
      "epoch": 3.247213114754098,
      "grad_norm": 0.642340098064921,
      "learning_rate": 1.798947138782131e-05,
      "loss": 11.6735,
      "step": 18570
    },
    {
      "epoch": 3.2473879781420765,
      "grad_norm": 0.6475388521108472,
      "learning_rate": 1.7981369171746398e-05,
      "loss": 11.4908,
      "step": 18571
    },
    {
      "epoch": 3.247562841530055,
      "grad_norm": 0.6766025812684561,
      "learning_rate": 1.7973268600396887e-05,
      "loss": 11.1901,
      "step": 18572
    },
    {
      "epoch": 3.2477377049180327,
      "grad_norm": 0.7386770681106553,
      "learning_rate": 1.796516967393517e-05,
      "loss": 11.2454,
      "step": 18573
    },
    {
      "epoch": 3.247912568306011,
      "grad_norm": 0.6771403769146704,
      "learning_rate": 1.7957072392523644e-05,
      "loss": 11.366,
      "step": 18574
    },
    {
      "epoch": 3.248087431693989,
      "grad_norm": 0.7124639272491651,
      "learning_rate": 1.7948976756324753e-05,
      "loss": 11.3841,
      "step": 18575
    },
    {
      "epoch": 3.2482622950819673,
      "grad_norm": 0.7262699646984648,
      "learning_rate": 1.7940882765500812e-05,
      "loss": 11.6423,
      "step": 18576
    },
    {
      "epoch": 3.248437158469945,
      "grad_norm": 0.6758146257631672,
      "learning_rate": 1.793279042021415e-05,
      "loss": 11.4137,
      "step": 18577
    },
    {
      "epoch": 3.2486120218579235,
      "grad_norm": 0.6618308759965568,
      "learning_rate": 1.792469972062699e-05,
      "loss": 11.4485,
      "step": 18578
    },
    {
      "epoch": 3.2487868852459014,
      "grad_norm": 0.770312182237066,
      "learning_rate": 1.791661066690159e-05,
      "loss": 11.3221,
      "step": 18579
    },
    {
      "epoch": 3.2489617486338798,
      "grad_norm": 0.7493918822586296,
      "learning_rate": 1.7908523259200192e-05,
      "loss": 11.2733,
      "step": 18580
    },
    {
      "epoch": 3.249136612021858,
      "grad_norm": 0.6712241728480381,
      "learning_rate": 1.7900437497684986e-05,
      "loss": 11.439,
      "step": 18581
    },
    {
      "epoch": 3.249311475409836,
      "grad_norm": 0.7147456366612766,
      "learning_rate": 1.7892353382518056e-05,
      "loss": 11.3634,
      "step": 18582
    },
    {
      "epoch": 3.2494863387978143,
      "grad_norm": 0.6389692413784834,
      "learning_rate": 1.7884270913861555e-05,
      "loss": 11.2482,
      "step": 18583
    },
    {
      "epoch": 3.2496612021857922,
      "grad_norm": 0.7510244344116181,
      "learning_rate": 1.787619009187751e-05,
      "loss": 11.4103,
      "step": 18584
    },
    {
      "epoch": 3.2498360655737706,
      "grad_norm": 0.727739422494916,
      "learning_rate": 1.786811091672804e-05,
      "loss": 11.1661,
      "step": 18585
    },
    {
      "epoch": 3.2500109289617485,
      "grad_norm": 0.6993535764387403,
      "learning_rate": 1.786003338857516e-05,
      "loss": 11.4917,
      "step": 18586
    },
    {
      "epoch": 3.250185792349727,
      "grad_norm": 0.6777687451675656,
      "learning_rate": 1.7851957507580774e-05,
      "loss": 11.641,
      "step": 18587
    },
    {
      "epoch": 3.2503606557377047,
      "grad_norm": 0.7143619131314908,
      "learning_rate": 1.784388327390687e-05,
      "loss": 11.5513,
      "step": 18588
    },
    {
      "epoch": 3.250535519125683,
      "grad_norm": 0.6761054707510651,
      "learning_rate": 1.7835810687715337e-05,
      "loss": 11.6522,
      "step": 18589
    },
    {
      "epoch": 3.2507103825136614,
      "grad_norm": 0.6688798113853718,
      "learning_rate": 1.7827739749168128e-05,
      "loss": 11.3018,
      "step": 18590
    },
    {
      "epoch": 3.2508852459016393,
      "grad_norm": 0.6555829364689371,
      "learning_rate": 1.7819670458427007e-05,
      "loss": 11.5172,
      "step": 18591
    },
    {
      "epoch": 3.2510601092896176,
      "grad_norm": 0.6888963672810745,
      "learning_rate": 1.781160281565383e-05,
      "loss": 11.4793,
      "step": 18592
    },
    {
      "epoch": 3.2512349726775955,
      "grad_norm": 0.7062761329871853,
      "learning_rate": 1.7803536821010335e-05,
      "loss": 11.0866,
      "step": 18593
    },
    {
      "epoch": 3.251409836065574,
      "grad_norm": 0.6980526467200888,
      "learning_rate": 1.7795472474658327e-05,
      "loss": 11.551,
      "step": 18594
    },
    {
      "epoch": 3.2515846994535518,
      "grad_norm": 0.6725351556757508,
      "learning_rate": 1.7787409776759523e-05,
      "loss": 11.139,
      "step": 18595
    },
    {
      "epoch": 3.25175956284153,
      "grad_norm": 0.6685695531199627,
      "learning_rate": 1.7779348727475554e-05,
      "loss": 11.3977,
      "step": 18596
    },
    {
      "epoch": 3.251934426229508,
      "grad_norm": 0.6590057827747201,
      "learning_rate": 1.7771289326968098e-05,
      "loss": 11.5095,
      "step": 18597
    },
    {
      "epoch": 3.2521092896174864,
      "grad_norm": 0.7020222607223597,
      "learning_rate": 1.7763231575398742e-05,
      "loss": 11.4277,
      "step": 18598
    },
    {
      "epoch": 3.2522841530054647,
      "grad_norm": 0.676115656410024,
      "learning_rate": 1.775517547292914e-05,
      "loss": 11.4032,
      "step": 18599
    },
    {
      "epoch": 3.2524590163934426,
      "grad_norm": 0.6692955208915133,
      "learning_rate": 1.774712101972077e-05,
      "loss": 11.4764,
      "step": 18600
    },
    {
      "epoch": 3.252633879781421,
      "grad_norm": 0.7025920244494545,
      "learning_rate": 1.7739068215935173e-05,
      "loss": 11.2716,
      "step": 18601
    },
    {
      "epoch": 3.252808743169399,
      "grad_norm": 0.7994008434605531,
      "learning_rate": 1.7731017061733835e-05,
      "loss": 11.5491,
      "step": 18602
    },
    {
      "epoch": 3.252983606557377,
      "grad_norm": 0.6471549702931549,
      "learning_rate": 1.772296755727817e-05,
      "loss": 11.3547,
      "step": 18603
    },
    {
      "epoch": 3.253158469945355,
      "grad_norm": 0.8711349022533355,
      "learning_rate": 1.771491970272968e-05,
      "loss": 11.3028,
      "step": 18604
    },
    {
      "epoch": 3.2533333333333334,
      "grad_norm": 0.6446958110414439,
      "learning_rate": 1.7706873498249687e-05,
      "loss": 11.5239,
      "step": 18605
    },
    {
      "epoch": 3.2535081967213113,
      "grad_norm": 0.635196609085357,
      "learning_rate": 1.7698828943999545e-05,
      "loss": 11.6591,
      "step": 18606
    },
    {
      "epoch": 3.2536830601092896,
      "grad_norm": 0.6200765455678928,
      "learning_rate": 1.769078604014057e-05,
      "loss": 11.5044,
      "step": 18607
    },
    {
      "epoch": 3.253857923497268,
      "grad_norm": 0.6933996212455135,
      "learning_rate": 1.768274478683407e-05,
      "loss": 11.4402,
      "step": 18608
    },
    {
      "epoch": 3.254032786885246,
      "grad_norm": 0.706260686365335,
      "learning_rate": 1.767470518424129e-05,
      "loss": 11.0865,
      "step": 18609
    },
    {
      "epoch": 3.2542076502732242,
      "grad_norm": 0.6974117785331708,
      "learning_rate": 1.7666667232523427e-05,
      "loss": 11.5556,
      "step": 18610
    },
    {
      "epoch": 3.254382513661202,
      "grad_norm": 0.6099903903017183,
      "learning_rate": 1.765863093184169e-05,
      "loss": 11.7041,
      "step": 18611
    },
    {
      "epoch": 3.2545573770491805,
      "grad_norm": 0.6851018808847973,
      "learning_rate": 1.7650596282357222e-05,
      "loss": 11.2551,
      "step": 18612
    },
    {
      "epoch": 3.2547322404371584,
      "grad_norm": 0.6398342064139314,
      "learning_rate": 1.764256328423115e-05,
      "loss": 11.2289,
      "step": 18613
    },
    {
      "epoch": 3.2549071038251367,
      "grad_norm": 0.667689647760158,
      "learning_rate": 1.7634531937624554e-05,
      "loss": 11.6493,
      "step": 18614
    },
    {
      "epoch": 3.2550819672131146,
      "grad_norm": 0.771460433123265,
      "learning_rate": 1.7626502242698484e-05,
      "loss": 11.5655,
      "step": 18615
    },
    {
      "epoch": 3.255256830601093,
      "grad_norm": 0.6730919064474415,
      "learning_rate": 1.7618474199613966e-05,
      "loss": 11.3858,
      "step": 18616
    },
    {
      "epoch": 3.2554316939890713,
      "grad_norm": 0.7103064375697575,
      "learning_rate": 1.761044780853198e-05,
      "loss": 11.1313,
      "step": 18617
    },
    {
      "epoch": 3.255606557377049,
      "grad_norm": 0.7069177123812033,
      "learning_rate": 1.7602423069613493e-05,
      "loss": 11.446,
      "step": 18618
    },
    {
      "epoch": 3.255781420765027,
      "grad_norm": 0.6703532474267897,
      "learning_rate": 1.7594399983019417e-05,
      "loss": 11.3872,
      "step": 18619
    },
    {
      "epoch": 3.2559562841530054,
      "grad_norm": 0.7173534325603477,
      "learning_rate": 1.7586378548910633e-05,
      "loss": 11.1254,
      "step": 18620
    },
    {
      "epoch": 3.2561311475409838,
      "grad_norm": 0.7078419871924169,
      "learning_rate": 1.7578358767448e-05,
      "loss": 11.373,
      "step": 18621
    },
    {
      "epoch": 3.2563060109289617,
      "grad_norm": 0.6736976697914153,
      "learning_rate": 1.757034063879235e-05,
      "loss": 11.3478,
      "step": 18622
    },
    {
      "epoch": 3.25648087431694,
      "grad_norm": 0.6366823050724632,
      "learning_rate": 1.756232416310446e-05,
      "loss": 11.4419,
      "step": 18623
    },
    {
      "epoch": 3.256655737704918,
      "grad_norm": 0.669114225487459,
      "learning_rate": 1.7554309340545084e-05,
      "loss": 11.523,
      "step": 18624
    },
    {
      "epoch": 3.2568306010928962,
      "grad_norm": 0.7087107253285816,
      "learning_rate": 1.7546296171274945e-05,
      "loss": 11.3685,
      "step": 18625
    },
    {
      "epoch": 3.257005464480874,
      "grad_norm": 0.7009875496744409,
      "learning_rate": 1.753828465545474e-05,
      "loss": 11.071,
      "step": 18626
    },
    {
      "epoch": 3.2571803278688525,
      "grad_norm": 0.6475792419957104,
      "learning_rate": 1.7530274793245114e-05,
      "loss": 11.0824,
      "step": 18627
    },
    {
      "epoch": 3.2573551912568304,
      "grad_norm": 0.7028595460741341,
      "learning_rate": 1.7522266584806702e-05,
      "loss": 11.2682,
      "step": 18628
    },
    {
      "epoch": 3.2575300546448087,
      "grad_norm": 0.7320878137751674,
      "learning_rate": 1.751426003030008e-05,
      "loss": 11.4783,
      "step": 18629
    },
    {
      "epoch": 3.257704918032787,
      "grad_norm": 0.7135548696110473,
      "learning_rate": 1.7506255129885806e-05,
      "loss": 11.4163,
      "step": 18630
    },
    {
      "epoch": 3.257879781420765,
      "grad_norm": 0.7007714202551522,
      "learning_rate": 1.7498251883724403e-05,
      "loss": 11.4788,
      "step": 18631
    },
    {
      "epoch": 3.2580546448087433,
      "grad_norm": 0.6768075400697257,
      "learning_rate": 1.7490250291976374e-05,
      "loss": 11.484,
      "step": 18632
    },
    {
      "epoch": 3.258229508196721,
      "grad_norm": 0.6533532052913279,
      "learning_rate": 1.7482250354802156e-05,
      "loss": 11.6071,
      "step": 18633
    },
    {
      "epoch": 3.2584043715846995,
      "grad_norm": 0.6492293777892232,
      "learning_rate": 1.7474252072362186e-05,
      "loss": 11.5947,
      "step": 18634
    },
    {
      "epoch": 3.2585792349726774,
      "grad_norm": 0.699106929955606,
      "learning_rate": 1.7466255444816848e-05,
      "loss": 11.3288,
      "step": 18635
    },
    {
      "epoch": 3.2587540983606558,
      "grad_norm": 0.6987951363195645,
      "learning_rate": 1.7458260472326526e-05,
      "loss": 11.3921,
      "step": 18636
    },
    {
      "epoch": 3.2589289617486337,
      "grad_norm": 0.7220653329692909,
      "learning_rate": 1.745026715505146e-05,
      "loss": 11.4872,
      "step": 18637
    },
    {
      "epoch": 3.259103825136612,
      "grad_norm": 0.6810031472623459,
      "learning_rate": 1.7442275493152037e-05,
      "loss": 11.2467,
      "step": 18638
    },
    {
      "epoch": 3.2592786885245904,
      "grad_norm": 0.6787675941442249,
      "learning_rate": 1.7434285486788472e-05,
      "loss": 11.1288,
      "step": 18639
    },
    {
      "epoch": 3.2594535519125682,
      "grad_norm": 0.6814391971082104,
      "learning_rate": 1.7426297136120984e-05,
      "loss": 11.2543,
      "step": 18640
    },
    {
      "epoch": 3.2596284153005466,
      "grad_norm": 0.6681364064584254,
      "learning_rate": 1.7418310441309783e-05,
      "loss": 11.4886,
      "step": 18641
    },
    {
      "epoch": 3.2598032786885245,
      "grad_norm": 0.6703093021587847,
      "learning_rate": 1.7410325402515003e-05,
      "loss": 11.5042,
      "step": 18642
    },
    {
      "epoch": 3.259978142076503,
      "grad_norm": 0.7622550254205165,
      "learning_rate": 1.74023420198968e-05,
      "loss": 11.2926,
      "step": 18643
    },
    {
      "epoch": 3.2601530054644807,
      "grad_norm": 0.6739155797203763,
      "learning_rate": 1.739436029361524e-05,
      "loss": 11.2995,
      "step": 18644
    },
    {
      "epoch": 3.260327868852459,
      "grad_norm": 0.7028355254421983,
      "learning_rate": 1.7386380223830413e-05,
      "loss": 11.619,
      "step": 18645
    },
    {
      "epoch": 3.260502732240437,
      "grad_norm": 0.6875172390697786,
      "learning_rate": 1.7378401810702273e-05,
      "loss": 11.5105,
      "step": 18646
    },
    {
      "epoch": 3.2606775956284153,
      "grad_norm": 0.695885393698569,
      "learning_rate": 1.737042505439087e-05,
      "loss": 11.0636,
      "step": 18647
    },
    {
      "epoch": 3.2608524590163936,
      "grad_norm": 0.6943888633070748,
      "learning_rate": 1.736244995505617e-05,
      "loss": 11.2896,
      "step": 18648
    },
    {
      "epoch": 3.2610273224043715,
      "grad_norm": 0.6901584137811975,
      "learning_rate": 1.7354476512858066e-05,
      "loss": 11.6298,
      "step": 18649
    },
    {
      "epoch": 3.26120218579235,
      "grad_norm": 0.7086746954593574,
      "learning_rate": 1.734650472795649e-05,
      "loss": 11.6862,
      "step": 18650
    },
    {
      "epoch": 3.261377049180328,
      "grad_norm": 0.6506814018173026,
      "learning_rate": 1.7338534600511224e-05,
      "loss": 11.6453,
      "step": 18651
    },
    {
      "epoch": 3.261551912568306,
      "grad_norm": 0.6920056245708462,
      "learning_rate": 1.7330566130682168e-05,
      "loss": 11.5376,
      "step": 18652
    },
    {
      "epoch": 3.261726775956284,
      "grad_norm": 0.7401224213565225,
      "learning_rate": 1.7322599318629095e-05,
      "loss": 11.371,
      "step": 18653
    },
    {
      "epoch": 3.2619016393442624,
      "grad_norm": 0.6365560295899626,
      "learning_rate": 1.7314634164511757e-05,
      "loss": 11.2885,
      "step": 18654
    },
    {
      "epoch": 3.2620765027322403,
      "grad_norm": 0.6510643668547302,
      "learning_rate": 1.73066706684899e-05,
      "loss": 11.1482,
      "step": 18655
    },
    {
      "epoch": 3.2622513661202186,
      "grad_norm": 0.6975286826563787,
      "learning_rate": 1.7298708830723155e-05,
      "loss": 11.4924,
      "step": 18656
    },
    {
      "epoch": 3.262426229508197,
      "grad_norm": 0.6763774678972907,
      "learning_rate": 1.729074865137125e-05,
      "loss": 11.391,
      "step": 18657
    },
    {
      "epoch": 3.262601092896175,
      "grad_norm": 0.6537351619058999,
      "learning_rate": 1.7282790130593784e-05,
      "loss": 11.426,
      "step": 18658
    },
    {
      "epoch": 3.262775956284153,
      "grad_norm": 0.6904128959606755,
      "learning_rate": 1.7274833268550373e-05,
      "loss": 11.3649,
      "step": 18659
    },
    {
      "epoch": 3.262950819672131,
      "grad_norm": 0.8181065427560829,
      "learning_rate": 1.7266878065400527e-05,
      "loss": 11.5409,
      "step": 18660
    },
    {
      "epoch": 3.2631256830601094,
      "grad_norm": 0.6304943276846585,
      "learning_rate": 1.7258924521303775e-05,
      "loss": 10.9395,
      "step": 18661
    },
    {
      "epoch": 3.2633005464480873,
      "grad_norm": 0.7298359815131777,
      "learning_rate": 1.725097263641967e-05,
      "loss": 11.2049,
      "step": 18662
    },
    {
      "epoch": 3.2634754098360657,
      "grad_norm": 0.6459087338657956,
      "learning_rate": 1.7243022410907627e-05,
      "loss": 11.3688,
      "step": 18663
    },
    {
      "epoch": 3.2636502732240436,
      "grad_norm": 0.6995500922826992,
      "learning_rate": 1.7235073844927108e-05,
      "loss": 11.4897,
      "step": 18664
    },
    {
      "epoch": 3.263825136612022,
      "grad_norm": 0.6321105932307818,
      "learning_rate": 1.722712693863744e-05,
      "loss": 11.6958,
      "step": 18665
    },
    {
      "epoch": 3.2640000000000002,
      "grad_norm": 0.6728122520617045,
      "learning_rate": 1.721918169219804e-05,
      "loss": 11.4633,
      "step": 18666
    },
    {
      "epoch": 3.264174863387978,
      "grad_norm": 0.624603037714942,
      "learning_rate": 1.7211238105768214e-05,
      "loss": 11.7023,
      "step": 18667
    },
    {
      "epoch": 3.2643497267759565,
      "grad_norm": 0.6909174856602922,
      "learning_rate": 1.720329617950729e-05,
      "loss": 11.6067,
      "step": 18668
    },
    {
      "epoch": 3.2645245901639344,
      "grad_norm": 0.8404172521227928,
      "learning_rate": 1.719535591357446e-05,
      "loss": 11.3078,
      "step": 18669
    },
    {
      "epoch": 3.2646994535519127,
      "grad_norm": 0.6574026363037665,
      "learning_rate": 1.7187417308128984e-05,
      "loss": 11.6753,
      "step": 18670
    },
    {
      "epoch": 3.2648743169398906,
      "grad_norm": 0.719541953945352,
      "learning_rate": 1.717948036333007e-05,
      "loss": 11.6008,
      "step": 18671
    },
    {
      "epoch": 3.265049180327869,
      "grad_norm": 0.6645859191206629,
      "learning_rate": 1.717154507933687e-05,
      "loss": 11.7126,
      "step": 18672
    },
    {
      "epoch": 3.265224043715847,
      "grad_norm": 0.7306962721580067,
      "learning_rate": 1.7163611456308527e-05,
      "loss": 11.4884,
      "step": 18673
    },
    {
      "epoch": 3.265398907103825,
      "grad_norm": 0.7848503252763629,
      "learning_rate": 1.7155679494404097e-05,
      "loss": 11.18,
      "step": 18674
    },
    {
      "epoch": 3.265573770491803,
      "grad_norm": 0.6451846326484596,
      "learning_rate": 1.7147749193782635e-05,
      "loss": 11.6616,
      "step": 18675
    },
    {
      "epoch": 3.2657486338797814,
      "grad_norm": 0.6857895954256809,
      "learning_rate": 1.7139820554603214e-05,
      "loss": 11.261,
      "step": 18676
    },
    {
      "epoch": 3.2659234972677593,
      "grad_norm": 0.6783121070158612,
      "learning_rate": 1.7131893577024826e-05,
      "loss": 11.351,
      "step": 18677
    },
    {
      "epoch": 3.2660983606557377,
      "grad_norm": 0.6616208929752255,
      "learning_rate": 1.712396826120639e-05,
      "loss": 11.5045,
      "step": 18678
    },
    {
      "epoch": 3.266273224043716,
      "grad_norm": 0.6709799026897318,
      "learning_rate": 1.711604460730685e-05,
      "loss": 11.3453,
      "step": 18679
    },
    {
      "epoch": 3.266448087431694,
      "grad_norm": 0.6796020430123746,
      "learning_rate": 1.710812261548508e-05,
      "loss": 11.4248,
      "step": 18680
    },
    {
      "epoch": 3.2666229508196722,
      "grad_norm": 0.6687588881584093,
      "learning_rate": 1.710020228589999e-05,
      "loss": 11.5416,
      "step": 18681
    },
    {
      "epoch": 3.26679781420765,
      "grad_norm": 0.6666600452711496,
      "learning_rate": 1.7092283618710404e-05,
      "loss": 11.3053,
      "step": 18682
    },
    {
      "epoch": 3.2669726775956285,
      "grad_norm": 0.7032970606542706,
      "learning_rate": 1.7084366614075064e-05,
      "loss": 11.4907,
      "step": 18683
    },
    {
      "epoch": 3.2671475409836064,
      "grad_norm": 0.7077594946333993,
      "learning_rate": 1.707645127215276e-05,
      "loss": 11.531,
      "step": 18684
    },
    {
      "epoch": 3.2673224043715847,
      "grad_norm": 0.6527489441535176,
      "learning_rate": 1.7068537593102186e-05,
      "loss": 11.526,
      "step": 18685
    },
    {
      "epoch": 3.2674972677595626,
      "grad_norm": 0.6999125160345748,
      "learning_rate": 1.7060625577082123e-05,
      "loss": 11.4895,
      "step": 18686
    },
    {
      "epoch": 3.267672131147541,
      "grad_norm": 0.7359811068929741,
      "learning_rate": 1.7052715224251147e-05,
      "loss": 11.3268,
      "step": 18687
    },
    {
      "epoch": 3.2678469945355193,
      "grad_norm": 0.6416947552870961,
      "learning_rate": 1.7044806534767908e-05,
      "loss": 11.2458,
      "step": 18688
    },
    {
      "epoch": 3.268021857923497,
      "grad_norm": 0.6837480840626342,
      "learning_rate": 1.7036899508790994e-05,
      "loss": 11.7231,
      "step": 18689
    },
    {
      "epoch": 3.2681967213114755,
      "grad_norm": 0.7015754326138633,
      "learning_rate": 1.7028994146478982e-05,
      "loss": 11.3243,
      "step": 18690
    },
    {
      "epoch": 3.2683715846994534,
      "grad_norm": 0.6497412306863392,
      "learning_rate": 1.7021090447990428e-05,
      "loss": 11.4127,
      "step": 18691
    },
    {
      "epoch": 3.2685464480874318,
      "grad_norm": 0.7248907174945463,
      "learning_rate": 1.701318841348376e-05,
      "loss": 11.0732,
      "step": 18692
    },
    {
      "epoch": 3.2687213114754097,
      "grad_norm": 0.681442517510825,
      "learning_rate": 1.7005288043117472e-05,
      "loss": 11.4111,
      "step": 18693
    },
    {
      "epoch": 3.268896174863388,
      "grad_norm": 0.6670447987513323,
      "learning_rate": 1.6997389337049952e-05,
      "loss": 11.0651,
      "step": 18694
    },
    {
      "epoch": 3.269071038251366,
      "grad_norm": 0.7363086173140129,
      "learning_rate": 1.698949229543969e-05,
      "loss": 11.1572,
      "step": 18695
    },
    {
      "epoch": 3.2692459016393443,
      "grad_norm": 0.673309523629511,
      "learning_rate": 1.6981596918444953e-05,
      "loss": 11.4398,
      "step": 18696
    },
    {
      "epoch": 3.2694207650273226,
      "grad_norm": 0.668877235084399,
      "learning_rate": 1.697370320622409e-05,
      "loss": 11.6023,
      "step": 18697
    },
    {
      "epoch": 3.2695956284153005,
      "grad_norm": 0.6987811754484948,
      "learning_rate": 1.696581115893541e-05,
      "loss": 11.5571,
      "step": 18698
    },
    {
      "epoch": 3.269770491803279,
      "grad_norm": 0.6986090895938591,
      "learning_rate": 1.6957920776737135e-05,
      "loss": 11.4157,
      "step": 18699
    },
    {
      "epoch": 3.2699453551912567,
      "grad_norm": 0.6672057189143312,
      "learning_rate": 1.695003205978757e-05,
      "loss": 11.4823,
      "step": 18700
    },
    {
      "epoch": 3.270120218579235,
      "grad_norm": 0.65997168010814,
      "learning_rate": 1.6942145008244837e-05,
      "loss": 11.2608,
      "step": 18701
    },
    {
      "epoch": 3.270295081967213,
      "grad_norm": 0.6142177359983465,
      "learning_rate": 1.6934259622267123e-05,
      "loss": 11.2838,
      "step": 18702
    },
    {
      "epoch": 3.2704699453551913,
      "grad_norm": 0.666862698031295,
      "learning_rate": 1.6926375902012547e-05,
      "loss": 11.5907,
      "step": 18703
    },
    {
      "epoch": 3.270644808743169,
      "grad_norm": 0.6671214855749686,
      "learning_rate": 1.6918493847639194e-05,
      "loss": 11.614,
      "step": 18704
    },
    {
      "epoch": 3.2708196721311475,
      "grad_norm": 0.6532477849270971,
      "learning_rate": 1.6910613459305146e-05,
      "loss": 11.3629,
      "step": 18705
    },
    {
      "epoch": 3.270994535519126,
      "grad_norm": 0.6958181728124367,
      "learning_rate": 1.6902734737168403e-05,
      "loss": 11.5094,
      "step": 18706
    },
    {
      "epoch": 3.271169398907104,
      "grad_norm": 0.691655545375823,
      "learning_rate": 1.6894857681386976e-05,
      "loss": 11.5697,
      "step": 18707
    },
    {
      "epoch": 3.271344262295082,
      "grad_norm": 0.7214728893437061,
      "learning_rate": 1.6886982292118824e-05,
      "loss": 11.2753,
      "step": 18708
    },
    {
      "epoch": 3.27151912568306,
      "grad_norm": 0.6374547334056477,
      "learning_rate": 1.687910856952186e-05,
      "loss": 11.4394,
      "step": 18709
    },
    {
      "epoch": 3.2716939890710384,
      "grad_norm": 0.6384734649993875,
      "learning_rate": 1.6871236513753998e-05,
      "loss": 11.6123,
      "step": 18710
    },
    {
      "epoch": 3.2718688524590163,
      "grad_norm": 0.7167280064431356,
      "learning_rate": 1.686336612497307e-05,
      "loss": 11.3142,
      "step": 18711
    },
    {
      "epoch": 3.2720437158469946,
      "grad_norm": 0.6680009653539883,
      "learning_rate": 1.6855497403336916e-05,
      "loss": 11.5372,
      "step": 18712
    },
    {
      "epoch": 3.2722185792349725,
      "grad_norm": 0.6522164359187761,
      "learning_rate": 1.684763034900333e-05,
      "loss": 11.4168,
      "step": 18713
    },
    {
      "epoch": 3.272393442622951,
      "grad_norm": 0.7091558197870927,
      "learning_rate": 1.6839764962130057e-05,
      "loss": 11.3771,
      "step": 18714
    },
    {
      "epoch": 3.272568306010929,
      "grad_norm": 0.6718631570195488,
      "learning_rate": 1.6831901242874836e-05,
      "loss": 11.1516,
      "step": 18715
    },
    {
      "epoch": 3.272743169398907,
      "grad_norm": 0.6499412773795009,
      "learning_rate": 1.6824039191395358e-05,
      "loss": 11.4656,
      "step": 18716
    },
    {
      "epoch": 3.2729180327868854,
      "grad_norm": 0.6692815877594728,
      "learning_rate": 1.681617880784928e-05,
      "loss": 11.5964,
      "step": 18717
    },
    {
      "epoch": 3.2730928961748633,
      "grad_norm": 0.7354278271662881,
      "learning_rate": 1.6808320092394226e-05,
      "loss": 11.1525,
      "step": 18718
    },
    {
      "epoch": 3.2732677595628417,
      "grad_norm": 0.7427537378869334,
      "learning_rate": 1.6800463045187776e-05,
      "loss": 11.4622,
      "step": 18719
    },
    {
      "epoch": 3.2734426229508196,
      "grad_norm": 0.7455228727781569,
      "learning_rate": 1.6792607666387506e-05,
      "loss": 11.2096,
      "step": 18720
    },
    {
      "epoch": 3.273617486338798,
      "grad_norm": 0.6676716971415698,
      "learning_rate": 1.6784753956150934e-05,
      "loss": 11.4733,
      "step": 18721
    },
    {
      "epoch": 3.273792349726776,
      "grad_norm": 0.6675862197287509,
      "learning_rate": 1.6776901914635546e-05,
      "loss": 11.3152,
      "step": 18722
    },
    {
      "epoch": 3.273967213114754,
      "grad_norm": 0.6776289694589411,
      "learning_rate": 1.6769051541998803e-05,
      "loss": 11.5522,
      "step": 18723
    },
    {
      "epoch": 3.2741420765027325,
      "grad_norm": 0.6525870811241827,
      "learning_rate": 1.6761202838398128e-05,
      "loss": 11.2478,
      "step": 18724
    },
    {
      "epoch": 3.2743169398907104,
      "grad_norm": 0.6904166643857717,
      "learning_rate": 1.6753355803990912e-05,
      "loss": 11.2853,
      "step": 18725
    },
    {
      "epoch": 3.2744918032786887,
      "grad_norm": 0.6848799072338283,
      "learning_rate": 1.674551043893452e-05,
      "loss": 11.3123,
      "step": 18726
    },
    {
      "epoch": 3.2746666666666666,
      "grad_norm": 0.6554542343401797,
      "learning_rate": 1.673766674338626e-05,
      "loss": 11.2396,
      "step": 18727
    },
    {
      "epoch": 3.274841530054645,
      "grad_norm": 0.666579536627245,
      "learning_rate": 1.672982471750344e-05,
      "loss": 11.5751,
      "step": 18728
    },
    {
      "epoch": 3.275016393442623,
      "grad_norm": 0.724588846438528,
      "learning_rate": 1.67219843614433e-05,
      "loss": 11.4126,
      "step": 18729
    },
    {
      "epoch": 3.275191256830601,
      "grad_norm": 0.7245997380127961,
      "learning_rate": 1.671414567536307e-05,
      "loss": 11.2889,
      "step": 18730
    },
    {
      "epoch": 3.275366120218579,
      "grad_norm": 0.7127324071644905,
      "learning_rate": 1.6706308659419945e-05,
      "loss": 11.4445,
      "step": 18731
    },
    {
      "epoch": 3.2755409836065574,
      "grad_norm": 0.6298922205759574,
      "learning_rate": 1.669847331377109e-05,
      "loss": 11.2829,
      "step": 18732
    },
    {
      "epoch": 3.2757158469945353,
      "grad_norm": 0.7074288362887919,
      "learning_rate": 1.669063963857358e-05,
      "loss": 11.5677,
      "step": 18733
    },
    {
      "epoch": 3.2758907103825137,
      "grad_norm": 0.6346969982830034,
      "learning_rate": 1.6682807633984555e-05,
      "loss": 11.4446,
      "step": 18734
    },
    {
      "epoch": 3.2760655737704916,
      "grad_norm": 0.6425862013302809,
      "learning_rate": 1.6674977300161056e-05,
      "loss": 11.6037,
      "step": 18735
    },
    {
      "epoch": 3.27624043715847,
      "grad_norm": 0.6691402521495831,
      "learning_rate": 1.666714863726011e-05,
      "loss": 11.5747,
      "step": 18736
    },
    {
      "epoch": 3.2764153005464483,
      "grad_norm": 0.6367433536392675,
      "learning_rate": 1.6659321645438685e-05,
      "loss": 11.5683,
      "step": 18737
    },
    {
      "epoch": 3.276590163934426,
      "grad_norm": 0.6717593645140302,
      "learning_rate": 1.6651496324853765e-05,
      "loss": 11.4528,
      "step": 18738
    },
    {
      "epoch": 3.2767650273224045,
      "grad_norm": 0.6326577196329022,
      "learning_rate": 1.664367267566225e-05,
      "loss": 11.3964,
      "step": 18739
    },
    {
      "epoch": 3.2769398907103824,
      "grad_norm": 0.6745846470239375,
      "learning_rate": 1.663585069802104e-05,
      "loss": 11.334,
      "step": 18740
    },
    {
      "epoch": 3.2771147540983607,
      "grad_norm": 0.6938382509674009,
      "learning_rate": 1.6628030392087e-05,
      "loss": 11.4038,
      "step": 18741
    },
    {
      "epoch": 3.2772896174863386,
      "grad_norm": 0.7126043426601363,
      "learning_rate": 1.66202117580169e-05,
      "loss": 11.4359,
      "step": 18742
    },
    {
      "epoch": 3.277464480874317,
      "grad_norm": 0.6688548254101454,
      "learning_rate": 1.6612394795967578e-05,
      "loss": 11.2275,
      "step": 18743
    },
    {
      "epoch": 3.277639344262295,
      "grad_norm": 0.6731821347155972,
      "learning_rate": 1.6604579506095775e-05,
      "loss": 11.6187,
      "step": 18744
    },
    {
      "epoch": 3.277814207650273,
      "grad_norm": 0.6635054828684874,
      "learning_rate": 1.659676588855822e-05,
      "loss": 11.5443,
      "step": 18745
    },
    {
      "epoch": 3.2779890710382515,
      "grad_norm": 0.7163466414901365,
      "learning_rate": 1.658895394351161e-05,
      "loss": 11.2849,
      "step": 18746
    },
    {
      "epoch": 3.2781639344262294,
      "grad_norm": 0.6504848313544602,
      "learning_rate": 1.658114367111252e-05,
      "loss": 11.4243,
      "step": 18747
    },
    {
      "epoch": 3.278338797814208,
      "grad_norm": 0.6502789060862801,
      "learning_rate": 1.657333507151766e-05,
      "loss": 11.2453,
      "step": 18748
    },
    {
      "epoch": 3.2785136612021857,
      "grad_norm": 0.7735304876279008,
      "learning_rate": 1.6565528144883592e-05,
      "loss": 11.4669,
      "step": 18749
    },
    {
      "epoch": 3.278688524590164,
      "grad_norm": 0.6104435939878401,
      "learning_rate": 1.6557722891366878e-05,
      "loss": 11.5423,
      "step": 18750
    },
    {
      "epoch": 3.278863387978142,
      "grad_norm": 0.6874664301540945,
      "learning_rate": 1.6549919311124008e-05,
      "loss": 11.273,
      "step": 18751
    },
    {
      "epoch": 3.2790382513661203,
      "grad_norm": 0.7128048437452447,
      "learning_rate": 1.654211740431144e-05,
      "loss": 11.1431,
      "step": 18752
    },
    {
      "epoch": 3.279213114754098,
      "grad_norm": 0.7109333218990658,
      "learning_rate": 1.6534317171085702e-05,
      "loss": 11.3307,
      "step": 18753
    },
    {
      "epoch": 3.2793879781420765,
      "grad_norm": 0.6264175125109779,
      "learning_rate": 1.652651861160318e-05,
      "loss": 11.295,
      "step": 18754
    },
    {
      "epoch": 3.279562841530055,
      "grad_norm": 0.6579312298201568,
      "learning_rate": 1.651872172602028e-05,
      "loss": 11.4754,
      "step": 18755
    },
    {
      "epoch": 3.2797377049180327,
      "grad_norm": 0.6903482191264763,
      "learning_rate": 1.6510926514493307e-05,
      "loss": 11.0395,
      "step": 18756
    },
    {
      "epoch": 3.279912568306011,
      "grad_norm": 0.6601224880364758,
      "learning_rate": 1.650313297717857e-05,
      "loss": 11.484,
      "step": 18757
    },
    {
      "epoch": 3.280087431693989,
      "grad_norm": 0.7032372491842018,
      "learning_rate": 1.6495341114232422e-05,
      "loss": 11.165,
      "step": 18758
    },
    {
      "epoch": 3.2802622950819673,
      "grad_norm": 0.6704093779275833,
      "learning_rate": 1.6487550925811092e-05,
      "loss": 11.4844,
      "step": 18759
    },
    {
      "epoch": 3.280437158469945,
      "grad_norm": 0.6829926376822131,
      "learning_rate": 1.6479762412070765e-05,
      "loss": 11.035,
      "step": 18760
    },
    {
      "epoch": 3.2806120218579236,
      "grad_norm": 0.7154002377166248,
      "learning_rate": 1.6471975573167618e-05,
      "loss": 11.5353,
      "step": 18761
    },
    {
      "epoch": 3.2807868852459015,
      "grad_norm": 0.7165909920986341,
      "learning_rate": 1.646419040925784e-05,
      "loss": 11.3098,
      "step": 18762
    },
    {
      "epoch": 3.28096174863388,
      "grad_norm": 0.6553266312364335,
      "learning_rate": 1.6456406920497548e-05,
      "loss": 11.179,
      "step": 18763
    },
    {
      "epoch": 3.281136612021858,
      "grad_norm": 0.8011259682924575,
      "learning_rate": 1.6448625107042825e-05,
      "loss": 11.4619,
      "step": 18764
    },
    {
      "epoch": 3.281311475409836,
      "grad_norm": 0.6971444463126913,
      "learning_rate": 1.6440844969049684e-05,
      "loss": 11.3163,
      "step": 18765
    },
    {
      "epoch": 3.2814863387978144,
      "grad_norm": 0.6418552662509788,
      "learning_rate": 1.6433066506674134e-05,
      "loss": 11.0307,
      "step": 18766
    },
    {
      "epoch": 3.2816612021857923,
      "grad_norm": 0.6485598831375881,
      "learning_rate": 1.6425289720072212e-05,
      "loss": 11.2579,
      "step": 18767
    },
    {
      "epoch": 3.2818360655737706,
      "grad_norm": 0.6386503665862506,
      "learning_rate": 1.6417514609399865e-05,
      "loss": 11.5382,
      "step": 18768
    },
    {
      "epoch": 3.2820109289617485,
      "grad_norm": 0.6731401513058611,
      "learning_rate": 1.6409741174812955e-05,
      "loss": 11.0209,
      "step": 18769
    },
    {
      "epoch": 3.282185792349727,
      "grad_norm": 0.717352507108002,
      "learning_rate": 1.640196941646739e-05,
      "loss": 11.4855,
      "step": 18770
    },
    {
      "epoch": 3.2823606557377047,
      "grad_norm": 0.6676653391361819,
      "learning_rate": 1.639419933451899e-05,
      "loss": 11.4944,
      "step": 18771
    },
    {
      "epoch": 3.282535519125683,
      "grad_norm": 0.7134379084578103,
      "learning_rate": 1.6386430929123618e-05,
      "loss": 11.1703,
      "step": 18772
    },
    {
      "epoch": 3.2827103825136614,
      "grad_norm": 0.7606730002814955,
      "learning_rate": 1.6378664200437067e-05,
      "loss": 11.5346,
      "step": 18773
    },
    {
      "epoch": 3.2828852459016393,
      "grad_norm": 0.6213334442421969,
      "learning_rate": 1.6370899148615027e-05,
      "loss": 11.743,
      "step": 18774
    },
    {
      "epoch": 3.2830601092896177,
      "grad_norm": 0.6913039282692974,
      "learning_rate": 1.6363135773813232e-05,
      "loss": 11.5769,
      "step": 18775
    },
    {
      "epoch": 3.2832349726775956,
      "grad_norm": 0.7190897646321089,
      "learning_rate": 1.6355374076187335e-05,
      "loss": 11.2777,
      "step": 18776
    },
    {
      "epoch": 3.283409836065574,
      "grad_norm": 0.7190582821816006,
      "learning_rate": 1.6347614055893055e-05,
      "loss": 11.4638,
      "step": 18777
    },
    {
      "epoch": 3.283584699453552,
      "grad_norm": 0.7284914417405901,
      "learning_rate": 1.6339855713085948e-05,
      "loss": 11.4098,
      "step": 18778
    },
    {
      "epoch": 3.28375956284153,
      "grad_norm": 0.7191088399461449,
      "learning_rate": 1.63320990479216e-05,
      "loss": 11.3756,
      "step": 18779
    },
    {
      "epoch": 3.283934426229508,
      "grad_norm": 0.6908735547268772,
      "learning_rate": 1.6324344060555554e-05,
      "loss": 11.7853,
      "step": 18780
    },
    {
      "epoch": 3.2841092896174864,
      "grad_norm": 0.6387859253677555,
      "learning_rate": 1.6316590751143314e-05,
      "loss": 11.5554,
      "step": 18781
    },
    {
      "epoch": 3.2842841530054647,
      "grad_norm": 0.7438637962731889,
      "learning_rate": 1.6308839119840412e-05,
      "loss": 11.3877,
      "step": 18782
    },
    {
      "epoch": 3.2844590163934426,
      "grad_norm": 0.6115008248418907,
      "learning_rate": 1.630108916680223e-05,
      "loss": 11.786,
      "step": 18783
    },
    {
      "epoch": 3.2846338797814205,
      "grad_norm": 0.7119865439410737,
      "learning_rate": 1.6293340892184196e-05,
      "loss": 11.1325,
      "step": 18784
    },
    {
      "epoch": 3.284808743169399,
      "grad_norm": 0.7251443033555335,
      "learning_rate": 1.628559429614167e-05,
      "loss": 11.4217,
      "step": 18785
    },
    {
      "epoch": 3.284983606557377,
      "grad_norm": 0.75714866414989,
      "learning_rate": 1.6277849378830057e-05,
      "loss": 11.3751,
      "step": 18786
    },
    {
      "epoch": 3.285158469945355,
      "grad_norm": 0.9607196469476373,
      "learning_rate": 1.6270106140404607e-05,
      "loss": 11.456,
      "step": 18787
    },
    {
      "epoch": 3.2853333333333334,
      "grad_norm": 0.675108758912322,
      "learning_rate": 1.6262364581020617e-05,
      "loss": 11.1628,
      "step": 18788
    },
    {
      "epoch": 3.2855081967213113,
      "grad_norm": 0.7240213043605557,
      "learning_rate": 1.625462470083332e-05,
      "loss": 11.401,
      "step": 18789
    },
    {
      "epoch": 3.2856830601092897,
      "grad_norm": 0.6987099695475293,
      "learning_rate": 1.6246886499997904e-05,
      "loss": 11.3481,
      "step": 18790
    },
    {
      "epoch": 3.2858579234972676,
      "grad_norm": 0.6517203317873995,
      "learning_rate": 1.6239149978669622e-05,
      "loss": 11.2898,
      "step": 18791
    },
    {
      "epoch": 3.286032786885246,
      "grad_norm": 0.6750016261843516,
      "learning_rate": 1.6231415137003537e-05,
      "loss": 11.36,
      "step": 18792
    },
    {
      "epoch": 3.286207650273224,
      "grad_norm": 0.7331217064487405,
      "learning_rate": 1.622368197515478e-05,
      "loss": 11.4279,
      "step": 18793
    },
    {
      "epoch": 3.286382513661202,
      "grad_norm": 0.6547668848912558,
      "learning_rate": 1.621595049327843e-05,
      "loss": 11.3612,
      "step": 18794
    },
    {
      "epoch": 3.2865573770491805,
      "grad_norm": 0.6500793849544231,
      "learning_rate": 1.620822069152952e-05,
      "loss": 11.4813,
      "step": 18795
    },
    {
      "epoch": 3.2867322404371584,
      "grad_norm": 0.7086586411799504,
      "learning_rate": 1.6200492570063063e-05,
      "loss": 11.6076,
      "step": 18796
    },
    {
      "epoch": 3.2869071038251367,
      "grad_norm": 0.7179554481966769,
      "learning_rate": 1.619276612903402e-05,
      "loss": 11.4673,
      "step": 18797
    },
    {
      "epoch": 3.2870819672131146,
      "grad_norm": 0.6493251302683531,
      "learning_rate": 1.618504136859734e-05,
      "loss": 11.3852,
      "step": 18798
    },
    {
      "epoch": 3.287256830601093,
      "grad_norm": 0.6806131502930557,
      "learning_rate": 1.617731828890794e-05,
      "loss": 11.2102,
      "step": 18799
    },
    {
      "epoch": 3.287431693989071,
      "grad_norm": 0.7475971601728141,
      "learning_rate": 1.6169596890120664e-05,
      "loss": 11.3001,
      "step": 18800
    },
    {
      "epoch": 3.287606557377049,
      "grad_norm": 0.6433186351073309,
      "learning_rate": 1.616187717239037e-05,
      "loss": 11.4597,
      "step": 18801
    },
    {
      "epoch": 3.287781420765027,
      "grad_norm": 0.6627562852900747,
      "learning_rate": 1.615415913587185e-05,
      "loss": 11.4501,
      "step": 18802
    },
    {
      "epoch": 3.2879562841530054,
      "grad_norm": 0.7975992267292021,
      "learning_rate": 1.6146442780719883e-05,
      "loss": 11.2498,
      "step": 18803
    },
    {
      "epoch": 3.288131147540984,
      "grad_norm": 0.6968185447345534,
      "learning_rate": 1.613872810708921e-05,
      "loss": 11.5076,
      "step": 18804
    },
    {
      "epoch": 3.2883060109289617,
      "grad_norm": 0.6604043264289541,
      "learning_rate": 1.6131015115134518e-05,
      "loss": 11.3403,
      "step": 18805
    },
    {
      "epoch": 3.28848087431694,
      "grad_norm": 0.667617214098815,
      "learning_rate": 1.612330380501048e-05,
      "loss": 11.6131,
      "step": 18806
    },
    {
      "epoch": 3.288655737704918,
      "grad_norm": 0.741547555424233,
      "learning_rate": 1.611559417687174e-05,
      "loss": 11.4713,
      "step": 18807
    },
    {
      "epoch": 3.2888306010928963,
      "grad_norm": 0.6673142608429674,
      "learning_rate": 1.6107886230872905e-05,
      "loss": 11.5258,
      "step": 18808
    },
    {
      "epoch": 3.289005464480874,
      "grad_norm": 0.6908773000352842,
      "learning_rate": 1.6100179967168526e-05,
      "loss": 11.4212,
      "step": 18809
    },
    {
      "epoch": 3.2891803278688525,
      "grad_norm": 0.7205988294784906,
      "learning_rate": 1.609247538591314e-05,
      "loss": 11.2555,
      "step": 18810
    },
    {
      "epoch": 3.2893551912568304,
      "grad_norm": 0.6962994429088754,
      "learning_rate": 1.608477248726126e-05,
      "loss": 11.5585,
      "step": 18811
    },
    {
      "epoch": 3.2895300546448087,
      "grad_norm": 0.7209479369936439,
      "learning_rate": 1.607707127136734e-05,
      "loss": 11.5964,
      "step": 18812
    },
    {
      "epoch": 3.289704918032787,
      "grad_norm": 0.6591940389884315,
      "learning_rate": 1.606937173838582e-05,
      "loss": 11.2648,
      "step": 18813
    },
    {
      "epoch": 3.289879781420765,
      "grad_norm": 0.5670715629462366,
      "learning_rate": 1.606167388847111e-05,
      "loss": 11.2677,
      "step": 18814
    },
    {
      "epoch": 3.2900546448087433,
      "grad_norm": 0.6834695296253437,
      "learning_rate": 1.6053977721777547e-05,
      "loss": 11.4259,
      "step": 18815
    },
    {
      "epoch": 3.290229508196721,
      "grad_norm": 0.6791189255121085,
      "learning_rate": 1.604628323845948e-05,
      "loss": 11.1134,
      "step": 18816
    },
    {
      "epoch": 3.2904043715846996,
      "grad_norm": 0.6216618302762318,
      "learning_rate": 1.603859043867121e-05,
      "loss": 11.7342,
      "step": 18817
    },
    {
      "epoch": 3.2905792349726775,
      "grad_norm": 0.6949325320760368,
      "learning_rate": 1.6030899322566995e-05,
      "loss": 11.2955,
      "step": 18818
    },
    {
      "epoch": 3.290754098360656,
      "grad_norm": 0.6419701568740875,
      "learning_rate": 1.602320989030107e-05,
      "loss": 11.0566,
      "step": 18819
    },
    {
      "epoch": 3.2909289617486337,
      "grad_norm": 0.7457505911570477,
      "learning_rate": 1.6015522142027628e-05,
      "loss": 11.3081,
      "step": 18820
    },
    {
      "epoch": 3.291103825136612,
      "grad_norm": 0.6772837661590858,
      "learning_rate": 1.600783607790083e-05,
      "loss": 11.3349,
      "step": 18821
    },
    {
      "epoch": 3.2912786885245904,
      "grad_norm": 0.6274895069356653,
      "learning_rate": 1.6000151698074816e-05,
      "loss": 11.385,
      "step": 18822
    },
    {
      "epoch": 3.2914535519125683,
      "grad_norm": 0.7664780818932974,
      "learning_rate": 1.599246900270367e-05,
      "loss": 11.2542,
      "step": 18823
    },
    {
      "epoch": 3.2916284153005466,
      "grad_norm": 0.6772849918745887,
      "learning_rate": 1.598478799194145e-05,
      "loss": 11.302,
      "step": 18824
    },
    {
      "epoch": 3.2918032786885245,
      "grad_norm": 0.6827101831280837,
      "learning_rate": 1.5977108665942208e-05,
      "loss": 11.5031,
      "step": 18825
    },
    {
      "epoch": 3.291978142076503,
      "grad_norm": 0.716308410495947,
      "learning_rate": 1.5969431024859903e-05,
      "loss": 11.5426,
      "step": 18826
    },
    {
      "epoch": 3.2921530054644808,
      "grad_norm": 0.6575423066035727,
      "learning_rate": 1.5961755068848526e-05,
      "loss": 11.317,
      "step": 18827
    },
    {
      "epoch": 3.292327868852459,
      "grad_norm": 0.6572914165301884,
      "learning_rate": 1.595408079806201e-05,
      "loss": 11.5894,
      "step": 18828
    },
    {
      "epoch": 3.292502732240437,
      "grad_norm": 0.6921080743864003,
      "learning_rate": 1.5946408212654183e-05,
      "loss": 11.343,
      "step": 18829
    },
    {
      "epoch": 3.2926775956284153,
      "grad_norm": 0.7646490460008546,
      "learning_rate": 1.5938737312778972e-05,
      "loss": 11.2632,
      "step": 18830
    },
    {
      "epoch": 3.2928524590163937,
      "grad_norm": 0.6409419511540864,
      "learning_rate": 1.5931068098590186e-05,
      "loss": 11.5439,
      "step": 18831
    },
    {
      "epoch": 3.2930273224043716,
      "grad_norm": 0.6444976216867216,
      "learning_rate": 1.592340057024161e-05,
      "loss": 11.5397,
      "step": 18832
    },
    {
      "epoch": 3.29320218579235,
      "grad_norm": 0.6484538775662085,
      "learning_rate": 1.591573472788699e-05,
      "loss": 11.6195,
      "step": 18833
    },
    {
      "epoch": 3.293377049180328,
      "grad_norm": 0.7387693250027392,
      "learning_rate": 1.590807057168007e-05,
      "loss": 11.3358,
      "step": 18834
    },
    {
      "epoch": 3.293551912568306,
      "grad_norm": 0.7014713451639708,
      "learning_rate": 1.5900408101774532e-05,
      "loss": 11.5305,
      "step": 18835
    },
    {
      "epoch": 3.293726775956284,
      "grad_norm": 0.6253561744813491,
      "learning_rate": 1.589274731832403e-05,
      "loss": 11.1408,
      "step": 18836
    },
    {
      "epoch": 3.2939016393442624,
      "grad_norm": 0.7160226579272951,
      "learning_rate": 1.5885088221482204e-05,
      "loss": 11.4747,
      "step": 18837
    },
    {
      "epoch": 3.2940765027322403,
      "grad_norm": 0.7084141073336915,
      "learning_rate": 1.587743081140258e-05,
      "loss": 11.3783,
      "step": 18838
    },
    {
      "epoch": 3.2942513661202186,
      "grad_norm": 0.7055674832930916,
      "learning_rate": 1.5869775088238793e-05,
      "loss": 11.355,
      "step": 18839
    },
    {
      "epoch": 3.294426229508197,
      "grad_norm": 0.6822890043629717,
      "learning_rate": 1.586212105214432e-05,
      "loss": 11.4981,
      "step": 18840
    },
    {
      "epoch": 3.294601092896175,
      "grad_norm": 0.6684328640214545,
      "learning_rate": 1.5854468703272663e-05,
      "loss": 11.45,
      "step": 18841
    },
    {
      "epoch": 3.2947759562841528,
      "grad_norm": 0.6419543236270397,
      "learning_rate": 1.584681804177729e-05,
      "loss": 11.2779,
      "step": 18842
    },
    {
      "epoch": 3.294950819672131,
      "grad_norm": 0.6804208319813507,
      "learning_rate": 1.5839169067811555e-05,
      "loss": 11.5021,
      "step": 18843
    },
    {
      "epoch": 3.2951256830601094,
      "grad_norm": 0.7032954241996331,
      "learning_rate": 1.58315217815289e-05,
      "loss": 11.3717,
      "step": 18844
    },
    {
      "epoch": 3.2953005464480873,
      "grad_norm": 0.6576149496637763,
      "learning_rate": 1.5823876183082675e-05,
      "loss": 11.6077,
      "step": 18845
    },
    {
      "epoch": 3.2954754098360657,
      "grad_norm": 0.6307267643087887,
      "learning_rate": 1.5816232272626208e-05,
      "loss": 11.3541,
      "step": 18846
    },
    {
      "epoch": 3.2956502732240436,
      "grad_norm": 0.6647992982675274,
      "learning_rate": 1.5808590050312743e-05,
      "loss": 11.3988,
      "step": 18847
    },
    {
      "epoch": 3.295825136612022,
      "grad_norm": 0.6725015452916938,
      "learning_rate": 1.5800949516295516e-05,
      "loss": 11.2974,
      "step": 18848
    },
    {
      "epoch": 3.296,
      "grad_norm": 0.6485754229313603,
      "learning_rate": 1.5793310670727814e-05,
      "loss": 11.5234,
      "step": 18849
    },
    {
      "epoch": 3.296174863387978,
      "grad_norm": 0.6665965304803172,
      "learning_rate": 1.5785673513762765e-05,
      "loss": 11.4802,
      "step": 18850
    },
    {
      "epoch": 3.296349726775956,
      "grad_norm": 0.7530070998201044,
      "learning_rate": 1.577803804555357e-05,
      "loss": 11.3661,
      "step": 18851
    },
    {
      "epoch": 3.2965245901639344,
      "grad_norm": 0.6882783123059848,
      "learning_rate": 1.577040426625328e-05,
      "loss": 11.4283,
      "step": 18852
    },
    {
      "epoch": 3.2966994535519127,
      "grad_norm": 0.65667061427827,
      "learning_rate": 1.5762772176014982e-05,
      "loss": 11.5824,
      "step": 18853
    },
    {
      "epoch": 3.2968743169398906,
      "grad_norm": 0.6651859457764548,
      "learning_rate": 1.5755141774991765e-05,
      "loss": 11.3318,
      "step": 18854
    },
    {
      "epoch": 3.297049180327869,
      "grad_norm": 0.743488499566041,
      "learning_rate": 1.5747513063336638e-05,
      "loss": 11.3382,
      "step": 18855
    },
    {
      "epoch": 3.297224043715847,
      "grad_norm": 0.7660010488344083,
      "learning_rate": 1.5739886041202545e-05,
      "loss": 11.4085,
      "step": 18856
    },
    {
      "epoch": 3.297398907103825,
      "grad_norm": 0.634899020210012,
      "learning_rate": 1.573226070874243e-05,
      "loss": 11.7776,
      "step": 18857
    },
    {
      "epoch": 3.297573770491803,
      "grad_norm": 0.8763173212562584,
      "learning_rate": 1.5724637066109248e-05,
      "loss": 11.5825,
      "step": 18858
    },
    {
      "epoch": 3.2977486338797815,
      "grad_norm": 0.6887940244235647,
      "learning_rate": 1.5717015113455845e-05,
      "loss": 11.5223,
      "step": 18859
    },
    {
      "epoch": 3.2979234972677594,
      "grad_norm": 0.7466251789116801,
      "learning_rate": 1.57093948509351e-05,
      "loss": 11.4741,
      "step": 18860
    },
    {
      "epoch": 3.2980983606557377,
      "grad_norm": 0.69473305684572,
      "learning_rate": 1.5701776278699766e-05,
      "loss": 11.3625,
      "step": 18861
    },
    {
      "epoch": 3.298273224043716,
      "grad_norm": 0.7134812656933427,
      "learning_rate": 1.569415939690263e-05,
      "loss": 11.2964,
      "step": 18862
    },
    {
      "epoch": 3.298448087431694,
      "grad_norm": 0.6586943266359309,
      "learning_rate": 1.5686544205696473e-05,
      "loss": 11.3641,
      "step": 18863
    },
    {
      "epoch": 3.2986229508196723,
      "grad_norm": 0.6665981755147853,
      "learning_rate": 1.5678930705234006e-05,
      "loss": 11.2762,
      "step": 18864
    },
    {
      "epoch": 3.29879781420765,
      "grad_norm": 0.658827710586091,
      "learning_rate": 1.5671318895667854e-05,
      "loss": 11.4091,
      "step": 18865
    },
    {
      "epoch": 3.2989726775956285,
      "grad_norm": 0.6956807601833653,
      "learning_rate": 1.5663708777150675e-05,
      "loss": 11.3779,
      "step": 18866
    },
    {
      "epoch": 3.2991475409836064,
      "grad_norm": 0.7381799416505541,
      "learning_rate": 1.5656100349835057e-05,
      "loss": 11.3311,
      "step": 18867
    },
    {
      "epoch": 3.2993224043715847,
      "grad_norm": 0.675783548446341,
      "learning_rate": 1.564849361387363e-05,
      "loss": 11.5711,
      "step": 18868
    },
    {
      "epoch": 3.2994972677595626,
      "grad_norm": 0.731865246089537,
      "learning_rate": 1.564088856941892e-05,
      "loss": 11.3134,
      "step": 18869
    },
    {
      "epoch": 3.299672131147541,
      "grad_norm": 0.6588595357168193,
      "learning_rate": 1.5633285216623385e-05,
      "loss": 11.5508,
      "step": 18870
    },
    {
      "epoch": 3.2998469945355193,
      "grad_norm": 0.7790102428873787,
      "learning_rate": 1.5625683555639514e-05,
      "loss": 11.6893,
      "step": 18871
    },
    {
      "epoch": 3.3000218579234972,
      "grad_norm": 0.7368430512287792,
      "learning_rate": 1.5618083586619738e-05,
      "loss": 11.1042,
      "step": 18872
    },
    {
      "epoch": 3.3001967213114756,
      "grad_norm": 0.7061060166946884,
      "learning_rate": 1.5610485309716514e-05,
      "loss": 11.3593,
      "step": 18873
    },
    {
      "epoch": 3.3003715846994535,
      "grad_norm": 0.6472978111891475,
      "learning_rate": 1.5602888725082154e-05,
      "loss": 11.2232,
      "step": 18874
    },
    {
      "epoch": 3.300546448087432,
      "grad_norm": 0.6810927041472581,
      "learning_rate": 1.5595293832868995e-05,
      "loss": 11.2832,
      "step": 18875
    },
    {
      "epoch": 3.3007213114754097,
      "grad_norm": 0.7181861017261931,
      "learning_rate": 1.5587700633229363e-05,
      "loss": 11.5698,
      "step": 18876
    },
    {
      "epoch": 3.300896174863388,
      "grad_norm": 0.7100291915653912,
      "learning_rate": 1.558010912631549e-05,
      "loss": 11.5315,
      "step": 18877
    },
    {
      "epoch": 3.301071038251366,
      "grad_norm": 0.6966232171741187,
      "learning_rate": 1.5572519312279664e-05,
      "loss": 11.5945,
      "step": 18878
    },
    {
      "epoch": 3.3012459016393443,
      "grad_norm": 0.6749479617512558,
      "learning_rate": 1.5564931191274045e-05,
      "loss": 11.5044,
      "step": 18879
    },
    {
      "epoch": 3.3014207650273226,
      "grad_norm": 0.7196912715642313,
      "learning_rate": 1.5557344763450797e-05,
      "loss": 11.5305,
      "step": 18880
    },
    {
      "epoch": 3.3015956284153005,
      "grad_norm": 0.6977413616231362,
      "learning_rate": 1.5549760028962034e-05,
      "loss": 11.3131,
      "step": 18881
    },
    {
      "epoch": 3.301770491803279,
      "grad_norm": 0.716242473842014,
      "learning_rate": 1.5542176987959934e-05,
      "loss": 11.4411,
      "step": 18882
    },
    {
      "epoch": 3.3019453551912568,
      "grad_norm": 0.7356388294694485,
      "learning_rate": 1.5534595640596485e-05,
      "loss": 11.5691,
      "step": 18883
    },
    {
      "epoch": 3.302120218579235,
      "grad_norm": 0.6776148189719862,
      "learning_rate": 1.552701598702372e-05,
      "loss": 11.5213,
      "step": 18884
    },
    {
      "epoch": 3.302295081967213,
      "grad_norm": 0.6559222121824878,
      "learning_rate": 1.5519438027393662e-05,
      "loss": 11.5904,
      "step": 18885
    },
    {
      "epoch": 3.3024699453551913,
      "grad_norm": 0.7283553289403063,
      "learning_rate": 1.5511861761858238e-05,
      "loss": 11.4265,
      "step": 18886
    },
    {
      "epoch": 3.3026448087431692,
      "grad_norm": 0.7328999837779715,
      "learning_rate": 1.5504287190569444e-05,
      "loss": 11.3324,
      "step": 18887
    },
    {
      "epoch": 3.3028196721311476,
      "grad_norm": 0.6831771109354264,
      "learning_rate": 1.5496714313679094e-05,
      "loss": 11.4907,
      "step": 18888
    },
    {
      "epoch": 3.302994535519126,
      "grad_norm": 0.6863108967878705,
      "learning_rate": 1.548914313133909e-05,
      "loss": 11.3487,
      "step": 18889
    },
    {
      "epoch": 3.303169398907104,
      "grad_norm": 0.611592803238787,
      "learning_rate": 1.5481573643701252e-05,
      "loss": 11.3386,
      "step": 18890
    },
    {
      "epoch": 3.303344262295082,
      "grad_norm": 0.6369979738426939,
      "learning_rate": 1.547400585091736e-05,
      "loss": 11.2892,
      "step": 18891
    },
    {
      "epoch": 3.30351912568306,
      "grad_norm": 0.6419710274689934,
      "learning_rate": 1.5466439753139184e-05,
      "loss": 11.2466,
      "step": 18892
    },
    {
      "epoch": 3.3036939890710384,
      "grad_norm": 0.6941698354524679,
      "learning_rate": 1.5458875350518443e-05,
      "loss": 11.505,
      "step": 18893
    },
    {
      "epoch": 3.3038688524590163,
      "grad_norm": 0.6600430413089398,
      "learning_rate": 1.5451312643206827e-05,
      "loss": 11.5515,
      "step": 18894
    },
    {
      "epoch": 3.3040437158469946,
      "grad_norm": 0.6866345610918285,
      "learning_rate": 1.5443751631355996e-05,
      "loss": 11.3164,
      "step": 18895
    },
    {
      "epoch": 3.3042185792349725,
      "grad_norm": 0.6797876893822717,
      "learning_rate": 1.543619231511756e-05,
      "loss": 11.6346,
      "step": 18896
    },
    {
      "epoch": 3.304393442622951,
      "grad_norm": 0.6586532520291494,
      "learning_rate": 1.5428634694643117e-05,
      "loss": 11.3828,
      "step": 18897
    },
    {
      "epoch": 3.3045683060109288,
      "grad_norm": 0.6758883504107678,
      "learning_rate": 1.542107877008421e-05,
      "loss": 11.4283,
      "step": 18898
    },
    {
      "epoch": 3.304743169398907,
      "grad_norm": 0.6787238881534176,
      "learning_rate": 1.541352454159237e-05,
      "loss": 11.4482,
      "step": 18899
    },
    {
      "epoch": 3.304918032786885,
      "grad_norm": 0.750466871328221,
      "learning_rate": 1.540597200931908e-05,
      "loss": 11.3987,
      "step": 18900
    },
    {
      "epoch": 3.3050928961748633,
      "grad_norm": 0.6862586362656045,
      "learning_rate": 1.5398421173415778e-05,
      "loss": 11.5555,
      "step": 18901
    },
    {
      "epoch": 3.3052677595628417,
      "grad_norm": 0.6861497589250264,
      "learning_rate": 1.5390872034033898e-05,
      "loss": 11.3233,
      "step": 18902
    },
    {
      "epoch": 3.3054426229508196,
      "grad_norm": 0.7548946341658542,
      "learning_rate": 1.538332459132482e-05,
      "loss": 11.2939,
      "step": 18903
    },
    {
      "epoch": 3.305617486338798,
      "grad_norm": 0.6857441085291571,
      "learning_rate": 1.537577884543989e-05,
      "loss": 11.5758,
      "step": 18904
    },
    {
      "epoch": 3.305792349726776,
      "grad_norm": 0.6807489697541272,
      "learning_rate": 1.5368234796530433e-05,
      "loss": 11.5102,
      "step": 18905
    },
    {
      "epoch": 3.305967213114754,
      "grad_norm": 0.7157977601161781,
      "learning_rate": 1.5360692444747715e-05,
      "loss": 11.4135,
      "step": 18906
    },
    {
      "epoch": 3.306142076502732,
      "grad_norm": 0.7390280374749808,
      "learning_rate": 1.535315179024298e-05,
      "loss": 11.4756,
      "step": 18907
    },
    {
      "epoch": 3.3063169398907104,
      "grad_norm": 0.6818163236864326,
      "learning_rate": 1.5345612833167466e-05,
      "loss": 11.492,
      "step": 18908
    },
    {
      "epoch": 3.3064918032786883,
      "grad_norm": 0.6656328637262843,
      "learning_rate": 1.5338075573672338e-05,
      "loss": 11.492,
      "step": 18909
    },
    {
      "epoch": 3.3066666666666666,
      "grad_norm": 0.6886787925659513,
      "learning_rate": 1.5330540011908735e-05,
      "loss": 11.2965,
      "step": 18910
    },
    {
      "epoch": 3.306841530054645,
      "grad_norm": 0.6363011853290248,
      "learning_rate": 1.5323006148027775e-05,
      "loss": 11.5558,
      "step": 18911
    },
    {
      "epoch": 3.307016393442623,
      "grad_norm": 0.6413160936935106,
      "learning_rate": 1.531547398218053e-05,
      "loss": 11.3392,
      "step": 18912
    },
    {
      "epoch": 3.3071912568306012,
      "grad_norm": 0.6836411858340646,
      "learning_rate": 1.5307943514518052e-05,
      "loss": 11.5188,
      "step": 18913
    },
    {
      "epoch": 3.307366120218579,
      "grad_norm": 0.6729481028260427,
      "learning_rate": 1.530041474519135e-05,
      "loss": 11.4831,
      "step": 18914
    },
    {
      "epoch": 3.3075409836065575,
      "grad_norm": 0.6549834300435409,
      "learning_rate": 1.5292887674351385e-05,
      "loss": 11.6052,
      "step": 18915
    },
    {
      "epoch": 3.3077158469945354,
      "grad_norm": 0.7350149251734136,
      "learning_rate": 1.5285362302149108e-05,
      "loss": 11.6488,
      "step": 18916
    },
    {
      "epoch": 3.3078907103825137,
      "grad_norm": 0.6683793294977454,
      "learning_rate": 1.5277838628735432e-05,
      "loss": 11.3101,
      "step": 18917
    },
    {
      "epoch": 3.3080655737704916,
      "grad_norm": 0.6270606062315579,
      "learning_rate": 1.5270316654261217e-05,
      "loss": 11.5965,
      "step": 18918
    },
    {
      "epoch": 3.30824043715847,
      "grad_norm": 0.651324100710125,
      "learning_rate": 1.526279637887731e-05,
      "loss": 11.5745,
      "step": 18919
    },
    {
      "epoch": 3.3084153005464483,
      "grad_norm": 0.6791630688196876,
      "learning_rate": 1.5255277802734513e-05,
      "loss": 11.2632,
      "step": 18920
    },
    {
      "epoch": 3.308590163934426,
      "grad_norm": 0.7114936371614106,
      "learning_rate": 1.5247760925983601e-05,
      "loss": 11.4562,
      "step": 18921
    },
    {
      "epoch": 3.3087650273224045,
      "grad_norm": 0.6304261771939095,
      "learning_rate": 1.5240245748775294e-05,
      "loss": 11.4037,
      "step": 18922
    },
    {
      "epoch": 3.3089398907103824,
      "grad_norm": 0.6917752991489055,
      "learning_rate": 1.523273227126032e-05,
      "loss": 11.4466,
      "step": 18923
    },
    {
      "epoch": 3.3091147540983608,
      "grad_norm": 0.6966368448310326,
      "learning_rate": 1.5225220493589354e-05,
      "loss": 11.591,
      "step": 18924
    },
    {
      "epoch": 3.3092896174863387,
      "grad_norm": 0.7431883621894613,
      "learning_rate": 1.5217710415912956e-05,
      "loss": 11.408,
      "step": 18925
    },
    {
      "epoch": 3.309464480874317,
      "grad_norm": 0.7043581980929791,
      "learning_rate": 1.5210202038381805e-05,
      "loss": 11.4628,
      "step": 18926
    },
    {
      "epoch": 3.309639344262295,
      "grad_norm": 0.700857594226178,
      "learning_rate": 1.5202695361146447e-05,
      "loss": 11.3236,
      "step": 18927
    },
    {
      "epoch": 3.3098142076502732,
      "grad_norm": 0.6720318729712527,
      "learning_rate": 1.5195190384357404e-05,
      "loss": 11.4945,
      "step": 18928
    },
    {
      "epoch": 3.3099890710382516,
      "grad_norm": 0.6881411859180349,
      "learning_rate": 1.5187687108165182e-05,
      "loss": 11.4952,
      "step": 18929
    },
    {
      "epoch": 3.3101639344262295,
      "grad_norm": 0.6450472900005433,
      "learning_rate": 1.5180185532720237e-05,
      "loss": 11.5789,
      "step": 18930
    },
    {
      "epoch": 3.310338797814208,
      "grad_norm": 0.6662129730524593,
      "learning_rate": 1.5172685658173003e-05,
      "loss": 11.3534,
      "step": 18931
    },
    {
      "epoch": 3.3105136612021857,
      "grad_norm": 0.7988713676167795,
      "learning_rate": 1.5165187484673882e-05,
      "loss": 11.1992,
      "step": 18932
    },
    {
      "epoch": 3.310688524590164,
      "grad_norm": 0.6785333252795556,
      "learning_rate": 1.5157691012373253e-05,
      "loss": 11.3215,
      "step": 18933
    },
    {
      "epoch": 3.310863387978142,
      "grad_norm": 0.6378013365735539,
      "learning_rate": 1.5150196241421366e-05,
      "loss": 11.2533,
      "step": 18934
    },
    {
      "epoch": 3.3110382513661203,
      "grad_norm": 0.6551522572432249,
      "learning_rate": 1.5142703171968598e-05,
      "loss": 11.5155,
      "step": 18935
    },
    {
      "epoch": 3.311213114754098,
      "grad_norm": 0.7482480724219888,
      "learning_rate": 1.5135211804165173e-05,
      "loss": 11.3994,
      "step": 18936
    },
    {
      "epoch": 3.3113879781420765,
      "grad_norm": 0.7343095275496587,
      "learning_rate": 1.5127722138161315e-05,
      "loss": 11.5037,
      "step": 18937
    },
    {
      "epoch": 3.311562841530055,
      "grad_norm": 0.6499169603776975,
      "learning_rate": 1.512023417410725e-05,
      "loss": 11.6432,
      "step": 18938
    },
    {
      "epoch": 3.3117377049180328,
      "grad_norm": 0.6640682448057602,
      "learning_rate": 1.5112747912153057e-05,
      "loss": 11.3665,
      "step": 18939
    },
    {
      "epoch": 3.311912568306011,
      "grad_norm": 0.8154064836521709,
      "learning_rate": 1.5105263352448928e-05,
      "loss": 11.3385,
      "step": 18940
    },
    {
      "epoch": 3.312087431693989,
      "grad_norm": 0.6970041404950283,
      "learning_rate": 1.509778049514493e-05,
      "loss": 11.2275,
      "step": 18941
    },
    {
      "epoch": 3.3122622950819673,
      "grad_norm": 0.6916855914244823,
      "learning_rate": 1.5090299340391144e-05,
      "loss": 11.3621,
      "step": 18942
    },
    {
      "epoch": 3.3124371584699452,
      "grad_norm": 0.6729132827203436,
      "learning_rate": 1.5082819888337541e-05,
      "loss": 11.6235,
      "step": 18943
    },
    {
      "epoch": 3.3126120218579236,
      "grad_norm": 0.6736986624495233,
      "learning_rate": 1.5075342139134107e-05,
      "loss": 11.3054,
      "step": 18944
    },
    {
      "epoch": 3.3127868852459015,
      "grad_norm": 0.7605199138990784,
      "learning_rate": 1.5067866092930838e-05,
      "loss": 11.0821,
      "step": 18945
    },
    {
      "epoch": 3.31296174863388,
      "grad_norm": 0.6801298130812263,
      "learning_rate": 1.5060391749877622e-05,
      "loss": 11.579,
      "step": 18946
    },
    {
      "epoch": 3.313136612021858,
      "grad_norm": 0.6797898178658375,
      "learning_rate": 1.5052919110124374e-05,
      "loss": 11.5208,
      "step": 18947
    },
    {
      "epoch": 3.313311475409836,
      "grad_norm": 0.6768454151199373,
      "learning_rate": 1.5045448173820908e-05,
      "loss": 11.0835,
      "step": 18948
    },
    {
      "epoch": 3.3134863387978144,
      "grad_norm": 0.7932254266413467,
      "learning_rate": 1.5037978941117015e-05,
      "loss": 11.6079,
      "step": 18949
    },
    {
      "epoch": 3.3136612021857923,
      "grad_norm": 0.6498227093603526,
      "learning_rate": 1.503051141216255e-05,
      "loss": 11.3898,
      "step": 18950
    },
    {
      "epoch": 3.3138360655737706,
      "grad_norm": 0.7808860668094021,
      "learning_rate": 1.5023045587107231e-05,
      "loss": 11.2929,
      "step": 18951
    },
    {
      "epoch": 3.3140109289617485,
      "grad_norm": 0.7038926997464783,
      "learning_rate": 1.5015581466100747e-05,
      "loss": 11.5344,
      "step": 18952
    },
    {
      "epoch": 3.314185792349727,
      "grad_norm": 0.6629714889615861,
      "learning_rate": 1.5008119049292757e-05,
      "loss": 11.5949,
      "step": 18953
    },
    {
      "epoch": 3.3143606557377048,
      "grad_norm": 0.6976503686422867,
      "learning_rate": 1.5000658336832974e-05,
      "loss": 11.524,
      "step": 18954
    },
    {
      "epoch": 3.314535519125683,
      "grad_norm": 0.7176952689757968,
      "learning_rate": 1.4993199328870967e-05,
      "loss": 11.5088,
      "step": 18955
    },
    {
      "epoch": 3.314710382513661,
      "grad_norm": 0.7072438976816485,
      "learning_rate": 1.4985742025556349e-05,
      "loss": 11.4005,
      "step": 18956
    },
    {
      "epoch": 3.3148852459016394,
      "grad_norm": 0.6852457415403855,
      "learning_rate": 1.4978286427038601e-05,
      "loss": 11.5707,
      "step": 18957
    },
    {
      "epoch": 3.3150601092896173,
      "grad_norm": 0.7049388096779992,
      "learning_rate": 1.4970832533467239e-05,
      "loss": 11.3649,
      "step": 18958
    },
    {
      "epoch": 3.3152349726775956,
      "grad_norm": 0.6577975868410111,
      "learning_rate": 1.4963380344991785e-05,
      "loss": 11.0582,
      "step": 18959
    },
    {
      "epoch": 3.315409836065574,
      "grad_norm": 0.6596410212753993,
      "learning_rate": 1.4955929861761676e-05,
      "loss": 11.3797,
      "step": 18960
    },
    {
      "epoch": 3.315584699453552,
      "grad_norm": 0.6781888313700152,
      "learning_rate": 1.494848108392627e-05,
      "loss": 11.3445,
      "step": 18961
    },
    {
      "epoch": 3.31575956284153,
      "grad_norm": 0.9655205630890787,
      "learning_rate": 1.4941034011634947e-05,
      "loss": 11.5036,
      "step": 18962
    },
    {
      "epoch": 3.315934426229508,
      "grad_norm": 0.6492649878786156,
      "learning_rate": 1.4933588645037044e-05,
      "loss": 11.594,
      "step": 18963
    },
    {
      "epoch": 3.3161092896174864,
      "grad_norm": 0.6360864032396228,
      "learning_rate": 1.4926144984281908e-05,
      "loss": 11.6315,
      "step": 18964
    },
    {
      "epoch": 3.3162841530054643,
      "grad_norm": 0.7218641747986476,
      "learning_rate": 1.4918703029518788e-05,
      "loss": 11.3013,
      "step": 18965
    },
    {
      "epoch": 3.3164590163934426,
      "grad_norm": 0.7050830647550541,
      "learning_rate": 1.4911262780896884e-05,
      "loss": 11.29,
      "step": 18966
    },
    {
      "epoch": 3.3166338797814205,
      "grad_norm": 0.6652284087887431,
      "learning_rate": 1.4903824238565422e-05,
      "loss": 11.1885,
      "step": 18967
    },
    {
      "epoch": 3.316808743169399,
      "grad_norm": 0.6659925328376591,
      "learning_rate": 1.489638740267354e-05,
      "loss": 11.2849,
      "step": 18968
    },
    {
      "epoch": 3.3169836065573772,
      "grad_norm": 0.6424107977000806,
      "learning_rate": 1.4888952273370437e-05,
      "loss": 11.4995,
      "step": 18969
    },
    {
      "epoch": 3.317158469945355,
      "grad_norm": 0.702077017042473,
      "learning_rate": 1.4881518850805153e-05,
      "loss": 11.4975,
      "step": 18970
    },
    {
      "epoch": 3.3173333333333335,
      "grad_norm": 0.6594438409586698,
      "learning_rate": 1.4874087135126758e-05,
      "loss": 11.5703,
      "step": 18971
    },
    {
      "epoch": 3.3175081967213114,
      "grad_norm": 0.7188053859117232,
      "learning_rate": 1.4866657126484284e-05,
      "loss": 11.1684,
      "step": 18972
    },
    {
      "epoch": 3.3176830601092897,
      "grad_norm": 0.661571461103919,
      "learning_rate": 1.4859228825026706e-05,
      "loss": 11.5232,
      "step": 18973
    },
    {
      "epoch": 3.3178579234972676,
      "grad_norm": 0.6845780718284138,
      "learning_rate": 1.4851802230903044e-05,
      "loss": 11.1166,
      "step": 18974
    },
    {
      "epoch": 3.318032786885246,
      "grad_norm": 0.6678749808880573,
      "learning_rate": 1.4844377344262172e-05,
      "loss": 11.6142,
      "step": 18975
    },
    {
      "epoch": 3.318207650273224,
      "grad_norm": 0.7408236036993064,
      "learning_rate": 1.4836954165253004e-05,
      "loss": 11.4964,
      "step": 18976
    },
    {
      "epoch": 3.318382513661202,
      "grad_norm": 0.6867626360361582,
      "learning_rate": 1.482953269402435e-05,
      "loss": 11.5192,
      "step": 18977
    },
    {
      "epoch": 3.3185573770491805,
      "grad_norm": 0.7489037378102513,
      "learning_rate": 1.482211293072513e-05,
      "loss": 11.3844,
      "step": 18978
    },
    {
      "epoch": 3.3187322404371584,
      "grad_norm": 0.6953919376591301,
      "learning_rate": 1.4814694875504053e-05,
      "loss": 11.4001,
      "step": 18979
    },
    {
      "epoch": 3.3189071038251368,
      "grad_norm": 0.6711313799742006,
      "learning_rate": 1.4807278528509893e-05,
      "loss": 11.6948,
      "step": 18980
    },
    {
      "epoch": 3.3190819672131147,
      "grad_norm": 0.6965181795201567,
      "learning_rate": 1.4799863889891385e-05,
      "loss": 11.6971,
      "step": 18981
    },
    {
      "epoch": 3.319256830601093,
      "grad_norm": 0.6700085438011776,
      "learning_rate": 1.4792450959797165e-05,
      "loss": 11.4187,
      "step": 18982
    },
    {
      "epoch": 3.319431693989071,
      "grad_norm": 0.6688206683637464,
      "learning_rate": 1.4785039738375983e-05,
      "loss": 11.2976,
      "step": 18983
    },
    {
      "epoch": 3.3196065573770492,
      "grad_norm": 0.678169338924187,
      "learning_rate": 1.4777630225776374e-05,
      "loss": 11.8414,
      "step": 18984
    },
    {
      "epoch": 3.319781420765027,
      "grad_norm": 0.66453538685186,
      "learning_rate": 1.4770222422146952e-05,
      "loss": 11.4036,
      "step": 18985
    },
    {
      "epoch": 3.3199562841530055,
      "grad_norm": 0.641724327129624,
      "learning_rate": 1.4762816327636241e-05,
      "loss": 11.1157,
      "step": 18986
    },
    {
      "epoch": 3.320131147540984,
      "grad_norm": 0.6061248760894691,
      "learning_rate": 1.4755411942392793e-05,
      "loss": 11.2756,
      "step": 18987
    },
    {
      "epoch": 3.3203060109289617,
      "grad_norm": 0.6315730140499464,
      "learning_rate": 1.4748009266565066e-05,
      "loss": 11.4077,
      "step": 18988
    },
    {
      "epoch": 3.32048087431694,
      "grad_norm": 0.6841757571063136,
      "learning_rate": 1.4740608300301506e-05,
      "loss": 11.6196,
      "step": 18989
    },
    {
      "epoch": 3.320655737704918,
      "grad_norm": 0.6488890676977045,
      "learning_rate": 1.473320904375054e-05,
      "loss": 11.2253,
      "step": 18990
    },
    {
      "epoch": 3.3208306010928963,
      "grad_norm": 0.6091432493311636,
      "learning_rate": 1.4725811497060527e-05,
      "loss": 11.6155,
      "step": 18991
    },
    {
      "epoch": 3.321005464480874,
      "grad_norm": 0.6855937603656833,
      "learning_rate": 1.4718415660379826e-05,
      "loss": 11.5071,
      "step": 18992
    },
    {
      "epoch": 3.3211803278688525,
      "grad_norm": 0.6801558508523874,
      "learning_rate": 1.4711021533856728e-05,
      "loss": 11.4788,
      "step": 18993
    },
    {
      "epoch": 3.3213551912568304,
      "grad_norm": 0.7239373147299146,
      "learning_rate": 1.4703629117639527e-05,
      "loss": 11.5545,
      "step": 18994
    },
    {
      "epoch": 3.3215300546448088,
      "grad_norm": 0.6795068865062261,
      "learning_rate": 1.4696238411876461e-05,
      "loss": 11.5934,
      "step": 18995
    },
    {
      "epoch": 3.321704918032787,
      "grad_norm": 0.7333035678674135,
      "learning_rate": 1.4688849416715734e-05,
      "loss": 11.4269,
      "step": 18996
    },
    {
      "epoch": 3.321879781420765,
      "grad_norm": 0.6399225718414845,
      "learning_rate": 1.46814621323055e-05,
      "loss": 11.2961,
      "step": 18997
    },
    {
      "epoch": 3.3220546448087434,
      "grad_norm": 0.7278302232278688,
      "learning_rate": 1.4674076558793925e-05,
      "loss": 11.4724,
      "step": 18998
    },
    {
      "epoch": 3.3222295081967212,
      "grad_norm": 0.6826479750931769,
      "learning_rate": 1.4666692696329099e-05,
      "loss": 11.0019,
      "step": 18999
    },
    {
      "epoch": 3.3224043715846996,
      "grad_norm": 0.708254331154218,
      "learning_rate": 1.465931054505909e-05,
      "loss": 11.1661,
      "step": 19000
    },
    {
      "epoch": 3.3225792349726775,
      "grad_norm": 0.6719203353776267,
      "learning_rate": 1.4651930105131939e-05,
      "loss": 11.1929,
      "step": 19001
    },
    {
      "epoch": 3.322754098360656,
      "grad_norm": 0.7287418148766267,
      "learning_rate": 1.4644551376695636e-05,
      "loss": 11.54,
      "step": 19002
    },
    {
      "epoch": 3.3229289617486337,
      "grad_norm": 0.6725627122160254,
      "learning_rate": 1.4637174359898165e-05,
      "loss": 11.5108,
      "step": 19003
    },
    {
      "epoch": 3.323103825136612,
      "grad_norm": 0.7090139593936159,
      "learning_rate": 1.462979905488744e-05,
      "loss": 11.379,
      "step": 19004
    },
    {
      "epoch": 3.3232786885245904,
      "grad_norm": 0.7436545900075834,
      "learning_rate": 1.4622425461811363e-05,
      "loss": 11.4065,
      "step": 19005
    },
    {
      "epoch": 3.3234535519125683,
      "grad_norm": 0.6492900663505379,
      "learning_rate": 1.4615053580817795e-05,
      "loss": 11.3161,
      "step": 19006
    },
    {
      "epoch": 3.323628415300546,
      "grad_norm": 0.6548844016080321,
      "learning_rate": 1.4607683412054574e-05,
      "loss": 11.2393,
      "step": 19007
    },
    {
      "epoch": 3.3238032786885245,
      "grad_norm": 0.7125021638371669,
      "learning_rate": 1.4600314955669491e-05,
      "loss": 11.5203,
      "step": 19008
    },
    {
      "epoch": 3.323978142076503,
      "grad_norm": 0.7275380903742333,
      "learning_rate": 1.4592948211810297e-05,
      "loss": 11.448,
      "step": 19009
    },
    {
      "epoch": 3.324153005464481,
      "grad_norm": 0.7016284659277938,
      "learning_rate": 1.458558318062473e-05,
      "loss": 11.0971,
      "step": 19010
    },
    {
      "epoch": 3.324327868852459,
      "grad_norm": 0.7035750669830376,
      "learning_rate": 1.4578219862260478e-05,
      "loss": 11.2812,
      "step": 19011
    },
    {
      "epoch": 3.324502732240437,
      "grad_norm": 0.6903022903790625,
      "learning_rate": 1.4570858256865194e-05,
      "loss": 11.6272,
      "step": 19012
    },
    {
      "epoch": 3.3246775956284154,
      "grad_norm": 0.6165226071271652,
      "learning_rate": 1.4563498364586514e-05,
      "loss": 11.5812,
      "step": 19013
    },
    {
      "epoch": 3.3248524590163933,
      "grad_norm": 0.7674435476909984,
      "learning_rate": 1.4556140185572009e-05,
      "loss": 11.2147,
      "step": 19014
    },
    {
      "epoch": 3.3250273224043716,
      "grad_norm": 0.691585152213077,
      "learning_rate": 1.4548783719969239e-05,
      "loss": 11.5876,
      "step": 19015
    },
    {
      "epoch": 3.3252021857923495,
      "grad_norm": 0.7345731940321671,
      "learning_rate": 1.4541428967925729e-05,
      "loss": 11.3992,
      "step": 19016
    },
    {
      "epoch": 3.325377049180328,
      "grad_norm": 0.7091080715943345,
      "learning_rate": 1.4534075929588953e-05,
      "loss": 11.3874,
      "step": 19017
    },
    {
      "epoch": 3.325551912568306,
      "grad_norm": 0.6894908292646146,
      "learning_rate": 1.4526724605106378e-05,
      "loss": 11.2228,
      "step": 19018
    },
    {
      "epoch": 3.325726775956284,
      "grad_norm": 0.6949649555235108,
      "learning_rate": 1.4519374994625402e-05,
      "loss": 11.4698,
      "step": 19019
    },
    {
      "epoch": 3.3259016393442624,
      "grad_norm": 0.6503361471660393,
      "learning_rate": 1.4512027098293445e-05,
      "loss": 11.5667,
      "step": 19020
    },
    {
      "epoch": 3.3260765027322403,
      "grad_norm": 0.6874453026789995,
      "learning_rate": 1.4504680916257784e-05,
      "loss": 11.4599,
      "step": 19021
    },
    {
      "epoch": 3.3262513661202187,
      "grad_norm": 0.692650704107385,
      "learning_rate": 1.44973364486658e-05,
      "loss": 11.1207,
      "step": 19022
    },
    {
      "epoch": 3.3264262295081966,
      "grad_norm": 0.6269363607873093,
      "learning_rate": 1.448999369566475e-05,
      "loss": 11.5665,
      "step": 19023
    },
    {
      "epoch": 3.326601092896175,
      "grad_norm": 0.6803378700206038,
      "learning_rate": 1.4482652657401874e-05,
      "loss": 11.5635,
      "step": 19024
    },
    {
      "epoch": 3.326775956284153,
      "grad_norm": 0.7157379402815278,
      "learning_rate": 1.4475313334024376e-05,
      "loss": 11.7249,
      "step": 19025
    },
    {
      "epoch": 3.326950819672131,
      "grad_norm": 0.6458052211567196,
      "learning_rate": 1.446797572567945e-05,
      "loss": 11.4847,
      "step": 19026
    },
    {
      "epoch": 3.3271256830601095,
      "grad_norm": 0.6424890880728343,
      "learning_rate": 1.4460639832514222e-05,
      "loss": 11.2046,
      "step": 19027
    },
    {
      "epoch": 3.3273005464480874,
      "grad_norm": 0.6826349580423988,
      "learning_rate": 1.445330565467582e-05,
      "loss": 11.158,
      "step": 19028
    },
    {
      "epoch": 3.3274754098360657,
      "grad_norm": 0.6553181073120802,
      "learning_rate": 1.4445973192311312e-05,
      "loss": 11.3803,
      "step": 19029
    },
    {
      "epoch": 3.3276502732240436,
      "grad_norm": 0.6685744233240912,
      "learning_rate": 1.4438642445567685e-05,
      "loss": 11.674,
      "step": 19030
    },
    {
      "epoch": 3.327825136612022,
      "grad_norm": 0.6683132360398404,
      "learning_rate": 1.4431313414592007e-05,
      "loss": 11.5424,
      "step": 19031
    },
    {
      "epoch": 3.328,
      "grad_norm": 0.6561646397742411,
      "learning_rate": 1.4423986099531228e-05,
      "loss": 11.4413,
      "step": 19032
    },
    {
      "epoch": 3.328174863387978,
      "grad_norm": 0.7430523028555657,
      "learning_rate": 1.4416660500532274e-05,
      "loss": 11.2751,
      "step": 19033
    },
    {
      "epoch": 3.328349726775956,
      "grad_norm": 0.7505461233826908,
      "learning_rate": 1.4409336617742087e-05,
      "loss": 11.5127,
      "step": 19034
    },
    {
      "epoch": 3.3285245901639344,
      "grad_norm": 0.6583587356759375,
      "learning_rate": 1.4402014451307444e-05,
      "loss": 11.4886,
      "step": 19035
    },
    {
      "epoch": 3.3286994535519128,
      "grad_norm": 0.6771584093418443,
      "learning_rate": 1.4394694001375253e-05,
      "loss": 11.2531,
      "step": 19036
    },
    {
      "epoch": 3.3288743169398907,
      "grad_norm": 0.6734449142323449,
      "learning_rate": 1.4387375268092296e-05,
      "loss": 11.5349,
      "step": 19037
    },
    {
      "epoch": 3.329049180327869,
      "grad_norm": 0.6711734222692061,
      "learning_rate": 1.4380058251605343e-05,
      "loss": 11.3993,
      "step": 19038
    },
    {
      "epoch": 3.329224043715847,
      "grad_norm": 0.7224824482519487,
      "learning_rate": 1.4372742952061103e-05,
      "loss": 11.3386,
      "step": 19039
    },
    {
      "epoch": 3.3293989071038252,
      "grad_norm": 0.63872446409018,
      "learning_rate": 1.4365429369606242e-05,
      "loss": 11.1692,
      "step": 19040
    },
    {
      "epoch": 3.329573770491803,
      "grad_norm": 0.7049652157128262,
      "learning_rate": 1.435811750438748e-05,
      "loss": 11.305,
      "step": 19041
    },
    {
      "epoch": 3.3297486338797815,
      "grad_norm": 0.7459184759933777,
      "learning_rate": 1.4350807356551422e-05,
      "loss": 11.4807,
      "step": 19042
    },
    {
      "epoch": 3.3299234972677594,
      "grad_norm": 0.6914813217589298,
      "learning_rate": 1.434349892624467e-05,
      "loss": 11.6678,
      "step": 19043
    },
    {
      "epoch": 3.3300983606557377,
      "grad_norm": 0.6184468584672693,
      "learning_rate": 1.4336192213613742e-05,
      "loss": 11.2917,
      "step": 19044
    },
    {
      "epoch": 3.330273224043716,
      "grad_norm": 0.7281581574645171,
      "learning_rate": 1.4328887218805165e-05,
      "loss": 11.2831,
      "step": 19045
    },
    {
      "epoch": 3.330448087431694,
      "grad_norm": 0.7079499911944278,
      "learning_rate": 1.4321583941965466e-05,
      "loss": 11.6126,
      "step": 19046
    },
    {
      "epoch": 3.3306229508196723,
      "grad_norm": 0.660373357360574,
      "learning_rate": 1.4314282383241096e-05,
      "loss": 11.3561,
      "step": 19047
    },
    {
      "epoch": 3.33079781420765,
      "grad_norm": 0.6629038074054551,
      "learning_rate": 1.4306982542778424e-05,
      "loss": 11.4513,
      "step": 19048
    },
    {
      "epoch": 3.3309726775956285,
      "grad_norm": 0.6975325020895616,
      "learning_rate": 1.4299684420723858e-05,
      "loss": 11.2417,
      "step": 19049
    },
    {
      "epoch": 3.3311475409836064,
      "grad_norm": 0.6477407651281998,
      "learning_rate": 1.4292388017223767e-05,
      "loss": 11.6325,
      "step": 19050
    },
    {
      "epoch": 3.3313224043715848,
      "grad_norm": 0.7438559922663607,
      "learning_rate": 1.4285093332424449e-05,
      "loss": 11.4699,
      "step": 19051
    },
    {
      "epoch": 3.3314972677595627,
      "grad_norm": 0.6892703030341047,
      "learning_rate": 1.4277800366472227e-05,
      "loss": 11.4838,
      "step": 19052
    },
    {
      "epoch": 3.331672131147541,
      "grad_norm": 0.7373606143650226,
      "learning_rate": 1.4270509119513275e-05,
      "loss": 11.6054,
      "step": 19053
    },
    {
      "epoch": 3.3318469945355194,
      "grad_norm": 0.7725537629000312,
      "learning_rate": 1.4263219591693811e-05,
      "loss": 11.5237,
      "step": 19054
    },
    {
      "epoch": 3.3320218579234973,
      "grad_norm": 0.6870691173713287,
      "learning_rate": 1.4255931783160082e-05,
      "loss": 11.196,
      "step": 19055
    },
    {
      "epoch": 3.3321967213114756,
      "grad_norm": 0.7018471036632938,
      "learning_rate": 1.4248645694058193e-05,
      "loss": 11.171,
      "step": 19056
    },
    {
      "epoch": 3.3323715846994535,
      "grad_norm": 0.6635888772371712,
      "learning_rate": 1.4241361324534242e-05,
      "loss": 11.5577,
      "step": 19057
    },
    {
      "epoch": 3.332546448087432,
      "grad_norm": 0.6954526746663429,
      "learning_rate": 1.4234078674734297e-05,
      "loss": 11.3992,
      "step": 19058
    },
    {
      "epoch": 3.3327213114754097,
      "grad_norm": 0.6814917191362505,
      "learning_rate": 1.42267977448044e-05,
      "loss": 11.4808,
      "step": 19059
    },
    {
      "epoch": 3.332896174863388,
      "grad_norm": 0.7586586401559146,
      "learning_rate": 1.4219518534890575e-05,
      "loss": 11.4966,
      "step": 19060
    },
    {
      "epoch": 3.333071038251366,
      "grad_norm": 0.6460392273978351,
      "learning_rate": 1.4212241045138819e-05,
      "loss": 11.3553,
      "step": 19061
    },
    {
      "epoch": 3.3332459016393443,
      "grad_norm": 0.6556526866904548,
      "learning_rate": 1.4204965275695004e-05,
      "loss": 11.0727,
      "step": 19062
    },
    {
      "epoch": 3.3334207650273227,
      "grad_norm": 0.6396087178014795,
      "learning_rate": 1.4197691226705068e-05,
      "loss": 11.1321,
      "step": 19063
    },
    {
      "epoch": 3.3335956284153005,
      "grad_norm": 0.6907352511611443,
      "learning_rate": 1.4190418898314839e-05,
      "loss": 11.3817,
      "step": 19064
    },
    {
      "epoch": 3.3337704918032784,
      "grad_norm": 0.6566980324064308,
      "learning_rate": 1.4183148290670223e-05,
      "loss": 11.2136,
      "step": 19065
    },
    {
      "epoch": 3.333945355191257,
      "grad_norm": 0.6263477157323037,
      "learning_rate": 1.417587940391697e-05,
      "loss": 11.6526,
      "step": 19066
    },
    {
      "epoch": 3.334120218579235,
      "grad_norm": 0.660653039466868,
      "learning_rate": 1.416861223820084e-05,
      "loss": 11.5347,
      "step": 19067
    },
    {
      "epoch": 3.334295081967213,
      "grad_norm": 0.67486397909476,
      "learning_rate": 1.4161346793667563e-05,
      "loss": 11.6765,
      "step": 19068
    },
    {
      "epoch": 3.3344699453551914,
      "grad_norm": 0.6519808917078862,
      "learning_rate": 1.4154083070462832e-05,
      "loss": 11.2456,
      "step": 19069
    },
    {
      "epoch": 3.3346448087431693,
      "grad_norm": 0.680071082895929,
      "learning_rate": 1.4146821068732352e-05,
      "loss": 11.5191,
      "step": 19070
    },
    {
      "epoch": 3.3348196721311476,
      "grad_norm": 0.7275646159231669,
      "learning_rate": 1.4139560788621697e-05,
      "loss": 11.4082,
      "step": 19071
    },
    {
      "epoch": 3.3349945355191255,
      "grad_norm": 0.6327533738796914,
      "learning_rate": 1.4132302230276483e-05,
      "loss": 11.4776,
      "step": 19072
    },
    {
      "epoch": 3.335169398907104,
      "grad_norm": 0.7402802496344262,
      "learning_rate": 1.4125045393842219e-05,
      "loss": 11.3019,
      "step": 19073
    },
    {
      "epoch": 3.3353442622950817,
      "grad_norm": 0.6979097075334346,
      "learning_rate": 1.4117790279464526e-05,
      "loss": 11.3286,
      "step": 19074
    },
    {
      "epoch": 3.33551912568306,
      "grad_norm": 0.7470174322030567,
      "learning_rate": 1.4110536887288795e-05,
      "loss": 11.3296,
      "step": 19075
    },
    {
      "epoch": 3.3356939890710384,
      "grad_norm": 0.7203778792206696,
      "learning_rate": 1.4103285217460527e-05,
      "loss": 11.5201,
      "step": 19076
    },
    {
      "epoch": 3.3358688524590163,
      "grad_norm": 0.7000500243291498,
      "learning_rate": 1.4096035270125118e-05,
      "loss": 11.4548,
      "step": 19077
    },
    {
      "epoch": 3.3360437158469947,
      "grad_norm": 0.696939268067629,
      "learning_rate": 1.4088787045427943e-05,
      "loss": 11.4796,
      "step": 19078
    },
    {
      "epoch": 3.3362185792349726,
      "grad_norm": 0.6932684500375407,
      "learning_rate": 1.4081540543514415e-05,
      "loss": 11.4538,
      "step": 19079
    },
    {
      "epoch": 3.336393442622951,
      "grad_norm": 0.7268437050700645,
      "learning_rate": 1.4074295764529776e-05,
      "loss": 11.4085,
      "step": 19080
    },
    {
      "epoch": 3.336568306010929,
      "grad_norm": 0.6675675357849726,
      "learning_rate": 1.4067052708619344e-05,
      "loss": 11.1147,
      "step": 19081
    },
    {
      "epoch": 3.336743169398907,
      "grad_norm": 0.7243166483931778,
      "learning_rate": 1.4059811375928344e-05,
      "loss": 11.4724,
      "step": 19082
    },
    {
      "epoch": 3.336918032786885,
      "grad_norm": 0.6908616260062447,
      "learning_rate": 1.4052571766601996e-05,
      "loss": 11.5882,
      "step": 19083
    },
    {
      "epoch": 3.3370928961748634,
      "grad_norm": 0.7166942037527501,
      "learning_rate": 1.4045333880785483e-05,
      "loss": 11.6677,
      "step": 19084
    },
    {
      "epoch": 3.3372677595628417,
      "grad_norm": 0.7022975287366406,
      "learning_rate": 1.4038097718623944e-05,
      "loss": 11.4294,
      "step": 19085
    },
    {
      "epoch": 3.3374426229508196,
      "grad_norm": 0.7184730278981618,
      "learning_rate": 1.4030863280262474e-05,
      "loss": 11.4903,
      "step": 19086
    },
    {
      "epoch": 3.337617486338798,
      "grad_norm": 0.7152281355093822,
      "learning_rate": 1.4023630565846169e-05,
      "loss": 11.3507,
      "step": 19087
    },
    {
      "epoch": 3.337792349726776,
      "grad_norm": 0.689396047800501,
      "learning_rate": 1.4016399575520045e-05,
      "loss": 11.0513,
      "step": 19088
    },
    {
      "epoch": 3.337967213114754,
      "grad_norm": 0.7951079124016668,
      "learning_rate": 1.4009170309429109e-05,
      "loss": 11.2829,
      "step": 19089
    },
    {
      "epoch": 3.338142076502732,
      "grad_norm": 0.6190940642298812,
      "learning_rate": 1.4001942767718346e-05,
      "loss": 11.448,
      "step": 19090
    },
    {
      "epoch": 3.3383169398907104,
      "grad_norm": 0.7197388557778365,
      "learning_rate": 1.3994716950532671e-05,
      "loss": 11.2593,
      "step": 19091
    },
    {
      "epoch": 3.3384918032786883,
      "grad_norm": 0.703502260149619,
      "learning_rate": 1.3987492858016994e-05,
      "loss": 11.4007,
      "step": 19092
    },
    {
      "epoch": 3.3386666666666667,
      "grad_norm": 0.7135125314237698,
      "learning_rate": 1.3980270490316172e-05,
      "loss": 11.5072,
      "step": 19093
    },
    {
      "epoch": 3.338841530054645,
      "grad_norm": 0.7515774500763136,
      "learning_rate": 1.3973049847575049e-05,
      "loss": 11.549,
      "step": 19094
    },
    {
      "epoch": 3.339016393442623,
      "grad_norm": 0.7124618857203335,
      "learning_rate": 1.3965830929938418e-05,
      "loss": 11.4771,
      "step": 19095
    },
    {
      "epoch": 3.3391912568306013,
      "grad_norm": 0.699176750686219,
      "learning_rate": 1.395861373755103e-05,
      "loss": 11.379,
      "step": 19096
    },
    {
      "epoch": 3.339366120218579,
      "grad_norm": 0.6470139528165667,
      "learning_rate": 1.3951398270557613e-05,
      "loss": 11.5239,
      "step": 19097
    },
    {
      "epoch": 3.3395409836065575,
      "grad_norm": 0.7730128715410606,
      "learning_rate": 1.3944184529102877e-05,
      "loss": 11.5295,
      "step": 19098
    },
    {
      "epoch": 3.3397158469945354,
      "grad_norm": 0.7104837690428899,
      "learning_rate": 1.3936972513331459e-05,
      "loss": 11.3069,
      "step": 19099
    },
    {
      "epoch": 3.3398907103825137,
      "grad_norm": 0.7061515882897995,
      "learning_rate": 1.3929762223387999e-05,
      "loss": 11.5177,
      "step": 19100
    },
    {
      "epoch": 3.3400655737704916,
      "grad_norm": 0.645306884708876,
      "learning_rate": 1.392255365941707e-05,
      "loss": 11.2113,
      "step": 19101
    },
    {
      "epoch": 3.34024043715847,
      "grad_norm": 0.6863427583289444,
      "learning_rate": 1.3915346821563235e-05,
      "loss": 11.3908,
      "step": 19102
    },
    {
      "epoch": 3.3404153005464483,
      "grad_norm": 0.6808940151947596,
      "learning_rate": 1.3908141709971e-05,
      "loss": 11.6263,
      "step": 19103
    },
    {
      "epoch": 3.340590163934426,
      "grad_norm": 0.7343091378949619,
      "learning_rate": 1.3900938324784873e-05,
      "loss": 11.5597,
      "step": 19104
    },
    {
      "epoch": 3.3407650273224045,
      "grad_norm": 0.7411515768173124,
      "learning_rate": 1.3893736666149292e-05,
      "loss": 11.5718,
      "step": 19105
    },
    {
      "epoch": 3.3409398907103824,
      "grad_norm": 0.621597550022438,
      "learning_rate": 1.3886536734208665e-05,
      "loss": 11.4298,
      "step": 19106
    },
    {
      "epoch": 3.341114754098361,
      "grad_norm": 0.661088284749748,
      "learning_rate": 1.3879338529107389e-05,
      "loss": 11.285,
      "step": 19107
    },
    {
      "epoch": 3.3412896174863387,
      "grad_norm": 0.640461840152806,
      "learning_rate": 1.387214205098979e-05,
      "loss": 11.7901,
      "step": 19108
    },
    {
      "epoch": 3.341464480874317,
      "grad_norm": 0.6295455371198289,
      "learning_rate": 1.3864947300000197e-05,
      "loss": 11.7697,
      "step": 19109
    },
    {
      "epoch": 3.341639344262295,
      "grad_norm": 0.6681832294840705,
      "learning_rate": 1.3857754276282875e-05,
      "loss": 11.5993,
      "step": 19110
    },
    {
      "epoch": 3.3418142076502733,
      "grad_norm": 0.6577743282336169,
      "learning_rate": 1.3850562979982095e-05,
      "loss": 11.4041,
      "step": 19111
    },
    {
      "epoch": 3.3419890710382516,
      "grad_norm": 0.6604483106605011,
      "learning_rate": 1.3843373411242e-05,
      "loss": 11.491,
      "step": 19112
    },
    {
      "epoch": 3.3421639344262295,
      "grad_norm": 0.7491394178026084,
      "learning_rate": 1.3836185570206817e-05,
      "loss": 11.4962,
      "step": 19113
    },
    {
      "epoch": 3.342338797814208,
      "grad_norm": 0.8477591226901298,
      "learning_rate": 1.3828999457020675e-05,
      "loss": 11.4106,
      "step": 19114
    },
    {
      "epoch": 3.3425136612021857,
      "grad_norm": 0.7085128705381283,
      "learning_rate": 1.3821815071827682e-05,
      "loss": 11.393,
      "step": 19115
    },
    {
      "epoch": 3.342688524590164,
      "grad_norm": 0.6691915056717129,
      "learning_rate": 1.3814632414771911e-05,
      "loss": 11.5839,
      "step": 19116
    },
    {
      "epoch": 3.342863387978142,
      "grad_norm": 0.6087610154604384,
      "learning_rate": 1.3807451485997346e-05,
      "loss": 11.4016,
      "step": 19117
    },
    {
      "epoch": 3.3430382513661203,
      "grad_norm": 0.7716282526179475,
      "learning_rate": 1.3800272285648042e-05,
      "loss": 11.3389,
      "step": 19118
    },
    {
      "epoch": 3.343213114754098,
      "grad_norm": 0.622751199047797,
      "learning_rate": 1.3793094813867947e-05,
      "loss": 11.2866,
      "step": 19119
    },
    {
      "epoch": 3.3433879781420766,
      "grad_norm": 0.6963202618552539,
      "learning_rate": 1.3785919070801012e-05,
      "loss": 11.1301,
      "step": 19120
    },
    {
      "epoch": 3.3435628415300545,
      "grad_norm": 0.66375482018379,
      "learning_rate": 1.377874505659107e-05,
      "loss": 11.7332,
      "step": 19121
    },
    {
      "epoch": 3.343737704918033,
      "grad_norm": 0.7385796352894355,
      "learning_rate": 1.3771572771382047e-05,
      "loss": 11.411,
      "step": 19122
    },
    {
      "epoch": 3.3439125683060107,
      "grad_norm": 0.6529463237490354,
      "learning_rate": 1.376440221531774e-05,
      "loss": 11.1163,
      "step": 19123
    },
    {
      "epoch": 3.344087431693989,
      "grad_norm": 0.6834999636029049,
      "learning_rate": 1.3757233388541958e-05,
      "loss": 11.2034,
      "step": 19124
    },
    {
      "epoch": 3.3442622950819674,
      "grad_norm": 0.7210243892007154,
      "learning_rate": 1.375006629119846e-05,
      "loss": 11.5836,
      "step": 19125
    },
    {
      "epoch": 3.3444371584699453,
      "grad_norm": 0.685299101468,
      "learning_rate": 1.3742900923430913e-05,
      "loss": 11.4532,
      "step": 19126
    },
    {
      "epoch": 3.3446120218579236,
      "grad_norm": 0.632211370399255,
      "learning_rate": 1.3735737285383067e-05,
      "loss": 11.5126,
      "step": 19127
    },
    {
      "epoch": 3.3447868852459015,
      "grad_norm": 0.6538852650065563,
      "learning_rate": 1.372857537719855e-05,
      "loss": 11.3822,
      "step": 19128
    },
    {
      "epoch": 3.34496174863388,
      "grad_norm": 0.6590840005814537,
      "learning_rate": 1.3721415199020992e-05,
      "loss": 11.5626,
      "step": 19129
    },
    {
      "epoch": 3.3451366120218577,
      "grad_norm": 0.6815031149776943,
      "learning_rate": 1.3714256750993992e-05,
      "loss": 11.3008,
      "step": 19130
    },
    {
      "epoch": 3.345311475409836,
      "grad_norm": 0.7144630081817628,
      "learning_rate": 1.3707100033261034e-05,
      "loss": 11.37,
      "step": 19131
    },
    {
      "epoch": 3.345486338797814,
      "grad_norm": 0.6424213480314561,
      "learning_rate": 1.369994504596569e-05,
      "loss": 11.4507,
      "step": 19132
    },
    {
      "epoch": 3.3456612021857923,
      "grad_norm": 0.6637832807475906,
      "learning_rate": 1.3692791789251425e-05,
      "loss": 11.1015,
      "step": 19133
    },
    {
      "epoch": 3.3458360655737707,
      "grad_norm": 0.6768594796564428,
      "learning_rate": 1.36856402632617e-05,
      "loss": 10.8105,
      "step": 19134
    },
    {
      "epoch": 3.3460109289617486,
      "grad_norm": 0.6711919913293131,
      "learning_rate": 1.3678490468139893e-05,
      "loss": 11.2515,
      "step": 19135
    },
    {
      "epoch": 3.346185792349727,
      "grad_norm": 0.6510234658361035,
      "learning_rate": 1.3671342404029375e-05,
      "loss": 11.3356,
      "step": 19136
    },
    {
      "epoch": 3.346360655737705,
      "grad_norm": 0.666490189239975,
      "learning_rate": 1.3664196071073521e-05,
      "loss": 11.2187,
      "step": 19137
    },
    {
      "epoch": 3.346535519125683,
      "grad_norm": 0.7415773065110179,
      "learning_rate": 1.3657051469415626e-05,
      "loss": 11.1184,
      "step": 19138
    },
    {
      "epoch": 3.346710382513661,
      "grad_norm": 0.7644971894140796,
      "learning_rate": 1.3649908599198969e-05,
      "loss": 11.4229,
      "step": 19139
    },
    {
      "epoch": 3.3468852459016394,
      "grad_norm": 0.7475931993352749,
      "learning_rate": 1.3642767460566763e-05,
      "loss": 11.4655,
      "step": 19140
    },
    {
      "epoch": 3.3470601092896173,
      "grad_norm": 0.6435011743288687,
      "learning_rate": 1.3635628053662197e-05,
      "loss": 11.3523,
      "step": 19141
    },
    {
      "epoch": 3.3472349726775956,
      "grad_norm": 0.8075052268928423,
      "learning_rate": 1.3628490378628479e-05,
      "loss": 11.5822,
      "step": 19142
    },
    {
      "epoch": 3.347409836065574,
      "grad_norm": 0.6780756652045757,
      "learning_rate": 1.3621354435608747e-05,
      "loss": 11.5418,
      "step": 19143
    },
    {
      "epoch": 3.347584699453552,
      "grad_norm": 0.6543811365210762,
      "learning_rate": 1.3614220224746054e-05,
      "loss": 11.4735,
      "step": 19144
    },
    {
      "epoch": 3.34775956284153,
      "grad_norm": 0.6981933730496882,
      "learning_rate": 1.3607087746183466e-05,
      "loss": 11.4764,
      "step": 19145
    },
    {
      "epoch": 3.347934426229508,
      "grad_norm": 0.7680083780900614,
      "learning_rate": 1.3599957000064057e-05,
      "loss": 11.2496,
      "step": 19146
    },
    {
      "epoch": 3.3481092896174864,
      "grad_norm": 0.7248594471302541,
      "learning_rate": 1.3592827986530787e-05,
      "loss": 11.3193,
      "step": 19147
    },
    {
      "epoch": 3.3482841530054643,
      "grad_norm": 0.6795185426079143,
      "learning_rate": 1.3585700705726645e-05,
      "loss": 11.5553,
      "step": 19148
    },
    {
      "epoch": 3.3484590163934427,
      "grad_norm": 0.6714159008456421,
      "learning_rate": 1.3578575157794504e-05,
      "loss": 11.7329,
      "step": 19149
    },
    {
      "epoch": 3.3486338797814206,
      "grad_norm": 0.7142169985672464,
      "learning_rate": 1.3571451342877261e-05,
      "loss": 11.6333,
      "step": 19150
    },
    {
      "epoch": 3.348808743169399,
      "grad_norm": 0.7809395332885958,
      "learning_rate": 1.3564329261117804e-05,
      "loss": 11.2754,
      "step": 19151
    },
    {
      "epoch": 3.3489836065573773,
      "grad_norm": 0.6856418766554683,
      "learning_rate": 1.355720891265897e-05,
      "loss": 11.1687,
      "step": 19152
    },
    {
      "epoch": 3.349158469945355,
      "grad_norm": 0.6417182635813726,
      "learning_rate": 1.3550090297643482e-05,
      "loss": 11.6745,
      "step": 19153
    },
    {
      "epoch": 3.3493333333333335,
      "grad_norm": 0.6625523936242077,
      "learning_rate": 1.3542973416214121e-05,
      "loss": 11.3644,
      "step": 19154
    },
    {
      "epoch": 3.3495081967213114,
      "grad_norm": 0.6977387114743642,
      "learning_rate": 1.353585826851358e-05,
      "loss": 11.5056,
      "step": 19155
    },
    {
      "epoch": 3.3496830601092897,
      "grad_norm": 0.644081760109978,
      "learning_rate": 1.3528744854684582e-05,
      "loss": 11.6636,
      "step": 19156
    },
    {
      "epoch": 3.3498579234972676,
      "grad_norm": 0.6548162227827262,
      "learning_rate": 1.3521633174869774e-05,
      "loss": 11.4595,
      "step": 19157
    },
    {
      "epoch": 3.350032786885246,
      "grad_norm": 0.6931425805401902,
      "learning_rate": 1.3514523229211729e-05,
      "loss": 11.2662,
      "step": 19158
    },
    {
      "epoch": 3.350207650273224,
      "grad_norm": 0.6906435357776148,
      "learning_rate": 1.3507415017853032e-05,
      "loss": 11.4694,
      "step": 19159
    },
    {
      "epoch": 3.350382513661202,
      "grad_norm": 0.6454453286585775,
      "learning_rate": 1.3500308540936201e-05,
      "loss": 11.4857,
      "step": 19160
    },
    {
      "epoch": 3.3505573770491806,
      "grad_norm": 0.6785484103104282,
      "learning_rate": 1.3493203798603826e-05,
      "loss": 11.5019,
      "step": 19161
    },
    {
      "epoch": 3.3507322404371584,
      "grad_norm": 0.6653094619205046,
      "learning_rate": 1.3486100790998302e-05,
      "loss": 11.6383,
      "step": 19162
    },
    {
      "epoch": 3.350907103825137,
      "grad_norm": 0.7169232289727354,
      "learning_rate": 1.3478999518262093e-05,
      "loss": 11.3538,
      "step": 19163
    },
    {
      "epoch": 3.3510819672131147,
      "grad_norm": 0.710963467204058,
      "learning_rate": 1.3471899980537594e-05,
      "loss": 11.3472,
      "step": 19164
    },
    {
      "epoch": 3.351256830601093,
      "grad_norm": 0.6866076028543225,
      "learning_rate": 1.3464802177967163e-05,
      "loss": 11.4738,
      "step": 19165
    },
    {
      "epoch": 3.351431693989071,
      "grad_norm": 0.6763694653912522,
      "learning_rate": 1.345770611069318e-05,
      "loss": 11.606,
      "step": 19166
    },
    {
      "epoch": 3.3516065573770493,
      "grad_norm": 0.6818665193139513,
      "learning_rate": 1.345061177885789e-05,
      "loss": 11.5319,
      "step": 19167
    },
    {
      "epoch": 3.351781420765027,
      "grad_norm": 0.698137885388258,
      "learning_rate": 1.344351918260357e-05,
      "loss": 11.5278,
      "step": 19168
    },
    {
      "epoch": 3.3519562841530055,
      "grad_norm": 0.7225863217055538,
      "learning_rate": 1.3436428322072447e-05,
      "loss": 11.2195,
      "step": 19169
    },
    {
      "epoch": 3.352131147540984,
      "grad_norm": 0.6310497548569705,
      "learning_rate": 1.3429339197406755e-05,
      "loss": 11.4802,
      "step": 19170
    },
    {
      "epoch": 3.3523060109289617,
      "grad_norm": 0.6547991501286959,
      "learning_rate": 1.3422251808748597e-05,
      "loss": 11.265,
      "step": 19171
    },
    {
      "epoch": 3.35248087431694,
      "grad_norm": 0.7176775842554479,
      "learning_rate": 1.3415166156240111e-05,
      "loss": 11.2992,
      "step": 19172
    },
    {
      "epoch": 3.352655737704918,
      "grad_norm": 0.6425166248433873,
      "learning_rate": 1.3408082240023412e-05,
      "loss": 11.344,
      "step": 19173
    },
    {
      "epoch": 3.3528306010928963,
      "grad_norm": 0.6976921996487182,
      "learning_rate": 1.3401000060240498e-05,
      "loss": 11.4397,
      "step": 19174
    },
    {
      "epoch": 3.353005464480874,
      "grad_norm": 0.6865532105098883,
      "learning_rate": 1.3393919617033479e-05,
      "loss": 11.3748,
      "step": 19175
    },
    {
      "epoch": 3.3531803278688526,
      "grad_norm": 0.6643013991112718,
      "learning_rate": 1.3386840910544263e-05,
      "loss": 11.6052,
      "step": 19176
    },
    {
      "epoch": 3.3533551912568305,
      "grad_norm": 0.7431184487684118,
      "learning_rate": 1.3379763940914824e-05,
      "loss": 11.4197,
      "step": 19177
    },
    {
      "epoch": 3.353530054644809,
      "grad_norm": 0.6789250660734368,
      "learning_rate": 1.3372688708287074e-05,
      "loss": 11.0438,
      "step": 19178
    },
    {
      "epoch": 3.3537049180327867,
      "grad_norm": 0.6628749270025261,
      "learning_rate": 1.3365615212802907e-05,
      "loss": 11.3847,
      "step": 19179
    },
    {
      "epoch": 3.353879781420765,
      "grad_norm": 0.6441007938928442,
      "learning_rate": 1.3358543454604145e-05,
      "loss": 11.4106,
      "step": 19180
    },
    {
      "epoch": 3.354054644808743,
      "grad_norm": 0.7173692627759191,
      "learning_rate": 1.3351473433832617e-05,
      "loss": 11.3707,
      "step": 19181
    },
    {
      "epoch": 3.3542295081967213,
      "grad_norm": 0.6992611083298252,
      "learning_rate": 1.334440515063009e-05,
      "loss": 11.1354,
      "step": 19182
    },
    {
      "epoch": 3.3544043715846996,
      "grad_norm": 0.7171065795505779,
      "learning_rate": 1.3337338605138306e-05,
      "loss": 11.2891,
      "step": 19183
    },
    {
      "epoch": 3.3545792349726775,
      "grad_norm": 0.6804641066248115,
      "learning_rate": 1.3330273797498982e-05,
      "loss": 11.3273,
      "step": 19184
    },
    {
      "epoch": 3.354754098360656,
      "grad_norm": 0.6785950435982473,
      "learning_rate": 1.3323210727853774e-05,
      "loss": 11.3959,
      "step": 19185
    },
    {
      "epoch": 3.3549289617486338,
      "grad_norm": 0.6684270114362515,
      "learning_rate": 1.3316149396344324e-05,
      "loss": 11.5401,
      "step": 19186
    },
    {
      "epoch": 3.355103825136612,
      "grad_norm": 0.6526146221877075,
      "learning_rate": 1.330908980311223e-05,
      "loss": 11.2069,
      "step": 19187
    },
    {
      "epoch": 3.35527868852459,
      "grad_norm": 0.6622067060327522,
      "learning_rate": 1.3302031948299054e-05,
      "loss": 11.5061,
      "step": 19188
    },
    {
      "epoch": 3.3554535519125683,
      "grad_norm": 0.753617093759627,
      "learning_rate": 1.3294975832046353e-05,
      "loss": 11.5557,
      "step": 19189
    },
    {
      "epoch": 3.3556284153005462,
      "grad_norm": 0.7229088453516199,
      "learning_rate": 1.3287921454495588e-05,
      "loss": 11.4231,
      "step": 19190
    },
    {
      "epoch": 3.3558032786885246,
      "grad_norm": 0.6986190163027558,
      "learning_rate": 1.3280868815788249e-05,
      "loss": 11.6359,
      "step": 19191
    },
    {
      "epoch": 3.355978142076503,
      "grad_norm": 0.7025068152100151,
      "learning_rate": 1.3273817916065756e-05,
      "loss": 11.3101,
      "step": 19192
    },
    {
      "epoch": 3.356153005464481,
      "grad_norm": 0.6355467170716171,
      "learning_rate": 1.3266768755469494e-05,
      "loss": 11.3778,
      "step": 19193
    },
    {
      "epoch": 3.356327868852459,
      "grad_norm": 0.7203362202696904,
      "learning_rate": 1.3259721334140818e-05,
      "loss": 11.3752,
      "step": 19194
    },
    {
      "epoch": 3.356502732240437,
      "grad_norm": 0.691447316449041,
      "learning_rate": 1.325267565222107e-05,
      "loss": 11.414,
      "step": 19195
    },
    {
      "epoch": 3.3566775956284154,
      "grad_norm": 0.7116169734789952,
      "learning_rate": 1.3245631709851513e-05,
      "loss": 11.2896,
      "step": 19196
    },
    {
      "epoch": 3.3568524590163933,
      "grad_norm": 0.7110342083770032,
      "learning_rate": 1.3238589507173416e-05,
      "loss": 11.1517,
      "step": 19197
    },
    {
      "epoch": 3.3570273224043716,
      "grad_norm": 0.6874436675677232,
      "learning_rate": 1.3231549044327995e-05,
      "loss": 11.2313,
      "step": 19198
    },
    {
      "epoch": 3.3572021857923495,
      "grad_norm": 0.6648207676493006,
      "learning_rate": 1.322451032145643e-05,
      "loss": 11.3542,
      "step": 19199
    },
    {
      "epoch": 3.357377049180328,
      "grad_norm": 0.6677477849755444,
      "learning_rate": 1.3217473338699859e-05,
      "loss": 11.3318,
      "step": 19200
    },
    {
      "epoch": 3.357551912568306,
      "grad_norm": 0.766246039676615,
      "learning_rate": 1.3210438096199418e-05,
      "loss": 11.577,
      "step": 19201
    },
    {
      "epoch": 3.357726775956284,
      "grad_norm": 0.6752213861789329,
      "learning_rate": 1.3203404594096159e-05,
      "loss": 11.3912,
      "step": 19202
    },
    {
      "epoch": 3.3579016393442624,
      "grad_norm": 0.6582444764172956,
      "learning_rate": 1.3196372832531145e-05,
      "loss": 11.3355,
      "step": 19203
    },
    {
      "epoch": 3.3580765027322403,
      "grad_norm": 0.6972753413406882,
      "learning_rate": 1.3189342811645377e-05,
      "loss": 11.3748,
      "step": 19204
    },
    {
      "epoch": 3.3582513661202187,
      "grad_norm": 0.6724073956990066,
      "learning_rate": 1.3182314531579832e-05,
      "loss": 11.4139,
      "step": 19205
    },
    {
      "epoch": 3.3584262295081966,
      "grad_norm": 0.6434708292007749,
      "learning_rate": 1.3175287992475438e-05,
      "loss": 11.5156,
      "step": 19206
    },
    {
      "epoch": 3.358601092896175,
      "grad_norm": 0.6451805143166551,
      "learning_rate": 1.316826319447314e-05,
      "loss": 11.4194,
      "step": 19207
    },
    {
      "epoch": 3.358775956284153,
      "grad_norm": 0.6981147424471428,
      "learning_rate": 1.3161240137713727e-05,
      "loss": 11.4154,
      "step": 19208
    },
    {
      "epoch": 3.358950819672131,
      "grad_norm": 0.6836731742116988,
      "learning_rate": 1.3154218822338094e-05,
      "loss": 11.3788,
      "step": 19209
    },
    {
      "epoch": 3.3591256830601095,
      "grad_norm": 0.6757743926375488,
      "learning_rate": 1.314719924848703e-05,
      "loss": 11.362,
      "step": 19210
    },
    {
      "epoch": 3.3593005464480874,
      "grad_norm": 0.6758341458979845,
      "learning_rate": 1.3140181416301289e-05,
      "loss": 11.4882,
      "step": 19211
    },
    {
      "epoch": 3.3594754098360657,
      "grad_norm": 0.6511784009738858,
      "learning_rate": 1.3133165325921636e-05,
      "loss": 11.5238,
      "step": 19212
    },
    {
      "epoch": 3.3596502732240436,
      "grad_norm": 0.6704509079472551,
      "learning_rate": 1.3126150977488682e-05,
      "loss": 11.3842,
      "step": 19213
    },
    {
      "epoch": 3.359825136612022,
      "grad_norm": 0.6672775249974755,
      "learning_rate": 1.311913837114317e-05,
      "loss": 11.0097,
      "step": 19214
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.6654493241628231,
      "learning_rate": 1.3112127507025684e-05,
      "loss": 11.4344,
      "step": 19215
    },
    {
      "epoch": 3.360174863387978,
      "grad_norm": 0.7226447733439499,
      "learning_rate": 1.310511838527685e-05,
      "loss": 11.3832,
      "step": 19216
    },
    {
      "epoch": 3.360349726775956,
      "grad_norm": 0.7241509711707511,
      "learning_rate": 1.309811100603715e-05,
      "loss": 11.1609,
      "step": 19217
    },
    {
      "epoch": 3.3605245901639345,
      "grad_norm": 0.7304881966514295,
      "learning_rate": 1.3091105369447165e-05,
      "loss": 11.4871,
      "step": 19218
    },
    {
      "epoch": 3.360699453551913,
      "grad_norm": 0.6408961282233937,
      "learning_rate": 1.308410147564737e-05,
      "loss": 11.4807,
      "step": 19219
    },
    {
      "epoch": 3.3608743169398907,
      "grad_norm": 0.6751704881584546,
      "learning_rate": 1.3077099324778208e-05,
      "loss": 11.1646,
      "step": 19220
    },
    {
      "epoch": 3.361049180327869,
      "grad_norm": 0.6582551972815973,
      "learning_rate": 1.30700989169801e-05,
      "loss": 11.3885,
      "step": 19221
    },
    {
      "epoch": 3.361224043715847,
      "grad_norm": 0.6631824817410653,
      "learning_rate": 1.306310025239339e-05,
      "loss": 11.1808,
      "step": 19222
    },
    {
      "epoch": 3.3613989071038253,
      "grad_norm": 0.6655095594370178,
      "learning_rate": 1.3056103331158464e-05,
      "loss": 11.6588,
      "step": 19223
    },
    {
      "epoch": 3.361573770491803,
      "grad_norm": 0.7478911706213001,
      "learning_rate": 1.3049108153415634e-05,
      "loss": 11.1961,
      "step": 19224
    },
    {
      "epoch": 3.3617486338797815,
      "grad_norm": 0.6733057440275563,
      "learning_rate": 1.3042114719305165e-05,
      "loss": 11.5161,
      "step": 19225
    },
    {
      "epoch": 3.3619234972677594,
      "grad_norm": 0.6293066897398808,
      "learning_rate": 1.3035123028967278e-05,
      "loss": 11.4564,
      "step": 19226
    },
    {
      "epoch": 3.3620983606557377,
      "grad_norm": 0.6575399804558028,
      "learning_rate": 1.3028133082542172e-05,
      "loss": 11.586,
      "step": 19227
    },
    {
      "epoch": 3.362273224043716,
      "grad_norm": 0.7076429219881679,
      "learning_rate": 1.3021144880170043e-05,
      "loss": 11.4321,
      "step": 19228
    },
    {
      "epoch": 3.362448087431694,
      "grad_norm": 0.6893548854935475,
      "learning_rate": 1.301415842199104e-05,
      "loss": 11.4709,
      "step": 19229
    },
    {
      "epoch": 3.362622950819672,
      "grad_norm": 0.6933407009838591,
      "learning_rate": 1.3007173708145249e-05,
      "loss": 11.4662,
      "step": 19230
    },
    {
      "epoch": 3.3627978142076502,
      "grad_norm": 0.6778572186787385,
      "learning_rate": 1.3000190738772711e-05,
      "loss": 11.3028,
      "step": 19231
    },
    {
      "epoch": 3.3629726775956286,
      "grad_norm": 0.6845413289726883,
      "learning_rate": 1.2993209514013461e-05,
      "loss": 11.4598,
      "step": 19232
    },
    {
      "epoch": 3.3631475409836065,
      "grad_norm": 0.6653653577356513,
      "learning_rate": 1.2986230034007519e-05,
      "loss": 11.4917,
      "step": 19233
    },
    {
      "epoch": 3.363322404371585,
      "grad_norm": 0.7517142714427265,
      "learning_rate": 1.297925229889485e-05,
      "loss": 11.4704,
      "step": 19234
    },
    {
      "epoch": 3.3634972677595627,
      "grad_norm": 0.6644927576966235,
      "learning_rate": 1.2972276308815345e-05,
      "loss": 11.3744,
      "step": 19235
    },
    {
      "epoch": 3.363672131147541,
      "grad_norm": 0.6781652523303218,
      "learning_rate": 1.2965302063908902e-05,
      "loss": 11.2209,
      "step": 19236
    },
    {
      "epoch": 3.363846994535519,
      "grad_norm": 0.760138633183042,
      "learning_rate": 1.2958329564315374e-05,
      "loss": 11.3294,
      "step": 19237
    },
    {
      "epoch": 3.3640218579234973,
      "grad_norm": 0.6526695430961797,
      "learning_rate": 1.2951358810174608e-05,
      "loss": 11.4374,
      "step": 19238
    },
    {
      "epoch": 3.364196721311475,
      "grad_norm": 0.6652004088846489,
      "learning_rate": 1.2944389801626388e-05,
      "loss": 11.5695,
      "step": 19239
    },
    {
      "epoch": 3.3643715846994535,
      "grad_norm": 0.6428480284544567,
      "learning_rate": 1.293742253881044e-05,
      "loss": 11.3812,
      "step": 19240
    },
    {
      "epoch": 3.364546448087432,
      "grad_norm": 0.6843670826029558,
      "learning_rate": 1.293045702186646e-05,
      "loss": 11.5028,
      "step": 19241
    },
    {
      "epoch": 3.3647213114754098,
      "grad_norm": 0.6432497039822308,
      "learning_rate": 1.2923493250934171e-05,
      "loss": 11.5571,
      "step": 19242
    },
    {
      "epoch": 3.364896174863388,
      "grad_norm": 0.6946559084747513,
      "learning_rate": 1.2916531226153229e-05,
      "loss": 11.4272,
      "step": 19243
    },
    {
      "epoch": 3.365071038251366,
      "grad_norm": 0.7344738783515882,
      "learning_rate": 1.2909570947663196e-05,
      "loss": 11.3878,
      "step": 19244
    },
    {
      "epoch": 3.3652459016393443,
      "grad_norm": 0.6752776614686915,
      "learning_rate": 1.2902612415603665e-05,
      "loss": 11.4301,
      "step": 19245
    },
    {
      "epoch": 3.3654207650273222,
      "grad_norm": 0.6381298435742087,
      "learning_rate": 1.2895655630114167e-05,
      "loss": 11.6572,
      "step": 19246
    },
    {
      "epoch": 3.3655956284153006,
      "grad_norm": 0.660334845454869,
      "learning_rate": 1.2888700591334223e-05,
      "loss": 11.6566,
      "step": 19247
    },
    {
      "epoch": 3.3657704918032785,
      "grad_norm": 0.6765322606415091,
      "learning_rate": 1.2881747299403335e-05,
      "loss": 11.5135,
      "step": 19248
    },
    {
      "epoch": 3.365945355191257,
      "grad_norm": 0.6498844636605677,
      "learning_rate": 1.2874795754460878e-05,
      "loss": 11.2433,
      "step": 19249
    },
    {
      "epoch": 3.366120218579235,
      "grad_norm": 0.7302002969574201,
      "learning_rate": 1.2867845956646274e-05,
      "loss": 11.0778,
      "step": 19250
    },
    {
      "epoch": 3.366295081967213,
      "grad_norm": 0.6938846623154334,
      "learning_rate": 1.2860897906098868e-05,
      "loss": 11.4526,
      "step": 19251
    },
    {
      "epoch": 3.3664699453551914,
      "grad_norm": 0.715380481052021,
      "learning_rate": 1.2853951602958059e-05,
      "loss": 11.0885,
      "step": 19252
    },
    {
      "epoch": 3.3666448087431693,
      "grad_norm": 0.7167019027422304,
      "learning_rate": 1.2847007047363058e-05,
      "loss": 11.0986,
      "step": 19253
    },
    {
      "epoch": 3.3668196721311476,
      "grad_norm": 0.6184796891538331,
      "learning_rate": 1.2840064239453176e-05,
      "loss": 11.6521,
      "step": 19254
    },
    {
      "epoch": 3.3669945355191255,
      "grad_norm": 0.6710945363027245,
      "learning_rate": 1.2833123179367613e-05,
      "loss": 11.3394,
      "step": 19255
    },
    {
      "epoch": 3.367169398907104,
      "grad_norm": 0.6928208186811978,
      "learning_rate": 1.2826183867245557e-05,
      "loss": 11.5128,
      "step": 19256
    },
    {
      "epoch": 3.3673442622950818,
      "grad_norm": 0.6398123001754984,
      "learning_rate": 1.2819246303226206e-05,
      "loss": 11.0204,
      "step": 19257
    },
    {
      "epoch": 3.36751912568306,
      "grad_norm": 0.6252219240714044,
      "learning_rate": 1.281231048744862e-05,
      "loss": 11.4111,
      "step": 19258
    },
    {
      "epoch": 3.3676939890710385,
      "grad_norm": 0.7305404481345995,
      "learning_rate": 1.2805376420051918e-05,
      "loss": 11.3221,
      "step": 19259
    },
    {
      "epoch": 3.3678688524590163,
      "grad_norm": 0.7234189528718311,
      "learning_rate": 1.2798444101175144e-05,
      "loss": 11.2903,
      "step": 19260
    },
    {
      "epoch": 3.3680437158469947,
      "grad_norm": 0.6796401936614406,
      "learning_rate": 1.279151353095731e-05,
      "loss": 11.4692,
      "step": 19261
    },
    {
      "epoch": 3.3682185792349726,
      "grad_norm": 0.686815673671348,
      "learning_rate": 1.2784584709537395e-05,
      "loss": 11.0708,
      "step": 19262
    },
    {
      "epoch": 3.368393442622951,
      "grad_norm": 0.6604908212488001,
      "learning_rate": 1.277765763705434e-05,
      "loss": 11.4509,
      "step": 19263
    },
    {
      "epoch": 3.368568306010929,
      "grad_norm": 0.7360430235046779,
      "learning_rate": 1.277073231364705e-05,
      "loss": 11.3115,
      "step": 19264
    },
    {
      "epoch": 3.368743169398907,
      "grad_norm": 0.6348372244009932,
      "learning_rate": 1.2763808739454398e-05,
      "loss": 11.516,
      "step": 19265
    },
    {
      "epoch": 3.368918032786885,
      "grad_norm": 0.60373746092339,
      "learning_rate": 1.2756886914615263e-05,
      "loss": 11.2322,
      "step": 19266
    },
    {
      "epoch": 3.3690928961748634,
      "grad_norm": 0.6221304960232924,
      "learning_rate": 1.2749966839268401e-05,
      "loss": 11.4056,
      "step": 19267
    },
    {
      "epoch": 3.3692677595628417,
      "grad_norm": 0.7346960258377198,
      "learning_rate": 1.2743048513552603e-05,
      "loss": 11.3132,
      "step": 19268
    },
    {
      "epoch": 3.3694426229508196,
      "grad_norm": 0.6579037954148986,
      "learning_rate": 1.2736131937606588e-05,
      "loss": 11.6366,
      "step": 19269
    },
    {
      "epoch": 3.369617486338798,
      "grad_norm": 0.656782360956674,
      "learning_rate": 1.2729217111569059e-05,
      "loss": 11.499,
      "step": 19270
    },
    {
      "epoch": 3.369792349726776,
      "grad_norm": 0.7037139842188581,
      "learning_rate": 1.2722304035578692e-05,
      "loss": 11.229,
      "step": 19271
    },
    {
      "epoch": 3.3699672131147542,
      "grad_norm": 0.7022662884074626,
      "learning_rate": 1.2715392709774099e-05,
      "loss": 11.4047,
      "step": 19272
    },
    {
      "epoch": 3.370142076502732,
      "grad_norm": 0.6432199732792045,
      "learning_rate": 1.2708483134293891e-05,
      "loss": 11.5047,
      "step": 19273
    },
    {
      "epoch": 3.3703169398907105,
      "grad_norm": 0.7673112250257679,
      "learning_rate": 1.2701575309276603e-05,
      "loss": 11.4092,
      "step": 19274
    },
    {
      "epoch": 3.3704918032786884,
      "grad_norm": 0.6859402155797942,
      "learning_rate": 1.2694669234860778e-05,
      "loss": 11.3971,
      "step": 19275
    },
    {
      "epoch": 3.3706666666666667,
      "grad_norm": 0.693748377795415,
      "learning_rate": 1.2687764911184907e-05,
      "loss": 11.2606,
      "step": 19276
    },
    {
      "epoch": 3.370841530054645,
      "grad_norm": 0.6864346259058342,
      "learning_rate": 1.2680862338387423e-05,
      "loss": 11.5279,
      "step": 19277
    },
    {
      "epoch": 3.371016393442623,
      "grad_norm": 0.6735862575996576,
      "learning_rate": 1.2673961516606758e-05,
      "loss": 11.5539,
      "step": 19278
    },
    {
      "epoch": 3.3711912568306013,
      "grad_norm": 0.7206296899372393,
      "learning_rate": 1.2667062445981282e-05,
      "loss": 11.6715,
      "step": 19279
    },
    {
      "epoch": 3.371366120218579,
      "grad_norm": 0.6470526896472627,
      "learning_rate": 1.266016512664936e-05,
      "loss": 11.5314,
      "step": 19280
    },
    {
      "epoch": 3.3715409836065575,
      "grad_norm": 0.6721530946748707,
      "learning_rate": 1.2653269558749292e-05,
      "loss": 11.2938,
      "step": 19281
    },
    {
      "epoch": 3.3717158469945354,
      "grad_norm": 0.7062935099304224,
      "learning_rate": 1.2646375742419358e-05,
      "loss": 11.4296,
      "step": 19282
    },
    {
      "epoch": 3.3718907103825138,
      "grad_norm": 0.6399984080881794,
      "learning_rate": 1.2639483677797814e-05,
      "loss": 11.5665,
      "step": 19283
    },
    {
      "epoch": 3.3720655737704917,
      "grad_norm": 0.6593956745685823,
      "learning_rate": 1.2632593365022837e-05,
      "loss": 11.367,
      "step": 19284
    },
    {
      "epoch": 3.37224043715847,
      "grad_norm": 0.7309015824538397,
      "learning_rate": 1.262570480423263e-05,
      "loss": 11.5386,
      "step": 19285
    },
    {
      "epoch": 3.3724153005464483,
      "grad_norm": 0.6857204337555651,
      "learning_rate": 1.2618817995565303e-05,
      "loss": 11.3911,
      "step": 19286
    },
    {
      "epoch": 3.3725901639344262,
      "grad_norm": 0.6896051823860476,
      "learning_rate": 1.2611932939158978e-05,
      "loss": 11.4671,
      "step": 19287
    },
    {
      "epoch": 3.372765027322404,
      "grad_norm": 0.6791899269346331,
      "learning_rate": 1.260504963515171e-05,
      "loss": 11.3812,
      "step": 19288
    },
    {
      "epoch": 3.3729398907103825,
      "grad_norm": 0.6271864088644705,
      "learning_rate": 1.2598168083681539e-05,
      "loss": 11.5198,
      "step": 19289
    },
    {
      "epoch": 3.373114754098361,
      "grad_norm": 0.6399637893240119,
      "learning_rate": 1.259128828488646e-05,
      "loss": 11.309,
      "step": 19290
    },
    {
      "epoch": 3.3732896174863387,
      "grad_norm": 0.6718901368946903,
      "learning_rate": 1.2584410238904432e-05,
      "loss": 11.239,
      "step": 19291
    },
    {
      "epoch": 3.373464480874317,
      "grad_norm": 0.6450467360422808,
      "learning_rate": 1.2577533945873366e-05,
      "loss": 11.3265,
      "step": 19292
    },
    {
      "epoch": 3.373639344262295,
      "grad_norm": 0.6485978863110407,
      "learning_rate": 1.2570659405931173e-05,
      "loss": 11.3929,
      "step": 19293
    },
    {
      "epoch": 3.3738142076502733,
      "grad_norm": 0.6618937838004697,
      "learning_rate": 1.2563786619215711e-05,
      "loss": 11.4552,
      "step": 19294
    },
    {
      "epoch": 3.373989071038251,
      "grad_norm": 0.6240498547648243,
      "learning_rate": 1.2556915585864781e-05,
      "loss": 11.632,
      "step": 19295
    },
    {
      "epoch": 3.3741639344262295,
      "grad_norm": 0.7969221626933622,
      "learning_rate": 1.2550046306016183e-05,
      "loss": 11.5589,
      "step": 19296
    },
    {
      "epoch": 3.3743387978142074,
      "grad_norm": 0.6573279030329141,
      "learning_rate": 1.2543178779807663e-05,
      "loss": 11.3538,
      "step": 19297
    },
    {
      "epoch": 3.3745136612021858,
      "grad_norm": 0.6670710534412506,
      "learning_rate": 1.2536313007376943e-05,
      "loss": 11.5029,
      "step": 19298
    },
    {
      "epoch": 3.374688524590164,
      "grad_norm": 0.6231253059474169,
      "learning_rate": 1.252944898886168e-05,
      "loss": 11.5924,
      "step": 19299
    },
    {
      "epoch": 3.374863387978142,
      "grad_norm": 0.6994424127242699,
      "learning_rate": 1.2522586724399554e-05,
      "loss": 11.455,
      "step": 19300
    },
    {
      "epoch": 3.3750382513661203,
      "grad_norm": 0.7252716362809437,
      "learning_rate": 1.2515726214128142e-05,
      "loss": 11.3817,
      "step": 19301
    },
    {
      "epoch": 3.3752131147540982,
      "grad_norm": 0.6431026107607943,
      "learning_rate": 1.2508867458185037e-05,
      "loss": 11.4498,
      "step": 19302
    },
    {
      "epoch": 3.3753879781420766,
      "grad_norm": 0.6530182186873575,
      "learning_rate": 1.250201045670779e-05,
      "loss": 11.5646,
      "step": 19303
    },
    {
      "epoch": 3.3755628415300545,
      "grad_norm": 0.6253690529743954,
      "learning_rate": 1.2495155209833843e-05,
      "loss": 11.2043,
      "step": 19304
    },
    {
      "epoch": 3.375737704918033,
      "grad_norm": 0.7369193747510633,
      "learning_rate": 1.2488301717700735e-05,
      "loss": 11.1954,
      "step": 19305
    },
    {
      "epoch": 3.3759125683060107,
      "grad_norm": 0.6786891048160469,
      "learning_rate": 1.2481449980445869e-05,
      "loss": 11.047,
      "step": 19306
    },
    {
      "epoch": 3.376087431693989,
      "grad_norm": 0.6139077842701199,
      "learning_rate": 1.247459999820666e-05,
      "loss": 11.4592,
      "step": 19307
    },
    {
      "epoch": 3.3762622950819674,
      "grad_norm": 0.5961102061249004,
      "learning_rate": 1.2467751771120462e-05,
      "loss": 11.312,
      "step": 19308
    },
    {
      "epoch": 3.3764371584699453,
      "grad_norm": 0.6669335960892877,
      "learning_rate": 1.2460905299324577e-05,
      "loss": 11.4971,
      "step": 19309
    },
    {
      "epoch": 3.3766120218579236,
      "grad_norm": 0.6047877646649171,
      "learning_rate": 1.2454060582956328e-05,
      "loss": 11.5368,
      "step": 19310
    },
    {
      "epoch": 3.3767868852459015,
      "grad_norm": 0.7111851595724814,
      "learning_rate": 1.2447217622152974e-05,
      "loss": 11.0444,
      "step": 19311
    },
    {
      "epoch": 3.37696174863388,
      "grad_norm": 0.699720353312032,
      "learning_rate": 1.2440376417051747e-05,
      "loss": 11.4305,
      "step": 19312
    },
    {
      "epoch": 3.3771366120218578,
      "grad_norm": 0.7008289518277488,
      "learning_rate": 1.2433536967789771e-05,
      "loss": 11.3982,
      "step": 19313
    },
    {
      "epoch": 3.377311475409836,
      "grad_norm": 0.7404382304672382,
      "learning_rate": 1.2426699274504272e-05,
      "loss": 11.2522,
      "step": 19314
    },
    {
      "epoch": 3.377486338797814,
      "grad_norm": 0.721402078150004,
      "learning_rate": 1.2419863337332326e-05,
      "loss": 11.4198,
      "step": 19315
    },
    {
      "epoch": 3.3776612021857924,
      "grad_norm": 0.6531173831843384,
      "learning_rate": 1.2413029156411038e-05,
      "loss": 11.1225,
      "step": 19316
    },
    {
      "epoch": 3.3778360655737707,
      "grad_norm": 0.64681638217275,
      "learning_rate": 1.2406196731877462e-05,
      "loss": 11.4781,
      "step": 19317
    },
    {
      "epoch": 3.3780109289617486,
      "grad_norm": 0.6835988386090905,
      "learning_rate": 1.2399366063868545e-05,
      "loss": 11.2709,
      "step": 19318
    },
    {
      "epoch": 3.378185792349727,
      "grad_norm": 0.6765997956418197,
      "learning_rate": 1.2392537152521321e-05,
      "loss": 11.2801,
      "step": 19319
    },
    {
      "epoch": 3.378360655737705,
      "grad_norm": 0.672260439213729,
      "learning_rate": 1.2385709997972728e-05,
      "loss": 11.5692,
      "step": 19320
    },
    {
      "epoch": 3.378535519125683,
      "grad_norm": 0.694575374424022,
      "learning_rate": 1.2378884600359686e-05,
      "loss": 11.252,
      "step": 19321
    },
    {
      "epoch": 3.378710382513661,
      "grad_norm": 0.676212278850996,
      "learning_rate": 1.237206095981901e-05,
      "loss": 11.1931,
      "step": 19322
    },
    {
      "epoch": 3.3788852459016394,
      "grad_norm": 0.7119559282145296,
      "learning_rate": 1.2365239076487544e-05,
      "loss": 11.4938,
      "step": 19323
    },
    {
      "epoch": 3.3790601092896173,
      "grad_norm": 0.7364552860341972,
      "learning_rate": 1.2358418950502126e-05,
      "loss": 11.4601,
      "step": 19324
    },
    {
      "epoch": 3.3792349726775956,
      "grad_norm": 0.6791883100271972,
      "learning_rate": 1.2351600581999512e-05,
      "loss": 11.5502,
      "step": 19325
    },
    {
      "epoch": 3.379409836065574,
      "grad_norm": 0.6213890914356214,
      "learning_rate": 1.2344783971116436e-05,
      "loss": 11.4531,
      "step": 19326
    },
    {
      "epoch": 3.379584699453552,
      "grad_norm": 0.6682282495648276,
      "learning_rate": 1.2337969117989567e-05,
      "loss": 11.0801,
      "step": 19327
    },
    {
      "epoch": 3.3797595628415302,
      "grad_norm": 0.6740886270866389,
      "learning_rate": 1.2331156022755541e-05,
      "loss": 11.2207,
      "step": 19328
    },
    {
      "epoch": 3.379934426229508,
      "grad_norm": 0.6858722704299499,
      "learning_rate": 1.2324344685551048e-05,
      "loss": 11.5121,
      "step": 19329
    },
    {
      "epoch": 3.3801092896174865,
      "grad_norm": 0.6943547791371149,
      "learning_rate": 1.2317535106512656e-05,
      "loss": 11.3193,
      "step": 19330
    },
    {
      "epoch": 3.3802841530054644,
      "grad_norm": 0.7084321345277614,
      "learning_rate": 1.2310727285776891e-05,
      "loss": 11.5688,
      "step": 19331
    },
    {
      "epoch": 3.3804590163934427,
      "grad_norm": 0.6932021361213156,
      "learning_rate": 1.2303921223480285e-05,
      "loss": 11.2891,
      "step": 19332
    },
    {
      "epoch": 3.3806338797814206,
      "grad_norm": 0.597314231898004,
      "learning_rate": 1.22971169197593e-05,
      "loss": 11.7182,
      "step": 19333
    },
    {
      "epoch": 3.380808743169399,
      "grad_norm": 0.7834623642913676,
      "learning_rate": 1.2290314374750422e-05,
      "loss": 11.1935,
      "step": 19334
    },
    {
      "epoch": 3.3809836065573773,
      "grad_norm": 0.7515618795483668,
      "learning_rate": 1.2283513588590067e-05,
      "loss": 11.45,
      "step": 19335
    },
    {
      "epoch": 3.381158469945355,
      "grad_norm": 0.7075607937141306,
      "learning_rate": 1.227671456141456e-05,
      "loss": 11.3486,
      "step": 19336
    },
    {
      "epoch": 3.3813333333333335,
      "grad_norm": 0.7178777308112954,
      "learning_rate": 1.2269917293360267e-05,
      "loss": 11.4228,
      "step": 19337
    },
    {
      "epoch": 3.3815081967213114,
      "grad_norm": 0.6561042306843388,
      "learning_rate": 1.2263121784563502e-05,
      "loss": 11.4691,
      "step": 19338
    },
    {
      "epoch": 3.3816830601092898,
      "grad_norm": 0.7084785522428909,
      "learning_rate": 1.225632803516057e-05,
      "loss": 11.4579,
      "step": 19339
    },
    {
      "epoch": 3.3818579234972677,
      "grad_norm": 0.7087468194343014,
      "learning_rate": 1.2249536045287647e-05,
      "loss": 11.4359,
      "step": 19340
    },
    {
      "epoch": 3.382032786885246,
      "grad_norm": 0.6836347888877207,
      "learning_rate": 1.2242745815080947e-05,
      "loss": 11.5073,
      "step": 19341
    },
    {
      "epoch": 3.382207650273224,
      "grad_norm": 0.6779822280030696,
      "learning_rate": 1.2235957344676629e-05,
      "loss": 11.4628,
      "step": 19342
    },
    {
      "epoch": 3.3823825136612022,
      "grad_norm": 0.7046125901903225,
      "learning_rate": 1.2229170634210852e-05,
      "loss": 11.0976,
      "step": 19343
    },
    {
      "epoch": 3.38255737704918,
      "grad_norm": 0.7725660160342849,
      "learning_rate": 1.2222385683819714e-05,
      "loss": 11.5423,
      "step": 19344
    },
    {
      "epoch": 3.3827322404371585,
      "grad_norm": 0.647651631136806,
      "learning_rate": 1.2215602493639244e-05,
      "loss": 11.3948,
      "step": 19345
    },
    {
      "epoch": 3.3829071038251364,
      "grad_norm": 0.6760232502952727,
      "learning_rate": 1.2208821063805476e-05,
      "loss": 11.3446,
      "step": 19346
    },
    {
      "epoch": 3.3830819672131147,
      "grad_norm": 0.7402660243745517,
      "learning_rate": 1.2202041394454378e-05,
      "loss": 11.2923,
      "step": 19347
    },
    {
      "epoch": 3.383256830601093,
      "grad_norm": 0.7962610242866834,
      "learning_rate": 1.2195263485721964e-05,
      "loss": 11.59,
      "step": 19348
    },
    {
      "epoch": 3.383431693989071,
      "grad_norm": 0.7286994223055319,
      "learning_rate": 1.2188487337744103e-05,
      "loss": 11.2994,
      "step": 19349
    },
    {
      "epoch": 3.3836065573770493,
      "grad_norm": 0.6934382180620486,
      "learning_rate": 1.2181712950656676e-05,
      "loss": 11.3718,
      "step": 19350
    },
    {
      "epoch": 3.383781420765027,
      "grad_norm": 0.6525451938022316,
      "learning_rate": 1.2174940324595542e-05,
      "loss": 11.4851,
      "step": 19351
    },
    {
      "epoch": 3.3839562841530055,
      "grad_norm": 0.8000727611825218,
      "learning_rate": 1.2168169459696498e-05,
      "loss": 11.0308,
      "step": 19352
    },
    {
      "epoch": 3.3841311475409834,
      "grad_norm": 0.7321569587804829,
      "learning_rate": 1.2161400356095375e-05,
      "loss": 11.3864,
      "step": 19353
    },
    {
      "epoch": 3.3843060109289618,
      "grad_norm": 0.6837569994612879,
      "learning_rate": 1.2154633013927852e-05,
      "loss": 11.4034,
      "step": 19354
    },
    {
      "epoch": 3.3844808743169397,
      "grad_norm": 0.6938120094953052,
      "learning_rate": 1.2147867433329663e-05,
      "loss": 11.4159,
      "step": 19355
    },
    {
      "epoch": 3.384655737704918,
      "grad_norm": 0.6262263317183047,
      "learning_rate": 1.2141103614436466e-05,
      "loss": 11.6041,
      "step": 19356
    },
    {
      "epoch": 3.3848306010928964,
      "grad_norm": 0.721373387218655,
      "learning_rate": 1.213434155738391e-05,
      "loss": 11.5375,
      "step": 19357
    },
    {
      "epoch": 3.3850054644808742,
      "grad_norm": 0.7047012605067732,
      "learning_rate": 1.2127581262307586e-05,
      "loss": 11.4543,
      "step": 19358
    },
    {
      "epoch": 3.3851803278688526,
      "grad_norm": 0.6086812879733698,
      "learning_rate": 1.2120822729343062e-05,
      "loss": 11.7327,
      "step": 19359
    },
    {
      "epoch": 3.3853551912568305,
      "grad_norm": 0.6743437269758963,
      "learning_rate": 1.2114065958625864e-05,
      "loss": 11.2944,
      "step": 19360
    },
    {
      "epoch": 3.385530054644809,
      "grad_norm": 0.66468760318516,
      "learning_rate": 1.2107310950291474e-05,
      "loss": 11.6288,
      "step": 19361
    },
    {
      "epoch": 3.3857049180327867,
      "grad_norm": 0.6565225857856414,
      "learning_rate": 1.2100557704475401e-05,
      "loss": 11.5716,
      "step": 19362
    },
    {
      "epoch": 3.385879781420765,
      "grad_norm": 0.6912859929303748,
      "learning_rate": 1.2093806221313008e-05,
      "loss": 11.5675,
      "step": 19363
    },
    {
      "epoch": 3.386054644808743,
      "grad_norm": 0.7161669585310798,
      "learning_rate": 1.208705650093972e-05,
      "loss": 11.2614,
      "step": 19364
    },
    {
      "epoch": 3.3862295081967213,
      "grad_norm": 0.6464272875255495,
      "learning_rate": 1.208030854349087e-05,
      "loss": 11.2703,
      "step": 19365
    },
    {
      "epoch": 3.3864043715846996,
      "grad_norm": 0.7146528059016518,
      "learning_rate": 1.2073562349101785e-05,
      "loss": 11.4069,
      "step": 19366
    },
    {
      "epoch": 3.3865792349726775,
      "grad_norm": 0.7504548107749396,
      "learning_rate": 1.2066817917907735e-05,
      "loss": 11.4136,
      "step": 19367
    },
    {
      "epoch": 3.386754098360656,
      "grad_norm": 0.6447803724884722,
      "learning_rate": 1.2060075250043989e-05,
      "loss": 11.4271,
      "step": 19368
    },
    {
      "epoch": 3.386928961748634,
      "grad_norm": 0.6506192323833202,
      "learning_rate": 1.205333434564574e-05,
      "loss": 11.3744,
      "step": 19369
    },
    {
      "epoch": 3.387103825136612,
      "grad_norm": 0.648952989630813,
      "learning_rate": 1.2046595204848176e-05,
      "loss": 11.6808,
      "step": 19370
    },
    {
      "epoch": 3.38727868852459,
      "grad_norm": 0.6835619328483019,
      "learning_rate": 1.2039857827786416e-05,
      "loss": 11.5746,
      "step": 19371
    },
    {
      "epoch": 3.3874535519125684,
      "grad_norm": 0.6636728574923418,
      "learning_rate": 1.2033122214595594e-05,
      "loss": 11.4935,
      "step": 19372
    },
    {
      "epoch": 3.3876284153005463,
      "grad_norm": 0.6281570663382087,
      "learning_rate": 1.2026388365410757e-05,
      "loss": 11.2131,
      "step": 19373
    },
    {
      "epoch": 3.3878032786885246,
      "grad_norm": 0.6447156168693511,
      "learning_rate": 1.2019656280366953e-05,
      "loss": 11.2136,
      "step": 19374
    },
    {
      "epoch": 3.387978142076503,
      "grad_norm": 0.6741636211271649,
      "learning_rate": 1.2012925959599175e-05,
      "loss": 11.3472,
      "step": 19375
    },
    {
      "epoch": 3.388153005464481,
      "grad_norm": 0.7062640798132473,
      "learning_rate": 1.2006197403242381e-05,
      "loss": 11.2655,
      "step": 19376
    },
    {
      "epoch": 3.388327868852459,
      "grad_norm": 0.7629085634938527,
      "learning_rate": 1.1999470611431508e-05,
      "loss": 11.4796,
      "step": 19377
    },
    {
      "epoch": 3.388502732240437,
      "grad_norm": 0.659127118590849,
      "learning_rate": 1.1992745584301446e-05,
      "loss": 11.5485,
      "step": 19378
    },
    {
      "epoch": 3.3886775956284154,
      "grad_norm": 0.728179902044679,
      "learning_rate": 1.1986022321987056e-05,
      "loss": 11.058,
      "step": 19379
    },
    {
      "epoch": 3.3888524590163933,
      "grad_norm": 0.652806320798972,
      "learning_rate": 1.1979300824623163e-05,
      "loss": 11.5428,
      "step": 19380
    },
    {
      "epoch": 3.3890273224043717,
      "grad_norm": 0.7284223092797238,
      "learning_rate": 1.1972581092344537e-05,
      "loss": 11.3361,
      "step": 19381
    },
    {
      "epoch": 3.3892021857923496,
      "grad_norm": 0.7072125835727872,
      "learning_rate": 1.1965863125285948e-05,
      "loss": 11.3351,
      "step": 19382
    },
    {
      "epoch": 3.389377049180328,
      "grad_norm": 0.7325159547068884,
      "learning_rate": 1.19591469235821e-05,
      "loss": 11.0377,
      "step": 19383
    },
    {
      "epoch": 3.3895519125683062,
      "grad_norm": 0.6538752030716208,
      "learning_rate": 1.195243248736767e-05,
      "loss": 11.5156,
      "step": 19384
    },
    {
      "epoch": 3.389726775956284,
      "grad_norm": 0.6609453451061148,
      "learning_rate": 1.1945719816777312e-05,
      "loss": 11.2379,
      "step": 19385
    },
    {
      "epoch": 3.3899016393442625,
      "grad_norm": 0.6370843234400574,
      "learning_rate": 1.1939008911945638e-05,
      "loss": 11.2849,
      "step": 19386
    },
    {
      "epoch": 3.3900765027322404,
      "grad_norm": 0.6560545264547976,
      "learning_rate": 1.1932299773007228e-05,
      "loss": 11.597,
      "step": 19387
    },
    {
      "epoch": 3.3902513661202187,
      "grad_norm": 0.6629582180135284,
      "learning_rate": 1.1925592400096597e-05,
      "loss": 11.4075,
      "step": 19388
    },
    {
      "epoch": 3.3904262295081966,
      "grad_norm": 0.7494846688895455,
      "learning_rate": 1.191888679334826e-05,
      "loss": 11.3204,
      "step": 19389
    },
    {
      "epoch": 3.390601092896175,
      "grad_norm": 0.7051681567219457,
      "learning_rate": 1.19121829528967e-05,
      "loss": 11.504,
      "step": 19390
    },
    {
      "epoch": 3.390775956284153,
      "grad_norm": 0.699821013836029,
      "learning_rate": 1.190548087887633e-05,
      "loss": 11.6011,
      "step": 19391
    },
    {
      "epoch": 3.390950819672131,
      "grad_norm": 0.6692187198961949,
      "learning_rate": 1.1898780571421552e-05,
      "loss": 11.2979,
      "step": 19392
    },
    {
      "epoch": 3.3911256830601095,
      "grad_norm": 0.6696842812600494,
      "learning_rate": 1.189208203066673e-05,
      "loss": 11.3877,
      "step": 19393
    },
    {
      "epoch": 3.3913005464480874,
      "grad_norm": 0.7331644316512508,
      "learning_rate": 1.1885385256746195e-05,
      "loss": 11.3724,
      "step": 19394
    },
    {
      "epoch": 3.3914754098360653,
      "grad_norm": 0.7496197817231822,
      "learning_rate": 1.1878690249794234e-05,
      "loss": 11.4453,
      "step": 19395
    },
    {
      "epoch": 3.3916502732240437,
      "grad_norm": 0.7575350075606282,
      "learning_rate": 1.1871997009945091e-05,
      "loss": 11.3897,
      "step": 19396
    },
    {
      "epoch": 3.391825136612022,
      "grad_norm": 0.6418549545770694,
      "learning_rate": 1.1865305537333004e-05,
      "loss": 11.6002,
      "step": 19397
    },
    {
      "epoch": 3.392,
      "grad_norm": 0.6587616774588586,
      "learning_rate": 1.1858615832092156e-05,
      "loss": 11.3,
      "step": 19398
    },
    {
      "epoch": 3.3921748633879782,
      "grad_norm": 0.7333187195210042,
      "learning_rate": 1.1851927894356696e-05,
      "loss": 11.6435,
      "step": 19399
    },
    {
      "epoch": 3.392349726775956,
      "grad_norm": 0.7360926457415249,
      "learning_rate": 1.1845241724260702e-05,
      "loss": 11.378,
      "step": 19400
    },
    {
      "epoch": 3.3925245901639345,
      "grad_norm": 0.7395623046202121,
      "learning_rate": 1.1838557321938303e-05,
      "loss": 11.4792,
      "step": 19401
    },
    {
      "epoch": 3.3926994535519124,
      "grad_norm": 0.6607382039371538,
      "learning_rate": 1.1831874687523514e-05,
      "loss": 11.31,
      "step": 19402
    },
    {
      "epoch": 3.3928743169398907,
      "grad_norm": 0.7446671669834687,
      "learning_rate": 1.1825193821150349e-05,
      "loss": 11.2042,
      "step": 19403
    },
    {
      "epoch": 3.3930491803278686,
      "grad_norm": 0.6633421594062854,
      "learning_rate": 1.181851472295281e-05,
      "loss": 11.2238,
      "step": 19404
    },
    {
      "epoch": 3.393224043715847,
      "grad_norm": 0.6547544722237943,
      "learning_rate": 1.181183739306475e-05,
      "loss": 11.507,
      "step": 19405
    },
    {
      "epoch": 3.3933989071038253,
      "grad_norm": 0.6750194239906674,
      "learning_rate": 1.180516183162016e-05,
      "loss": 11.481,
      "step": 19406
    },
    {
      "epoch": 3.393573770491803,
      "grad_norm": 0.6861913078999456,
      "learning_rate": 1.1798488038752853e-05,
      "loss": 11.3469,
      "step": 19407
    },
    {
      "epoch": 3.3937486338797815,
      "grad_norm": 0.723151387799635,
      "learning_rate": 1.1791816014596702e-05,
      "loss": 10.9889,
      "step": 19408
    },
    {
      "epoch": 3.3939234972677594,
      "grad_norm": 0.668628843448093,
      "learning_rate": 1.1785145759285431e-05,
      "loss": 11.5184,
      "step": 19409
    },
    {
      "epoch": 3.394098360655738,
      "grad_norm": 0.6701294112828234,
      "learning_rate": 1.177847727295287e-05,
      "loss": 11.4898,
      "step": 19410
    },
    {
      "epoch": 3.3942732240437157,
      "grad_norm": 0.6952727114693293,
      "learning_rate": 1.177181055573271e-05,
      "loss": 11.5299,
      "step": 19411
    },
    {
      "epoch": 3.394448087431694,
      "grad_norm": 0.6607167278720015,
      "learning_rate": 1.1765145607758644e-05,
      "loss": 11.4076,
      "step": 19412
    },
    {
      "epoch": 3.394622950819672,
      "grad_norm": 0.6473228862289245,
      "learning_rate": 1.1758482429164352e-05,
      "loss": 11.5943,
      "step": 19413
    },
    {
      "epoch": 3.3947978142076503,
      "grad_norm": 0.7139445834994442,
      "learning_rate": 1.1751821020083375e-05,
      "loss": 11.2585,
      "step": 19414
    },
    {
      "epoch": 3.3949726775956286,
      "grad_norm": 0.6873403287702757,
      "learning_rate": 1.1745161380649372e-05,
      "loss": 11.4626,
      "step": 19415
    },
    {
      "epoch": 3.3951475409836065,
      "grad_norm": 0.6584384137833166,
      "learning_rate": 1.1738503510995857e-05,
      "loss": 11.2571,
      "step": 19416
    },
    {
      "epoch": 3.395322404371585,
      "grad_norm": 0.7188916204315516,
      "learning_rate": 1.173184741125637e-05,
      "loss": 11.0351,
      "step": 19417
    },
    {
      "epoch": 3.3954972677595627,
      "grad_norm": 0.6476659158131516,
      "learning_rate": 1.1725193081564335e-05,
      "loss": 11.543,
      "step": 19418
    },
    {
      "epoch": 3.395672131147541,
      "grad_norm": 0.7421393416052126,
      "learning_rate": 1.1718540522053201e-05,
      "loss": 11.5815,
      "step": 19419
    },
    {
      "epoch": 3.395846994535519,
      "grad_norm": 0.711620620920621,
      "learning_rate": 1.1711889732856406e-05,
      "loss": 11.4616,
      "step": 19420
    },
    {
      "epoch": 3.3960218579234973,
      "grad_norm": 0.7137611095772337,
      "learning_rate": 1.1705240714107302e-05,
      "loss": 11.4881,
      "step": 19421
    },
    {
      "epoch": 3.396196721311475,
      "grad_norm": 0.7070873328217485,
      "learning_rate": 1.1698593465939246e-05,
      "loss": 11.4046,
      "step": 19422
    },
    {
      "epoch": 3.3963715846994535,
      "grad_norm": 0.6779137862297658,
      "learning_rate": 1.1691947988485485e-05,
      "loss": 11.2222,
      "step": 19423
    },
    {
      "epoch": 3.396546448087432,
      "grad_norm": 0.66276225330041,
      "learning_rate": 1.1685304281879294e-05,
      "loss": 11.5867,
      "step": 19424
    },
    {
      "epoch": 3.39672131147541,
      "grad_norm": 0.6676053606953074,
      "learning_rate": 1.1678662346253933e-05,
      "loss": 11.4187,
      "step": 19425
    },
    {
      "epoch": 3.396896174863388,
      "grad_norm": 0.7452378983588804,
      "learning_rate": 1.1672022181742592e-05,
      "loss": 11.2131,
      "step": 19426
    },
    {
      "epoch": 3.397071038251366,
      "grad_norm": 0.6929592359593548,
      "learning_rate": 1.1665383788478402e-05,
      "loss": 11.4399,
      "step": 19427
    },
    {
      "epoch": 3.3972459016393444,
      "grad_norm": 0.7077272688023193,
      "learning_rate": 1.1658747166594475e-05,
      "loss": 11.5328,
      "step": 19428
    },
    {
      "epoch": 3.3974207650273223,
      "grad_norm": 0.6329506992508964,
      "learning_rate": 1.1652112316223895e-05,
      "loss": 11.6594,
      "step": 19429
    },
    {
      "epoch": 3.3975956284153006,
      "grad_norm": 0.6857259537048779,
      "learning_rate": 1.1645479237499746e-05,
      "loss": 11.1528,
      "step": 19430
    },
    {
      "epoch": 3.3977704918032785,
      "grad_norm": 0.702162882270825,
      "learning_rate": 1.163884793055504e-05,
      "loss": 11.4702,
      "step": 19431
    },
    {
      "epoch": 3.397945355191257,
      "grad_norm": 0.6671291403980789,
      "learning_rate": 1.163221839552272e-05,
      "loss": 11.3733,
      "step": 19432
    },
    {
      "epoch": 3.398120218579235,
      "grad_norm": 0.5772651727666426,
      "learning_rate": 1.1625590632535721e-05,
      "loss": 11.612,
      "step": 19433
    },
    {
      "epoch": 3.398295081967213,
      "grad_norm": 0.6746163108721499,
      "learning_rate": 1.1618964641727004e-05,
      "loss": 11.3388,
      "step": 19434
    },
    {
      "epoch": 3.3984699453551914,
      "grad_norm": 0.6846565813103219,
      "learning_rate": 1.161234042322943e-05,
      "loss": 11.5575,
      "step": 19435
    },
    {
      "epoch": 3.3986448087431693,
      "grad_norm": 0.66375748884509,
      "learning_rate": 1.160571797717579e-05,
      "loss": 11.6462,
      "step": 19436
    },
    {
      "epoch": 3.3988196721311477,
      "grad_norm": 0.7023540802672463,
      "learning_rate": 1.1599097303698914e-05,
      "loss": 11.3934,
      "step": 19437
    },
    {
      "epoch": 3.3989945355191256,
      "grad_norm": 0.5848008624070283,
      "learning_rate": 1.159247840293154e-05,
      "loss": 11.6382,
      "step": 19438
    },
    {
      "epoch": 3.399169398907104,
      "grad_norm": 0.6901140606861,
      "learning_rate": 1.1585861275006448e-05,
      "loss": 11.3705,
      "step": 19439
    },
    {
      "epoch": 3.399344262295082,
      "grad_norm": 0.6326062439067981,
      "learning_rate": 1.1579245920056326e-05,
      "loss": 11.3939,
      "step": 19440
    },
    {
      "epoch": 3.39951912568306,
      "grad_norm": 0.7140128070695113,
      "learning_rate": 1.1572632338213785e-05,
      "loss": 11.1673,
      "step": 19441
    },
    {
      "epoch": 3.3996939890710385,
      "grad_norm": 0.6582470355365088,
      "learning_rate": 1.1566020529611466e-05,
      "loss": 11.4264,
      "step": 19442
    },
    {
      "epoch": 3.3998688524590164,
      "grad_norm": 0.6324497591536227,
      "learning_rate": 1.1559410494381951e-05,
      "loss": 11.4694,
      "step": 19443
    },
    {
      "epoch": 3.4000437158469947,
      "grad_norm": 0.7346852479249281,
      "learning_rate": 1.1552802232657844e-05,
      "loss": 11.4255,
      "step": 19444
    },
    {
      "epoch": 3.4002185792349726,
      "grad_norm": 0.6785644111619534,
      "learning_rate": 1.1546195744571608e-05,
      "loss": 11.4327,
      "step": 19445
    },
    {
      "epoch": 3.400393442622951,
      "grad_norm": 0.7141625647876263,
      "learning_rate": 1.1539591030255725e-05,
      "loss": 11.3905,
      "step": 19446
    },
    {
      "epoch": 3.400568306010929,
      "grad_norm": 0.693897775496001,
      "learning_rate": 1.1532988089842655e-05,
      "loss": 11.5894,
      "step": 19447
    },
    {
      "epoch": 3.400743169398907,
      "grad_norm": 0.6380423227136287,
      "learning_rate": 1.1526386923464783e-05,
      "loss": 11.4106,
      "step": 19448
    },
    {
      "epoch": 3.400918032786885,
      "grad_norm": 0.7072063970940682,
      "learning_rate": 1.1519787531254545e-05,
      "loss": 11.5762,
      "step": 19449
    },
    {
      "epoch": 3.4010928961748634,
      "grad_norm": 0.6690137763398986,
      "learning_rate": 1.1513189913344214e-05,
      "loss": 11.4124,
      "step": 19450
    },
    {
      "epoch": 3.4012677595628418,
      "grad_norm": 0.7912548164321923,
      "learning_rate": 1.1506594069866116e-05,
      "loss": 11.3282,
      "step": 19451
    },
    {
      "epoch": 3.4014426229508197,
      "grad_norm": 0.6557463339286027,
      "learning_rate": 1.1500000000952516e-05,
      "loss": 11.6792,
      "step": 19452
    },
    {
      "epoch": 3.4016174863387976,
      "grad_norm": 0.7138811992311642,
      "learning_rate": 1.1493407706735648e-05,
      "loss": 11.4856,
      "step": 19453
    },
    {
      "epoch": 3.401792349726776,
      "grad_norm": 0.7657260723625198,
      "learning_rate": 1.148681718734771e-05,
      "loss": 11.1897,
      "step": 19454
    },
    {
      "epoch": 3.4019672131147543,
      "grad_norm": 0.736281415664722,
      "learning_rate": 1.1480228442920849e-05,
      "loss": 11.4388,
      "step": 19455
    },
    {
      "epoch": 3.402142076502732,
      "grad_norm": 0.6887729340548734,
      "learning_rate": 1.1473641473587204e-05,
      "loss": 11.3879,
      "step": 19456
    },
    {
      "epoch": 3.4023169398907105,
      "grad_norm": 0.7199315532149402,
      "learning_rate": 1.1467056279478838e-05,
      "loss": 11.5068,
      "step": 19457
    },
    {
      "epoch": 3.4024918032786884,
      "grad_norm": 0.6754192516631112,
      "learning_rate": 1.1460472860727877e-05,
      "loss": 11.1669,
      "step": 19458
    },
    {
      "epoch": 3.4026666666666667,
      "grad_norm": 0.6952418293949691,
      "learning_rate": 1.145389121746625e-05,
      "loss": 11.0973,
      "step": 19459
    },
    {
      "epoch": 3.4028415300546446,
      "grad_norm": 0.7015242887929444,
      "learning_rate": 1.1447311349825984e-05,
      "loss": 11.3858,
      "step": 19460
    },
    {
      "epoch": 3.403016393442623,
      "grad_norm": 0.7729910454910052,
      "learning_rate": 1.1440733257939018e-05,
      "loss": 11.4408,
      "step": 19461
    },
    {
      "epoch": 3.403191256830601,
      "grad_norm": 0.7118621047839907,
      "learning_rate": 1.143415694193727e-05,
      "loss": 11.5537,
      "step": 19462
    },
    {
      "epoch": 3.403366120218579,
      "grad_norm": 0.6348179281581242,
      "learning_rate": 1.1427582401952597e-05,
      "loss": 11.4845,
      "step": 19463
    },
    {
      "epoch": 3.4035409836065575,
      "grad_norm": 0.6662473243240871,
      "learning_rate": 1.1421009638116852e-05,
      "loss": 11.4644,
      "step": 19464
    },
    {
      "epoch": 3.4037158469945354,
      "grad_norm": 0.6169733428797939,
      "learning_rate": 1.1414438650561843e-05,
      "loss": 11.2678,
      "step": 19465
    },
    {
      "epoch": 3.403890710382514,
      "grad_norm": 0.6826658939576553,
      "learning_rate": 1.1407869439419327e-05,
      "loss": 11.3818,
      "step": 19466
    },
    {
      "epoch": 3.4040655737704917,
      "grad_norm": 0.6627129985072872,
      "learning_rate": 1.1401302004821046e-05,
      "loss": 11.3578,
      "step": 19467
    },
    {
      "epoch": 3.40424043715847,
      "grad_norm": 0.7218311234702995,
      "learning_rate": 1.1394736346898694e-05,
      "loss": 11.4175,
      "step": 19468
    },
    {
      "epoch": 3.404415300546448,
      "grad_norm": 0.6724452877216651,
      "learning_rate": 1.138817246578393e-05,
      "loss": 11.3432,
      "step": 19469
    },
    {
      "epoch": 3.4045901639344263,
      "grad_norm": 0.7770163330094362,
      "learning_rate": 1.1381610361608374e-05,
      "loss": 11.2563,
      "step": 19470
    },
    {
      "epoch": 3.404765027322404,
      "grad_norm": 0.723273694946376,
      "learning_rate": 1.1375050034503632e-05,
      "loss": 11.422,
      "step": 19471
    },
    {
      "epoch": 3.4049398907103825,
      "grad_norm": 0.7011629725437588,
      "learning_rate": 1.136849148460125e-05,
      "loss": 11.2896,
      "step": 19472
    },
    {
      "epoch": 3.405114754098361,
      "grad_norm": 0.6688774819811485,
      "learning_rate": 1.1361934712032752e-05,
      "loss": 11.4195,
      "step": 19473
    },
    {
      "epoch": 3.4052896174863387,
      "grad_norm": 0.681340862375607,
      "learning_rate": 1.1355379716929616e-05,
      "loss": 11.2239,
      "step": 19474
    },
    {
      "epoch": 3.405464480874317,
      "grad_norm": 0.6736201955297877,
      "learning_rate": 1.1348826499423292e-05,
      "loss": 11.2977,
      "step": 19475
    },
    {
      "epoch": 3.405639344262295,
      "grad_norm": 0.682748459668812,
      "learning_rate": 1.134227505964519e-05,
      "loss": 11.5804,
      "step": 19476
    },
    {
      "epoch": 3.4058142076502733,
      "grad_norm": 0.6488070710924543,
      "learning_rate": 1.1335725397726693e-05,
      "loss": 11.344,
      "step": 19477
    },
    {
      "epoch": 3.405989071038251,
      "grad_norm": 0.7572770348069102,
      "learning_rate": 1.1329177513799127e-05,
      "loss": 11.3608,
      "step": 19478
    },
    {
      "epoch": 3.4061639344262296,
      "grad_norm": 0.6716765857696008,
      "learning_rate": 1.1322631407993811e-05,
      "loss": 11.6116,
      "step": 19479
    },
    {
      "epoch": 3.4063387978142075,
      "grad_norm": 0.6901883587012855,
      "learning_rate": 1.1316087080442017e-05,
      "loss": 11.2858,
      "step": 19480
    },
    {
      "epoch": 3.406513661202186,
      "grad_norm": 0.6701266417863553,
      "learning_rate": 1.1309544531274963e-05,
      "loss": 11.3692,
      "step": 19481
    },
    {
      "epoch": 3.406688524590164,
      "grad_norm": 0.6926646956055877,
      "learning_rate": 1.1303003760623865e-05,
      "loss": 11.3259,
      "step": 19482
    },
    {
      "epoch": 3.406863387978142,
      "grad_norm": 0.7093765784471575,
      "learning_rate": 1.1296464768619864e-05,
      "loss": 11.6271,
      "step": 19483
    },
    {
      "epoch": 3.4070382513661204,
      "grad_norm": 0.7041626533585098,
      "learning_rate": 1.1289927555394109e-05,
      "loss": 11.3122,
      "step": 19484
    },
    {
      "epoch": 3.4072131147540983,
      "grad_norm": 0.6395771531219043,
      "learning_rate": 1.1283392121077686e-05,
      "loss": 11.3489,
      "step": 19485
    },
    {
      "epoch": 3.4073879781420766,
      "grad_norm": 0.7978042037154187,
      "learning_rate": 1.1276858465801643e-05,
      "loss": 11.4461,
      "step": 19486
    },
    {
      "epoch": 3.4075628415300545,
      "grad_norm": 0.7200336989796909,
      "learning_rate": 1.1270326589697e-05,
      "loss": 11.4652,
      "step": 19487
    },
    {
      "epoch": 3.407737704918033,
      "grad_norm": 0.7686524832470755,
      "learning_rate": 1.1263796492894751e-05,
      "loss": 11.2442,
      "step": 19488
    },
    {
      "epoch": 3.4079125683060107,
      "grad_norm": 0.6931903559645086,
      "learning_rate": 1.1257268175525837e-05,
      "loss": 11.3857,
      "step": 19489
    },
    {
      "epoch": 3.408087431693989,
      "grad_norm": 0.7742339595600555,
      "learning_rate": 1.1250741637721163e-05,
      "loss": 11.5713,
      "step": 19490
    },
    {
      "epoch": 3.4082622950819674,
      "grad_norm": 0.7097121568628112,
      "learning_rate": 1.1244216879611624e-05,
      "loss": 11.3935,
      "step": 19491
    },
    {
      "epoch": 3.4084371584699453,
      "grad_norm": 0.6765951573015256,
      "learning_rate": 1.123769390132805e-05,
      "loss": 11.6649,
      "step": 19492
    },
    {
      "epoch": 3.4086120218579237,
      "grad_norm": 0.6804989039792027,
      "learning_rate": 1.1231172703001258e-05,
      "loss": 11.4418,
      "step": 19493
    },
    {
      "epoch": 3.4087868852459016,
      "grad_norm": 0.6590975269839899,
      "learning_rate": 1.1224653284761998e-05,
      "loss": 11.4742,
      "step": 19494
    },
    {
      "epoch": 3.40896174863388,
      "grad_norm": 0.6679270234937882,
      "learning_rate": 1.1218135646741046e-05,
      "loss": 11.2716,
      "step": 19495
    },
    {
      "epoch": 3.409136612021858,
      "grad_norm": 0.7170377498149204,
      "learning_rate": 1.1211619789069017e-05,
      "loss": 11.3904,
      "step": 19496
    },
    {
      "epoch": 3.409311475409836,
      "grad_norm": 0.6950640927738144,
      "learning_rate": 1.1205105711876651e-05,
      "loss": 11.2533,
      "step": 19497
    },
    {
      "epoch": 3.409486338797814,
      "grad_norm": 0.7273203231291375,
      "learning_rate": 1.1198593415294556e-05,
      "loss": 11.278,
      "step": 19498
    },
    {
      "epoch": 3.4096612021857924,
      "grad_norm": 0.6548127590743505,
      "learning_rate": 1.1192082899453326e-05,
      "loss": 11.3075,
      "step": 19499
    },
    {
      "epoch": 3.4098360655737707,
      "grad_norm": 0.7262783533007604,
      "learning_rate": 1.1185574164483514e-05,
      "loss": 11.4226,
      "step": 19500
    },
    {
      "epoch": 3.4100109289617486,
      "grad_norm": 0.7099718071122522,
      "learning_rate": 1.1179067210515603e-05,
      "loss": 11.53,
      "step": 19501
    },
    {
      "epoch": 3.410185792349727,
      "grad_norm": 0.6647166355207599,
      "learning_rate": 1.1172562037680135e-05,
      "loss": 11.5389,
      "step": 19502
    },
    {
      "epoch": 3.410360655737705,
      "grad_norm": 0.6883306195278168,
      "learning_rate": 1.1166058646107535e-05,
      "loss": 11.0434,
      "step": 19503
    },
    {
      "epoch": 3.410535519125683,
      "grad_norm": 0.7033768297117562,
      "learning_rate": 1.1159557035928225e-05,
      "loss": 11.4589,
      "step": 19504
    },
    {
      "epoch": 3.410710382513661,
      "grad_norm": 0.6798727690167561,
      "learning_rate": 1.1153057207272544e-05,
      "loss": 11.6267,
      "step": 19505
    },
    {
      "epoch": 3.4108852459016394,
      "grad_norm": 0.6441663407198627,
      "learning_rate": 1.1146559160270875e-05,
      "loss": 11.0145,
      "step": 19506
    },
    {
      "epoch": 3.4110601092896173,
      "grad_norm": 0.7022666293776063,
      "learning_rate": 1.1140062895053516e-05,
      "loss": 11.4025,
      "step": 19507
    },
    {
      "epoch": 3.4112349726775957,
      "grad_norm": 0.670038270875124,
      "learning_rate": 1.1133568411750727e-05,
      "loss": 11.3582,
      "step": 19508
    },
    {
      "epoch": 3.411409836065574,
      "grad_norm": 0.6550389445095026,
      "learning_rate": 1.1127075710492763e-05,
      "loss": 11.4911,
      "step": 19509
    },
    {
      "epoch": 3.411584699453552,
      "grad_norm": 0.7172079011750498,
      "learning_rate": 1.1120584791409772e-05,
      "loss": 11.4903,
      "step": 19510
    },
    {
      "epoch": 3.41175956284153,
      "grad_norm": 0.6449186575157126,
      "learning_rate": 1.1114095654631974e-05,
      "loss": 11.5586,
      "step": 19511
    },
    {
      "epoch": 3.411934426229508,
      "grad_norm": 0.673520643965724,
      "learning_rate": 1.1107608300289474e-05,
      "loss": 11.4866,
      "step": 19512
    },
    {
      "epoch": 3.4121092896174865,
      "grad_norm": 0.7022163297445154,
      "learning_rate": 1.110112272851238e-05,
      "loss": 11.3095,
      "step": 19513
    },
    {
      "epoch": 3.4122841530054644,
      "grad_norm": 0.6714906321564845,
      "learning_rate": 1.1094638939430702e-05,
      "loss": 11.4915,
      "step": 19514
    },
    {
      "epoch": 3.4124590163934427,
      "grad_norm": 0.6710333537968866,
      "learning_rate": 1.1088156933174487e-05,
      "loss": 11.5205,
      "step": 19515
    },
    {
      "epoch": 3.4126338797814206,
      "grad_norm": 0.6480662170797242,
      "learning_rate": 1.1081676709873722e-05,
      "loss": 11.2873,
      "step": 19516
    },
    {
      "epoch": 3.412808743169399,
      "grad_norm": 0.743327260661386,
      "learning_rate": 1.1075198269658371e-05,
      "loss": 11.561,
      "step": 19517
    },
    {
      "epoch": 3.412983606557377,
      "grad_norm": 0.6879108923026066,
      "learning_rate": 1.106872161265834e-05,
      "loss": 11.5083,
      "step": 19518
    },
    {
      "epoch": 3.413158469945355,
      "grad_norm": 0.7307594168314456,
      "learning_rate": 1.106224673900348e-05,
      "loss": 11.4312,
      "step": 19519
    },
    {
      "epoch": 3.413333333333333,
      "grad_norm": 0.6774553246458158,
      "learning_rate": 1.105577364882362e-05,
      "loss": 11.2511,
      "step": 19520
    },
    {
      "epoch": 3.4135081967213114,
      "grad_norm": 0.731042547778099,
      "learning_rate": 1.1049302342248612e-05,
      "loss": 11.2899,
      "step": 19521
    },
    {
      "epoch": 3.41368306010929,
      "grad_norm": 0.6504206031719764,
      "learning_rate": 1.1042832819408245e-05,
      "loss": 11.2836,
      "step": 19522
    },
    {
      "epoch": 3.4138579234972677,
      "grad_norm": 0.6312827723091757,
      "learning_rate": 1.1036365080432176e-05,
      "loss": 11.4139,
      "step": 19523
    },
    {
      "epoch": 3.414032786885246,
      "grad_norm": 0.7180761869166852,
      "learning_rate": 1.102989912545015e-05,
      "loss": 11.2238,
      "step": 19524
    },
    {
      "epoch": 3.414207650273224,
      "grad_norm": 0.7117059996791504,
      "learning_rate": 1.1023434954591793e-05,
      "loss": 11.3674,
      "step": 19525
    },
    {
      "epoch": 3.4143825136612023,
      "grad_norm": 0.7483156785065226,
      "learning_rate": 1.1016972567986782e-05,
      "loss": 11.2793,
      "step": 19526
    },
    {
      "epoch": 3.41455737704918,
      "grad_norm": 0.6624315970429766,
      "learning_rate": 1.1010511965764714e-05,
      "loss": 11.2305,
      "step": 19527
    },
    {
      "epoch": 3.4147322404371585,
      "grad_norm": 0.681255851590091,
      "learning_rate": 1.1004053148055083e-05,
      "loss": 11.5799,
      "step": 19528
    },
    {
      "epoch": 3.4149071038251364,
      "grad_norm": 0.7015090082951186,
      "learning_rate": 1.099759611498743e-05,
      "loss": 11.3466,
      "step": 19529
    },
    {
      "epoch": 3.4150819672131147,
      "grad_norm": 0.6494855289941975,
      "learning_rate": 1.0991140866691262e-05,
      "loss": 11.5018,
      "step": 19530
    },
    {
      "epoch": 3.415256830601093,
      "grad_norm": 0.6768711610835918,
      "learning_rate": 1.0984687403296035e-05,
      "loss": 11.4737,
      "step": 19531
    },
    {
      "epoch": 3.415431693989071,
      "grad_norm": 0.6700220764063055,
      "learning_rate": 1.0978235724931107e-05,
      "loss": 11.5861,
      "step": 19532
    },
    {
      "epoch": 3.4156065573770493,
      "grad_norm": 0.6557798342855979,
      "learning_rate": 1.0971785831725901e-05,
      "loss": 11.519,
      "step": 19533
    },
    {
      "epoch": 3.415781420765027,
      "grad_norm": 0.6665268180476345,
      "learning_rate": 1.096533772380971e-05,
      "loss": 11.4573,
      "step": 19534
    },
    {
      "epoch": 3.4159562841530056,
      "grad_norm": 0.6838892208632652,
      "learning_rate": 1.09588914013119e-05,
      "loss": 11.3648,
      "step": 19535
    },
    {
      "epoch": 3.4161311475409835,
      "grad_norm": 0.7155239275758993,
      "learning_rate": 1.0952446864361721e-05,
      "loss": 11.4416,
      "step": 19536
    },
    {
      "epoch": 3.416306010928962,
      "grad_norm": 0.6811972402137717,
      "learning_rate": 1.0946004113088381e-05,
      "loss": 11.3347,
      "step": 19537
    },
    {
      "epoch": 3.4164808743169397,
      "grad_norm": 0.6994730687923699,
      "learning_rate": 1.093956314762109e-05,
      "loss": 11.4032,
      "step": 19538
    },
    {
      "epoch": 3.416655737704918,
      "grad_norm": 0.6582623886956042,
      "learning_rate": 1.0933123968088987e-05,
      "loss": 11.5031,
      "step": 19539
    },
    {
      "epoch": 3.4168306010928964,
      "grad_norm": 0.7017519575051054,
      "learning_rate": 1.092668657462126e-05,
      "loss": 11.4842,
      "step": 19540
    },
    {
      "epoch": 3.4170054644808743,
      "grad_norm": 0.7077169738153811,
      "learning_rate": 1.0920250967346935e-05,
      "loss": 11.2897,
      "step": 19541
    },
    {
      "epoch": 3.4171803278688526,
      "grad_norm": 0.6789575938731134,
      "learning_rate": 1.0913817146395088e-05,
      "loss": 11.0736,
      "step": 19542
    },
    {
      "epoch": 3.4173551912568305,
      "grad_norm": 0.6790968047654906,
      "learning_rate": 1.0907385111894752e-05,
      "loss": 11.3597,
      "step": 19543
    },
    {
      "epoch": 3.417530054644809,
      "grad_norm": 0.632815484429048,
      "learning_rate": 1.0900954863974855e-05,
      "loss": 11.5681,
      "step": 19544
    },
    {
      "epoch": 3.4177049180327868,
      "grad_norm": 0.6616990833785903,
      "learning_rate": 1.0894526402764437e-05,
      "loss": 11.4433,
      "step": 19545
    },
    {
      "epoch": 3.417879781420765,
      "grad_norm": 0.6720302191672118,
      "learning_rate": 1.0888099728392331e-05,
      "loss": 11.3183,
      "step": 19546
    },
    {
      "epoch": 3.418054644808743,
      "grad_norm": 0.7266159137659904,
      "learning_rate": 1.0881674840987433e-05,
      "loss": 11.3943,
      "step": 19547
    },
    {
      "epoch": 3.4182295081967213,
      "grad_norm": 0.6710576596801453,
      "learning_rate": 1.0875251740678582e-05,
      "loss": 11.3191,
      "step": 19548
    },
    {
      "epoch": 3.4184043715846997,
      "grad_norm": 0.7029693987044007,
      "learning_rate": 1.086883042759459e-05,
      "loss": 10.9725,
      "step": 19549
    },
    {
      "epoch": 3.4185792349726776,
      "grad_norm": 0.7141439988226523,
      "learning_rate": 1.0862410901864206e-05,
      "loss": 11.4079,
      "step": 19550
    },
    {
      "epoch": 3.418754098360656,
      "grad_norm": 0.6936218013199325,
      "learning_rate": 1.0855993163616174e-05,
      "loss": 11.282,
      "step": 19551
    },
    {
      "epoch": 3.418928961748634,
      "grad_norm": 0.7537490254290147,
      "learning_rate": 1.084957721297919e-05,
      "loss": 11.5176,
      "step": 19552
    },
    {
      "epoch": 3.419103825136612,
      "grad_norm": 0.7131826375210213,
      "learning_rate": 1.0843163050081896e-05,
      "loss": 11.2476,
      "step": 19553
    },
    {
      "epoch": 3.41927868852459,
      "grad_norm": 0.6846712020052236,
      "learning_rate": 1.0836750675052964e-05,
      "loss": 11.3727,
      "step": 19554
    },
    {
      "epoch": 3.4194535519125684,
      "grad_norm": 0.6223212162333804,
      "learning_rate": 1.083034008802093e-05,
      "loss": 11.517,
      "step": 19555
    },
    {
      "epoch": 3.4196284153005463,
      "grad_norm": 0.6941979957949274,
      "learning_rate": 1.0823931289114364e-05,
      "loss": 11.2961,
      "step": 19556
    },
    {
      "epoch": 3.4198032786885246,
      "grad_norm": 0.6403019194690904,
      "learning_rate": 1.0817524278461776e-05,
      "loss": 11.712,
      "step": 19557
    },
    {
      "epoch": 3.419978142076503,
      "grad_norm": 0.6783849061166551,
      "learning_rate": 1.0811119056191665e-05,
      "loss": 11.6418,
      "step": 19558
    },
    {
      "epoch": 3.420153005464481,
      "grad_norm": 0.7552603563364919,
      "learning_rate": 1.0804715622432449e-05,
      "loss": 11.3322,
      "step": 19559
    },
    {
      "epoch": 3.420327868852459,
      "grad_norm": 0.6563104415427056,
      "learning_rate": 1.0798313977312557e-05,
      "loss": 11.5527,
      "step": 19560
    },
    {
      "epoch": 3.420502732240437,
      "grad_norm": 0.6540030230738342,
      "learning_rate": 1.0791914120960356e-05,
      "loss": 11.4492,
      "step": 19561
    },
    {
      "epoch": 3.4206775956284154,
      "grad_norm": 0.6954187687284048,
      "learning_rate": 1.0785516053504185e-05,
      "loss": 11.5827,
      "step": 19562
    },
    {
      "epoch": 3.4208524590163933,
      "grad_norm": 0.7235946840314683,
      "learning_rate": 1.0779119775072333e-05,
      "loss": 11.4926,
      "step": 19563
    },
    {
      "epoch": 3.4210273224043717,
      "grad_norm": 0.7164412333624183,
      "learning_rate": 1.0772725285793073e-05,
      "loss": 11.5725,
      "step": 19564
    },
    {
      "epoch": 3.4212021857923496,
      "grad_norm": 0.7047236864168718,
      "learning_rate": 1.0766332585794647e-05,
      "loss": 11.4258,
      "step": 19565
    },
    {
      "epoch": 3.421377049180328,
      "grad_norm": 0.6618104328971143,
      "learning_rate": 1.0759941675205221e-05,
      "loss": 11.3892,
      "step": 19566
    },
    {
      "epoch": 3.421551912568306,
      "grad_norm": 0.6339275128396816,
      "learning_rate": 1.075355255415298e-05,
      "loss": 11.6,
      "step": 19567
    },
    {
      "epoch": 3.421726775956284,
      "grad_norm": 0.6866710930758841,
      "learning_rate": 1.074716522276602e-05,
      "loss": 11.4826,
      "step": 19568
    },
    {
      "epoch": 3.421901639344262,
      "grad_norm": 0.6445500509731057,
      "learning_rate": 1.0740779681172453e-05,
      "loss": 11.4623,
      "step": 19569
    },
    {
      "epoch": 3.4220765027322404,
      "grad_norm": 0.6676467083671458,
      "learning_rate": 1.0734395929500307e-05,
      "loss": 11.3113,
      "step": 19570
    },
    {
      "epoch": 3.4222513661202187,
      "grad_norm": 0.7300695678032524,
      "learning_rate": 1.0728013967877603e-05,
      "loss": 11.4675,
      "step": 19571
    },
    {
      "epoch": 3.4224262295081966,
      "grad_norm": 0.6788522663874985,
      "learning_rate": 1.0721633796432328e-05,
      "loss": 11.6518,
      "step": 19572
    },
    {
      "epoch": 3.422601092896175,
      "grad_norm": 0.7137296495721487,
      "learning_rate": 1.0715255415292413e-05,
      "loss": 11.4464,
      "step": 19573
    },
    {
      "epoch": 3.422775956284153,
      "grad_norm": 0.711284086090492,
      "learning_rate": 1.0708878824585766e-05,
      "loss": 11.5262,
      "step": 19574
    },
    {
      "epoch": 3.422950819672131,
      "grad_norm": 0.7406491737992926,
      "learning_rate": 1.0702504024440252e-05,
      "loss": 11.35,
      "step": 19575
    },
    {
      "epoch": 3.423125683060109,
      "grad_norm": 0.7064426616469401,
      "learning_rate": 1.0696131014983712e-05,
      "loss": 11.512,
      "step": 19576
    },
    {
      "epoch": 3.4233005464480875,
      "grad_norm": 0.6589113696094194,
      "learning_rate": 1.0689759796343956e-05,
      "loss": 11.3637,
      "step": 19577
    },
    {
      "epoch": 3.4234754098360654,
      "grad_norm": 0.8513942318919431,
      "learning_rate": 1.0683390368648726e-05,
      "loss": 11.0524,
      "step": 19578
    },
    {
      "epoch": 3.4236502732240437,
      "grad_norm": 0.6865366629602023,
      "learning_rate": 1.0677022732025754e-05,
      "loss": 11.3548,
      "step": 19579
    },
    {
      "epoch": 3.423825136612022,
      "grad_norm": 0.6528475894189336,
      "learning_rate": 1.0670656886602748e-05,
      "loss": 11.3969,
      "step": 19580
    },
    {
      "epoch": 3.424,
      "grad_norm": 0.6574347833218649,
      "learning_rate": 1.0664292832507339e-05,
      "loss": 11.2999,
      "step": 19581
    },
    {
      "epoch": 3.4241748633879783,
      "grad_norm": 0.7861709224034834,
      "learning_rate": 1.0657930569867147e-05,
      "loss": 11.2636,
      "step": 19582
    },
    {
      "epoch": 3.424349726775956,
      "grad_norm": 0.716744988116514,
      "learning_rate": 1.065157009880977e-05,
      "loss": 11.4542,
      "step": 19583
    },
    {
      "epoch": 3.4245245901639345,
      "grad_norm": 0.6708487225132239,
      "learning_rate": 1.0645211419462753e-05,
      "loss": 11.2653,
      "step": 19584
    },
    {
      "epoch": 3.4246994535519124,
      "grad_norm": 0.6669776788815551,
      "learning_rate": 1.0638854531953601e-05,
      "loss": 11.6354,
      "step": 19585
    },
    {
      "epoch": 3.4248743169398908,
      "grad_norm": 0.7233336335818117,
      "learning_rate": 1.0632499436409803e-05,
      "loss": 11.4358,
      "step": 19586
    },
    {
      "epoch": 3.4250491803278686,
      "grad_norm": 0.7540102400751276,
      "learning_rate": 1.0626146132958759e-05,
      "loss": 11.3399,
      "step": 19587
    },
    {
      "epoch": 3.425224043715847,
      "grad_norm": 0.6671396867538152,
      "learning_rate": 1.0619794621727907e-05,
      "loss": 11.4516,
      "step": 19588
    },
    {
      "epoch": 3.4253989071038253,
      "grad_norm": 0.6443695808730046,
      "learning_rate": 1.0613444902844617e-05,
      "loss": 11.3227,
      "step": 19589
    },
    {
      "epoch": 3.4255737704918032,
      "grad_norm": 0.7228698789628702,
      "learning_rate": 1.0607096976436193e-05,
      "loss": 11.3886,
      "step": 19590
    },
    {
      "epoch": 3.4257486338797816,
      "grad_norm": 0.7956314415788527,
      "learning_rate": 1.060075084262998e-05,
      "loss": 11.1651,
      "step": 19591
    },
    {
      "epoch": 3.4259234972677595,
      "grad_norm": 0.7305733517836608,
      "learning_rate": 1.0594406501553166e-05,
      "loss": 11.4521,
      "step": 19592
    },
    {
      "epoch": 3.426098360655738,
      "grad_norm": 0.7096303283426766,
      "learning_rate": 1.0588063953333027e-05,
      "loss": 11.4522,
      "step": 19593
    },
    {
      "epoch": 3.4262732240437157,
      "grad_norm": 0.6764465679899642,
      "learning_rate": 1.0581723198096737e-05,
      "loss": 11.5726,
      "step": 19594
    },
    {
      "epoch": 3.426448087431694,
      "grad_norm": 0.7641993921615087,
      "learning_rate": 1.0575384235971465e-05,
      "loss": 11.2924,
      "step": 19595
    },
    {
      "epoch": 3.426622950819672,
      "grad_norm": 0.6822828545262367,
      "learning_rate": 1.0569047067084293e-05,
      "loss": 11.3178,
      "step": 19596
    },
    {
      "epoch": 3.4267978142076503,
      "grad_norm": 0.6751311698714064,
      "learning_rate": 1.0562711691562287e-05,
      "loss": 11.4842,
      "step": 19597
    },
    {
      "epoch": 3.4269726775956286,
      "grad_norm": 0.6479596149596567,
      "learning_rate": 1.055637810953255e-05,
      "loss": 11.5695,
      "step": 19598
    },
    {
      "epoch": 3.4271475409836065,
      "grad_norm": 0.6364812841276576,
      "learning_rate": 1.0550046321122054e-05,
      "loss": 11.3692,
      "step": 19599
    },
    {
      "epoch": 3.427322404371585,
      "grad_norm": 0.6779466069386964,
      "learning_rate": 1.0543716326457808e-05,
      "loss": 11.2666,
      "step": 19600
    },
    {
      "epoch": 3.4274972677595628,
      "grad_norm": 0.7023483993642834,
      "learning_rate": 1.053738812566667e-05,
      "loss": 11.4387,
      "step": 19601
    },
    {
      "epoch": 3.427672131147541,
      "grad_norm": 0.677683855167753,
      "learning_rate": 1.0531061718875602e-05,
      "loss": 11.2133,
      "step": 19602
    },
    {
      "epoch": 3.427846994535519,
      "grad_norm": 0.6848249016663641,
      "learning_rate": 1.0524737106211446e-05,
      "loss": 11.4036,
      "step": 19603
    },
    {
      "epoch": 3.4280218579234973,
      "grad_norm": 0.6955927858333998,
      "learning_rate": 1.0518414287801049e-05,
      "loss": 11.417,
      "step": 19604
    },
    {
      "epoch": 3.4281967213114752,
      "grad_norm": 0.7365615586165807,
      "learning_rate": 1.0512093263771206e-05,
      "loss": 11.0989,
      "step": 19605
    },
    {
      "epoch": 3.4283715846994536,
      "grad_norm": 0.6451773839056599,
      "learning_rate": 1.0505774034248616e-05,
      "loss": 11.5406,
      "step": 19606
    },
    {
      "epoch": 3.428546448087432,
      "grad_norm": 0.6660236246404461,
      "learning_rate": 1.049945659936007e-05,
      "loss": 11.3445,
      "step": 19607
    },
    {
      "epoch": 3.42872131147541,
      "grad_norm": 0.6651950430497459,
      "learning_rate": 1.0493140959232206e-05,
      "loss": 11.6132,
      "step": 19608
    },
    {
      "epoch": 3.428896174863388,
      "grad_norm": 0.709494724805886,
      "learning_rate": 1.0486827113991715e-05,
      "loss": 11.5617,
      "step": 19609
    },
    {
      "epoch": 3.429071038251366,
      "grad_norm": 0.6265708130380256,
      "learning_rate": 1.0480515063765162e-05,
      "loss": 11.628,
      "step": 19610
    },
    {
      "epoch": 3.4292459016393444,
      "grad_norm": 0.7546900814421224,
      "learning_rate": 1.0474204808679122e-05,
      "loss": 11.5869,
      "step": 19611
    },
    {
      "epoch": 3.4294207650273223,
      "grad_norm": 0.7011563231360226,
      "learning_rate": 1.0467896348860185e-05,
      "loss": 11.3472,
      "step": 19612
    },
    {
      "epoch": 3.4295956284153006,
      "grad_norm": 0.6780103946325036,
      "learning_rate": 1.0461589684434814e-05,
      "loss": 11.3815,
      "step": 19613
    },
    {
      "epoch": 3.4297704918032785,
      "grad_norm": 0.6792571036152125,
      "learning_rate": 1.045528481552951e-05,
      "loss": 11.4416,
      "step": 19614
    },
    {
      "epoch": 3.429945355191257,
      "grad_norm": 0.7097847524210654,
      "learning_rate": 1.044898174227067e-05,
      "loss": 11.5465,
      "step": 19615
    },
    {
      "epoch": 3.430120218579235,
      "grad_norm": 0.647119985753044,
      "learning_rate": 1.0442680464784682e-05,
      "loss": 11.469,
      "step": 19616
    },
    {
      "epoch": 3.430295081967213,
      "grad_norm": 0.7351976639425237,
      "learning_rate": 1.0436380983197946e-05,
      "loss": 11.6279,
      "step": 19617
    },
    {
      "epoch": 3.430469945355191,
      "grad_norm": 0.6599178385039851,
      "learning_rate": 1.0430083297636783e-05,
      "loss": 11.6935,
      "step": 19618
    },
    {
      "epoch": 3.4306448087431693,
      "grad_norm": 0.7122474916522585,
      "learning_rate": 1.0423787408227447e-05,
      "loss": 11.3512,
      "step": 19619
    },
    {
      "epoch": 3.4308196721311477,
      "grad_norm": 0.6466893357908444,
      "learning_rate": 1.0417493315096216e-05,
      "loss": 11.3396,
      "step": 19620
    },
    {
      "epoch": 3.4309945355191256,
      "grad_norm": 0.6773779602938272,
      "learning_rate": 1.0411201018369277e-05,
      "loss": 11.3179,
      "step": 19621
    },
    {
      "epoch": 3.431169398907104,
      "grad_norm": 0.6923641935381091,
      "learning_rate": 1.040491051817285e-05,
      "loss": 11.4427,
      "step": 19622
    },
    {
      "epoch": 3.431344262295082,
      "grad_norm": 0.6317719270130805,
      "learning_rate": 1.039862181463307e-05,
      "loss": 11.251,
      "step": 19623
    },
    {
      "epoch": 3.43151912568306,
      "grad_norm": 0.6253039067191304,
      "learning_rate": 1.0392334907876022e-05,
      "loss": 11.4309,
      "step": 19624
    },
    {
      "epoch": 3.431693989071038,
      "grad_norm": 0.7147907785974644,
      "learning_rate": 1.0386049798027776e-05,
      "loss": 11.2973,
      "step": 19625
    },
    {
      "epoch": 3.4318688524590164,
      "grad_norm": 0.6797609560580775,
      "learning_rate": 1.0379766485214403e-05,
      "loss": 11.4554,
      "step": 19626
    },
    {
      "epoch": 3.4320437158469943,
      "grad_norm": 0.6840443194932276,
      "learning_rate": 1.0373484969561897e-05,
      "loss": 11.5185,
      "step": 19627
    },
    {
      "epoch": 3.4322185792349726,
      "grad_norm": 0.7026949154685266,
      "learning_rate": 1.0367205251196187e-05,
      "loss": 11.1211,
      "step": 19628
    },
    {
      "epoch": 3.432393442622951,
      "grad_norm": 0.6646273647208899,
      "learning_rate": 1.036092733024322e-05,
      "loss": 11.5967,
      "step": 19629
    },
    {
      "epoch": 3.432568306010929,
      "grad_norm": 0.6649015397246024,
      "learning_rate": 1.035465120682887e-05,
      "loss": 11.4825,
      "step": 19630
    },
    {
      "epoch": 3.4327431693989072,
      "grad_norm": 0.6979671973723209,
      "learning_rate": 1.0348376881079037e-05,
      "loss": 11.3972,
      "step": 19631
    },
    {
      "epoch": 3.432918032786885,
      "grad_norm": 0.7156774858735617,
      "learning_rate": 1.034210435311952e-05,
      "loss": 11.3872,
      "step": 19632
    },
    {
      "epoch": 3.4330928961748635,
      "grad_norm": 0.6712527554720589,
      "learning_rate": 1.0335833623076086e-05,
      "loss": 11.6091,
      "step": 19633
    },
    {
      "epoch": 3.4332677595628414,
      "grad_norm": 0.7136171449994265,
      "learning_rate": 1.032956469107449e-05,
      "loss": 11.4565,
      "step": 19634
    },
    {
      "epoch": 3.4334426229508197,
      "grad_norm": 0.6508549739294689,
      "learning_rate": 1.032329755724043e-05,
      "loss": 11.3452,
      "step": 19635
    },
    {
      "epoch": 3.4336174863387976,
      "grad_norm": 0.7504386523089761,
      "learning_rate": 1.0317032221699641e-05,
      "loss": 11.2002,
      "step": 19636
    },
    {
      "epoch": 3.433792349726776,
      "grad_norm": 0.6250149957283074,
      "learning_rate": 1.0310768684577687e-05,
      "loss": 11.4625,
      "step": 19637
    },
    {
      "epoch": 3.4339672131147543,
      "grad_norm": 0.6610546089847854,
      "learning_rate": 1.0304506946000215e-05,
      "loss": 11.2092,
      "step": 19638
    },
    {
      "epoch": 3.434142076502732,
      "grad_norm": 0.7324584045524787,
      "learning_rate": 1.0298247006092765e-05,
      "loss": 11.5082,
      "step": 19639
    },
    {
      "epoch": 3.4343169398907105,
      "grad_norm": 0.699834838887423,
      "learning_rate": 1.0291988864980872e-05,
      "loss": 11.5142,
      "step": 19640
    },
    {
      "epoch": 3.4344918032786884,
      "grad_norm": 0.6849862930315422,
      "learning_rate": 1.0285732522790092e-05,
      "loss": 11.5527,
      "step": 19641
    },
    {
      "epoch": 3.4346666666666668,
      "grad_norm": 0.6882339887533795,
      "learning_rate": 1.0279477979645802e-05,
      "loss": 11.6263,
      "step": 19642
    },
    {
      "epoch": 3.4348415300546447,
      "grad_norm": 0.6657883995256113,
      "learning_rate": 1.0273225235673457e-05,
      "loss": 11.598,
      "step": 19643
    },
    {
      "epoch": 3.435016393442623,
      "grad_norm": 0.7231566852246751,
      "learning_rate": 1.0266974290998444e-05,
      "loss": 11.2597,
      "step": 19644
    },
    {
      "epoch": 3.435191256830601,
      "grad_norm": 0.6849146035227425,
      "learning_rate": 1.026072514574612e-05,
      "loss": 11.3546,
      "step": 19645
    },
    {
      "epoch": 3.4353661202185792,
      "grad_norm": 0.6049322743324276,
      "learning_rate": 1.0254477800041784e-05,
      "loss": 11.0229,
      "step": 19646
    },
    {
      "epoch": 3.4355409836065576,
      "grad_norm": 0.6559550386848629,
      "learning_rate": 1.0248232254010726e-05,
      "loss": 11.0223,
      "step": 19647
    },
    {
      "epoch": 3.4357158469945355,
      "grad_norm": 0.6423930887541298,
      "learning_rate": 1.0241988507778178e-05,
      "loss": 11.6089,
      "step": 19648
    },
    {
      "epoch": 3.435890710382514,
      "grad_norm": 0.7004825995248811,
      "learning_rate": 1.023574656146935e-05,
      "loss": 11.1604,
      "step": 19649
    },
    {
      "epoch": 3.4360655737704917,
      "grad_norm": 0.6223483262561084,
      "learning_rate": 1.0229506415209444e-05,
      "loss": 11.2871,
      "step": 19650
    },
    {
      "epoch": 3.43624043715847,
      "grad_norm": 0.6774384926507471,
      "learning_rate": 1.0223268069123548e-05,
      "loss": 11.5064,
      "step": 19651
    },
    {
      "epoch": 3.436415300546448,
      "grad_norm": 0.692015149179145,
      "learning_rate": 1.0217031523336784e-05,
      "loss": 11.3465,
      "step": 19652
    },
    {
      "epoch": 3.4365901639344263,
      "grad_norm": 0.6703111357331845,
      "learning_rate": 1.0210796777974197e-05,
      "loss": 11.2252,
      "step": 19653
    },
    {
      "epoch": 3.436765027322404,
      "grad_norm": 0.7036633809834459,
      "learning_rate": 1.0204563833160829e-05,
      "loss": 11.0675,
      "step": 19654
    },
    {
      "epoch": 3.4369398907103825,
      "grad_norm": 0.6526912803971358,
      "learning_rate": 1.0198332689021661e-05,
      "loss": 11.7226,
      "step": 19655
    },
    {
      "epoch": 3.437114754098361,
      "grad_norm": 0.6370189482714445,
      "learning_rate": 1.0192103345681659e-05,
      "loss": 11.4325,
      "step": 19656
    },
    {
      "epoch": 3.4372896174863388,
      "grad_norm": 0.6547135434099725,
      "learning_rate": 1.0185875803265722e-05,
      "loss": 11.4555,
      "step": 19657
    },
    {
      "epoch": 3.437464480874317,
      "grad_norm": 0.652743977828125,
      "learning_rate": 1.0179650061898749e-05,
      "loss": 11.3158,
      "step": 19658
    },
    {
      "epoch": 3.437639344262295,
      "grad_norm": 0.691614101332998,
      "learning_rate": 1.0173426121705576e-05,
      "loss": 11.4919,
      "step": 19659
    },
    {
      "epoch": 3.4378142076502733,
      "grad_norm": 0.6321204230468687,
      "learning_rate": 1.0167203982811002e-05,
      "loss": 11.6469,
      "step": 19660
    },
    {
      "epoch": 3.4379890710382512,
      "grad_norm": 0.5994470482459944,
      "learning_rate": 1.0160983645339828e-05,
      "loss": 11.3482,
      "step": 19661
    },
    {
      "epoch": 3.4381639344262296,
      "grad_norm": 0.7686076427004831,
      "learning_rate": 1.0154765109416764e-05,
      "loss": 11.3207,
      "step": 19662
    },
    {
      "epoch": 3.4383387978142075,
      "grad_norm": 0.7208162370374523,
      "learning_rate": 1.0148548375166523e-05,
      "loss": 11.5502,
      "step": 19663
    },
    {
      "epoch": 3.438513661202186,
      "grad_norm": 0.7453000749454984,
      "learning_rate": 1.0142333442713759e-05,
      "loss": 11.3888,
      "step": 19664
    },
    {
      "epoch": 3.438688524590164,
      "grad_norm": 0.6715294934421908,
      "learning_rate": 1.0136120312183118e-05,
      "loss": 11.5486,
      "step": 19665
    },
    {
      "epoch": 3.438863387978142,
      "grad_norm": 0.7138077163967845,
      "learning_rate": 1.0129908983699177e-05,
      "loss": 11.5501,
      "step": 19666
    },
    {
      "epoch": 3.4390382513661204,
      "grad_norm": 0.6888009619595845,
      "learning_rate": 1.0123699457386504e-05,
      "loss": 10.9183,
      "step": 19667
    },
    {
      "epoch": 3.4392131147540983,
      "grad_norm": 0.694781474743157,
      "learning_rate": 1.0117491733369611e-05,
      "loss": 11.5953,
      "step": 19668
    },
    {
      "epoch": 3.4393879781420766,
      "grad_norm": 0.7296143968612876,
      "learning_rate": 1.0111285811772987e-05,
      "loss": 11.4339,
      "step": 19669
    },
    {
      "epoch": 3.4395628415300545,
      "grad_norm": 0.6153744330150476,
      "learning_rate": 1.0105081692721075e-05,
      "loss": 11.5238,
      "step": 19670
    },
    {
      "epoch": 3.439737704918033,
      "grad_norm": 0.8152183972073233,
      "learning_rate": 1.0098879376338289e-05,
      "loss": 11.2187,
      "step": 19671
    },
    {
      "epoch": 3.4399125683060108,
      "grad_norm": 0.6199263298757888,
      "learning_rate": 1.0092678862749006e-05,
      "loss": 11.4982,
      "step": 19672
    },
    {
      "epoch": 3.440087431693989,
      "grad_norm": 0.6567524767111017,
      "learning_rate": 1.008648015207756e-05,
      "loss": 11.1825,
      "step": 19673
    },
    {
      "epoch": 3.4402622950819675,
      "grad_norm": 0.6624906190662622,
      "learning_rate": 1.0080283244448251e-05,
      "loss": 11.368,
      "step": 19674
    },
    {
      "epoch": 3.4404371584699454,
      "grad_norm": 0.6917636864295744,
      "learning_rate": 1.007408813998536e-05,
      "loss": 11.3544,
      "step": 19675
    },
    {
      "epoch": 3.4406120218579233,
      "grad_norm": 0.647925099353744,
      "learning_rate": 1.0067894838813118e-05,
      "loss": 11.6082,
      "step": 19676
    },
    {
      "epoch": 3.4407868852459016,
      "grad_norm": 0.6638081170882918,
      "learning_rate": 1.0061703341055706e-05,
      "loss": 11.553,
      "step": 19677
    },
    {
      "epoch": 3.44096174863388,
      "grad_norm": 0.6774268654458513,
      "learning_rate": 1.0055513646837289e-05,
      "loss": 11.3785,
      "step": 19678
    },
    {
      "epoch": 3.441136612021858,
      "grad_norm": 0.6337788416633541,
      "learning_rate": 1.004932575628199e-05,
      "loss": 11.4596,
      "step": 19679
    },
    {
      "epoch": 3.441311475409836,
      "grad_norm": 0.7011964880850062,
      "learning_rate": 1.004313966951389e-05,
      "loss": 11.3885,
      "step": 19680
    },
    {
      "epoch": 3.441486338797814,
      "grad_norm": 0.6864034017993512,
      "learning_rate": 1.0036955386657043e-05,
      "loss": 11.3249,
      "step": 19681
    },
    {
      "epoch": 3.4416612021857924,
      "grad_norm": 0.626874497962871,
      "learning_rate": 1.0030772907835483e-05,
      "loss": 11.4655,
      "step": 19682
    },
    {
      "epoch": 3.4418360655737703,
      "grad_norm": 0.6350981759225945,
      "learning_rate": 1.0024592233173136e-05,
      "loss": 11.5595,
      "step": 19683
    },
    {
      "epoch": 3.4420109289617487,
      "grad_norm": 0.6444125006874779,
      "learning_rate": 1.0018413362793989e-05,
      "loss": 11.203,
      "step": 19684
    },
    {
      "epoch": 3.4421857923497265,
      "grad_norm": 0.7309317297554806,
      "learning_rate": 1.0012236296821941e-05,
      "loss": 11.4982,
      "step": 19685
    },
    {
      "epoch": 3.442360655737705,
      "grad_norm": 0.6789300355244275,
      "learning_rate": 1.0006061035380843e-05,
      "loss": 11.2872,
      "step": 19686
    },
    {
      "epoch": 3.4425355191256832,
      "grad_norm": 0.6682334299173938,
      "learning_rate": 9.999887578594569e-06,
      "loss": 11.423,
      "step": 19687
    },
    {
      "epoch": 3.442710382513661,
      "grad_norm": 0.7019004242263547,
      "learning_rate": 9.993715926586844e-06,
      "loss": 11.4259,
      "step": 19688
    },
    {
      "epoch": 3.4428852459016395,
      "grad_norm": 0.6960081860002605,
      "learning_rate": 9.98754607948149e-06,
      "loss": 11.5619,
      "step": 19689
    },
    {
      "epoch": 3.4430601092896174,
      "grad_norm": 0.6501395113574187,
      "learning_rate": 9.98137803740221e-06,
      "loss": 11.4219,
      "step": 19690
    },
    {
      "epoch": 3.4432349726775957,
      "grad_norm": 0.652200518021958,
      "learning_rate": 9.975211800472706e-06,
      "loss": 11.1345,
      "step": 19691
    },
    {
      "epoch": 3.4434098360655736,
      "grad_norm": 0.7028617841631852,
      "learning_rate": 9.969047368816608e-06,
      "loss": 11.1177,
      "step": 19692
    },
    {
      "epoch": 3.443584699453552,
      "grad_norm": 0.6452608645071348,
      "learning_rate": 9.96288474255752e-06,
      "loss": 11.4905,
      "step": 19693
    },
    {
      "epoch": 3.44375956284153,
      "grad_norm": 0.7051224838584028,
      "learning_rate": 9.956723921819067e-06,
      "loss": 11.3966,
      "step": 19694
    },
    {
      "epoch": 3.443934426229508,
      "grad_norm": 0.7072655618764867,
      "learning_rate": 9.950564906724757e-06,
      "loss": 11.4036,
      "step": 19695
    },
    {
      "epoch": 3.4441092896174865,
      "grad_norm": 0.7036781230362811,
      "learning_rate": 9.94440769739814e-06,
      "loss": 11.258,
      "step": 19696
    },
    {
      "epoch": 3.4442841530054644,
      "grad_norm": 0.6334074764419195,
      "learning_rate": 9.938252293962603e-06,
      "loss": 11.5043,
      "step": 19697
    },
    {
      "epoch": 3.4444590163934428,
      "grad_norm": 0.6737764721005145,
      "learning_rate": 9.93209869654167e-06,
      "loss": 11.5823,
      "step": 19698
    },
    {
      "epoch": 3.4446338797814207,
      "grad_norm": 0.6628582189313917,
      "learning_rate": 9.925946905258687e-06,
      "loss": 11.5904,
      "step": 19699
    },
    {
      "epoch": 3.444808743169399,
      "grad_norm": 0.6909125133986561,
      "learning_rate": 9.919796920237057e-06,
      "loss": 11.4459,
      "step": 19700
    },
    {
      "epoch": 3.444983606557377,
      "grad_norm": 0.6588307292296744,
      "learning_rate": 9.913648741600069e-06,
      "loss": 11.5186,
      "step": 19701
    },
    {
      "epoch": 3.4451584699453552,
      "grad_norm": 0.6864860986268867,
      "learning_rate": 9.907502369471e-06,
      "loss": 11.4301,
      "step": 19702
    },
    {
      "epoch": 3.445333333333333,
      "grad_norm": 0.7178833047195109,
      "learning_rate": 9.901357803973144e-06,
      "loss": 11.3566,
      "step": 19703
    },
    {
      "epoch": 3.4455081967213115,
      "grad_norm": 0.6722411303809227,
      "learning_rate": 9.89521504522971e-06,
      "loss": 11.5159,
      "step": 19704
    },
    {
      "epoch": 3.44568306010929,
      "grad_norm": 0.7225511418151249,
      "learning_rate": 9.889074093363892e-06,
      "loss": 11.4755,
      "step": 19705
    },
    {
      "epoch": 3.4458579234972677,
      "grad_norm": 0.6344504497710458,
      "learning_rate": 9.88293494849879e-06,
      "loss": 11.7671,
      "step": 19706
    },
    {
      "epoch": 3.446032786885246,
      "grad_norm": 0.7234580341927749,
      "learning_rate": 9.876797610757527e-06,
      "loss": 11.3763,
      "step": 19707
    },
    {
      "epoch": 3.446207650273224,
      "grad_norm": 0.7104529398903718,
      "learning_rate": 9.870662080263205e-06,
      "loss": 11.526,
      "step": 19708
    },
    {
      "epoch": 3.4463825136612023,
      "grad_norm": 0.6930951787530426,
      "learning_rate": 9.864528357138847e-06,
      "loss": 11.3415,
      "step": 19709
    },
    {
      "epoch": 3.44655737704918,
      "grad_norm": 0.6805745094617837,
      "learning_rate": 9.858396441507444e-06,
      "loss": 11.2603,
      "step": 19710
    },
    {
      "epoch": 3.4467322404371585,
      "grad_norm": 0.6717224723775358,
      "learning_rate": 9.852266333491954e-06,
      "loss": 11.2758,
      "step": 19711
    },
    {
      "epoch": 3.4469071038251364,
      "grad_norm": 0.6391840999063606,
      "learning_rate": 9.846138033215302e-06,
      "loss": 11.4124,
      "step": 19712
    },
    {
      "epoch": 3.4470819672131148,
      "grad_norm": 0.7338688178822784,
      "learning_rate": 9.840011540800409e-06,
      "loss": 11.4534,
      "step": 19713
    },
    {
      "epoch": 3.447256830601093,
      "grad_norm": 0.6774341035125035,
      "learning_rate": 9.833886856370123e-06,
      "loss": 11.1942,
      "step": 19714
    },
    {
      "epoch": 3.447431693989071,
      "grad_norm": 0.6996929161975736,
      "learning_rate": 9.827763980047244e-06,
      "loss": 11.5002,
      "step": 19715
    },
    {
      "epoch": 3.4476065573770494,
      "grad_norm": 0.7072932007316258,
      "learning_rate": 9.821642911954554e-06,
      "loss": 11.2722,
      "step": 19716
    },
    {
      "epoch": 3.4477814207650272,
      "grad_norm": 0.6782910465362243,
      "learning_rate": 9.815523652214798e-06,
      "loss": 11.2917,
      "step": 19717
    },
    {
      "epoch": 3.4479562841530056,
      "grad_norm": 0.7212206282652455,
      "learning_rate": 9.809406200950721e-06,
      "loss": 11.5724,
      "step": 19718
    },
    {
      "epoch": 3.4481311475409835,
      "grad_norm": 0.6388427515810693,
      "learning_rate": 9.803290558284961e-06,
      "loss": 11.5156,
      "step": 19719
    },
    {
      "epoch": 3.448306010928962,
      "grad_norm": 0.6564882085030408,
      "learning_rate": 9.797176724340163e-06,
      "loss": 11.2762,
      "step": 19720
    },
    {
      "epoch": 3.4484808743169397,
      "grad_norm": 0.6694619061379483,
      "learning_rate": 9.791064699238905e-06,
      "loss": 11.4982,
      "step": 19721
    },
    {
      "epoch": 3.448655737704918,
      "grad_norm": 0.6625917121813398,
      "learning_rate": 9.784954483103803e-06,
      "loss": 11.3094,
      "step": 19722
    },
    {
      "epoch": 3.4488306010928964,
      "grad_norm": 0.6833597216182711,
      "learning_rate": 9.77884607605738e-06,
      "loss": 11.3561,
      "step": 19723
    },
    {
      "epoch": 3.4490054644808743,
      "grad_norm": 0.7307747600697725,
      "learning_rate": 9.772739478222092e-06,
      "loss": 11.3142,
      "step": 19724
    },
    {
      "epoch": 3.4491803278688526,
      "grad_norm": 0.7125088525356889,
      "learning_rate": 9.766634689720411e-06,
      "loss": 11.5467,
      "step": 19725
    },
    {
      "epoch": 3.4493551912568305,
      "grad_norm": 0.6792203427491904,
      "learning_rate": 9.760531710674736e-06,
      "loss": 11.3524,
      "step": 19726
    },
    {
      "epoch": 3.449530054644809,
      "grad_norm": 0.6753306055445705,
      "learning_rate": 9.754430541207505e-06,
      "loss": 11.3902,
      "step": 19727
    },
    {
      "epoch": 3.449704918032787,
      "grad_norm": 0.6684618163438608,
      "learning_rate": 9.748331181441018e-06,
      "loss": 11.5579,
      "step": 19728
    },
    {
      "epoch": 3.449879781420765,
      "grad_norm": 0.6748729169189595,
      "learning_rate": 9.742233631497599e-06,
      "loss": 11.3542,
      "step": 19729
    },
    {
      "epoch": 3.450054644808743,
      "grad_norm": 0.6467200051365503,
      "learning_rate": 9.73613789149953e-06,
      "loss": 11.4773,
      "step": 19730
    },
    {
      "epoch": 3.4502295081967214,
      "grad_norm": 0.6298201888086886,
      "learning_rate": 9.730043961569013e-06,
      "loss": 11.5327,
      "step": 19731
    },
    {
      "epoch": 3.4504043715846997,
      "grad_norm": 0.6902561539499217,
      "learning_rate": 9.723951841828328e-06,
      "loss": 11.1091,
      "step": 19732
    },
    {
      "epoch": 3.4505792349726776,
      "grad_norm": 0.7382230234740569,
      "learning_rate": 9.717861532399564e-06,
      "loss": 11.1652,
      "step": 19733
    },
    {
      "epoch": 3.4507540983606555,
      "grad_norm": 0.7338865465208093,
      "learning_rate": 9.71177303340488e-06,
      "loss": 11.3368,
      "step": 19734
    },
    {
      "epoch": 3.450928961748634,
      "grad_norm": 0.7174343462055868,
      "learning_rate": 9.70568634496637e-06,
      "loss": 11.5807,
      "step": 19735
    },
    {
      "epoch": 3.451103825136612,
      "grad_norm": 0.6626087957893152,
      "learning_rate": 9.699601467206088e-06,
      "loss": 11.501,
      "step": 19736
    },
    {
      "epoch": 3.45127868852459,
      "grad_norm": 0.675772498916895,
      "learning_rate": 9.69351840024606e-06,
      "loss": 10.9265,
      "step": 19737
    },
    {
      "epoch": 3.4514535519125684,
      "grad_norm": 0.6921281157214348,
      "learning_rate": 9.687437144208267e-06,
      "loss": 11.5491,
      "step": 19738
    },
    {
      "epoch": 3.4516284153005463,
      "grad_norm": 0.680812175906376,
      "learning_rate": 9.681357699214643e-06,
      "loss": 11.2908,
      "step": 19739
    },
    {
      "epoch": 3.4518032786885247,
      "grad_norm": 0.6887127390493099,
      "learning_rate": 9.675280065387116e-06,
      "loss": 11.2863,
      "step": 19740
    },
    {
      "epoch": 3.4519781420765026,
      "grad_norm": 0.7301690427919844,
      "learning_rate": 9.669204242847563e-06,
      "loss": 11.5883,
      "step": 19741
    },
    {
      "epoch": 3.452153005464481,
      "grad_norm": 0.7253996693556163,
      "learning_rate": 9.66313023171781e-06,
      "loss": 11.5057,
      "step": 19742
    },
    {
      "epoch": 3.452327868852459,
      "grad_norm": 0.6605803077833688,
      "learning_rate": 9.65705803211967e-06,
      "loss": 11.2593,
      "step": 19743
    },
    {
      "epoch": 3.452502732240437,
      "grad_norm": 0.6408999798441998,
      "learning_rate": 9.650987644174913e-06,
      "loss": 11.5125,
      "step": 19744
    },
    {
      "epoch": 3.4526775956284155,
      "grad_norm": 0.6385723732921125,
      "learning_rate": 9.644919068005253e-06,
      "loss": 11.4523,
      "step": 19745
    },
    {
      "epoch": 3.4528524590163934,
      "grad_norm": 0.7061043972285737,
      "learning_rate": 9.638852303732393e-06,
      "loss": 11.0753,
      "step": 19746
    },
    {
      "epoch": 3.4530273224043717,
      "grad_norm": 0.6426013572807424,
      "learning_rate": 9.63278735147799e-06,
      "loss": 11.5473,
      "step": 19747
    },
    {
      "epoch": 3.4532021857923496,
      "grad_norm": 0.7870999071971368,
      "learning_rate": 9.62672421136367e-06,
      "loss": 11.4119,
      "step": 19748
    },
    {
      "epoch": 3.453377049180328,
      "grad_norm": 0.6454624553616134,
      "learning_rate": 9.620662883511e-06,
      "loss": 11.3194,
      "step": 19749
    },
    {
      "epoch": 3.453551912568306,
      "grad_norm": 0.6827245550055205,
      "learning_rate": 9.614603368041553e-06,
      "loss": 11.2287,
      "step": 19750
    },
    {
      "epoch": 3.453726775956284,
      "grad_norm": 0.6577778268286589,
      "learning_rate": 9.608545665076819e-06,
      "loss": 11.4577,
      "step": 19751
    },
    {
      "epoch": 3.453901639344262,
      "grad_norm": 0.7045580993952655,
      "learning_rate": 9.602489774738289e-06,
      "loss": 11.5033,
      "step": 19752
    },
    {
      "epoch": 3.4540765027322404,
      "grad_norm": 0.7160070406787015,
      "learning_rate": 9.596435697147388e-06,
      "loss": 11.6816,
      "step": 19753
    },
    {
      "epoch": 3.4542513661202188,
      "grad_norm": 0.673437217732017,
      "learning_rate": 9.590383432425531e-06,
      "loss": 11.4365,
      "step": 19754
    },
    {
      "epoch": 3.4544262295081967,
      "grad_norm": 0.6595831965441247,
      "learning_rate": 9.584332980694077e-06,
      "loss": 11.4989,
      "step": 19755
    },
    {
      "epoch": 3.454601092896175,
      "grad_norm": 0.7257597030156613,
      "learning_rate": 9.578284342074361e-06,
      "loss": 11.273,
      "step": 19756
    },
    {
      "epoch": 3.454775956284153,
      "grad_norm": 0.6692156762461907,
      "learning_rate": 9.572237516687676e-06,
      "loss": 11.2595,
      "step": 19757
    },
    {
      "epoch": 3.4549508196721312,
      "grad_norm": 0.7097932759464528,
      "learning_rate": 9.56619250465528e-06,
      "loss": 11.3368,
      "step": 19758
    },
    {
      "epoch": 3.455125683060109,
      "grad_norm": 0.7546706669744768,
      "learning_rate": 9.560149306098397e-06,
      "loss": 11.2594,
      "step": 19759
    },
    {
      "epoch": 3.4553005464480875,
      "grad_norm": 0.7213376812086862,
      "learning_rate": 9.5541079211382e-06,
      "loss": 11.3565,
      "step": 19760
    },
    {
      "epoch": 3.4554754098360654,
      "grad_norm": 0.6361583205080596,
      "learning_rate": 9.548068349895844e-06,
      "loss": 11.4811,
      "step": 19761
    },
    {
      "epoch": 3.4556502732240437,
      "grad_norm": 0.7469804474075982,
      "learning_rate": 9.542030592492447e-06,
      "loss": 11.2444,
      "step": 19762
    },
    {
      "epoch": 3.455825136612022,
      "grad_norm": 0.6551670556412795,
      "learning_rate": 9.535994649049074e-06,
      "loss": 11.5345,
      "step": 19763
    },
    {
      "epoch": 3.456,
      "grad_norm": 0.6630286869273349,
      "learning_rate": 9.529960519686776e-06,
      "loss": 11.1997,
      "step": 19764
    },
    {
      "epoch": 3.4561748633879783,
      "grad_norm": 0.6474598251119904,
      "learning_rate": 9.523928204526544e-06,
      "loss": 11.5703,
      "step": 19765
    },
    {
      "epoch": 3.456349726775956,
      "grad_norm": 0.7008247923499434,
      "learning_rate": 9.517897703689349e-06,
      "loss": 11.415,
      "step": 19766
    },
    {
      "epoch": 3.4565245901639345,
      "grad_norm": 0.7555165183136157,
      "learning_rate": 9.511869017296116e-06,
      "loss": 11.1887,
      "step": 19767
    },
    {
      "epoch": 3.4566994535519124,
      "grad_norm": 0.6667170699091092,
      "learning_rate": 9.505842145467757e-06,
      "loss": 11.4176,
      "step": 19768
    },
    {
      "epoch": 3.456874316939891,
      "grad_norm": 0.7119294692278948,
      "learning_rate": 9.499817088325102e-06,
      "loss": 11.6096,
      "step": 19769
    },
    {
      "epoch": 3.4570491803278687,
      "grad_norm": 0.7270843197052976,
      "learning_rate": 9.493793845988996e-06,
      "loss": 11.6314,
      "step": 19770
    },
    {
      "epoch": 3.457224043715847,
      "grad_norm": 0.6185035231211398,
      "learning_rate": 9.48777241858022e-06,
      "loss": 11.7327,
      "step": 19771
    },
    {
      "epoch": 3.4573989071038254,
      "grad_norm": 0.6625313853117925,
      "learning_rate": 9.481752806219502e-06,
      "loss": 11.7151,
      "step": 19772
    },
    {
      "epoch": 3.4575737704918033,
      "grad_norm": 0.7408313278116323,
      "learning_rate": 9.475735009027576e-06,
      "loss": 11.5243,
      "step": 19773
    },
    {
      "epoch": 3.4577486338797816,
      "grad_norm": 0.6575197987176027,
      "learning_rate": 9.469719027125113e-06,
      "loss": 11.564,
      "step": 19774
    },
    {
      "epoch": 3.4579234972677595,
      "grad_norm": 0.6500641030604041,
      "learning_rate": 9.463704860632739e-06,
      "loss": 11.2298,
      "step": 19775
    },
    {
      "epoch": 3.458098360655738,
      "grad_norm": 0.7292048929494419,
      "learning_rate": 9.457692509671069e-06,
      "loss": 11.3908,
      "step": 19776
    },
    {
      "epoch": 3.4582732240437157,
      "grad_norm": 0.7062370008612465,
      "learning_rate": 9.45168197436066e-06,
      "loss": 11.217,
      "step": 19777
    },
    {
      "epoch": 3.458448087431694,
      "grad_norm": 0.6910213841377641,
      "learning_rate": 9.445673254822074e-06,
      "loss": 11.2489,
      "step": 19778
    },
    {
      "epoch": 3.458622950819672,
      "grad_norm": 0.7267775339401398,
      "learning_rate": 9.439666351175736e-06,
      "loss": 11.6289,
      "step": 19779
    },
    {
      "epoch": 3.4587978142076503,
      "grad_norm": 0.6821943515592785,
      "learning_rate": 9.433661263542171e-06,
      "loss": 11.4249,
      "step": 19780
    },
    {
      "epoch": 3.4589726775956287,
      "grad_norm": 0.6336641758468537,
      "learning_rate": 9.427657992041761e-06,
      "loss": 11.5867,
      "step": 19781
    },
    {
      "epoch": 3.4591475409836066,
      "grad_norm": 0.7572313954554557,
      "learning_rate": 9.421656536794909e-06,
      "loss": 11.3126,
      "step": 19782
    },
    {
      "epoch": 3.459322404371585,
      "grad_norm": 0.6885244907976872,
      "learning_rate": 9.415656897921977e-06,
      "loss": 11.2067,
      "step": 19783
    },
    {
      "epoch": 3.459497267759563,
      "grad_norm": 0.7001046442080187,
      "learning_rate": 9.40965907554321e-06,
      "loss": 11.2937,
      "step": 19784
    },
    {
      "epoch": 3.459672131147541,
      "grad_norm": 0.7291172071420629,
      "learning_rate": 9.403663069778945e-06,
      "loss": 11.1851,
      "step": 19785
    },
    {
      "epoch": 3.459846994535519,
      "grad_norm": 0.6669572382639136,
      "learning_rate": 9.397668880749411e-06,
      "loss": 11.4894,
      "step": 19786
    },
    {
      "epoch": 3.4600218579234974,
      "grad_norm": 0.6687157298344749,
      "learning_rate": 9.39167650857482e-06,
      "loss": 11.2505,
      "step": 19787
    },
    {
      "epoch": 3.4601967213114753,
      "grad_norm": 0.66551126136634,
      "learning_rate": 9.385685953375289e-06,
      "loss": 11.4074,
      "step": 19788
    },
    {
      "epoch": 3.4603715846994536,
      "grad_norm": 0.6458184753495206,
      "learning_rate": 9.379697215270967e-06,
      "loss": 11.3132,
      "step": 19789
    },
    {
      "epoch": 3.4605464480874315,
      "grad_norm": 0.8020073951744034,
      "learning_rate": 9.373710294381976e-06,
      "loss": 11.3082,
      "step": 19790
    },
    {
      "epoch": 3.46072131147541,
      "grad_norm": 0.6483963831218837,
      "learning_rate": 9.367725190828348e-06,
      "loss": 11.414,
      "step": 19791
    },
    {
      "epoch": 3.4608961748633877,
      "grad_norm": 0.674543562380337,
      "learning_rate": 9.361741904730137e-06,
      "loss": 11.6934,
      "step": 19792
    },
    {
      "epoch": 3.461071038251366,
      "grad_norm": 0.6969476387827234,
      "learning_rate": 9.35576043620725e-06,
      "loss": 11.4069,
      "step": 19793
    },
    {
      "epoch": 3.4612459016393444,
      "grad_norm": 0.8179972269683576,
      "learning_rate": 9.349780785379703e-06,
      "loss": 11.6292,
      "step": 19794
    },
    {
      "epoch": 3.4614207650273223,
      "grad_norm": 0.6951622921600692,
      "learning_rate": 9.343802952367375e-06,
      "loss": 11.434,
      "step": 19795
    },
    {
      "epoch": 3.4615956284153007,
      "grad_norm": 0.7471282191258393,
      "learning_rate": 9.337826937290173e-06,
      "loss": 11.3174,
      "step": 19796
    },
    {
      "epoch": 3.4617704918032786,
      "grad_norm": 0.6961476525285702,
      "learning_rate": 9.331852740267898e-06,
      "loss": 11.2971,
      "step": 19797
    },
    {
      "epoch": 3.461945355191257,
      "grad_norm": 0.7044629824055463,
      "learning_rate": 9.325880361420336e-06,
      "loss": 11.4994,
      "step": 19798
    },
    {
      "epoch": 3.462120218579235,
      "grad_norm": 0.728161544166233,
      "learning_rate": 9.319909800867299e-06,
      "loss": 11.3874,
      "step": 19799
    },
    {
      "epoch": 3.462295081967213,
      "grad_norm": 0.6674982545908605,
      "learning_rate": 9.313941058728504e-06,
      "loss": 11.6575,
      "step": 19800
    },
    {
      "epoch": 3.462469945355191,
      "grad_norm": 0.6240385023618911,
      "learning_rate": 9.307974135123643e-06,
      "loss": 11.2499,
      "step": 19801
    },
    {
      "epoch": 3.4626448087431694,
      "grad_norm": 0.694703963994535,
      "learning_rate": 9.302009030172343e-06,
      "loss": 11.3333,
      "step": 19802
    },
    {
      "epoch": 3.4628196721311477,
      "grad_norm": 0.7477003829464717,
      "learning_rate": 9.29604574399423e-06,
      "loss": 11.5391,
      "step": 19803
    },
    {
      "epoch": 3.4629945355191256,
      "grad_norm": 0.6914289768349168,
      "learning_rate": 9.29008427670891e-06,
      "loss": 11.3277,
      "step": 19804
    },
    {
      "epoch": 3.463169398907104,
      "grad_norm": 0.7261493737664115,
      "learning_rate": 9.284124628435941e-06,
      "loss": 11.2897,
      "step": 19805
    },
    {
      "epoch": 3.463344262295082,
      "grad_norm": 0.7047699274895247,
      "learning_rate": 9.278166799294786e-06,
      "loss": 11.3846,
      "step": 19806
    },
    {
      "epoch": 3.46351912568306,
      "grad_norm": 0.6154739255497778,
      "learning_rate": 9.272210789404933e-06,
      "loss": 11.7174,
      "step": 19807
    },
    {
      "epoch": 3.463693989071038,
      "grad_norm": 0.6380272312506066,
      "learning_rate": 9.266256598885814e-06,
      "loss": 11.5387,
      "step": 19808
    },
    {
      "epoch": 3.4638688524590164,
      "grad_norm": 0.6980116899257153,
      "learning_rate": 9.260304227856853e-06,
      "loss": 11.2505,
      "step": 19809
    },
    {
      "epoch": 3.4640437158469943,
      "grad_norm": 0.6899502437719737,
      "learning_rate": 9.254353676437422e-06,
      "loss": 11.3416,
      "step": 19810
    },
    {
      "epoch": 3.4642185792349727,
      "grad_norm": 0.6404947681921864,
      "learning_rate": 9.248404944746802e-06,
      "loss": 11.5328,
      "step": 19811
    },
    {
      "epoch": 3.464393442622951,
      "grad_norm": 0.734924347354425,
      "learning_rate": 9.242458032904311e-06,
      "loss": 11.4679,
      "step": 19812
    },
    {
      "epoch": 3.464568306010929,
      "grad_norm": 0.6632290913528089,
      "learning_rate": 9.236512941029174e-06,
      "loss": 11.4522,
      "step": 19813
    },
    {
      "epoch": 3.4647431693989073,
      "grad_norm": 0.6524137890255747,
      "learning_rate": 9.230569669240674e-06,
      "loss": 11.5371,
      "step": 19814
    },
    {
      "epoch": 3.464918032786885,
      "grad_norm": 0.6503486548480274,
      "learning_rate": 9.224628217657938e-06,
      "loss": 11.6682,
      "step": 19815
    },
    {
      "epoch": 3.4650928961748635,
      "grad_norm": 0.7000070071418203,
      "learning_rate": 9.218688586400116e-06,
      "loss": 11.2677,
      "step": 19816
    },
    {
      "epoch": 3.4652677595628414,
      "grad_norm": 0.6780319678658517,
      "learning_rate": 9.21275077558631e-06,
      "loss": 11.4943,
      "step": 19817
    },
    {
      "epoch": 3.4654426229508197,
      "grad_norm": 0.7448775563336901,
      "learning_rate": 9.206814785335616e-06,
      "loss": 11.5328,
      "step": 19818
    },
    {
      "epoch": 3.4656174863387976,
      "grad_norm": 0.7582504113164201,
      "learning_rate": 9.200880615767083e-06,
      "loss": 11.5836,
      "step": 19819
    },
    {
      "epoch": 3.465792349726776,
      "grad_norm": 0.6665657098216644,
      "learning_rate": 9.19494826699966e-06,
      "loss": 11.3999,
      "step": 19820
    },
    {
      "epoch": 3.4659672131147543,
      "grad_norm": 0.6351128659173256,
      "learning_rate": 9.189017739152328e-06,
      "loss": 11.3293,
      "step": 19821
    },
    {
      "epoch": 3.466142076502732,
      "grad_norm": 0.6695778807984335,
      "learning_rate": 9.183089032344006e-06,
      "loss": 11.2557,
      "step": 19822
    },
    {
      "epoch": 3.4663169398907105,
      "grad_norm": 0.7168168074212401,
      "learning_rate": 9.17716214669363e-06,
      "loss": 11.3417,
      "step": 19823
    },
    {
      "epoch": 3.4664918032786884,
      "grad_norm": 0.6442939189767309,
      "learning_rate": 9.171237082319994e-06,
      "loss": 11.4417,
      "step": 19824
    },
    {
      "epoch": 3.466666666666667,
      "grad_norm": 0.666027512432187,
      "learning_rate": 9.165313839341938e-06,
      "loss": 11.2483,
      "step": 19825
    },
    {
      "epoch": 3.4668415300546447,
      "grad_norm": 0.6290853971645681,
      "learning_rate": 9.159392417878243e-06,
      "loss": 11.7485,
      "step": 19826
    },
    {
      "epoch": 3.467016393442623,
      "grad_norm": 0.6757431036783857,
      "learning_rate": 9.153472818047625e-06,
      "loss": 11.3476,
      "step": 19827
    },
    {
      "epoch": 3.467191256830601,
      "grad_norm": 0.6773615933362018,
      "learning_rate": 9.147555039968848e-06,
      "loss": 11.3438,
      "step": 19828
    },
    {
      "epoch": 3.4673661202185793,
      "grad_norm": 0.6721119432000564,
      "learning_rate": 9.141639083760522e-06,
      "loss": 11.4568,
      "step": 19829
    },
    {
      "epoch": 3.4675409836065576,
      "grad_norm": 0.6842040008422555,
      "learning_rate": 9.135724949541314e-06,
      "loss": 11.339,
      "step": 19830
    },
    {
      "epoch": 3.4677158469945355,
      "grad_norm": 0.7420686830608597,
      "learning_rate": 9.129812637429803e-06,
      "loss": 11.4025,
      "step": 19831
    },
    {
      "epoch": 3.467890710382514,
      "grad_norm": 0.6459581666236072,
      "learning_rate": 9.12390214754456e-06,
      "loss": 11.6216,
      "step": 19832
    },
    {
      "epoch": 3.4680655737704917,
      "grad_norm": 0.5978674649153921,
      "learning_rate": 9.117993480004105e-06,
      "loss": 11.3196,
      "step": 19833
    },
    {
      "epoch": 3.46824043715847,
      "grad_norm": 0.6664154703028428,
      "learning_rate": 9.112086634926931e-06,
      "loss": 11.517,
      "step": 19834
    },
    {
      "epoch": 3.468415300546448,
      "grad_norm": 0.7058902662274564,
      "learning_rate": 9.106181612431474e-06,
      "loss": 11.495,
      "step": 19835
    },
    {
      "epoch": 3.4685901639344263,
      "grad_norm": 0.7046709897140903,
      "learning_rate": 9.100278412636166e-06,
      "loss": 11.3939,
      "step": 19836
    },
    {
      "epoch": 3.468765027322404,
      "grad_norm": 0.7203866540647013,
      "learning_rate": 9.094377035659363e-06,
      "loss": 11.3242,
      "step": 19837
    },
    {
      "epoch": 3.4689398907103826,
      "grad_norm": 0.6368861723665784,
      "learning_rate": 9.088477481619429e-06,
      "loss": 11.4377,
      "step": 19838
    },
    {
      "epoch": 3.469114754098361,
      "grad_norm": 0.6553228180469842,
      "learning_rate": 9.082579750634646e-06,
      "loss": 11.4504,
      "step": 19839
    },
    {
      "epoch": 3.469289617486339,
      "grad_norm": 0.7373917420428591,
      "learning_rate": 9.076683842823298e-06,
      "loss": 11.392,
      "step": 19840
    },
    {
      "epoch": 3.4694644808743167,
      "grad_norm": 0.7048745660494522,
      "learning_rate": 9.070789758303611e-06,
      "loss": 11.3315,
      "step": 19841
    },
    {
      "epoch": 3.469639344262295,
      "grad_norm": 0.6631173216555549,
      "learning_rate": 9.06489749719377e-06,
      "loss": 11.249,
      "step": 19842
    },
    {
      "epoch": 3.4698142076502734,
      "grad_norm": 0.6279157524150926,
      "learning_rate": 9.059007059611946e-06,
      "loss": 11.5394,
      "step": 19843
    },
    {
      "epoch": 3.4699890710382513,
      "grad_norm": 0.6100633674890759,
      "learning_rate": 9.053118445676267e-06,
      "loss": 11.463,
      "step": 19844
    },
    {
      "epoch": 3.4701639344262296,
      "grad_norm": 0.6503253341963969,
      "learning_rate": 9.047231655504795e-06,
      "loss": 11.6262,
      "step": 19845
    },
    {
      "epoch": 3.4703387978142075,
      "grad_norm": 0.6464774214039871,
      "learning_rate": 9.041346689215591e-06,
      "loss": 11.5778,
      "step": 19846
    },
    {
      "epoch": 3.470513661202186,
      "grad_norm": 0.6709904391536633,
      "learning_rate": 9.03546354692667e-06,
      "loss": 11.6463,
      "step": 19847
    },
    {
      "epoch": 3.4706885245901637,
      "grad_norm": 0.6860093807628237,
      "learning_rate": 9.029582228755996e-06,
      "loss": 11.5089,
      "step": 19848
    },
    {
      "epoch": 3.470863387978142,
      "grad_norm": 0.7441702316279782,
      "learning_rate": 9.023702734821516e-06,
      "loss": 11.1391,
      "step": 19849
    },
    {
      "epoch": 3.47103825136612,
      "grad_norm": 0.6541155553402144,
      "learning_rate": 9.017825065241136e-06,
      "loss": 11.3178,
      "step": 19850
    },
    {
      "epoch": 3.4712131147540983,
      "grad_norm": 0.6609294749701519,
      "learning_rate": 9.01194922013271e-06,
      "loss": 11.3676,
      "step": 19851
    },
    {
      "epoch": 3.4713879781420767,
      "grad_norm": 0.6986106094291971,
      "learning_rate": 9.006075199614073e-06,
      "loss": 11.4148,
      "step": 19852
    },
    {
      "epoch": 3.4715628415300546,
      "grad_norm": 0.654336284387832,
      "learning_rate": 9.000203003803021e-06,
      "loss": 11.5173,
      "step": 19853
    },
    {
      "epoch": 3.471737704918033,
      "grad_norm": 0.8311977957189121,
      "learning_rate": 8.994332632817304e-06,
      "loss": 11.4373,
      "step": 19854
    },
    {
      "epoch": 3.471912568306011,
      "grad_norm": 0.7012945038436513,
      "learning_rate": 8.988464086774628e-06,
      "loss": 11.3743,
      "step": 19855
    },
    {
      "epoch": 3.472087431693989,
      "grad_norm": 0.7034679519474475,
      "learning_rate": 8.982597365792711e-06,
      "loss": 11.411,
      "step": 19856
    },
    {
      "epoch": 3.472262295081967,
      "grad_norm": 0.6734023739281377,
      "learning_rate": 8.976732469989157e-06,
      "loss": 11.4235,
      "step": 19857
    },
    {
      "epoch": 3.4724371584699454,
      "grad_norm": 0.6448219725533326,
      "learning_rate": 8.970869399481608e-06,
      "loss": 11.5491,
      "step": 19858
    },
    {
      "epoch": 3.4726120218579233,
      "grad_norm": 0.6418958139835965,
      "learning_rate": 8.965008154387621e-06,
      "loss": 11.5315,
      "step": 19859
    },
    {
      "epoch": 3.4727868852459016,
      "grad_norm": 0.6306534960044415,
      "learning_rate": 8.95914873482473e-06,
      "loss": 11.3387,
      "step": 19860
    },
    {
      "epoch": 3.47296174863388,
      "grad_norm": 0.6724006031940306,
      "learning_rate": 8.953291140910448e-06,
      "loss": 11.3583,
      "step": 19861
    },
    {
      "epoch": 3.473136612021858,
      "grad_norm": 0.6763212898858975,
      "learning_rate": 8.947435372762215e-06,
      "loss": 11.3597,
      "step": 19862
    },
    {
      "epoch": 3.473311475409836,
      "grad_norm": 0.6087263996522122,
      "learning_rate": 8.941581430497481e-06,
      "loss": 11.4479,
      "step": 19863
    },
    {
      "epoch": 3.473486338797814,
      "grad_norm": 0.7345862765569735,
      "learning_rate": 8.93572931423362e-06,
      "loss": 11.3563,
      "step": 19864
    },
    {
      "epoch": 3.4736612021857924,
      "grad_norm": 0.6557450869140541,
      "learning_rate": 8.929879024087994e-06,
      "loss": 11.4492,
      "step": 19865
    },
    {
      "epoch": 3.4738360655737703,
      "grad_norm": 0.6510402568501277,
      "learning_rate": 8.924030560177921e-06,
      "loss": 11.3013,
      "step": 19866
    },
    {
      "epoch": 3.4740109289617487,
      "grad_norm": 0.7291365928285698,
      "learning_rate": 8.91818392262067e-06,
      "loss": 11.391,
      "step": 19867
    },
    {
      "epoch": 3.4741857923497266,
      "grad_norm": 0.6702096274795527,
      "learning_rate": 8.912339111533485e-06,
      "loss": 11.4447,
      "step": 19868
    },
    {
      "epoch": 3.474360655737705,
      "grad_norm": 0.7187150131071662,
      "learning_rate": 8.90649612703358e-06,
      "loss": 11.4683,
      "step": 19869
    },
    {
      "epoch": 3.4745355191256833,
      "grad_norm": 0.6885067814715911,
      "learning_rate": 8.900654969238121e-06,
      "loss": 11.3429,
      "step": 19870
    },
    {
      "epoch": 3.474710382513661,
      "grad_norm": 0.6613061715381547,
      "learning_rate": 8.894815638264243e-06,
      "loss": 11.564,
      "step": 19871
    },
    {
      "epoch": 3.4748852459016395,
      "grad_norm": 0.6865064749542474,
      "learning_rate": 8.888978134229042e-06,
      "loss": 11.2295,
      "step": 19872
    },
    {
      "epoch": 3.4750601092896174,
      "grad_norm": 0.7133797757133942,
      "learning_rate": 8.883142457249572e-06,
      "loss": 11.3157,
      "step": 19873
    },
    {
      "epoch": 3.4752349726775957,
      "grad_norm": 0.6777907189017169,
      "learning_rate": 8.87730860744289e-06,
      "loss": 11.4364,
      "step": 19874
    },
    {
      "epoch": 3.4754098360655736,
      "grad_norm": 0.675381616591403,
      "learning_rate": 8.871476584925909e-06,
      "loss": 11.2363,
      "step": 19875
    },
    {
      "epoch": 3.475584699453552,
      "grad_norm": 0.6781059320735329,
      "learning_rate": 8.865646389815641e-06,
      "loss": 11.4425,
      "step": 19876
    },
    {
      "epoch": 3.47575956284153,
      "grad_norm": 0.6638518801520533,
      "learning_rate": 8.859818022228982e-06,
      "loss": 11.1423,
      "step": 19877
    },
    {
      "epoch": 3.475934426229508,
      "grad_norm": 0.7074326545355818,
      "learning_rate": 8.853991482282809e-06,
      "loss": 11.2238,
      "step": 19878
    },
    {
      "epoch": 3.4761092896174866,
      "grad_norm": 0.6552031944194799,
      "learning_rate": 8.84816677009399e-06,
      "loss": 11.4818,
      "step": 19879
    },
    {
      "epoch": 3.4762841530054645,
      "grad_norm": 0.6463829439573823,
      "learning_rate": 8.842343885779247e-06,
      "loss": 11.4617,
      "step": 19880
    },
    {
      "epoch": 3.476459016393443,
      "grad_norm": 0.6623599974884179,
      "learning_rate": 8.836522829455429e-06,
      "loss": 11.5627,
      "step": 19881
    },
    {
      "epoch": 3.4766338797814207,
      "grad_norm": 0.6762356706662503,
      "learning_rate": 8.830703601239243e-06,
      "loss": 11.4422,
      "step": 19882
    },
    {
      "epoch": 3.476808743169399,
      "grad_norm": 0.7190085562626567,
      "learning_rate": 8.824886201247384e-06,
      "loss": 11.4732,
      "step": 19883
    },
    {
      "epoch": 3.476983606557377,
      "grad_norm": 0.6893026129427178,
      "learning_rate": 8.819070629596482e-06,
      "loss": 11.7151,
      "step": 19884
    },
    {
      "epoch": 3.4771584699453553,
      "grad_norm": 0.6368287458052004,
      "learning_rate": 8.813256886403164e-06,
      "loss": 11.5535,
      "step": 19885
    },
    {
      "epoch": 3.477333333333333,
      "grad_norm": 0.6765111318141344,
      "learning_rate": 8.80744497178404e-06,
      "loss": 11.2016,
      "step": 19886
    },
    {
      "epoch": 3.4775081967213115,
      "grad_norm": 0.6905291968832027,
      "learning_rate": 8.801634885855637e-06,
      "loss": 11.6108,
      "step": 19887
    },
    {
      "epoch": 3.47768306010929,
      "grad_norm": 0.6809517523348989,
      "learning_rate": 8.795826628734494e-06,
      "loss": 11.4812,
      "step": 19888
    },
    {
      "epoch": 3.4778579234972677,
      "grad_norm": 0.66987948345716,
      "learning_rate": 8.790020200537031e-06,
      "loss": 11.322,
      "step": 19889
    },
    {
      "epoch": 3.478032786885246,
      "grad_norm": 0.7439457423195567,
      "learning_rate": 8.784215601379719e-06,
      "loss": 11.4867,
      "step": 19890
    },
    {
      "epoch": 3.478207650273224,
      "grad_norm": 0.6436146795441133,
      "learning_rate": 8.778412831378969e-06,
      "loss": 11.5931,
      "step": 19891
    },
    {
      "epoch": 3.4783825136612023,
      "grad_norm": 0.6555308423004599,
      "learning_rate": 8.772611890651139e-06,
      "loss": 11.4474,
      "step": 19892
    },
    {
      "epoch": 3.47855737704918,
      "grad_norm": 0.7024740386364979,
      "learning_rate": 8.766812779312528e-06,
      "loss": 11.4125,
      "step": 19893
    },
    {
      "epoch": 3.4787322404371586,
      "grad_norm": 0.6644283181545886,
      "learning_rate": 8.761015497479418e-06,
      "loss": 11.4134,
      "step": 19894
    },
    {
      "epoch": 3.4789071038251365,
      "grad_norm": 0.6860161249875201,
      "learning_rate": 8.75522004526812e-06,
      "loss": 11.2811,
      "step": 19895
    },
    {
      "epoch": 3.479081967213115,
      "grad_norm": 0.7086026371528515,
      "learning_rate": 8.749426422794804e-06,
      "loss": 11.4241,
      "step": 19896
    },
    {
      "epoch": 3.479256830601093,
      "grad_norm": 0.6824714532827882,
      "learning_rate": 8.74363463017569e-06,
      "loss": 11.5885,
      "step": 19897
    },
    {
      "epoch": 3.479431693989071,
      "grad_norm": 0.7016594374115809,
      "learning_rate": 8.737844667526874e-06,
      "loss": 11.6705,
      "step": 19898
    },
    {
      "epoch": 3.479606557377049,
      "grad_norm": 0.6387487407578589,
      "learning_rate": 8.732056534964473e-06,
      "loss": 11.5173,
      "step": 19899
    },
    {
      "epoch": 3.4797814207650273,
      "grad_norm": 0.7626431136886299,
      "learning_rate": 8.726270232604583e-06,
      "loss": 11.6789,
      "step": 19900
    },
    {
      "epoch": 3.4799562841530056,
      "grad_norm": 0.7069455912321156,
      "learning_rate": 8.720485760563234e-06,
      "loss": 11.1362,
      "step": 19901
    },
    {
      "epoch": 3.4801311475409835,
      "grad_norm": 0.6216970002453948,
      "learning_rate": 8.714703118956402e-06,
      "loss": 11.7002,
      "step": 19902
    },
    {
      "epoch": 3.480306010928962,
      "grad_norm": 0.6292625844633217,
      "learning_rate": 8.708922307900047e-06,
      "loss": 11.4731,
      "step": 19903
    },
    {
      "epoch": 3.4804808743169398,
      "grad_norm": 0.7025843228549298,
      "learning_rate": 8.703143327510089e-06,
      "loss": 11.4107,
      "step": 19904
    },
    {
      "epoch": 3.480655737704918,
      "grad_norm": 0.6426196135707164,
      "learning_rate": 8.697366177902433e-06,
      "loss": 11.4219,
      "step": 19905
    },
    {
      "epoch": 3.480830601092896,
      "grad_norm": 0.6929935818437202,
      "learning_rate": 8.691590859192945e-06,
      "loss": 11.2765,
      "step": 19906
    },
    {
      "epoch": 3.4810054644808743,
      "grad_norm": 0.6901669738087431,
      "learning_rate": 8.685817371497396e-06,
      "loss": 11.392,
      "step": 19907
    },
    {
      "epoch": 3.4811803278688522,
      "grad_norm": 0.6781352588315729,
      "learning_rate": 8.680045714931584e-06,
      "loss": 11.2471,
      "step": 19908
    },
    {
      "epoch": 3.4813551912568306,
      "grad_norm": 0.6967001326974772,
      "learning_rate": 8.674275889611216e-06,
      "loss": 11.3313,
      "step": 19909
    },
    {
      "epoch": 3.481530054644809,
      "grad_norm": 0.6634024732076983,
      "learning_rate": 8.668507895652067e-06,
      "loss": 11.4733,
      "step": 19910
    },
    {
      "epoch": 3.481704918032787,
      "grad_norm": 0.6606847333410969,
      "learning_rate": 8.662741733169743e-06,
      "loss": 11.3098,
      "step": 19911
    },
    {
      "epoch": 3.481879781420765,
      "grad_norm": 0.6686565880230592,
      "learning_rate": 8.656977402279886e-06,
      "loss": 11.4159,
      "step": 19912
    },
    {
      "epoch": 3.482054644808743,
      "grad_norm": 0.6613261846666553,
      "learning_rate": 8.651214903098081e-06,
      "loss": 11.4289,
      "step": 19913
    },
    {
      "epoch": 3.4822295081967214,
      "grad_norm": 0.7407244673252835,
      "learning_rate": 8.645454235739903e-06,
      "loss": 11.516,
      "step": 19914
    },
    {
      "epoch": 3.4824043715846993,
      "grad_norm": 0.6791794472815137,
      "learning_rate": 8.639695400320879e-06,
      "loss": 11.4604,
      "step": 19915
    },
    {
      "epoch": 3.4825792349726776,
      "grad_norm": 0.6445561021033254,
      "learning_rate": 8.633938396956464e-06,
      "loss": 11.3439,
      "step": 19916
    },
    {
      "epoch": 3.4827540983606555,
      "grad_norm": 0.63134729949677,
      "learning_rate": 8.62818322576212e-06,
      "loss": 11.5654,
      "step": 19917
    },
    {
      "epoch": 3.482928961748634,
      "grad_norm": 0.6764600995349316,
      "learning_rate": 8.622429886853223e-06,
      "loss": 11.0497,
      "step": 19918
    },
    {
      "epoch": 3.483103825136612,
      "grad_norm": 0.7555383295323385,
      "learning_rate": 8.616678380345212e-06,
      "loss": 11.4512,
      "step": 19919
    },
    {
      "epoch": 3.48327868852459,
      "grad_norm": 0.6751964688618943,
      "learning_rate": 8.61092870635336e-06,
      "loss": 11.2213,
      "step": 19920
    },
    {
      "epoch": 3.4834535519125684,
      "grad_norm": 0.7327751025633087,
      "learning_rate": 8.605180864993e-06,
      "loss": 11.5121,
      "step": 19921
    },
    {
      "epoch": 3.4836284153005463,
      "grad_norm": 0.7283719051165108,
      "learning_rate": 8.599434856379374e-06,
      "loss": 11.4338,
      "step": 19922
    },
    {
      "epoch": 3.4838032786885247,
      "grad_norm": 0.6855520832427598,
      "learning_rate": 8.593690680627697e-06,
      "loss": 11.5043,
      "step": 19923
    },
    {
      "epoch": 3.4839781420765026,
      "grad_norm": 0.608365878221417,
      "learning_rate": 8.587948337853224e-06,
      "loss": 11.1587,
      "step": 19924
    },
    {
      "epoch": 3.484153005464481,
      "grad_norm": 0.6884916492766182,
      "learning_rate": 8.58220782817103e-06,
      "loss": 11.5563,
      "step": 19925
    },
    {
      "epoch": 3.484327868852459,
      "grad_norm": 0.764499168104412,
      "learning_rate": 8.576469151696254e-06,
      "loss": 11.488,
      "step": 19926
    },
    {
      "epoch": 3.484502732240437,
      "grad_norm": 0.6396049490498548,
      "learning_rate": 8.570732308543994e-06,
      "loss": 11.3989,
      "step": 19927
    },
    {
      "epoch": 3.4846775956284155,
      "grad_norm": 0.6823514553872329,
      "learning_rate": 8.564997298829258e-06,
      "loss": 11.4969,
      "step": 19928
    },
    {
      "epoch": 3.4848524590163934,
      "grad_norm": 0.6804216242817113,
      "learning_rate": 8.559264122667087e-06,
      "loss": 11.4278,
      "step": 19929
    },
    {
      "epoch": 3.4850273224043717,
      "grad_norm": 0.6340359492227958,
      "learning_rate": 8.553532780172413e-06,
      "loss": 11.2966,
      "step": 19930
    },
    {
      "epoch": 3.4852021857923496,
      "grad_norm": 0.7146843161417883,
      "learning_rate": 8.547803271460197e-06,
      "loss": 11.2495,
      "step": 19931
    },
    {
      "epoch": 3.485377049180328,
      "grad_norm": 0.6315013716925083,
      "learning_rate": 8.542075596645316e-06,
      "loss": 11.735,
      "step": 19932
    },
    {
      "epoch": 3.485551912568306,
      "grad_norm": 0.7155178878328879,
      "learning_rate": 8.536349755842633e-06,
      "loss": 11.3422,
      "step": 19933
    },
    {
      "epoch": 3.485726775956284,
      "grad_norm": 0.6480513989866838,
      "learning_rate": 8.530625749166965e-06,
      "loss": 11.2493,
      "step": 19934
    },
    {
      "epoch": 3.485901639344262,
      "grad_norm": 0.7056108730258467,
      "learning_rate": 8.5249035767331e-06,
      "loss": 11.5104,
      "step": 19935
    },
    {
      "epoch": 3.4860765027322405,
      "grad_norm": 0.6727907956885809,
      "learning_rate": 8.519183238655781e-06,
      "loss": 11.4439,
      "step": 19936
    },
    {
      "epoch": 3.486251366120219,
      "grad_norm": 0.7167192433341524,
      "learning_rate": 8.513464735049725e-06,
      "loss": 11.5278,
      "step": 19937
    },
    {
      "epoch": 3.4864262295081967,
      "grad_norm": 0.7060017230215284,
      "learning_rate": 8.507748066029597e-06,
      "loss": 11.4487,
      "step": 19938
    },
    {
      "epoch": 3.486601092896175,
      "grad_norm": 0.6563548653253191,
      "learning_rate": 8.502033231710038e-06,
      "loss": 11.0837,
      "step": 19939
    },
    {
      "epoch": 3.486775956284153,
      "grad_norm": 0.6456378532220753,
      "learning_rate": 8.496320232205656e-06,
      "loss": 11.4372,
      "step": 19940
    },
    {
      "epoch": 3.4869508196721313,
      "grad_norm": 0.6603797083379593,
      "learning_rate": 8.490609067631006e-06,
      "loss": 11.3945,
      "step": 19941
    },
    {
      "epoch": 3.487125683060109,
      "grad_norm": 0.6554126846280337,
      "learning_rate": 8.484899738100605e-06,
      "loss": 11.4029,
      "step": 19942
    },
    {
      "epoch": 3.4873005464480875,
      "grad_norm": 0.6478430099883231,
      "learning_rate": 8.479192243728962e-06,
      "loss": 11.1614,
      "step": 19943
    },
    {
      "epoch": 3.4874754098360654,
      "grad_norm": 0.6852852015506831,
      "learning_rate": 8.473486584630519e-06,
      "loss": 11.2691,
      "step": 19944
    },
    {
      "epoch": 3.4876502732240438,
      "grad_norm": 0.6658506416507658,
      "learning_rate": 8.467782760919696e-06,
      "loss": 11.6371,
      "step": 19945
    },
    {
      "epoch": 3.487825136612022,
      "grad_norm": 0.7104212765141177,
      "learning_rate": 8.462080772710856e-06,
      "loss": 11.5772,
      "step": 19946
    },
    {
      "epoch": 3.488,
      "grad_norm": 0.7071720624794446,
      "learning_rate": 8.456380620118365e-06,
      "loss": 11.3991,
      "step": 19947
    },
    {
      "epoch": 3.4881748633879783,
      "grad_norm": 0.7100228776173297,
      "learning_rate": 8.450682303256519e-06,
      "loss": 11.6371,
      "step": 19948
    },
    {
      "epoch": 3.4883497267759562,
      "grad_norm": 0.709320260713258,
      "learning_rate": 8.444985822239582e-06,
      "loss": 11.3115,
      "step": 19949
    },
    {
      "epoch": 3.4885245901639346,
      "grad_norm": 0.6725182575062479,
      "learning_rate": 8.439291177181796e-06,
      "loss": 11.3806,
      "step": 19950
    },
    {
      "epoch": 3.4886994535519125,
      "grad_norm": 0.6808355604740209,
      "learning_rate": 8.433598368197348e-06,
      "loss": 11.225,
      "step": 19951
    },
    {
      "epoch": 3.488874316939891,
      "grad_norm": 0.6937779854629536,
      "learning_rate": 8.427907395400393e-06,
      "loss": 11.5672,
      "step": 19952
    },
    {
      "epoch": 3.4890491803278687,
      "grad_norm": 0.6922069534918399,
      "learning_rate": 8.422218258905068e-06,
      "loss": 11.3337,
      "step": 19953
    },
    {
      "epoch": 3.489224043715847,
      "grad_norm": 0.6941578840027959,
      "learning_rate": 8.416530958825442e-06,
      "loss": 11.4403,
      "step": 19954
    },
    {
      "epoch": 3.489398907103825,
      "grad_norm": 0.7247451283461132,
      "learning_rate": 8.410845495275566e-06,
      "loss": 11.5172,
      "step": 19955
    },
    {
      "epoch": 3.4895737704918033,
      "grad_norm": 0.6825559889175544,
      "learning_rate": 8.405161868369448e-06,
      "loss": 11.2962,
      "step": 19956
    },
    {
      "epoch": 3.489748633879781,
      "grad_norm": 0.6532244475236779,
      "learning_rate": 8.399480078221079e-06,
      "loss": 11.6041,
      "step": 19957
    },
    {
      "epoch": 3.4899234972677595,
      "grad_norm": 0.6907429336365469,
      "learning_rate": 8.393800124944384e-06,
      "loss": 11.3748,
      "step": 19958
    },
    {
      "epoch": 3.490098360655738,
      "grad_norm": 0.7099514425784694,
      "learning_rate": 8.388122008653265e-06,
      "loss": 11.0629,
      "step": 19959
    },
    {
      "epoch": 3.4902732240437158,
      "grad_norm": 0.6626478406729922,
      "learning_rate": 8.382445729461575e-06,
      "loss": 11.049,
      "step": 19960
    },
    {
      "epoch": 3.490448087431694,
      "grad_norm": 0.7095311857446615,
      "learning_rate": 8.376771287483165e-06,
      "loss": 11.6109,
      "step": 19961
    },
    {
      "epoch": 3.490622950819672,
      "grad_norm": 0.7494805055647512,
      "learning_rate": 8.371098682831802e-06,
      "loss": 11.3958,
      "step": 19962
    },
    {
      "epoch": 3.4907978142076503,
      "grad_norm": 0.6853392665243346,
      "learning_rate": 8.365427915621249e-06,
      "loss": 11.4177,
      "step": 19963
    },
    {
      "epoch": 3.4909726775956282,
      "grad_norm": 0.6369100661469118,
      "learning_rate": 8.359758985965217e-06,
      "loss": 11.6827,
      "step": 19964
    },
    {
      "epoch": 3.4911475409836066,
      "grad_norm": 0.6906967755039624,
      "learning_rate": 8.354091893977401e-06,
      "loss": 11.4001,
      "step": 19965
    },
    {
      "epoch": 3.4913224043715845,
      "grad_norm": 0.719592273966948,
      "learning_rate": 8.348426639771423e-06,
      "loss": 11.6144,
      "step": 19966
    },
    {
      "epoch": 3.491497267759563,
      "grad_norm": 0.679091535620373,
      "learning_rate": 8.342763223460903e-06,
      "loss": 11.6187,
      "step": 19967
    },
    {
      "epoch": 3.491672131147541,
      "grad_norm": 0.7058132739792964,
      "learning_rate": 8.337101645159407e-06,
      "loss": 11.3379,
      "step": 19968
    },
    {
      "epoch": 3.491846994535519,
      "grad_norm": 0.7590258043357403,
      "learning_rate": 8.331441904980463e-06,
      "loss": 11.3033,
      "step": 19969
    },
    {
      "epoch": 3.4920218579234974,
      "grad_norm": 0.6317595034455932,
      "learning_rate": 8.325784003037584e-06,
      "loss": 11.3446,
      "step": 19970
    },
    {
      "epoch": 3.4921967213114753,
      "grad_norm": 0.6496958225175731,
      "learning_rate": 8.320127939444178e-06,
      "loss": 11.465,
      "step": 19971
    },
    {
      "epoch": 3.4923715846994536,
      "grad_norm": 0.705708384001276,
      "learning_rate": 8.314473714313719e-06,
      "loss": 11.4115,
      "step": 19972
    },
    {
      "epoch": 3.4925464480874315,
      "grad_norm": 0.6879350081129232,
      "learning_rate": 8.308821327759563e-06,
      "loss": 11.3288,
      "step": 19973
    },
    {
      "epoch": 3.49272131147541,
      "grad_norm": 0.7405972283354795,
      "learning_rate": 8.303170779895086e-06,
      "loss": 11.3869,
      "step": 19974
    },
    {
      "epoch": 3.4928961748633878,
      "grad_norm": 0.6395490553379024,
      "learning_rate": 8.297522070833586e-06,
      "loss": 11.5959,
      "step": 19975
    },
    {
      "epoch": 3.493071038251366,
      "grad_norm": 0.8100249108436806,
      "learning_rate": 8.291875200688303e-06,
      "loss": 11.2989,
      "step": 19976
    },
    {
      "epoch": 3.4932459016393445,
      "grad_norm": 0.6649467491265911,
      "learning_rate": 8.286230169572517e-06,
      "loss": 11.579,
      "step": 19977
    },
    {
      "epoch": 3.4934207650273223,
      "grad_norm": 0.6547074382984749,
      "learning_rate": 8.280586977599425e-06,
      "loss": 11.5563,
      "step": 19978
    },
    {
      "epoch": 3.4935956284153007,
      "grad_norm": 0.6422087645257031,
      "learning_rate": 8.274945624882191e-06,
      "loss": 11.4517,
      "step": 19979
    },
    {
      "epoch": 3.4937704918032786,
      "grad_norm": 0.6434890431106041,
      "learning_rate": 8.269306111533904e-06,
      "loss": 11.4663,
      "step": 19980
    },
    {
      "epoch": 3.493945355191257,
      "grad_norm": 0.7289205272595806,
      "learning_rate": 8.263668437667671e-06,
      "loss": 11.5619,
      "step": 19981
    },
    {
      "epoch": 3.494120218579235,
      "grad_norm": 0.7248648791630976,
      "learning_rate": 8.25803260339657e-06,
      "loss": 11.4618,
      "step": 19982
    },
    {
      "epoch": 3.494295081967213,
      "grad_norm": 0.6684693520755074,
      "learning_rate": 8.2523986088336e-06,
      "loss": 11.2932,
      "step": 19983
    },
    {
      "epoch": 3.494469945355191,
      "grad_norm": 0.760816658267923,
      "learning_rate": 8.246766454091758e-06,
      "loss": 11.1798,
      "step": 19984
    },
    {
      "epoch": 3.4946448087431694,
      "grad_norm": 0.6489188909599581,
      "learning_rate": 8.241136139283934e-06,
      "loss": 11.5211,
      "step": 19985
    },
    {
      "epoch": 3.4948196721311477,
      "grad_norm": 0.6394158096758226,
      "learning_rate": 8.235507664523078e-06,
      "loss": 11.1146,
      "step": 19986
    },
    {
      "epoch": 3.4949945355191256,
      "grad_norm": 0.7148968634121926,
      "learning_rate": 8.229881029922049e-06,
      "loss": 11.5084,
      "step": 19987
    },
    {
      "epoch": 3.495169398907104,
      "grad_norm": 0.7092125550835529,
      "learning_rate": 8.224256235593696e-06,
      "loss": 11.5463,
      "step": 19988
    },
    {
      "epoch": 3.495344262295082,
      "grad_norm": 0.667864124325641,
      "learning_rate": 8.218633281650779e-06,
      "loss": 11.4743,
      "step": 19989
    },
    {
      "epoch": 3.4955191256830602,
      "grad_norm": 0.6160919547087947,
      "learning_rate": 8.213012168206048e-06,
      "loss": 11.316,
      "step": 19990
    },
    {
      "epoch": 3.495693989071038,
      "grad_norm": 0.6250530480129016,
      "learning_rate": 8.207392895372267e-06,
      "loss": 11.2412,
      "step": 19991
    },
    {
      "epoch": 3.4958688524590165,
      "grad_norm": 0.6638100604288301,
      "learning_rate": 8.201775463262107e-06,
      "loss": 11.3614,
      "step": 19992
    },
    {
      "epoch": 3.4960437158469944,
      "grad_norm": 0.6981600915199778,
      "learning_rate": 8.196159871988207e-06,
      "loss": 11.4493,
      "step": 19993
    },
    {
      "epoch": 3.4962185792349727,
      "grad_norm": 0.7695420054445441,
      "learning_rate": 8.190546121663167e-06,
      "loss": 11.3381,
      "step": 19994
    },
    {
      "epoch": 3.496393442622951,
      "grad_norm": 0.6223923352908143,
      "learning_rate": 8.184934212399552e-06,
      "loss": 11.644,
      "step": 19995
    },
    {
      "epoch": 3.496568306010929,
      "grad_norm": 0.6348121663868634,
      "learning_rate": 8.17932414430994e-06,
      "loss": 11.2696,
      "step": 19996
    },
    {
      "epoch": 3.4967431693989073,
      "grad_norm": 0.7097990763632934,
      "learning_rate": 8.173715917506819e-06,
      "loss": 11.2487,
      "step": 19997
    },
    {
      "epoch": 3.496918032786885,
      "grad_norm": 0.639080414552194,
      "learning_rate": 8.168109532102608e-06,
      "loss": 11.6555,
      "step": 19998
    },
    {
      "epoch": 3.4970928961748635,
      "grad_norm": 0.6443629458847763,
      "learning_rate": 8.162504988209773e-06,
      "loss": 11.6054,
      "step": 19999
    },
    {
      "epoch": 3.4972677595628414,
      "grad_norm": 0.7151301597258146,
      "learning_rate": 8.156902285940671e-06,
      "loss": 11.2756,
      "step": 20000
    },
    {
      "epoch": 3.4974426229508198,
      "grad_norm": 0.7027922045940591,
      "learning_rate": 8.151301425407699e-06,
      "loss": 11.3681,
      "step": 20001
    },
    {
      "epoch": 3.4976174863387977,
      "grad_norm": 0.6691036067466091,
      "learning_rate": 8.145702406723155e-06,
      "loss": 11.3692,
      "step": 20002
    },
    {
      "epoch": 3.497792349726776,
      "grad_norm": 0.7525520017391865,
      "learning_rate": 8.140105229999283e-06,
      "loss": 11.2944,
      "step": 20003
    },
    {
      "epoch": 3.4979672131147543,
      "grad_norm": 0.7063259433130284,
      "learning_rate": 8.134509895348363e-06,
      "loss": 11.4786,
      "step": 20004
    },
    {
      "epoch": 3.4981420765027322,
      "grad_norm": 0.7097322644954009,
      "learning_rate": 8.128916402882558e-06,
      "loss": 11.3391,
      "step": 20005
    },
    {
      "epoch": 3.4983169398907106,
      "grad_norm": 0.674802208023494,
      "learning_rate": 8.12332475271409e-06,
      "loss": 11.551,
      "step": 20006
    },
    {
      "epoch": 3.4984918032786885,
      "grad_norm": 0.7270791028119342,
      "learning_rate": 8.117734944955058e-06,
      "loss": 11.2242,
      "step": 20007
    },
    {
      "epoch": 3.498666666666667,
      "grad_norm": 0.6905467697491849,
      "learning_rate": 8.112146979717539e-06,
      "loss": 11.2457,
      "step": 20008
    },
    {
      "epoch": 3.4988415300546447,
      "grad_norm": 0.6857542894465408,
      "learning_rate": 8.106560857113598e-06,
      "loss": 11.5137,
      "step": 20009
    },
    {
      "epoch": 3.499016393442623,
      "grad_norm": 0.6736362602476279,
      "learning_rate": 8.100976577255281e-06,
      "loss": 11.2998,
      "step": 20010
    },
    {
      "epoch": 3.499191256830601,
      "grad_norm": 0.7353100956394724,
      "learning_rate": 8.095394140254564e-06,
      "loss": 11.3675,
      "step": 20011
    },
    {
      "epoch": 3.4993661202185793,
      "grad_norm": 0.6739479517157441,
      "learning_rate": 8.089813546223357e-06,
      "loss": 11.4229,
      "step": 20012
    },
    {
      "epoch": 3.499540983606557,
      "grad_norm": 0.6405419714624951,
      "learning_rate": 8.084234795273594e-06,
      "loss": 11.5446,
      "step": 20013
    },
    {
      "epoch": 3.4997158469945355,
      "grad_norm": 0.7128851416036689,
      "learning_rate": 8.078657887517128e-06,
      "loss": 11.2472,
      "step": 20014
    },
    {
      "epoch": 3.4998907103825134,
      "grad_norm": 0.6507080461608706,
      "learning_rate": 8.073082823065836e-06,
      "loss": 11.4711,
      "step": 20015
    },
    {
      "epoch": 3.5000655737704918,
      "grad_norm": 0.7066629015614273,
      "learning_rate": 8.067509602031464e-06,
      "loss": 11.3609,
      "step": 20016
    },
    {
      "epoch": 3.50024043715847,
      "grad_norm": 0.6838191699551982,
      "learning_rate": 8.06193822452581e-06,
      "loss": 11.5157,
      "step": 20017
    },
    {
      "epoch": 3.500415300546448,
      "grad_norm": 0.7980891346587842,
      "learning_rate": 8.056368690660566e-06,
      "loss": 11.1254,
      "step": 20018
    },
    {
      "epoch": 3.5005901639344263,
      "grad_norm": 0.728958647495142,
      "learning_rate": 8.050801000547426e-06,
      "loss": 11.1541,
      "step": 20019
    },
    {
      "epoch": 3.5007650273224042,
      "grad_norm": 0.6862422937481731,
      "learning_rate": 8.045235154298081e-06,
      "loss": 11.4117,
      "step": 20020
    },
    {
      "epoch": 3.5009398907103826,
      "grad_norm": 0.6754509702662612,
      "learning_rate": 8.039671152024087e-06,
      "loss": 11.5833,
      "step": 20021
    },
    {
      "epoch": 3.501114754098361,
      "grad_norm": 0.6318543166097454,
      "learning_rate": 8.034108993837042e-06,
      "loss": 11.3777,
      "step": 20022
    },
    {
      "epoch": 3.501289617486339,
      "grad_norm": 0.6752788437759782,
      "learning_rate": 8.02854867984849e-06,
      "loss": 11.4796,
      "step": 20023
    },
    {
      "epoch": 3.5014644808743167,
      "grad_norm": 0.6459880988317716,
      "learning_rate": 8.022990210169923e-06,
      "loss": 11.5575,
      "step": 20024
    },
    {
      "epoch": 3.501639344262295,
      "grad_norm": 0.6654987403840527,
      "learning_rate": 8.017433584912814e-06,
      "loss": 11.1047,
      "step": 20025
    },
    {
      "epoch": 3.5018142076502734,
      "grad_norm": 0.6515572668050743,
      "learning_rate": 8.011878804188578e-06,
      "loss": 11.1479,
      "step": 20026
    },
    {
      "epoch": 3.5019890710382513,
      "grad_norm": 0.7214069078674022,
      "learning_rate": 8.006325868108611e-06,
      "loss": 10.8172,
      "step": 20027
    },
    {
      "epoch": 3.5021639344262296,
      "grad_norm": 0.6625621135779949,
      "learning_rate": 8.00077477678427e-06,
      "loss": 11.5361,
      "step": 20028
    },
    {
      "epoch": 3.5023387978142075,
      "grad_norm": 0.6784294555965769,
      "learning_rate": 7.995225530326877e-06,
      "loss": 11.3838,
      "step": 20029
    },
    {
      "epoch": 3.502513661202186,
      "grad_norm": 0.678216875319249,
      "learning_rate": 7.9896781288477e-06,
      "loss": 11.2708,
      "step": 20030
    },
    {
      "epoch": 3.5026885245901638,
      "grad_norm": 0.6284186998269777,
      "learning_rate": 7.984132572457981e-06,
      "loss": 11.4785,
      "step": 20031
    },
    {
      "epoch": 3.502863387978142,
      "grad_norm": 0.8243482577160219,
      "learning_rate": 7.978588861268932e-06,
      "loss": 11.2139,
      "step": 20032
    },
    {
      "epoch": 3.50303825136612,
      "grad_norm": 0.683611317354256,
      "learning_rate": 7.973046995391719e-06,
      "loss": 11.2671,
      "step": 20033
    },
    {
      "epoch": 3.5032131147540984,
      "grad_norm": 0.6304856712662404,
      "learning_rate": 7.967506974937466e-06,
      "loss": 11.2772,
      "step": 20034
    },
    {
      "epoch": 3.5033879781420767,
      "grad_norm": 0.6876180322583704,
      "learning_rate": 7.961968800017283e-06,
      "loss": 11.5245,
      "step": 20035
    },
    {
      "epoch": 3.5035628415300546,
      "grad_norm": 0.6348145341493465,
      "learning_rate": 7.956432470742226e-06,
      "loss": 11.0493,
      "step": 20036
    },
    {
      "epoch": 3.503737704918033,
      "grad_norm": 0.7365140633032866,
      "learning_rate": 7.950897987223304e-06,
      "loss": 11.1881,
      "step": 20037
    },
    {
      "epoch": 3.503912568306011,
      "grad_norm": 0.6793573499316481,
      "learning_rate": 7.945365349571498e-06,
      "loss": 11.0786,
      "step": 20038
    },
    {
      "epoch": 3.504087431693989,
      "grad_norm": 0.681506179977519,
      "learning_rate": 7.939834557897762e-06,
      "loss": 11.4883,
      "step": 20039
    },
    {
      "epoch": 3.504262295081967,
      "grad_norm": 0.6475670589152159,
      "learning_rate": 7.93430561231302e-06,
      "loss": 11.2998,
      "step": 20040
    },
    {
      "epoch": 3.5044371584699454,
      "grad_norm": 0.7280715621648685,
      "learning_rate": 7.928778512928114e-06,
      "loss": 11.5078,
      "step": 20041
    },
    {
      "epoch": 3.5046120218579233,
      "grad_norm": 0.6503947158222957,
      "learning_rate": 7.923253259853902e-06,
      "loss": 11.6124,
      "step": 20042
    },
    {
      "epoch": 3.5047868852459017,
      "grad_norm": 0.7326590093127514,
      "learning_rate": 7.917729853201173e-06,
      "loss": 11.6288,
      "step": 20043
    },
    {
      "epoch": 3.50496174863388,
      "grad_norm": 0.6301476346769239,
      "learning_rate": 7.912208293080691e-06,
      "loss": 11.5898,
      "step": 20044
    },
    {
      "epoch": 3.505136612021858,
      "grad_norm": 0.6556090539718644,
      "learning_rate": 7.906688579603172e-06,
      "loss": 11.5774,
      "step": 20045
    },
    {
      "epoch": 3.5053114754098362,
      "grad_norm": 0.69743174605612,
      "learning_rate": 7.901170712879325e-06,
      "loss": 11.6144,
      "step": 20046
    },
    {
      "epoch": 3.505486338797814,
      "grad_norm": 0.7339313285257046,
      "learning_rate": 7.895654693019771e-06,
      "loss": 11.4794,
      "step": 20047
    },
    {
      "epoch": 3.5056612021857925,
      "grad_norm": 0.7082102056201268,
      "learning_rate": 7.890140520135147e-06,
      "loss": 11.6248,
      "step": 20048
    },
    {
      "epoch": 3.5058360655737704,
      "grad_norm": 0.6926155348892384,
      "learning_rate": 7.884628194336018e-06,
      "loss": 11.4388,
      "step": 20049
    },
    {
      "epoch": 3.5060109289617487,
      "grad_norm": 0.7189522279823798,
      "learning_rate": 7.87911771573292e-06,
      "loss": 11.0903,
      "step": 20050
    },
    {
      "epoch": 3.5061857923497266,
      "grad_norm": 0.7067475308612154,
      "learning_rate": 7.873609084436362e-06,
      "loss": 11.444,
      "step": 20051
    },
    {
      "epoch": 3.506360655737705,
      "grad_norm": 0.6833236579372131,
      "learning_rate": 7.868102300556835e-06,
      "loss": 11.5718,
      "step": 20052
    },
    {
      "epoch": 3.5065355191256833,
      "grad_norm": 0.6677494189460516,
      "learning_rate": 7.862597364204694e-06,
      "loss": 11.3624,
      "step": 20053
    },
    {
      "epoch": 3.506710382513661,
      "grad_norm": 0.6571989868218334,
      "learning_rate": 7.857094275490395e-06,
      "loss": 11.0624,
      "step": 20054
    },
    {
      "epoch": 3.506885245901639,
      "grad_norm": 0.6845878366612844,
      "learning_rate": 7.851593034524262e-06,
      "loss": 11.4057,
      "step": 20055
    },
    {
      "epoch": 3.5070601092896174,
      "grad_norm": 0.6589795726151374,
      "learning_rate": 7.846093641416641e-06,
      "loss": 11.5909,
      "step": 20056
    },
    {
      "epoch": 3.5072349726775958,
      "grad_norm": 0.6140157872798978,
      "learning_rate": 7.840596096277774e-06,
      "loss": 11.4369,
      "step": 20057
    },
    {
      "epoch": 3.5074098360655737,
      "grad_norm": 0.6476907069820783,
      "learning_rate": 7.835100399217931e-06,
      "loss": 11.405,
      "step": 20058
    },
    {
      "epoch": 3.507584699453552,
      "grad_norm": 0.6444916762223746,
      "learning_rate": 7.829606550347313e-06,
      "loss": 11.631,
      "step": 20059
    },
    {
      "epoch": 3.50775956284153,
      "grad_norm": 0.7193429142596856,
      "learning_rate": 7.824114549776074e-06,
      "loss": 11.4298,
      "step": 20060
    },
    {
      "epoch": 3.5079344262295082,
      "grad_norm": 0.6724229159543943,
      "learning_rate": 7.818624397614371e-06,
      "loss": 11.5383,
      "step": 20061
    },
    {
      "epoch": 3.5081092896174866,
      "grad_norm": 0.6092302566929059,
      "learning_rate": 7.813136093972263e-06,
      "loss": 11.5893,
      "step": 20062
    },
    {
      "epoch": 3.5082841530054645,
      "grad_norm": 0.7467302643979892,
      "learning_rate": 7.807649638959824e-06,
      "loss": 11.4,
      "step": 20063
    },
    {
      "epoch": 3.5084590163934424,
      "grad_norm": 0.6214228161341581,
      "learning_rate": 7.802165032687092e-06,
      "loss": 11.4668,
      "step": 20064
    },
    {
      "epoch": 3.5086338797814207,
      "grad_norm": 0.6125286044073931,
      "learning_rate": 7.796682275264034e-06,
      "loss": 11.2031,
      "step": 20065
    },
    {
      "epoch": 3.508808743169399,
      "grad_norm": 0.6330308788642528,
      "learning_rate": 7.791201366800615e-06,
      "loss": 11.7548,
      "step": 20066
    },
    {
      "epoch": 3.508983606557377,
      "grad_norm": 0.7367997421292585,
      "learning_rate": 7.785722307406684e-06,
      "loss": 11.1572,
      "step": 20067
    },
    {
      "epoch": 3.5091584699453553,
      "grad_norm": 0.6632418692058116,
      "learning_rate": 7.780245097192184e-06,
      "loss": 11.2947,
      "step": 20068
    },
    {
      "epoch": 3.509333333333333,
      "grad_norm": 0.7472286634278994,
      "learning_rate": 7.774769736266919e-06,
      "loss": 11.2922,
      "step": 20069
    },
    {
      "epoch": 3.5095081967213115,
      "grad_norm": 0.6938118512536203,
      "learning_rate": 7.769296224740697e-06,
      "loss": 11.3375,
      "step": 20070
    },
    {
      "epoch": 3.50968306010929,
      "grad_norm": 0.7756851012169593,
      "learning_rate": 7.763824562723265e-06,
      "loss": 11.3911,
      "step": 20071
    },
    {
      "epoch": 3.5098579234972678,
      "grad_norm": 0.6712622462477694,
      "learning_rate": 7.758354750324337e-06,
      "loss": 11.3226,
      "step": 20072
    },
    {
      "epoch": 3.5100327868852457,
      "grad_norm": 0.6826219214448561,
      "learning_rate": 7.752886787653624e-06,
      "loss": 11.3522,
      "step": 20073
    },
    {
      "epoch": 3.510207650273224,
      "grad_norm": 0.8177819639540818,
      "learning_rate": 7.74742067482077e-06,
      "loss": 11.2011,
      "step": 20074
    },
    {
      "epoch": 3.5103825136612024,
      "grad_norm": 0.6906907441912529,
      "learning_rate": 7.741956411935403e-06,
      "loss": 11.3593,
      "step": 20075
    },
    {
      "epoch": 3.5105573770491802,
      "grad_norm": 0.6920396210121484,
      "learning_rate": 7.736493999107064e-06,
      "loss": 11.6072,
      "step": 20076
    },
    {
      "epoch": 3.5107322404371586,
      "grad_norm": 0.715804926682898,
      "learning_rate": 7.73103343644528e-06,
      "loss": 11.4704,
      "step": 20077
    },
    {
      "epoch": 3.5109071038251365,
      "grad_norm": 0.648246303741708,
      "learning_rate": 7.725574724059603e-06,
      "loss": 11.2562,
      "step": 20078
    },
    {
      "epoch": 3.511081967213115,
      "grad_norm": 0.6615737941941533,
      "learning_rate": 7.720117862059495e-06,
      "loss": 11.4881,
      "step": 20079
    },
    {
      "epoch": 3.511256830601093,
      "grad_norm": 0.6980094652952192,
      "learning_rate": 7.714662850554333e-06,
      "loss": 11.6294,
      "step": 20080
    },
    {
      "epoch": 3.511431693989071,
      "grad_norm": 0.6944372538615752,
      "learning_rate": 7.709209689653518e-06,
      "loss": 11.5004,
      "step": 20081
    },
    {
      "epoch": 3.511606557377049,
      "grad_norm": 0.6469072567153159,
      "learning_rate": 7.703758379466441e-06,
      "loss": 11.1802,
      "step": 20082
    },
    {
      "epoch": 3.5117814207650273,
      "grad_norm": 0.6533996517200104,
      "learning_rate": 7.698308920102393e-06,
      "loss": 11.2975,
      "step": 20083
    },
    {
      "epoch": 3.5119562841530056,
      "grad_norm": 0.6634761643549836,
      "learning_rate": 7.692861311670674e-06,
      "loss": 11.4563,
      "step": 20084
    },
    {
      "epoch": 3.5121311475409835,
      "grad_norm": 0.6479871475565211,
      "learning_rate": 7.687415554280486e-06,
      "loss": 11.623,
      "step": 20085
    },
    {
      "epoch": 3.512306010928962,
      "grad_norm": 0.6787757621932798,
      "learning_rate": 7.681971648041031e-06,
      "loss": 11.2905,
      "step": 20086
    },
    {
      "epoch": 3.51248087431694,
      "grad_norm": 0.713059495003022,
      "learning_rate": 7.676529593061522e-06,
      "loss": 11.5953,
      "step": 20087
    },
    {
      "epoch": 3.512655737704918,
      "grad_norm": 0.715303701260387,
      "learning_rate": 7.671089389451058e-06,
      "loss": 11.0893,
      "step": 20088
    },
    {
      "epoch": 3.512830601092896,
      "grad_norm": 0.6806829453373032,
      "learning_rate": 7.665651037318767e-06,
      "loss": 11.571,
      "step": 20089
    },
    {
      "epoch": 3.5130054644808744,
      "grad_norm": 0.77368724509142,
      "learning_rate": 7.660214536773647e-06,
      "loss": 11.1409,
      "step": 20090
    },
    {
      "epoch": 3.5131803278688523,
      "grad_norm": 0.7113679586034964,
      "learning_rate": 7.654779887924734e-06,
      "loss": 11.5471,
      "step": 20091
    },
    {
      "epoch": 3.5133551912568306,
      "grad_norm": 0.6294099736446382,
      "learning_rate": 7.649347090881043e-06,
      "loss": 11.4805,
      "step": 20092
    },
    {
      "epoch": 3.513530054644809,
      "grad_norm": 0.6114473110895564,
      "learning_rate": 7.643916145751506e-06,
      "loss": 11.4613,
      "step": 20093
    },
    {
      "epoch": 3.513704918032787,
      "grad_norm": 0.6636213649560477,
      "learning_rate": 7.638487052645016e-06,
      "loss": 11.259,
      "step": 20094
    },
    {
      "epoch": 3.513879781420765,
      "grad_norm": 0.7630923871204446,
      "learning_rate": 7.633059811670429e-06,
      "loss": 11.2076,
      "step": 20095
    },
    {
      "epoch": 3.514054644808743,
      "grad_norm": 0.7440531490354164,
      "learning_rate": 7.627634422936592e-06,
      "loss": 11.3638,
      "step": 20096
    },
    {
      "epoch": 3.5142295081967214,
      "grad_norm": 0.7095846480858468,
      "learning_rate": 7.622210886552328e-06,
      "loss": 11.2302,
      "step": 20097
    },
    {
      "epoch": 3.5144043715846993,
      "grad_norm": 0.6944730168280437,
      "learning_rate": 7.616789202626373e-06,
      "loss": 11.42,
      "step": 20098
    },
    {
      "epoch": 3.5145792349726777,
      "grad_norm": 0.7354049165354051,
      "learning_rate": 7.611369371267441e-06,
      "loss": 11.2658,
      "step": 20099
    },
    {
      "epoch": 3.5147540983606556,
      "grad_norm": 0.6894014655133404,
      "learning_rate": 7.605951392584221e-06,
      "loss": 11.4062,
      "step": 20100
    },
    {
      "epoch": 3.514928961748634,
      "grad_norm": 0.7056175329825453,
      "learning_rate": 7.600535266685349e-06,
      "loss": 11.4449,
      "step": 20101
    },
    {
      "epoch": 3.5151038251366122,
      "grad_norm": 0.7395515463855038,
      "learning_rate": 7.595120993679483e-06,
      "loss": 11.3406,
      "step": 20102
    },
    {
      "epoch": 3.51527868852459,
      "grad_norm": 0.7121108742844762,
      "learning_rate": 7.589708573675136e-06,
      "loss": 11.2465,
      "step": 20103
    },
    {
      "epoch": 3.515453551912568,
      "grad_norm": 0.6842331590019438,
      "learning_rate": 7.584298006780876e-06,
      "loss": 11.4532,
      "step": 20104
    },
    {
      "epoch": 3.5156284153005464,
      "grad_norm": 0.6571724550856821,
      "learning_rate": 7.578889293105185e-06,
      "loss": 11.4278,
      "step": 20105
    },
    {
      "epoch": 3.5158032786885247,
      "grad_norm": 0.628805362205554,
      "learning_rate": 7.57348243275654e-06,
      "loss": 11.4866,
      "step": 20106
    },
    {
      "epoch": 3.5159781420765026,
      "grad_norm": 0.6890121908982038,
      "learning_rate": 7.568077425843378e-06,
      "loss": 11.3043,
      "step": 20107
    },
    {
      "epoch": 3.516153005464481,
      "grad_norm": 0.6645613656106202,
      "learning_rate": 7.5626742724740465e-06,
      "loss": 11.2893,
      "step": 20108
    },
    {
      "epoch": 3.516327868852459,
      "grad_norm": 0.6994524160809978,
      "learning_rate": 7.557272972756923e-06,
      "loss": 11.2891,
      "step": 20109
    },
    {
      "epoch": 3.516502732240437,
      "grad_norm": 0.6987877406298383,
      "learning_rate": 7.551873526800291e-06,
      "loss": 11.3968,
      "step": 20110
    },
    {
      "epoch": 3.5166775956284155,
      "grad_norm": 0.7082239476662868,
      "learning_rate": 7.5464759347124824e-06,
      "loss": 11.2862,
      "step": 20111
    },
    {
      "epoch": 3.5168524590163934,
      "grad_norm": 0.6984025167507101,
      "learning_rate": 7.54108019660168e-06,
      "loss": 11.3837,
      "step": 20112
    },
    {
      "epoch": 3.5170273224043713,
      "grad_norm": 0.7324819642650121,
      "learning_rate": 7.535686312576107e-06,
      "loss": 11.4555,
      "step": 20113
    },
    {
      "epoch": 3.5172021857923497,
      "grad_norm": 0.648808144543881,
      "learning_rate": 7.530294282743933e-06,
      "loss": 11.5868,
      "step": 20114
    },
    {
      "epoch": 3.517377049180328,
      "grad_norm": 0.6463741776581747,
      "learning_rate": 7.5249041072132595e-06,
      "loss": 11.5189,
      "step": 20115
    },
    {
      "epoch": 3.517551912568306,
      "grad_norm": 0.6603787660445561,
      "learning_rate": 7.519515786092224e-06,
      "loss": 11.3391,
      "step": 20116
    },
    {
      "epoch": 3.5177267759562842,
      "grad_norm": 0.6846607219235907,
      "learning_rate": 7.514129319488839e-06,
      "loss": 11.2486,
      "step": 20117
    },
    {
      "epoch": 3.517901639344262,
      "grad_norm": 0.6618122433051499,
      "learning_rate": 7.508744707511117e-06,
      "loss": 11.4507,
      "step": 20118
    },
    {
      "epoch": 3.5180765027322405,
      "grad_norm": 0.7428585078060294,
      "learning_rate": 7.503361950267063e-06,
      "loss": 11.3483,
      "step": 20119
    },
    {
      "epoch": 3.518251366120219,
      "grad_norm": 0.6906806853212103,
      "learning_rate": 7.497981047864599e-06,
      "loss": 11.2823,
      "step": 20120
    },
    {
      "epoch": 3.5184262295081967,
      "grad_norm": 0.6339055236747031,
      "learning_rate": 7.492602000411631e-06,
      "loss": 11.5031,
      "step": 20121
    },
    {
      "epoch": 3.5186010928961746,
      "grad_norm": 0.7444408177147336,
      "learning_rate": 7.4872248080160245e-06,
      "loss": 11.4047,
      "step": 20122
    },
    {
      "epoch": 3.518775956284153,
      "grad_norm": 0.7023900222046495,
      "learning_rate": 7.481849470785618e-06,
      "loss": 11.3485,
      "step": 20123
    },
    {
      "epoch": 3.5189508196721313,
      "grad_norm": 0.6991314177088297,
      "learning_rate": 7.476475988828202e-06,
      "loss": 11.4272,
      "step": 20124
    },
    {
      "epoch": 3.519125683060109,
      "grad_norm": 0.7063489454122558,
      "learning_rate": 7.471104362251525e-06,
      "loss": 11.3164,
      "step": 20125
    },
    {
      "epoch": 3.5193005464480875,
      "grad_norm": 0.7710794227933601,
      "learning_rate": 7.465734591163298e-06,
      "loss": 11.3225,
      "step": 20126
    },
    {
      "epoch": 3.5194754098360654,
      "grad_norm": 0.6801501320500666,
      "learning_rate": 7.460366675671215e-06,
      "loss": 11.565,
      "step": 20127
    },
    {
      "epoch": 3.519650273224044,
      "grad_norm": 0.6695837918390286,
      "learning_rate": 7.455000615882912e-06,
      "loss": 11.5134,
      "step": 20128
    },
    {
      "epoch": 3.519825136612022,
      "grad_norm": 0.6804525574674395,
      "learning_rate": 7.449636411906002e-06,
      "loss": 11.2689,
      "step": 20129
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7272874917698132,
      "learning_rate": 7.444274063848045e-06,
      "loss": 11.1837,
      "step": 20130
    },
    {
      "epoch": 3.520174863387978,
      "grad_norm": 0.68573546289733,
      "learning_rate": 7.438913571816586e-06,
      "loss": 11.4008,
      "step": 20131
    },
    {
      "epoch": 3.5203497267759563,
      "grad_norm": 0.6671874029108007,
      "learning_rate": 7.433554935919096e-06,
      "loss": 11.6072,
      "step": 20132
    },
    {
      "epoch": 3.5205245901639346,
      "grad_norm": 0.6870060924333197,
      "learning_rate": 7.428198156263055e-06,
      "loss": 11.5659,
      "step": 20133
    },
    {
      "epoch": 3.5206994535519125,
      "grad_norm": 0.6959809588169863,
      "learning_rate": 7.422843232955867e-06,
      "loss": 11.3957,
      "step": 20134
    },
    {
      "epoch": 3.520874316939891,
      "grad_norm": 0.7068245835803821,
      "learning_rate": 7.4174901661049235e-06,
      "loss": 11.5576,
      "step": 20135
    },
    {
      "epoch": 3.5210491803278687,
      "grad_norm": 0.6687213399927847,
      "learning_rate": 7.412138955817571e-06,
      "loss": 11.4717,
      "step": 20136
    },
    {
      "epoch": 3.521224043715847,
      "grad_norm": 0.6750317825029039,
      "learning_rate": 7.406789602201114e-06,
      "loss": 11.3813,
      "step": 20137
    },
    {
      "epoch": 3.521398907103825,
      "grad_norm": 0.7229226943374133,
      "learning_rate": 7.401442105362821e-06,
      "loss": 11.3736,
      "step": 20138
    },
    {
      "epoch": 3.5215737704918033,
      "grad_norm": 0.7290267615718861,
      "learning_rate": 7.39609646540993e-06,
      "loss": 11.4136,
      "step": 20139
    },
    {
      "epoch": 3.521748633879781,
      "grad_norm": 0.7125537954057699,
      "learning_rate": 7.390752682449642e-06,
      "loss": 11.6445,
      "step": 20140
    },
    {
      "epoch": 3.5219234972677596,
      "grad_norm": 0.6777492347432302,
      "learning_rate": 7.3854107565891064e-06,
      "loss": 11.419,
      "step": 20141
    },
    {
      "epoch": 3.522098360655738,
      "grad_norm": 0.7017796340161525,
      "learning_rate": 7.380070687935448e-06,
      "loss": 11.5591,
      "step": 20142
    },
    {
      "epoch": 3.522273224043716,
      "grad_norm": 0.6896053460063722,
      "learning_rate": 7.3747324765957474e-06,
      "loss": 11.5667,
      "step": 20143
    },
    {
      "epoch": 3.522448087431694,
      "grad_norm": 0.6815227065705803,
      "learning_rate": 7.369396122677063e-06,
      "loss": 11.4929,
      "step": 20144
    },
    {
      "epoch": 3.522622950819672,
      "grad_norm": 0.7065755746760083,
      "learning_rate": 7.3640616262864e-06,
      "loss": 11.257,
      "step": 20145
    },
    {
      "epoch": 3.5227978142076504,
      "grad_norm": 0.6671775530225341,
      "learning_rate": 7.358728987530728e-06,
      "loss": 11.4732,
      "step": 20146
    },
    {
      "epoch": 3.5229726775956283,
      "grad_norm": 0.6839127248880151,
      "learning_rate": 7.353398206516993e-06,
      "loss": 11.1073,
      "step": 20147
    },
    {
      "epoch": 3.5231475409836066,
      "grad_norm": 0.6579201279227853,
      "learning_rate": 7.348069283352099e-06,
      "loss": 11.4131,
      "step": 20148
    },
    {
      "epoch": 3.5233224043715845,
      "grad_norm": 0.7002360955746615,
      "learning_rate": 7.342742218142862e-06,
      "loss": 11.5945,
      "step": 20149
    },
    {
      "epoch": 3.523497267759563,
      "grad_norm": 0.6674605318170949,
      "learning_rate": 7.3374170109961505e-06,
      "loss": 11.4385,
      "step": 20150
    },
    {
      "epoch": 3.523672131147541,
      "grad_norm": 0.6758792513864397,
      "learning_rate": 7.332093662018746e-06,
      "loss": 11.1426,
      "step": 20151
    },
    {
      "epoch": 3.523846994535519,
      "grad_norm": 0.6788626826976455,
      "learning_rate": 7.3267721713173865e-06,
      "loss": 11.3345,
      "step": 20152
    },
    {
      "epoch": 3.5240218579234974,
      "grad_norm": 0.6869425206126266,
      "learning_rate": 7.321452538998785e-06,
      "loss": 11.4839,
      "step": 20153
    },
    {
      "epoch": 3.5241967213114753,
      "grad_norm": 0.6269995953104908,
      "learning_rate": 7.316134765169635e-06,
      "loss": 11.5966,
      "step": 20154
    },
    {
      "epoch": 3.5243715846994537,
      "grad_norm": 0.6802552746875693,
      "learning_rate": 7.310818849936551e-06,
      "loss": 11.2875,
      "step": 20155
    },
    {
      "epoch": 3.5245464480874316,
      "grad_norm": 0.7472137252071603,
      "learning_rate": 7.305504793406137e-06,
      "loss": 11.4947,
      "step": 20156
    },
    {
      "epoch": 3.52472131147541,
      "grad_norm": 0.7147092796457569,
      "learning_rate": 7.300192595684985e-06,
      "loss": 11.6417,
      "step": 20157
    },
    {
      "epoch": 3.524896174863388,
      "grad_norm": 0.717269140218281,
      "learning_rate": 7.294882256879565e-06,
      "loss": 11.429,
      "step": 20158
    },
    {
      "epoch": 3.525071038251366,
      "grad_norm": 0.6432151740475187,
      "learning_rate": 7.289573777096415e-06,
      "loss": 11.3935,
      "step": 20159
    },
    {
      "epoch": 3.5252459016393445,
      "grad_norm": 0.6917988141149872,
      "learning_rate": 7.2842671564419595e-06,
      "loss": 11.4464,
      "step": 20160
    },
    {
      "epoch": 3.5254207650273224,
      "grad_norm": 0.7204796197618117,
      "learning_rate": 7.278962395022637e-06,
      "loss": 11.5203,
      "step": 20161
    },
    {
      "epoch": 3.5255956284153003,
      "grad_norm": 0.6087185096633275,
      "learning_rate": 7.273659492944817e-06,
      "loss": 11.4956,
      "step": 20162
    },
    {
      "epoch": 3.5257704918032786,
      "grad_norm": 0.6771698772710462,
      "learning_rate": 7.268358450314794e-06,
      "loss": 11.4837,
      "step": 20163
    },
    {
      "epoch": 3.525945355191257,
      "grad_norm": 0.6531612284243976,
      "learning_rate": 7.263059267238937e-06,
      "loss": 11.4543,
      "step": 20164
    },
    {
      "epoch": 3.526120218579235,
      "grad_norm": 0.763414959254627,
      "learning_rate": 7.257761943823471e-06,
      "loss": 11.0084,
      "step": 20165
    },
    {
      "epoch": 3.526295081967213,
      "grad_norm": 0.7101590987035192,
      "learning_rate": 7.252466480174658e-06,
      "loss": 11.2587,
      "step": 20166
    },
    {
      "epoch": 3.526469945355191,
      "grad_norm": 0.6422695406555864,
      "learning_rate": 7.247172876398633e-06,
      "loss": 11.3571,
      "step": 20167
    },
    {
      "epoch": 3.5266448087431694,
      "grad_norm": 0.6291879706486571,
      "learning_rate": 7.24188113260158e-06,
      "loss": 11.6867,
      "step": 20168
    },
    {
      "epoch": 3.5268196721311478,
      "grad_norm": 0.767872785844734,
      "learning_rate": 7.236591248889623e-06,
      "loss": 11.3118,
      "step": 20169
    },
    {
      "epoch": 3.5269945355191257,
      "grad_norm": 0.6404299119417859,
      "learning_rate": 7.231303225368835e-06,
      "loss": 11.3453,
      "step": 20170
    },
    {
      "epoch": 3.5271693989071036,
      "grad_norm": 0.7030859231075729,
      "learning_rate": 7.226017062145274e-06,
      "loss": 11.4948,
      "step": 20171
    },
    {
      "epoch": 3.527344262295082,
      "grad_norm": 0.6960944806357422,
      "learning_rate": 7.220732759324911e-06,
      "loss": 11.4653,
      "step": 20172
    },
    {
      "epoch": 3.5275191256830603,
      "grad_norm": 0.6168162830391645,
      "learning_rate": 7.215450317013706e-06,
      "loss": 11.5478,
      "step": 20173
    },
    {
      "epoch": 3.527693989071038,
      "grad_norm": 0.6851448908524117,
      "learning_rate": 7.210169735317618e-06,
      "loss": 10.9394,
      "step": 20174
    },
    {
      "epoch": 3.5278688524590165,
      "grad_norm": 0.7150972064550885,
      "learning_rate": 7.204891014342552e-06,
      "loss": 11.296,
      "step": 20175
    },
    {
      "epoch": 3.5280437158469944,
      "grad_norm": 0.6562891633883589,
      "learning_rate": 7.199614154194323e-06,
      "loss": 11.3402,
      "step": 20176
    },
    {
      "epoch": 3.5282185792349727,
      "grad_norm": 0.6824562512975034,
      "learning_rate": 7.194339154978747e-06,
      "loss": 11.3051,
      "step": 20177
    },
    {
      "epoch": 3.528393442622951,
      "grad_norm": 0.7162979400072013,
      "learning_rate": 7.189066016801638e-06,
      "loss": 11.2309,
      "step": 20178
    },
    {
      "epoch": 3.528568306010929,
      "grad_norm": 0.9080445925488937,
      "learning_rate": 7.183794739768712e-06,
      "loss": 11.4196,
      "step": 20179
    },
    {
      "epoch": 3.528743169398907,
      "grad_norm": 0.6677045084117553,
      "learning_rate": 7.178525323985719e-06,
      "loss": 11.5183,
      "step": 20180
    },
    {
      "epoch": 3.528918032786885,
      "grad_norm": 0.7489696363683965,
      "learning_rate": 7.173257769558262e-06,
      "loss": 11.3445,
      "step": 20181
    },
    {
      "epoch": 3.5290928961748635,
      "grad_norm": 0.6908171041835242,
      "learning_rate": 7.167992076592001e-06,
      "loss": 11.5075,
      "step": 20182
    },
    {
      "epoch": 3.5292677595628414,
      "grad_norm": 0.6515718703595406,
      "learning_rate": 7.16272824519253e-06,
      "loss": 11.4636,
      "step": 20183
    },
    {
      "epoch": 3.52944262295082,
      "grad_norm": 0.6855798931916863,
      "learning_rate": 7.15746627546543e-06,
      "loss": 11.2063,
      "step": 20184
    },
    {
      "epoch": 3.5296174863387977,
      "grad_norm": 0.6296938153261801,
      "learning_rate": 7.152206167516184e-06,
      "loss": 11.63,
      "step": 20185
    },
    {
      "epoch": 3.529792349726776,
      "grad_norm": 0.6635658480373569,
      "learning_rate": 7.146947921450276e-06,
      "loss": 11.4919,
      "step": 20186
    },
    {
      "epoch": 3.5299672131147544,
      "grad_norm": 0.7188079487617995,
      "learning_rate": 7.141691537373152e-06,
      "loss": 11.2463,
      "step": 20187
    },
    {
      "epoch": 3.5301420765027323,
      "grad_norm": 0.6086820682282127,
      "learning_rate": 7.136437015390229e-06,
      "loss": 11.5957,
      "step": 20188
    },
    {
      "epoch": 3.53031693989071,
      "grad_norm": 0.7420189919003632,
      "learning_rate": 7.131184355606901e-06,
      "loss": 11.1116,
      "step": 20189
    },
    {
      "epoch": 3.5304918032786885,
      "grad_norm": 0.6543469517055667,
      "learning_rate": 7.125933558128451e-06,
      "loss": 11.447,
      "step": 20190
    },
    {
      "epoch": 3.530666666666667,
      "grad_norm": 0.8273247181472366,
      "learning_rate": 7.120684623060192e-06,
      "loss": 11.5006,
      "step": 20191
    },
    {
      "epoch": 3.5308415300546447,
      "grad_norm": 0.6726111230681544,
      "learning_rate": 7.115437550507365e-06,
      "loss": 11.1845,
      "step": 20192
    },
    {
      "epoch": 3.531016393442623,
      "grad_norm": 0.7027032608016808,
      "learning_rate": 7.1101923405752394e-06,
      "loss": 11.1752,
      "step": 20193
    },
    {
      "epoch": 3.531191256830601,
      "grad_norm": 0.7201589796585007,
      "learning_rate": 7.104948993368954e-06,
      "loss": 11.2046,
      "step": 20194
    },
    {
      "epoch": 3.5313661202185793,
      "grad_norm": 0.7282582573950956,
      "learning_rate": 7.099707508993669e-06,
      "loss": 11.1586,
      "step": 20195
    },
    {
      "epoch": 3.531540983606557,
      "grad_norm": 0.6838343481506021,
      "learning_rate": 7.09446788755449e-06,
      "loss": 11.3978,
      "step": 20196
    },
    {
      "epoch": 3.5317158469945356,
      "grad_norm": 0.6448372757879174,
      "learning_rate": 7.089230129156466e-06,
      "loss": 11.1968,
      "step": 20197
    },
    {
      "epoch": 3.5318907103825135,
      "grad_norm": 0.6991568530989236,
      "learning_rate": 7.083994233904679e-06,
      "loss": 11.5983,
      "step": 20198
    },
    {
      "epoch": 3.532065573770492,
      "grad_norm": 0.6802715844542685,
      "learning_rate": 7.078760201904089e-06,
      "loss": 11.5938,
      "step": 20199
    },
    {
      "epoch": 3.53224043715847,
      "grad_norm": 0.6299401007350952,
      "learning_rate": 7.073528033259669e-06,
      "loss": 11.1228,
      "step": 20200
    },
    {
      "epoch": 3.532415300546448,
      "grad_norm": 0.6553942271725296,
      "learning_rate": 7.068297728076301e-06,
      "loss": 11.321,
      "step": 20201
    },
    {
      "epoch": 3.5325901639344264,
      "grad_norm": 0.6190007895041016,
      "learning_rate": 7.063069286458945e-06,
      "loss": 11.6221,
      "step": 20202
    },
    {
      "epoch": 3.5327650273224043,
      "grad_norm": 0.7168664480289363,
      "learning_rate": 7.057842708512385e-06,
      "loss": 11.6227,
      "step": 20203
    },
    {
      "epoch": 3.5329398907103826,
      "grad_norm": 0.6520876484401793,
      "learning_rate": 7.052617994341448e-06,
      "loss": 11.4723,
      "step": 20204
    },
    {
      "epoch": 3.5331147540983605,
      "grad_norm": 0.657289782559986,
      "learning_rate": 7.047395144050906e-06,
      "loss": 11.1569,
      "step": 20205
    },
    {
      "epoch": 3.533289617486339,
      "grad_norm": 0.6733583617001216,
      "learning_rate": 7.042174157745485e-06,
      "loss": 11.2086,
      "step": 20206
    },
    {
      "epoch": 3.5334644808743167,
      "grad_norm": 0.6682336987814238,
      "learning_rate": 7.036955035529913e-06,
      "loss": 11.2828,
      "step": 20207
    },
    {
      "epoch": 3.533639344262295,
      "grad_norm": 0.6156362295330696,
      "learning_rate": 7.031737777508818e-06,
      "loss": 11.5458,
      "step": 20208
    },
    {
      "epoch": 3.5338142076502734,
      "grad_norm": 0.7455008789331922,
      "learning_rate": 7.026522383786827e-06,
      "loss": 11.2213,
      "step": 20209
    },
    {
      "epoch": 3.5339890710382513,
      "grad_norm": 0.6659540460495047,
      "learning_rate": 7.021308854468522e-06,
      "loss": 11.5155,
      "step": 20210
    },
    {
      "epoch": 3.5341639344262297,
      "grad_norm": 0.6005069606039075,
      "learning_rate": 7.016097189658466e-06,
      "loss": 11.0299,
      "step": 20211
    },
    {
      "epoch": 3.5343387978142076,
      "grad_norm": 0.644867551625218,
      "learning_rate": 7.010887389461152e-06,
      "loss": 11.5605,
      "step": 20212
    },
    {
      "epoch": 3.534513661202186,
      "grad_norm": 0.7346111098077341,
      "learning_rate": 7.005679453981062e-06,
      "loss": 11.3793,
      "step": 20213
    },
    {
      "epoch": 3.534688524590164,
      "grad_norm": 0.6860413549910852,
      "learning_rate": 7.000473383322625e-06,
      "loss": 11.6386,
      "step": 20214
    },
    {
      "epoch": 3.534863387978142,
      "grad_norm": 0.6418068782260477,
      "learning_rate": 6.9952691775902465e-06,
      "loss": 11.443,
      "step": 20215
    },
    {
      "epoch": 3.53503825136612,
      "grad_norm": 0.6191872900751468,
      "learning_rate": 6.9900668368882874e-06,
      "loss": 11.6278,
      "step": 20216
    },
    {
      "epoch": 3.5352131147540984,
      "grad_norm": 0.6460401161942868,
      "learning_rate": 6.984866361321063e-06,
      "loss": 11.2995,
      "step": 20217
    },
    {
      "epoch": 3.5353879781420767,
      "grad_norm": 0.6903222965028601,
      "learning_rate": 6.979667750992869e-06,
      "loss": 11.3413,
      "step": 20218
    },
    {
      "epoch": 3.5355628415300546,
      "grad_norm": 0.6621485093139988,
      "learning_rate": 6.974471006007932e-06,
      "loss": 11.2339,
      "step": 20219
    },
    {
      "epoch": 3.5357377049180325,
      "grad_norm": 0.6949853971755305,
      "learning_rate": 6.9692761264704916e-06,
      "loss": 11.2503,
      "step": 20220
    },
    {
      "epoch": 3.535912568306011,
      "grad_norm": 0.6507383084511441,
      "learning_rate": 6.964083112484698e-06,
      "loss": 11.3743,
      "step": 20221
    },
    {
      "epoch": 3.536087431693989,
      "grad_norm": 0.7026359792055432,
      "learning_rate": 6.95889196415469e-06,
      "loss": 11.3776,
      "step": 20222
    },
    {
      "epoch": 3.536262295081967,
      "grad_norm": 0.6572977609946913,
      "learning_rate": 6.953702681584584e-06,
      "loss": 11.1512,
      "step": 20223
    },
    {
      "epoch": 3.5364371584699454,
      "grad_norm": 0.7278457285484501,
      "learning_rate": 6.948515264878419e-06,
      "loss": 11.4124,
      "step": 20224
    },
    {
      "epoch": 3.5366120218579233,
      "grad_norm": 0.6278725659883498,
      "learning_rate": 6.943329714140234e-06,
      "loss": 11.2994,
      "step": 20225
    },
    {
      "epoch": 3.5367868852459017,
      "grad_norm": 0.67445695990395,
      "learning_rate": 6.938146029474013e-06,
      "loss": 11.5305,
      "step": 20226
    },
    {
      "epoch": 3.53696174863388,
      "grad_norm": 0.6439257456723306,
      "learning_rate": 6.932964210983695e-06,
      "loss": 11.4743,
      "step": 20227
    },
    {
      "epoch": 3.537136612021858,
      "grad_norm": 0.6941618606915472,
      "learning_rate": 6.927784258773207e-06,
      "loss": 11.4863,
      "step": 20228
    },
    {
      "epoch": 3.537311475409836,
      "grad_norm": 0.6642481649164635,
      "learning_rate": 6.922606172946411e-06,
      "loss": 11.5151,
      "step": 20229
    },
    {
      "epoch": 3.537486338797814,
      "grad_norm": 0.7264888095496739,
      "learning_rate": 6.917429953607135e-06,
      "loss": 11.3693,
      "step": 20230
    },
    {
      "epoch": 3.5376612021857925,
      "grad_norm": 0.6914286814781346,
      "learning_rate": 6.912255600859197e-06,
      "loss": 11.4114,
      "step": 20231
    },
    {
      "epoch": 3.5378360655737704,
      "grad_norm": 0.6935323058167006,
      "learning_rate": 6.907083114806357e-06,
      "loss": 11.1241,
      "step": 20232
    },
    {
      "epoch": 3.5380109289617487,
      "grad_norm": 0.6916129442463659,
      "learning_rate": 6.901912495552332e-06,
      "loss": 11.4501,
      "step": 20233
    },
    {
      "epoch": 3.5381857923497266,
      "grad_norm": 0.683258581225812,
      "learning_rate": 6.8967437432008065e-06,
      "loss": 11.4247,
      "step": 20234
    },
    {
      "epoch": 3.538360655737705,
      "grad_norm": 0.618793757404532,
      "learning_rate": 6.891576857855431e-06,
      "loss": 10.9887,
      "step": 20235
    },
    {
      "epoch": 3.5385355191256833,
      "grad_norm": 0.6301115787787662,
      "learning_rate": 6.8864118396198215e-06,
      "loss": 11.3683,
      "step": 20236
    },
    {
      "epoch": 3.538710382513661,
      "grad_norm": 0.6652263308991888,
      "learning_rate": 6.881248688597553e-06,
      "loss": 11.2437,
      "step": 20237
    },
    {
      "epoch": 3.538885245901639,
      "grad_norm": 0.7447177958013826,
      "learning_rate": 6.876087404892151e-06,
      "loss": 11.5374,
      "step": 20238
    },
    {
      "epoch": 3.5390601092896175,
      "grad_norm": 0.6907072146282551,
      "learning_rate": 6.870927988607135e-06,
      "loss": 11.4152,
      "step": 20239
    },
    {
      "epoch": 3.539234972677596,
      "grad_norm": 0.6761492057715187,
      "learning_rate": 6.865770439845953e-06,
      "loss": 11.446,
      "step": 20240
    },
    {
      "epoch": 3.5394098360655737,
      "grad_norm": 0.6473810233168316,
      "learning_rate": 6.860614758712036e-06,
      "loss": 11.4726,
      "step": 20241
    },
    {
      "epoch": 3.539584699453552,
      "grad_norm": 0.7246633271338265,
      "learning_rate": 6.855460945308767e-06,
      "loss": 11.1926,
      "step": 20242
    },
    {
      "epoch": 3.53975956284153,
      "grad_norm": 0.6598776463631498,
      "learning_rate": 6.850308999739497e-06,
      "loss": 11.4928,
      "step": 20243
    },
    {
      "epoch": 3.5399344262295083,
      "grad_norm": 0.6663619414799132,
      "learning_rate": 6.845158922107553e-06,
      "loss": 11.3824,
      "step": 20244
    },
    {
      "epoch": 3.5401092896174866,
      "grad_norm": 0.6737301772506952,
      "learning_rate": 6.8400107125161646e-06,
      "loss": 10.9754,
      "step": 20245
    },
    {
      "epoch": 3.5402841530054645,
      "grad_norm": 0.6851904391524486,
      "learning_rate": 6.8348643710686165e-06,
      "loss": 11.3291,
      "step": 20246
    },
    {
      "epoch": 3.5404590163934424,
      "grad_norm": 0.710683350095471,
      "learning_rate": 6.829719897868103e-06,
      "loss": 11.0939,
      "step": 20247
    },
    {
      "epoch": 3.5406338797814207,
      "grad_norm": 0.7558554474935447,
      "learning_rate": 6.824577293017764e-06,
      "loss": 11.5767,
      "step": 20248
    },
    {
      "epoch": 3.540808743169399,
      "grad_norm": 0.6671549909767309,
      "learning_rate": 6.81943655662074e-06,
      "loss": 11.5178,
      "step": 20249
    },
    {
      "epoch": 3.540983606557377,
      "grad_norm": 0.8853936835195537,
      "learning_rate": 6.814297688780113e-06,
      "loss": 11.258,
      "step": 20250
    },
    {
      "epoch": 3.5411584699453553,
      "grad_norm": 0.6989580045975519,
      "learning_rate": 6.809160689598926e-06,
      "loss": 11.218,
      "step": 20251
    },
    {
      "epoch": 3.541333333333333,
      "grad_norm": 0.7262285160472457,
      "learning_rate": 6.804025559180216e-06,
      "loss": 11.1772,
      "step": 20252
    },
    {
      "epoch": 3.5415081967213116,
      "grad_norm": 0.792359408214972,
      "learning_rate": 6.798892297626946e-06,
      "loss": 11.6532,
      "step": 20253
    },
    {
      "epoch": 3.5416830601092895,
      "grad_norm": 0.7120991683953108,
      "learning_rate": 6.793760905042035e-06,
      "loss": 11.5636,
      "step": 20254
    },
    {
      "epoch": 3.541857923497268,
      "grad_norm": 0.7189619372640501,
      "learning_rate": 6.7886313815283985e-06,
      "loss": 11.4317,
      "step": 20255
    },
    {
      "epoch": 3.5420327868852457,
      "grad_norm": 0.7209705815014296,
      "learning_rate": 6.783503727188911e-06,
      "loss": 11.2583,
      "step": 20256
    },
    {
      "epoch": 3.542207650273224,
      "grad_norm": 0.7345600372689529,
      "learning_rate": 6.778377942126379e-06,
      "loss": 11.2647,
      "step": 20257
    },
    {
      "epoch": 3.5423825136612024,
      "grad_norm": 0.6953787638357116,
      "learning_rate": 6.77325402644361e-06,
      "loss": 11.4819,
      "step": 20258
    },
    {
      "epoch": 3.5425573770491803,
      "grad_norm": 0.6787951555483461,
      "learning_rate": 6.7681319802433195e-06,
      "loss": 11.3722,
      "step": 20259
    },
    {
      "epoch": 3.5427322404371586,
      "grad_norm": 0.6829810102139829,
      "learning_rate": 6.7630118036282494e-06,
      "loss": 11.5105,
      "step": 20260
    },
    {
      "epoch": 3.5429071038251365,
      "grad_norm": 0.687320604807956,
      "learning_rate": 6.757893496701073e-06,
      "loss": 11.1064,
      "step": 20261
    },
    {
      "epoch": 3.543081967213115,
      "grad_norm": 0.7145461202049253,
      "learning_rate": 6.75277705956443e-06,
      "loss": 11.3374,
      "step": 20262
    },
    {
      "epoch": 3.5432568306010928,
      "grad_norm": 0.6226841373461331,
      "learning_rate": 6.747662492320905e-06,
      "loss": 11.5006,
      "step": 20263
    },
    {
      "epoch": 3.543431693989071,
      "grad_norm": 0.7118268963389616,
      "learning_rate": 6.7425497950730605e-06,
      "loss": 11.4013,
      "step": 20264
    },
    {
      "epoch": 3.543606557377049,
      "grad_norm": 0.6396746859650334,
      "learning_rate": 6.7374389679234375e-06,
      "loss": 11.2419,
      "step": 20265
    },
    {
      "epoch": 3.5437814207650273,
      "grad_norm": 0.6769565255731619,
      "learning_rate": 6.732330010974519e-06,
      "loss": 11.3845,
      "step": 20266
    },
    {
      "epoch": 3.5439562841530057,
      "grad_norm": 0.6177434693748319,
      "learning_rate": 6.7272229243287685e-06,
      "loss": 11.5954,
      "step": 20267
    },
    {
      "epoch": 3.5441311475409836,
      "grad_norm": 0.6689748339219436,
      "learning_rate": 6.72211770808856e-06,
      "loss": 11.3464,
      "step": 20268
    },
    {
      "epoch": 3.544306010928962,
      "grad_norm": 0.6830154880184259,
      "learning_rate": 6.717014362356289e-06,
      "loss": 11.5985,
      "step": 20269
    },
    {
      "epoch": 3.54448087431694,
      "grad_norm": 0.6600079461116182,
      "learning_rate": 6.711912887234306e-06,
      "loss": 11.2434,
      "step": 20270
    },
    {
      "epoch": 3.544655737704918,
      "grad_norm": 0.6462317826779674,
      "learning_rate": 6.70681328282492e-06,
      "loss": 11.3225,
      "step": 20271
    },
    {
      "epoch": 3.544830601092896,
      "grad_norm": 0.6955445332946077,
      "learning_rate": 6.701715549230358e-06,
      "loss": 11.5667,
      "step": 20272
    },
    {
      "epoch": 3.5450054644808744,
      "grad_norm": 0.7373319535255841,
      "learning_rate": 6.696619686552841e-06,
      "loss": 11.3399,
      "step": 20273
    },
    {
      "epoch": 3.5451803278688523,
      "grad_norm": 0.6787379099369841,
      "learning_rate": 6.691525694894607e-06,
      "loss": 11.5422,
      "step": 20274
    },
    {
      "epoch": 3.5453551912568306,
      "grad_norm": 0.7012974532548502,
      "learning_rate": 6.686433574357765e-06,
      "loss": 11.3858,
      "step": 20275
    },
    {
      "epoch": 3.545530054644809,
      "grad_norm": 0.6780807101642142,
      "learning_rate": 6.681343325044464e-06,
      "loss": 11.3881,
      "step": 20276
    },
    {
      "epoch": 3.545704918032787,
      "grad_norm": 0.6698104581935107,
      "learning_rate": 6.676254947056737e-06,
      "loss": 11.4528,
      "step": 20277
    },
    {
      "epoch": 3.5458797814207648,
      "grad_norm": 0.6673587526130461,
      "learning_rate": 6.671168440496623e-06,
      "loss": 11.6355,
      "step": 20278
    },
    {
      "epoch": 3.546054644808743,
      "grad_norm": 0.694858534664297,
      "learning_rate": 6.66608380546615e-06,
      "loss": 11.265,
      "step": 20279
    },
    {
      "epoch": 3.5462295081967214,
      "grad_norm": 0.6474697522980225,
      "learning_rate": 6.661001042067294e-06,
      "loss": 11.1368,
      "step": 20280
    },
    {
      "epoch": 3.5464043715846993,
      "grad_norm": 0.6562399631160665,
      "learning_rate": 6.65592015040194e-06,
      "loss": 11.5379,
      "step": 20281
    },
    {
      "epoch": 3.5465792349726777,
      "grad_norm": 0.6603423452668437,
      "learning_rate": 6.650841130571983e-06,
      "loss": 11.357,
      "step": 20282
    },
    {
      "epoch": 3.5467540983606556,
      "grad_norm": 0.6844497441376463,
      "learning_rate": 6.645763982679276e-06,
      "loss": 11.3856,
      "step": 20283
    },
    {
      "epoch": 3.546928961748634,
      "grad_norm": 0.7613895270473942,
      "learning_rate": 6.640688706825637e-06,
      "loss": 11.3991,
      "step": 20284
    },
    {
      "epoch": 3.5471038251366123,
      "grad_norm": 0.6716712118934869,
      "learning_rate": 6.6356153031128745e-06,
      "loss": 11.4334,
      "step": 20285
    },
    {
      "epoch": 3.54727868852459,
      "grad_norm": 0.6680944130588576,
      "learning_rate": 6.630543771642661e-06,
      "loss": 11.3935,
      "step": 20286
    },
    {
      "epoch": 3.547453551912568,
      "grad_norm": 0.730180488525848,
      "learning_rate": 6.625474112516728e-06,
      "loss": 11.356,
      "step": 20287
    },
    {
      "epoch": 3.5476284153005464,
      "grad_norm": 0.7336722512212784,
      "learning_rate": 6.6204063258367255e-06,
      "loss": 11.4097,
      "step": 20288
    },
    {
      "epoch": 3.5478032786885247,
      "grad_norm": 0.6557890972090425,
      "learning_rate": 6.615340411704318e-06,
      "loss": 11.3912,
      "step": 20289
    },
    {
      "epoch": 3.5479781420765026,
      "grad_norm": 0.7249705355964517,
      "learning_rate": 6.6102763702210465e-06,
      "loss": 11.3232,
      "step": 20290
    },
    {
      "epoch": 3.548153005464481,
      "grad_norm": 0.6602594387745971,
      "learning_rate": 6.605214201488486e-06,
      "loss": 11.2667,
      "step": 20291
    },
    {
      "epoch": 3.548327868852459,
      "grad_norm": 0.6751240658903019,
      "learning_rate": 6.600153905608131e-06,
      "loss": 11.3461,
      "step": 20292
    },
    {
      "epoch": 3.548502732240437,
      "grad_norm": 0.7495628209935037,
      "learning_rate": 6.595095482681457e-06,
      "loss": 11.3321,
      "step": 20293
    },
    {
      "epoch": 3.5486775956284156,
      "grad_norm": 0.7050256689210967,
      "learning_rate": 6.590038932809939e-06,
      "loss": 11.5453,
      "step": 20294
    },
    {
      "epoch": 3.5488524590163935,
      "grad_norm": 0.8089713285406913,
      "learning_rate": 6.584984256094928e-06,
      "loss": 11.3493,
      "step": 20295
    },
    {
      "epoch": 3.5490273224043714,
      "grad_norm": 0.7103977250577547,
      "learning_rate": 6.579931452637811e-06,
      "loss": 11.4061,
      "step": 20296
    },
    {
      "epoch": 3.5492021857923497,
      "grad_norm": 0.6959904095538176,
      "learning_rate": 6.574880522539883e-06,
      "loss": 11.5715,
      "step": 20297
    },
    {
      "epoch": 3.549377049180328,
      "grad_norm": 0.672902453078155,
      "learning_rate": 6.569831465902488e-06,
      "loss": 11.2962,
      "step": 20298
    },
    {
      "epoch": 3.549551912568306,
      "grad_norm": 0.7174209347002249,
      "learning_rate": 6.564784282826819e-06,
      "loss": 11.2704,
      "step": 20299
    },
    {
      "epoch": 3.5497267759562843,
      "grad_norm": 0.7055457027712776,
      "learning_rate": 6.55973897341412e-06,
      "loss": 11.1536,
      "step": 20300
    },
    {
      "epoch": 3.549901639344262,
      "grad_norm": 0.6556126445179038,
      "learning_rate": 6.554695537765543e-06,
      "loss": 11.4469,
      "step": 20301
    },
    {
      "epoch": 3.5500765027322405,
      "grad_norm": 0.6680335956862784,
      "learning_rate": 6.5496539759822175e-06,
      "loss": 11.4456,
      "step": 20302
    },
    {
      "epoch": 3.550251366120219,
      "grad_norm": 0.7358198626714627,
      "learning_rate": 6.544614288165307e-06,
      "loss": 11.4452,
      "step": 20303
    },
    {
      "epoch": 3.5504262295081968,
      "grad_norm": 0.7076789414133945,
      "learning_rate": 6.539576474415798e-06,
      "loss": 11.6328,
      "step": 20304
    },
    {
      "epoch": 3.5506010928961746,
      "grad_norm": 0.735225673227651,
      "learning_rate": 6.534540534834743e-06,
      "loss": 11.4382,
      "step": 20305
    },
    {
      "epoch": 3.550775956284153,
      "grad_norm": 0.6852897334421568,
      "learning_rate": 6.529506469523128e-06,
      "loss": 11.1306,
      "step": 20306
    },
    {
      "epoch": 3.5509508196721313,
      "grad_norm": 0.6743146592666128,
      "learning_rate": 6.524474278581905e-06,
      "loss": 11.4347,
      "step": 20307
    },
    {
      "epoch": 3.5511256830601092,
      "grad_norm": 0.7068957105621559,
      "learning_rate": 6.519443962111971e-06,
      "loss": 11.414,
      "step": 20308
    },
    {
      "epoch": 3.5513005464480876,
      "grad_norm": 0.70282269929621,
      "learning_rate": 6.514415520214212e-06,
      "loss": 11.6937,
      "step": 20309
    },
    {
      "epoch": 3.5514754098360655,
      "grad_norm": 0.6787134914978981,
      "learning_rate": 6.5093889529894705e-06,
      "loss": 11.106,
      "step": 20310
    },
    {
      "epoch": 3.551650273224044,
      "grad_norm": 0.6954698435668465,
      "learning_rate": 6.50436426053852e-06,
      "loss": 11.1712,
      "step": 20311
    },
    {
      "epoch": 3.5518251366120217,
      "grad_norm": 0.7002004555068183,
      "learning_rate": 6.499341442962137e-06,
      "loss": 11.419,
      "step": 20312
    },
    {
      "epoch": 3.552,
      "grad_norm": 0.6470526206170838,
      "learning_rate": 6.49432050036105e-06,
      "loss": 11.4076,
      "step": 20313
    },
    {
      "epoch": 3.552174863387978,
      "grad_norm": 0.6824333266360895,
      "learning_rate": 6.489301432835937e-06,
      "loss": 11.5409,
      "step": 20314
    },
    {
      "epoch": 3.5523497267759563,
      "grad_norm": 0.6936446276616985,
      "learning_rate": 6.484284240487437e-06,
      "loss": 11.1951,
      "step": 20315
    },
    {
      "epoch": 3.5525245901639346,
      "grad_norm": 0.7312490294227932,
      "learning_rate": 6.479268923416182e-06,
      "loss": 11.5005,
      "step": 20316
    },
    {
      "epoch": 3.5526994535519125,
      "grad_norm": 0.6816777241570794,
      "learning_rate": 6.474255481722724e-06,
      "loss": 11.6811,
      "step": 20317
    },
    {
      "epoch": 3.552874316939891,
      "grad_norm": 0.6533461154364696,
      "learning_rate": 6.469243915507606e-06,
      "loss": 11.4464,
      "step": 20318
    },
    {
      "epoch": 3.5530491803278688,
      "grad_norm": 0.6114102830202739,
      "learning_rate": 6.464234224871324e-06,
      "loss": 11.2769,
      "step": 20319
    },
    {
      "epoch": 3.553224043715847,
      "grad_norm": 0.6419151282376172,
      "learning_rate": 6.459226409914332e-06,
      "loss": 11.369,
      "step": 20320
    },
    {
      "epoch": 3.553398907103825,
      "grad_norm": 0.7080190073147021,
      "learning_rate": 6.454220470737071e-06,
      "loss": 11.6254,
      "step": 20321
    },
    {
      "epoch": 3.5535737704918033,
      "grad_norm": 0.6560973031663189,
      "learning_rate": 6.4492164074399065e-06,
      "loss": 11.6628,
      "step": 20322
    },
    {
      "epoch": 3.5537486338797812,
      "grad_norm": 0.6492929711938284,
      "learning_rate": 6.44421422012319e-06,
      "loss": 11.4654,
      "step": 20323
    },
    {
      "epoch": 3.5539234972677596,
      "grad_norm": 0.6522471599791609,
      "learning_rate": 6.439213908887232e-06,
      "loss": 11.4893,
      "step": 20324
    },
    {
      "epoch": 3.554098360655738,
      "grad_norm": 0.6636920342898374,
      "learning_rate": 6.4342154738323054e-06,
      "loss": 11.264,
      "step": 20325
    },
    {
      "epoch": 3.554273224043716,
      "grad_norm": 0.749964160701998,
      "learning_rate": 6.429218915058644e-06,
      "loss": 11.4542,
      "step": 20326
    },
    {
      "epoch": 3.5544480874316937,
      "grad_norm": 0.6993992904231855,
      "learning_rate": 6.424224232666454e-06,
      "loss": 11.5936,
      "step": 20327
    },
    {
      "epoch": 3.554622950819672,
      "grad_norm": 0.6841831309652177,
      "learning_rate": 6.419231426755867e-06,
      "loss": 11.4725,
      "step": 20328
    },
    {
      "epoch": 3.5547978142076504,
      "grad_norm": 0.6848356757302868,
      "learning_rate": 6.414240497427027e-06,
      "loss": 11.3915,
      "step": 20329
    },
    {
      "epoch": 3.5549726775956283,
      "grad_norm": 0.6703979056153834,
      "learning_rate": 6.409251444780018e-06,
      "loss": 11.3505,
      "step": 20330
    },
    {
      "epoch": 3.5551475409836066,
      "grad_norm": 0.700935030799026,
      "learning_rate": 6.404264268914861e-06,
      "loss": 11.2792,
      "step": 20331
    },
    {
      "epoch": 3.5553224043715845,
      "grad_norm": 0.6594155380273637,
      "learning_rate": 6.399278969931599e-06,
      "loss": 11.4547,
      "step": 20332
    },
    {
      "epoch": 3.555497267759563,
      "grad_norm": 0.6937276392398721,
      "learning_rate": 6.3942955479301844e-06,
      "loss": 11.3323,
      "step": 20333
    },
    {
      "epoch": 3.555672131147541,
      "grad_norm": 0.6354740906208409,
      "learning_rate": 6.389314003010538e-06,
      "loss": 11.6424,
      "step": 20334
    },
    {
      "epoch": 3.555846994535519,
      "grad_norm": 0.6606271182351952,
      "learning_rate": 6.384334335272579e-06,
      "loss": 11.4593,
      "step": 20335
    },
    {
      "epoch": 3.556021857923497,
      "grad_norm": 0.7025269385768448,
      "learning_rate": 6.379356544816151e-06,
      "loss": 11.6709,
      "step": 20336
    },
    {
      "epoch": 3.5561967213114754,
      "grad_norm": 0.7099366073553259,
      "learning_rate": 6.374380631741084e-06,
      "loss": 11.3148,
      "step": 20337
    },
    {
      "epoch": 3.5563715846994537,
      "grad_norm": 0.6849607881784651,
      "learning_rate": 6.3694065961471425e-06,
      "loss": 11.3555,
      "step": 20338
    },
    {
      "epoch": 3.5565464480874316,
      "grad_norm": 0.6658442975731371,
      "learning_rate": 6.364434438134092e-06,
      "loss": 11.5032,
      "step": 20339
    },
    {
      "epoch": 3.55672131147541,
      "grad_norm": 0.6681520375927397,
      "learning_rate": 6.359464157801642e-06,
      "loss": 11.0979,
      "step": 20340
    },
    {
      "epoch": 3.556896174863388,
      "grad_norm": 0.7979084812189711,
      "learning_rate": 6.354495755249423e-06,
      "loss": 11.4387,
      "step": 20341
    },
    {
      "epoch": 3.557071038251366,
      "grad_norm": 0.6761585489575289,
      "learning_rate": 6.3495292305771e-06,
      "loss": 11.1604,
      "step": 20342
    },
    {
      "epoch": 3.5572459016393445,
      "grad_norm": 0.6983721441592488,
      "learning_rate": 6.344564583884271e-06,
      "loss": 11.4246,
      "step": 20343
    },
    {
      "epoch": 3.5574207650273224,
      "grad_norm": 0.6992174705684105,
      "learning_rate": 6.339601815270468e-06,
      "loss": 11.3027,
      "step": 20344
    },
    {
      "epoch": 3.5575956284153003,
      "grad_norm": 0.6678724325844629,
      "learning_rate": 6.334640924835233e-06,
      "loss": 11.4463,
      "step": 20345
    },
    {
      "epoch": 3.5577704918032786,
      "grad_norm": 0.7002977926582297,
      "learning_rate": 6.3296819126780424e-06,
      "loss": 11.3424,
      "step": 20346
    },
    {
      "epoch": 3.557945355191257,
      "grad_norm": 0.7065158264113084,
      "learning_rate": 6.324724778898328e-06,
      "loss": 11.7362,
      "step": 20347
    },
    {
      "epoch": 3.558120218579235,
      "grad_norm": 0.6590213282418679,
      "learning_rate": 6.319769523595509e-06,
      "loss": 11.5666,
      "step": 20348
    },
    {
      "epoch": 3.5582950819672132,
      "grad_norm": 0.6468930036280289,
      "learning_rate": 6.314816146868952e-06,
      "loss": 11.6258,
      "step": 20349
    },
    {
      "epoch": 3.558469945355191,
      "grad_norm": 0.6810998547968466,
      "learning_rate": 6.3098646488179645e-06,
      "loss": 11.2392,
      "step": 20350
    },
    {
      "epoch": 3.5586448087431695,
      "grad_norm": 0.6842705787876837,
      "learning_rate": 6.304915029541858e-06,
      "loss": 11.4767,
      "step": 20351
    },
    {
      "epoch": 3.558819672131148,
      "grad_norm": 0.6975897525518338,
      "learning_rate": 6.299967289139896e-06,
      "loss": 11.4997,
      "step": 20352
    },
    {
      "epoch": 3.5589945355191257,
      "grad_norm": 0.6735749262720225,
      "learning_rate": 6.29502142771129e-06,
      "loss": 11.581,
      "step": 20353
    },
    {
      "epoch": 3.5591693989071036,
      "grad_norm": 0.6588434231418818,
      "learning_rate": 6.290077445355225e-06,
      "loss": 11.4799,
      "step": 20354
    },
    {
      "epoch": 3.559344262295082,
      "grad_norm": 0.6464037711314295,
      "learning_rate": 6.285135342170812e-06,
      "loss": 11.5615,
      "step": 20355
    },
    {
      "epoch": 3.5595191256830603,
      "grad_norm": 0.6689692655932938,
      "learning_rate": 6.280195118257182e-06,
      "loss": 11.4622,
      "step": 20356
    },
    {
      "epoch": 3.559693989071038,
      "grad_norm": 0.6931116262126702,
      "learning_rate": 6.2752567737134005e-06,
      "loss": 11.4888,
      "step": 20357
    },
    {
      "epoch": 3.5598688524590165,
      "grad_norm": 0.680626421849092,
      "learning_rate": 6.270320308638511e-06,
      "loss": 11.0792,
      "step": 20358
    },
    {
      "epoch": 3.5600437158469944,
      "grad_norm": 0.6593421020096756,
      "learning_rate": 6.265385723131478e-06,
      "loss": 11.5506,
      "step": 20359
    },
    {
      "epoch": 3.5602185792349728,
      "grad_norm": 0.7008124952882735,
      "learning_rate": 6.260453017291245e-06,
      "loss": 11.3342,
      "step": 20360
    },
    {
      "epoch": 3.5603934426229507,
      "grad_norm": 0.7636364295781389,
      "learning_rate": 6.255522191216756e-06,
      "loss": 11.3669,
      "step": 20361
    },
    {
      "epoch": 3.560568306010929,
      "grad_norm": 0.663316255793433,
      "learning_rate": 6.250593245006886e-06,
      "loss": 11.4796,
      "step": 20362
    },
    {
      "epoch": 3.560743169398907,
      "grad_norm": 0.6949387594926444,
      "learning_rate": 6.245666178760479e-06,
      "loss": 11.4234,
      "step": 20363
    },
    {
      "epoch": 3.5609180327868852,
      "grad_norm": 0.6741896396103343,
      "learning_rate": 6.240740992576322e-06,
      "loss": 11.6129,
      "step": 20364
    },
    {
      "epoch": 3.5610928961748636,
      "grad_norm": 0.6222917985154879,
      "learning_rate": 6.235817686553169e-06,
      "loss": 11.4692,
      "step": 20365
    },
    {
      "epoch": 3.5612677595628415,
      "grad_norm": 0.6386586111095399,
      "learning_rate": 6.2308962607897755e-06,
      "loss": 11.4454,
      "step": 20366
    },
    {
      "epoch": 3.56144262295082,
      "grad_norm": 0.676206973866466,
      "learning_rate": 6.225976715384841e-06,
      "loss": 11.3849,
      "step": 20367
    },
    {
      "epoch": 3.5616174863387977,
      "grad_norm": 0.6621346509631941,
      "learning_rate": 6.221059050436983e-06,
      "loss": 11.6101,
      "step": 20368
    },
    {
      "epoch": 3.561792349726776,
      "grad_norm": 0.6317082541749678,
      "learning_rate": 6.216143266044816e-06,
      "loss": 11.4759,
      "step": 20369
    },
    {
      "epoch": 3.561967213114754,
      "grad_norm": 0.7428351813245013,
      "learning_rate": 6.211229362306947e-06,
      "loss": 11.2822,
      "step": 20370
    },
    {
      "epoch": 3.5621420765027323,
      "grad_norm": 0.7129322751092751,
      "learning_rate": 6.2063173393218874e-06,
      "loss": 11.2758,
      "step": 20371
    },
    {
      "epoch": 3.56231693989071,
      "grad_norm": 0.646990301600292,
      "learning_rate": 6.2014071971881804e-06,
      "loss": 11.289,
      "step": 20372
    },
    {
      "epoch": 3.5624918032786885,
      "grad_norm": 0.647397624403573,
      "learning_rate": 6.196498936004235e-06,
      "loss": 11.335,
      "step": 20373
    },
    {
      "epoch": 3.562666666666667,
      "grad_norm": 0.7767045372606638,
      "learning_rate": 6.191592555868486e-06,
      "loss": 11.3377,
      "step": 20374
    },
    {
      "epoch": 3.5628415300546448,
      "grad_norm": 0.6858357119109362,
      "learning_rate": 6.186688056879353e-06,
      "loss": 11.1183,
      "step": 20375
    },
    {
      "epoch": 3.563016393442623,
      "grad_norm": 0.6711907855608167,
      "learning_rate": 6.181785439135179e-06,
      "loss": 11.1633,
      "step": 20376
    },
    {
      "epoch": 3.563191256830601,
      "grad_norm": 0.6620032562416832,
      "learning_rate": 6.176884702734254e-06,
      "loss": 11.2891,
      "step": 20377
    },
    {
      "epoch": 3.5633661202185793,
      "grad_norm": 0.6754313648366531,
      "learning_rate": 6.171985847774864e-06,
      "loss": 11.0471,
      "step": 20378
    },
    {
      "epoch": 3.5635409836065572,
      "grad_norm": 0.6697742421126,
      "learning_rate": 6.167088874355231e-06,
      "loss": 11.6295,
      "step": 20379
    },
    {
      "epoch": 3.5637158469945356,
      "grad_norm": 0.6311179130556145,
      "learning_rate": 6.1621937825735994e-06,
      "loss": 11.5655,
      "step": 20380
    },
    {
      "epoch": 3.5638907103825135,
      "grad_norm": 0.6855517960910434,
      "learning_rate": 6.157300572528102e-06,
      "loss": 11.2781,
      "step": 20381
    },
    {
      "epoch": 3.564065573770492,
      "grad_norm": 0.6735282182129008,
      "learning_rate": 6.152409244316859e-06,
      "loss": 11.5855,
      "step": 20382
    },
    {
      "epoch": 3.56424043715847,
      "grad_norm": 0.6493061302163639,
      "learning_rate": 6.147519798037949e-06,
      "loss": 11.3052,
      "step": 20383
    },
    {
      "epoch": 3.564415300546448,
      "grad_norm": 0.6574338189370336,
      "learning_rate": 6.142632233789425e-06,
      "loss": 11.4942,
      "step": 20384
    },
    {
      "epoch": 3.564590163934426,
      "grad_norm": 0.6795078728403049,
      "learning_rate": 6.137746551669332e-06,
      "loss": 11.5337,
      "step": 20385
    },
    {
      "epoch": 3.5647650273224043,
      "grad_norm": 0.6910693579052575,
      "learning_rate": 6.132862751775603e-06,
      "loss": 11.5292,
      "step": 20386
    },
    {
      "epoch": 3.5649398907103826,
      "grad_norm": 0.6706465613541851,
      "learning_rate": 6.1279808342061815e-06,
      "loss": 11.4491,
      "step": 20387
    },
    {
      "epoch": 3.5651147540983605,
      "grad_norm": 0.6583819183475238,
      "learning_rate": 6.123100799058978e-06,
      "loss": 11.3423,
      "step": 20388
    },
    {
      "epoch": 3.565289617486339,
      "grad_norm": 0.6983082395409771,
      "learning_rate": 6.118222646431826e-06,
      "loss": 11.51,
      "step": 20389
    },
    {
      "epoch": 3.5654644808743168,
      "grad_norm": 0.6784636255175301,
      "learning_rate": 6.113346376422591e-06,
      "loss": 11.2782,
      "step": 20390
    },
    {
      "epoch": 3.565639344262295,
      "grad_norm": 0.6435756208200683,
      "learning_rate": 6.108471989129017e-06,
      "loss": 11.5984,
      "step": 20391
    },
    {
      "epoch": 3.5658142076502735,
      "grad_norm": 0.6772907190339349,
      "learning_rate": 6.103599484648858e-06,
      "loss": 11.3979,
      "step": 20392
    },
    {
      "epoch": 3.5659890710382514,
      "grad_norm": 0.7254282551453209,
      "learning_rate": 6.098728863079828e-06,
      "loss": 11.4761,
      "step": 20393
    },
    {
      "epoch": 3.5661639344262293,
      "grad_norm": 0.7114196078045301,
      "learning_rate": 6.093860124519612e-06,
      "loss": 11.27,
      "step": 20394
    },
    {
      "epoch": 3.5663387978142076,
      "grad_norm": 0.6103954925269571,
      "learning_rate": 6.0889932690658345e-06,
      "loss": 11.4985,
      "step": 20395
    },
    {
      "epoch": 3.566513661202186,
      "grad_norm": 0.7123702752345289,
      "learning_rate": 6.084128296816071e-06,
      "loss": 11.2965,
      "step": 20396
    },
    {
      "epoch": 3.566688524590164,
      "grad_norm": 0.6784218032799318,
      "learning_rate": 6.079265207867901e-06,
      "loss": 11.6663,
      "step": 20397
    },
    {
      "epoch": 3.566863387978142,
      "grad_norm": 0.6422086155007892,
      "learning_rate": 6.074404002318823e-06,
      "loss": 11.4981,
      "step": 20398
    },
    {
      "epoch": 3.56703825136612,
      "grad_norm": 0.6338055524210654,
      "learning_rate": 6.069544680266359e-06,
      "loss": 11.677,
      "step": 20399
    },
    {
      "epoch": 3.5672131147540984,
      "grad_norm": 0.6939853363957597,
      "learning_rate": 6.064687241807921e-06,
      "loss": 11.716,
      "step": 20400
    },
    {
      "epoch": 3.5673879781420768,
      "grad_norm": 0.6620464568417016,
      "learning_rate": 6.059831687040918e-06,
      "loss": 11.405,
      "step": 20401
    },
    {
      "epoch": 3.5675628415300547,
      "grad_norm": 0.6634435615394995,
      "learning_rate": 6.054978016062729e-06,
      "loss": 11.1223,
      "step": 20402
    },
    {
      "epoch": 3.5677377049180325,
      "grad_norm": 0.6626029108332527,
      "learning_rate": 6.050126228970676e-06,
      "loss": 11.1514,
      "step": 20403
    },
    {
      "epoch": 3.567912568306011,
      "grad_norm": 0.7098427850630891,
      "learning_rate": 6.045276325862059e-06,
      "loss": 11.4741,
      "step": 20404
    },
    {
      "epoch": 3.5680874316939892,
      "grad_norm": 0.6872152062432273,
      "learning_rate": 6.040428306834134e-06,
      "loss": 11.4163,
      "step": 20405
    },
    {
      "epoch": 3.568262295081967,
      "grad_norm": 0.7046796798595995,
      "learning_rate": 6.0355821719841e-06,
      "loss": 11.4678,
      "step": 20406
    },
    {
      "epoch": 3.5684371584699455,
      "grad_norm": 0.6937206535824598,
      "learning_rate": 6.030737921409169e-06,
      "loss": 11.3952,
      "step": 20407
    },
    {
      "epoch": 3.5686120218579234,
      "grad_norm": 0.6873701596407337,
      "learning_rate": 6.025895555206462e-06,
      "loss": 11.59,
      "step": 20408
    },
    {
      "epoch": 3.5687868852459017,
      "grad_norm": 0.6523204626005985,
      "learning_rate": 6.021055073473081e-06,
      "loss": 11.2187,
      "step": 20409
    },
    {
      "epoch": 3.56896174863388,
      "grad_norm": 0.7417754363007906,
      "learning_rate": 6.016216476306103e-06,
      "loss": 11.3662,
      "step": 20410
    },
    {
      "epoch": 3.569136612021858,
      "grad_norm": 0.707552699625468,
      "learning_rate": 6.0113797638025494e-06,
      "loss": 11.3834,
      "step": 20411
    },
    {
      "epoch": 3.569311475409836,
      "grad_norm": 0.685160951270097,
      "learning_rate": 6.0065449360594214e-06,
      "loss": 11.4229,
      "step": 20412
    },
    {
      "epoch": 3.569486338797814,
      "grad_norm": 0.6475561542637508,
      "learning_rate": 6.001711993173664e-06,
      "loss": 11.2203,
      "step": 20413
    },
    {
      "epoch": 3.5696612021857925,
      "grad_norm": 0.6474810805098989,
      "learning_rate": 5.996880935242188e-06,
      "loss": 11.5551,
      "step": 20414
    },
    {
      "epoch": 3.5698360655737704,
      "grad_norm": 0.7331128016840389,
      "learning_rate": 5.992051762361883e-06,
      "loss": 11.3999,
      "step": 20415
    },
    {
      "epoch": 3.5700109289617488,
      "grad_norm": 0.7062035594587056,
      "learning_rate": 5.987224474629571e-06,
      "loss": 11.4496,
      "step": 20416
    },
    {
      "epoch": 3.5701857923497267,
      "grad_norm": 0.6601985248088277,
      "learning_rate": 5.982399072142075e-06,
      "loss": 11.5123,
      "step": 20417
    },
    {
      "epoch": 3.570360655737705,
      "grad_norm": 0.6504134418722495,
      "learning_rate": 5.977575554996151e-06,
      "loss": 11.3832,
      "step": 20418
    },
    {
      "epoch": 3.570535519125683,
      "grad_norm": 0.6803864454175399,
      "learning_rate": 5.972753923288521e-06,
      "loss": 11.3998,
      "step": 20419
    },
    {
      "epoch": 3.5707103825136612,
      "grad_norm": 0.6931782646778363,
      "learning_rate": 5.967934177115875e-06,
      "loss": 11.4137,
      "step": 20420
    },
    {
      "epoch": 3.570885245901639,
      "grad_norm": 0.6601685810759098,
      "learning_rate": 5.963116316574869e-06,
      "loss": 11.5018,
      "step": 20421
    },
    {
      "epoch": 3.5710601092896175,
      "grad_norm": 0.6550307638807861,
      "learning_rate": 5.958300341762113e-06,
      "loss": 11.3587,
      "step": 20422
    },
    {
      "epoch": 3.571234972677596,
      "grad_norm": 0.5981371614818138,
      "learning_rate": 5.953486252774177e-06,
      "loss": 11.3872,
      "step": 20423
    },
    {
      "epoch": 3.5714098360655737,
      "grad_norm": 0.6697453555589403,
      "learning_rate": 5.948674049707603e-06,
      "loss": 11.4383,
      "step": 20424
    },
    {
      "epoch": 3.571584699453552,
      "grad_norm": 0.7267808243706845,
      "learning_rate": 5.943863732658894e-06,
      "loss": 11.3271,
      "step": 20425
    },
    {
      "epoch": 3.57175956284153,
      "grad_norm": 0.7054226701851218,
      "learning_rate": 5.939055301724505e-06,
      "loss": 11.0999,
      "step": 20426
    },
    {
      "epoch": 3.5719344262295083,
      "grad_norm": 0.6409824631779769,
      "learning_rate": 5.93424875700086e-06,
      "loss": 11.6553,
      "step": 20427
    },
    {
      "epoch": 3.572109289617486,
      "grad_norm": 0.7194552699529817,
      "learning_rate": 5.929444098584347e-06,
      "loss": 11.3014,
      "step": 20428
    },
    {
      "epoch": 3.5722841530054645,
      "grad_norm": 0.6814069834504957,
      "learning_rate": 5.924641326571312e-06,
      "loss": 11.0478,
      "step": 20429
    },
    {
      "epoch": 3.5724590163934424,
      "grad_norm": 0.6643010656182192,
      "learning_rate": 5.919840441058067e-06,
      "loss": 11.4163,
      "step": 20430
    },
    {
      "epoch": 3.5726338797814208,
      "grad_norm": 0.7304212831820466,
      "learning_rate": 5.91504144214089e-06,
      "loss": 11.3968,
      "step": 20431
    },
    {
      "epoch": 3.572808743169399,
      "grad_norm": 0.7084842244417023,
      "learning_rate": 5.910244329916015e-06,
      "loss": 11.6199,
      "step": 20432
    },
    {
      "epoch": 3.572983606557377,
      "grad_norm": 0.7199860002416383,
      "learning_rate": 5.905449104479632e-06,
      "loss": 11.5398,
      "step": 20433
    },
    {
      "epoch": 3.5731584699453554,
      "grad_norm": 0.6346064772825282,
      "learning_rate": 5.900655765927898e-06,
      "loss": 11.2182,
      "step": 20434
    },
    {
      "epoch": 3.5733333333333333,
      "grad_norm": 0.7082063067121461,
      "learning_rate": 5.895864314356936e-06,
      "loss": 11.3356,
      "step": 20435
    },
    {
      "epoch": 3.5735081967213116,
      "grad_norm": 0.6515711816314227,
      "learning_rate": 5.891074749862857e-06,
      "loss": 11.3772,
      "step": 20436
    },
    {
      "epoch": 3.5736830601092895,
      "grad_norm": 0.646374622876439,
      "learning_rate": 5.886287072541652e-06,
      "loss": 11.3079,
      "step": 20437
    },
    {
      "epoch": 3.573857923497268,
      "grad_norm": 0.7234105818445503,
      "learning_rate": 5.881501282489365e-06,
      "loss": 11.6462,
      "step": 20438
    },
    {
      "epoch": 3.5740327868852457,
      "grad_norm": 0.7234039988740731,
      "learning_rate": 5.876717379801966e-06,
      "loss": 11.3851,
      "step": 20439
    },
    {
      "epoch": 3.574207650273224,
      "grad_norm": 0.6837521410081148,
      "learning_rate": 5.871935364575376e-06,
      "loss": 11.5168,
      "step": 20440
    },
    {
      "epoch": 3.5743825136612024,
      "grad_norm": 0.6712415284501726,
      "learning_rate": 5.867155236905486e-06,
      "loss": 11.4677,
      "step": 20441
    },
    {
      "epoch": 3.5745573770491803,
      "grad_norm": 0.7542986799102701,
      "learning_rate": 5.862376996888175e-06,
      "loss": 11.2229,
      "step": 20442
    },
    {
      "epoch": 3.574732240437158,
      "grad_norm": 0.7164832618278657,
      "learning_rate": 5.857600644619233e-06,
      "loss": 11.3183,
      "step": 20443
    },
    {
      "epoch": 3.5749071038251365,
      "grad_norm": 0.6975817716063266,
      "learning_rate": 5.85282618019446e-06,
      "loss": 11.22,
      "step": 20444
    },
    {
      "epoch": 3.575081967213115,
      "grad_norm": 0.7426022016776465,
      "learning_rate": 5.848053603709602e-06,
      "loss": 11.5139,
      "step": 20445
    },
    {
      "epoch": 3.575256830601093,
      "grad_norm": 0.6698604702692535,
      "learning_rate": 5.843282915260328e-06,
      "loss": 11.2146,
      "step": 20446
    },
    {
      "epoch": 3.575431693989071,
      "grad_norm": 0.7184525552820832,
      "learning_rate": 5.83851411494234e-06,
      "loss": 11.3367,
      "step": 20447
    },
    {
      "epoch": 3.575606557377049,
      "grad_norm": 0.7430587861391329,
      "learning_rate": 5.8337472028512584e-06,
      "loss": 11.3611,
      "step": 20448
    },
    {
      "epoch": 3.5757814207650274,
      "grad_norm": 0.6399231907649614,
      "learning_rate": 5.828982179082676e-06,
      "loss": 11.5712,
      "step": 20449
    },
    {
      "epoch": 3.5759562841530057,
      "grad_norm": 0.7565118344640128,
      "learning_rate": 5.82421904373216e-06,
      "loss": 11.2535,
      "step": 20450
    },
    {
      "epoch": 3.5761311475409836,
      "grad_norm": 0.6710735361391414,
      "learning_rate": 5.819457796895189e-06,
      "loss": 11.4203,
      "step": 20451
    },
    {
      "epoch": 3.5763060109289615,
      "grad_norm": 0.6579036784930333,
      "learning_rate": 5.8146984386672655e-06,
      "loss": 11.5329,
      "step": 20452
    },
    {
      "epoch": 3.57648087431694,
      "grad_norm": 0.6607935917315276,
      "learning_rate": 5.8099409691438236e-06,
      "loss": 11.2455,
      "step": 20453
    },
    {
      "epoch": 3.576655737704918,
      "grad_norm": 0.6378902244709801,
      "learning_rate": 5.805185388420287e-06,
      "loss": 11.576,
      "step": 20454
    },
    {
      "epoch": 3.576830601092896,
      "grad_norm": 0.6559786775105855,
      "learning_rate": 5.800431696591979e-06,
      "loss": 11.2931,
      "step": 20455
    },
    {
      "epoch": 3.5770054644808744,
      "grad_norm": 0.7050589102757214,
      "learning_rate": 5.7956798937542355e-06,
      "loss": 11.3327,
      "step": 20456
    },
    {
      "epoch": 3.5771803278688523,
      "grad_norm": 0.6993967894255338,
      "learning_rate": 5.790929980002369e-06,
      "loss": 11.1941,
      "step": 20457
    },
    {
      "epoch": 3.5773551912568307,
      "grad_norm": 0.6669630202479544,
      "learning_rate": 5.786181955431613e-06,
      "loss": 11.3705,
      "step": 20458
    },
    {
      "epoch": 3.577530054644809,
      "grad_norm": 0.6853982913444824,
      "learning_rate": 5.781435820137204e-06,
      "loss": 11.6064,
      "step": 20459
    },
    {
      "epoch": 3.577704918032787,
      "grad_norm": 0.7797742109153685,
      "learning_rate": 5.776691574214277e-06,
      "loss": 11.4423,
      "step": 20460
    },
    {
      "epoch": 3.577879781420765,
      "grad_norm": 0.6894166702442399,
      "learning_rate": 5.771949217757977e-06,
      "loss": 11.6293,
      "step": 20461
    },
    {
      "epoch": 3.578054644808743,
      "grad_norm": 0.7354363691414313,
      "learning_rate": 5.767208750863428e-06,
      "loss": 11.3907,
      "step": 20462
    },
    {
      "epoch": 3.5782295081967215,
      "grad_norm": 0.6880742312855017,
      "learning_rate": 5.762470173625689e-06,
      "loss": 11.3609,
      "step": 20463
    },
    {
      "epoch": 3.5784043715846994,
      "grad_norm": 0.6122271364376366,
      "learning_rate": 5.757733486139749e-06,
      "loss": 11.4017,
      "step": 20464
    },
    {
      "epoch": 3.5785792349726777,
      "grad_norm": 0.7453952529524933,
      "learning_rate": 5.75299868850061e-06,
      "loss": 11.2812,
      "step": 20465
    },
    {
      "epoch": 3.5787540983606556,
      "grad_norm": 0.6018315136464468,
      "learning_rate": 5.74826578080323e-06,
      "loss": 11.2247,
      "step": 20466
    },
    {
      "epoch": 3.578928961748634,
      "grad_norm": 0.7007841754997227,
      "learning_rate": 5.743534763142522e-06,
      "loss": 11.2937,
      "step": 20467
    },
    {
      "epoch": 3.5791038251366123,
      "grad_norm": 0.6808217736383543,
      "learning_rate": 5.738805635613354e-06,
      "loss": 11.4853,
      "step": 20468
    },
    {
      "epoch": 3.57927868852459,
      "grad_norm": 0.6394792940772692,
      "learning_rate": 5.734078398310538e-06,
      "loss": 11.4099,
      "step": 20469
    },
    {
      "epoch": 3.579453551912568,
      "grad_norm": 0.6195664492264327,
      "learning_rate": 5.729353051328867e-06,
      "loss": 11.2229,
      "step": 20470
    },
    {
      "epoch": 3.5796284153005464,
      "grad_norm": 0.6781106513826826,
      "learning_rate": 5.724629594763143e-06,
      "loss": 11.3498,
      "step": 20471
    },
    {
      "epoch": 3.5798032786885248,
      "grad_norm": 0.6542057916300353,
      "learning_rate": 5.719908028708065e-06,
      "loss": 11.2479,
      "step": 20472
    },
    {
      "epoch": 3.5799781420765027,
      "grad_norm": 0.6516127065675513,
      "learning_rate": 5.715188353258294e-06,
      "loss": 11.4727,
      "step": 20473
    },
    {
      "epoch": 3.580153005464481,
      "grad_norm": 0.8760334016215836,
      "learning_rate": 5.7104705685084855e-06,
      "loss": 11.3762,
      "step": 20474
    },
    {
      "epoch": 3.580327868852459,
      "grad_norm": 0.6606146624234076,
      "learning_rate": 5.705754674553243e-06,
      "loss": 11.4175,
      "step": 20475
    },
    {
      "epoch": 3.5805027322404372,
      "grad_norm": 0.7006280465479154,
      "learning_rate": 5.7010406714871566e-06,
      "loss": 11.111,
      "step": 20476
    },
    {
      "epoch": 3.580677595628415,
      "grad_norm": 0.6352722893925354,
      "learning_rate": 5.69632855940474e-06,
      "loss": 11.5125,
      "step": 20477
    },
    {
      "epoch": 3.5808524590163935,
      "grad_norm": 0.7221247923431339,
      "learning_rate": 5.691618338400484e-06,
      "loss": 11.34,
      "step": 20478
    },
    {
      "epoch": 3.5810273224043714,
      "grad_norm": 0.667854082273962,
      "learning_rate": 5.686910008568847e-06,
      "loss": 11.2163,
      "step": 20479
    },
    {
      "epoch": 3.5812021857923497,
      "grad_norm": 0.6351542254980926,
      "learning_rate": 5.682203570004219e-06,
      "loss": 11.5941,
      "step": 20480
    },
    {
      "epoch": 3.581377049180328,
      "grad_norm": 0.6908286849654249,
      "learning_rate": 5.677499022801047e-06,
      "loss": 11.0231,
      "step": 20481
    },
    {
      "epoch": 3.581551912568306,
      "grad_norm": 0.6545111251184038,
      "learning_rate": 5.672796367053601e-06,
      "loss": 11.327,
      "step": 20482
    },
    {
      "epoch": 3.5817267759562843,
      "grad_norm": 0.6903180299500489,
      "learning_rate": 5.668095602856216e-06,
      "loss": 11.5184,
      "step": 20483
    },
    {
      "epoch": 3.581901639344262,
      "grad_norm": 0.7567199647349337,
      "learning_rate": 5.663396730303139e-06,
      "loss": 11.3241,
      "step": 20484
    },
    {
      "epoch": 3.5820765027322405,
      "grad_norm": 0.6943331591126969,
      "learning_rate": 5.658699749488605e-06,
      "loss": 11.4606,
      "step": 20485
    },
    {
      "epoch": 3.5822513661202184,
      "grad_norm": 0.6651597867640967,
      "learning_rate": 5.65400466050684e-06,
      "loss": 11.2481,
      "step": 20486
    },
    {
      "epoch": 3.582426229508197,
      "grad_norm": 0.6706911284472685,
      "learning_rate": 5.6493114634519455e-06,
      "loss": 11.3575,
      "step": 20487
    },
    {
      "epoch": 3.5826010928961747,
      "grad_norm": 0.6779192787487948,
      "learning_rate": 5.644620158418046e-06,
      "loss": 11.6402,
      "step": 20488
    },
    {
      "epoch": 3.582775956284153,
      "grad_norm": 0.7681374120255288,
      "learning_rate": 5.6399307454992e-06,
      "loss": 11.6169,
      "step": 20489
    },
    {
      "epoch": 3.5829508196721314,
      "grad_norm": 0.7554999200734093,
      "learning_rate": 5.63524322478951e-06,
      "loss": 11.3946,
      "step": 20490
    },
    {
      "epoch": 3.5831256830601093,
      "grad_norm": 0.7243785506085029,
      "learning_rate": 5.630557596382913e-06,
      "loss": 11.4252,
      "step": 20491
    },
    {
      "epoch": 3.5833005464480876,
      "grad_norm": 0.7001224282843261,
      "learning_rate": 5.625873860373377e-06,
      "loss": 11.4264,
      "step": 20492
    },
    {
      "epoch": 3.5834754098360655,
      "grad_norm": 0.6867745685503603,
      "learning_rate": 5.621192016854848e-06,
      "loss": 11.2243,
      "step": 20493
    },
    {
      "epoch": 3.583650273224044,
      "grad_norm": 0.6547336172207882,
      "learning_rate": 5.616512065921187e-06,
      "loss": 11.1097,
      "step": 20494
    },
    {
      "epoch": 3.5838251366120217,
      "grad_norm": 0.6557943000576542,
      "learning_rate": 5.611834007666272e-06,
      "loss": 11.0475,
      "step": 20495
    },
    {
      "epoch": 3.584,
      "grad_norm": 0.6535388961294476,
      "learning_rate": 5.607157842183896e-06,
      "loss": 11.326,
      "step": 20496
    },
    {
      "epoch": 3.584174863387978,
      "grad_norm": 0.6528606822540476,
      "learning_rate": 5.602483569567818e-06,
      "loss": 11.2889,
      "step": 20497
    },
    {
      "epoch": 3.5843497267759563,
      "grad_norm": 0.711121968005079,
      "learning_rate": 5.597811189911783e-06,
      "loss": 11.2116,
      "step": 20498
    },
    {
      "epoch": 3.5845245901639347,
      "grad_norm": 0.6304806369407162,
      "learning_rate": 5.593140703309496e-06,
      "loss": 11.3324,
      "step": 20499
    },
    {
      "epoch": 3.5846994535519126,
      "grad_norm": 0.6828754654487703,
      "learning_rate": 5.588472109854592e-06,
      "loss": 11.3874,
      "step": 20500
    },
    {
      "epoch": 3.5848743169398904,
      "grad_norm": 0.6712444782861752,
      "learning_rate": 5.583805409640708e-06,
      "loss": 11.5856,
      "step": 20501
    },
    {
      "epoch": 3.585049180327869,
      "grad_norm": 0.7019231064463577,
      "learning_rate": 5.5791406027614235e-06,
      "loss": 11.2238,
      "step": 20502
    },
    {
      "epoch": 3.585224043715847,
      "grad_norm": 0.7227598101149244,
      "learning_rate": 5.574477689310286e-06,
      "loss": 11.4794,
      "step": 20503
    },
    {
      "epoch": 3.585398907103825,
      "grad_norm": 0.6661901937532644,
      "learning_rate": 5.569816669380789e-06,
      "loss": 11.4285,
      "step": 20504
    },
    {
      "epoch": 3.5855737704918034,
      "grad_norm": 0.7532071917327466,
      "learning_rate": 5.565157543066402e-06,
      "loss": 11.5116,
      "step": 20505
    },
    {
      "epoch": 3.5857486338797813,
      "grad_norm": 0.7115359604790543,
      "learning_rate": 5.56050031046057e-06,
      "loss": 11.3785,
      "step": 20506
    },
    {
      "epoch": 3.5859234972677596,
      "grad_norm": 0.6983832402857887,
      "learning_rate": 5.555844971656665e-06,
      "loss": 11.1954,
      "step": 20507
    },
    {
      "epoch": 3.586098360655738,
      "grad_norm": 0.6974110252988365,
      "learning_rate": 5.551191526748057e-06,
      "loss": 11.3481,
      "step": 20508
    },
    {
      "epoch": 3.586273224043716,
      "grad_norm": 0.6989586087158282,
      "learning_rate": 5.546539975828047e-06,
      "loss": 11.4583,
      "step": 20509
    },
    {
      "epoch": 3.5864480874316937,
      "grad_norm": 0.6047135093922964,
      "learning_rate": 5.541890318989928e-06,
      "loss": 11.5064,
      "step": 20510
    },
    {
      "epoch": 3.586622950819672,
      "grad_norm": 0.6909452551047933,
      "learning_rate": 5.537242556326927e-06,
      "loss": 11.4344,
      "step": 20511
    },
    {
      "epoch": 3.5867978142076504,
      "grad_norm": 0.7050285022062817,
      "learning_rate": 5.532596687932257e-06,
      "loss": 11.1314,
      "step": 20512
    },
    {
      "epoch": 3.5869726775956283,
      "grad_norm": 0.6107085114627948,
      "learning_rate": 5.527952713899065e-06,
      "loss": 11.4932,
      "step": 20513
    },
    {
      "epoch": 3.5871475409836067,
      "grad_norm": 0.6510254677384882,
      "learning_rate": 5.5233106343205e-06,
      "loss": 11.4933,
      "step": 20514
    },
    {
      "epoch": 3.5873224043715846,
      "grad_norm": 0.6918458046105288,
      "learning_rate": 5.518670449289631e-06,
      "loss": 11.4031,
      "step": 20515
    },
    {
      "epoch": 3.587497267759563,
      "grad_norm": 0.6984145579832103,
      "learning_rate": 5.514032158899518e-06,
      "loss": 11.1636,
      "step": 20516
    },
    {
      "epoch": 3.5876721311475412,
      "grad_norm": 0.6795063612870262,
      "learning_rate": 5.509395763243175e-06,
      "loss": 11.5518,
      "step": 20517
    },
    {
      "epoch": 3.587846994535519,
      "grad_norm": 0.7554141179442222,
      "learning_rate": 5.504761262413571e-06,
      "loss": 11.4289,
      "step": 20518
    },
    {
      "epoch": 3.588021857923497,
      "grad_norm": 0.6725136026256421,
      "learning_rate": 5.500128656503634e-06,
      "loss": 11.1526,
      "step": 20519
    },
    {
      "epoch": 3.5881967213114754,
      "grad_norm": 0.7179158016635784,
      "learning_rate": 5.495497945606276e-06,
      "loss": 11.2539,
      "step": 20520
    },
    {
      "epoch": 3.5883715846994537,
      "grad_norm": 0.6743558609422902,
      "learning_rate": 5.490869129814346e-06,
      "loss": 11.6729,
      "step": 20521
    },
    {
      "epoch": 3.5885464480874316,
      "grad_norm": 0.7044483871063769,
      "learning_rate": 5.486242209220682e-06,
      "loss": 11.2645,
      "step": 20522
    },
    {
      "epoch": 3.58872131147541,
      "grad_norm": 0.6956640418617794,
      "learning_rate": 5.481617183918053e-06,
      "loss": 11.5478,
      "step": 20523
    },
    {
      "epoch": 3.588896174863388,
      "grad_norm": 0.6405390008849037,
      "learning_rate": 5.476994053999207e-06,
      "loss": 11.6047,
      "step": 20524
    },
    {
      "epoch": 3.589071038251366,
      "grad_norm": 0.7349346666455415,
      "learning_rate": 5.472372819556848e-06,
      "loss": 11.3777,
      "step": 20525
    },
    {
      "epoch": 3.5892459016393445,
      "grad_norm": 0.7085276047997834,
      "learning_rate": 5.4677534806836574e-06,
      "loss": 11.3019,
      "step": 20526
    },
    {
      "epoch": 3.5894207650273224,
      "grad_norm": 0.6873299210212597,
      "learning_rate": 5.463136037472272e-06,
      "loss": 11.4104,
      "step": 20527
    },
    {
      "epoch": 3.5895956284153003,
      "grad_norm": 0.7370042226907633,
      "learning_rate": 5.458520490015251e-06,
      "loss": 10.8819,
      "step": 20528
    },
    {
      "epoch": 3.5897704918032787,
      "grad_norm": 0.694453107686203,
      "learning_rate": 5.4539068384051876e-06,
      "loss": 11.5486,
      "step": 20529
    },
    {
      "epoch": 3.589945355191257,
      "grad_norm": 0.680677632081373,
      "learning_rate": 5.4492950827345955e-06,
      "loss": 11.5216,
      "step": 20530
    },
    {
      "epoch": 3.590120218579235,
      "grad_norm": 0.6822488002386142,
      "learning_rate": 5.444685223095936e-06,
      "loss": 11.2752,
      "step": 20531
    },
    {
      "epoch": 3.5902950819672133,
      "grad_norm": 0.7346817980085245,
      "learning_rate": 5.4400772595816774e-06,
      "loss": 11.4871,
      "step": 20532
    },
    {
      "epoch": 3.590469945355191,
      "grad_norm": 0.7019205069588728,
      "learning_rate": 5.43547119228417e-06,
      "loss": 11.436,
      "step": 20533
    },
    {
      "epoch": 3.5906448087431695,
      "grad_norm": 0.7246472661507004,
      "learning_rate": 5.430867021295838e-06,
      "loss": 11.4572,
      "step": 20534
    },
    {
      "epoch": 3.5908196721311474,
      "grad_norm": 0.7162794869232403,
      "learning_rate": 5.426264746708987e-06,
      "loss": 11.3621,
      "step": 20535
    },
    {
      "epoch": 3.5909945355191257,
      "grad_norm": 0.7012147707140508,
      "learning_rate": 5.421664368615931e-06,
      "loss": 11.4531,
      "step": 20536
    },
    {
      "epoch": 3.5911693989071036,
      "grad_norm": 0.7061889353560973,
      "learning_rate": 5.417065887108852e-06,
      "loss": 11.2169,
      "step": 20537
    },
    {
      "epoch": 3.591344262295082,
      "grad_norm": 0.6749188368901958,
      "learning_rate": 5.412469302280044e-06,
      "loss": 11.2071,
      "step": 20538
    },
    {
      "epoch": 3.5915191256830603,
      "grad_norm": 0.6425860763217961,
      "learning_rate": 5.40787461422163e-06,
      "loss": 11.5275,
      "step": 20539
    },
    {
      "epoch": 3.591693989071038,
      "grad_norm": 0.6897124663669717,
      "learning_rate": 5.403281823025775e-06,
      "loss": 11.3224,
      "step": 20540
    },
    {
      "epoch": 3.5918688524590165,
      "grad_norm": 0.7181485646691262,
      "learning_rate": 5.398690928784578e-06,
      "loss": 11.1415,
      "step": 20541
    },
    {
      "epoch": 3.5920437158469944,
      "grad_norm": 0.6610578501836314,
      "learning_rate": 5.394101931590068e-06,
      "loss": 11.4845,
      "step": 20542
    },
    {
      "epoch": 3.592218579234973,
      "grad_norm": 0.6690356220333954,
      "learning_rate": 5.389514831534315e-06,
      "loss": 11.5923,
      "step": 20543
    },
    {
      "epoch": 3.5923934426229507,
      "grad_norm": 0.6953748873031981,
      "learning_rate": 5.3849296287092675e-06,
      "loss": 11.5232,
      "step": 20544
    },
    {
      "epoch": 3.592568306010929,
      "grad_norm": 0.6083169594341136,
      "learning_rate": 5.3803463232069085e-06,
      "loss": 11.6352,
      "step": 20545
    },
    {
      "epoch": 3.592743169398907,
      "grad_norm": 0.6708473201019965,
      "learning_rate": 5.375764915119108e-06,
      "loss": 11.4144,
      "step": 20546
    },
    {
      "epoch": 3.5929180327868853,
      "grad_norm": 0.6831796756788646,
      "learning_rate": 5.3711854045377485e-06,
      "loss": 11.3525,
      "step": 20547
    },
    {
      "epoch": 3.5930928961748636,
      "grad_norm": 0.6761788974660012,
      "learning_rate": 5.366607791554678e-06,
      "loss": 11.4124,
      "step": 20548
    },
    {
      "epoch": 3.5932677595628415,
      "grad_norm": 0.6742553236468181,
      "learning_rate": 5.36203207626168e-06,
      "loss": 11.3266,
      "step": 20549
    },
    {
      "epoch": 3.5934426229508194,
      "grad_norm": 0.7114346245675356,
      "learning_rate": 5.357458258750547e-06,
      "loss": 11.4942,
      "step": 20550
    },
    {
      "epoch": 3.5936174863387977,
      "grad_norm": 0.6811286604257083,
      "learning_rate": 5.352886339112939e-06,
      "loss": 11.2773,
      "step": 20551
    },
    {
      "epoch": 3.593792349726776,
      "grad_norm": 0.6992954270719022,
      "learning_rate": 5.348316317440549e-06,
      "loss": 11.463,
      "step": 20552
    },
    {
      "epoch": 3.593967213114754,
      "grad_norm": 0.6803751682141337,
      "learning_rate": 5.34374819382506e-06,
      "loss": 11.3562,
      "step": 20553
    },
    {
      "epoch": 3.5941420765027323,
      "grad_norm": 0.6942058833971101,
      "learning_rate": 5.339181968358054e-06,
      "loss": 11.6014,
      "step": 20554
    },
    {
      "epoch": 3.59431693989071,
      "grad_norm": 0.6704026897394255,
      "learning_rate": 5.33461764113109e-06,
      "loss": 11.3808,
      "step": 20555
    },
    {
      "epoch": 3.5944918032786886,
      "grad_norm": 0.6289610176314814,
      "learning_rate": 5.330055212235707e-06,
      "loss": 11.2676,
      "step": 20556
    },
    {
      "epoch": 3.594666666666667,
      "grad_norm": 0.8172469540630385,
      "learning_rate": 5.325494681763365e-06,
      "loss": 11.4511,
      "step": 20557
    },
    {
      "epoch": 3.594841530054645,
      "grad_norm": 0.6606503019283138,
      "learning_rate": 5.320936049805569e-06,
      "loss": 11.3773,
      "step": 20558
    },
    {
      "epoch": 3.5950163934426227,
      "grad_norm": 0.6982582480685291,
      "learning_rate": 5.316379316453713e-06,
      "loss": 11.2779,
      "step": 20559
    },
    {
      "epoch": 3.595191256830601,
      "grad_norm": 0.699499898889848,
      "learning_rate": 5.311824481799166e-06,
      "loss": 11.4519,
      "step": 20560
    },
    {
      "epoch": 3.5953661202185794,
      "grad_norm": 0.6546932640857263,
      "learning_rate": 5.307271545933245e-06,
      "loss": 11.1887,
      "step": 20561
    },
    {
      "epoch": 3.5955409836065573,
      "grad_norm": 0.6675289318392631,
      "learning_rate": 5.30272050894729e-06,
      "loss": 11.4219,
      "step": 20562
    },
    {
      "epoch": 3.5957158469945356,
      "grad_norm": 0.6810281978044279,
      "learning_rate": 5.298171370932558e-06,
      "loss": 11.1862,
      "step": 20563
    },
    {
      "epoch": 3.5958907103825135,
      "grad_norm": 0.657599575749701,
      "learning_rate": 5.293624131980268e-06,
      "loss": 11.4681,
      "step": 20564
    },
    {
      "epoch": 3.596065573770492,
      "grad_norm": 0.6702993672825601,
      "learning_rate": 5.2890787921815895e-06,
      "loss": 11.5985,
      "step": 20565
    },
    {
      "epoch": 3.59624043715847,
      "grad_norm": 0.668243030768679,
      "learning_rate": 5.2845353516276616e-06,
      "loss": 11.0053,
      "step": 20566
    },
    {
      "epoch": 3.596415300546448,
      "grad_norm": 0.6454923220920559,
      "learning_rate": 5.279993810409633e-06,
      "loss": 11.3771,
      "step": 20567
    },
    {
      "epoch": 3.596590163934426,
      "grad_norm": 0.7231566224712127,
      "learning_rate": 5.275454168618577e-06,
      "loss": 11.4814,
      "step": 20568
    },
    {
      "epoch": 3.5967650273224043,
      "grad_norm": 0.749575907044823,
      "learning_rate": 5.270916426345484e-06,
      "loss": 11.258,
      "step": 20569
    },
    {
      "epoch": 3.5969398907103827,
      "grad_norm": 0.6997466482605952,
      "learning_rate": 5.266380583681374e-06,
      "loss": 11.5282,
      "step": 20570
    },
    {
      "epoch": 3.5971147540983606,
      "grad_norm": 0.6246900850261646,
      "learning_rate": 5.261846640717194e-06,
      "loss": 11.6327,
      "step": 20571
    },
    {
      "epoch": 3.597289617486339,
      "grad_norm": 0.643737651680527,
      "learning_rate": 5.257314597543883e-06,
      "loss": 11.521,
      "step": 20572
    },
    {
      "epoch": 3.597464480874317,
      "grad_norm": 0.7392639869598888,
      "learning_rate": 5.252784454252335e-06,
      "loss": 11.3926,
      "step": 20573
    },
    {
      "epoch": 3.597639344262295,
      "grad_norm": 0.6628022690479212,
      "learning_rate": 5.248256210933355e-06,
      "loss": 11.518,
      "step": 20574
    },
    {
      "epoch": 3.5978142076502735,
      "grad_norm": 0.6483200998427381,
      "learning_rate": 5.2437298676777605e-06,
      "loss": 11.4435,
      "step": 20575
    },
    {
      "epoch": 3.5979890710382514,
      "grad_norm": 0.677497586431336,
      "learning_rate": 5.239205424576299e-06,
      "loss": 11.3317,
      "step": 20576
    },
    {
      "epoch": 3.5981639344262293,
      "grad_norm": 0.8390688962127367,
      "learning_rate": 5.2346828817197655e-06,
      "loss": 11.2485,
      "step": 20577
    },
    {
      "epoch": 3.5983387978142076,
      "grad_norm": 0.7057335484049473,
      "learning_rate": 5.230162239198788e-06,
      "loss": 11.3451,
      "step": 20578
    },
    {
      "epoch": 3.598513661202186,
      "grad_norm": 0.7046934256334537,
      "learning_rate": 5.225643497104038e-06,
      "loss": 11.541,
      "step": 20579
    },
    {
      "epoch": 3.598688524590164,
      "grad_norm": 0.6818366628083167,
      "learning_rate": 5.221126655526132e-06,
      "loss": 11.1525,
      "step": 20580
    },
    {
      "epoch": 3.598863387978142,
      "grad_norm": 0.6756340230353121,
      "learning_rate": 5.21661171455563e-06,
      "loss": 11.3453,
      "step": 20581
    },
    {
      "epoch": 3.59903825136612,
      "grad_norm": 0.7203315066137109,
      "learning_rate": 5.212098674283117e-06,
      "loss": 11.2178,
      "step": 20582
    },
    {
      "epoch": 3.5992131147540984,
      "grad_norm": 0.6901551545140087,
      "learning_rate": 5.2075875347990526e-06,
      "loss": 11.3031,
      "step": 20583
    },
    {
      "epoch": 3.5993879781420763,
      "grad_norm": 0.7153395980865249,
      "learning_rate": 5.203078296193908e-06,
      "loss": 11.4894,
      "step": 20584
    },
    {
      "epoch": 3.5995628415300547,
      "grad_norm": 0.7346469067898649,
      "learning_rate": 5.198570958558091e-06,
      "loss": 11.4557,
      "step": 20585
    },
    {
      "epoch": 3.5997377049180326,
      "grad_norm": 0.6710878898802682,
      "learning_rate": 5.194065521982028e-06,
      "loss": 11.2054,
      "step": 20586
    },
    {
      "epoch": 3.599912568306011,
      "grad_norm": 0.7068273567902561,
      "learning_rate": 5.189561986556035e-06,
      "loss": 11.3255,
      "step": 20587
    },
    {
      "epoch": 3.6000874316939893,
      "grad_norm": 0.6773867585178498,
      "learning_rate": 5.18506035237043e-06,
      "loss": 11.3991,
      "step": 20588
    },
    {
      "epoch": 3.600262295081967,
      "grad_norm": 0.6605069001118589,
      "learning_rate": 5.180560619515484e-06,
      "loss": 11.5591,
      "step": 20589
    },
    {
      "epoch": 3.6004371584699455,
      "grad_norm": 0.6473297553479594,
      "learning_rate": 5.176062788081426e-06,
      "loss": 11.3181,
      "step": 20590
    },
    {
      "epoch": 3.6006120218579234,
      "grad_norm": 0.6980960411071551,
      "learning_rate": 5.171566858158472e-06,
      "loss": 11.514,
      "step": 20591
    },
    {
      "epoch": 3.6007868852459017,
      "grad_norm": 0.7132553853128425,
      "learning_rate": 5.167072829836761e-06,
      "loss": 11.3613,
      "step": 20592
    },
    {
      "epoch": 3.6009617486338796,
      "grad_norm": 0.7459065049047447,
      "learning_rate": 5.1625807032064096e-06,
      "loss": 10.9519,
      "step": 20593
    },
    {
      "epoch": 3.601136612021858,
      "grad_norm": 0.7532294113309961,
      "learning_rate": 5.158090478357502e-06,
      "loss": 11.3368,
      "step": 20594
    },
    {
      "epoch": 3.601311475409836,
      "grad_norm": 0.7316200235565752,
      "learning_rate": 5.153602155380089e-06,
      "loss": 11.4164,
      "step": 20595
    },
    {
      "epoch": 3.601486338797814,
      "grad_norm": 0.6977896430522582,
      "learning_rate": 5.149115734364162e-06,
      "loss": 11.6382,
      "step": 20596
    },
    {
      "epoch": 3.6016612021857926,
      "grad_norm": 0.711999106021076,
      "learning_rate": 5.144631215399698e-06,
      "loss": 11.1661,
      "step": 20597
    },
    {
      "epoch": 3.6018360655737705,
      "grad_norm": 0.7205719748508337,
      "learning_rate": 5.1401485985766215e-06,
      "loss": 11.2826,
      "step": 20598
    },
    {
      "epoch": 3.602010928961749,
      "grad_norm": 0.69357967550045,
      "learning_rate": 5.135667883984829e-06,
      "loss": 11.1926,
      "step": 20599
    },
    {
      "epoch": 3.6021857923497267,
      "grad_norm": 0.7053482696154206,
      "learning_rate": 5.13118907171416e-06,
      "loss": 11.3767,
      "step": 20600
    },
    {
      "epoch": 3.602360655737705,
      "grad_norm": 0.6639007992962144,
      "learning_rate": 5.126712161854441e-06,
      "loss": 11.5025,
      "step": 20601
    },
    {
      "epoch": 3.602535519125683,
      "grad_norm": 0.6218522392766122,
      "learning_rate": 5.122237154495446e-06,
      "loss": 11.4505,
      "step": 20602
    },
    {
      "epoch": 3.6027103825136613,
      "grad_norm": 0.6458267785371303,
      "learning_rate": 5.117764049726903e-06,
      "loss": 11.2454,
      "step": 20603
    },
    {
      "epoch": 3.602885245901639,
      "grad_norm": 0.7018360264071861,
      "learning_rate": 5.113292847638518e-06,
      "loss": 11.3865,
      "step": 20604
    },
    {
      "epoch": 3.6030601092896175,
      "grad_norm": 0.6214226081163313,
      "learning_rate": 5.108823548319952e-06,
      "loss": 11.6366,
      "step": 20605
    },
    {
      "epoch": 3.603234972677596,
      "grad_norm": 0.6530458875129569,
      "learning_rate": 5.104356151860823e-06,
      "loss": 11.3191,
      "step": 20606
    },
    {
      "epoch": 3.6034098360655737,
      "grad_norm": 0.9809507240838029,
      "learning_rate": 5.099890658350726e-06,
      "loss": 11.1954,
      "step": 20607
    },
    {
      "epoch": 3.6035846994535516,
      "grad_norm": 0.6889746568621041,
      "learning_rate": 5.0954270678792104e-06,
      "loss": 11.4018,
      "step": 20608
    },
    {
      "epoch": 3.60375956284153,
      "grad_norm": 0.6632841853499918,
      "learning_rate": 5.090965380535772e-06,
      "loss": 11.6249,
      "step": 20609
    },
    {
      "epoch": 3.6039344262295083,
      "grad_norm": 0.6699224426910525,
      "learning_rate": 5.086505596409885e-06,
      "loss": 11.5083,
      "step": 20610
    },
    {
      "epoch": 3.6041092896174862,
      "grad_norm": 0.7347424645095331,
      "learning_rate": 5.082047715590976e-06,
      "loss": 11.0308,
      "step": 20611
    },
    {
      "epoch": 3.6042841530054646,
      "grad_norm": 0.6090844861875756,
      "learning_rate": 5.077591738168453e-06,
      "loss": 11.6801,
      "step": 20612
    },
    {
      "epoch": 3.6044590163934425,
      "grad_norm": 0.6667860249415443,
      "learning_rate": 5.073137664231675e-06,
      "loss": 11.4359,
      "step": 20613
    },
    {
      "epoch": 3.604633879781421,
      "grad_norm": 0.6800794285068404,
      "learning_rate": 5.06868549386994e-06,
      "loss": 11.3903,
      "step": 20614
    },
    {
      "epoch": 3.604808743169399,
      "grad_norm": 0.7209174541764901,
      "learning_rate": 5.064235227172542e-06,
      "loss": 11.4061,
      "step": 20615
    },
    {
      "epoch": 3.604983606557377,
      "grad_norm": 0.641909896033241,
      "learning_rate": 5.059786864228711e-06,
      "loss": 11.6314,
      "step": 20616
    },
    {
      "epoch": 3.605158469945355,
      "grad_norm": 0.7791251689432742,
      "learning_rate": 5.055340405127673e-06,
      "loss": 11.2035,
      "step": 20617
    },
    {
      "epoch": 3.6053333333333333,
      "grad_norm": 0.6205789569765655,
      "learning_rate": 5.05089584995857e-06,
      "loss": 11.5948,
      "step": 20618
    },
    {
      "epoch": 3.6055081967213116,
      "grad_norm": 0.6168526630323768,
      "learning_rate": 5.04645319881053e-06,
      "loss": 11.3756,
      "step": 20619
    },
    {
      "epoch": 3.6056830601092895,
      "grad_norm": 0.7104771172004178,
      "learning_rate": 5.04201245177266e-06,
      "loss": 11.2408,
      "step": 20620
    },
    {
      "epoch": 3.605857923497268,
      "grad_norm": 0.6573220498781213,
      "learning_rate": 5.0375736089339985e-06,
      "loss": 11.4342,
      "step": 20621
    },
    {
      "epoch": 3.6060327868852458,
      "grad_norm": 0.6983856529433933,
      "learning_rate": 5.033136670383554e-06,
      "loss": 11.413,
      "step": 20622
    },
    {
      "epoch": 3.606207650273224,
      "grad_norm": 0.6387494661441481,
      "learning_rate": 5.02870163621032e-06,
      "loss": 11.5659,
      "step": 20623
    },
    {
      "epoch": 3.6063825136612024,
      "grad_norm": 0.6685196400546757,
      "learning_rate": 5.024268506503194e-06,
      "loss": 11.3647,
      "step": 20624
    },
    {
      "epoch": 3.6065573770491803,
      "grad_norm": 0.6963483140760186,
      "learning_rate": 5.0198372813511144e-06,
      "loss": 11.5017,
      "step": 20625
    },
    {
      "epoch": 3.6067322404371582,
      "grad_norm": 0.6681927082908145,
      "learning_rate": 5.015407960842922e-06,
      "loss": 11.4341,
      "step": 20626
    },
    {
      "epoch": 3.6069071038251366,
      "grad_norm": 0.6292233331899143,
      "learning_rate": 5.010980545067434e-06,
      "loss": 11.5553,
      "step": 20627
    },
    {
      "epoch": 3.607081967213115,
      "grad_norm": 0.6781768752271935,
      "learning_rate": 5.006555034113458e-06,
      "loss": 11.5509,
      "step": 20628
    },
    {
      "epoch": 3.607256830601093,
      "grad_norm": 0.6499727582906473,
      "learning_rate": 5.002131428069701e-06,
      "loss": 11.4134,
      "step": 20629
    },
    {
      "epoch": 3.607431693989071,
      "grad_norm": 0.6715006580154654,
      "learning_rate": 4.997709727024902e-06,
      "loss": 11.1836,
      "step": 20630
    },
    {
      "epoch": 3.607606557377049,
      "grad_norm": 0.6501200016622355,
      "learning_rate": 4.993289931067713e-06,
      "loss": 11.3654,
      "step": 20631
    },
    {
      "epoch": 3.6077814207650274,
      "grad_norm": 0.6749024824866585,
      "learning_rate": 4.988872040286785e-06,
      "loss": 11.5495,
      "step": 20632
    },
    {
      "epoch": 3.6079562841530057,
      "grad_norm": 0.7088939861450023,
      "learning_rate": 4.98445605477067e-06,
      "loss": 11.3349,
      "step": 20633
    },
    {
      "epoch": 3.6081311475409836,
      "grad_norm": 0.6771379057590167,
      "learning_rate": 4.980041974607963e-06,
      "loss": 11.5557,
      "step": 20634
    },
    {
      "epoch": 3.6083060109289615,
      "grad_norm": 0.6382336173298802,
      "learning_rate": 4.975629799887161e-06,
      "loss": 11.164,
      "step": 20635
    },
    {
      "epoch": 3.60848087431694,
      "grad_norm": 0.6656160874841216,
      "learning_rate": 4.971219530696747e-06,
      "loss": 11.2294,
      "step": 20636
    },
    {
      "epoch": 3.608655737704918,
      "grad_norm": 0.6138778227531407,
      "learning_rate": 4.966811167125174e-06,
      "loss": 11.3543,
      "step": 20637
    },
    {
      "epoch": 3.608830601092896,
      "grad_norm": 0.8807610629614601,
      "learning_rate": 4.962404709260793e-06,
      "loss": 11.028,
      "step": 20638
    },
    {
      "epoch": 3.6090054644808744,
      "grad_norm": 0.6332373811628391,
      "learning_rate": 4.958000157192022e-06,
      "loss": 11.5456,
      "step": 20639
    },
    {
      "epoch": 3.6091803278688523,
      "grad_norm": 0.6569592513415493,
      "learning_rate": 4.953597511007158e-06,
      "loss": 11.2578,
      "step": 20640
    },
    {
      "epoch": 3.6093551912568307,
      "grad_norm": 0.6702603196424044,
      "learning_rate": 4.949196770794507e-06,
      "loss": 11.6189,
      "step": 20641
    },
    {
      "epoch": 3.6095300546448086,
      "grad_norm": 0.6537627263996731,
      "learning_rate": 4.944797936642298e-06,
      "loss": 11.3867,
      "step": 20642
    },
    {
      "epoch": 3.609704918032787,
      "grad_norm": 0.7078320586169893,
      "learning_rate": 4.9404010086387285e-06,
      "loss": 11.3088,
      "step": 20643
    },
    {
      "epoch": 3.609879781420765,
      "grad_norm": 0.6546679683015213,
      "learning_rate": 4.936005986872016e-06,
      "loss": 11.6102,
      "step": 20644
    },
    {
      "epoch": 3.610054644808743,
      "grad_norm": 0.6318089722319531,
      "learning_rate": 4.931612871430258e-06,
      "loss": 11.2737,
      "step": 20645
    },
    {
      "epoch": 3.6102295081967215,
      "grad_norm": 0.6589188991809561,
      "learning_rate": 4.927221662401571e-06,
      "loss": 11.3184,
      "step": 20646
    },
    {
      "epoch": 3.6104043715846994,
      "grad_norm": 0.6295917285359459,
      "learning_rate": 4.922832359873996e-06,
      "loss": 11.5036,
      "step": 20647
    },
    {
      "epoch": 3.6105792349726777,
      "grad_norm": 0.6222442637631291,
      "learning_rate": 4.918444963935543e-06,
      "loss": 11.4956,
      "step": 20648
    },
    {
      "epoch": 3.6107540983606556,
      "grad_norm": 0.6371244285798852,
      "learning_rate": 4.914059474674216e-06,
      "loss": 11.1683,
      "step": 20649
    },
    {
      "epoch": 3.610928961748634,
      "grad_norm": 0.6318433014581143,
      "learning_rate": 4.9096758921779675e-06,
      "loss": 11.5154,
      "step": 20650
    },
    {
      "epoch": 3.611103825136612,
      "grad_norm": 0.6417720300928148,
      "learning_rate": 4.905294216534661e-06,
      "loss": 11.5946,
      "step": 20651
    },
    {
      "epoch": 3.61127868852459,
      "grad_norm": 0.7207767076340712,
      "learning_rate": 4.900914447832194e-06,
      "loss": 11.3772,
      "step": 20652
    },
    {
      "epoch": 3.611453551912568,
      "grad_norm": 0.6457707219370878,
      "learning_rate": 4.896536586158362e-06,
      "loss": 11.3914,
      "step": 20653
    },
    {
      "epoch": 3.6116284153005465,
      "grad_norm": 0.6934572510369746,
      "learning_rate": 4.892160631600995e-06,
      "loss": 11.109,
      "step": 20654
    },
    {
      "epoch": 3.611803278688525,
      "grad_norm": 0.7263804900533171,
      "learning_rate": 4.887786584247844e-06,
      "loss": 10.9949,
      "step": 20655
    },
    {
      "epoch": 3.6119781420765027,
      "grad_norm": 0.6823683198871299,
      "learning_rate": 4.883414444186584e-06,
      "loss": 11.5011,
      "step": 20656
    },
    {
      "epoch": 3.612153005464481,
      "grad_norm": 0.6501176407494372,
      "learning_rate": 4.879044211504902e-06,
      "loss": 11.4164,
      "step": 20657
    },
    {
      "epoch": 3.612327868852459,
      "grad_norm": 0.7328384368288021,
      "learning_rate": 4.874675886290459e-06,
      "loss": 11.4367,
      "step": 20658
    },
    {
      "epoch": 3.6125027322404373,
      "grad_norm": 0.6872475998617453,
      "learning_rate": 4.870309468630851e-06,
      "loss": 11.4903,
      "step": 20659
    },
    {
      "epoch": 3.612677595628415,
      "grad_norm": 0.6611507573137154,
      "learning_rate": 4.865944958613611e-06,
      "loss": 11.5069,
      "step": 20660
    },
    {
      "epoch": 3.6128524590163935,
      "grad_norm": 0.7141388761279769,
      "learning_rate": 4.861582356326266e-06,
      "loss": 11.2858,
      "step": 20661
    },
    {
      "epoch": 3.6130273224043714,
      "grad_norm": 0.6916462533418505,
      "learning_rate": 4.857221661856304e-06,
      "loss": 11.3258,
      "step": 20662
    },
    {
      "epoch": 3.6132021857923498,
      "grad_norm": 0.6713180111090462,
      "learning_rate": 4.852862875291186e-06,
      "loss": 11.277,
      "step": 20663
    },
    {
      "epoch": 3.613377049180328,
      "grad_norm": 0.6041988277003986,
      "learning_rate": 4.8485059967183225e-06,
      "loss": 11.5453,
      "step": 20664
    },
    {
      "epoch": 3.613551912568306,
      "grad_norm": 0.7423265208188636,
      "learning_rate": 4.8441510262250526e-06,
      "loss": 11.3215,
      "step": 20665
    },
    {
      "epoch": 3.613726775956284,
      "grad_norm": 0.7064214781924966,
      "learning_rate": 4.839797963898718e-06,
      "loss": 11.4634,
      "step": 20666
    },
    {
      "epoch": 3.6139016393442622,
      "grad_norm": 0.6635743085715551,
      "learning_rate": 4.835446809826604e-06,
      "loss": 11.5167,
      "step": 20667
    },
    {
      "epoch": 3.6140765027322406,
      "grad_norm": 0.6747489603673988,
      "learning_rate": 4.831097564095999e-06,
      "loss": 11.3099,
      "step": 20668
    },
    {
      "epoch": 3.6142513661202185,
      "grad_norm": 0.6640510243921918,
      "learning_rate": 4.8267502267940855e-06,
      "loss": 11.5614,
      "step": 20669
    },
    {
      "epoch": 3.614426229508197,
      "grad_norm": 0.66977315782981,
      "learning_rate": 4.82240479800804e-06,
      "loss": 11.5176,
      "step": 20670
    },
    {
      "epoch": 3.6146010928961747,
      "grad_norm": 0.6574707106441685,
      "learning_rate": 4.818061277825026e-06,
      "loss": 11.4772,
      "step": 20671
    },
    {
      "epoch": 3.614775956284153,
      "grad_norm": 0.7278567921679109,
      "learning_rate": 4.813719666332106e-06,
      "loss": 11.4739,
      "step": 20672
    },
    {
      "epoch": 3.6149508196721314,
      "grad_norm": 0.6909391212502795,
      "learning_rate": 4.809379963616401e-06,
      "loss": 11.6788,
      "step": 20673
    },
    {
      "epoch": 3.6151256830601093,
      "grad_norm": 0.7241572636760556,
      "learning_rate": 4.805042169764884e-06,
      "loss": 11.4628,
      "step": 20674
    },
    {
      "epoch": 3.615300546448087,
      "grad_norm": 0.6607320379374143,
      "learning_rate": 4.800706284864565e-06,
      "loss": 11.2263,
      "step": 20675
    },
    {
      "epoch": 3.6154754098360655,
      "grad_norm": 0.6193213790566231,
      "learning_rate": 4.796372309002372e-06,
      "loss": 11.456,
      "step": 20676
    },
    {
      "epoch": 3.615650273224044,
      "grad_norm": 0.6730715686409053,
      "learning_rate": 4.792040242265239e-06,
      "loss": 11.4772,
      "step": 20677
    },
    {
      "epoch": 3.6158251366120218,
      "grad_norm": 0.713902353277044,
      "learning_rate": 4.787710084740027e-06,
      "loss": 11.4758,
      "step": 20678
    },
    {
      "epoch": 3.616,
      "grad_norm": 0.6611975043884698,
      "learning_rate": 4.7833818365135565e-06,
      "loss": 11.5352,
      "step": 20679
    },
    {
      "epoch": 3.616174863387978,
      "grad_norm": 0.6844112689073706,
      "learning_rate": 4.779055497672646e-06,
      "loss": 11.2469,
      "step": 20680
    },
    {
      "epoch": 3.6163497267759563,
      "grad_norm": 0.6380343477710326,
      "learning_rate": 4.774731068304017e-06,
      "loss": 10.9217,
      "step": 20681
    },
    {
      "epoch": 3.6165245901639347,
      "grad_norm": 0.643043684291732,
      "learning_rate": 4.770408548494443e-06,
      "loss": 11.099,
      "step": 20682
    },
    {
      "epoch": 3.6166994535519126,
      "grad_norm": 0.720372416862012,
      "learning_rate": 4.766087938330555e-06,
      "loss": 11.2329,
      "step": 20683
    },
    {
      "epoch": 3.6168743169398905,
      "grad_norm": 0.6329318021734569,
      "learning_rate": 4.761769237899005e-06,
      "loss": 11.4001,
      "step": 20684
    },
    {
      "epoch": 3.617049180327869,
      "grad_norm": 0.6211657798658915,
      "learning_rate": 4.757452447286415e-06,
      "loss": 11.688,
      "step": 20685
    },
    {
      "epoch": 3.617224043715847,
      "grad_norm": 0.6869089399005136,
      "learning_rate": 4.753137566579324e-06,
      "loss": 11.5809,
      "step": 20686
    },
    {
      "epoch": 3.617398907103825,
      "grad_norm": 0.7065310907889302,
      "learning_rate": 4.7488245958642766e-06,
      "loss": 11.3223,
      "step": 20687
    },
    {
      "epoch": 3.6175737704918034,
      "grad_norm": 0.6730280023393236,
      "learning_rate": 4.744513535227757e-06,
      "loss": 11.6006,
      "step": 20688
    },
    {
      "epoch": 3.6177486338797813,
      "grad_norm": 0.6639352972253543,
      "learning_rate": 4.740204384756208e-06,
      "loss": 11.4438,
      "step": 20689
    },
    {
      "epoch": 3.6179234972677596,
      "grad_norm": 0.7032413697608471,
      "learning_rate": 4.73589714453605e-06,
      "loss": 11.539,
      "step": 20690
    },
    {
      "epoch": 3.618098360655738,
      "grad_norm": 0.637370953778899,
      "learning_rate": 4.731591814653658e-06,
      "loss": 11.5721,
      "step": 20691
    },
    {
      "epoch": 3.618273224043716,
      "grad_norm": 0.6581726797020778,
      "learning_rate": 4.72728839519535e-06,
      "loss": 11.5083,
      "step": 20692
    },
    {
      "epoch": 3.6184480874316938,
      "grad_norm": 0.616596059610861,
      "learning_rate": 4.722986886247449e-06,
      "loss": 11.5263,
      "step": 20693
    },
    {
      "epoch": 3.618622950819672,
      "grad_norm": 0.6461758001883267,
      "learning_rate": 4.718687287896195e-06,
      "loss": 11.3925,
      "step": 20694
    },
    {
      "epoch": 3.6187978142076505,
      "grad_norm": 0.6605182950398318,
      "learning_rate": 4.714389600227808e-06,
      "loss": 11.3893,
      "step": 20695
    },
    {
      "epoch": 3.6189726775956284,
      "grad_norm": 0.6370468096425856,
      "learning_rate": 4.710093823328476e-06,
      "loss": 11.5213,
      "step": 20696
    },
    {
      "epoch": 3.6191475409836067,
      "grad_norm": 0.6676127781976634,
      "learning_rate": 4.705799957284351e-06,
      "loss": 11.3954,
      "step": 20697
    },
    {
      "epoch": 3.6193224043715846,
      "grad_norm": 0.6812127751250208,
      "learning_rate": 4.701508002181521e-06,
      "loss": 11.4129,
      "step": 20698
    },
    {
      "epoch": 3.619497267759563,
      "grad_norm": 0.635809870509958,
      "learning_rate": 4.697217958106059e-06,
      "loss": 11.4644,
      "step": 20699
    },
    {
      "epoch": 3.619672131147541,
      "grad_norm": 0.6913245179347725,
      "learning_rate": 4.692929825143999e-06,
      "loss": 11.3436,
      "step": 20700
    },
    {
      "epoch": 3.619846994535519,
      "grad_norm": 0.7095345584001688,
      "learning_rate": 4.688643603381326e-06,
      "loss": 11.1357,
      "step": 20701
    },
    {
      "epoch": 3.620021857923497,
      "grad_norm": 1.1651080609871993,
      "learning_rate": 4.684359292903995e-06,
      "loss": 11.5699,
      "step": 20702
    },
    {
      "epoch": 3.6201967213114754,
      "grad_norm": 0.6458293515309672,
      "learning_rate": 4.680076893797914e-06,
      "loss": 11.6083,
      "step": 20703
    },
    {
      "epoch": 3.6203715846994537,
      "grad_norm": 0.6821648255028119,
      "learning_rate": 4.675796406148958e-06,
      "loss": 11.4211,
      "step": 20704
    },
    {
      "epoch": 3.6205464480874316,
      "grad_norm": 0.6760672161534634,
      "learning_rate": 4.671517830042971e-06,
      "loss": 11.4081,
      "step": 20705
    },
    {
      "epoch": 3.62072131147541,
      "grad_norm": 0.6451296069946961,
      "learning_rate": 4.66724116556575e-06,
      "loss": 11.326,
      "step": 20706
    },
    {
      "epoch": 3.620896174863388,
      "grad_norm": 0.6730146654217966,
      "learning_rate": 4.662966412803061e-06,
      "loss": 11.5864,
      "step": 20707
    },
    {
      "epoch": 3.6210710382513662,
      "grad_norm": 0.6358113205596585,
      "learning_rate": 4.658693571840611e-06,
      "loss": 11.6513,
      "step": 20708
    },
    {
      "epoch": 3.621245901639344,
      "grad_norm": 0.6725284701133715,
      "learning_rate": 4.654422642764089e-06,
      "loss": 11.3204,
      "step": 20709
    },
    {
      "epoch": 3.6214207650273225,
      "grad_norm": 0.8235940770717823,
      "learning_rate": 4.650153625659137e-06,
      "loss": 11.2876,
      "step": 20710
    },
    {
      "epoch": 3.6215956284153004,
      "grad_norm": 0.7396713097550192,
      "learning_rate": 4.645886520611376e-06,
      "loss": 11.5654,
      "step": 20711
    },
    {
      "epoch": 3.6217704918032787,
      "grad_norm": 0.7554667424389433,
      "learning_rate": 4.641621327706369e-06,
      "loss": 11.4922,
      "step": 20712
    },
    {
      "epoch": 3.621945355191257,
      "grad_norm": 0.6514745025826372,
      "learning_rate": 4.637358047029638e-06,
      "loss": 11.4171,
      "step": 20713
    },
    {
      "epoch": 3.622120218579235,
      "grad_norm": 0.7702899844815465,
      "learning_rate": 4.633096678666682e-06,
      "loss": 11.129,
      "step": 20714
    },
    {
      "epoch": 3.622295081967213,
      "grad_norm": 0.6917194074537102,
      "learning_rate": 4.628837222702953e-06,
      "loss": 10.922,
      "step": 20715
    },
    {
      "epoch": 3.622469945355191,
      "grad_norm": 0.6304856411341777,
      "learning_rate": 4.624579679223862e-06,
      "loss": 11.3148,
      "step": 20716
    },
    {
      "epoch": 3.6226448087431695,
      "grad_norm": 0.6866005957214895,
      "learning_rate": 4.620324048314794e-06,
      "loss": 11.3652,
      "step": 20717
    },
    {
      "epoch": 3.6228196721311474,
      "grad_norm": 0.6778641392501918,
      "learning_rate": 4.616070330061073e-06,
      "loss": 11.4997,
      "step": 20718
    },
    {
      "epoch": 3.6229945355191258,
      "grad_norm": 0.6904444045030841,
      "learning_rate": 4.611818524548028e-06,
      "loss": 11.5172,
      "step": 20719
    },
    {
      "epoch": 3.6231693989071037,
      "grad_norm": 0.7047940581964235,
      "learning_rate": 4.607568631860881e-06,
      "loss": 11.3795,
      "step": 20720
    },
    {
      "epoch": 3.623344262295082,
      "grad_norm": 0.6299025707077313,
      "learning_rate": 4.603320652084886e-06,
      "loss": 11.3374,
      "step": 20721
    },
    {
      "epoch": 3.6235191256830603,
      "grad_norm": 0.6871613597873326,
      "learning_rate": 4.599074585305218e-06,
      "loss": 11.3682,
      "step": 20722
    },
    {
      "epoch": 3.6236939890710382,
      "grad_norm": 0.6940378990150391,
      "learning_rate": 4.594830431607022e-06,
      "loss": 11.1602,
      "step": 20723
    },
    {
      "epoch": 3.623868852459016,
      "grad_norm": 0.6445351599005454,
      "learning_rate": 4.590588191075429e-06,
      "loss": 11.5428,
      "step": 20724
    },
    {
      "epoch": 3.6240437158469945,
      "grad_norm": 0.6643521004015234,
      "learning_rate": 4.586347863795448e-06,
      "loss": 11.5586,
      "step": 20725
    },
    {
      "epoch": 3.624218579234973,
      "grad_norm": 0.7011807347202963,
      "learning_rate": 4.582109449852168e-06,
      "loss": 11.3526,
      "step": 20726
    },
    {
      "epoch": 3.6243934426229507,
      "grad_norm": 0.649990069624239,
      "learning_rate": 4.577872949330575e-06,
      "loss": 11.3288,
      "step": 20727
    },
    {
      "epoch": 3.624568306010929,
      "grad_norm": 0.7003207714366084,
      "learning_rate": 4.5736383623156135e-06,
      "loss": 11.2343,
      "step": 20728
    },
    {
      "epoch": 3.624743169398907,
      "grad_norm": 0.6898313206260414,
      "learning_rate": 4.569405688892181e-06,
      "loss": 11.4123,
      "step": 20729
    },
    {
      "epoch": 3.6249180327868853,
      "grad_norm": 0.6876503887273488,
      "learning_rate": 4.565174929145188e-06,
      "loss": 11.4332,
      "step": 20730
    },
    {
      "epoch": 3.6250928961748636,
      "grad_norm": 0.701757825097562,
      "learning_rate": 4.560946083159468e-06,
      "loss": 11.4709,
      "step": 20731
    },
    {
      "epoch": 3.6252677595628415,
      "grad_norm": 0.7329408970520049,
      "learning_rate": 4.556719151019806e-06,
      "loss": 11.1056,
      "step": 20732
    },
    {
      "epoch": 3.6254426229508194,
      "grad_norm": 0.6027699220629816,
      "learning_rate": 4.552494132810991e-06,
      "loss": 11.2728,
      "step": 20733
    },
    {
      "epoch": 3.6256174863387978,
      "grad_norm": 0.6744998839428789,
      "learning_rate": 4.548271028617701e-06,
      "loss": 11.3297,
      "step": 20734
    },
    {
      "epoch": 3.625792349726776,
      "grad_norm": 0.705892694004161,
      "learning_rate": 4.544049838524666e-06,
      "loss": 11.5908,
      "step": 20735
    },
    {
      "epoch": 3.625967213114754,
      "grad_norm": 0.6461119341255125,
      "learning_rate": 4.53983056261652e-06,
      "loss": 11.3074,
      "step": 20736
    },
    {
      "epoch": 3.6261420765027323,
      "grad_norm": 0.6521828300959853,
      "learning_rate": 4.535613200977895e-06,
      "loss": 11.393,
      "step": 20737
    },
    {
      "epoch": 3.6263169398907102,
      "grad_norm": 0.673548773733516,
      "learning_rate": 4.531397753693312e-06,
      "loss": 11.5905,
      "step": 20738
    },
    {
      "epoch": 3.6264918032786886,
      "grad_norm": 0.6998759426136425,
      "learning_rate": 4.527184220847325e-06,
      "loss": 11.2775,
      "step": 20739
    },
    {
      "epoch": 3.626666666666667,
      "grad_norm": 0.6046305782487543,
      "learning_rate": 4.522972602524445e-06,
      "loss": 11.4028,
      "step": 20740
    },
    {
      "epoch": 3.626841530054645,
      "grad_norm": 0.662372685664302,
      "learning_rate": 4.518762898809125e-06,
      "loss": 11.1902,
      "step": 20741
    },
    {
      "epoch": 3.6270163934426227,
      "grad_norm": 0.6719034404987984,
      "learning_rate": 4.5145551097857785e-06,
      "loss": 11.2023,
      "step": 20742
    },
    {
      "epoch": 3.627191256830601,
      "grad_norm": 0.6455562672779042,
      "learning_rate": 4.510349235538769e-06,
      "loss": 11.323,
      "step": 20743
    },
    {
      "epoch": 3.6273661202185794,
      "grad_norm": 0.7337618128905793,
      "learning_rate": 4.50614527615244e-06,
      "loss": 11.5622,
      "step": 20744
    },
    {
      "epoch": 3.6275409836065573,
      "grad_norm": 0.6948404558011295,
      "learning_rate": 4.5019432317111145e-06,
      "loss": 10.9497,
      "step": 20745
    },
    {
      "epoch": 3.6277158469945356,
      "grad_norm": 0.6982099214175747,
      "learning_rate": 4.497743102299057e-06,
      "loss": 11.2553,
      "step": 20746
    },
    {
      "epoch": 3.6278907103825135,
      "grad_norm": 0.6278991906435065,
      "learning_rate": 4.493544888000467e-06,
      "loss": 11.3721,
      "step": 20747
    },
    {
      "epoch": 3.628065573770492,
      "grad_norm": 0.6748154875835786,
      "learning_rate": 4.489348588899556e-06,
      "loss": 11.1961,
      "step": 20748
    },
    {
      "epoch": 3.62824043715847,
      "grad_norm": 0.7139828589801748,
      "learning_rate": 4.485154205080433e-06,
      "loss": 11.3448,
      "step": 20749
    },
    {
      "epoch": 3.628415300546448,
      "grad_norm": 0.7115683762351399,
      "learning_rate": 4.480961736627254e-06,
      "loss": 11.6347,
      "step": 20750
    },
    {
      "epoch": 3.628590163934426,
      "grad_norm": 0.619634968455174,
      "learning_rate": 4.476771183624096e-06,
      "loss": 11.5924,
      "step": 20751
    },
    {
      "epoch": 3.6287650273224044,
      "grad_norm": 0.6451491762056636,
      "learning_rate": 4.472582546154946e-06,
      "loss": 11.4124,
      "step": 20752
    },
    {
      "epoch": 3.6289398907103827,
      "grad_norm": 0.671290256582847,
      "learning_rate": 4.468395824303817e-06,
      "loss": 11.4462,
      "step": 20753
    },
    {
      "epoch": 3.6291147540983606,
      "grad_norm": 0.7119550896532642,
      "learning_rate": 4.464211018154685e-06,
      "loss": 11.5349,
      "step": 20754
    },
    {
      "epoch": 3.629289617486339,
      "grad_norm": 0.673278125071149,
      "learning_rate": 4.4600281277914715e-06,
      "loss": 11.3621,
      "step": 20755
    },
    {
      "epoch": 3.629464480874317,
      "grad_norm": 1.0239926798964791,
      "learning_rate": 4.455847153298021e-06,
      "loss": 11.532,
      "step": 20756
    },
    {
      "epoch": 3.629639344262295,
      "grad_norm": 0.7068826726340949,
      "learning_rate": 4.451668094758199e-06,
      "loss": 11.3151,
      "step": 20757
    },
    {
      "epoch": 3.629814207650273,
      "grad_norm": 0.671533174848821,
      "learning_rate": 4.447490952255795e-06,
      "loss": 11.3531,
      "step": 20758
    },
    {
      "epoch": 3.6299890710382514,
      "grad_norm": 0.6580954297191504,
      "learning_rate": 4.443315725874586e-06,
      "loss": 11.4896,
      "step": 20759
    },
    {
      "epoch": 3.6301639344262293,
      "grad_norm": 0.6473037907359017,
      "learning_rate": 4.439142415698306e-06,
      "loss": 11.6278,
      "step": 20760
    },
    {
      "epoch": 3.6303387978142077,
      "grad_norm": 0.6717361934575325,
      "learning_rate": 4.434971021810619e-06,
      "loss": 11.3686,
      "step": 20761
    },
    {
      "epoch": 3.630513661202186,
      "grad_norm": 0.6587085252586218,
      "learning_rate": 4.430801544295193e-06,
      "loss": 11.3712,
      "step": 20762
    },
    {
      "epoch": 3.630688524590164,
      "grad_norm": 0.6519171282720092,
      "learning_rate": 4.426633983235618e-06,
      "loss": 11.2225,
      "step": 20763
    },
    {
      "epoch": 3.6308633879781422,
      "grad_norm": 0.7570562563358366,
      "learning_rate": 4.422468338715502e-06,
      "loss": 11.2774,
      "step": 20764
    },
    {
      "epoch": 3.63103825136612,
      "grad_norm": 0.6787394940013362,
      "learning_rate": 4.418304610818347e-06,
      "loss": 11.6258,
      "step": 20765
    },
    {
      "epoch": 3.6312131147540985,
      "grad_norm": 0.6713737548276124,
      "learning_rate": 4.414142799627663e-06,
      "loss": 11.3857,
      "step": 20766
    },
    {
      "epoch": 3.6313879781420764,
      "grad_norm": 0.6968703110429461,
      "learning_rate": 4.409982905226895e-06,
      "loss": 11.3986,
      "step": 20767
    },
    {
      "epoch": 3.6315628415300547,
      "grad_norm": 0.7013329955341935,
      "learning_rate": 4.405824927699453e-06,
      "loss": 11.3378,
      "step": 20768
    },
    {
      "epoch": 3.6317377049180326,
      "grad_norm": 0.7053167536419166,
      "learning_rate": 4.401668867128772e-06,
      "loss": 11.2812,
      "step": 20769
    },
    {
      "epoch": 3.631912568306011,
      "grad_norm": 0.6571383266600088,
      "learning_rate": 4.39751472359814e-06,
      "loss": 11.3204,
      "step": 20770
    },
    {
      "epoch": 3.6320874316939893,
      "grad_norm": 0.7636238058840853,
      "learning_rate": 4.393362497190867e-06,
      "loss": 11.1875,
      "step": 20771
    },
    {
      "epoch": 3.632262295081967,
      "grad_norm": 0.6795182888119572,
      "learning_rate": 4.389212187990233e-06,
      "loss": 11.5717,
      "step": 20772
    },
    {
      "epoch": 3.632437158469945,
      "grad_norm": 0.6696812878075362,
      "learning_rate": 4.385063796079459e-06,
      "loss": 11.4407,
      "step": 20773
    },
    {
      "epoch": 3.6326120218579234,
      "grad_norm": 0.5990241417191977,
      "learning_rate": 4.380917321541733e-06,
      "loss": 11.2141,
      "step": 20774
    },
    {
      "epoch": 3.6327868852459018,
      "grad_norm": 0.706563101009174,
      "learning_rate": 4.3767727644602015e-06,
      "loss": 11.5252,
      "step": 20775
    },
    {
      "epoch": 3.6329617486338797,
      "grad_norm": 0.6413925647166034,
      "learning_rate": 4.372630124917976e-06,
      "loss": 11.1717,
      "step": 20776
    },
    {
      "epoch": 3.633136612021858,
      "grad_norm": 0.7449811467594166,
      "learning_rate": 4.368489402998122e-06,
      "loss": 11.3839,
      "step": 20777
    },
    {
      "epoch": 3.633311475409836,
      "grad_norm": 0.6691039417765128,
      "learning_rate": 4.364350598783706e-06,
      "loss": 11.6486,
      "step": 20778
    },
    {
      "epoch": 3.6334863387978142,
      "grad_norm": 0.6446534006895229,
      "learning_rate": 4.360213712357697e-06,
      "loss": 11.6388,
      "step": 20779
    },
    {
      "epoch": 3.6336612021857926,
      "grad_norm": 0.6564914425414361,
      "learning_rate": 4.356078743803049e-06,
      "loss": 11.6061,
      "step": 20780
    },
    {
      "epoch": 3.6338360655737705,
      "grad_norm": 0.6867012100472881,
      "learning_rate": 4.351945693202686e-06,
      "loss": 11.5495,
      "step": 20781
    },
    {
      "epoch": 3.6340109289617484,
      "grad_norm": 0.7352755117006794,
      "learning_rate": 4.347814560639485e-06,
      "loss": 11.0058,
      "step": 20782
    },
    {
      "epoch": 3.6341857923497267,
      "grad_norm": 0.6844717245482815,
      "learning_rate": 4.343685346196302e-06,
      "loss": 11.506,
      "step": 20783
    },
    {
      "epoch": 3.634360655737705,
      "grad_norm": 0.6201348652959261,
      "learning_rate": 4.339558049955927e-06,
      "loss": 11.4098,
      "step": 20784
    },
    {
      "epoch": 3.634535519125683,
      "grad_norm": 0.716266637192909,
      "learning_rate": 4.335432672001127e-06,
      "loss": 11.6111,
      "step": 20785
    },
    {
      "epoch": 3.6347103825136613,
      "grad_norm": 0.6939991892605448,
      "learning_rate": 4.331309212414636e-06,
      "loss": 11.1394,
      "step": 20786
    },
    {
      "epoch": 3.634885245901639,
      "grad_norm": 0.6609256953292704,
      "learning_rate": 4.327187671279132e-06,
      "loss": 11.3227,
      "step": 20787
    },
    {
      "epoch": 3.6350601092896175,
      "grad_norm": 0.697842696501607,
      "learning_rate": 4.32306804867727e-06,
      "loss": 11.3514,
      "step": 20788
    },
    {
      "epoch": 3.635234972677596,
      "grad_norm": 0.6235493174080282,
      "learning_rate": 4.318950344691663e-06,
      "loss": 11.5242,
      "step": 20789
    },
    {
      "epoch": 3.6354098360655738,
      "grad_norm": 0.7749563805945598,
      "learning_rate": 4.314834559404879e-06,
      "loss": 11.2648,
      "step": 20790
    },
    {
      "epoch": 3.6355846994535517,
      "grad_norm": 0.6956864859279391,
      "learning_rate": 4.310720692899461e-06,
      "loss": 11.4855,
      "step": 20791
    },
    {
      "epoch": 3.63575956284153,
      "grad_norm": 0.6230850638535899,
      "learning_rate": 4.30660874525789e-06,
      "loss": 11.3395,
      "step": 20792
    },
    {
      "epoch": 3.6359344262295084,
      "grad_norm": 0.6756340822364714,
      "learning_rate": 4.3024987165626305e-06,
      "loss": 11.5027,
      "step": 20793
    },
    {
      "epoch": 3.6361092896174863,
      "grad_norm": 0.7038112322274025,
      "learning_rate": 4.298390606896108e-06,
      "loss": 11.186,
      "step": 20794
    },
    {
      "epoch": 3.6362841530054646,
      "grad_norm": 0.6668777094000192,
      "learning_rate": 4.2942844163406884e-06,
      "loss": 11.4852,
      "step": 20795
    },
    {
      "epoch": 3.6364590163934425,
      "grad_norm": 0.7224303320565367,
      "learning_rate": 4.290180144978728e-06,
      "loss": 11.4621,
      "step": 20796
    },
    {
      "epoch": 3.636633879781421,
      "grad_norm": 0.7224453376522942,
      "learning_rate": 4.286077792892518e-06,
      "loss": 11.1315,
      "step": 20797
    },
    {
      "epoch": 3.636808743169399,
      "grad_norm": 0.6530337920068243,
      "learning_rate": 4.281977360164335e-06,
      "loss": 11.167,
      "step": 20798
    },
    {
      "epoch": 3.636983606557377,
      "grad_norm": 0.6430830175855372,
      "learning_rate": 4.277878846876393e-06,
      "loss": 11.2433,
      "step": 20799
    },
    {
      "epoch": 3.637158469945355,
      "grad_norm": 0.6894887482800995,
      "learning_rate": 4.273782253110891e-06,
      "loss": 11.3621,
      "step": 20800
    },
    {
      "epoch": 3.6373333333333333,
      "grad_norm": 0.687479116820304,
      "learning_rate": 4.269687578949966e-06,
      "loss": 11.2431,
      "step": 20801
    },
    {
      "epoch": 3.6375081967213116,
      "grad_norm": 0.6032786928581065,
      "learning_rate": 4.265594824475738e-06,
      "loss": 11.1221,
      "step": 20802
    },
    {
      "epoch": 3.6376830601092895,
      "grad_norm": 0.672464156132955,
      "learning_rate": 4.261503989770277e-06,
      "loss": 11.2326,
      "step": 20803
    },
    {
      "epoch": 3.637857923497268,
      "grad_norm": 0.8508022018498207,
      "learning_rate": 4.257415074915627e-06,
      "loss": 11.5499,
      "step": 20804
    },
    {
      "epoch": 3.638032786885246,
      "grad_norm": 0.623668146471543,
      "learning_rate": 4.253328079993757e-06,
      "loss": 11.506,
      "step": 20805
    },
    {
      "epoch": 3.638207650273224,
      "grad_norm": 0.6903298411343073,
      "learning_rate": 4.249243005086645e-06,
      "loss": 11.5052,
      "step": 20806
    },
    {
      "epoch": 3.638382513661202,
      "grad_norm": 0.6998128491762726,
      "learning_rate": 4.245159850276214e-06,
      "loss": 11.3737,
      "step": 20807
    },
    {
      "epoch": 3.6385573770491804,
      "grad_norm": 0.6783285085912645,
      "learning_rate": 4.241078615644323e-06,
      "loss": 11.1728,
      "step": 20808
    },
    {
      "epoch": 3.6387322404371583,
      "grad_norm": 0.6677917034693784,
      "learning_rate": 4.236999301272826e-06,
      "loss": 11.5722,
      "step": 20809
    },
    {
      "epoch": 3.6389071038251366,
      "grad_norm": 0.816241435701146,
      "learning_rate": 4.232921907243526e-06,
      "loss": 11.3897,
      "step": 20810
    },
    {
      "epoch": 3.639081967213115,
      "grad_norm": 0.7461152453526523,
      "learning_rate": 4.22884643363819e-06,
      "loss": 11.3817,
      "step": 20811
    },
    {
      "epoch": 3.639256830601093,
      "grad_norm": 0.6318761884245515,
      "learning_rate": 4.22477288053853e-06,
      "loss": 11.6127,
      "step": 20812
    },
    {
      "epoch": 3.639431693989071,
      "grad_norm": 0.636284348058208,
      "learning_rate": 4.220701248026248e-06,
      "loss": 11.4014,
      "step": 20813
    },
    {
      "epoch": 3.639606557377049,
      "grad_norm": 0.685793559072118,
      "learning_rate": 4.216631536182991e-06,
      "loss": 11.5869,
      "step": 20814
    },
    {
      "epoch": 3.6397814207650274,
      "grad_norm": 0.670650896267476,
      "learning_rate": 4.21256374509037e-06,
      "loss": 11.4401,
      "step": 20815
    },
    {
      "epoch": 3.6399562841530053,
      "grad_norm": 0.601448081459686,
      "learning_rate": 4.208497874829931e-06,
      "loss": 11.0687,
      "step": 20816
    },
    {
      "epoch": 3.6401311475409837,
      "grad_norm": 0.650199833702703,
      "learning_rate": 4.204433925483231e-06,
      "loss": 11.2386,
      "step": 20817
    },
    {
      "epoch": 3.6403060109289616,
      "grad_norm": 0.7159087872290557,
      "learning_rate": 4.200371897131772e-06,
      "loss": 11.4079,
      "step": 20818
    },
    {
      "epoch": 3.64048087431694,
      "grad_norm": 0.7408480752020133,
      "learning_rate": 4.196311789856999e-06,
      "loss": 11.1399,
      "step": 20819
    },
    {
      "epoch": 3.6406557377049182,
      "grad_norm": 0.6487995522920129,
      "learning_rate": 4.192253603740337e-06,
      "loss": 11.4034,
      "step": 20820
    },
    {
      "epoch": 3.640830601092896,
      "grad_norm": 0.6549980554673935,
      "learning_rate": 4.188197338863142e-06,
      "loss": 11.2805,
      "step": 20821
    },
    {
      "epoch": 3.6410054644808745,
      "grad_norm": 0.7175679055291391,
      "learning_rate": 4.184142995306783e-06,
      "loss": 11.2021,
      "step": 20822
    },
    {
      "epoch": 3.6411803278688524,
      "grad_norm": 0.639481367413216,
      "learning_rate": 4.180090573152551e-06,
      "loss": 11.2298,
      "step": 20823
    },
    {
      "epoch": 3.6413551912568307,
      "grad_norm": 0.7029127369394288,
      "learning_rate": 4.176040072481713e-06,
      "loss": 11.3887,
      "step": 20824
    },
    {
      "epoch": 3.6415300546448086,
      "grad_norm": 0.756604566004702,
      "learning_rate": 4.1719914933754715e-06,
      "loss": 11.2764,
      "step": 20825
    },
    {
      "epoch": 3.641704918032787,
      "grad_norm": 0.6466602069181709,
      "learning_rate": 4.167944835915038e-06,
      "loss": 11.1874,
      "step": 20826
    },
    {
      "epoch": 3.641879781420765,
      "grad_norm": 0.6868758076560182,
      "learning_rate": 4.1639001001815615e-06,
      "loss": 11.4888,
      "step": 20827
    },
    {
      "epoch": 3.642054644808743,
      "grad_norm": 0.6239119984965877,
      "learning_rate": 4.15985728625613e-06,
      "loss": 11.5196,
      "step": 20828
    },
    {
      "epoch": 3.6422295081967215,
      "grad_norm": 0.6572864329753678,
      "learning_rate": 4.155816394219858e-06,
      "loss": 11.5247,
      "step": 20829
    },
    {
      "epoch": 3.6424043715846994,
      "grad_norm": 0.6924945810433394,
      "learning_rate": 4.151777424153714e-06,
      "loss": 11.2533,
      "step": 20830
    },
    {
      "epoch": 3.6425792349726773,
      "grad_norm": 0.6263722804546645,
      "learning_rate": 4.147740376138731e-06,
      "loss": 11.4533,
      "step": 20831
    },
    {
      "epoch": 3.6427540983606557,
      "grad_norm": 0.6932018537088472,
      "learning_rate": 4.143705250255869e-06,
      "loss": 11.2627,
      "step": 20832
    },
    {
      "epoch": 3.642928961748634,
      "grad_norm": 0.7137140763778905,
      "learning_rate": 4.139672046586041e-06,
      "loss": 11.4888,
      "step": 20833
    },
    {
      "epoch": 3.643103825136612,
      "grad_norm": 0.6578612071596761,
      "learning_rate": 4.1356407652101135e-06,
      "loss": 11.4088,
      "step": 20834
    },
    {
      "epoch": 3.6432786885245902,
      "grad_norm": 0.6768892360160587,
      "learning_rate": 4.131611406208913e-06,
      "loss": 11.2815,
      "step": 20835
    },
    {
      "epoch": 3.643453551912568,
      "grad_norm": 0.7327569394650104,
      "learning_rate": 4.127583969663273e-06,
      "loss": 11.3832,
      "step": 20836
    },
    {
      "epoch": 3.6436284153005465,
      "grad_norm": 0.7062431404900957,
      "learning_rate": 4.123558455653931e-06,
      "loss": 11.261,
      "step": 20837
    },
    {
      "epoch": 3.643803278688525,
      "grad_norm": 0.6307974002926322,
      "learning_rate": 4.119534864261643e-06,
      "loss": 11.6222,
      "step": 20838
    },
    {
      "epoch": 3.6439781420765027,
      "grad_norm": 0.6926661059336731,
      "learning_rate": 4.115513195567067e-06,
      "loss": 11.3176,
      "step": 20839
    },
    {
      "epoch": 3.6441530054644806,
      "grad_norm": 0.8058895727415636,
      "learning_rate": 4.111493449650827e-06,
      "loss": 11.3329,
      "step": 20840
    },
    {
      "epoch": 3.644327868852459,
      "grad_norm": 0.6935504553076447,
      "learning_rate": 4.1074756265935824e-06,
      "loss": 11.1473,
      "step": 20841
    },
    {
      "epoch": 3.6445027322404373,
      "grad_norm": 0.6460941479221114,
      "learning_rate": 4.103459726475889e-06,
      "loss": 11.757,
      "step": 20842
    },
    {
      "epoch": 3.644677595628415,
      "grad_norm": 0.6922776624475666,
      "learning_rate": 4.099445749378261e-06,
      "loss": 11.3879,
      "step": 20843
    },
    {
      "epoch": 3.6448524590163935,
      "grad_norm": 0.632945926255722,
      "learning_rate": 4.095433695381201e-06,
      "loss": 11.3971,
      "step": 20844
    },
    {
      "epoch": 3.6450273224043714,
      "grad_norm": 0.7108945798682511,
      "learning_rate": 4.0914235645651445e-06,
      "loss": 11.4788,
      "step": 20845
    },
    {
      "epoch": 3.64520218579235,
      "grad_norm": 0.6400741797079396,
      "learning_rate": 4.087415357010527e-06,
      "loss": 11.339,
      "step": 20846
    },
    {
      "epoch": 3.645377049180328,
      "grad_norm": 0.6658586573986266,
      "learning_rate": 4.0834090727977505e-06,
      "loss": 11.5544,
      "step": 20847
    },
    {
      "epoch": 3.645551912568306,
      "grad_norm": 0.686417278697631,
      "learning_rate": 4.079404712007095e-06,
      "loss": 11.2397,
      "step": 20848
    },
    {
      "epoch": 3.645726775956284,
      "grad_norm": 0.608579619638943,
      "learning_rate": 4.0754022747188866e-06,
      "loss": 11.5621,
      "step": 20849
    },
    {
      "epoch": 3.6459016393442623,
      "grad_norm": 0.6520047867789444,
      "learning_rate": 4.071401761013405e-06,
      "loss": 11.3257,
      "step": 20850
    },
    {
      "epoch": 3.6460765027322406,
      "grad_norm": 0.6521385205776672,
      "learning_rate": 4.067403170970874e-06,
      "loss": 11.3099,
      "step": 20851
    },
    {
      "epoch": 3.6462513661202185,
      "grad_norm": 0.6739403066577654,
      "learning_rate": 4.063406504671441e-06,
      "loss": 11.4103,
      "step": 20852
    },
    {
      "epoch": 3.646426229508197,
      "grad_norm": 0.6460707652535054,
      "learning_rate": 4.0594117621952755e-06,
      "loss": 11.4569,
      "step": 20853
    },
    {
      "epoch": 3.6466010928961747,
      "grad_norm": 0.6232948243249297,
      "learning_rate": 4.055418943622457e-06,
      "loss": 11.758,
      "step": 20854
    },
    {
      "epoch": 3.646775956284153,
      "grad_norm": 0.707211782681084,
      "learning_rate": 4.0514280490331015e-06,
      "loss": 11.3225,
      "step": 20855
    },
    {
      "epoch": 3.6469508196721314,
      "grad_norm": 0.6652771475150435,
      "learning_rate": 4.04743907850722e-06,
      "loss": 11.3242,
      "step": 20856
    },
    {
      "epoch": 3.6471256830601093,
      "grad_norm": 0.6527049541026058,
      "learning_rate": 4.043452032124795e-06,
      "loss": 11.569,
      "step": 20857
    },
    {
      "epoch": 3.647300546448087,
      "grad_norm": 0.6673633030179886,
      "learning_rate": 4.039466909965772e-06,
      "loss": 11.3258,
      "step": 20858
    },
    {
      "epoch": 3.6474754098360656,
      "grad_norm": 0.705817176811131,
      "learning_rate": 4.035483712110067e-06,
      "loss": 11.5262,
      "step": 20859
    },
    {
      "epoch": 3.647650273224044,
      "grad_norm": 0.7276274462017542,
      "learning_rate": 4.031502438637591e-06,
      "loss": 11.5323,
      "step": 20860
    },
    {
      "epoch": 3.647825136612022,
      "grad_norm": 0.6447378229768059,
      "learning_rate": 4.027523089628138e-06,
      "loss": 11.5985,
      "step": 20861
    },
    {
      "epoch": 3.648,
      "grad_norm": 0.6971584928768501,
      "learning_rate": 4.023545665161521e-06,
      "loss": 11.2364,
      "step": 20862
    },
    {
      "epoch": 3.648174863387978,
      "grad_norm": 0.7805641488052942,
      "learning_rate": 4.019570165317499e-06,
      "loss": 11.5572,
      "step": 20863
    },
    {
      "epoch": 3.6483497267759564,
      "grad_norm": 0.6914677216739815,
      "learning_rate": 4.0155965901757854e-06,
      "loss": 11.3775,
      "step": 20864
    },
    {
      "epoch": 3.6485245901639343,
      "grad_norm": 0.7358184124133201,
      "learning_rate": 4.011624939816094e-06,
      "loss": 11.3356,
      "step": 20865
    },
    {
      "epoch": 3.6486994535519126,
      "grad_norm": 0.6344505225176965,
      "learning_rate": 4.007655214318029e-06,
      "loss": 11.4462,
      "step": 20866
    },
    {
      "epoch": 3.6488743169398905,
      "grad_norm": 0.6528814277320709,
      "learning_rate": 4.003687413761214e-06,
      "loss": 11.2637,
      "step": 20867
    },
    {
      "epoch": 3.649049180327869,
      "grad_norm": 0.6220321252477617,
      "learning_rate": 3.999721538225221e-06,
      "loss": 11.5646,
      "step": 20868
    },
    {
      "epoch": 3.649224043715847,
      "grad_norm": 0.7014261773514326,
      "learning_rate": 3.995757587789561e-06,
      "loss": 11.0433,
      "step": 20869
    },
    {
      "epoch": 3.649398907103825,
      "grad_norm": 0.6294734286816739,
      "learning_rate": 3.99179556253374e-06,
      "loss": 11.4514,
      "step": 20870
    },
    {
      "epoch": 3.6495737704918034,
      "grad_norm": 0.6391035392214249,
      "learning_rate": 3.987835462537193e-06,
      "loss": 11.3098,
      "step": 20871
    },
    {
      "epoch": 3.6497486338797813,
      "grad_norm": 0.6538171951131072,
      "learning_rate": 3.983877287879345e-06,
      "loss": 11.534,
      "step": 20872
    },
    {
      "epoch": 3.6499234972677597,
      "grad_norm": 0.6195507824646131,
      "learning_rate": 3.979921038639556e-06,
      "loss": 11.6502,
      "step": 20873
    },
    {
      "epoch": 3.6500983606557376,
      "grad_norm": 0.7262220051017215,
      "learning_rate": 3.975966714897195e-06,
      "loss": 11.3031,
      "step": 20874
    },
    {
      "epoch": 3.650273224043716,
      "grad_norm": 0.7069836217214005,
      "learning_rate": 3.972014316731521e-06,
      "loss": 11.4406,
      "step": 20875
    },
    {
      "epoch": 3.650448087431694,
      "grad_norm": 0.6446928606262956,
      "learning_rate": 3.968063844221803e-06,
      "loss": 11.3023,
      "step": 20876
    },
    {
      "epoch": 3.650622950819672,
      "grad_norm": 0.6797361248419524,
      "learning_rate": 3.964115297447257e-06,
      "loss": 11.5835,
      "step": 20877
    },
    {
      "epoch": 3.6507978142076505,
      "grad_norm": 0.7103599721009771,
      "learning_rate": 3.9601686764870745e-06,
      "loss": 11.255,
      "step": 20878
    },
    {
      "epoch": 3.6509726775956284,
      "grad_norm": 0.696992419409764,
      "learning_rate": 3.956223981420393e-06,
      "loss": 11.4601,
      "step": 20879
    },
    {
      "epoch": 3.6511475409836067,
      "grad_norm": 0.7091393825240324,
      "learning_rate": 3.952281212326314e-06,
      "loss": 11.517,
      "step": 20880
    },
    {
      "epoch": 3.6513224043715846,
      "grad_norm": 0.6905131484461887,
      "learning_rate": 3.9483403692838985e-06,
      "loss": 11.2758,
      "step": 20881
    },
    {
      "epoch": 3.651497267759563,
      "grad_norm": 0.66214453714463,
      "learning_rate": 3.944401452372182e-06,
      "loss": 11.5161,
      "step": 20882
    },
    {
      "epoch": 3.651672131147541,
      "grad_norm": 0.6807560263886443,
      "learning_rate": 3.940464461670135e-06,
      "loss": 11.41,
      "step": 20883
    },
    {
      "epoch": 3.651846994535519,
      "grad_norm": 0.6816333034007737,
      "learning_rate": 3.936529397256727e-06,
      "loss": 11.5134,
      "step": 20884
    },
    {
      "epoch": 3.652021857923497,
      "grad_norm": 0.7151038516414591,
      "learning_rate": 3.932596259210852e-06,
      "loss": 11.3831,
      "step": 20885
    },
    {
      "epoch": 3.6521967213114754,
      "grad_norm": 0.6546403095718187,
      "learning_rate": 3.928665047611391e-06,
      "loss": 11.456,
      "step": 20886
    },
    {
      "epoch": 3.6523715846994538,
      "grad_norm": 0.7090582692531003,
      "learning_rate": 3.924735762537169e-06,
      "loss": 11.3627,
      "step": 20887
    },
    {
      "epoch": 3.6525464480874317,
      "grad_norm": 0.6909998337127563,
      "learning_rate": 3.920808404066989e-06,
      "loss": 11.4448,
      "step": 20888
    },
    {
      "epoch": 3.6527213114754096,
      "grad_norm": 0.8896613572321257,
      "learning_rate": 3.916882972279601e-06,
      "loss": 11.3795,
      "step": 20889
    },
    {
      "epoch": 3.652896174863388,
      "grad_norm": 0.6607172915864689,
      "learning_rate": 3.912959467253719e-06,
      "loss": 11.5902,
      "step": 20890
    },
    {
      "epoch": 3.6530710382513663,
      "grad_norm": 0.7155200258122307,
      "learning_rate": 3.909037889068034e-06,
      "loss": 11.4123,
      "step": 20891
    },
    {
      "epoch": 3.653245901639344,
      "grad_norm": 0.6428273280729313,
      "learning_rate": 3.9051182378011755e-06,
      "loss": 11.3358,
      "step": 20892
    },
    {
      "epoch": 3.6534207650273225,
      "grad_norm": 0.6321936500191302,
      "learning_rate": 3.901200513531744e-06,
      "loss": 11.3325,
      "step": 20893
    },
    {
      "epoch": 3.6535956284153004,
      "grad_norm": 0.6833987878436308,
      "learning_rate": 3.897284716338301e-06,
      "loss": 11.3864,
      "step": 20894
    },
    {
      "epoch": 3.6537704918032787,
      "grad_norm": 0.6113707635858608,
      "learning_rate": 3.893370846299371e-06,
      "loss": 11.4484,
      "step": 20895
    },
    {
      "epoch": 3.653945355191257,
      "grad_norm": 0.6992157309446646,
      "learning_rate": 3.889458903493448e-06,
      "loss": 11.4188,
      "step": 20896
    },
    {
      "epoch": 3.654120218579235,
      "grad_norm": 0.6256130957779557,
      "learning_rate": 3.88554888799898e-06,
      "loss": 11.2141,
      "step": 20897
    },
    {
      "epoch": 3.654295081967213,
      "grad_norm": 0.6469462166248411,
      "learning_rate": 3.88164079989436e-06,
      "loss": 11.1711,
      "step": 20898
    },
    {
      "epoch": 3.654469945355191,
      "grad_norm": 0.7522022849160256,
      "learning_rate": 3.877734639257957e-06,
      "loss": 11.212,
      "step": 20899
    },
    {
      "epoch": 3.6546448087431695,
      "grad_norm": 0.7343443618809771,
      "learning_rate": 3.873830406168111e-06,
      "loss": 11.067,
      "step": 20900
    },
    {
      "epoch": 3.6548196721311474,
      "grad_norm": 0.651984991576274,
      "learning_rate": 3.8699281007031245e-06,
      "loss": 11.1581,
      "step": 20901
    },
    {
      "epoch": 3.654994535519126,
      "grad_norm": 0.6572907084034294,
      "learning_rate": 3.866027722941223e-06,
      "loss": 11.2756,
      "step": 20902
    },
    {
      "epoch": 3.6551693989071037,
      "grad_norm": 0.6380555950731281,
      "learning_rate": 3.8621292729606355e-06,
      "loss": 11.5648,
      "step": 20903
    },
    {
      "epoch": 3.655344262295082,
      "grad_norm": 0.6969064447252765,
      "learning_rate": 3.858232750839552e-06,
      "loss": 11.4764,
      "step": 20904
    },
    {
      "epoch": 3.6555191256830604,
      "grad_norm": 0.7142410454378542,
      "learning_rate": 3.854338156656079e-06,
      "loss": 11.3503,
      "step": 20905
    },
    {
      "epoch": 3.6556939890710383,
      "grad_norm": 0.6887378391597841,
      "learning_rate": 3.850445490488342e-06,
      "loss": 11.2571,
      "step": 20906
    },
    {
      "epoch": 3.655868852459016,
      "grad_norm": 0.6681654377786864,
      "learning_rate": 3.846554752414389e-06,
      "loss": 11.3737,
      "step": 20907
    },
    {
      "epoch": 3.6560437158469945,
      "grad_norm": 0.6687786475779376,
      "learning_rate": 3.8426659425122465e-06,
      "loss": 11.4944,
      "step": 20908
    },
    {
      "epoch": 3.656218579234973,
      "grad_norm": 0.7587416514523466,
      "learning_rate": 3.838779060859887e-06,
      "loss": 11.2919,
      "step": 20909
    },
    {
      "epoch": 3.6563934426229507,
      "grad_norm": 0.6382726333410222,
      "learning_rate": 3.834894107535269e-06,
      "loss": 11.2838,
      "step": 20910
    },
    {
      "epoch": 3.656568306010929,
      "grad_norm": 0.7194098568492774,
      "learning_rate": 3.831011082616298e-06,
      "loss": 11.4262,
      "step": 20911
    },
    {
      "epoch": 3.656743169398907,
      "grad_norm": 0.662339772130628,
      "learning_rate": 3.8271299861807995e-06,
      "loss": 11.3557,
      "step": 20912
    },
    {
      "epoch": 3.6569180327868853,
      "grad_norm": 0.6869307439982283,
      "learning_rate": 3.823250818306656e-06,
      "loss": 11.5892,
      "step": 20913
    },
    {
      "epoch": 3.6570928961748637,
      "grad_norm": 0.6255588241139141,
      "learning_rate": 3.819373579071639e-06,
      "loss": 11.2459,
      "step": 20914
    },
    {
      "epoch": 3.6572677595628416,
      "grad_norm": 0.6758782480349076,
      "learning_rate": 3.8154982685534854e-06,
      "loss": 11.5575,
      "step": 20915
    },
    {
      "epoch": 3.6574426229508195,
      "grad_norm": 0.6183259425633261,
      "learning_rate": 3.8116248868299342e-06,
      "loss": 11.3821,
      "step": 20916
    },
    {
      "epoch": 3.657617486338798,
      "grad_norm": 0.6682380635285037,
      "learning_rate": 3.807753433978611e-06,
      "loss": 11.4974,
      "step": 20917
    },
    {
      "epoch": 3.657792349726776,
      "grad_norm": 0.7036111925844103,
      "learning_rate": 3.8038839100771994e-06,
      "loss": 11.4491,
      "step": 20918
    },
    {
      "epoch": 3.657967213114754,
      "grad_norm": 0.6780068132131756,
      "learning_rate": 3.8000163152032697e-06,
      "loss": 11.3378,
      "step": 20919
    },
    {
      "epoch": 3.6581420765027324,
      "grad_norm": 0.6100455529393897,
      "learning_rate": 3.7961506494344044e-06,
      "loss": 11.3173,
      "step": 20920
    },
    {
      "epoch": 3.6583169398907103,
      "grad_norm": 0.7570350868146638,
      "learning_rate": 3.7922869128480753e-06,
      "loss": 11.5894,
      "step": 20921
    },
    {
      "epoch": 3.6584918032786886,
      "grad_norm": 0.6587474173371123,
      "learning_rate": 3.788425105521809e-06,
      "loss": 11.5759,
      "step": 20922
    },
    {
      "epoch": 3.6586666666666665,
      "grad_norm": 0.6666818868148977,
      "learning_rate": 3.7845652275330212e-06,
      "loss": 11.5971,
      "step": 20923
    },
    {
      "epoch": 3.658841530054645,
      "grad_norm": 0.6484563487683418,
      "learning_rate": 3.780707278959117e-06,
      "loss": 11.552,
      "step": 20924
    },
    {
      "epoch": 3.6590163934426227,
      "grad_norm": 0.7113873099102809,
      "learning_rate": 3.7768512598774784e-06,
      "loss": 11.4421,
      "step": 20925
    },
    {
      "epoch": 3.659191256830601,
      "grad_norm": 0.6521231462210688,
      "learning_rate": 3.7729971703654e-06,
      "loss": 11.5626,
      "step": 20926
    },
    {
      "epoch": 3.6593661202185794,
      "grad_norm": 0.6302550632612254,
      "learning_rate": 3.7691450105001857e-06,
      "loss": 11.4072,
      "step": 20927
    },
    {
      "epoch": 3.6595409836065573,
      "grad_norm": 0.7083936757612899,
      "learning_rate": 3.7652947803590855e-06,
      "loss": 11.4524,
      "step": 20928
    },
    {
      "epoch": 3.6597158469945357,
      "grad_norm": 0.6741547171487058,
      "learning_rate": 3.761446480019315e-06,
      "loss": 11.4449,
      "step": 20929
    },
    {
      "epoch": 3.6598907103825136,
      "grad_norm": 0.6828474213330419,
      "learning_rate": 3.7576001095580237e-06,
      "loss": 11.4834,
      "step": 20930
    },
    {
      "epoch": 3.660065573770492,
      "grad_norm": 0.7059022792739508,
      "learning_rate": 3.753755669052339e-06,
      "loss": 11.19,
      "step": 20931
    },
    {
      "epoch": 3.66024043715847,
      "grad_norm": 0.6910148222444269,
      "learning_rate": 3.749913158579388e-06,
      "loss": 11.3045,
      "step": 20932
    },
    {
      "epoch": 3.660415300546448,
      "grad_norm": 0.7249700689148786,
      "learning_rate": 3.746072578216198e-06,
      "loss": 11.0858,
      "step": 20933
    },
    {
      "epoch": 3.660590163934426,
      "grad_norm": 0.7174801004060507,
      "learning_rate": 3.7422339280397955e-06,
      "loss": 11.3974,
      "step": 20934
    },
    {
      "epoch": 3.6607650273224044,
      "grad_norm": 0.6583767655587731,
      "learning_rate": 3.7383972081271533e-06,
      "loss": 11.6342,
      "step": 20935
    },
    {
      "epoch": 3.6609398907103827,
      "grad_norm": 0.7112049870289096,
      "learning_rate": 3.734562418555199e-06,
      "loss": 11.3134,
      "step": 20936
    },
    {
      "epoch": 3.6611147540983606,
      "grad_norm": 0.6612219080524708,
      "learning_rate": 3.7307295594008472e-06,
      "loss": 11.61,
      "step": 20937
    },
    {
      "epoch": 3.6612896174863385,
      "grad_norm": 0.7282972466084265,
      "learning_rate": 3.7268986307409604e-06,
      "loss": 11.2006,
      "step": 20938
    },
    {
      "epoch": 3.661464480874317,
      "grad_norm": 0.6312481666026566,
      "learning_rate": 3.7230696326523428e-06,
      "loss": 11.3895,
      "step": 20939
    },
    {
      "epoch": 3.661639344262295,
      "grad_norm": 0.6649954354710575,
      "learning_rate": 3.7192425652118e-06,
      "loss": 11.2571,
      "step": 20940
    },
    {
      "epoch": 3.661814207650273,
      "grad_norm": 0.6737734646184979,
      "learning_rate": 3.7154174284960375e-06,
      "loss": 11.6564,
      "step": 20941
    },
    {
      "epoch": 3.6619890710382514,
      "grad_norm": 0.6384686183073548,
      "learning_rate": 3.7115942225818047e-06,
      "loss": 11.4315,
      "step": 20942
    },
    {
      "epoch": 3.6621639344262293,
      "grad_norm": 0.6269132166337009,
      "learning_rate": 3.7077729475457735e-06,
      "loss": 11.2402,
      "step": 20943
    },
    {
      "epoch": 3.6623387978142077,
      "grad_norm": 0.7210605093096362,
      "learning_rate": 3.7039536034645272e-06,
      "loss": 11.3077,
      "step": 20944
    },
    {
      "epoch": 3.662513661202186,
      "grad_norm": 0.7290590148249799,
      "learning_rate": 3.7001361904146715e-06,
      "loss": 11.3563,
      "step": 20945
    },
    {
      "epoch": 3.662688524590164,
      "grad_norm": 0.6966561556594887,
      "learning_rate": 3.696320708472778e-06,
      "loss": 11.2523,
      "step": 20946
    },
    {
      "epoch": 3.662863387978142,
      "grad_norm": 0.6743437663047507,
      "learning_rate": 3.6925071577153526e-06,
      "loss": 11.1435,
      "step": 20947
    },
    {
      "epoch": 3.66303825136612,
      "grad_norm": 0.6687508132015024,
      "learning_rate": 3.688695538218856e-06,
      "loss": 11.4967,
      "step": 20948
    },
    {
      "epoch": 3.6632131147540985,
      "grad_norm": 0.6862349697151745,
      "learning_rate": 3.684885850059727e-06,
      "loss": 11.254,
      "step": 20949
    },
    {
      "epoch": 3.6633879781420764,
      "grad_norm": 0.7120197279376641,
      "learning_rate": 3.6810780933143384e-06,
      "loss": 11.3415,
      "step": 20950
    },
    {
      "epoch": 3.6635628415300547,
      "grad_norm": 0.7129331480606172,
      "learning_rate": 3.6772722680590955e-06,
      "loss": 11.0432,
      "step": 20951
    },
    {
      "epoch": 3.6637377049180326,
      "grad_norm": 0.6928402392494882,
      "learning_rate": 3.6734683743703036e-06,
      "loss": 11.5848,
      "step": 20952
    },
    {
      "epoch": 3.663912568306011,
      "grad_norm": 0.6541322140761556,
      "learning_rate": 3.6696664123242242e-06,
      "loss": 11.3576,
      "step": 20953
    },
    {
      "epoch": 3.6640874316939893,
      "grad_norm": 0.6458894564155891,
      "learning_rate": 3.665866381997107e-06,
      "loss": 11.5959,
      "step": 20954
    },
    {
      "epoch": 3.664262295081967,
      "grad_norm": 0.6456993458226626,
      "learning_rate": 3.6620682834651366e-06,
      "loss": 11.4173,
      "step": 20955
    },
    {
      "epoch": 3.664437158469945,
      "grad_norm": 0.6791906276417822,
      "learning_rate": 3.6582721168045177e-06,
      "loss": 11.035,
      "step": 20956
    },
    {
      "epoch": 3.6646120218579235,
      "grad_norm": 0.6714918009747871,
      "learning_rate": 3.654477882091334e-06,
      "loss": 11.4572,
      "step": 20957
    },
    {
      "epoch": 3.664786885245902,
      "grad_norm": 0.6990833932646643,
      "learning_rate": 3.6506855794016913e-06,
      "loss": 11.3339,
      "step": 20958
    },
    {
      "epoch": 3.6649617486338797,
      "grad_norm": 0.7074092956593268,
      "learning_rate": 3.6468952088116403e-06,
      "loss": 11.4131,
      "step": 20959
    },
    {
      "epoch": 3.665136612021858,
      "grad_norm": 0.6853441780058104,
      "learning_rate": 3.6431067703971645e-06,
      "loss": 11.5417,
      "step": 20960
    },
    {
      "epoch": 3.665311475409836,
      "grad_norm": 0.6678603772103399,
      "learning_rate": 3.6393202642342807e-06,
      "loss": 11.6208,
      "step": 20961
    },
    {
      "epoch": 3.6654863387978143,
      "grad_norm": 0.7252431881498038,
      "learning_rate": 3.6355356903988723e-06,
      "loss": 11.4378,
      "step": 20962
    },
    {
      "epoch": 3.6656612021857926,
      "grad_norm": 0.6542891964894296,
      "learning_rate": 3.6317530489668572e-06,
      "loss": 11.4348,
      "step": 20963
    },
    {
      "epoch": 3.6658360655737705,
      "grad_norm": 0.6683760897960844,
      "learning_rate": 3.627972340014085e-06,
      "loss": 11.3225,
      "step": 20964
    },
    {
      "epoch": 3.6660109289617484,
      "grad_norm": 0.733637831441806,
      "learning_rate": 3.6241935636163626e-06,
      "loss": 11.3516,
      "step": 20965
    },
    {
      "epoch": 3.6661857923497267,
      "grad_norm": 0.6427191406086151,
      "learning_rate": 3.620416719849484e-06,
      "loss": 11.5464,
      "step": 20966
    },
    {
      "epoch": 3.666360655737705,
      "grad_norm": 0.7557571048761749,
      "learning_rate": 3.6166418087891563e-06,
      "loss": 11.2679,
      "step": 20967
    },
    {
      "epoch": 3.666535519125683,
      "grad_norm": 0.6861720196976475,
      "learning_rate": 3.612868830511107e-06,
      "loss": 11.4964,
      "step": 20968
    },
    {
      "epoch": 3.6667103825136613,
      "grad_norm": 0.7010825331317975,
      "learning_rate": 3.609097785090976e-06,
      "loss": 11.4696,
      "step": 20969
    },
    {
      "epoch": 3.6668852459016392,
      "grad_norm": 0.683456347744738,
      "learning_rate": 3.6053286726044133e-06,
      "loss": 11.3886,
      "step": 20970
    },
    {
      "epoch": 3.6670601092896176,
      "grad_norm": 0.6696084346669436,
      "learning_rate": 3.60156149312697e-06,
      "loss": 11.3915,
      "step": 20971
    },
    {
      "epoch": 3.667234972677596,
      "grad_norm": 0.6550239102052222,
      "learning_rate": 3.5977962467341974e-06,
      "loss": 11.54,
      "step": 20972
    },
    {
      "epoch": 3.667409836065574,
      "grad_norm": 0.684591954275792,
      "learning_rate": 3.594032933501601e-06,
      "loss": 11.3056,
      "step": 20973
    },
    {
      "epoch": 3.6675846994535517,
      "grad_norm": 0.6814278746820239,
      "learning_rate": 3.590271553504654e-06,
      "loss": 11.4504,
      "step": 20974
    },
    {
      "epoch": 3.66775956284153,
      "grad_norm": 0.6679711518558635,
      "learning_rate": 3.5865121068187848e-06,
      "loss": 11.304,
      "step": 20975
    },
    {
      "epoch": 3.6679344262295084,
      "grad_norm": 0.6507704085423326,
      "learning_rate": 3.5827545935193665e-06,
      "loss": 11.5037,
      "step": 20976
    },
    {
      "epoch": 3.6681092896174863,
      "grad_norm": 0.6562937492087023,
      "learning_rate": 3.578999013681761e-06,
      "loss": 11.5071,
      "step": 20977
    },
    {
      "epoch": 3.6682841530054646,
      "grad_norm": 0.6319747985666123,
      "learning_rate": 3.5752453673812745e-06,
      "loss": 11.5674,
      "step": 20978
    },
    {
      "epoch": 3.6684590163934425,
      "grad_norm": 0.753607161953644,
      "learning_rate": 3.571493654693192e-06,
      "loss": 11.2684,
      "step": 20979
    },
    {
      "epoch": 3.668633879781421,
      "grad_norm": 0.6596135957652408,
      "learning_rate": 3.5677438756927304e-06,
      "loss": 11.3747,
      "step": 20980
    },
    {
      "epoch": 3.6688087431693988,
      "grad_norm": 0.7307346454821029,
      "learning_rate": 3.5639960304550857e-06,
      "loss": 10.9555,
      "step": 20981
    },
    {
      "epoch": 3.668983606557377,
      "grad_norm": 0.7054083709068519,
      "learning_rate": 3.5602501190554193e-06,
      "loss": 11.5588,
      "step": 20982
    },
    {
      "epoch": 3.669158469945355,
      "grad_norm": 0.6697378313326537,
      "learning_rate": 3.5565061415688494e-06,
      "loss": 11.6176,
      "step": 20983
    },
    {
      "epoch": 3.6693333333333333,
      "grad_norm": 0.6712023222789748,
      "learning_rate": 3.552764098070449e-06,
      "loss": 11.062,
      "step": 20984
    },
    {
      "epoch": 3.6695081967213117,
      "grad_norm": 0.7031598086995297,
      "learning_rate": 3.549023988635258e-06,
      "loss": 11.4854,
      "step": 20985
    },
    {
      "epoch": 3.6696830601092896,
      "grad_norm": 0.6609626390748053,
      "learning_rate": 3.5452858133382836e-06,
      "loss": 11.4973,
      "step": 20986
    },
    {
      "epoch": 3.669857923497268,
      "grad_norm": 0.9288162588343111,
      "learning_rate": 3.541549572254488e-06,
      "loss": 11.4605,
      "step": 20987
    },
    {
      "epoch": 3.670032786885246,
      "grad_norm": 0.5966254056225466,
      "learning_rate": 3.5378152654587882e-06,
      "loss": 11.6187,
      "step": 20988
    },
    {
      "epoch": 3.670207650273224,
      "grad_norm": 0.6828502892166883,
      "learning_rate": 3.5340828930260694e-06,
      "loss": 11.2353,
      "step": 20989
    },
    {
      "epoch": 3.670382513661202,
      "grad_norm": 0.7011214874244862,
      "learning_rate": 3.5303524550311718e-06,
      "loss": 11.5043,
      "step": 20990
    },
    {
      "epoch": 3.6705573770491804,
      "grad_norm": 0.6391511701286228,
      "learning_rate": 3.526623951548913e-06,
      "loss": 11.5401,
      "step": 20991
    },
    {
      "epoch": 3.6707322404371583,
      "grad_norm": 0.6818977186449856,
      "learning_rate": 3.5228973826540558e-06,
      "loss": 11.2919,
      "step": 20992
    },
    {
      "epoch": 3.6709071038251366,
      "grad_norm": 0.68809176343445,
      "learning_rate": 3.519172748421329e-06,
      "loss": 11.2362,
      "step": 20993
    },
    {
      "epoch": 3.671081967213115,
      "grad_norm": 0.696192473949071,
      "learning_rate": 3.515450048925417e-06,
      "loss": 11.3424,
      "step": 20994
    },
    {
      "epoch": 3.671256830601093,
      "grad_norm": 0.6900657496050453,
      "learning_rate": 3.5117292842409945e-06,
      "loss": 11.3906,
      "step": 20995
    },
    {
      "epoch": 3.6714316939890708,
      "grad_norm": 0.7492661449899339,
      "learning_rate": 3.508010454442645e-06,
      "loss": 11.4742,
      "step": 20996
    },
    {
      "epoch": 3.671606557377049,
      "grad_norm": 0.6369191317865015,
      "learning_rate": 3.5042935596049544e-06,
      "loss": 11.5629,
      "step": 20997
    },
    {
      "epoch": 3.6717814207650274,
      "grad_norm": 0.6537267853457516,
      "learning_rate": 3.500578599802462e-06,
      "loss": 11.5076,
      "step": 20998
    },
    {
      "epoch": 3.6719562841530053,
      "grad_norm": 0.6609299257641115,
      "learning_rate": 3.4968655751096646e-06,
      "loss": 11.2984,
      "step": 20999
    },
    {
      "epoch": 3.6721311475409837,
      "grad_norm": 0.6141062803016328,
      "learning_rate": 3.4931544856010133e-06,
      "loss": 11.4852,
      "step": 21000
    },
    {
      "epoch": 3.6723060109289616,
      "grad_norm": 0.6641355775691782,
      "learning_rate": 3.489445331350927e-06,
      "loss": 11.2443,
      "step": 21001
    },
    {
      "epoch": 3.67248087431694,
      "grad_norm": 0.6410474003586608,
      "learning_rate": 3.485738112433801e-06,
      "loss": 11.4764,
      "step": 21002
    },
    {
      "epoch": 3.6726557377049183,
      "grad_norm": 0.6805826679384273,
      "learning_rate": 3.482032828923942e-06,
      "loss": 11.2628,
      "step": 21003
    },
    {
      "epoch": 3.672830601092896,
      "grad_norm": 0.709421499982236,
      "learning_rate": 3.478329480895681e-06,
      "loss": 11.2086,
      "step": 21004
    },
    {
      "epoch": 3.673005464480874,
      "grad_norm": 0.6554329486047368,
      "learning_rate": 3.47462806842328e-06,
      "loss": 11.5025,
      "step": 21005
    },
    {
      "epoch": 3.6731803278688524,
      "grad_norm": 0.6512676277271813,
      "learning_rate": 3.4709285915809576e-06,
      "loss": 11.1757,
      "step": 21006
    },
    {
      "epoch": 3.6733551912568307,
      "grad_norm": 0.6296717116354549,
      "learning_rate": 3.4672310504429095e-06,
      "loss": 11.3291,
      "step": 21007
    },
    {
      "epoch": 3.6735300546448086,
      "grad_norm": 0.7144632723043483,
      "learning_rate": 3.4635354450832545e-06,
      "loss": 11.6213,
      "step": 21008
    },
    {
      "epoch": 3.673704918032787,
      "grad_norm": 0.6971950192196741,
      "learning_rate": 3.4598417755761225e-06,
      "loss": 11.5,
      "step": 21009
    },
    {
      "epoch": 3.673879781420765,
      "grad_norm": 0.6566753539384009,
      "learning_rate": 3.4561500419955873e-06,
      "loss": 11.423,
      "step": 21010
    },
    {
      "epoch": 3.674054644808743,
      "grad_norm": 0.6520950822839582,
      "learning_rate": 3.4524602444156782e-06,
      "loss": 11.6211,
      "step": 21011
    },
    {
      "epoch": 3.6742295081967216,
      "grad_norm": 0.681590218592105,
      "learning_rate": 3.44877238291037e-06,
      "loss": 11.4463,
      "step": 21012
    },
    {
      "epoch": 3.6744043715846995,
      "grad_norm": 0.6332945991504851,
      "learning_rate": 3.4450864575536256e-06,
      "loss": 11.2253,
      "step": 21013
    },
    {
      "epoch": 3.6745792349726774,
      "grad_norm": 0.6615590299548335,
      "learning_rate": 3.4414024684193637e-06,
      "loss": 11.5381,
      "step": 21014
    },
    {
      "epoch": 3.6747540983606557,
      "grad_norm": 0.6783473502007087,
      "learning_rate": 3.437720415581458e-06,
      "loss": 11.5227,
      "step": 21015
    },
    {
      "epoch": 3.674928961748634,
      "grad_norm": 0.6882384161563448,
      "learning_rate": 3.43404029911375e-06,
      "loss": 11.136,
      "step": 21016
    },
    {
      "epoch": 3.675103825136612,
      "grad_norm": 0.6419456124495331,
      "learning_rate": 3.430362119090014e-06,
      "loss": 11.5115,
      "step": 21017
    },
    {
      "epoch": 3.6752786885245903,
      "grad_norm": 0.6529828765427208,
      "learning_rate": 3.4266858755840346e-06,
      "loss": 11.3143,
      "step": 21018
    },
    {
      "epoch": 3.675453551912568,
      "grad_norm": 0.7268513235296182,
      "learning_rate": 3.423011568669521e-06,
      "loss": 11.3547,
      "step": 21019
    },
    {
      "epoch": 3.6756284153005465,
      "grad_norm": 0.6837848767193596,
      "learning_rate": 3.4193391984201683e-06,
      "loss": 11.2093,
      "step": 21020
    },
    {
      "epoch": 3.675803278688525,
      "grad_norm": 0.6768924175898529,
      "learning_rate": 3.4156687649095854e-06,
      "loss": 11.3115,
      "step": 21021
    },
    {
      "epoch": 3.6759781420765028,
      "grad_norm": 0.6885282214785705,
      "learning_rate": 3.4120002682114014e-06,
      "loss": 11.5122,
      "step": 21022
    },
    {
      "epoch": 3.6761530054644806,
      "grad_norm": 0.6481689743185431,
      "learning_rate": 3.4083337083991807e-06,
      "loss": 11.4354,
      "step": 21023
    },
    {
      "epoch": 3.676327868852459,
      "grad_norm": 0.696891610927649,
      "learning_rate": 3.4046690855464415e-06,
      "loss": 11.5569,
      "step": 21024
    },
    {
      "epoch": 3.6765027322404373,
      "grad_norm": 0.6390416259645292,
      "learning_rate": 3.4010063997266807e-06,
      "loss": 11.5435,
      "step": 21025
    },
    {
      "epoch": 3.6766775956284152,
      "grad_norm": 0.6721162633333773,
      "learning_rate": 3.3973456510133393e-06,
      "loss": 11.2236,
      "step": 21026
    },
    {
      "epoch": 3.6768524590163936,
      "grad_norm": 0.6997837757206053,
      "learning_rate": 3.393686839479815e-06,
      "loss": 11.0989,
      "step": 21027
    },
    {
      "epoch": 3.6770273224043715,
      "grad_norm": 0.6976284820919382,
      "learning_rate": 3.3900299651994927e-06,
      "loss": 11.2471,
      "step": 21028
    },
    {
      "epoch": 3.67720218579235,
      "grad_norm": 0.6232158014857037,
      "learning_rate": 3.386375028245714e-06,
      "loss": 11.5011,
      "step": 21029
    },
    {
      "epoch": 3.6773770491803277,
      "grad_norm": 0.6447997809693466,
      "learning_rate": 3.3827220286917537e-06,
      "loss": 11.257,
      "step": 21030
    },
    {
      "epoch": 3.677551912568306,
      "grad_norm": 0.6713574368811625,
      "learning_rate": 3.3790709666108644e-06,
      "loss": 11.2606,
      "step": 21031
    },
    {
      "epoch": 3.677726775956284,
      "grad_norm": 0.681920654851512,
      "learning_rate": 3.3754218420762653e-06,
      "loss": 11.1979,
      "step": 21032
    },
    {
      "epoch": 3.6779016393442623,
      "grad_norm": 0.7402448776252443,
      "learning_rate": 3.371774655161142e-06,
      "loss": 11.0965,
      "step": 21033
    },
    {
      "epoch": 3.6780765027322406,
      "grad_norm": 0.6662619117531948,
      "learning_rate": 3.368129405938625e-06,
      "loss": 11.5517,
      "step": 21034
    },
    {
      "epoch": 3.6782513661202185,
      "grad_norm": 0.7038529075999956,
      "learning_rate": 3.3644860944818113e-06,
      "loss": 11.3798,
      "step": 21035
    },
    {
      "epoch": 3.678426229508197,
      "grad_norm": 0.7034190366179605,
      "learning_rate": 3.360844720863765e-06,
      "loss": 11.3581,
      "step": 21036
    },
    {
      "epoch": 3.6786010928961748,
      "grad_norm": 0.6164860491123495,
      "learning_rate": 3.357205285157483e-06,
      "loss": 11.4379,
      "step": 21037
    },
    {
      "epoch": 3.678775956284153,
      "grad_norm": 0.6648450611498081,
      "learning_rate": 3.353567787435985e-06,
      "loss": 11.2577,
      "step": 21038
    },
    {
      "epoch": 3.678950819672131,
      "grad_norm": 0.6987184649035336,
      "learning_rate": 3.349932227772201e-06,
      "loss": 11.2339,
      "step": 21039
    },
    {
      "epoch": 3.6791256830601093,
      "grad_norm": 0.6147068919966009,
      "learning_rate": 3.3462986062390287e-06,
      "loss": 11.1218,
      "step": 21040
    },
    {
      "epoch": 3.6793005464480872,
      "grad_norm": 0.6883096687138147,
      "learning_rate": 3.34266692290931e-06,
      "loss": 11.5078,
      "step": 21041
    },
    {
      "epoch": 3.6794754098360656,
      "grad_norm": 0.6259967860809464,
      "learning_rate": 3.3390371778559194e-06,
      "loss": 11.2537,
      "step": 21042
    },
    {
      "epoch": 3.679650273224044,
      "grad_norm": 0.6748685139029067,
      "learning_rate": 3.335409371151632e-06,
      "loss": 11.5281,
      "step": 21043
    },
    {
      "epoch": 3.679825136612022,
      "grad_norm": 0.7328205528610405,
      "learning_rate": 3.331783502869168e-06,
      "loss": 11.0263,
      "step": 21044
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.6319232028573011,
      "learning_rate": 3.3281595730812575e-06,
      "loss": 11.4967,
      "step": 21045
    },
    {
      "epoch": 3.680174863387978,
      "grad_norm": 0.7284845356688041,
      "learning_rate": 3.324537581860565e-06,
      "loss": 11.5098,
      "step": 21046
    },
    {
      "epoch": 3.6803497267759564,
      "grad_norm": 0.7205846322297585,
      "learning_rate": 3.3209175292797324e-06,
      "loss": 11.1909,
      "step": 21047
    },
    {
      "epoch": 3.6805245901639343,
      "grad_norm": 0.6582828559956249,
      "learning_rate": 3.317299415411357e-06,
      "loss": 11.3867,
      "step": 21048
    },
    {
      "epoch": 3.6806994535519126,
      "grad_norm": 0.680162904585262,
      "learning_rate": 3.3136832403279694e-06,
      "loss": 11.2102,
      "step": 21049
    },
    {
      "epoch": 3.6808743169398905,
      "grad_norm": 0.703174960288891,
      "learning_rate": 3.310069004102112e-06,
      "loss": 11.3177,
      "step": 21050
    },
    {
      "epoch": 3.681049180327869,
      "grad_norm": 0.7048172927558982,
      "learning_rate": 3.306456706806227e-06,
      "loss": 11.3541,
      "step": 21051
    },
    {
      "epoch": 3.681224043715847,
      "grad_norm": 0.6731057474594639,
      "learning_rate": 3.302846348512789e-06,
      "loss": 11.1648,
      "step": 21052
    },
    {
      "epoch": 3.681398907103825,
      "grad_norm": 0.6960784771791723,
      "learning_rate": 3.2992379292941855e-06,
      "loss": 11.4197,
      "step": 21053
    },
    {
      "epoch": 3.681573770491803,
      "grad_norm": 0.7347542877369614,
      "learning_rate": 3.295631449222758e-06,
      "loss": 11.3275,
      "step": 21054
    },
    {
      "epoch": 3.6817486338797814,
      "grad_norm": 0.6995161134739877,
      "learning_rate": 3.292026908370849e-06,
      "loss": 11.56,
      "step": 21055
    },
    {
      "epoch": 3.6819234972677597,
      "grad_norm": 0.699314583304045,
      "learning_rate": 3.2884243068107223e-06,
      "loss": 11.4317,
      "step": 21056
    },
    {
      "epoch": 3.6820983606557376,
      "grad_norm": 0.5858425924417027,
      "learning_rate": 3.2848236446146428e-06,
      "loss": 11.6864,
      "step": 21057
    },
    {
      "epoch": 3.682273224043716,
      "grad_norm": 0.6852395586402056,
      "learning_rate": 3.2812249218548084e-06,
      "loss": 11.4561,
      "step": 21058
    },
    {
      "epoch": 3.682448087431694,
      "grad_norm": 0.6371415116909551,
      "learning_rate": 3.2776281386033725e-06,
      "loss": 11.3889,
      "step": 21059
    },
    {
      "epoch": 3.682622950819672,
      "grad_norm": 0.6666708406207484,
      "learning_rate": 3.2740332949324657e-06,
      "loss": 11.4715,
      "step": 21060
    },
    {
      "epoch": 3.6827978142076505,
      "grad_norm": 0.6426250894777257,
      "learning_rate": 3.270440390914187e-06,
      "loss": 11.5568,
      "step": 21061
    },
    {
      "epoch": 3.6829726775956284,
      "grad_norm": 0.6011436489347736,
      "learning_rate": 3.266849426620566e-06,
      "loss": 11.5602,
      "step": 21062
    },
    {
      "epoch": 3.6831475409836063,
      "grad_norm": 0.691436613877571,
      "learning_rate": 3.2632604021236358e-06,
      "loss": 11.1701,
      "step": 21063
    },
    {
      "epoch": 3.6833224043715846,
      "grad_norm": 0.7329527515484354,
      "learning_rate": 3.2596733174953487e-06,
      "loss": 11.3857,
      "step": 21064
    },
    {
      "epoch": 3.683497267759563,
      "grad_norm": 0.6639139677732511,
      "learning_rate": 3.2560881728076365e-06,
      "loss": 11.5761,
      "step": 21065
    },
    {
      "epoch": 3.683672131147541,
      "grad_norm": 0.7433932919179124,
      "learning_rate": 3.252504968132419e-06,
      "loss": 11.3618,
      "step": 21066
    },
    {
      "epoch": 3.6838469945355192,
      "grad_norm": 0.7591133568694595,
      "learning_rate": 3.2489237035415175e-06,
      "loss": 11.2472,
      "step": 21067
    },
    {
      "epoch": 3.684021857923497,
      "grad_norm": 0.6697221982059034,
      "learning_rate": 3.245344379106763e-06,
      "loss": 11.1975,
      "step": 21068
    },
    {
      "epoch": 3.6841967213114755,
      "grad_norm": 0.6440545239061523,
      "learning_rate": 3.241766994899931e-06,
      "loss": 11.5407,
      "step": 21069
    },
    {
      "epoch": 3.684371584699454,
      "grad_norm": 0.6883356533269858,
      "learning_rate": 3.238191550992753e-06,
      "loss": 11.3432,
      "step": 21070
    },
    {
      "epoch": 3.6845464480874317,
      "grad_norm": 0.6655172647223494,
      "learning_rate": 3.2346180474569385e-06,
      "loss": 11.5934,
      "step": 21071
    },
    {
      "epoch": 3.6847213114754096,
      "grad_norm": 0.7029892319251543,
      "learning_rate": 3.2310464843641307e-06,
      "loss": 10.8472,
      "step": 21072
    },
    {
      "epoch": 3.684896174863388,
      "grad_norm": 0.6944494311127489,
      "learning_rate": 3.227476861785972e-06,
      "loss": 11.4993,
      "step": 21073
    },
    {
      "epoch": 3.6850710382513663,
      "grad_norm": 0.6421070608809678,
      "learning_rate": 3.223909179794027e-06,
      "loss": 11.3833,
      "step": 21074
    },
    {
      "epoch": 3.685245901639344,
      "grad_norm": 0.6576345206943094,
      "learning_rate": 3.2203434384598497e-06,
      "loss": 11.5556,
      "step": 21075
    },
    {
      "epoch": 3.6854207650273225,
      "grad_norm": 0.6411605252603841,
      "learning_rate": 3.2167796378549277e-06,
      "loss": 11.4081,
      "step": 21076
    },
    {
      "epoch": 3.6855956284153004,
      "grad_norm": 0.6991795244738955,
      "learning_rate": 3.2132177780507478e-06,
      "loss": 11.7281,
      "step": 21077
    },
    {
      "epoch": 3.6857704918032788,
      "grad_norm": 0.6470465176705005,
      "learning_rate": 3.20965785911872e-06,
      "loss": 11.1472,
      "step": 21078
    },
    {
      "epoch": 3.685945355191257,
      "grad_norm": 0.6272026344304293,
      "learning_rate": 3.2060998811302423e-06,
      "loss": 11.5609,
      "step": 21079
    },
    {
      "epoch": 3.686120218579235,
      "grad_norm": 0.6606088440913969,
      "learning_rate": 3.202543844156658e-06,
      "loss": 11.357,
      "step": 21080
    },
    {
      "epoch": 3.686295081967213,
      "grad_norm": 0.7340386043870543,
      "learning_rate": 3.1989897482692767e-06,
      "loss": 11.4071,
      "step": 21081
    },
    {
      "epoch": 3.6864699453551912,
      "grad_norm": 0.6522773372489723,
      "learning_rate": 3.1954375935393633e-06,
      "loss": 11.4186,
      "step": 21082
    },
    {
      "epoch": 3.6866448087431696,
      "grad_norm": 0.7106619279098112,
      "learning_rate": 3.1918873800381612e-06,
      "loss": 11.2372,
      "step": 21083
    },
    {
      "epoch": 3.6868196721311475,
      "grad_norm": 0.6329724148953304,
      "learning_rate": 3.1883391078368572e-06,
      "loss": 11.6339,
      "step": 21084
    },
    {
      "epoch": 3.686994535519126,
      "grad_norm": 0.6955169183822598,
      "learning_rate": 3.184792777006607e-06,
      "loss": 11.4709,
      "step": 21085
    },
    {
      "epoch": 3.6871693989071037,
      "grad_norm": 0.6535100095805987,
      "learning_rate": 3.1812483876185183e-06,
      "loss": 11.1339,
      "step": 21086
    },
    {
      "epoch": 3.687344262295082,
      "grad_norm": 0.6337855848319859,
      "learning_rate": 3.1777059397436692e-06,
      "loss": 11.1322,
      "step": 21087
    },
    {
      "epoch": 3.68751912568306,
      "grad_norm": 0.651263905499177,
      "learning_rate": 3.1741654334531024e-06,
      "loss": 11.3325,
      "step": 21088
    },
    {
      "epoch": 3.6876939890710383,
      "grad_norm": 0.6414995504234738,
      "learning_rate": 3.1706268688178165e-06,
      "loss": 11.233,
      "step": 21089
    },
    {
      "epoch": 3.687868852459016,
      "grad_norm": 0.6821970369961586,
      "learning_rate": 3.1670902459087658e-06,
      "loss": 11.5079,
      "step": 21090
    },
    {
      "epoch": 3.6880437158469945,
      "grad_norm": 0.7596799405033792,
      "learning_rate": 3.163555564796872e-06,
      "loss": 10.8323,
      "step": 21091
    },
    {
      "epoch": 3.688218579234973,
      "grad_norm": 0.6383899108857872,
      "learning_rate": 3.160022825553011e-06,
      "loss": 11.4177,
      "step": 21092
    },
    {
      "epoch": 3.6883934426229508,
      "grad_norm": 0.6441889753970241,
      "learning_rate": 3.156492028248037e-06,
      "loss": 11.3652,
      "step": 21093
    },
    {
      "epoch": 3.688568306010929,
      "grad_norm": 0.615965517517037,
      "learning_rate": 3.152963172952739e-06,
      "loss": 11.4618,
      "step": 21094
    },
    {
      "epoch": 3.688743169398907,
      "grad_norm": 0.5970562393520907,
      "learning_rate": 3.1494362597378925e-06,
      "loss": 11.1754,
      "step": 21095
    },
    {
      "epoch": 3.6889180327868853,
      "grad_norm": 0.6523255024416824,
      "learning_rate": 3.1459112886742193e-06,
      "loss": 11.4406,
      "step": 21096
    },
    {
      "epoch": 3.6890928961748632,
      "grad_norm": 0.6404486008355039,
      "learning_rate": 3.1423882598324073e-06,
      "loss": 11.1268,
      "step": 21097
    },
    {
      "epoch": 3.6892677595628416,
      "grad_norm": 0.7089501721531595,
      "learning_rate": 3.138867173283111e-06,
      "loss": 11.404,
      "step": 21098
    },
    {
      "epoch": 3.6894426229508195,
      "grad_norm": 0.6496257475660542,
      "learning_rate": 3.1353480290969183e-06,
      "loss": 11.7317,
      "step": 21099
    },
    {
      "epoch": 3.689617486338798,
      "grad_norm": 0.6652884091649816,
      "learning_rate": 3.1318308273444065e-06,
      "loss": 11.1897,
      "step": 21100
    },
    {
      "epoch": 3.689792349726776,
      "grad_norm": 0.8275322472286782,
      "learning_rate": 3.1283155680961187e-06,
      "loss": 11.2975,
      "step": 21101
    },
    {
      "epoch": 3.689967213114754,
      "grad_norm": 0.7122510432315098,
      "learning_rate": 3.1248022514225427e-06,
      "loss": 11.5335,
      "step": 21102
    },
    {
      "epoch": 3.6901420765027324,
      "grad_norm": 0.7582528733975904,
      "learning_rate": 3.121290877394134e-06,
      "loss": 11.3581,
      "step": 21103
    },
    {
      "epoch": 3.6903169398907103,
      "grad_norm": 0.6796454403171533,
      "learning_rate": 3.1177814460812803e-06,
      "loss": 11.4628,
      "step": 21104
    },
    {
      "epoch": 3.6904918032786886,
      "grad_norm": 0.6669318076293915,
      "learning_rate": 3.1142739575543812e-06,
      "loss": 11.2361,
      "step": 21105
    },
    {
      "epoch": 3.6906666666666665,
      "grad_norm": 0.6767052518535354,
      "learning_rate": 3.11076841188378e-06,
      "loss": 11.4563,
      "step": 21106
    },
    {
      "epoch": 3.690841530054645,
      "grad_norm": 0.6938162544622705,
      "learning_rate": 3.107264809139765e-06,
      "loss": 11.3146,
      "step": 21107
    },
    {
      "epoch": 3.691016393442623,
      "grad_norm": 0.6825545175791495,
      "learning_rate": 3.103763149392569e-06,
      "loss": 11.2326,
      "step": 21108
    },
    {
      "epoch": 3.691191256830601,
      "grad_norm": 0.6572914166029505,
      "learning_rate": 3.1002634327124357e-06,
      "loss": 11.1738,
      "step": 21109
    },
    {
      "epoch": 3.6913661202185795,
      "grad_norm": 0.7512332171342122,
      "learning_rate": 3.0967656591695427e-06,
      "loss": 11.1302,
      "step": 21110
    },
    {
      "epoch": 3.6915409836065574,
      "grad_norm": 0.6607506592852695,
      "learning_rate": 3.0932698288340445e-06,
      "loss": 11.4079,
      "step": 21111
    },
    {
      "epoch": 3.6917158469945353,
      "grad_norm": 0.7436496616049424,
      "learning_rate": 3.08977594177603e-06,
      "loss": 11.3425,
      "step": 21112
    },
    {
      "epoch": 3.6918907103825136,
      "grad_norm": 0.9334844657281889,
      "learning_rate": 3.0862839980655313e-06,
      "loss": 11.5176,
      "step": 21113
    },
    {
      "epoch": 3.692065573770492,
      "grad_norm": 0.6562437860201995,
      "learning_rate": 3.0827939977726263e-06,
      "loss": 11.2958,
      "step": 21114
    },
    {
      "epoch": 3.69224043715847,
      "grad_norm": 0.6830554188089442,
      "learning_rate": 3.07930594096727e-06,
      "loss": 11.3088,
      "step": 21115
    },
    {
      "epoch": 3.692415300546448,
      "grad_norm": 0.7127573420388608,
      "learning_rate": 3.075819827719428e-06,
      "loss": 11.3916,
      "step": 21116
    },
    {
      "epoch": 3.692590163934426,
      "grad_norm": 0.6426072137702489,
      "learning_rate": 3.0723356580989904e-06,
      "loss": 11.4819,
      "step": 21117
    },
    {
      "epoch": 3.6927650273224044,
      "grad_norm": 0.6599356319117281,
      "learning_rate": 3.068853432175811e-06,
      "loss": 11.1625,
      "step": 21118
    },
    {
      "epoch": 3.6929398907103828,
      "grad_norm": 0.6897987120803921,
      "learning_rate": 3.0653731500197567e-06,
      "loss": 11.2038,
      "step": 21119
    },
    {
      "epoch": 3.6931147540983607,
      "grad_norm": 0.7110902090532076,
      "learning_rate": 3.0618948117005943e-06,
      "loss": 11.4794,
      "step": 21120
    },
    {
      "epoch": 3.6932896174863385,
      "grad_norm": 0.6227724275289114,
      "learning_rate": 3.058418417288089e-06,
      "loss": 11.3458,
      "step": 21121
    },
    {
      "epoch": 3.693464480874317,
      "grad_norm": 0.7029649148535282,
      "learning_rate": 3.0549439668519307e-06,
      "loss": 11.3616,
      "step": 21122
    },
    {
      "epoch": 3.6936393442622952,
      "grad_norm": 0.866238371126358,
      "learning_rate": 3.0514714604617966e-06,
      "loss": 11.5531,
      "step": 21123
    },
    {
      "epoch": 3.693814207650273,
      "grad_norm": 0.6532240207004036,
      "learning_rate": 3.048000898187342e-06,
      "loss": 11.5994,
      "step": 21124
    },
    {
      "epoch": 3.6939890710382515,
      "grad_norm": 0.782847021232879,
      "learning_rate": 3.044532280098167e-06,
      "loss": 11.477,
      "step": 21125
    },
    {
      "epoch": 3.6941639344262294,
      "grad_norm": 0.7082391197973785,
      "learning_rate": 3.041065606263804e-06,
      "loss": 11.3841,
      "step": 21126
    },
    {
      "epoch": 3.6943387978142077,
      "grad_norm": 0.6369473574259029,
      "learning_rate": 3.0376008767537655e-06,
      "loss": 11.2907,
      "step": 21127
    },
    {
      "epoch": 3.694513661202186,
      "grad_norm": 0.7013258005691015,
      "learning_rate": 3.0341380916375394e-06,
      "loss": 11.3077,
      "step": 21128
    },
    {
      "epoch": 3.694688524590164,
      "grad_norm": 0.6553873486566683,
      "learning_rate": 3.0306772509845817e-06,
      "loss": 11.3098,
      "step": 21129
    },
    {
      "epoch": 3.694863387978142,
      "grad_norm": 0.6384781977029842,
      "learning_rate": 3.027218354864292e-06,
      "loss": 11.1776,
      "step": 21130
    },
    {
      "epoch": 3.69503825136612,
      "grad_norm": 0.748292854751835,
      "learning_rate": 3.0237614033460037e-06,
      "loss": 11.6621,
      "step": 21131
    },
    {
      "epoch": 3.6952131147540985,
      "grad_norm": 0.6983282598145828,
      "learning_rate": 3.0203063964990617e-06,
      "loss": 11.4443,
      "step": 21132
    },
    {
      "epoch": 3.6953879781420764,
      "grad_norm": 0.6899633937878299,
      "learning_rate": 3.016853334392722e-06,
      "loss": 11.1809,
      "step": 21133
    },
    {
      "epoch": 3.6955628415300548,
      "grad_norm": 0.690746018854031,
      "learning_rate": 3.013402217096284e-06,
      "loss": 11.5896,
      "step": 21134
    },
    {
      "epoch": 3.6957377049180327,
      "grad_norm": 0.689840092649128,
      "learning_rate": 3.0099530446789036e-06,
      "loss": 11.279,
      "step": 21135
    },
    {
      "epoch": 3.695912568306011,
      "grad_norm": 0.7667155037096272,
      "learning_rate": 3.0065058172097594e-06,
      "loss": 11.0273,
      "step": 21136
    },
    {
      "epoch": 3.6960874316939893,
      "grad_norm": 0.6463769157715572,
      "learning_rate": 3.0030605347579842e-06,
      "loss": 11.4513,
      "step": 21137
    },
    {
      "epoch": 3.6962622950819672,
      "grad_norm": 0.8207832502830958,
      "learning_rate": 2.9996171973926566e-06,
      "loss": 11.0866,
      "step": 21138
    },
    {
      "epoch": 3.696437158469945,
      "grad_norm": 0.6414446592613835,
      "learning_rate": 2.996175805182866e-06,
      "loss": 11.476,
      "step": 21139
    },
    {
      "epoch": 3.6966120218579235,
      "grad_norm": 0.6529669323727292,
      "learning_rate": 2.9927363581975676e-06,
      "loss": 11.4053,
      "step": 21140
    },
    {
      "epoch": 3.696786885245902,
      "grad_norm": 0.712500167008896,
      "learning_rate": 2.989298856505762e-06,
      "loss": 11.0943,
      "step": 21141
    },
    {
      "epoch": 3.6969617486338797,
      "grad_norm": 0.6497184518510847,
      "learning_rate": 2.9858633001763615e-06,
      "loss": 11.5522,
      "step": 21142
    },
    {
      "epoch": 3.697136612021858,
      "grad_norm": 0.6793177647931825,
      "learning_rate": 2.9824296892782986e-06,
      "loss": 11.4407,
      "step": 21143
    },
    {
      "epoch": 3.697311475409836,
      "grad_norm": 0.7201915746611843,
      "learning_rate": 2.978998023880386e-06,
      "loss": 11.5045,
      "step": 21144
    },
    {
      "epoch": 3.6974863387978143,
      "grad_norm": 0.662090801016377,
      "learning_rate": 2.975568304051468e-06,
      "loss": 11.3089,
      "step": 21145
    },
    {
      "epoch": 3.697661202185792,
      "grad_norm": 0.6308949564770688,
      "learning_rate": 2.972140529860301e-06,
      "loss": 11.2681,
      "step": 21146
    },
    {
      "epoch": 3.6978360655737705,
      "grad_norm": 0.686415822878143,
      "learning_rate": 2.968714701375619e-06,
      "loss": 11.5858,
      "step": 21147
    },
    {
      "epoch": 3.6980109289617484,
      "grad_norm": 0.6399839887504885,
      "learning_rate": 2.9652908186661556e-06,
      "loss": 11.38,
      "step": 21148
    },
    {
      "epoch": 3.6981857923497268,
      "grad_norm": 0.7042891011355985,
      "learning_rate": 2.961868881800545e-06,
      "loss": 11.391,
      "step": 21149
    },
    {
      "epoch": 3.698360655737705,
      "grad_norm": 0.6336342405908894,
      "learning_rate": 2.9584488908473985e-06,
      "loss": 11.2614,
      "step": 21150
    },
    {
      "epoch": 3.698535519125683,
      "grad_norm": 0.660339338330295,
      "learning_rate": 2.955030845875317e-06,
      "loss": 11.4253,
      "step": 21151
    },
    {
      "epoch": 3.6987103825136614,
      "grad_norm": 0.6748911232566307,
      "learning_rate": 2.9516147469528246e-06,
      "loss": 11.303,
      "step": 21152
    },
    {
      "epoch": 3.6988852459016393,
      "grad_norm": 0.6268204268326637,
      "learning_rate": 2.9482005941484423e-06,
      "loss": 11.3153,
      "step": 21153
    },
    {
      "epoch": 3.6990601092896176,
      "grad_norm": 0.672040084034425,
      "learning_rate": 2.944788387530628e-06,
      "loss": 11.4333,
      "step": 21154
    },
    {
      "epoch": 3.6992349726775955,
      "grad_norm": 0.6868405780415969,
      "learning_rate": 2.9413781271677933e-06,
      "loss": 11.204,
      "step": 21155
    },
    {
      "epoch": 3.699409836065574,
      "grad_norm": 0.6678607102896231,
      "learning_rate": 2.9379698131283494e-06,
      "loss": 11.5526,
      "step": 21156
    },
    {
      "epoch": 3.6995846994535517,
      "grad_norm": 0.701962086015211,
      "learning_rate": 2.93456344548062e-06,
      "loss": 11.3875,
      "step": 21157
    },
    {
      "epoch": 3.69975956284153,
      "grad_norm": 0.6799397546750666,
      "learning_rate": 2.9311590242929286e-06,
      "loss": 11.5842,
      "step": 21158
    },
    {
      "epoch": 3.6999344262295084,
      "grad_norm": 0.6683283471466506,
      "learning_rate": 2.9277565496335425e-06,
      "loss": 11.6447,
      "step": 21159
    },
    {
      "epoch": 3.7001092896174863,
      "grad_norm": 0.6467755362951495,
      "learning_rate": 2.9243560215706846e-06,
      "loss": 11.6387,
      "step": 21160
    },
    {
      "epoch": 3.700284153005464,
      "grad_norm": 0.7323015483300319,
      "learning_rate": 2.9209574401725557e-06,
      "loss": 11.4666,
      "step": 21161
    },
    {
      "epoch": 3.7004590163934425,
      "grad_norm": 0.6893829152053476,
      "learning_rate": 2.9175608055072913e-06,
      "loss": 11.7669,
      "step": 21162
    },
    {
      "epoch": 3.700633879781421,
      "grad_norm": 0.6971916681860326,
      "learning_rate": 2.9141661176430244e-06,
      "loss": 11.4871,
      "step": 21163
    },
    {
      "epoch": 3.700808743169399,
      "grad_norm": 0.6717144730084816,
      "learning_rate": 2.910773376647813e-06,
      "loss": 11.6068,
      "step": 21164
    },
    {
      "epoch": 3.700983606557377,
      "grad_norm": 0.7114008373443043,
      "learning_rate": 2.907382582589702e-06,
      "loss": 11.2276,
      "step": 21165
    },
    {
      "epoch": 3.701158469945355,
      "grad_norm": 0.6762688948540945,
      "learning_rate": 2.903993735536681e-06,
      "loss": 11.2744,
      "step": 21166
    },
    {
      "epoch": 3.7013333333333334,
      "grad_norm": 0.6569025374982318,
      "learning_rate": 2.9006068355567075e-06,
      "loss": 11.418,
      "step": 21167
    },
    {
      "epoch": 3.7015081967213117,
      "grad_norm": 0.7309613083505777,
      "learning_rate": 2.8972218827177045e-06,
      "loss": 11.1369,
      "step": 21168
    },
    {
      "epoch": 3.7016830601092896,
      "grad_norm": 0.8511925703840454,
      "learning_rate": 2.8938388770875515e-06,
      "loss": 11.5227,
      "step": 21169
    },
    {
      "epoch": 3.7018579234972675,
      "grad_norm": 0.7080722774652617,
      "learning_rate": 2.8904578187340826e-06,
      "loss": 11.365,
      "step": 21170
    },
    {
      "epoch": 3.702032786885246,
      "grad_norm": 0.7338873112495635,
      "learning_rate": 2.8870787077250994e-06,
      "loss": 11.2579,
      "step": 21171
    },
    {
      "epoch": 3.702207650273224,
      "grad_norm": 0.7226697585238925,
      "learning_rate": 2.8837015441283586e-06,
      "loss": 11.4319,
      "step": 21172
    },
    {
      "epoch": 3.702382513661202,
      "grad_norm": 0.590984322151267,
      "learning_rate": 2.8803263280115844e-06,
      "loss": 11.582,
      "step": 21173
    },
    {
      "epoch": 3.7025573770491804,
      "grad_norm": 0.6232416120945211,
      "learning_rate": 2.8769530594424775e-06,
      "loss": 11.2542,
      "step": 21174
    },
    {
      "epoch": 3.7027322404371583,
      "grad_norm": 0.6892477565587705,
      "learning_rate": 2.873581738488662e-06,
      "loss": 11.4135,
      "step": 21175
    },
    {
      "epoch": 3.7029071038251367,
      "grad_norm": 0.7364915916249167,
      "learning_rate": 2.8702123652177505e-06,
      "loss": 11.5107,
      "step": 21176
    },
    {
      "epoch": 3.703081967213115,
      "grad_norm": 0.732292587617256,
      "learning_rate": 2.866844939697311e-06,
      "loss": 11.4749,
      "step": 21177
    },
    {
      "epoch": 3.703256830601093,
      "grad_norm": 0.6845547576909479,
      "learning_rate": 2.863479461994867e-06,
      "loss": 11.3043,
      "step": 21178
    },
    {
      "epoch": 3.703431693989071,
      "grad_norm": 0.6569687568713865,
      "learning_rate": 2.8601159321778982e-06,
      "loss": 11.3843,
      "step": 21179
    },
    {
      "epoch": 3.703606557377049,
      "grad_norm": 0.943965111208733,
      "learning_rate": 2.856754350313873e-06,
      "loss": 11.5886,
      "step": 21180
    },
    {
      "epoch": 3.7037814207650275,
      "grad_norm": 0.7291365169513426,
      "learning_rate": 2.8533947164701926e-06,
      "loss": 11.3553,
      "step": 21181
    },
    {
      "epoch": 3.7039562841530054,
      "grad_norm": 0.6546826595588872,
      "learning_rate": 2.8500370307142255e-06,
      "loss": 11.0454,
      "step": 21182
    },
    {
      "epoch": 3.7041311475409837,
      "grad_norm": 0.6313536459303474,
      "learning_rate": 2.8466812931133068e-06,
      "loss": 11.6102,
      "step": 21183
    },
    {
      "epoch": 3.7043060109289616,
      "grad_norm": 0.6692963453913018,
      "learning_rate": 2.8433275037347273e-06,
      "loss": 11.3449,
      "step": 21184
    },
    {
      "epoch": 3.70448087431694,
      "grad_norm": 0.6785077130657156,
      "learning_rate": 2.839975662645744e-06,
      "loss": 11.3826,
      "step": 21185
    },
    {
      "epoch": 3.7046557377049183,
      "grad_norm": 0.6428968402887233,
      "learning_rate": 2.836625769913559e-06,
      "loss": 11.2233,
      "step": 21186
    },
    {
      "epoch": 3.704830601092896,
      "grad_norm": 0.6034238926120757,
      "learning_rate": 2.833277825605374e-06,
      "loss": 11.1967,
      "step": 21187
    },
    {
      "epoch": 3.705005464480874,
      "grad_norm": 0.6744232081660723,
      "learning_rate": 2.8299318297883013e-06,
      "loss": 11.3816,
      "step": 21188
    },
    {
      "epoch": 3.7051803278688524,
      "grad_norm": 0.6521500379755825,
      "learning_rate": 2.826587782529444e-06,
      "loss": 11.3701,
      "step": 21189
    },
    {
      "epoch": 3.7053551912568308,
      "grad_norm": 0.7075534842258452,
      "learning_rate": 2.82324568389587e-06,
      "loss": 11.4864,
      "step": 21190
    },
    {
      "epoch": 3.7055300546448087,
      "grad_norm": 0.7107219240546891,
      "learning_rate": 2.819905533954581e-06,
      "loss": 11.2196,
      "step": 21191
    },
    {
      "epoch": 3.705704918032787,
      "grad_norm": 0.6469933394343899,
      "learning_rate": 2.8165673327725795e-06,
      "loss": 11.637,
      "step": 21192
    },
    {
      "epoch": 3.705879781420765,
      "grad_norm": 0.6826311463026605,
      "learning_rate": 2.8132310804167894e-06,
      "loss": 11.4454,
      "step": 21193
    },
    {
      "epoch": 3.7060546448087432,
      "grad_norm": 0.6839002009318433,
      "learning_rate": 2.809896776954124e-06,
      "loss": 11.1854,
      "step": 21194
    },
    {
      "epoch": 3.706229508196721,
      "grad_norm": 0.718570908308841,
      "learning_rate": 2.80656442245143e-06,
      "loss": 11.4357,
      "step": 21195
    },
    {
      "epoch": 3.7064043715846995,
      "grad_norm": 0.5999965846989423,
      "learning_rate": 2.8032340169755535e-06,
      "loss": 11.7936,
      "step": 21196
    },
    {
      "epoch": 3.7065792349726774,
      "grad_norm": 0.6963498026934576,
      "learning_rate": 2.7999055605932635e-06,
      "loss": 11.5091,
      "step": 21197
    },
    {
      "epoch": 3.7067540983606557,
      "grad_norm": 0.7025544810201263,
      "learning_rate": 2.7965790533713064e-06,
      "loss": 11.3839,
      "step": 21198
    },
    {
      "epoch": 3.706928961748634,
      "grad_norm": 0.7737596132915558,
      "learning_rate": 2.7932544953764073e-06,
      "loss": 11.4828,
      "step": 21199
    },
    {
      "epoch": 3.707103825136612,
      "grad_norm": 0.6683360741506724,
      "learning_rate": 2.7899318866752013e-06,
      "loss": 11.6206,
      "step": 21200
    },
    {
      "epoch": 3.7072786885245903,
      "grad_norm": 0.6291167780082939,
      "learning_rate": 2.786611227334346e-06,
      "loss": 11.5759,
      "step": 21201
    },
    {
      "epoch": 3.707453551912568,
      "grad_norm": 0.6955142894935662,
      "learning_rate": 2.783292517420422e-06,
      "loss": 11.1979,
      "step": 21202
    },
    {
      "epoch": 3.7076284153005465,
      "grad_norm": 0.6764441314862264,
      "learning_rate": 2.779975756999975e-06,
      "loss": 11.4989,
      "step": 21203
    },
    {
      "epoch": 3.7078032786885244,
      "grad_norm": 0.6623905868305725,
      "learning_rate": 2.77666094613952e-06,
      "loss": 11.4776,
      "step": 21204
    },
    {
      "epoch": 3.707978142076503,
      "grad_norm": 0.6709071121059328,
      "learning_rate": 2.773348084905514e-06,
      "loss": 11.3373,
      "step": 21205
    },
    {
      "epoch": 3.7081530054644807,
      "grad_norm": 0.718042023011541,
      "learning_rate": 2.770037173364415e-06,
      "loss": 11.6175,
      "step": 21206
    },
    {
      "epoch": 3.708327868852459,
      "grad_norm": 0.7044123353818402,
      "learning_rate": 2.7667282115826033e-06,
      "loss": 11.177,
      "step": 21207
    },
    {
      "epoch": 3.7085027322404374,
      "grad_norm": 0.6591672944955114,
      "learning_rate": 2.7634211996264482e-06,
      "loss": 11.5106,
      "step": 21208
    },
    {
      "epoch": 3.7086775956284153,
      "grad_norm": 0.6646941664117043,
      "learning_rate": 2.760116137562241e-06,
      "loss": 11.4205,
      "step": 21209
    },
    {
      "epoch": 3.7088524590163936,
      "grad_norm": 0.7019949106759867,
      "learning_rate": 2.756813025456273e-06,
      "loss": 11.3468,
      "step": 21210
    },
    {
      "epoch": 3.7090273224043715,
      "grad_norm": 0.6419643717095472,
      "learning_rate": 2.7535118633747802e-06,
      "loss": 11.0835,
      "step": 21211
    },
    {
      "epoch": 3.70920218579235,
      "grad_norm": 0.6707609631205568,
      "learning_rate": 2.7502126513839653e-06,
      "loss": 11.5381,
      "step": 21212
    },
    {
      "epoch": 3.7093770491803277,
      "grad_norm": 0.5951922354754257,
      "learning_rate": 2.7469153895499757e-06,
      "loss": 11.468,
      "step": 21213
    },
    {
      "epoch": 3.709551912568306,
      "grad_norm": 0.733588828336494,
      "learning_rate": 2.7436200779389353e-06,
      "loss": 11.3247,
      "step": 21214
    },
    {
      "epoch": 3.709726775956284,
      "grad_norm": 0.6438853872247959,
      "learning_rate": 2.7403267166169366e-06,
      "loss": 11.344,
      "step": 21215
    },
    {
      "epoch": 3.7099016393442623,
      "grad_norm": 0.6785279740915582,
      "learning_rate": 2.737035305650004e-06,
      "loss": 11.3908,
      "step": 21216
    },
    {
      "epoch": 3.7100765027322407,
      "grad_norm": 0.6513420902185075,
      "learning_rate": 2.7337458451041744e-06,
      "loss": 11.5907,
      "step": 21217
    },
    {
      "epoch": 3.7102513661202186,
      "grad_norm": 0.6997460446706265,
      "learning_rate": 2.7304583350453714e-06,
      "loss": 11.4832,
      "step": 21218
    },
    {
      "epoch": 3.7104262295081964,
      "grad_norm": 0.6460824552043857,
      "learning_rate": 2.7271727755395214e-06,
      "loss": 10.9885,
      "step": 21219
    },
    {
      "epoch": 3.710601092896175,
      "grad_norm": 0.6811394942956229,
      "learning_rate": 2.723889166652538e-06,
      "loss": 11.4448,
      "step": 21220
    },
    {
      "epoch": 3.710775956284153,
      "grad_norm": 0.6584368376740242,
      "learning_rate": 2.720607508450268e-06,
      "loss": 11.4439,
      "step": 21221
    },
    {
      "epoch": 3.710950819672131,
      "grad_norm": 0.6499129268550742,
      "learning_rate": 2.7173278009984814e-06,
      "loss": 11.3789,
      "step": 21222
    },
    {
      "epoch": 3.7111256830601094,
      "grad_norm": 0.6394965967254702,
      "learning_rate": 2.714050044362981e-06,
      "loss": 11.1137,
      "step": 21223
    },
    {
      "epoch": 3.7113005464480873,
      "grad_norm": 0.6367447422110902,
      "learning_rate": 2.71077423860947e-06,
      "loss": 11.5129,
      "step": 21224
    },
    {
      "epoch": 3.7114754098360656,
      "grad_norm": 0.6840317811416678,
      "learning_rate": 2.707500383803663e-06,
      "loss": 11.4778,
      "step": 21225
    },
    {
      "epoch": 3.711650273224044,
      "grad_norm": 0.6402355765127327,
      "learning_rate": 2.704228480011206e-06,
      "loss": 11.3547,
      "step": 21226
    },
    {
      "epoch": 3.711825136612022,
      "grad_norm": 0.6690869475438027,
      "learning_rate": 2.7009585272977034e-06,
      "loss": 11.3509,
      "step": 21227
    },
    {
      "epoch": 3.7119999999999997,
      "grad_norm": 0.6900925033327554,
      "learning_rate": 2.6976905257287243e-06,
      "loss": 11.2871,
      "step": 21228
    },
    {
      "epoch": 3.712174863387978,
      "grad_norm": 0.7026580889362635,
      "learning_rate": 2.6944244753698056e-06,
      "loss": 11.2666,
      "step": 21229
    },
    {
      "epoch": 3.7123497267759564,
      "grad_norm": 0.6999165759622213,
      "learning_rate": 2.69116037628645e-06,
      "loss": 11.4103,
      "step": 21230
    },
    {
      "epoch": 3.7125245901639343,
      "grad_norm": 0.6305044570190361,
      "learning_rate": 2.6878982285441056e-06,
      "loss": 11.2362,
      "step": 21231
    },
    {
      "epoch": 3.7126994535519127,
      "grad_norm": 0.68395971046322,
      "learning_rate": 2.6846380322081867e-06,
      "loss": 11.3769,
      "step": 21232
    },
    {
      "epoch": 3.7128743169398906,
      "grad_norm": 0.7073834193529903,
      "learning_rate": 2.6813797873440626e-06,
      "loss": 11.57,
      "step": 21233
    },
    {
      "epoch": 3.713049180327869,
      "grad_norm": 0.7048812083406854,
      "learning_rate": 2.678123494017093e-06,
      "loss": 11.5162,
      "step": 21234
    },
    {
      "epoch": 3.7132240437158472,
      "grad_norm": 0.6701578113854657,
      "learning_rate": 2.6748691522925808e-06,
      "loss": 11.271,
      "step": 21235
    },
    {
      "epoch": 3.713398907103825,
      "grad_norm": 0.6782356484357434,
      "learning_rate": 2.6716167622357513e-06,
      "loss": 11.4694,
      "step": 21236
    },
    {
      "epoch": 3.713573770491803,
      "grad_norm": 0.7168631617350897,
      "learning_rate": 2.6683663239118416e-06,
      "loss": 11.4462,
      "step": 21237
    },
    {
      "epoch": 3.7137486338797814,
      "grad_norm": 0.6493847297409044,
      "learning_rate": 2.665117837386022e-06,
      "loss": 11.3525,
      "step": 21238
    },
    {
      "epoch": 3.7139234972677597,
      "grad_norm": 0.6837995117474343,
      "learning_rate": 2.661871302723473e-06,
      "loss": 11.3693,
      "step": 21239
    },
    {
      "epoch": 3.7140983606557376,
      "grad_norm": 0.6731469025192517,
      "learning_rate": 2.6586267199892658e-06,
      "loss": 11.1436,
      "step": 21240
    },
    {
      "epoch": 3.714273224043716,
      "grad_norm": 0.6602449450015446,
      "learning_rate": 2.6553840892484583e-06,
      "loss": 11.1367,
      "step": 21241
    },
    {
      "epoch": 3.714448087431694,
      "grad_norm": 0.7406723353831105,
      "learning_rate": 2.6521434105660993e-06,
      "loss": 11.4874,
      "step": 21242
    },
    {
      "epoch": 3.714622950819672,
      "grad_norm": 0.7271088166306814,
      "learning_rate": 2.6489046840071475e-06,
      "loss": 11.3933,
      "step": 21243
    },
    {
      "epoch": 3.7147978142076505,
      "grad_norm": 0.6981049915623166,
      "learning_rate": 2.6456679096365845e-06,
      "loss": 11.4025,
      "step": 21244
    },
    {
      "epoch": 3.7149726775956284,
      "grad_norm": 0.655323776244891,
      "learning_rate": 2.6424330875192805e-06,
      "loss": 11.6643,
      "step": 21245
    },
    {
      "epoch": 3.7151475409836063,
      "grad_norm": 0.6914629286710853,
      "learning_rate": 2.639200217720128e-06,
      "loss": 11.4076,
      "step": 21246
    },
    {
      "epoch": 3.7153224043715847,
      "grad_norm": 0.651091707388968,
      "learning_rate": 2.635969300303942e-06,
      "loss": 11.3716,
      "step": 21247
    },
    {
      "epoch": 3.715497267759563,
      "grad_norm": 0.6999897365345267,
      "learning_rate": 2.6327403353355264e-06,
      "loss": 11.2183,
      "step": 21248
    },
    {
      "epoch": 3.715672131147541,
      "grad_norm": 0.7248312561907078,
      "learning_rate": 2.6295133228796177e-06,
      "loss": 11.285,
      "step": 21249
    },
    {
      "epoch": 3.7158469945355193,
      "grad_norm": 0.7227858598095199,
      "learning_rate": 2.626288263000931e-06,
      "loss": 11.3966,
      "step": 21250
    },
    {
      "epoch": 3.716021857923497,
      "grad_norm": 0.7510836697747643,
      "learning_rate": 2.6230651557641485e-06,
      "loss": 11.4225,
      "step": 21251
    },
    {
      "epoch": 3.7161967213114755,
      "grad_norm": 0.7341764430006616,
      "learning_rate": 2.619844001233884e-06,
      "loss": 11.5016,
      "step": 21252
    },
    {
      "epoch": 3.7163715846994534,
      "grad_norm": 0.6247740695058335,
      "learning_rate": 2.616624799474754e-06,
      "loss": 11.6266,
      "step": 21253
    },
    {
      "epoch": 3.7165464480874317,
      "grad_norm": 0.6135283795373245,
      "learning_rate": 2.6134075505513057e-06,
      "loss": 11.7058,
      "step": 21254
    },
    {
      "epoch": 3.7167213114754096,
      "grad_norm": 0.6185334037847324,
      "learning_rate": 2.6101922545280434e-06,
      "loss": 11.216,
      "step": 21255
    },
    {
      "epoch": 3.716896174863388,
      "grad_norm": 0.6718171016559146,
      "learning_rate": 2.60697891146946e-06,
      "loss": 11.484,
      "step": 21256
    },
    {
      "epoch": 3.7170710382513663,
      "grad_norm": 0.6978113443538528,
      "learning_rate": 2.6037675214399813e-06,
      "loss": 11.498,
      "step": 21257
    },
    {
      "epoch": 3.717245901639344,
      "grad_norm": 0.7216140477173337,
      "learning_rate": 2.6005580845040123e-06,
      "loss": 11.3818,
      "step": 21258
    },
    {
      "epoch": 3.7174207650273225,
      "grad_norm": 0.708525160185481,
      "learning_rate": 2.5973506007259007e-06,
      "loss": 11.2237,
      "step": 21259
    },
    {
      "epoch": 3.7175956284153004,
      "grad_norm": 0.6637549543306561,
      "learning_rate": 2.5941450701699844e-06,
      "loss": 11.3923,
      "step": 21260
    },
    {
      "epoch": 3.717770491803279,
      "grad_norm": 0.6431093333228916,
      "learning_rate": 2.590941492900534e-06,
      "loss": 11.2295,
      "step": 21261
    },
    {
      "epoch": 3.7179453551912567,
      "grad_norm": 0.7482533762839588,
      "learning_rate": 2.5877398689817866e-06,
      "loss": 11.4748,
      "step": 21262
    },
    {
      "epoch": 3.718120218579235,
      "grad_norm": 0.6440245879310188,
      "learning_rate": 2.584540198477947e-06,
      "loss": 11.5461,
      "step": 21263
    },
    {
      "epoch": 3.718295081967213,
      "grad_norm": 0.6764498429008597,
      "learning_rate": 2.581342481453186e-06,
      "loss": 11.5852,
      "step": 21264
    },
    {
      "epoch": 3.7184699453551913,
      "grad_norm": 0.6298796458105058,
      "learning_rate": 2.57814671797163e-06,
      "loss": 11.5126,
      "step": 21265
    },
    {
      "epoch": 3.7186448087431696,
      "grad_norm": 0.6995286709874081,
      "learning_rate": 2.57495290809735e-06,
      "loss": 11.3123,
      "step": 21266
    },
    {
      "epoch": 3.7188196721311475,
      "grad_norm": 0.6649302279157037,
      "learning_rate": 2.5717610518943947e-06,
      "loss": 11.4756,
      "step": 21267
    },
    {
      "epoch": 3.718994535519126,
      "grad_norm": 0.69799609378449,
      "learning_rate": 2.5685711494267794e-06,
      "loss": 11.2042,
      "step": 21268
    },
    {
      "epoch": 3.7191693989071037,
      "grad_norm": 0.6970071652549858,
      "learning_rate": 2.5653832007584643e-06,
      "loss": 11.5358,
      "step": 21269
    },
    {
      "epoch": 3.719344262295082,
      "grad_norm": 0.6081393338578628,
      "learning_rate": 2.562197205953376e-06,
      "loss": 11.7425,
      "step": 21270
    },
    {
      "epoch": 3.71951912568306,
      "grad_norm": 0.6545666900587743,
      "learning_rate": 2.559013165075419e-06,
      "loss": 11.3139,
      "step": 21271
    },
    {
      "epoch": 3.7196939890710383,
      "grad_norm": 0.6503906236025065,
      "learning_rate": 2.55583107818842e-06,
      "loss": 11.5655,
      "step": 21272
    },
    {
      "epoch": 3.719868852459016,
      "grad_norm": 0.6850898419663725,
      "learning_rate": 2.5526509453562053e-06,
      "loss": 11.3084,
      "step": 21273
    },
    {
      "epoch": 3.7200437158469946,
      "grad_norm": 0.6395926465970305,
      "learning_rate": 2.5494727666425355e-06,
      "loss": 11.2827,
      "step": 21274
    },
    {
      "epoch": 3.720218579234973,
      "grad_norm": 0.6769285741196878,
      "learning_rate": 2.5462965421111596e-06,
      "loss": 11.2986,
      "step": 21275
    },
    {
      "epoch": 3.720393442622951,
      "grad_norm": 0.6547783943456537,
      "learning_rate": 2.5431222718257596e-06,
      "loss": 11.2324,
      "step": 21276
    },
    {
      "epoch": 3.7205683060109287,
      "grad_norm": 0.6927534653845785,
      "learning_rate": 2.539949955849985e-06,
      "loss": 11.4369,
      "step": 21277
    },
    {
      "epoch": 3.720743169398907,
      "grad_norm": 0.6374279487351582,
      "learning_rate": 2.5367795942474517e-06,
      "loss": 11.5278,
      "step": 21278
    },
    {
      "epoch": 3.7209180327868854,
      "grad_norm": 0.6976635549210682,
      "learning_rate": 2.5336111870817414e-06,
      "loss": 11.2511,
      "step": 21279
    },
    {
      "epoch": 3.7210928961748633,
      "grad_norm": 0.7194527089106372,
      "learning_rate": 2.5304447344163927e-06,
      "loss": 11.1499,
      "step": 21280
    },
    {
      "epoch": 3.7212677595628416,
      "grad_norm": 0.7584014061412901,
      "learning_rate": 2.5272802363148885e-06,
      "loss": 11.2387,
      "step": 21281
    },
    {
      "epoch": 3.7214426229508195,
      "grad_norm": 0.6527044177402763,
      "learning_rate": 2.5241176928407106e-06,
      "loss": 11.3984,
      "step": 21282
    },
    {
      "epoch": 3.721617486338798,
      "grad_norm": 0.681698311079585,
      "learning_rate": 2.5209571040572534e-06,
      "loss": 11.3519,
      "step": 21283
    },
    {
      "epoch": 3.721792349726776,
      "grad_norm": 0.6983395994749383,
      "learning_rate": 2.51779847002791e-06,
      "loss": 11.358,
      "step": 21284
    },
    {
      "epoch": 3.721967213114754,
      "grad_norm": 0.6351425124000084,
      "learning_rate": 2.514641790816008e-06,
      "loss": 11.1791,
      "step": 21285
    },
    {
      "epoch": 3.722142076502732,
      "grad_norm": 0.6533274056294586,
      "learning_rate": 2.511487066484863e-06,
      "loss": 11.3338,
      "step": 21286
    },
    {
      "epoch": 3.7223169398907103,
      "grad_norm": 0.6828747615321556,
      "learning_rate": 2.5083342970977366e-06,
      "loss": 11.4922,
      "step": 21287
    },
    {
      "epoch": 3.7224918032786887,
      "grad_norm": 0.6318508611980272,
      "learning_rate": 2.5051834827178432e-06,
      "loss": 11.422,
      "step": 21288
    },
    {
      "epoch": 3.7226666666666666,
      "grad_norm": 0.6702439785845103,
      "learning_rate": 2.5020346234083667e-06,
      "loss": 11.6012,
      "step": 21289
    },
    {
      "epoch": 3.722841530054645,
      "grad_norm": 0.6923210573201489,
      "learning_rate": 2.4988877192324677e-06,
      "loss": 11.1034,
      "step": 21290
    },
    {
      "epoch": 3.723016393442623,
      "grad_norm": 0.6453390972596514,
      "learning_rate": 2.4957427702532176e-06,
      "loss": 11.5555,
      "step": 21291
    },
    {
      "epoch": 3.723191256830601,
      "grad_norm": 0.727287785470698,
      "learning_rate": 2.4925997765337105e-06,
      "loss": 11.492,
      "step": 21292
    },
    {
      "epoch": 3.7233661202185795,
      "grad_norm": 0.6743989755055148,
      "learning_rate": 2.4894587381369628e-06,
      "loss": 11.3343,
      "step": 21293
    },
    {
      "epoch": 3.7235409836065574,
      "grad_norm": 0.6946810357268357,
      "learning_rate": 2.4863196551259683e-06,
      "loss": 11.0763,
      "step": 21294
    },
    {
      "epoch": 3.7237158469945353,
      "grad_norm": 0.6640839307374414,
      "learning_rate": 2.4831825275636767e-06,
      "loss": 11.4524,
      "step": 21295
    },
    {
      "epoch": 3.7238907103825136,
      "grad_norm": 0.7077245636658941,
      "learning_rate": 2.480047355512971e-06,
      "loss": 11.3587,
      "step": 21296
    },
    {
      "epoch": 3.724065573770492,
      "grad_norm": 0.6851862454655051,
      "learning_rate": 2.4769141390367567e-06,
      "loss": 11.1872,
      "step": 21297
    },
    {
      "epoch": 3.72424043715847,
      "grad_norm": 0.7052056500181966,
      "learning_rate": 2.473782878197839e-06,
      "loss": 11.1649,
      "step": 21298
    },
    {
      "epoch": 3.724415300546448,
      "grad_norm": 0.6327658940649656,
      "learning_rate": 2.4706535730590453e-06,
      "loss": 11.0743,
      "step": 21299
    },
    {
      "epoch": 3.724590163934426,
      "grad_norm": 0.721984420726789,
      "learning_rate": 2.467526223683081e-06,
      "loss": 11.2918,
      "step": 21300
    },
    {
      "epoch": 3.7247650273224044,
      "grad_norm": 0.6506658911290788,
      "learning_rate": 2.4644008301326736e-06,
      "loss": 11.327,
      "step": 21301
    },
    {
      "epoch": 3.724939890710383,
      "grad_norm": 0.7101367337262076,
      "learning_rate": 2.4612773924705067e-06,
      "loss": 11.3644,
      "step": 21302
    },
    {
      "epoch": 3.7251147540983607,
      "grad_norm": 0.6791475650136739,
      "learning_rate": 2.458155910759219e-06,
      "loss": 11.5684,
      "step": 21303
    },
    {
      "epoch": 3.7252896174863386,
      "grad_norm": 0.7108798900631323,
      "learning_rate": 2.4550363850614043e-06,
      "loss": 11.2863,
      "step": 21304
    },
    {
      "epoch": 3.725464480874317,
      "grad_norm": 0.6569935287879903,
      "learning_rate": 2.451918815439591e-06,
      "loss": 11.6678,
      "step": 21305
    },
    {
      "epoch": 3.7256393442622953,
      "grad_norm": 0.6137635054448826,
      "learning_rate": 2.4488032019563402e-06,
      "loss": 11.4255,
      "step": 21306
    },
    {
      "epoch": 3.725814207650273,
      "grad_norm": 0.7628303609453353,
      "learning_rate": 2.4456895446740903e-06,
      "loss": 11.1643,
      "step": 21307
    },
    {
      "epoch": 3.7259890710382515,
      "grad_norm": 0.651563966374777,
      "learning_rate": 2.4425778436553137e-06,
      "loss": 11.4736,
      "step": 21308
    },
    {
      "epoch": 3.7261639344262294,
      "grad_norm": 0.5916055381938062,
      "learning_rate": 2.439468098962383e-06,
      "loss": 10.9407,
      "step": 21309
    },
    {
      "epoch": 3.7263387978142077,
      "grad_norm": 0.6705458612804799,
      "learning_rate": 2.436360310657648e-06,
      "loss": 11.3872,
      "step": 21310
    },
    {
      "epoch": 3.7265136612021856,
      "grad_norm": 0.7330027459803885,
      "learning_rate": 2.4332544788034708e-06,
      "loss": 11.3718,
      "step": 21311
    },
    {
      "epoch": 3.726688524590164,
      "grad_norm": 0.713402046624095,
      "learning_rate": 2.4301506034621e-06,
      "loss": 11.3818,
      "step": 21312
    },
    {
      "epoch": 3.726863387978142,
      "grad_norm": 0.6479202332861501,
      "learning_rate": 2.4270486846958094e-06,
      "loss": 11.2786,
      "step": 21313
    },
    {
      "epoch": 3.72703825136612,
      "grad_norm": 0.7215317736569159,
      "learning_rate": 2.42394872256676e-06,
      "loss": 11.5318,
      "step": 21314
    },
    {
      "epoch": 3.7272131147540986,
      "grad_norm": 0.6588122445854055,
      "learning_rate": 2.4208507171371353e-06,
      "loss": 11.4825,
      "step": 21315
    },
    {
      "epoch": 3.7273879781420765,
      "grad_norm": 0.6181403480909198,
      "learning_rate": 2.4177546684690634e-06,
      "loss": 11.4702,
      "step": 21316
    },
    {
      "epoch": 3.727562841530055,
      "grad_norm": 0.6351625176963754,
      "learning_rate": 2.414660576624639e-06,
      "loss": 11.5363,
      "step": 21317
    },
    {
      "epoch": 3.7277377049180327,
      "grad_norm": 0.6785931096075013,
      "learning_rate": 2.41156844166589e-06,
      "loss": 11.4634,
      "step": 21318
    },
    {
      "epoch": 3.727912568306011,
      "grad_norm": 0.6830518901693439,
      "learning_rate": 2.4084782636548343e-06,
      "loss": 11.4188,
      "step": 21319
    },
    {
      "epoch": 3.728087431693989,
      "grad_norm": 0.6776154378722778,
      "learning_rate": 2.4053900426534215e-06,
      "loss": 11.4985,
      "step": 21320
    },
    {
      "epoch": 3.7282622950819673,
      "grad_norm": 0.6359410453391726,
      "learning_rate": 2.4023037787235912e-06,
      "loss": 11.0262,
      "step": 21321
    },
    {
      "epoch": 3.728437158469945,
      "grad_norm": 0.6585222855702035,
      "learning_rate": 2.3992194719272608e-06,
      "loss": 11.455,
      "step": 21322
    },
    {
      "epoch": 3.7286120218579235,
      "grad_norm": 0.7556558581686259,
      "learning_rate": 2.3961371223262364e-06,
      "loss": 11.2691,
      "step": 21323
    },
    {
      "epoch": 3.728786885245902,
      "grad_norm": 0.6553630692835865,
      "learning_rate": 2.3930567299823457e-06,
      "loss": 11.4464,
      "step": 21324
    },
    {
      "epoch": 3.7289617486338797,
      "grad_norm": 0.6153373606326136,
      "learning_rate": 2.389978294957351e-06,
      "loss": 11.147,
      "step": 21325
    },
    {
      "epoch": 3.729136612021858,
      "grad_norm": 0.8194038556988311,
      "learning_rate": 2.3869018173130143e-06,
      "loss": 11.4602,
      "step": 21326
    },
    {
      "epoch": 3.729311475409836,
      "grad_norm": 0.6683806067448789,
      "learning_rate": 2.3838272971109966e-06,
      "loss": 11.5442,
      "step": 21327
    },
    {
      "epoch": 3.7294863387978143,
      "grad_norm": 0.6579180551931301,
      "learning_rate": 2.38075473441296e-06,
      "loss": 11.5647,
      "step": 21328
    },
    {
      "epoch": 3.7296612021857922,
      "grad_norm": 0.6765105289751278,
      "learning_rate": 2.3776841292805107e-06,
      "loss": 11.4161,
      "step": 21329
    },
    {
      "epoch": 3.7298360655737706,
      "grad_norm": 0.7351930398074985,
      "learning_rate": 2.374615481775244e-06,
      "loss": 11.343,
      "step": 21330
    },
    {
      "epoch": 3.7300109289617485,
      "grad_norm": 0.6834121257418109,
      "learning_rate": 2.3715487919587e-06,
      "loss": 11.3139,
      "step": 21331
    },
    {
      "epoch": 3.730185792349727,
      "grad_norm": 0.6546217827506211,
      "learning_rate": 2.3684840598923506e-06,
      "loss": 11.2318,
      "step": 21332
    },
    {
      "epoch": 3.730360655737705,
      "grad_norm": 0.6266420518041371,
      "learning_rate": 2.36542128563767e-06,
      "loss": 11.3902,
      "step": 21333
    },
    {
      "epoch": 3.730535519125683,
      "grad_norm": 0.6470501438588482,
      "learning_rate": 2.362360469256053e-06,
      "loss": 11.31,
      "step": 21334
    },
    {
      "epoch": 3.730710382513661,
      "grad_norm": 0.6620759782114255,
      "learning_rate": 2.359301610808917e-06,
      "loss": 11.4158,
      "step": 21335
    },
    {
      "epoch": 3.7308852459016393,
      "grad_norm": 0.8670343439492945,
      "learning_rate": 2.356244710357569e-06,
      "loss": 11.2247,
      "step": 21336
    },
    {
      "epoch": 3.7310601092896176,
      "grad_norm": 0.6498025358768814,
      "learning_rate": 2.3531897679633264e-06,
      "loss": 11.4002,
      "step": 21337
    },
    {
      "epoch": 3.7312349726775955,
      "grad_norm": 0.6775296242878049,
      "learning_rate": 2.35013678368744e-06,
      "loss": 11.4997,
      "step": 21338
    },
    {
      "epoch": 3.731409836065574,
      "grad_norm": 0.7343822199320967,
      "learning_rate": 2.3470857575911276e-06,
      "loss": 11.4652,
      "step": 21339
    },
    {
      "epoch": 3.7315846994535518,
      "grad_norm": 0.6183312902233945,
      "learning_rate": 2.344036689735596e-06,
      "loss": 11.4988,
      "step": 21340
    },
    {
      "epoch": 3.73175956284153,
      "grad_norm": 0.641750609191814,
      "learning_rate": 2.3409895801819626e-06,
      "loss": 11.3641,
      "step": 21341
    },
    {
      "epoch": 3.7319344262295084,
      "grad_norm": 0.8017126993005264,
      "learning_rate": 2.3379444289913342e-06,
      "loss": 11.3796,
      "step": 21342
    },
    {
      "epoch": 3.7321092896174863,
      "grad_norm": 0.7289385591331982,
      "learning_rate": 2.334901236224796e-06,
      "loss": 11.2065,
      "step": 21343
    },
    {
      "epoch": 3.7322841530054642,
      "grad_norm": 0.7974281366870347,
      "learning_rate": 2.3318600019433424e-06,
      "loss": 11.4893,
      "step": 21344
    },
    {
      "epoch": 3.7324590163934426,
      "grad_norm": 0.6598071989041584,
      "learning_rate": 2.328820726207992e-06,
      "loss": 11.0576,
      "step": 21345
    },
    {
      "epoch": 3.732633879781421,
      "grad_norm": 0.6861928094836124,
      "learning_rate": 2.3257834090796627e-06,
      "loss": 11.1285,
      "step": 21346
    },
    {
      "epoch": 3.732808743169399,
      "grad_norm": 0.6275160325171527,
      "learning_rate": 2.3227480506192834e-06,
      "loss": 11.4645,
      "step": 21347
    },
    {
      "epoch": 3.732983606557377,
      "grad_norm": 0.6186968624143899,
      "learning_rate": 2.3197146508877164e-06,
      "loss": 11.6547,
      "step": 21348
    },
    {
      "epoch": 3.733158469945355,
      "grad_norm": 0.6151189836965423,
      "learning_rate": 2.3166832099457802e-06,
      "loss": 11.5605,
      "step": 21349
    },
    {
      "epoch": 3.7333333333333334,
      "grad_norm": 0.7145098004332249,
      "learning_rate": 2.3136537278542814e-06,
      "loss": 11.5153,
      "step": 21350
    },
    {
      "epoch": 3.7335081967213117,
      "grad_norm": 0.6329883571653245,
      "learning_rate": 2.31062620467396e-06,
      "loss": 11.4177,
      "step": 21351
    },
    {
      "epoch": 3.7336830601092896,
      "grad_norm": 0.7139878063333839,
      "learning_rate": 2.3076006404655237e-06,
      "loss": 11.4806,
      "step": 21352
    },
    {
      "epoch": 3.7338579234972675,
      "grad_norm": 0.6790114149067165,
      "learning_rate": 2.3045770352896566e-06,
      "loss": 11.4311,
      "step": 21353
    },
    {
      "epoch": 3.734032786885246,
      "grad_norm": 0.6317786290967141,
      "learning_rate": 2.3015553892069776e-06,
      "loss": 11.1976,
      "step": 21354
    },
    {
      "epoch": 3.734207650273224,
      "grad_norm": 0.6789435631032672,
      "learning_rate": 2.298535702278093e-06,
      "loss": 11.4286,
      "step": 21355
    },
    {
      "epoch": 3.734382513661202,
      "grad_norm": 0.6565736084175833,
      "learning_rate": 2.2955179745635434e-06,
      "loss": 11.4104,
      "step": 21356
    },
    {
      "epoch": 3.7345573770491804,
      "grad_norm": 0.6988801603189113,
      "learning_rate": 2.292502206123859e-06,
      "loss": 11.6829,
      "step": 21357
    },
    {
      "epoch": 3.7347322404371583,
      "grad_norm": 0.679071158072601,
      "learning_rate": 2.2894883970195012e-06,
      "loss": 11.4119,
      "step": 21358
    },
    {
      "epoch": 3.7349071038251367,
      "grad_norm": 0.6779980166569808,
      "learning_rate": 2.2864765473109117e-06,
      "loss": 11.4594,
      "step": 21359
    },
    {
      "epoch": 3.735081967213115,
      "grad_norm": 0.6838513610970369,
      "learning_rate": 2.2834666570584862e-06,
      "loss": 11.3334,
      "step": 21360
    },
    {
      "epoch": 3.735256830601093,
      "grad_norm": 0.7361471316126931,
      "learning_rate": 2.2804587263225874e-06,
      "loss": 11.3634,
      "step": 21361
    },
    {
      "epoch": 3.735431693989071,
      "grad_norm": 0.6385592984492917,
      "learning_rate": 2.2774527551635337e-06,
      "loss": 11.3545,
      "step": 21362
    },
    {
      "epoch": 3.735606557377049,
      "grad_norm": 0.7383120846765256,
      "learning_rate": 2.2744487436415886e-06,
      "loss": 11.3851,
      "step": 21363
    },
    {
      "epoch": 3.7357814207650275,
      "grad_norm": 0.6572511042387568,
      "learning_rate": 2.271446691817014e-06,
      "loss": 11.4044,
      "step": 21364
    },
    {
      "epoch": 3.7359562841530054,
      "grad_norm": 0.6979923977337336,
      "learning_rate": 2.268446599749985e-06,
      "loss": 11.2111,
      "step": 21365
    },
    {
      "epoch": 3.7361311475409837,
      "grad_norm": 0.6375738655874085,
      "learning_rate": 2.265448467500686e-06,
      "loss": 11.038,
      "step": 21366
    },
    {
      "epoch": 3.7363060109289616,
      "grad_norm": 0.6781871564742274,
      "learning_rate": 2.262452295129236e-06,
      "loss": 11.4592,
      "step": 21367
    },
    {
      "epoch": 3.73648087431694,
      "grad_norm": 0.7399415481509574,
      "learning_rate": 2.2594580826956977e-06,
      "loss": 11.2908,
      "step": 21368
    },
    {
      "epoch": 3.736655737704918,
      "grad_norm": 0.6856209740464911,
      "learning_rate": 2.256465830260135e-06,
      "loss": 11.2349,
      "step": 21369
    },
    {
      "epoch": 3.736830601092896,
      "grad_norm": 0.6598064679070521,
      "learning_rate": 2.2534755378825433e-06,
      "loss": 11.3377,
      "step": 21370
    },
    {
      "epoch": 3.737005464480874,
      "grad_norm": 0.6853360734085522,
      "learning_rate": 2.250487205622898e-06,
      "loss": 11.2216,
      "step": 21371
    },
    {
      "epoch": 3.7371803278688525,
      "grad_norm": 0.6555160359751695,
      "learning_rate": 2.247500833541105e-06,
      "loss": 11.5415,
      "step": 21372
    },
    {
      "epoch": 3.737355191256831,
      "grad_norm": 0.7446275542148189,
      "learning_rate": 2.2445164216970514e-06,
      "loss": 11.3692,
      "step": 21373
    },
    {
      "epoch": 3.7375300546448087,
      "grad_norm": 0.6433660341893329,
      "learning_rate": 2.241533970150611e-06,
      "loss": 11.5191,
      "step": 21374
    },
    {
      "epoch": 3.737704918032787,
      "grad_norm": 0.7554289814288451,
      "learning_rate": 2.2385534789615582e-06,
      "loss": 11.3725,
      "step": 21375
    },
    {
      "epoch": 3.737879781420765,
      "grad_norm": 0.7060566108209428,
      "learning_rate": 2.235574948189678e-06,
      "loss": 11.4292,
      "step": 21376
    },
    {
      "epoch": 3.7380546448087433,
      "grad_norm": 0.6508869695440045,
      "learning_rate": 2.2325983778947014e-06,
      "loss": 11.5693,
      "step": 21377
    },
    {
      "epoch": 3.738229508196721,
      "grad_norm": 0.6265645680805058,
      "learning_rate": 2.229623768136313e-06,
      "loss": 11.5783,
      "step": 21378
    },
    {
      "epoch": 3.7384043715846995,
      "grad_norm": 0.6829622758624286,
      "learning_rate": 2.2266511189741545e-06,
      "loss": 11.5696,
      "step": 21379
    },
    {
      "epoch": 3.7385792349726774,
      "grad_norm": 0.6571097258222995,
      "learning_rate": 2.223680430467845e-06,
      "loss": 11.4801,
      "step": 21380
    },
    {
      "epoch": 3.7387540983606558,
      "grad_norm": 0.6452452322615853,
      "learning_rate": 2.2207117026769696e-06,
      "loss": 11.327,
      "step": 21381
    },
    {
      "epoch": 3.738928961748634,
      "grad_norm": 0.7606311751307457,
      "learning_rate": 2.217744935661037e-06,
      "loss": 11.3371,
      "step": 21382
    },
    {
      "epoch": 3.739103825136612,
      "grad_norm": 0.6785028213115383,
      "learning_rate": 2.2147801294795432e-06,
      "loss": 11.2695,
      "step": 21383
    },
    {
      "epoch": 3.73927868852459,
      "grad_norm": 0.6820055336870653,
      "learning_rate": 2.2118172841919527e-06,
      "loss": 11.5225,
      "step": 21384
    },
    {
      "epoch": 3.7394535519125682,
      "grad_norm": 0.7595323945081106,
      "learning_rate": 2.2088563998576726e-06,
      "loss": 11.4381,
      "step": 21385
    },
    {
      "epoch": 3.7396284153005466,
      "grad_norm": 0.7521083565222643,
      "learning_rate": 2.2058974765361008e-06,
      "loss": 11.4645,
      "step": 21386
    },
    {
      "epoch": 3.7398032786885245,
      "grad_norm": 0.7486722088012028,
      "learning_rate": 2.2029405142865224e-06,
      "loss": 11.5105,
      "step": 21387
    },
    {
      "epoch": 3.739978142076503,
      "grad_norm": 0.70573479810943,
      "learning_rate": 2.1999855131682787e-06,
      "loss": 11.2238,
      "step": 21388
    },
    {
      "epoch": 3.7401530054644807,
      "grad_norm": 0.7396401679623501,
      "learning_rate": 2.1970324732406012e-06,
      "loss": 11.5586,
      "step": 21389
    },
    {
      "epoch": 3.740327868852459,
      "grad_norm": 0.6399393729518904,
      "learning_rate": 2.194081394562719e-06,
      "loss": 11.639,
      "step": 21390
    },
    {
      "epoch": 3.7405027322404374,
      "grad_norm": 0.7501050755091146,
      "learning_rate": 2.19113227719383e-06,
      "loss": 11.183,
      "step": 21391
    },
    {
      "epoch": 3.7406775956284153,
      "grad_norm": 0.6750511296250356,
      "learning_rate": 2.1881851211930204e-06,
      "loss": 11.4187,
      "step": 21392
    },
    {
      "epoch": 3.740852459016393,
      "grad_norm": 0.7248662031284706,
      "learning_rate": 2.1852399266194314e-06,
      "loss": 11.2324,
      "step": 21393
    },
    {
      "epoch": 3.7410273224043715,
      "grad_norm": 0.7089426032917737,
      "learning_rate": 2.1822966935321154e-06,
      "loss": 11.2841,
      "step": 21394
    },
    {
      "epoch": 3.74120218579235,
      "grad_norm": 0.7304563590402248,
      "learning_rate": 2.1793554219901034e-06,
      "loss": 11.5147,
      "step": 21395
    },
    {
      "epoch": 3.7413770491803278,
      "grad_norm": 0.6667331222045934,
      "learning_rate": 2.1764161120523484e-06,
      "loss": 11.3378,
      "step": 21396
    },
    {
      "epoch": 3.741551912568306,
      "grad_norm": 0.741809473328492,
      "learning_rate": 2.173478763777803e-06,
      "loss": 11.3767,
      "step": 21397
    },
    {
      "epoch": 3.741726775956284,
      "grad_norm": 0.5894526305841766,
      "learning_rate": 2.170543377225387e-06,
      "loss": 11.1924,
      "step": 21398
    },
    {
      "epoch": 3.7419016393442623,
      "grad_norm": 0.6994963578895362,
      "learning_rate": 2.167609952453953e-06,
      "loss": 11.309,
      "step": 21399
    },
    {
      "epoch": 3.7420765027322407,
      "grad_norm": 0.6418841742655821,
      "learning_rate": 2.1646784895223203e-06,
      "loss": 11.4768,
      "step": 21400
    },
    {
      "epoch": 3.7422513661202186,
      "grad_norm": 0.6425220560227772,
      "learning_rate": 2.161748988489276e-06,
      "loss": 11.4308,
      "step": 21401
    },
    {
      "epoch": 3.7424262295081965,
      "grad_norm": 0.6490609764011408,
      "learning_rate": 2.1588214494135615e-06,
      "loss": 11.4017,
      "step": 21402
    },
    {
      "epoch": 3.742601092896175,
      "grad_norm": 0.7047411263919068,
      "learning_rate": 2.155895872353897e-06,
      "loss": 11.5221,
      "step": 21403
    },
    {
      "epoch": 3.742775956284153,
      "grad_norm": 0.6295354305741403,
      "learning_rate": 2.152972257368957e-06,
      "loss": 11.2594,
      "step": 21404
    },
    {
      "epoch": 3.742950819672131,
      "grad_norm": 0.7380748124336757,
      "learning_rate": 2.1500506045173287e-06,
      "loss": 11.4473,
      "step": 21405
    },
    {
      "epoch": 3.7431256830601094,
      "grad_norm": 0.6903633484848702,
      "learning_rate": 2.1471309138576314e-06,
      "loss": 11.3821,
      "step": 21406
    },
    {
      "epoch": 3.7433005464480873,
      "grad_norm": 0.7028347377498799,
      "learning_rate": 2.144213185448407e-06,
      "loss": 11.3281,
      "step": 21407
    },
    {
      "epoch": 3.7434754098360656,
      "grad_norm": 0.6758660140380502,
      "learning_rate": 2.141297419348165e-06,
      "loss": 11.6282,
      "step": 21408
    },
    {
      "epoch": 3.743650273224044,
      "grad_norm": 0.6517685089485477,
      "learning_rate": 2.13838361561538e-06,
      "loss": 11.3784,
      "step": 21409
    },
    {
      "epoch": 3.743825136612022,
      "grad_norm": 0.6422491253725318,
      "learning_rate": 2.135471774308473e-06,
      "loss": 11.6285,
      "step": 21410
    },
    {
      "epoch": 3.7439999999999998,
      "grad_norm": 0.6476690730626866,
      "learning_rate": 2.132561895485841e-06,
      "loss": 11.151,
      "step": 21411
    },
    {
      "epoch": 3.744174863387978,
      "grad_norm": 0.6925164326216564,
      "learning_rate": 2.1296539792058368e-06,
      "loss": 11.4762,
      "step": 21412
    },
    {
      "epoch": 3.7443497267759565,
      "grad_norm": 0.6074273454613417,
      "learning_rate": 2.12674802552677e-06,
      "loss": 11.4858,
      "step": 21413
    },
    {
      "epoch": 3.7445245901639344,
      "grad_norm": 0.7128632023728257,
      "learning_rate": 2.123844034506928e-06,
      "loss": 11.5744,
      "step": 21414
    },
    {
      "epoch": 3.7446994535519127,
      "grad_norm": 0.6754804809553201,
      "learning_rate": 2.1209420062045182e-06,
      "loss": 11.6367,
      "step": 21415
    },
    {
      "epoch": 3.7448743169398906,
      "grad_norm": 0.661143362834265,
      "learning_rate": 2.11804194067774e-06,
      "loss": 11.5127,
      "step": 21416
    },
    {
      "epoch": 3.745049180327869,
      "grad_norm": 0.7223109066584589,
      "learning_rate": 2.1151438379847677e-06,
      "loss": 11.1385,
      "step": 21417
    },
    {
      "epoch": 3.745224043715847,
      "grad_norm": 0.7354925686406553,
      "learning_rate": 2.1122476981837226e-06,
      "loss": 11.3455,
      "step": 21418
    },
    {
      "epoch": 3.745398907103825,
      "grad_norm": 0.6545976574529349,
      "learning_rate": 2.1093535213326577e-06,
      "loss": 11.4054,
      "step": 21419
    },
    {
      "epoch": 3.745573770491803,
      "grad_norm": 0.6763929445720714,
      "learning_rate": 2.1064613074896155e-06,
      "loss": 11.4322,
      "step": 21420
    },
    {
      "epoch": 3.7457486338797814,
      "grad_norm": 0.6497990831150771,
      "learning_rate": 2.103571056712583e-06,
      "loss": 11.3218,
      "step": 21421
    },
    {
      "epoch": 3.7459234972677597,
      "grad_norm": 0.6793330006474788,
      "learning_rate": 2.100682769059548e-06,
      "loss": 11.415,
      "step": 21422
    },
    {
      "epoch": 3.7460983606557376,
      "grad_norm": 0.7450719910602619,
      "learning_rate": 2.097796444588418e-06,
      "loss": 11.3586,
      "step": 21423
    },
    {
      "epoch": 3.746273224043716,
      "grad_norm": 0.6674027677590296,
      "learning_rate": 2.094912083357059e-06,
      "loss": 11.5453,
      "step": 21424
    },
    {
      "epoch": 3.746448087431694,
      "grad_norm": 0.62268940801914,
      "learning_rate": 2.0920296854233137e-06,
      "loss": 11.6584,
      "step": 21425
    },
    {
      "epoch": 3.7466229508196722,
      "grad_norm": 0.7127286928335199,
      "learning_rate": 2.0891492508450015e-06,
      "loss": 11.0587,
      "step": 21426
    },
    {
      "epoch": 3.74679781420765,
      "grad_norm": 0.7152363428172519,
      "learning_rate": 2.0862707796798776e-06,
      "loss": 11.5219,
      "step": 21427
    },
    {
      "epoch": 3.7469726775956285,
      "grad_norm": 0.6939044431187444,
      "learning_rate": 2.08339427198565e-06,
      "loss": 11.1971,
      "step": 21428
    },
    {
      "epoch": 3.7471475409836064,
      "grad_norm": 0.6609713370382929,
      "learning_rate": 2.0805197278200184e-06,
      "loss": 11.3131,
      "step": 21429
    },
    {
      "epoch": 3.7473224043715847,
      "grad_norm": 0.6982839609639528,
      "learning_rate": 2.077647147240602e-06,
      "loss": 11.5248,
      "step": 21430
    },
    {
      "epoch": 3.747497267759563,
      "grad_norm": 0.6653140328532673,
      "learning_rate": 2.0747765303050447e-06,
      "loss": 11.5921,
      "step": 21431
    },
    {
      "epoch": 3.747672131147541,
      "grad_norm": 0.7206718840038289,
      "learning_rate": 2.0719078770708777e-06,
      "loss": 11.4973,
      "step": 21432
    },
    {
      "epoch": 3.7478469945355193,
      "grad_norm": 0.6813454861868644,
      "learning_rate": 2.0690411875956327e-06,
      "loss": 11.4579,
      "step": 21433
    },
    {
      "epoch": 3.748021857923497,
      "grad_norm": 0.6972410624501166,
      "learning_rate": 2.0661764619367975e-06,
      "loss": 11.4797,
      "step": 21434
    },
    {
      "epoch": 3.7481967213114755,
      "grad_norm": 0.730814413685684,
      "learning_rate": 2.0633137001518145e-06,
      "loss": 11.3036,
      "step": 21435
    },
    {
      "epoch": 3.7483715846994534,
      "grad_norm": 0.7127708360455195,
      "learning_rate": 2.0604529022981154e-06,
      "loss": 11.4452,
      "step": 21436
    },
    {
      "epoch": 3.7485464480874318,
      "grad_norm": 0.6677761351705317,
      "learning_rate": 2.057594068433044e-06,
      "loss": 11.3941,
      "step": 21437
    },
    {
      "epoch": 3.7487213114754097,
      "grad_norm": 0.7048670028211015,
      "learning_rate": 2.054737198613932e-06,
      "loss": 11.3065,
      "step": 21438
    },
    {
      "epoch": 3.748896174863388,
      "grad_norm": 0.7212138806854235,
      "learning_rate": 2.051882292898066e-06,
      "loss": 11.4153,
      "step": 21439
    },
    {
      "epoch": 3.7490710382513663,
      "grad_norm": 0.6616445216143957,
      "learning_rate": 2.0490293513427015e-06,
      "loss": 11.2578,
      "step": 21440
    },
    {
      "epoch": 3.7492459016393442,
      "grad_norm": 0.7571381088352966,
      "learning_rate": 2.046178374005059e-06,
      "loss": 11.1347,
      "step": 21441
    },
    {
      "epoch": 3.749420765027322,
      "grad_norm": 0.6674880693389299,
      "learning_rate": 2.043329360942281e-06,
      "loss": 11.4211,
      "step": 21442
    },
    {
      "epoch": 3.7495956284153005,
      "grad_norm": 0.6357396811011687,
      "learning_rate": 2.0404823122115223e-06,
      "loss": 11.5632,
      "step": 21443
    },
    {
      "epoch": 3.749770491803279,
      "grad_norm": 0.6664140600008738,
      "learning_rate": 2.037637227869871e-06,
      "loss": 11.4777,
      "step": 21444
    },
    {
      "epoch": 3.7499453551912567,
      "grad_norm": 0.5913440156533735,
      "learning_rate": 2.0347941079743804e-06,
      "loss": 11.7514,
      "step": 21445
    },
    {
      "epoch": 3.750120218579235,
      "grad_norm": 0.7323948589166555,
      "learning_rate": 2.0319529525820613e-06,
      "loss": 11.5225,
      "step": 21446
    },
    {
      "epoch": 3.750295081967213,
      "grad_norm": 0.6908934645253103,
      "learning_rate": 2.029113761749879e-06,
      "loss": 11.482,
      "step": 21447
    },
    {
      "epoch": 3.7504699453551913,
      "grad_norm": 0.7168308732828845,
      "learning_rate": 2.026276535534777e-06,
      "loss": 11.5216,
      "step": 21448
    },
    {
      "epoch": 3.7506448087431696,
      "grad_norm": 0.7721920705942769,
      "learning_rate": 2.0234412739936536e-06,
      "loss": 11.1444,
      "step": 21449
    },
    {
      "epoch": 3.7508196721311475,
      "grad_norm": 0.7139445088608454,
      "learning_rate": 2.020607977183353e-06,
      "loss": 11.4283,
      "step": 21450
    },
    {
      "epoch": 3.7509945355191254,
      "grad_norm": 0.6624158801958772,
      "learning_rate": 2.017776645160707e-06,
      "loss": 11.3439,
      "step": 21451
    },
    {
      "epoch": 3.7511693989071038,
      "grad_norm": 0.70691829268174,
      "learning_rate": 2.014947277982482e-06,
      "loss": 11.3776,
      "step": 21452
    },
    {
      "epoch": 3.751344262295082,
      "grad_norm": 0.7668730549795119,
      "learning_rate": 2.0121198757054094e-06,
      "loss": 11.4244,
      "step": 21453
    },
    {
      "epoch": 3.75151912568306,
      "grad_norm": 0.6759364425297492,
      "learning_rate": 2.009294438386211e-06,
      "loss": 11.3353,
      "step": 21454
    },
    {
      "epoch": 3.7516939890710383,
      "grad_norm": 0.7034825912036812,
      "learning_rate": 2.0064709660815194e-06,
      "loss": 11.4225,
      "step": 21455
    },
    {
      "epoch": 3.7518688524590162,
      "grad_norm": 0.7013167390151493,
      "learning_rate": 2.003649458847967e-06,
      "loss": 11.5599,
      "step": 21456
    },
    {
      "epoch": 3.7520437158469946,
      "grad_norm": 0.7565713034451546,
      "learning_rate": 2.0008299167421305e-06,
      "loss": 11.5755,
      "step": 21457
    },
    {
      "epoch": 3.752218579234973,
      "grad_norm": 0.6390043421289789,
      "learning_rate": 1.998012339820554e-06,
      "loss": 11.6267,
      "step": 21458
    },
    {
      "epoch": 3.752393442622951,
      "grad_norm": 0.821077388691662,
      "learning_rate": 1.9951967281397257e-06,
      "loss": 11.5312,
      "step": 21459
    },
    {
      "epoch": 3.7525683060109287,
      "grad_norm": 0.7169877067046636,
      "learning_rate": 1.992383081756133e-06,
      "loss": 11.2978,
      "step": 21460
    },
    {
      "epoch": 3.752743169398907,
      "grad_norm": 0.6717367499162762,
      "learning_rate": 1.9895714007261647e-06,
      "loss": 11.5065,
      "step": 21461
    },
    {
      "epoch": 3.7529180327868854,
      "grad_norm": 0.706834000386293,
      "learning_rate": 1.986761685106231e-06,
      "loss": 11.2358,
      "step": 21462
    },
    {
      "epoch": 3.7530928961748633,
      "grad_norm": 0.7240527300201695,
      "learning_rate": 1.9839539349526647e-06,
      "loss": 10.9613,
      "step": 21463
    },
    {
      "epoch": 3.7532677595628416,
      "grad_norm": 0.6265020503640039,
      "learning_rate": 1.9811481503217767e-06,
      "loss": 11.0244,
      "step": 21464
    },
    {
      "epoch": 3.7534426229508195,
      "grad_norm": 0.6846709393463283,
      "learning_rate": 1.9783443312698212e-06,
      "loss": 11.3454,
      "step": 21465
    },
    {
      "epoch": 3.753617486338798,
      "grad_norm": 0.7390079612620508,
      "learning_rate": 1.975542477853032e-06,
      "loss": 11.3361,
      "step": 21466
    },
    {
      "epoch": 3.7537923497267762,
      "grad_norm": 0.6663856998268509,
      "learning_rate": 1.9727425901275855e-06,
      "loss": 11.3643,
      "step": 21467
    },
    {
      "epoch": 3.753967213114754,
      "grad_norm": 0.7099288017732652,
      "learning_rate": 1.969944668149637e-06,
      "loss": 11.1282,
      "step": 21468
    },
    {
      "epoch": 3.754142076502732,
      "grad_norm": 0.6176529796799461,
      "learning_rate": 1.967148711975275e-06,
      "loss": 11.5357,
      "step": 21469
    },
    {
      "epoch": 3.7543169398907104,
      "grad_norm": 0.7075651632733087,
      "learning_rate": 1.9643547216605995e-06,
      "loss": 11.4634,
      "step": 21470
    },
    {
      "epoch": 3.7544918032786887,
      "grad_norm": 0.6235502125361586,
      "learning_rate": 1.9615626972616096e-06,
      "loss": 11.1511,
      "step": 21471
    },
    {
      "epoch": 3.7546666666666666,
      "grad_norm": 0.6424076589894574,
      "learning_rate": 1.9587726388343163e-06,
      "loss": 11.2709,
      "step": 21472
    },
    {
      "epoch": 3.754841530054645,
      "grad_norm": 0.6652294440938422,
      "learning_rate": 1.9559845464346525e-06,
      "loss": 11.2422,
      "step": 21473
    },
    {
      "epoch": 3.755016393442623,
      "grad_norm": 0.661533755809313,
      "learning_rate": 1.9531984201185295e-06,
      "loss": 11.5422,
      "step": 21474
    },
    {
      "epoch": 3.755191256830601,
      "grad_norm": 0.6748971461364276,
      "learning_rate": 1.950414259941824e-06,
      "loss": 11.405,
      "step": 21475
    },
    {
      "epoch": 3.755366120218579,
      "grad_norm": 0.6688765108659914,
      "learning_rate": 1.9476320659603697e-06,
      "loss": 11.4001,
      "step": 21476
    },
    {
      "epoch": 3.7555409836065574,
      "grad_norm": 0.6172530081685796,
      "learning_rate": 1.9448518382299553e-06,
      "loss": 11.4297,
      "step": 21477
    },
    {
      "epoch": 3.7557158469945353,
      "grad_norm": 0.633754012954905,
      "learning_rate": 1.942073576806314e-06,
      "loss": 11.2907,
      "step": 21478
    },
    {
      "epoch": 3.7558907103825137,
      "grad_norm": 0.6710242584523609,
      "learning_rate": 1.939297281745178e-06,
      "loss": 11.5074,
      "step": 21479
    },
    {
      "epoch": 3.756065573770492,
      "grad_norm": 0.6122037888169032,
      "learning_rate": 1.9365229531022264e-06,
      "loss": 11.2968,
      "step": 21480
    },
    {
      "epoch": 3.75624043715847,
      "grad_norm": 0.6340209138529594,
      "learning_rate": 1.933750590933081e-06,
      "loss": 11.4411,
      "step": 21481
    },
    {
      "epoch": 3.7564153005464482,
      "grad_norm": 0.7316521620897845,
      "learning_rate": 1.9309801952933416e-06,
      "loss": 11.1801,
      "step": 21482
    },
    {
      "epoch": 3.756590163934426,
      "grad_norm": 0.666341015392755,
      "learning_rate": 1.9282117662385526e-06,
      "loss": 11.557,
      "step": 21483
    },
    {
      "epoch": 3.7567650273224045,
      "grad_norm": 0.8510529369230584,
      "learning_rate": 1.925445303824247e-06,
      "loss": 11.494,
      "step": 21484
    },
    {
      "epoch": 3.7569398907103824,
      "grad_norm": 0.6994600772019016,
      "learning_rate": 1.9226808081058813e-06,
      "loss": 11.3545,
      "step": 21485
    },
    {
      "epoch": 3.7571147540983607,
      "grad_norm": 0.6850422074856498,
      "learning_rate": 1.9199182791389214e-06,
      "loss": 11.1247,
      "step": 21486
    },
    {
      "epoch": 3.7572896174863386,
      "grad_norm": 0.6869821787518156,
      "learning_rate": 1.9171577169787347e-06,
      "loss": 11.5448,
      "step": 21487
    },
    {
      "epoch": 3.757464480874317,
      "grad_norm": 0.6221384552753642,
      "learning_rate": 1.914399121680677e-06,
      "loss": 11.6725,
      "step": 21488
    },
    {
      "epoch": 3.7576393442622953,
      "grad_norm": 0.6487608543452544,
      "learning_rate": 1.911642493300092e-06,
      "loss": 11.3205,
      "step": 21489
    },
    {
      "epoch": 3.757814207650273,
      "grad_norm": 0.6442929669856715,
      "learning_rate": 1.9088878318922366e-06,
      "loss": 11.0487,
      "step": 21490
    },
    {
      "epoch": 3.7579890710382515,
      "grad_norm": 0.6858951327720983,
      "learning_rate": 1.9061351375123771e-06,
      "loss": 11.3998,
      "step": 21491
    },
    {
      "epoch": 3.7581639344262294,
      "grad_norm": 0.6720472183278692,
      "learning_rate": 1.9033844102156917e-06,
      "loss": 11.5716,
      "step": 21492
    },
    {
      "epoch": 3.7583387978142078,
      "grad_norm": 0.6755443133249575,
      "learning_rate": 1.9006356500573253e-06,
      "loss": 11.5785,
      "step": 21493
    },
    {
      "epoch": 3.7585136612021857,
      "grad_norm": 0.6664989732042959,
      "learning_rate": 1.8978888570924335e-06,
      "loss": 11.4117,
      "step": 21494
    },
    {
      "epoch": 3.758688524590164,
      "grad_norm": 0.7124610802597832,
      "learning_rate": 1.8951440313760837e-06,
      "loss": 11.4765,
      "step": 21495
    },
    {
      "epoch": 3.758863387978142,
      "grad_norm": 0.6972327090250509,
      "learning_rate": 1.8924011729633207e-06,
      "loss": 11.3185,
      "step": 21496
    },
    {
      "epoch": 3.7590382513661202,
      "grad_norm": 0.6630352717885192,
      "learning_rate": 1.8896602819091335e-06,
      "loss": 11.5142,
      "step": 21497
    },
    {
      "epoch": 3.7592131147540986,
      "grad_norm": 0.6939803431373511,
      "learning_rate": 1.8869213582685008e-06,
      "loss": 11.4389,
      "step": 21498
    },
    {
      "epoch": 3.7593879781420765,
      "grad_norm": 0.6566190934088517,
      "learning_rate": 1.884184402096334e-06,
      "loss": 11.5148,
      "step": 21499
    },
    {
      "epoch": 3.7595628415300544,
      "grad_norm": 0.7401032949538423,
      "learning_rate": 1.8814494134475446e-06,
      "loss": 11.6657,
      "step": 21500
    },
    {
      "epoch": 3.7597377049180327,
      "grad_norm": 0.6935176976400149,
      "learning_rate": 1.8787163923769446e-06,
      "loss": 11.4011,
      "step": 21501
    },
    {
      "epoch": 3.759912568306011,
      "grad_norm": 0.6952802233771768,
      "learning_rate": 1.8759853389393457e-06,
      "loss": 11.3297,
      "step": 21502
    },
    {
      "epoch": 3.760087431693989,
      "grad_norm": 0.6054173156662545,
      "learning_rate": 1.8732562531895371e-06,
      "loss": 11.557,
      "step": 21503
    },
    {
      "epoch": 3.7602622950819673,
      "grad_norm": 0.6924309523883603,
      "learning_rate": 1.8705291351822307e-06,
      "loss": 11.5296,
      "step": 21504
    },
    {
      "epoch": 3.760437158469945,
      "grad_norm": 0.7052511246907718,
      "learning_rate": 1.8678039849720942e-06,
      "loss": 11.3563,
      "step": 21505
    },
    {
      "epoch": 3.7606120218579235,
      "grad_norm": 0.688020442653259,
      "learning_rate": 1.8650808026138056e-06,
      "loss": 11.2571,
      "step": 21506
    },
    {
      "epoch": 3.760786885245902,
      "grad_norm": 0.6435078287852349,
      "learning_rate": 1.862359588161955e-06,
      "loss": 11.4977,
      "step": 21507
    },
    {
      "epoch": 3.7609617486338798,
      "grad_norm": 0.7340450321545594,
      "learning_rate": 1.8596403416711206e-06,
      "loss": 11.2499,
      "step": 21508
    },
    {
      "epoch": 3.7611366120218577,
      "grad_norm": 0.6953458114334911,
      "learning_rate": 1.8569230631958256e-06,
      "loss": 11.6337,
      "step": 21509
    },
    {
      "epoch": 3.761311475409836,
      "grad_norm": 0.6770998996185141,
      "learning_rate": 1.854207752790571e-06,
      "loss": 11.4684,
      "step": 21510
    },
    {
      "epoch": 3.7614863387978144,
      "grad_norm": 0.6706572169763056,
      "learning_rate": 1.8514944105097908e-06,
      "loss": 11.358,
      "step": 21511
    },
    {
      "epoch": 3.7616612021857923,
      "grad_norm": 0.6956084841605262,
      "learning_rate": 1.8487830364078862e-06,
      "loss": 11.0388,
      "step": 21512
    },
    {
      "epoch": 3.7618360655737706,
      "grad_norm": 0.6710411326520045,
      "learning_rate": 1.84607363053928e-06,
      "loss": 11.3525,
      "step": 21513
    },
    {
      "epoch": 3.7620109289617485,
      "grad_norm": 0.6716510230504811,
      "learning_rate": 1.8433661929582513e-06,
      "loss": 11.4625,
      "step": 21514
    },
    {
      "epoch": 3.762185792349727,
      "grad_norm": 0.7180600517935983,
      "learning_rate": 1.840660723719112e-06,
      "loss": 11.4366,
      "step": 21515
    },
    {
      "epoch": 3.762360655737705,
      "grad_norm": 0.6254611476969246,
      "learning_rate": 1.8379572228761076e-06,
      "loss": 11.1775,
      "step": 21516
    },
    {
      "epoch": 3.762535519125683,
      "grad_norm": 0.6362072422534971,
      "learning_rate": 1.8352556904834506e-06,
      "loss": 11.3463,
      "step": 21517
    },
    {
      "epoch": 3.762710382513661,
      "grad_norm": 0.7540359027192931,
      "learning_rate": 1.832556126595353e-06,
      "loss": 11.6338,
      "step": 21518
    },
    {
      "epoch": 3.7628852459016393,
      "grad_norm": 0.9078894181729021,
      "learning_rate": 1.8298585312658932e-06,
      "loss": 11.2122,
      "step": 21519
    },
    {
      "epoch": 3.7630601092896176,
      "grad_norm": 0.6623024135174911,
      "learning_rate": 1.8271629045492066e-06,
      "loss": 11.3318,
      "step": 21520
    },
    {
      "epoch": 3.7632349726775955,
      "grad_norm": 0.6850662318727456,
      "learning_rate": 1.8244692464993163e-06,
      "loss": 11.6768,
      "step": 21521
    },
    {
      "epoch": 3.763409836065574,
      "grad_norm": 0.6977544190124761,
      "learning_rate": 1.8217775571702677e-06,
      "loss": 11.5331,
      "step": 21522
    },
    {
      "epoch": 3.763584699453552,
      "grad_norm": 0.6590388500959583,
      "learning_rate": 1.819087836616029e-06,
      "loss": 11.3293,
      "step": 21523
    },
    {
      "epoch": 3.76375956284153,
      "grad_norm": 0.6512229257558674,
      "learning_rate": 1.8164000848905348e-06,
      "loss": 11.4609,
      "step": 21524
    },
    {
      "epoch": 3.7639344262295085,
      "grad_norm": 0.6728145327142758,
      "learning_rate": 1.8137143020476753e-06,
      "loss": 10.9548,
      "step": 21525
    },
    {
      "epoch": 3.7641092896174864,
      "grad_norm": 0.6215214174199649,
      "learning_rate": 1.8110304881413075e-06,
      "loss": 11.5661,
      "step": 21526
    },
    {
      "epoch": 3.7642841530054643,
      "grad_norm": 0.7138325503695168,
      "learning_rate": 1.8083486432252772e-06,
      "loss": 11.1959,
      "step": 21527
    },
    {
      "epoch": 3.7644590163934426,
      "grad_norm": 0.6348156877161484,
      "learning_rate": 1.8056687673533413e-06,
      "loss": 11.5384,
      "step": 21528
    },
    {
      "epoch": 3.764633879781421,
      "grad_norm": 0.7093787826026086,
      "learning_rate": 1.8029908605792344e-06,
      "loss": 11.3212,
      "step": 21529
    },
    {
      "epoch": 3.764808743169399,
      "grad_norm": 0.643074027182809,
      "learning_rate": 1.8003149229566696e-06,
      "loss": 11.4093,
      "step": 21530
    },
    {
      "epoch": 3.764983606557377,
      "grad_norm": 0.6889665573226249,
      "learning_rate": 1.7976409545392924e-06,
      "loss": 11.2969,
      "step": 21531
    },
    {
      "epoch": 3.765158469945355,
      "grad_norm": 0.7397293937686599,
      "learning_rate": 1.7949689553807493e-06,
      "loss": 11.3199,
      "step": 21532
    },
    {
      "epoch": 3.7653333333333334,
      "grad_norm": 0.7357676318408418,
      "learning_rate": 1.7922989255345968e-06,
      "loss": 11.536,
      "step": 21533
    },
    {
      "epoch": 3.7655081967213113,
      "grad_norm": 0.6790089263576833,
      "learning_rate": 1.7896308650544037e-06,
      "loss": 11.2932,
      "step": 21534
    },
    {
      "epoch": 3.7656830601092897,
      "grad_norm": 0.681499829180873,
      "learning_rate": 1.7869647739936378e-06,
      "loss": 11.3768,
      "step": 21535
    },
    {
      "epoch": 3.7658579234972676,
      "grad_norm": 0.6280887172116271,
      "learning_rate": 1.7843006524058015e-06,
      "loss": 11.4694,
      "step": 21536
    },
    {
      "epoch": 3.766032786885246,
      "grad_norm": 0.6239330034002,
      "learning_rate": 1.7816385003442847e-06,
      "loss": 11.3759,
      "step": 21537
    },
    {
      "epoch": 3.7662076502732242,
      "grad_norm": 0.6806628173457208,
      "learning_rate": 1.7789783178624897e-06,
      "loss": 10.9768,
      "step": 21538
    },
    {
      "epoch": 3.766382513661202,
      "grad_norm": 0.6679882014921489,
      "learning_rate": 1.7763201050137512e-06,
      "loss": 11.308,
      "step": 21539
    },
    {
      "epoch": 3.7665573770491805,
      "grad_norm": 0.6410843648667973,
      "learning_rate": 1.7736638618513934e-06,
      "loss": 11.6117,
      "step": 21540
    },
    {
      "epoch": 3.7667322404371584,
      "grad_norm": 0.6568205238672742,
      "learning_rate": 1.7710095884286625e-06,
      "loss": 11.0701,
      "step": 21541
    },
    {
      "epoch": 3.7669071038251367,
      "grad_norm": 0.6532423296461041,
      "learning_rate": 1.768357284798794e-06,
      "loss": 11.2958,
      "step": 21542
    },
    {
      "epoch": 3.7670819672131146,
      "grad_norm": 0.6667030176860484,
      "learning_rate": 1.765706951014967e-06,
      "loss": 11.2217,
      "step": 21543
    },
    {
      "epoch": 3.767256830601093,
      "grad_norm": 0.7003109654300508,
      "learning_rate": 1.7630585871303397e-06,
      "loss": 11.4996,
      "step": 21544
    },
    {
      "epoch": 3.767431693989071,
      "grad_norm": 0.6569774321056765,
      "learning_rate": 1.7604121931980134e-06,
      "loss": 11.682,
      "step": 21545
    },
    {
      "epoch": 3.767606557377049,
      "grad_norm": 0.6059187181485112,
      "learning_rate": 1.7577677692710681e-06,
      "loss": 11.4225,
      "step": 21546
    },
    {
      "epoch": 3.7677814207650275,
      "grad_norm": 0.6620409602280161,
      "learning_rate": 1.755125315402506e-06,
      "loss": 10.874,
      "step": 21547
    },
    {
      "epoch": 3.7679562841530054,
      "grad_norm": 0.6773063388250433,
      "learning_rate": 1.75248483164534e-06,
      "loss": 11.5281,
      "step": 21548
    },
    {
      "epoch": 3.7681311475409838,
      "grad_norm": 0.6871279597863346,
      "learning_rate": 1.7498463180525172e-06,
      "loss": 11.4383,
      "step": 21549
    },
    {
      "epoch": 3.7683060109289617,
      "grad_norm": 0.65365663583171,
      "learning_rate": 1.7472097746769388e-06,
      "loss": 11.4784,
      "step": 21550
    },
    {
      "epoch": 3.76848087431694,
      "grad_norm": 0.7016035583276262,
      "learning_rate": 1.7445752015714744e-06,
      "loss": 11.1507,
      "step": 21551
    },
    {
      "epoch": 3.768655737704918,
      "grad_norm": 0.7529453224603021,
      "learning_rate": 1.7419425987889705e-06,
      "loss": 11.5593,
      "step": 21552
    },
    {
      "epoch": 3.7688306010928962,
      "grad_norm": 0.7024536016329234,
      "learning_rate": 1.7393119663822067e-06,
      "loss": 11.5285,
      "step": 21553
    },
    {
      "epoch": 3.769005464480874,
      "grad_norm": 0.7278878630242667,
      "learning_rate": 1.73668330440393e-06,
      "loss": 11.2957,
      "step": 21554
    },
    {
      "epoch": 3.7691803278688525,
      "grad_norm": 0.6419425521299015,
      "learning_rate": 1.7340566129068758e-06,
      "loss": 11.5567,
      "step": 21555
    },
    {
      "epoch": 3.769355191256831,
      "grad_norm": 0.7159777049088306,
      "learning_rate": 1.731431891943691e-06,
      "loss": 11.2516,
      "step": 21556
    },
    {
      "epoch": 3.7695300546448087,
      "grad_norm": 0.6302970374116071,
      "learning_rate": 1.7288091415670226e-06,
      "loss": 11.5696,
      "step": 21557
    },
    {
      "epoch": 3.7697049180327866,
      "grad_norm": 0.7298669986503871,
      "learning_rate": 1.7261883618294616e-06,
      "loss": 11.346,
      "step": 21558
    },
    {
      "epoch": 3.769879781420765,
      "grad_norm": 0.6980907554471797,
      "learning_rate": 1.723569552783566e-06,
      "loss": 11.324,
      "step": 21559
    },
    {
      "epoch": 3.7700546448087433,
      "grad_norm": 0.6968395824023499,
      "learning_rate": 1.7209527144818494e-06,
      "loss": 11.3563,
      "step": 21560
    },
    {
      "epoch": 3.770229508196721,
      "grad_norm": 0.6878251951797202,
      "learning_rate": 1.7183378469767918e-06,
      "loss": 11.4357,
      "step": 21561
    },
    {
      "epoch": 3.7704043715846995,
      "grad_norm": 0.6636480431192984,
      "learning_rate": 1.7157249503208184e-06,
      "loss": 11.3194,
      "step": 21562
    },
    {
      "epoch": 3.7705792349726774,
      "grad_norm": 0.6915449490010361,
      "learning_rate": 1.7131140245663423e-06,
      "loss": 11.2803,
      "step": 21563
    },
    {
      "epoch": 3.770754098360656,
      "grad_norm": 0.6817864609652529,
      "learning_rate": 1.710505069765711e-06,
      "loss": 11.4786,
      "step": 21564
    },
    {
      "epoch": 3.770928961748634,
      "grad_norm": 0.6390559740698334,
      "learning_rate": 1.7078980859712157e-06,
      "loss": 11.4592,
      "step": 21565
    },
    {
      "epoch": 3.771103825136612,
      "grad_norm": 0.6490675481529182,
      "learning_rate": 1.7052930732351812e-06,
      "loss": 11.3853,
      "step": 21566
    },
    {
      "epoch": 3.77127868852459,
      "grad_norm": 0.7053775229196716,
      "learning_rate": 1.7026900316098215e-06,
      "loss": 11.51,
      "step": 21567
    },
    {
      "epoch": 3.7714535519125683,
      "grad_norm": 0.6571011104206319,
      "learning_rate": 1.7000889611473392e-06,
      "loss": 11.1774,
      "step": 21568
    },
    {
      "epoch": 3.7716284153005466,
      "grad_norm": 0.6701957274447885,
      "learning_rate": 1.6974898618998925e-06,
      "loss": 11.3295,
      "step": 21569
    },
    {
      "epoch": 3.7718032786885245,
      "grad_norm": 0.7034320088192494,
      "learning_rate": 1.6948927339196064e-06,
      "loss": 11.1548,
      "step": 21570
    },
    {
      "epoch": 3.771978142076503,
      "grad_norm": 0.7286530886901558,
      "learning_rate": 1.6922975772585503e-06,
      "loss": 11.3212,
      "step": 21571
    },
    {
      "epoch": 3.7721530054644807,
      "grad_norm": 0.6845861347425851,
      "learning_rate": 1.6897043919687828e-06,
      "loss": 11.2945,
      "step": 21572
    },
    {
      "epoch": 3.772327868852459,
      "grad_norm": 0.6427617683764657,
      "learning_rate": 1.6871131781022954e-06,
      "loss": 11.3199,
      "step": 21573
    },
    {
      "epoch": 3.7725027322404374,
      "grad_norm": 0.7010891420916078,
      "learning_rate": 1.6845239357110243e-06,
      "loss": 11.5703,
      "step": 21574
    },
    {
      "epoch": 3.7726775956284153,
      "grad_norm": 0.634075440760912,
      "learning_rate": 1.6819366648469393e-06,
      "loss": 11.2213,
      "step": 21575
    },
    {
      "epoch": 3.772852459016393,
      "grad_norm": 0.6361562082315715,
      "learning_rate": 1.6793513655618986e-06,
      "loss": 11.6298,
      "step": 21576
    },
    {
      "epoch": 3.7730273224043716,
      "grad_norm": 0.7004054235837236,
      "learning_rate": 1.676768037907739e-06,
      "loss": 11.1838,
      "step": 21577
    },
    {
      "epoch": 3.77320218579235,
      "grad_norm": 0.6767156102570924,
      "learning_rate": 1.674186681936285e-06,
      "loss": 11.4672,
      "step": 21578
    },
    {
      "epoch": 3.773377049180328,
      "grad_norm": 0.6726295092935075,
      "learning_rate": 1.6716072976992737e-06,
      "loss": 11.5182,
      "step": 21579
    },
    {
      "epoch": 3.773551912568306,
      "grad_norm": 0.7303492068477676,
      "learning_rate": 1.669029885248441e-06,
      "loss": 11.1912,
      "step": 21580
    },
    {
      "epoch": 3.773726775956284,
      "grad_norm": 0.6581821435397665,
      "learning_rate": 1.6664544446354903e-06,
      "loss": 11.3111,
      "step": 21581
    },
    {
      "epoch": 3.7739016393442624,
      "grad_norm": 0.7269528826363139,
      "learning_rate": 1.6638809759120465e-06,
      "loss": 11.204,
      "step": 21582
    },
    {
      "epoch": 3.7740765027322407,
      "grad_norm": 0.7044674226710496,
      "learning_rate": 1.6613094791297246e-06,
      "loss": 11.6143,
      "step": 21583
    },
    {
      "epoch": 3.7742513661202186,
      "grad_norm": 0.6839859262092408,
      "learning_rate": 1.6587399543400716e-06,
      "loss": 11.5419,
      "step": 21584
    },
    {
      "epoch": 3.7744262295081965,
      "grad_norm": 0.7203936838218816,
      "learning_rate": 1.6561724015946356e-06,
      "loss": 11.5699,
      "step": 21585
    },
    {
      "epoch": 3.774601092896175,
      "grad_norm": 0.6802004758662061,
      "learning_rate": 1.6536068209449086e-06,
      "loss": 11.4432,
      "step": 21586
    },
    {
      "epoch": 3.774775956284153,
      "grad_norm": 0.6844523411704712,
      "learning_rate": 1.651043212442327e-06,
      "loss": 11.3744,
      "step": 21587
    },
    {
      "epoch": 3.774950819672131,
      "grad_norm": 0.6927932350344567,
      "learning_rate": 1.6484815761382943e-06,
      "loss": 11.3696,
      "step": 21588
    },
    {
      "epoch": 3.7751256830601094,
      "grad_norm": 0.6371405290609848,
      "learning_rate": 1.6459219120841695e-06,
      "loss": 11.4557,
      "step": 21589
    },
    {
      "epoch": 3.7753005464480873,
      "grad_norm": 0.7248644484837193,
      "learning_rate": 1.6433642203313115e-06,
      "loss": 11.3817,
      "step": 21590
    },
    {
      "epoch": 3.7754754098360657,
      "grad_norm": 0.6511405721477078,
      "learning_rate": 1.6408085009310126e-06,
      "loss": 11.1919,
      "step": 21591
    },
    {
      "epoch": 3.7756502732240436,
      "grad_norm": 0.6831821573344856,
      "learning_rate": 1.6382547539344873e-06,
      "loss": 11.474,
      "step": 21592
    },
    {
      "epoch": 3.775825136612022,
      "grad_norm": 0.6295552422088565,
      "learning_rate": 1.6357029793929612e-06,
      "loss": 11.4002,
      "step": 21593
    },
    {
      "epoch": 3.776,
      "grad_norm": 0.7035462326556745,
      "learning_rate": 1.6331531773576048e-06,
      "loss": 11.3756,
      "step": 21594
    },
    {
      "epoch": 3.776174863387978,
      "grad_norm": 0.62780767028213,
      "learning_rate": 1.6306053478795657e-06,
      "loss": 11.2823,
      "step": 21595
    },
    {
      "epoch": 3.7763497267759565,
      "grad_norm": 0.6783690462516025,
      "learning_rate": 1.6280594910099256e-06,
      "loss": 11.3935,
      "step": 21596
    },
    {
      "epoch": 3.7765245901639344,
      "grad_norm": 0.6501273653623753,
      "learning_rate": 1.6255156067997323e-06,
      "loss": 11.4731,
      "step": 21597
    },
    {
      "epoch": 3.7766994535519127,
      "grad_norm": 0.6912130654227829,
      "learning_rate": 1.6229736952999896e-06,
      "loss": 11.2315,
      "step": 21598
    },
    {
      "epoch": 3.7768743169398906,
      "grad_norm": 0.6803718570611734,
      "learning_rate": 1.620433756561679e-06,
      "loss": 11.0398,
      "step": 21599
    },
    {
      "epoch": 3.777049180327869,
      "grad_norm": 0.6392657951443943,
      "learning_rate": 1.6178957906357594e-06,
      "loss": 11.2816,
      "step": 21600
    },
    {
      "epoch": 3.777224043715847,
      "grad_norm": 0.6493935367633835,
      "learning_rate": 1.6153597975730794e-06,
      "loss": 11.549,
      "step": 21601
    },
    {
      "epoch": 3.777398907103825,
      "grad_norm": 0.6893180597619989,
      "learning_rate": 1.6128257774245315e-06,
      "loss": 11.5569,
      "step": 21602
    },
    {
      "epoch": 3.777573770491803,
      "grad_norm": 0.7214742827503464,
      "learning_rate": 1.6102937302408972e-06,
      "loss": 11.5251,
      "step": 21603
    },
    {
      "epoch": 3.7777486338797814,
      "grad_norm": 0.7495168562668619,
      "learning_rate": 1.6077636560729803e-06,
      "loss": 11.2512,
      "step": 21604
    },
    {
      "epoch": 3.77792349726776,
      "grad_norm": 0.7050120528443321,
      "learning_rate": 1.6052355549715181e-06,
      "loss": 11.3872,
      "step": 21605
    },
    {
      "epoch": 3.7780983606557377,
      "grad_norm": 0.6636958380995638,
      "learning_rate": 1.6027094269871811e-06,
      "loss": 11.2904,
      "step": 21606
    },
    {
      "epoch": 3.7782732240437156,
      "grad_norm": 0.7088707933398632,
      "learning_rate": 1.6001852721706402e-06,
      "loss": 11.1563,
      "step": 21607
    },
    {
      "epoch": 3.778448087431694,
      "grad_norm": 0.7079571941593267,
      "learning_rate": 1.5976630905724987e-06,
      "loss": 11.4106,
      "step": 21608
    },
    {
      "epoch": 3.7786229508196723,
      "grad_norm": 0.6898393210095863,
      "learning_rate": 1.5951428822433724e-06,
      "loss": 11.303,
      "step": 21609
    },
    {
      "epoch": 3.77879781420765,
      "grad_norm": 0.6713507689010229,
      "learning_rate": 1.5926246472337647e-06,
      "loss": 11.3682,
      "step": 21610
    },
    {
      "epoch": 3.7789726775956285,
      "grad_norm": 0.6769082989608749,
      "learning_rate": 1.5901083855941801e-06,
      "loss": 11.3616,
      "step": 21611
    },
    {
      "epoch": 3.7791475409836064,
      "grad_norm": 0.6977764198034261,
      "learning_rate": 1.587594097375078e-06,
      "loss": 11.2657,
      "step": 21612
    },
    {
      "epoch": 3.7793224043715847,
      "grad_norm": 0.6523044853906493,
      "learning_rate": 1.5850817826268737e-06,
      "loss": 11.3875,
      "step": 21613
    },
    {
      "epoch": 3.779497267759563,
      "grad_norm": 0.6195460477856113,
      "learning_rate": 1.5825714413999714e-06,
      "loss": 11.3662,
      "step": 21614
    },
    {
      "epoch": 3.779672131147541,
      "grad_norm": 0.6426613768112462,
      "learning_rate": 1.5800630737446865e-06,
      "loss": 11.2112,
      "step": 21615
    },
    {
      "epoch": 3.779846994535519,
      "grad_norm": 0.6968337846779231,
      "learning_rate": 1.5775566797113228e-06,
      "loss": 11.0583,
      "step": 21616
    },
    {
      "epoch": 3.780021857923497,
      "grad_norm": 0.7061575532074598,
      "learning_rate": 1.5750522593501404e-06,
      "loss": 11.3576,
      "step": 21617
    },
    {
      "epoch": 3.7801967213114755,
      "grad_norm": 0.6279458123987175,
      "learning_rate": 1.5725498127113769e-06,
      "loss": 11.2874,
      "step": 21618
    },
    {
      "epoch": 3.7803715846994534,
      "grad_norm": 0.6350759352149412,
      "learning_rate": 1.570049339845192e-06,
      "loss": 11.3374,
      "step": 21619
    },
    {
      "epoch": 3.780546448087432,
      "grad_norm": 0.6820347620513371,
      "learning_rate": 1.5675508408017347e-06,
      "loss": 11.4933,
      "step": 21620
    },
    {
      "epoch": 3.7807213114754097,
      "grad_norm": 0.6767392086404163,
      "learning_rate": 1.5650543156311205e-06,
      "loss": 11.4097,
      "step": 21621
    },
    {
      "epoch": 3.780896174863388,
      "grad_norm": 0.6837462544477698,
      "learning_rate": 1.5625597643833866e-06,
      "loss": 11.2201,
      "step": 21622
    },
    {
      "epoch": 3.7810710382513664,
      "grad_norm": 0.6796876136010264,
      "learning_rate": 1.5600671871085936e-06,
      "loss": 11.5639,
      "step": 21623
    },
    {
      "epoch": 3.7812459016393443,
      "grad_norm": 0.6984108849573015,
      "learning_rate": 1.5575765838566902e-06,
      "loss": 11.5857,
      "step": 21624
    },
    {
      "epoch": 3.781420765027322,
      "grad_norm": 0.6492843841294391,
      "learning_rate": 1.5550879546776364e-06,
      "loss": 11.3994,
      "step": 21625
    },
    {
      "epoch": 3.7815956284153005,
      "grad_norm": 0.6759240548610096,
      "learning_rate": 1.552601299621337e-06,
      "loss": 11.4221,
      "step": 21626
    },
    {
      "epoch": 3.781770491803279,
      "grad_norm": 0.6596801141090902,
      "learning_rate": 1.5501166187376515e-06,
      "loss": 11.3804,
      "step": 21627
    },
    {
      "epoch": 3.7819453551912567,
      "grad_norm": 0.6374436256709989,
      "learning_rate": 1.5476339120764072e-06,
      "loss": 11.3069,
      "step": 21628
    },
    {
      "epoch": 3.782120218579235,
      "grad_norm": 0.6674797572950522,
      "learning_rate": 1.5451531796873974e-06,
      "loss": 11.7213,
      "step": 21629
    },
    {
      "epoch": 3.782295081967213,
      "grad_norm": 0.6900058507452765,
      "learning_rate": 1.5426744216203493e-06,
      "loss": 11.624,
      "step": 21630
    },
    {
      "epoch": 3.7824699453551913,
      "grad_norm": 0.6960221251548724,
      "learning_rate": 1.5401976379250005e-06,
      "loss": 11.2025,
      "step": 21631
    },
    {
      "epoch": 3.7826448087431697,
      "grad_norm": 0.6995754168759281,
      "learning_rate": 1.5377228286509892e-06,
      "loss": 11.3925,
      "step": 21632
    },
    {
      "epoch": 3.7828196721311476,
      "grad_norm": 0.6864764641284659,
      "learning_rate": 1.5352499938479536e-06,
      "loss": 11.2244,
      "step": 21633
    },
    {
      "epoch": 3.7829945355191255,
      "grad_norm": 0.7807618044776586,
      "learning_rate": 1.5327791335654872e-06,
      "loss": 11.3279,
      "step": 21634
    },
    {
      "epoch": 3.783169398907104,
      "grad_norm": 0.6176381997439405,
      "learning_rate": 1.5303102478531284e-06,
      "loss": 11.4961,
      "step": 21635
    },
    {
      "epoch": 3.783344262295082,
      "grad_norm": 0.7220468801342399,
      "learning_rate": 1.5278433367603928e-06,
      "loss": 11.1604,
      "step": 21636
    },
    {
      "epoch": 3.78351912568306,
      "grad_norm": 0.717605699795073,
      "learning_rate": 1.5253784003367522e-06,
      "loss": 11.3734,
      "step": 21637
    },
    {
      "epoch": 3.7836939890710384,
      "grad_norm": 0.6547209212465848,
      "learning_rate": 1.5229154386316335e-06,
      "loss": 11.3342,
      "step": 21638
    },
    {
      "epoch": 3.7838688524590163,
      "grad_norm": 0.7525264684936224,
      "learning_rate": 1.5204544516944198e-06,
      "loss": 11.4215,
      "step": 21639
    },
    {
      "epoch": 3.7840437158469946,
      "grad_norm": 0.6228653233518013,
      "learning_rate": 1.51799543957446e-06,
      "loss": 11.4732,
      "step": 21640
    },
    {
      "epoch": 3.7842185792349725,
      "grad_norm": 0.6765258945745065,
      "learning_rate": 1.5155384023210817e-06,
      "loss": 11.4934,
      "step": 21641
    },
    {
      "epoch": 3.784393442622951,
      "grad_norm": 0.7967037338693838,
      "learning_rate": 1.5130833399835342e-06,
      "loss": 11.4806,
      "step": 21642
    },
    {
      "epoch": 3.7845683060109288,
      "grad_norm": 0.6480640234356637,
      "learning_rate": 1.5106302526110672e-06,
      "loss": 11.4532,
      "step": 21643
    },
    {
      "epoch": 3.784743169398907,
      "grad_norm": 0.6120776106354694,
      "learning_rate": 1.5081791402528634e-06,
      "loss": 11.1981,
      "step": 21644
    },
    {
      "epoch": 3.7849180327868854,
      "grad_norm": 0.6628513525208216,
      "learning_rate": 1.5057300029580834e-06,
      "loss": 11.2134,
      "step": 21645
    },
    {
      "epoch": 3.7850928961748633,
      "grad_norm": 0.6427047477459583,
      "learning_rate": 1.5032828407758215e-06,
      "loss": 11.357,
      "step": 21646
    },
    {
      "epoch": 3.7852677595628417,
      "grad_norm": 0.696625160677728,
      "learning_rate": 1.5008376537551716e-06,
      "loss": 11.14,
      "step": 21647
    },
    {
      "epoch": 3.7854426229508196,
      "grad_norm": 0.7428680651367991,
      "learning_rate": 1.4983944419451613e-06,
      "loss": 11.3974,
      "step": 21648
    },
    {
      "epoch": 3.785617486338798,
      "grad_norm": 0.6223343674744787,
      "learning_rate": 1.4959532053947844e-06,
      "loss": 11.565,
      "step": 21649
    },
    {
      "epoch": 3.785792349726776,
      "grad_norm": 0.7109211676228985,
      "learning_rate": 1.49351394415298e-06,
      "loss": 11.5817,
      "step": 21650
    },
    {
      "epoch": 3.785967213114754,
      "grad_norm": 0.6535840903360238,
      "learning_rate": 1.4910766582686863e-06,
      "loss": 11.4056,
      "step": 21651
    },
    {
      "epoch": 3.786142076502732,
      "grad_norm": 0.6761218348365886,
      "learning_rate": 1.4886413477907645e-06,
      "loss": 11.2241,
      "step": 21652
    },
    {
      "epoch": 3.7863169398907104,
      "grad_norm": 0.6902844882076269,
      "learning_rate": 1.4862080127680534e-06,
      "loss": 11.1422,
      "step": 21653
    },
    {
      "epoch": 3.7864918032786887,
      "grad_norm": 0.7248272748802413,
      "learning_rate": 1.4837766532493468e-06,
      "loss": 11.1203,
      "step": 21654
    },
    {
      "epoch": 3.7866666666666666,
      "grad_norm": 0.6957148050694456,
      "learning_rate": 1.4813472692834064e-06,
      "loss": 10.991,
      "step": 21655
    },
    {
      "epoch": 3.786841530054645,
      "grad_norm": 0.7047240618042053,
      "learning_rate": 1.4789198609189481e-06,
      "loss": 11.5558,
      "step": 21656
    },
    {
      "epoch": 3.787016393442623,
      "grad_norm": 0.6696598164465436,
      "learning_rate": 1.4764944282046445e-06,
      "loss": 11.4667,
      "step": 21657
    },
    {
      "epoch": 3.787191256830601,
      "grad_norm": 0.6676498999659775,
      "learning_rate": 1.4740709711891232e-06,
      "loss": 11.3759,
      "step": 21658
    },
    {
      "epoch": 3.787366120218579,
      "grad_norm": 0.6968339768691658,
      "learning_rate": 1.471649489921001e-06,
      "loss": 11.3679,
      "step": 21659
    },
    {
      "epoch": 3.7875409836065574,
      "grad_norm": 0.6928526554663633,
      "learning_rate": 1.469229984448839e-06,
      "loss": 11.4611,
      "step": 21660
    },
    {
      "epoch": 3.7877158469945353,
      "grad_norm": 0.6449539679987235,
      "learning_rate": 1.4668124548211315e-06,
      "loss": 11.3849,
      "step": 21661
    },
    {
      "epoch": 3.7878907103825137,
      "grad_norm": 0.6886932304504282,
      "learning_rate": 1.4643969010863734e-06,
      "loss": 11.3016,
      "step": 21662
    },
    {
      "epoch": 3.788065573770492,
      "grad_norm": 0.7394292335950609,
      "learning_rate": 1.4619833232930036e-06,
      "loss": 11.6761,
      "step": 21663
    },
    {
      "epoch": 3.78824043715847,
      "grad_norm": 0.7558896419805259,
      "learning_rate": 1.4595717214894167e-06,
      "loss": 11.4713,
      "step": 21664
    },
    {
      "epoch": 3.788415300546448,
      "grad_norm": 0.6881613250129629,
      "learning_rate": 1.457162095723974e-06,
      "loss": 11.462,
      "step": 21665
    },
    {
      "epoch": 3.788590163934426,
      "grad_norm": 0.6789247140722574,
      "learning_rate": 1.4547544460450035e-06,
      "loss": 11.4051,
      "step": 21666
    },
    {
      "epoch": 3.7887650273224045,
      "grad_norm": 0.7052932370917498,
      "learning_rate": 1.4523487725007779e-06,
      "loss": 11.1279,
      "step": 21667
    },
    {
      "epoch": 3.7889398907103824,
      "grad_norm": 0.7098004847401852,
      "learning_rate": 1.4499450751395472e-06,
      "loss": 11.2968,
      "step": 21668
    },
    {
      "epoch": 3.7891147540983607,
      "grad_norm": 0.750990115236456,
      "learning_rate": 1.4475433540095063e-06,
      "loss": 11.2289,
      "step": 21669
    },
    {
      "epoch": 3.7892896174863386,
      "grad_norm": 0.7120531978909953,
      "learning_rate": 1.4451436091587944e-06,
      "loss": 11.3815,
      "step": 21670
    },
    {
      "epoch": 3.789464480874317,
      "grad_norm": 0.6752494997387922,
      "learning_rate": 1.442745840635573e-06,
      "loss": 11.5358,
      "step": 21671
    },
    {
      "epoch": 3.7896393442622953,
      "grad_norm": 0.6024716652064119,
      "learning_rate": 1.440350048487904e-06,
      "loss": 11.3047,
      "step": 21672
    },
    {
      "epoch": 3.789814207650273,
      "grad_norm": 0.8089811643733722,
      "learning_rate": 1.4379562327638375e-06,
      "loss": 11.4015,
      "step": 21673
    },
    {
      "epoch": 3.789989071038251,
      "grad_norm": 0.7215622678280417,
      "learning_rate": 1.4355643935113794e-06,
      "loss": 11.045,
      "step": 21674
    },
    {
      "epoch": 3.7901639344262295,
      "grad_norm": 0.7176725169595767,
      "learning_rate": 1.4331745307784805e-06,
      "loss": 11.4566,
      "step": 21675
    },
    {
      "epoch": 3.790338797814208,
      "grad_norm": 0.659750836054806,
      "learning_rate": 1.430786644613069e-06,
      "loss": 11.3596,
      "step": 21676
    },
    {
      "epoch": 3.7905136612021857,
      "grad_norm": 0.7228817938246977,
      "learning_rate": 1.4284007350630402e-06,
      "loss": 11.2323,
      "step": 21677
    },
    {
      "epoch": 3.790688524590164,
      "grad_norm": 0.6713306632792738,
      "learning_rate": 1.4260168021762332e-06,
      "loss": 11.5507,
      "step": 21678
    },
    {
      "epoch": 3.790863387978142,
      "grad_norm": 0.7213297701481831,
      "learning_rate": 1.4236348460004434e-06,
      "loss": 11.3863,
      "step": 21679
    },
    {
      "epoch": 3.7910382513661203,
      "grad_norm": 0.6288655323877381,
      "learning_rate": 1.4212548665834435e-06,
      "loss": 11.3889,
      "step": 21680
    },
    {
      "epoch": 3.7912131147540986,
      "grad_norm": 0.6707911413069361,
      "learning_rate": 1.418876863972962e-06,
      "loss": 11.3876,
      "step": 21681
    },
    {
      "epoch": 3.7913879781420765,
      "grad_norm": 0.7116432865521126,
      "learning_rate": 1.4165008382166833e-06,
      "loss": 11.3806,
      "step": 21682
    },
    {
      "epoch": 3.7915628415300544,
      "grad_norm": 0.6451701994039539,
      "learning_rate": 1.414126789362269e-06,
      "loss": 11.6123,
      "step": 21683
    },
    {
      "epoch": 3.7917377049180327,
      "grad_norm": 0.7316558161567663,
      "learning_rate": 1.411754717457292e-06,
      "loss": 11.5505,
      "step": 21684
    },
    {
      "epoch": 3.791912568306011,
      "grad_norm": 0.6781105129701943,
      "learning_rate": 1.4093846225493368e-06,
      "loss": 11.4768,
      "step": 21685
    },
    {
      "epoch": 3.792087431693989,
      "grad_norm": 0.6627951729040945,
      "learning_rate": 1.407016504685943e-06,
      "loss": 11.3029,
      "step": 21686
    },
    {
      "epoch": 3.7922622950819673,
      "grad_norm": 0.652271365109785,
      "learning_rate": 1.4046503639145835e-06,
      "loss": 11.4174,
      "step": 21687
    },
    {
      "epoch": 3.7924371584699452,
      "grad_norm": 0.6248994317217885,
      "learning_rate": 1.4022862002827208e-06,
      "loss": 11.4906,
      "step": 21688
    },
    {
      "epoch": 3.7926120218579236,
      "grad_norm": 0.6845125482519265,
      "learning_rate": 1.399924013837739e-06,
      "loss": 11.387,
      "step": 21689
    },
    {
      "epoch": 3.792786885245902,
      "grad_norm": 0.6890442665170484,
      "learning_rate": 1.3975638046270335e-06,
      "loss": 11.2788,
      "step": 21690
    },
    {
      "epoch": 3.79296174863388,
      "grad_norm": 0.6761926690833275,
      "learning_rate": 1.3952055726979108e-06,
      "loss": 11.5405,
      "step": 21691
    },
    {
      "epoch": 3.7931366120218577,
      "grad_norm": 0.6719662993798963,
      "learning_rate": 1.3928493180976887e-06,
      "loss": 11.5356,
      "step": 21692
    },
    {
      "epoch": 3.793311475409836,
      "grad_norm": 0.8045407790974687,
      "learning_rate": 1.3904950408735962e-06,
      "loss": 11.2938,
      "step": 21693
    },
    {
      "epoch": 3.7934863387978144,
      "grad_norm": 0.6625080600633575,
      "learning_rate": 1.3881427410728398e-06,
      "loss": 11.2699,
      "step": 21694
    },
    {
      "epoch": 3.7936612021857923,
      "grad_norm": 0.661170128591541,
      "learning_rate": 1.3857924187426151e-06,
      "loss": 11.3874,
      "step": 21695
    },
    {
      "epoch": 3.7938360655737706,
      "grad_norm": 0.6713441858117372,
      "learning_rate": 1.3834440739300403e-06,
      "loss": 11.2832,
      "step": 21696
    },
    {
      "epoch": 3.7940109289617485,
      "grad_norm": 0.6921276004227417,
      "learning_rate": 1.3810977066821884e-06,
      "loss": 11.0332,
      "step": 21697
    },
    {
      "epoch": 3.794185792349727,
      "grad_norm": 0.750049584915794,
      "learning_rate": 1.378753317046133e-06,
      "loss": 11.4959,
      "step": 21698
    },
    {
      "epoch": 3.7943606557377048,
      "grad_norm": 0.6722468580319432,
      "learning_rate": 1.37641090506887e-06,
      "loss": 11.4719,
      "step": 21699
    },
    {
      "epoch": 3.794535519125683,
      "grad_norm": 0.7187187102941147,
      "learning_rate": 1.374070470797395e-06,
      "loss": 11.1004,
      "step": 21700
    },
    {
      "epoch": 3.794710382513661,
      "grad_norm": 0.6724676928197065,
      "learning_rate": 1.3717320142786372e-06,
      "loss": 11.4842,
      "step": 21701
    },
    {
      "epoch": 3.7948852459016393,
      "grad_norm": 0.6785189382367369,
      "learning_rate": 1.36939553555947e-06,
      "loss": 11.2188,
      "step": 21702
    },
    {
      "epoch": 3.7950601092896177,
      "grad_norm": 0.6934019108846353,
      "learning_rate": 1.3670610346867562e-06,
      "loss": 11.4581,
      "step": 21703
    },
    {
      "epoch": 3.7952349726775956,
      "grad_norm": 0.6404054155908028,
      "learning_rate": 1.3647285117073138e-06,
      "loss": 11.2045,
      "step": 21704
    },
    {
      "epoch": 3.795409836065574,
      "grad_norm": 0.6259701306875162,
      "learning_rate": 1.3623979666679165e-06,
      "loss": 11.2543,
      "step": 21705
    },
    {
      "epoch": 3.795584699453552,
      "grad_norm": 0.6546030271732121,
      "learning_rate": 1.3600693996152936e-06,
      "loss": 11.1812,
      "step": 21706
    },
    {
      "epoch": 3.79575956284153,
      "grad_norm": 0.6744093358387045,
      "learning_rate": 1.357742810596152e-06,
      "loss": 11.1692,
      "step": 21707
    },
    {
      "epoch": 3.795934426229508,
      "grad_norm": 0.7313658117273454,
      "learning_rate": 1.3554181996571324e-06,
      "loss": 11.3961,
      "step": 21708
    },
    {
      "epoch": 3.7961092896174864,
      "grad_norm": 0.6613646851616181,
      "learning_rate": 1.353095566844842e-06,
      "loss": 10.8328,
      "step": 21709
    },
    {
      "epoch": 3.7962841530054643,
      "grad_norm": 0.6872983924582722,
      "learning_rate": 1.3507749122058988e-06,
      "loss": 11.1788,
      "step": 21710
    },
    {
      "epoch": 3.7964590163934426,
      "grad_norm": 0.7201985741099554,
      "learning_rate": 1.3484562357867992e-06,
      "loss": 11.308,
      "step": 21711
    },
    {
      "epoch": 3.796633879781421,
      "grad_norm": 0.6202612966897861,
      "learning_rate": 1.3461395376340502e-06,
      "loss": 11.1519,
      "step": 21712
    },
    {
      "epoch": 3.796808743169399,
      "grad_norm": 0.6326874719650233,
      "learning_rate": 1.3438248177941038e-06,
      "loss": 11.547,
      "step": 21713
    },
    {
      "epoch": 3.796983606557377,
      "grad_norm": 0.6700513051026601,
      "learning_rate": 1.3415120763134005e-06,
      "loss": 11.2899,
      "step": 21714
    },
    {
      "epoch": 3.797158469945355,
      "grad_norm": 0.7393993855704422,
      "learning_rate": 1.339201313238292e-06,
      "loss": 11.4865,
      "step": 21715
    },
    {
      "epoch": 3.7973333333333334,
      "grad_norm": 0.6335611491278043,
      "learning_rate": 1.3368925286151302e-06,
      "loss": 11.4803,
      "step": 21716
    },
    {
      "epoch": 3.7975081967213113,
      "grad_norm": 0.6804502556876819,
      "learning_rate": 1.3345857224902004e-06,
      "loss": 11.2631,
      "step": 21717
    },
    {
      "epoch": 3.7976830601092897,
      "grad_norm": 0.7271495300457889,
      "learning_rate": 1.3322808949097653e-06,
      "loss": 11.2974,
      "step": 21718
    },
    {
      "epoch": 3.7978579234972676,
      "grad_norm": 0.7528030157364163,
      "learning_rate": 1.329978045920066e-06,
      "loss": 11.1852,
      "step": 21719
    },
    {
      "epoch": 3.798032786885246,
      "grad_norm": 0.6821737599986933,
      "learning_rate": 1.3276771755672545e-06,
      "loss": 11.4689,
      "step": 21720
    },
    {
      "epoch": 3.7982076502732243,
      "grad_norm": 0.6556370483381991,
      "learning_rate": 1.3253782838974826e-06,
      "loss": 11.2867,
      "step": 21721
    },
    {
      "epoch": 3.798382513661202,
      "grad_norm": 0.651999533341081,
      "learning_rate": 1.3230813709568468e-06,
      "loss": 11.5655,
      "step": 21722
    },
    {
      "epoch": 3.79855737704918,
      "grad_norm": 0.6680906177401587,
      "learning_rate": 1.320786436791399e-06,
      "loss": 11.3228,
      "step": 21723
    },
    {
      "epoch": 3.7987322404371584,
      "grad_norm": 0.6809657563108059,
      "learning_rate": 1.3184934814471694e-06,
      "loss": 11.4574,
      "step": 21724
    },
    {
      "epoch": 3.7989071038251367,
      "grad_norm": 0.6997344224263775,
      "learning_rate": 1.3162025049701432e-06,
      "loss": 11.2535,
      "step": 21725
    },
    {
      "epoch": 3.7990819672131146,
      "grad_norm": 0.6755629350222527,
      "learning_rate": 1.3139135074062614e-06,
      "loss": 11.5884,
      "step": 21726
    },
    {
      "epoch": 3.799256830601093,
      "grad_norm": 0.7627439936864124,
      "learning_rate": 1.3116264888014097e-06,
      "loss": 11.3006,
      "step": 21727
    },
    {
      "epoch": 3.799431693989071,
      "grad_norm": 0.6063918846176312,
      "learning_rate": 1.3093414492014621e-06,
      "loss": 11.2741,
      "step": 21728
    },
    {
      "epoch": 3.799606557377049,
      "grad_norm": 0.6911048036641151,
      "learning_rate": 1.307058388652238e-06,
      "loss": 11.5426,
      "step": 21729
    },
    {
      "epoch": 3.7997814207650276,
      "grad_norm": 0.6587285173186361,
      "learning_rate": 1.3047773071995228e-06,
      "loss": 11.2818,
      "step": 21730
    },
    {
      "epoch": 3.7999562841530055,
      "grad_norm": 0.6524558308589025,
      "learning_rate": 1.3024982048890577e-06,
      "loss": 11.4933,
      "step": 21731
    },
    {
      "epoch": 3.8001311475409834,
      "grad_norm": 0.6495182394443297,
      "learning_rate": 1.3002210817665394e-06,
      "loss": 11.3579,
      "step": 21732
    },
    {
      "epoch": 3.8003060109289617,
      "grad_norm": 0.6957911945774359,
      "learning_rate": 1.2979459378776426e-06,
      "loss": 11.6128,
      "step": 21733
    },
    {
      "epoch": 3.80048087431694,
      "grad_norm": 0.7318314435393588,
      "learning_rate": 1.2956727732679864e-06,
      "loss": 11.3336,
      "step": 21734
    },
    {
      "epoch": 3.800655737704918,
      "grad_norm": 0.6387338086158251,
      "learning_rate": 1.2934015879831451e-06,
      "loss": 11.591,
      "step": 21735
    },
    {
      "epoch": 3.8008306010928963,
      "grad_norm": 0.686506699016033,
      "learning_rate": 1.2911323820686717e-06,
      "loss": 11.4908,
      "step": 21736
    },
    {
      "epoch": 3.801005464480874,
      "grad_norm": 0.697369796038047,
      "learning_rate": 1.2888651555700737e-06,
      "loss": 11.2219,
      "step": 21737
    },
    {
      "epoch": 3.8011803278688525,
      "grad_norm": 0.6962411302433169,
      "learning_rate": 1.286599908532815e-06,
      "loss": 11.1495,
      "step": 21738
    },
    {
      "epoch": 3.801355191256831,
      "grad_norm": 0.6551368748028917,
      "learning_rate": 1.284336641002315e-06,
      "loss": 11.116,
      "step": 21739
    },
    {
      "epoch": 3.8015300546448088,
      "grad_norm": 0.7845861191658509,
      "learning_rate": 1.2820753530239592e-06,
      "loss": 11.3006,
      "step": 21740
    },
    {
      "epoch": 3.8017049180327867,
      "grad_norm": 0.7480791767358598,
      "learning_rate": 1.2798160446431006e-06,
      "loss": 11.3903,
      "step": 21741
    },
    {
      "epoch": 3.801879781420765,
      "grad_norm": 0.8399118702155723,
      "learning_rate": 1.2775587159050362e-06,
      "loss": 11.4717,
      "step": 21742
    },
    {
      "epoch": 3.8020546448087433,
      "grad_norm": 0.7482598893349096,
      "learning_rate": 1.2753033668550406e-06,
      "loss": 11.3672,
      "step": 21743
    },
    {
      "epoch": 3.8022295081967212,
      "grad_norm": 0.6620999774368092,
      "learning_rate": 1.2730499975383448e-06,
      "loss": 11.461,
      "step": 21744
    },
    {
      "epoch": 3.8024043715846996,
      "grad_norm": 0.658621321096771,
      "learning_rate": 1.2707986080001122e-06,
      "loss": 11.1595,
      "step": 21745
    },
    {
      "epoch": 3.8025792349726775,
      "grad_norm": 0.6160384626491685,
      "learning_rate": 1.268549198285518e-06,
      "loss": 11.4139,
      "step": 21746
    },
    {
      "epoch": 3.802754098360656,
      "grad_norm": 0.7396820014090675,
      "learning_rate": 1.2663017684396593e-06,
      "loss": 11.615,
      "step": 21747
    },
    {
      "epoch": 3.802928961748634,
      "grad_norm": 0.6504244103363483,
      "learning_rate": 1.2640563185075894e-06,
      "loss": 11.5016,
      "step": 21748
    },
    {
      "epoch": 3.803103825136612,
      "grad_norm": 0.6831855126706706,
      "learning_rate": 1.2618128485343606e-06,
      "loss": 11.332,
      "step": 21749
    },
    {
      "epoch": 3.80327868852459,
      "grad_norm": 0.6904788760426985,
      "learning_rate": 1.2595713585649482e-06,
      "loss": 11.1976,
      "step": 21750
    },
    {
      "epoch": 3.8034535519125683,
      "grad_norm": 0.7191212328072552,
      "learning_rate": 1.2573318486443053e-06,
      "loss": 11.3433,
      "step": 21751
    },
    {
      "epoch": 3.8036284153005466,
      "grad_norm": 0.6910612495406048,
      "learning_rate": 1.2550943188173292e-06,
      "loss": 11.5721,
      "step": 21752
    },
    {
      "epoch": 3.8038032786885245,
      "grad_norm": 0.6613241092621821,
      "learning_rate": 1.2528587691289063e-06,
      "loss": 11.3654,
      "step": 21753
    },
    {
      "epoch": 3.803978142076503,
      "grad_norm": 0.703450001981849,
      "learning_rate": 1.250625199623856e-06,
      "loss": 11.3181,
      "step": 21754
    },
    {
      "epoch": 3.8041530054644808,
      "grad_norm": 0.6662817321817479,
      "learning_rate": 1.2483936103469761e-06,
      "loss": 11.2554,
      "step": 21755
    },
    {
      "epoch": 3.804327868852459,
      "grad_norm": 0.6354317003892664,
      "learning_rate": 1.2461640013430087e-06,
      "loss": 11.687,
      "step": 21756
    },
    {
      "epoch": 3.804502732240437,
      "grad_norm": 0.653331816281828,
      "learning_rate": 1.2439363726566622e-06,
      "loss": 11.1585,
      "step": 21757
    },
    {
      "epoch": 3.8046775956284153,
      "grad_norm": 0.6761401784028742,
      "learning_rate": 1.241710724332612e-06,
      "loss": 11.3492,
      "step": 21758
    },
    {
      "epoch": 3.8048524590163932,
      "grad_norm": 0.7264177914373289,
      "learning_rate": 1.2394870564155004e-06,
      "loss": 11.3359,
      "step": 21759
    },
    {
      "epoch": 3.8050273224043716,
      "grad_norm": 0.6810387962854307,
      "learning_rate": 1.2372653689499025e-06,
      "loss": 11.2517,
      "step": 21760
    },
    {
      "epoch": 3.80520218579235,
      "grad_norm": 0.6833777676586809,
      "learning_rate": 1.2350456619803718e-06,
      "loss": 11.5441,
      "step": 21761
    },
    {
      "epoch": 3.805377049180328,
      "grad_norm": 0.6207779900196028,
      "learning_rate": 1.2328279355514283e-06,
      "loss": 11.4436,
      "step": 21762
    },
    {
      "epoch": 3.805551912568306,
      "grad_norm": 0.6450595837910299,
      "learning_rate": 1.2306121897075361e-06,
      "loss": 11.5046,
      "step": 21763
    },
    {
      "epoch": 3.805726775956284,
      "grad_norm": 0.7405187478216245,
      "learning_rate": 1.2283984244931269e-06,
      "loss": 11.4107,
      "step": 21764
    },
    {
      "epoch": 3.8059016393442624,
      "grad_norm": 0.7228318467206892,
      "learning_rate": 1.22618663995262e-06,
      "loss": 10.9054,
      "step": 21765
    },
    {
      "epoch": 3.8060765027322403,
      "grad_norm": 0.677210356547022,
      "learning_rate": 1.2239768361303138e-06,
      "loss": 11.2671,
      "step": 21766
    },
    {
      "epoch": 3.8062513661202186,
      "grad_norm": 0.6102452770156598,
      "learning_rate": 1.2217690130705727e-06,
      "loss": 11.4312,
      "step": 21767
    },
    {
      "epoch": 3.8064262295081965,
      "grad_norm": 0.7361174198732131,
      "learning_rate": 1.2195631708176502e-06,
      "loss": 11.2346,
      "step": 21768
    },
    {
      "epoch": 3.806601092896175,
      "grad_norm": 0.6479019417643606,
      "learning_rate": 1.2173593094157776e-06,
      "loss": 11.269,
      "step": 21769
    },
    {
      "epoch": 3.806775956284153,
      "grad_norm": 0.717120167806998,
      "learning_rate": 1.2151574289091749e-06,
      "loss": 11.4161,
      "step": 21770
    },
    {
      "epoch": 3.806950819672131,
      "grad_norm": 0.6490841379930231,
      "learning_rate": 1.2129575293419405e-06,
      "loss": 11.7051,
      "step": 21771
    },
    {
      "epoch": 3.8071256830601095,
      "grad_norm": 0.6468254867209429,
      "learning_rate": 1.2107596107582496e-06,
      "loss": 11.2074,
      "step": 21772
    },
    {
      "epoch": 3.8073005464480874,
      "grad_norm": 0.7251641696467509,
      "learning_rate": 1.2085636732021345e-06,
      "loss": 11.4169,
      "step": 21773
    },
    {
      "epoch": 3.8074754098360657,
      "grad_norm": 0.6764388597397106,
      "learning_rate": 1.20636971671767e-06,
      "loss": 11.5145,
      "step": 21774
    },
    {
      "epoch": 3.8076502732240436,
      "grad_norm": 0.7008256585066599,
      "learning_rate": 1.2041777413488108e-06,
      "loss": 11.372,
      "step": 21775
    },
    {
      "epoch": 3.807825136612022,
      "grad_norm": 0.6635918202680072,
      "learning_rate": 1.2019877471395214e-06,
      "loss": 11.2762,
      "step": 21776
    },
    {
      "epoch": 3.808,
      "grad_norm": 0.6846071233725062,
      "learning_rate": 1.1997997341337442e-06,
      "loss": 11.144,
      "step": 21777
    },
    {
      "epoch": 3.808174863387978,
      "grad_norm": 0.6777161631047565,
      "learning_rate": 1.1976137023753331e-06,
      "loss": 11.2256,
      "step": 21778
    },
    {
      "epoch": 3.8083497267759565,
      "grad_norm": 0.7556482613412091,
      "learning_rate": 1.195429651908142e-06,
      "loss": 11.3432,
      "step": 21779
    },
    {
      "epoch": 3.8085245901639344,
      "grad_norm": 0.664986770939662,
      "learning_rate": 1.193247582775947e-06,
      "loss": 11.3393,
      "step": 21780
    },
    {
      "epoch": 3.8086994535519123,
      "grad_norm": 0.6746569972127214,
      "learning_rate": 1.1910674950225021e-06,
      "loss": 11.4863,
      "step": 21781
    },
    {
      "epoch": 3.8088743169398906,
      "grad_norm": 0.6789887372583042,
      "learning_rate": 1.1888893886915497e-06,
      "loss": 11.2365,
      "step": 21782
    },
    {
      "epoch": 3.809049180327869,
      "grad_norm": 0.7163882301940908,
      "learning_rate": 1.1867132638267664e-06,
      "loss": 11.4787,
      "step": 21783
    },
    {
      "epoch": 3.809224043715847,
      "grad_norm": 0.6749814489216954,
      "learning_rate": 1.1845391204717614e-06,
      "loss": 11.4416,
      "step": 21784
    },
    {
      "epoch": 3.8093989071038252,
      "grad_norm": 0.6959623528693627,
      "learning_rate": 1.1823669586701558e-06,
      "loss": 11.5424,
      "step": 21785
    },
    {
      "epoch": 3.809573770491803,
      "grad_norm": 0.7056033463275159,
      "learning_rate": 1.1801967784655032e-06,
      "loss": 11.4328,
      "step": 21786
    },
    {
      "epoch": 3.8097486338797815,
      "grad_norm": 0.6702052809988763,
      "learning_rate": 1.1780285799013136e-06,
      "loss": 11.5038,
      "step": 21787
    },
    {
      "epoch": 3.80992349726776,
      "grad_norm": 0.6641672936507819,
      "learning_rate": 1.1758623630210964e-06,
      "loss": 11.3253,
      "step": 21788
    },
    {
      "epoch": 3.8100983606557377,
      "grad_norm": 0.6142080918224601,
      "learning_rate": 1.1736981278682502e-06,
      "loss": 11.5353,
      "step": 21789
    },
    {
      "epoch": 3.8102732240437156,
      "grad_norm": 0.6941362401972796,
      "learning_rate": 1.1715358744861849e-06,
      "loss": 11.1996,
      "step": 21790
    },
    {
      "epoch": 3.810448087431694,
      "grad_norm": 0.6206491707882267,
      "learning_rate": 1.1693756029182879e-06,
      "loss": 11.2426,
      "step": 21791
    },
    {
      "epoch": 3.8106229508196723,
      "grad_norm": 0.681811481875873,
      "learning_rate": 1.167217313207858e-06,
      "loss": 11.5119,
      "step": 21792
    },
    {
      "epoch": 3.81079781420765,
      "grad_norm": 0.7133301471125699,
      "learning_rate": 1.1650610053981715e-06,
      "loss": 11.3206,
      "step": 21793
    },
    {
      "epoch": 3.8109726775956285,
      "grad_norm": 0.662478751096236,
      "learning_rate": 1.1629066795324717e-06,
      "loss": 11.3565,
      "step": 21794
    },
    {
      "epoch": 3.8111475409836064,
      "grad_norm": 0.6612419859856687,
      "learning_rate": 1.1607543356539685e-06,
      "loss": 11.4695,
      "step": 21795
    },
    {
      "epoch": 3.8113224043715848,
      "grad_norm": 0.6149790415282377,
      "learning_rate": 1.1586039738058162e-06,
      "loss": 11.3739,
      "step": 21796
    },
    {
      "epoch": 3.811497267759563,
      "grad_norm": 0.6480930528040936,
      "learning_rate": 1.1564555940311362e-06,
      "loss": 11.42,
      "step": 21797
    },
    {
      "epoch": 3.811672131147541,
      "grad_norm": 0.6731751176312032,
      "learning_rate": 1.1543091963730157e-06,
      "loss": 11.4159,
      "step": 21798
    },
    {
      "epoch": 3.811846994535519,
      "grad_norm": 0.747384799324599,
      "learning_rate": 1.1521647808744873e-06,
      "loss": 11.4487,
      "step": 21799
    },
    {
      "epoch": 3.8120218579234972,
      "grad_norm": 0.652592047241871,
      "learning_rate": 1.1500223475785498e-06,
      "loss": 11.5179,
      "step": 21800
    },
    {
      "epoch": 3.8121967213114756,
      "grad_norm": 0.6519337599778324,
      "learning_rate": 1.1478818965281911e-06,
      "loss": 11.4858,
      "step": 21801
    },
    {
      "epoch": 3.8123715846994535,
      "grad_norm": 0.6425350547407582,
      "learning_rate": 1.1457434277662993e-06,
      "loss": 11.3941,
      "step": 21802
    },
    {
      "epoch": 3.812546448087432,
      "grad_norm": 0.6907542665938983,
      "learning_rate": 1.1436069413357842e-06,
      "loss": 11.5392,
      "step": 21803
    },
    {
      "epoch": 3.8127213114754097,
      "grad_norm": 0.5910087763947584,
      "learning_rate": 1.1414724372794784e-06,
      "loss": 11.4507,
      "step": 21804
    },
    {
      "epoch": 3.812896174863388,
      "grad_norm": 0.6568945218498917,
      "learning_rate": 1.1393399156401808e-06,
      "loss": 11.5176,
      "step": 21805
    },
    {
      "epoch": 3.8130710382513664,
      "grad_norm": 0.7816995090665815,
      "learning_rate": 1.1372093764606684e-06,
      "loss": 11.2558,
      "step": 21806
    },
    {
      "epoch": 3.8132459016393443,
      "grad_norm": 0.7473612296566138,
      "learning_rate": 1.1350808197836515e-06,
      "loss": 11.3406,
      "step": 21807
    },
    {
      "epoch": 3.813420765027322,
      "grad_norm": 0.6843527809777635,
      "learning_rate": 1.1329542456518183e-06,
      "loss": 11.2589,
      "step": 21808
    },
    {
      "epoch": 3.8135956284153005,
      "grad_norm": 0.6686442614133955,
      "learning_rate": 1.1308296541078123e-06,
      "loss": 11.3004,
      "step": 21809
    },
    {
      "epoch": 3.813770491803279,
      "grad_norm": 0.702348703746152,
      "learning_rate": 1.1287070451942438e-06,
      "loss": 11.4063,
      "step": 21810
    },
    {
      "epoch": 3.8139453551912568,
      "grad_norm": 0.5872511040949084,
      "learning_rate": 1.1265864189536678e-06,
      "loss": 11.5005,
      "step": 21811
    },
    {
      "epoch": 3.814120218579235,
      "grad_norm": 0.7406135235744911,
      "learning_rate": 1.1244677754286282e-06,
      "loss": 11.1102,
      "step": 21812
    },
    {
      "epoch": 3.814295081967213,
      "grad_norm": 0.7070905965420418,
      "learning_rate": 1.1223511146615905e-06,
      "loss": 11.2171,
      "step": 21813
    },
    {
      "epoch": 3.8144699453551913,
      "grad_norm": 0.6457552718537385,
      "learning_rate": 1.1202364366949991e-06,
      "loss": 11.6293,
      "step": 21814
    },
    {
      "epoch": 3.8146448087431692,
      "grad_norm": 0.6963240332607049,
      "learning_rate": 1.1181237415712752e-06,
      "loss": 11.4554,
      "step": 21815
    },
    {
      "epoch": 3.8148196721311476,
      "grad_norm": 0.651475187887694,
      "learning_rate": 1.1160130293327852e-06,
      "loss": 11.4486,
      "step": 21816
    },
    {
      "epoch": 3.8149945355191255,
      "grad_norm": 0.6517737390559879,
      "learning_rate": 1.1139043000218397e-06,
      "loss": 11.3955,
      "step": 21817
    },
    {
      "epoch": 3.815169398907104,
      "grad_norm": 0.7138551133264993,
      "learning_rate": 1.1117975536807267e-06,
      "loss": 11.6031,
      "step": 21818
    },
    {
      "epoch": 3.815344262295082,
      "grad_norm": 0.6890653408105044,
      "learning_rate": 1.109692790351713e-06,
      "loss": 11.1481,
      "step": 21819
    },
    {
      "epoch": 3.81551912568306,
      "grad_norm": 0.6869428171667036,
      "learning_rate": 1.1075900100769754e-06,
      "loss": 11.6398,
      "step": 21820
    },
    {
      "epoch": 3.8156939890710384,
      "grad_norm": 0.7452812401619588,
      "learning_rate": 1.1054892128987137e-06,
      "loss": 11.5646,
      "step": 21821
    },
    {
      "epoch": 3.8158688524590163,
      "grad_norm": 0.6391660069473795,
      "learning_rate": 1.1033903988590278e-06,
      "loss": 11.7126,
      "step": 21822
    },
    {
      "epoch": 3.8160437158469946,
      "grad_norm": 0.6719686127527603,
      "learning_rate": 1.101293568000028e-06,
      "loss": 11.689,
      "step": 21823
    },
    {
      "epoch": 3.8162185792349725,
      "grad_norm": 0.6701207326271683,
      "learning_rate": 1.0991987203637367e-06,
      "loss": 11.4147,
      "step": 21824
    },
    {
      "epoch": 3.816393442622951,
      "grad_norm": 0.6202938351268814,
      "learning_rate": 1.0971058559921864e-06,
      "loss": 11.5254,
      "step": 21825
    },
    {
      "epoch": 3.816568306010929,
      "grad_norm": 0.7149770221435652,
      "learning_rate": 1.0950149749273332e-06,
      "loss": 11.2549,
      "step": 21826
    },
    {
      "epoch": 3.816743169398907,
      "grad_norm": 0.5958317051260693,
      "learning_rate": 1.0929260772111094e-06,
      "loss": 11.4348,
      "step": 21827
    },
    {
      "epoch": 3.8169180327868855,
      "grad_norm": 0.6395490273673452,
      "learning_rate": 1.0908391628854041e-06,
      "loss": 11.4633,
      "step": 21828
    },
    {
      "epoch": 3.8170928961748634,
      "grad_norm": 0.6535148395278362,
      "learning_rate": 1.0887542319920619e-06,
      "loss": 11.3784,
      "step": 21829
    },
    {
      "epoch": 3.8172677595628413,
      "grad_norm": 0.6639234451737633,
      "learning_rate": 1.0866712845728932e-06,
      "loss": 11.3903,
      "step": 21830
    },
    {
      "epoch": 3.8174426229508196,
      "grad_norm": 0.669062134279637,
      "learning_rate": 1.084590320669676e-06,
      "loss": 11.2516,
      "step": 21831
    },
    {
      "epoch": 3.817617486338798,
      "grad_norm": 0.764564641056888,
      "learning_rate": 1.0825113403241217e-06,
      "loss": 11.3854,
      "step": 21832
    },
    {
      "epoch": 3.817792349726776,
      "grad_norm": 0.6424977021950138,
      "learning_rate": 1.0804343435779408e-06,
      "loss": 11.445,
      "step": 21833
    },
    {
      "epoch": 3.817967213114754,
      "grad_norm": 0.6796855829923563,
      "learning_rate": 1.0783593304727668e-06,
      "loss": 11.5764,
      "step": 21834
    },
    {
      "epoch": 3.818142076502732,
      "grad_norm": 0.7251559011046214,
      "learning_rate": 1.0762863010502222e-06,
      "loss": 11.334,
      "step": 21835
    },
    {
      "epoch": 3.8183169398907104,
      "grad_norm": 0.7106739286402776,
      "learning_rate": 1.0742152553518736e-06,
      "loss": 11.1841,
      "step": 21836
    },
    {
      "epoch": 3.8184918032786888,
      "grad_norm": 0.6437532217330445,
      "learning_rate": 1.0721461934192545e-06,
      "loss": 11.4392,
      "step": 21837
    },
    {
      "epoch": 3.8186666666666667,
      "grad_norm": 0.7000744819925142,
      "learning_rate": 1.0700791152938428e-06,
      "loss": 11.4429,
      "step": 21838
    },
    {
      "epoch": 3.8188415300546446,
      "grad_norm": 0.6535557870925043,
      "learning_rate": 1.0680140210171053e-06,
      "loss": 11.5049,
      "step": 21839
    },
    {
      "epoch": 3.819016393442623,
      "grad_norm": 0.7173959984898537,
      "learning_rate": 1.0659509106304532e-06,
      "loss": 11.1542,
      "step": 21840
    },
    {
      "epoch": 3.8191912568306012,
      "grad_norm": 0.6593651146981226,
      "learning_rate": 1.0638897841752427e-06,
      "loss": 11.521,
      "step": 21841
    },
    {
      "epoch": 3.819366120218579,
      "grad_norm": 0.6881676967142716,
      "learning_rate": 1.0618306416928181e-06,
      "loss": 11.1991,
      "step": 21842
    },
    {
      "epoch": 3.8195409836065575,
      "grad_norm": 0.674752200270326,
      "learning_rate": 1.0597734832244799e-06,
      "loss": 11.5598,
      "step": 21843
    },
    {
      "epoch": 3.8197158469945354,
      "grad_norm": 0.6822008026275035,
      "learning_rate": 1.0577183088114618e-06,
      "loss": 11.3574,
      "step": 21844
    },
    {
      "epoch": 3.8198907103825137,
      "grad_norm": 0.6369694069679117,
      "learning_rate": 1.0556651184949751e-06,
      "loss": 11.0288,
      "step": 21845
    },
    {
      "epoch": 3.820065573770492,
      "grad_norm": 0.7261939249919529,
      "learning_rate": 1.0536139123162093e-06,
      "loss": 11.4232,
      "step": 21846
    },
    {
      "epoch": 3.82024043715847,
      "grad_norm": 0.6969424818351744,
      "learning_rate": 1.0515646903162978e-06,
      "loss": 11.424,
      "step": 21847
    },
    {
      "epoch": 3.820415300546448,
      "grad_norm": 0.6607820046282541,
      "learning_rate": 1.0495174525363193e-06,
      "loss": 11.3134,
      "step": 21848
    },
    {
      "epoch": 3.820590163934426,
      "grad_norm": 0.7516250224363553,
      "learning_rate": 1.0474721990173296e-06,
      "loss": 11.1967,
      "step": 21849
    },
    {
      "epoch": 3.8207650273224045,
      "grad_norm": 0.649260477845565,
      "learning_rate": 1.0454289298003517e-06,
      "loss": 11.4728,
      "step": 21850
    },
    {
      "epoch": 3.8209398907103824,
      "grad_norm": 0.5996114807472576,
      "learning_rate": 1.0433876449263524e-06,
      "loss": 11.1795,
      "step": 21851
    },
    {
      "epoch": 3.8211147540983608,
      "grad_norm": 0.7255039795334298,
      "learning_rate": 1.0413483444362771e-06,
      "loss": 11.1875,
      "step": 21852
    },
    {
      "epoch": 3.8212896174863387,
      "grad_norm": 0.6562202181505602,
      "learning_rate": 1.0393110283709927e-06,
      "loss": 11.5368,
      "step": 21853
    },
    {
      "epoch": 3.821464480874317,
      "grad_norm": 0.7162703882097491,
      "learning_rate": 1.0372756967713782e-06,
      "loss": 11.5206,
      "step": 21854
    },
    {
      "epoch": 3.8216393442622953,
      "grad_norm": 0.6894720410393381,
      "learning_rate": 1.035242349678245e-06,
      "loss": 11.5382,
      "step": 21855
    },
    {
      "epoch": 3.8218142076502732,
      "grad_norm": 0.6795397402975056,
      "learning_rate": 1.0332109871323604e-06,
      "loss": 11.4665,
      "step": 21856
    },
    {
      "epoch": 3.821989071038251,
      "grad_norm": 0.6785798875839237,
      "learning_rate": 1.0311816091744698e-06,
      "loss": 11.3834,
      "step": 21857
    },
    {
      "epoch": 3.8221639344262295,
      "grad_norm": 0.6527066632873395,
      "learning_rate": 1.0291542158452517e-06,
      "loss": 11.3367,
      "step": 21858
    },
    {
      "epoch": 3.822338797814208,
      "grad_norm": 0.647935370314691,
      "learning_rate": 1.0271288071853847e-06,
      "loss": 11.4005,
      "step": 21859
    },
    {
      "epoch": 3.8225136612021857,
      "grad_norm": 0.5824886162456258,
      "learning_rate": 1.0251053832354695e-06,
      "loss": 11.0953,
      "step": 21860
    },
    {
      "epoch": 3.822688524590164,
      "grad_norm": 0.7144109993982138,
      "learning_rate": 1.023083944036085e-06,
      "loss": 11.2582,
      "step": 21861
    },
    {
      "epoch": 3.822863387978142,
      "grad_norm": 0.6617284165835662,
      "learning_rate": 1.0210644896277544e-06,
      "loss": 11.3108,
      "step": 21862
    },
    {
      "epoch": 3.8230382513661203,
      "grad_norm": 0.6857984451393883,
      "learning_rate": 1.0190470200509895e-06,
      "loss": 11.4163,
      "step": 21863
    },
    {
      "epoch": 3.823213114754098,
      "grad_norm": 0.7084793853358153,
      "learning_rate": 1.0170315353462466e-06,
      "loss": 11.3452,
      "step": 21864
    },
    {
      "epoch": 3.8233879781420765,
      "grad_norm": 0.7483352342961906,
      "learning_rate": 1.0150180355539385e-06,
      "loss": 11.4008,
      "step": 21865
    },
    {
      "epoch": 3.8235628415300544,
      "grad_norm": 0.6549274692915824,
      "learning_rate": 1.0130065207144545e-06,
      "loss": 11.5256,
      "step": 21866
    },
    {
      "epoch": 3.8237377049180328,
      "grad_norm": 0.6288509823580699,
      "learning_rate": 1.0109969908680961e-06,
      "loss": 11.2561,
      "step": 21867
    },
    {
      "epoch": 3.823912568306011,
      "grad_norm": 0.7171173005131254,
      "learning_rate": 1.0089894460551974e-06,
      "loss": 11.4333,
      "step": 21868
    },
    {
      "epoch": 3.824087431693989,
      "grad_norm": 0.7103900356030033,
      "learning_rate": 1.0069838863160041e-06,
      "loss": 11.4733,
      "step": 21869
    },
    {
      "epoch": 3.8242622950819674,
      "grad_norm": 0.5950495616304714,
      "learning_rate": 1.0049803116907397e-06,
      "loss": 11.4932,
      "step": 21870
    },
    {
      "epoch": 3.8244371584699453,
      "grad_norm": 0.6471858882324271,
      "learning_rate": 1.0029787222195608e-06,
      "loss": 11.448,
      "step": 21871
    },
    {
      "epoch": 3.8246120218579236,
      "grad_norm": 0.7027957549149252,
      "learning_rate": 1.0009791179426242e-06,
      "loss": 11.4205,
      "step": 21872
    },
    {
      "epoch": 3.8247868852459015,
      "grad_norm": 0.7270010702379193,
      "learning_rate": 9.9898149890002e-07,
      "loss": 11.3953,
      "step": 21873
    },
    {
      "epoch": 3.82496174863388,
      "grad_norm": 0.62920745175884,
      "learning_rate": 9.969858651318166e-07,
      "loss": 11.5393,
      "step": 21874
    },
    {
      "epoch": 3.8251366120218577,
      "grad_norm": 0.6279164157290268,
      "learning_rate": 9.949922166780257e-07,
      "loss": 11.3045,
      "step": 21875
    },
    {
      "epoch": 3.825311475409836,
      "grad_norm": 0.7045544225547441,
      "learning_rate": 9.930005535786157e-07,
      "loss": 11.3472,
      "step": 21876
    },
    {
      "epoch": 3.8254863387978144,
      "grad_norm": 0.6427430282591454,
      "learning_rate": 9.910108758735437e-07,
      "loss": 11.383,
      "step": 21877
    },
    {
      "epoch": 3.8256612021857923,
      "grad_norm": 0.7916529391595738,
      "learning_rate": 9.890231836026998e-07,
      "loss": 11.453,
      "step": 21878
    },
    {
      "epoch": 3.8258360655737707,
      "grad_norm": 0.7443004444773483,
      "learning_rate": 9.87037476805952e-07,
      "loss": 11.312,
      "step": 21879
    },
    {
      "epoch": 3.8260109289617485,
      "grad_norm": 0.7158477080378937,
      "learning_rate": 9.85053755523102e-07,
      "loss": 11.4665,
      "step": 21880
    },
    {
      "epoch": 3.826185792349727,
      "grad_norm": 0.6722908288205226,
      "learning_rate": 9.830720197939403e-07,
      "loss": 11.3238,
      "step": 21881
    },
    {
      "epoch": 3.826360655737705,
      "grad_norm": 0.6475942961699777,
      "learning_rate": 9.810922696582014e-07,
      "loss": 11.122,
      "step": 21882
    },
    {
      "epoch": 3.826535519125683,
      "grad_norm": 0.7086169699368753,
      "learning_rate": 9.791145051555983e-07,
      "loss": 11.4588,
      "step": 21883
    },
    {
      "epoch": 3.826710382513661,
      "grad_norm": 0.7260357779937077,
      "learning_rate": 9.771387263257992e-07,
      "loss": 11.2463,
      "step": 21884
    },
    {
      "epoch": 3.8268852459016394,
      "grad_norm": 0.6952044374001818,
      "learning_rate": 9.751649332083946e-07,
      "loss": 11.3036,
      "step": 21885
    },
    {
      "epoch": 3.8270601092896177,
      "grad_norm": 0.6362459467421392,
      "learning_rate": 9.731931258429638e-07,
      "loss": 11.0827,
      "step": 21886
    },
    {
      "epoch": 3.8272349726775956,
      "grad_norm": 0.6619188743509409,
      "learning_rate": 9.712233042690754e-07,
      "loss": 11.2647,
      "step": 21887
    },
    {
      "epoch": 3.8274098360655735,
      "grad_norm": 0.661752151209525,
      "learning_rate": 9.69255468526231e-07,
      "loss": 11.2018,
      "step": 21888
    },
    {
      "epoch": 3.827584699453552,
      "grad_norm": 0.6290143418374939,
      "learning_rate": 9.67289618653866e-07,
      "loss": 11.6535,
      "step": 21889
    },
    {
      "epoch": 3.82775956284153,
      "grad_norm": 0.6683418464507891,
      "learning_rate": 9.653257546914151e-07,
      "loss": 11.5237,
      "step": 21890
    },
    {
      "epoch": 3.827934426229508,
      "grad_norm": 0.6471235044716898,
      "learning_rate": 9.633638766782582e-07,
      "loss": 11.6072,
      "step": 21891
    },
    {
      "epoch": 3.8281092896174864,
      "grad_norm": 0.6950710050602903,
      "learning_rate": 9.614039846537416e-07,
      "loss": 11.3388,
      "step": 21892
    },
    {
      "epoch": 3.8282841530054643,
      "grad_norm": 0.6893561489794595,
      "learning_rate": 9.594460786571669e-07,
      "loss": 11.2935,
      "step": 21893
    },
    {
      "epoch": 3.8284590163934427,
      "grad_norm": 0.5966150194875364,
      "learning_rate": 9.574901587277917e-07,
      "loss": 11.5264,
      "step": 21894
    },
    {
      "epoch": 3.828633879781421,
      "grad_norm": 0.6201162912394348,
      "learning_rate": 9.555362249048295e-07,
      "loss": 11.4034,
      "step": 21895
    },
    {
      "epoch": 3.828808743169399,
      "grad_norm": 0.7156687670312107,
      "learning_rate": 9.535842772274706e-07,
      "loss": 11.4529,
      "step": 21896
    },
    {
      "epoch": 3.828983606557377,
      "grad_norm": 0.6721961677523566,
      "learning_rate": 9.516343157348729e-07,
      "loss": 11.4158,
      "step": 21897
    },
    {
      "epoch": 3.829158469945355,
      "grad_norm": 0.6310070390478066,
      "learning_rate": 9.496863404661272e-07,
      "loss": 11.395,
      "step": 21898
    },
    {
      "epoch": 3.8293333333333335,
      "grad_norm": 0.6982147012145064,
      "learning_rate": 9.4774035146028e-07,
      "loss": 11.4838,
      "step": 21899
    },
    {
      "epoch": 3.8295081967213114,
      "grad_norm": 0.6700226466280377,
      "learning_rate": 9.457963487563781e-07,
      "loss": 11.4358,
      "step": 21900
    },
    {
      "epoch": 3.8296830601092897,
      "grad_norm": 0.6341162781855317,
      "learning_rate": 9.438543323933902e-07,
      "loss": 11.4748,
      "step": 21901
    },
    {
      "epoch": 3.8298579234972676,
      "grad_norm": 0.6694147441225394,
      "learning_rate": 9.419143024102739e-07,
      "loss": 11.4376,
      "step": 21902
    },
    {
      "epoch": 3.830032786885246,
      "grad_norm": 0.6906725882505649,
      "learning_rate": 9.399762588459204e-07,
      "loss": 11.351,
      "step": 21903
    },
    {
      "epoch": 3.8302076502732243,
      "grad_norm": 0.6605029423590011,
      "learning_rate": 9.380402017391988e-07,
      "loss": 11.5485,
      "step": 21904
    },
    {
      "epoch": 3.830382513661202,
      "grad_norm": 0.6800461766552839,
      "learning_rate": 9.36106131128922e-07,
      "loss": 11.5399,
      "step": 21905
    },
    {
      "epoch": 3.83055737704918,
      "grad_norm": 0.6579948643798262,
      "learning_rate": 9.341740470538929e-07,
      "loss": 11.5737,
      "step": 21906
    },
    {
      "epoch": 3.8307322404371584,
      "grad_norm": 0.7204895265485367,
      "learning_rate": 9.322439495528468e-07,
      "loss": 11.4379,
      "step": 21907
    },
    {
      "epoch": 3.8309071038251368,
      "grad_norm": 0.7061751352252764,
      "learning_rate": 9.303158386644861e-07,
      "loss": 11.46,
      "step": 21908
    },
    {
      "epoch": 3.8310819672131147,
      "grad_norm": 0.6399086062422564,
      "learning_rate": 9.283897144274689e-07,
      "loss": 11.3479,
      "step": 21909
    },
    {
      "epoch": 3.831256830601093,
      "grad_norm": 0.7013442212246377,
      "learning_rate": 9.264655768804309e-07,
      "loss": 11.2307,
      "step": 21910
    },
    {
      "epoch": 3.831431693989071,
      "grad_norm": 0.6680799996161982,
      "learning_rate": 9.245434260619634e-07,
      "loss": 11.4888,
      "step": 21911
    },
    {
      "epoch": 3.8316065573770492,
      "grad_norm": 0.6908264057630462,
      "learning_rate": 9.226232620105913e-07,
      "loss": 11.3839,
      "step": 21912
    },
    {
      "epoch": 3.8317814207650276,
      "grad_norm": 0.6757625526382697,
      "learning_rate": 9.20705084764839e-07,
      "loss": 11.4187,
      "step": 21913
    },
    {
      "epoch": 3.8319562841530055,
      "grad_norm": 0.6283624567455902,
      "learning_rate": 9.18788894363154e-07,
      "loss": 11.4528,
      "step": 21914
    },
    {
      "epoch": 3.8321311475409834,
      "grad_norm": 0.6548899357920547,
      "learning_rate": 9.168746908439718e-07,
      "loss": 11.6573,
      "step": 21915
    },
    {
      "epoch": 3.8323060109289617,
      "grad_norm": 0.7255382263244254,
      "learning_rate": 9.149624742456842e-07,
      "loss": 11.431,
      "step": 21916
    },
    {
      "epoch": 3.83248087431694,
      "grad_norm": 0.691488779634624,
      "learning_rate": 9.13052244606627e-07,
      "loss": 11.4837,
      "step": 21917
    },
    {
      "epoch": 3.832655737704918,
      "grad_norm": 0.6247546518790479,
      "learning_rate": 9.11144001965103e-07,
      "loss": 11.4017,
      "step": 21918
    },
    {
      "epoch": 3.8328306010928963,
      "grad_norm": 0.6639091846589765,
      "learning_rate": 9.092377463593926e-07,
      "loss": 11.506,
      "step": 21919
    },
    {
      "epoch": 3.833005464480874,
      "grad_norm": 0.6374672028118458,
      "learning_rate": 9.073334778277099e-07,
      "loss": 11.4929,
      "step": 21920
    },
    {
      "epoch": 3.8331803278688525,
      "grad_norm": 0.6691729668672207,
      "learning_rate": 9.054311964082463e-07,
      "loss": 11.3834,
      "step": 21921
    },
    {
      "epoch": 3.8333551912568304,
      "grad_norm": 0.6529021397480438,
      "learning_rate": 9.035309021391492e-07,
      "loss": 11.6075,
      "step": 21922
    },
    {
      "epoch": 3.833530054644809,
      "grad_norm": 0.6264402585099426,
      "learning_rate": 9.016325950585213e-07,
      "loss": 11.264,
      "step": 21923
    },
    {
      "epoch": 3.8337049180327867,
      "grad_norm": 0.6673018229666562,
      "learning_rate": 8.997362752044325e-07,
      "loss": 11.7068,
      "step": 21924
    },
    {
      "epoch": 3.833879781420765,
      "grad_norm": 0.6166018429559383,
      "learning_rate": 8.978419426149187e-07,
      "loss": 11.2792,
      "step": 21925
    },
    {
      "epoch": 3.8340546448087434,
      "grad_norm": 0.6862107287729096,
      "learning_rate": 8.959495973279386e-07,
      "loss": 11.0948,
      "step": 21926
    },
    {
      "epoch": 3.8342295081967213,
      "grad_norm": 0.5966576519819649,
      "learning_rate": 8.940592393814728e-07,
      "loss": 11.3609,
      "step": 21927
    },
    {
      "epoch": 3.8344043715846996,
      "grad_norm": 0.751395561853647,
      "learning_rate": 8.921708688134134e-07,
      "loss": 11.1914,
      "step": 21928
    },
    {
      "epoch": 3.8345792349726775,
      "grad_norm": 0.6593722335954385,
      "learning_rate": 8.902844856616188e-07,
      "loss": 11.5174,
      "step": 21929
    },
    {
      "epoch": 3.834754098360656,
      "grad_norm": 0.6699267338082129,
      "learning_rate": 8.884000899639367e-07,
      "loss": 11.4122,
      "step": 21930
    },
    {
      "epoch": 3.8349289617486337,
      "grad_norm": 0.7055945491357224,
      "learning_rate": 8.865176817581477e-07,
      "loss": 11.2857,
      "step": 21931
    },
    {
      "epoch": 3.835103825136612,
      "grad_norm": 0.6256221786161617,
      "learning_rate": 8.846372610819886e-07,
      "loss": 11.6581,
      "step": 21932
    },
    {
      "epoch": 3.83527868852459,
      "grad_norm": 0.6743577617863361,
      "learning_rate": 8.827588279731846e-07,
      "loss": 11.2849,
      "step": 21933
    },
    {
      "epoch": 3.8354535519125683,
      "grad_norm": 0.722596082442665,
      "learning_rate": 8.808823824693835e-07,
      "loss": 11.4992,
      "step": 21934
    },
    {
      "epoch": 3.8356284153005467,
      "grad_norm": 0.6696931213940597,
      "learning_rate": 8.790079246082328e-07,
      "loss": 11.3826,
      "step": 21935
    },
    {
      "epoch": 3.8358032786885246,
      "grad_norm": 0.6808102640594208,
      "learning_rate": 8.771354544273247e-07,
      "loss": 11.3116,
      "step": 21936
    },
    {
      "epoch": 3.835978142076503,
      "grad_norm": 0.6417810917068674,
      "learning_rate": 8.752649719641848e-07,
      "loss": 11.4471,
      "step": 21937
    },
    {
      "epoch": 3.836153005464481,
      "grad_norm": 0.6046412597700564,
      "learning_rate": 8.733964772563385e-07,
      "loss": 11.1269,
      "step": 21938
    },
    {
      "epoch": 3.836327868852459,
      "grad_norm": 0.6898092336651087,
      "learning_rate": 8.71529970341256e-07,
      "loss": 11.5603,
      "step": 21939
    },
    {
      "epoch": 3.836502732240437,
      "grad_norm": 0.7196797086365407,
      "learning_rate": 8.696654512563629e-07,
      "loss": 11.5465,
      "step": 21940
    },
    {
      "epoch": 3.8366775956284154,
      "grad_norm": 0.6599462567931197,
      "learning_rate": 8.678029200390403e-07,
      "loss": 11.2892,
      "step": 21941
    },
    {
      "epoch": 3.8368524590163933,
      "grad_norm": 0.7156920383566882,
      "learning_rate": 8.659423767266473e-07,
      "loss": 11.2913,
      "step": 21942
    },
    {
      "epoch": 3.8370273224043716,
      "grad_norm": 0.6516435503734149,
      "learning_rate": 8.640838213564984e-07,
      "loss": 11.0341,
      "step": 21943
    },
    {
      "epoch": 3.83720218579235,
      "grad_norm": 0.7036174623426099,
      "learning_rate": 8.622272539658415e-07,
      "loss": 11.3417,
      "step": 21944
    },
    {
      "epoch": 3.837377049180328,
      "grad_norm": 0.7012389060074209,
      "learning_rate": 8.603726745919361e-07,
      "loss": 11.0425,
      "step": 21945
    },
    {
      "epoch": 3.8375519125683057,
      "grad_norm": 0.612678464094987,
      "learning_rate": 8.585200832719409e-07,
      "loss": 11.0823,
      "step": 21946
    },
    {
      "epoch": 3.837726775956284,
      "grad_norm": 0.6418117816859762,
      "learning_rate": 8.566694800430375e-07,
      "loss": 11.3845,
      "step": 21947
    },
    {
      "epoch": 3.8379016393442624,
      "grad_norm": 0.6869786478167403,
      "learning_rate": 8.548208649423184e-07,
      "loss": 11.4193,
      "step": 21948
    },
    {
      "epoch": 3.8380765027322403,
      "grad_norm": 0.6746338096200595,
      "learning_rate": 8.529742380068539e-07,
      "loss": 11.5766,
      "step": 21949
    },
    {
      "epoch": 3.8382513661202187,
      "grad_norm": 0.6658759141562953,
      "learning_rate": 8.511295992736701e-07,
      "loss": 11.3905,
      "step": 21950
    },
    {
      "epoch": 3.8384262295081966,
      "grad_norm": 0.655137047512902,
      "learning_rate": 8.492869487797705e-07,
      "loss": 11.532,
      "step": 21951
    },
    {
      "epoch": 3.838601092896175,
      "grad_norm": 0.658784480375302,
      "learning_rate": 8.474462865621147e-07,
      "loss": 11.6078,
      "step": 21952
    },
    {
      "epoch": 3.8387759562841532,
      "grad_norm": 0.7651780710665882,
      "learning_rate": 8.456076126575729e-07,
      "loss": 11.2297,
      "step": 21953
    },
    {
      "epoch": 3.838950819672131,
      "grad_norm": 0.6939312019505577,
      "learning_rate": 8.437709271030603e-07,
      "loss": 11.433,
      "step": 21954
    },
    {
      "epoch": 3.839125683060109,
      "grad_norm": 0.6521054889204686,
      "learning_rate": 8.419362299353806e-07,
      "loss": 11.4894,
      "step": 21955
    },
    {
      "epoch": 3.8393005464480874,
      "grad_norm": 0.666299052265028,
      "learning_rate": 8.401035211913488e-07,
      "loss": 11.3356,
      "step": 21956
    },
    {
      "epoch": 3.8394754098360657,
      "grad_norm": 0.6668890916922067,
      "learning_rate": 8.382728009076912e-07,
      "loss": 11.3296,
      "step": 21957
    },
    {
      "epoch": 3.8396502732240436,
      "grad_norm": 0.6353824553231889,
      "learning_rate": 8.364440691211339e-07,
      "loss": 10.9595,
      "step": 21958
    },
    {
      "epoch": 3.839825136612022,
      "grad_norm": 0.7451253665685288,
      "learning_rate": 8.346173258683365e-07,
      "loss": 11.1835,
      "step": 21959
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6707805165581373,
      "learning_rate": 8.327925711859475e-07,
      "loss": 11.3747,
      "step": 21960
    },
    {
      "epoch": 3.840174863387978,
      "grad_norm": 0.6621465047416782,
      "learning_rate": 8.309698051105596e-07,
      "loss": 11.5924,
      "step": 21961
    },
    {
      "epoch": 3.8403497267759565,
      "grad_norm": 0.7697625941060391,
      "learning_rate": 8.291490276786995e-07,
      "loss": 11.5118,
      "step": 21962
    },
    {
      "epoch": 3.8405245901639344,
      "grad_norm": 0.6403957304902199,
      "learning_rate": 8.273302389269044e-07,
      "loss": 11.4195,
      "step": 21963
    },
    {
      "epoch": 3.8406994535519123,
      "grad_norm": 0.6328139390609253,
      "learning_rate": 8.255134388916341e-07,
      "loss": 11.1828,
      "step": 21964
    },
    {
      "epoch": 3.8408743169398907,
      "grad_norm": 0.687546761771262,
      "learning_rate": 8.236986276093261e-07,
      "loss": 11.3077,
      "step": 21965
    },
    {
      "epoch": 3.841049180327869,
      "grad_norm": 0.6762937683929173,
      "learning_rate": 8.218858051163846e-07,
      "loss": 11.3056,
      "step": 21966
    },
    {
      "epoch": 3.841224043715847,
      "grad_norm": 0.7272923070570025,
      "learning_rate": 8.200749714491363e-07,
      "loss": 11.4185,
      "step": 21967
    },
    {
      "epoch": 3.8413989071038253,
      "grad_norm": 0.6774162278637093,
      "learning_rate": 8.182661266438962e-07,
      "loss": 11.622,
      "step": 21968
    },
    {
      "epoch": 3.841573770491803,
      "grad_norm": 0.7226161080227453,
      "learning_rate": 8.16459270736969e-07,
      "loss": 11.3038,
      "step": 21969
    },
    {
      "epoch": 3.8417486338797815,
      "grad_norm": 0.6783359580179432,
      "learning_rate": 8.146544037645587e-07,
      "loss": 11.4587,
      "step": 21970
    },
    {
      "epoch": 3.84192349726776,
      "grad_norm": 0.6709481582928194,
      "learning_rate": 8.1285152576287e-07,
      "loss": 11.1858,
      "step": 21971
    },
    {
      "epoch": 3.8420983606557377,
      "grad_norm": 0.6227974603330118,
      "learning_rate": 8.110506367680515e-07,
      "loss": 11.643,
      "step": 21972
    },
    {
      "epoch": 3.8422732240437156,
      "grad_norm": 0.6661412700441564,
      "learning_rate": 8.092517368162078e-07,
      "loss": 11.4723,
      "step": 21973
    },
    {
      "epoch": 3.842448087431694,
      "grad_norm": 0.6663345198540387,
      "learning_rate": 8.074548259434322e-07,
      "loss": 11.426,
      "step": 21974
    },
    {
      "epoch": 3.8426229508196723,
      "grad_norm": 0.6474652075445541,
      "learning_rate": 8.056599041857516e-07,
      "loss": 11.3809,
      "step": 21975
    },
    {
      "epoch": 3.84279781420765,
      "grad_norm": 0.760805464089859,
      "learning_rate": 8.038669715791481e-07,
      "loss": 11.3447,
      "step": 21976
    },
    {
      "epoch": 3.8429726775956286,
      "grad_norm": 0.7238962139610141,
      "learning_rate": 8.020760281595818e-07,
      "loss": 11.2488,
      "step": 21977
    },
    {
      "epoch": 3.8431475409836064,
      "grad_norm": 0.6901194485277039,
      "learning_rate": 8.002870739629687e-07,
      "loss": 11.4745,
      "step": 21978
    },
    {
      "epoch": 3.843322404371585,
      "grad_norm": 0.690369915900958,
      "learning_rate": 7.98500109025202e-07,
      "loss": 11.4595,
      "step": 21979
    },
    {
      "epoch": 3.8434972677595627,
      "grad_norm": 0.7172845636284333,
      "learning_rate": 7.967151333820755e-07,
      "loss": 11.0971,
      "step": 21980
    },
    {
      "epoch": 3.843672131147541,
      "grad_norm": 0.7099497985098622,
      "learning_rate": 7.94932147069416e-07,
      "loss": 11.2715,
      "step": 21981
    },
    {
      "epoch": 3.843846994535519,
      "grad_norm": 0.6842065530121006,
      "learning_rate": 7.931511501229616e-07,
      "loss": 11.4123,
      "step": 21982
    },
    {
      "epoch": 3.8440218579234973,
      "grad_norm": 0.7445348786153836,
      "learning_rate": 7.913721425784282e-07,
      "loss": 11.598,
      "step": 21983
    },
    {
      "epoch": 3.8441967213114756,
      "grad_norm": 0.6761115144019276,
      "learning_rate": 7.895951244715094e-07,
      "loss": 11.4799,
      "step": 21984
    },
    {
      "epoch": 3.8443715846994535,
      "grad_norm": 0.6263910839768817,
      "learning_rate": 7.878200958378102e-07,
      "loss": 11.4815,
      "step": 21985
    },
    {
      "epoch": 3.844546448087432,
      "grad_norm": 0.6596705165243563,
      "learning_rate": 7.860470567129464e-07,
      "loss": 11.5304,
      "step": 21986
    },
    {
      "epoch": 3.8447213114754097,
      "grad_norm": 0.6251464388249409,
      "learning_rate": 7.842760071324562e-07,
      "loss": 11.5548,
      "step": 21987
    },
    {
      "epoch": 3.844896174863388,
      "grad_norm": 0.6992885127184559,
      "learning_rate": 7.825069471318669e-07,
      "loss": 11.5929,
      "step": 21988
    },
    {
      "epoch": 3.845071038251366,
      "grad_norm": 0.6362380120720206,
      "learning_rate": 7.807398767466611e-07,
      "loss": 11.294,
      "step": 21989
    },
    {
      "epoch": 3.8452459016393443,
      "grad_norm": 0.6802040354462128,
      "learning_rate": 7.789747960122551e-07,
      "loss": 11.4465,
      "step": 21990
    },
    {
      "epoch": 3.845420765027322,
      "grad_norm": 0.6508789387684776,
      "learning_rate": 7.772117049640537e-07,
      "loss": 11.5234,
      "step": 21991
    },
    {
      "epoch": 3.8455956284153006,
      "grad_norm": 0.6863611255057128,
      "learning_rate": 7.754506036374065e-07,
      "loss": 11.3396,
      "step": 21992
    },
    {
      "epoch": 3.845770491803279,
      "grad_norm": 0.7058827513392745,
      "learning_rate": 7.736914920676519e-07,
      "loss": 11.3468,
      "step": 21993
    },
    {
      "epoch": 3.845945355191257,
      "grad_norm": 0.6226957715838419,
      "learning_rate": 7.719343702900395e-07,
      "loss": 11.4633,
      "step": 21994
    },
    {
      "epoch": 3.8461202185792347,
      "grad_norm": 0.7065501148171973,
      "learning_rate": 7.701792383398076e-07,
      "loss": 11.5211,
      "step": 21995
    },
    {
      "epoch": 3.846295081967213,
      "grad_norm": 0.6823680112657032,
      "learning_rate": 7.684260962521616e-07,
      "loss": 11.4577,
      "step": 21996
    },
    {
      "epoch": 3.8464699453551914,
      "grad_norm": 0.6391499687124038,
      "learning_rate": 7.6667494406224e-07,
      "loss": 11.338,
      "step": 21997
    },
    {
      "epoch": 3.8466448087431693,
      "grad_norm": 0.6583197016914861,
      "learning_rate": 7.649257818051925e-07,
      "loss": 11.4246,
      "step": 21998
    },
    {
      "epoch": 3.8468196721311476,
      "grad_norm": 0.7136202627953535,
      "learning_rate": 7.631786095160687e-07,
      "loss": 11.3457,
      "step": 21999
    },
    {
      "epoch": 3.8469945355191255,
      "grad_norm": 0.6263387468165091,
      "learning_rate": 7.614334272299072e-07,
      "loss": 11.3589,
      "step": 22000
    },
    {
      "epoch": 3.847169398907104,
      "grad_norm": 0.6479948096047935,
      "learning_rate": 7.596902349817026e-07,
      "loss": 11.6007,
      "step": 22001
    },
    {
      "epoch": 3.847344262295082,
      "grad_norm": 0.7397879634683296,
      "learning_rate": 7.579490328064265e-07,
      "loss": 11.1832,
      "step": 22002
    },
    {
      "epoch": 3.84751912568306,
      "grad_norm": 0.6571932175095075,
      "learning_rate": 7.562098207389845e-07,
      "loss": 11.4758,
      "step": 22003
    },
    {
      "epoch": 3.847693989071038,
      "grad_norm": 0.6406733847136634,
      "learning_rate": 7.544725988142487e-07,
      "loss": 11.7072,
      "step": 22004
    },
    {
      "epoch": 3.8478688524590163,
      "grad_norm": 0.7064113523990561,
      "learning_rate": 7.52737367067069e-07,
      "loss": 11.4953,
      "step": 22005
    },
    {
      "epoch": 3.8480437158469947,
      "grad_norm": 0.6477947457992583,
      "learning_rate": 7.510041255322175e-07,
      "loss": 11.391,
      "step": 22006
    },
    {
      "epoch": 3.8482185792349726,
      "grad_norm": 0.6652539757980348,
      "learning_rate": 7.492728742444888e-07,
      "loss": 11.0471,
      "step": 22007
    },
    {
      "epoch": 3.848393442622951,
      "grad_norm": 0.6560254715256809,
      "learning_rate": 7.47543613238566e-07,
      "loss": 11.4,
      "step": 22008
    },
    {
      "epoch": 3.848568306010929,
      "grad_norm": 0.6398492242111029,
      "learning_rate": 7.458163425491438e-07,
      "loss": 11.2504,
      "step": 22009
    },
    {
      "epoch": 3.848743169398907,
      "grad_norm": 0.6830880757768276,
      "learning_rate": 7.440910622108499e-07,
      "loss": 11.2915,
      "step": 22010
    },
    {
      "epoch": 3.8489180327868855,
      "grad_norm": 0.6820168906547182,
      "learning_rate": 7.423677722582901e-07,
      "loss": 11.5004,
      "step": 22011
    },
    {
      "epoch": 3.8490928961748634,
      "grad_norm": 0.7112802079904618,
      "learning_rate": 7.406464727260142e-07,
      "loss": 11.4492,
      "step": 22012
    },
    {
      "epoch": 3.8492677595628413,
      "grad_norm": 0.681049101306705,
      "learning_rate": 7.389271636485507e-07,
      "loss": 11.3309,
      "step": 22013
    },
    {
      "epoch": 3.8494426229508196,
      "grad_norm": 0.7154650518716641,
      "learning_rate": 7.372098450603604e-07,
      "loss": 11.2132,
      "step": 22014
    },
    {
      "epoch": 3.849617486338798,
      "grad_norm": 0.641768937082933,
      "learning_rate": 7.354945169958938e-07,
      "loss": 11.2455,
      "step": 22015
    },
    {
      "epoch": 3.849792349726776,
      "grad_norm": 0.6609705780102639,
      "learning_rate": 7.337811794895344e-07,
      "loss": 11.4425,
      "step": 22016
    },
    {
      "epoch": 3.849967213114754,
      "grad_norm": 0.7549238358639961,
      "learning_rate": 7.320698325756658e-07,
      "loss": 11.1664,
      "step": 22017
    },
    {
      "epoch": 3.850142076502732,
      "grad_norm": 0.6211930518844592,
      "learning_rate": 7.303604762885719e-07,
      "loss": 11.4477,
      "step": 22018
    },
    {
      "epoch": 3.8503169398907104,
      "grad_norm": 0.6763490774290648,
      "learning_rate": 7.286531106625582e-07,
      "loss": 11.1843,
      "step": 22019
    },
    {
      "epoch": 3.850491803278689,
      "grad_norm": 0.6480901514499499,
      "learning_rate": 7.269477357318532e-07,
      "loss": 11.5838,
      "step": 22020
    },
    {
      "epoch": 3.8506666666666667,
      "grad_norm": 0.6881006991153837,
      "learning_rate": 7.252443515306518e-07,
      "loss": 11.3271,
      "step": 22021
    },
    {
      "epoch": 3.8508415300546446,
      "grad_norm": 0.7074197877971741,
      "learning_rate": 7.235429580931152e-07,
      "loss": 11.5497,
      "step": 22022
    },
    {
      "epoch": 3.851016393442623,
      "grad_norm": 0.684903547374297,
      "learning_rate": 7.21843555453372e-07,
      "loss": 11.1631,
      "step": 22023
    },
    {
      "epoch": 3.8511912568306013,
      "grad_norm": 0.6622362716360745,
      "learning_rate": 7.201461436454727e-07,
      "loss": 11.488,
      "step": 22024
    },
    {
      "epoch": 3.851366120218579,
      "grad_norm": 0.7500827155124092,
      "learning_rate": 7.184507227034787e-07,
      "loss": 11.4261,
      "step": 22025
    },
    {
      "epoch": 3.8515409836065575,
      "grad_norm": 0.6473607228508905,
      "learning_rate": 7.167572926613853e-07,
      "loss": 11.4873,
      "step": 22026
    },
    {
      "epoch": 3.8517158469945354,
      "grad_norm": 0.6753613432623512,
      "learning_rate": 7.15065853553143e-07,
      "loss": 11.4014,
      "step": 22027
    },
    {
      "epoch": 3.8518907103825137,
      "grad_norm": 0.6898535268246914,
      "learning_rate": 7.133764054126802e-07,
      "loss": 11.0459,
      "step": 22028
    },
    {
      "epoch": 3.852065573770492,
      "grad_norm": 0.7009286732464997,
      "learning_rate": 7.116889482738698e-07,
      "loss": 11.4365,
      "step": 22029
    },
    {
      "epoch": 3.85224043715847,
      "grad_norm": 0.7094788147566088,
      "learning_rate": 7.100034821705515e-07,
      "loss": 11.2944,
      "step": 22030
    },
    {
      "epoch": 3.852415300546448,
      "grad_norm": 0.6508285545608977,
      "learning_rate": 7.083200071365203e-07,
      "loss": 11.6232,
      "step": 22031
    },
    {
      "epoch": 3.852590163934426,
      "grad_norm": 0.735760841763154,
      "learning_rate": 7.066385232055384e-07,
      "loss": 10.9729,
      "step": 22032
    },
    {
      "epoch": 3.8527650273224046,
      "grad_norm": 0.6711627708751032,
      "learning_rate": 7.04959030411323e-07,
      "loss": 11.5789,
      "step": 22033
    },
    {
      "epoch": 3.8529398907103825,
      "grad_norm": 0.6784014101157412,
      "learning_rate": 7.032815287875582e-07,
      "loss": 11.6522,
      "step": 22034
    },
    {
      "epoch": 3.853114754098361,
      "grad_norm": 0.6647372298998481,
      "learning_rate": 7.01606018367873e-07,
      "loss": 11.153,
      "step": 22035
    },
    {
      "epoch": 3.8532896174863387,
      "grad_norm": 0.7197709290276476,
      "learning_rate": 6.999324991858847e-07,
      "loss": 11.3586,
      "step": 22036
    },
    {
      "epoch": 3.853464480874317,
      "grad_norm": 0.6199569744537206,
      "learning_rate": 6.982609712751332e-07,
      "loss": 11.4574,
      "step": 22037
    },
    {
      "epoch": 3.853639344262295,
      "grad_norm": 0.671562509936662,
      "learning_rate": 6.965914346691471e-07,
      "loss": 11.5505,
      "step": 22038
    },
    {
      "epoch": 3.8538142076502733,
      "grad_norm": 0.663102616742214,
      "learning_rate": 6.94923889401411e-07,
      "loss": 11.5777,
      "step": 22039
    },
    {
      "epoch": 3.853989071038251,
      "grad_norm": 0.7599515159920017,
      "learning_rate": 6.932583355053535e-07,
      "loss": 11.3245,
      "step": 22040
    },
    {
      "epoch": 3.8541639344262295,
      "grad_norm": 0.6877832948293062,
      "learning_rate": 6.915947730143702e-07,
      "loss": 11.6029,
      "step": 22041
    },
    {
      "epoch": 3.854338797814208,
      "grad_norm": 0.6791963647973908,
      "learning_rate": 6.899332019618343e-07,
      "loss": 11.3601,
      "step": 22042
    },
    {
      "epoch": 3.8545136612021857,
      "grad_norm": 0.7150658833942669,
      "learning_rate": 6.882736223810638e-07,
      "loss": 11.2194,
      "step": 22043
    },
    {
      "epoch": 3.854688524590164,
      "grad_norm": 0.6816842976856098,
      "learning_rate": 6.866160343053318e-07,
      "loss": 11.404,
      "step": 22044
    },
    {
      "epoch": 3.854863387978142,
      "grad_norm": 0.7487335922469358,
      "learning_rate": 6.849604377678786e-07,
      "loss": 11.6025,
      "step": 22045
    },
    {
      "epoch": 3.8550382513661203,
      "grad_norm": 0.6101404617693988,
      "learning_rate": 6.833068328018998e-07,
      "loss": 11.3698,
      "step": 22046
    },
    {
      "epoch": 3.8552131147540982,
      "grad_norm": 0.6600869470100982,
      "learning_rate": 6.81655219440569e-07,
      "loss": 11.354,
      "step": 22047
    },
    {
      "epoch": 3.8553879781420766,
      "grad_norm": 0.6508310708080952,
      "learning_rate": 6.800055977170039e-07,
      "loss": 11.0952,
      "step": 22048
    },
    {
      "epoch": 3.8555628415300545,
      "grad_norm": 0.6631331326837686,
      "learning_rate": 6.783579676642671e-07,
      "loss": 11.4126,
      "step": 22049
    },
    {
      "epoch": 3.855737704918033,
      "grad_norm": 0.6040382323893769,
      "learning_rate": 6.767123293154209e-07,
      "loss": 11.4147,
      "step": 22050
    },
    {
      "epoch": 3.855912568306011,
      "grad_norm": 0.660770688253936,
      "learning_rate": 6.750686827034503e-07,
      "loss": 11.3416,
      "step": 22051
    },
    {
      "epoch": 3.856087431693989,
      "grad_norm": 0.6519930627556171,
      "learning_rate": 6.734270278613175e-07,
      "loss": 11.6686,
      "step": 22052
    },
    {
      "epoch": 3.856262295081967,
      "grad_norm": 0.645692548684563,
      "learning_rate": 6.71787364821952e-07,
      "loss": 11.4097,
      "step": 22053
    },
    {
      "epoch": 3.8564371584699453,
      "grad_norm": 0.6466212898126016,
      "learning_rate": 6.701496936182161e-07,
      "loss": 11.3789,
      "step": 22054
    },
    {
      "epoch": 3.8566120218579236,
      "grad_norm": 0.7933190671388267,
      "learning_rate": 6.685140142829727e-07,
      "loss": 11.5535,
      "step": 22055
    },
    {
      "epoch": 3.8567868852459015,
      "grad_norm": 0.680882671598621,
      "learning_rate": 6.668803268490064e-07,
      "loss": 11.1706,
      "step": 22056
    },
    {
      "epoch": 3.85696174863388,
      "grad_norm": 0.6057172293131582,
      "learning_rate": 6.652486313490913e-07,
      "loss": 11.5755,
      "step": 22057
    },
    {
      "epoch": 3.8571366120218578,
      "grad_norm": 0.6565944271968543,
      "learning_rate": 6.636189278159233e-07,
      "loss": 11.324,
      "step": 22058
    },
    {
      "epoch": 3.857311475409836,
      "grad_norm": 0.6155419008437166,
      "learning_rate": 6.619912162821984e-07,
      "loss": 11.1201,
      "step": 22059
    },
    {
      "epoch": 3.8574863387978144,
      "grad_norm": 0.7100682633627571,
      "learning_rate": 6.603654967805683e-07,
      "loss": 11.391,
      "step": 22060
    },
    {
      "epoch": 3.8576612021857923,
      "grad_norm": 0.6454866832941606,
      "learning_rate": 6.587417693436182e-07,
      "loss": 11.4685,
      "step": 22061
    },
    {
      "epoch": 3.8578360655737702,
      "grad_norm": 0.7773377648755078,
      "learning_rate": 6.571200340039218e-07,
      "loss": 11.2297,
      "step": 22062
    },
    {
      "epoch": 3.8580109289617486,
      "grad_norm": 0.6588947848478085,
      "learning_rate": 6.555002907939867e-07,
      "loss": 11.3723,
      "step": 22063
    },
    {
      "epoch": 3.858185792349727,
      "grad_norm": 0.6876212496363052,
      "learning_rate": 6.538825397462866e-07,
      "loss": 11.3013,
      "step": 22064
    },
    {
      "epoch": 3.858360655737705,
      "grad_norm": 0.6959682161243831,
      "learning_rate": 6.522667808932847e-07,
      "loss": 11.4681,
      "step": 22065
    },
    {
      "epoch": 3.858535519125683,
      "grad_norm": 0.6965621031232831,
      "learning_rate": 6.506530142673773e-07,
      "loss": 11.4246,
      "step": 22066
    },
    {
      "epoch": 3.858710382513661,
      "grad_norm": 0.8290043380508781,
      "learning_rate": 6.490412399009049e-07,
      "loss": 11.2185,
      "step": 22067
    },
    {
      "epoch": 3.8588852459016394,
      "grad_norm": 0.6705615125829036,
      "learning_rate": 6.474314578262198e-07,
      "loss": 11.4897,
      "step": 22068
    },
    {
      "epoch": 3.8590601092896177,
      "grad_norm": 0.6686423443641821,
      "learning_rate": 6.458236680755736e-07,
      "loss": 11.1883,
      "step": 22069
    },
    {
      "epoch": 3.8592349726775956,
      "grad_norm": 0.6742694741785096,
      "learning_rate": 6.442178706812186e-07,
      "loss": 11.4767,
      "step": 22070
    },
    {
      "epoch": 3.8594098360655735,
      "grad_norm": 0.6490278284064946,
      "learning_rate": 6.426140656753621e-07,
      "loss": 11.4194,
      "step": 22071
    },
    {
      "epoch": 3.859584699453552,
      "grad_norm": 0.6603271139747051,
      "learning_rate": 6.410122530901675e-07,
      "loss": 11.4654,
      "step": 22072
    },
    {
      "epoch": 3.85975956284153,
      "grad_norm": 0.6480174362186651,
      "learning_rate": 6.394124329577422e-07,
      "loss": 11.1792,
      "step": 22073
    },
    {
      "epoch": 3.859934426229508,
      "grad_norm": 0.645531343858674,
      "learning_rate": 6.378146053101719e-07,
      "loss": 11.3429,
      "step": 22074
    },
    {
      "epoch": 3.8601092896174864,
      "grad_norm": 0.6680674708194584,
      "learning_rate": 6.362187701795086e-07,
      "loss": 11.3501,
      "step": 22075
    },
    {
      "epoch": 3.8602841530054643,
      "grad_norm": 0.9155004043912357,
      "learning_rate": 6.346249275977378e-07,
      "loss": 11.2621,
      "step": 22076
    },
    {
      "epoch": 3.8604590163934427,
      "grad_norm": 0.679894964219641,
      "learning_rate": 6.330330775968341e-07,
      "loss": 11.5425,
      "step": 22077
    },
    {
      "epoch": 3.860633879781421,
      "grad_norm": 0.7058621671263934,
      "learning_rate": 6.314432202087162e-07,
      "loss": 11.4316,
      "step": 22078
    },
    {
      "epoch": 3.860808743169399,
      "grad_norm": 0.66143192347435,
      "learning_rate": 6.298553554652586e-07,
      "loss": 11.5334,
      "step": 22079
    },
    {
      "epoch": 3.860983606557377,
      "grad_norm": 0.8295994208904902,
      "learning_rate": 6.282694833983138e-07,
      "loss": 11.448,
      "step": 22080
    },
    {
      "epoch": 3.861158469945355,
      "grad_norm": 0.6420783153017303,
      "learning_rate": 6.266856040396673e-07,
      "loss": 11.2394,
      "step": 22081
    },
    {
      "epoch": 3.8613333333333335,
      "grad_norm": 0.649249911717091,
      "learning_rate": 6.251037174210938e-07,
      "loss": 11.5193,
      "step": 22082
    },
    {
      "epoch": 3.8615081967213114,
      "grad_norm": 0.6977528279359849,
      "learning_rate": 6.235238235743123e-07,
      "loss": 11.4014,
      "step": 22083
    },
    {
      "epoch": 3.8616830601092897,
      "grad_norm": 0.6712226198586307,
      "learning_rate": 6.219459225310087e-07,
      "loss": 11.3647,
      "step": 22084
    },
    {
      "epoch": 3.8618579234972676,
      "grad_norm": 0.6480243231540358,
      "learning_rate": 6.203700143228241e-07,
      "loss": 11.6033,
      "step": 22085
    },
    {
      "epoch": 3.862032786885246,
      "grad_norm": 0.712546845613772,
      "learning_rate": 6.187960989813446e-07,
      "loss": 11.351,
      "step": 22086
    },
    {
      "epoch": 3.862207650273224,
      "grad_norm": 0.6487372777540398,
      "learning_rate": 6.172241765381448e-07,
      "loss": 11.3449,
      "step": 22087
    },
    {
      "epoch": 3.862382513661202,
      "grad_norm": 0.6953212272361055,
      "learning_rate": 6.15654247024744e-07,
      "loss": 11.4639,
      "step": 22088
    },
    {
      "epoch": 3.86255737704918,
      "grad_norm": 0.7042679141499218,
      "learning_rate": 6.140863104726391e-07,
      "loss": 11.4661,
      "step": 22089
    },
    {
      "epoch": 3.8627322404371585,
      "grad_norm": 0.6638807097184897,
      "learning_rate": 6.125203669132496e-07,
      "loss": 11.5282,
      "step": 22090
    },
    {
      "epoch": 3.862907103825137,
      "grad_norm": 0.7474315156394189,
      "learning_rate": 6.109564163779835e-07,
      "loss": 11.3772,
      "step": 22091
    },
    {
      "epoch": 3.8630819672131147,
      "grad_norm": 0.6393177746748945,
      "learning_rate": 6.093944588982048e-07,
      "loss": 11.5746,
      "step": 22092
    },
    {
      "epoch": 3.863256830601093,
      "grad_norm": 0.6863274771720096,
      "learning_rate": 6.078344945052438e-07,
      "loss": 11.4711,
      "step": 22093
    },
    {
      "epoch": 3.863431693989071,
      "grad_norm": 0.6360214921436323,
      "learning_rate": 6.062765232303758e-07,
      "loss": 11.4153,
      "step": 22094
    },
    {
      "epoch": 3.8636065573770493,
      "grad_norm": 0.6581332852901384,
      "learning_rate": 6.04720545104831e-07,
      "loss": 11.446,
      "step": 22095
    },
    {
      "epoch": 3.863781420765027,
      "grad_norm": 0.7020007242646591,
      "learning_rate": 6.031665601598291e-07,
      "loss": 11.5219,
      "step": 22096
    },
    {
      "epoch": 3.8639562841530055,
      "grad_norm": 0.7055850916061026,
      "learning_rate": 6.016145684265229e-07,
      "loss": 11.1924,
      "step": 22097
    },
    {
      "epoch": 3.8641311475409834,
      "grad_norm": 0.65042437871336,
      "learning_rate": 6.000645699360541e-07,
      "loss": 11.2669,
      "step": 22098
    },
    {
      "epoch": 3.8643060109289618,
      "grad_norm": 0.666623448740973,
      "learning_rate": 5.985165647194757e-07,
      "loss": 11.6448,
      "step": 22099
    },
    {
      "epoch": 3.86448087431694,
      "grad_norm": 0.6921207993591725,
      "learning_rate": 5.969705528078406e-07,
      "loss": 11.5887,
      "step": 22100
    },
    {
      "epoch": 3.864655737704918,
      "grad_norm": 0.6125711203839248,
      "learning_rate": 5.954265342321574e-07,
      "loss": 11.253,
      "step": 22101
    },
    {
      "epoch": 3.8648306010928963,
      "grad_norm": 0.653487789016056,
      "learning_rate": 5.938845090233791e-07,
      "loss": 11.2579,
      "step": 22102
    },
    {
      "epoch": 3.8650054644808742,
      "grad_norm": 0.6736877822956242,
      "learning_rate": 5.923444772124365e-07,
      "loss": 11.2308,
      "step": 22103
    },
    {
      "epoch": 3.8651803278688526,
      "grad_norm": 0.6457556580587083,
      "learning_rate": 5.908064388302048e-07,
      "loss": 11.4945,
      "step": 22104
    },
    {
      "epoch": 3.8653551912568305,
      "grad_norm": 0.641413501085783,
      "learning_rate": 5.892703939075372e-07,
      "loss": 11.5303,
      "step": 22105
    },
    {
      "epoch": 3.865530054644809,
      "grad_norm": 0.6233720744058917,
      "learning_rate": 5.877363424752313e-07,
      "loss": 11.4359,
      "step": 22106
    },
    {
      "epoch": 3.8657049180327867,
      "grad_norm": 0.78457463388623,
      "learning_rate": 5.862042845640403e-07,
      "loss": 11.3772,
      "step": 22107
    },
    {
      "epoch": 3.865879781420765,
      "grad_norm": 0.6988771959270693,
      "learning_rate": 5.846742202046951e-07,
      "loss": 11.4859,
      "step": 22108
    },
    {
      "epoch": 3.8660546448087434,
      "grad_norm": 0.6601041301296979,
      "learning_rate": 5.831461494278823e-07,
      "loss": 11.3945,
      "step": 22109
    },
    {
      "epoch": 3.8662295081967213,
      "grad_norm": 0.6356149863795761,
      "learning_rate": 5.816200722642329e-07,
      "loss": 11.3408,
      "step": 22110
    },
    {
      "epoch": 3.866404371584699,
      "grad_norm": 0.5961356032941991,
      "learning_rate": 5.800959887443557e-07,
      "loss": 11.2879,
      "step": 22111
    },
    {
      "epoch": 3.8665792349726775,
      "grad_norm": 0.7019999600846578,
      "learning_rate": 5.785738988988154e-07,
      "loss": 11.2483,
      "step": 22112
    },
    {
      "epoch": 3.866754098360656,
      "grad_norm": 0.7746634000240669,
      "learning_rate": 5.770538027581318e-07,
      "loss": 11.3129,
      "step": 22113
    },
    {
      "epoch": 3.8669289617486338,
      "grad_norm": 0.6445632415226534,
      "learning_rate": 5.755357003527917e-07,
      "loss": 11.4353,
      "step": 22114
    },
    {
      "epoch": 3.867103825136612,
      "grad_norm": 0.6441147372993249,
      "learning_rate": 5.740195917132263e-07,
      "loss": 11.5954,
      "step": 22115
    },
    {
      "epoch": 3.86727868852459,
      "grad_norm": 0.6160177598356703,
      "learning_rate": 5.725054768698557e-07,
      "loss": 11.3395,
      "step": 22116
    },
    {
      "epoch": 3.8674535519125683,
      "grad_norm": 0.682574421180029,
      "learning_rate": 5.709933558530223e-07,
      "loss": 11.5849,
      "step": 22117
    },
    {
      "epoch": 3.8676284153005467,
      "grad_norm": 0.7223264633673742,
      "learning_rate": 5.694832286930685e-07,
      "loss": 11.3656,
      "step": 22118
    },
    {
      "epoch": 3.8678032786885246,
      "grad_norm": 0.637587367923843,
      "learning_rate": 5.67975095420259e-07,
      "loss": 11.3843,
      "step": 22119
    },
    {
      "epoch": 3.8679781420765025,
      "grad_norm": 0.7088761202032778,
      "learning_rate": 5.664689560648473e-07,
      "loss": 11.4483,
      "step": 22120
    },
    {
      "epoch": 3.868153005464481,
      "grad_norm": 0.7245958341002438,
      "learning_rate": 5.649648106570426e-07,
      "loss": 11.3236,
      "step": 22121
    },
    {
      "epoch": 3.868327868852459,
      "grad_norm": 0.6565182158771846,
      "learning_rate": 5.634626592269876e-07,
      "loss": 11.3936,
      "step": 22122
    },
    {
      "epoch": 3.868502732240437,
      "grad_norm": 0.7039782360118021,
      "learning_rate": 5.619625018048247e-07,
      "loss": 11.5381,
      "step": 22123
    },
    {
      "epoch": 3.8686775956284154,
      "grad_norm": 0.6361238826986333,
      "learning_rate": 5.604643384206299e-07,
      "loss": 11.5106,
      "step": 22124
    },
    {
      "epoch": 3.8688524590163933,
      "grad_norm": 0.64005860745197,
      "learning_rate": 5.589681691044346e-07,
      "loss": 11.2907,
      "step": 22125
    },
    {
      "epoch": 3.8690273224043716,
      "grad_norm": 0.644110877594191,
      "learning_rate": 5.574739938862594e-07,
      "loss": 11.1808,
      "step": 22126
    },
    {
      "epoch": 3.86920218579235,
      "grad_norm": 0.6916862692406524,
      "learning_rate": 5.559818127960581e-07,
      "loss": 11.2297,
      "step": 22127
    },
    {
      "epoch": 3.869377049180328,
      "grad_norm": 0.6611354818054083,
      "learning_rate": 5.544916258637511e-07,
      "loss": 11.3135,
      "step": 22128
    },
    {
      "epoch": 3.8695519125683058,
      "grad_norm": 0.6802740813334771,
      "learning_rate": 5.530034331192257e-07,
      "loss": 11.4754,
      "step": 22129
    },
    {
      "epoch": 3.869726775956284,
      "grad_norm": 0.6944439933817356,
      "learning_rate": 5.515172345923358e-07,
      "loss": 11.3992,
      "step": 22130
    },
    {
      "epoch": 3.8699016393442625,
      "grad_norm": 0.6372226985684787,
      "learning_rate": 5.500330303128575e-07,
      "loss": 11.489,
      "step": 22131
    },
    {
      "epoch": 3.8700765027322404,
      "grad_norm": 0.673390924252368,
      "learning_rate": 5.485508203105782e-07,
      "loss": 11.456,
      "step": 22132
    },
    {
      "epoch": 3.8702513661202187,
      "grad_norm": 0.6498566163532798,
      "learning_rate": 5.470706046151963e-07,
      "loss": 11.3113,
      "step": 22133
    },
    {
      "epoch": 3.8704262295081966,
      "grad_norm": 0.6228333193163349,
      "learning_rate": 5.455923832564214e-07,
      "loss": 11.575,
      "step": 22134
    },
    {
      "epoch": 3.870601092896175,
      "grad_norm": 0.6815202663419258,
      "learning_rate": 5.441161562638852e-07,
      "loss": 11.3242,
      "step": 22135
    },
    {
      "epoch": 3.8707759562841533,
      "grad_norm": 0.7260061691330374,
      "learning_rate": 5.426419236671865e-07,
      "loss": 11.4049,
      "step": 22136
    },
    {
      "epoch": 3.870950819672131,
      "grad_norm": 0.6789960089599949,
      "learning_rate": 5.411696854958903e-07,
      "loss": 11.2164,
      "step": 22137
    },
    {
      "epoch": 3.871125683060109,
      "grad_norm": 0.6382333028566971,
      "learning_rate": 5.396994417795176e-07,
      "loss": 11.58,
      "step": 22138
    },
    {
      "epoch": 3.8713005464480874,
      "grad_norm": 0.7274225228417497,
      "learning_rate": 5.382311925475669e-07,
      "loss": 11.1823,
      "step": 22139
    },
    {
      "epoch": 3.8714754098360658,
      "grad_norm": 0.7570292449301085,
      "learning_rate": 5.367649378294592e-07,
      "loss": 11.2473,
      "step": 22140
    },
    {
      "epoch": 3.8716502732240436,
      "grad_norm": 0.6584531084516144,
      "learning_rate": 5.353006776546043e-07,
      "loss": 11.4714,
      "step": 22141
    },
    {
      "epoch": 3.871825136612022,
      "grad_norm": 0.6362511818917546,
      "learning_rate": 5.338384120523676e-07,
      "loss": 11.2906,
      "step": 22142
    },
    {
      "epoch": 3.872,
      "grad_norm": 0.6166573396830205,
      "learning_rate": 5.323781410520812e-07,
      "loss": 11.512,
      "step": 22143
    },
    {
      "epoch": 3.8721748633879782,
      "grad_norm": 0.7000672135368448,
      "learning_rate": 5.309198646830104e-07,
      "loss": 11.4595,
      "step": 22144
    },
    {
      "epoch": 3.872349726775956,
      "grad_norm": 0.6578113954530407,
      "learning_rate": 5.294635829744099e-07,
      "loss": 11.454,
      "step": 22145
    },
    {
      "epoch": 3.8725245901639345,
      "grad_norm": 0.7193727276265808,
      "learning_rate": 5.280092959554784e-07,
      "loss": 11.1916,
      "step": 22146
    },
    {
      "epoch": 3.8726994535519124,
      "grad_norm": 0.6919381816786611,
      "learning_rate": 5.265570036553813e-07,
      "loss": 11.4477,
      "step": 22147
    },
    {
      "epoch": 3.8728743169398907,
      "grad_norm": 0.7973568792885202,
      "learning_rate": 5.2510670610324e-07,
      "loss": 10.8529,
      "step": 22148
    },
    {
      "epoch": 3.873049180327869,
      "grad_norm": 0.7414532121426319,
      "learning_rate": 5.236584033281311e-07,
      "loss": 11.4378,
      "step": 22149
    },
    {
      "epoch": 3.873224043715847,
      "grad_norm": 0.7185519682552268,
      "learning_rate": 5.222120953591092e-07,
      "loss": 11.4783,
      "step": 22150
    },
    {
      "epoch": 3.8733989071038253,
      "grad_norm": 0.683323679143907,
      "learning_rate": 5.207677822251733e-07,
      "loss": 11.5213,
      "step": 22151
    },
    {
      "epoch": 3.873573770491803,
      "grad_norm": 0.5818950658001669,
      "learning_rate": 5.19325463955278e-07,
      "loss": 11.6292,
      "step": 22152
    },
    {
      "epoch": 3.8737486338797815,
      "grad_norm": 0.7920030812865172,
      "learning_rate": 5.17885140578367e-07,
      "loss": 11.2762,
      "step": 22153
    },
    {
      "epoch": 3.8739234972677594,
      "grad_norm": 0.6920717208231202,
      "learning_rate": 5.164468121232946e-07,
      "loss": 11.3388,
      "step": 22154
    },
    {
      "epoch": 3.8740983606557378,
      "grad_norm": 0.740269261115674,
      "learning_rate": 5.150104786189158e-07,
      "loss": 11.2921,
      "step": 22155
    },
    {
      "epoch": 3.8742732240437157,
      "grad_norm": 0.7408425113085484,
      "learning_rate": 5.135761400940409e-07,
      "loss": 11.3282,
      "step": 22156
    },
    {
      "epoch": 3.874448087431694,
      "grad_norm": 0.6676373162153311,
      "learning_rate": 5.121437965774356e-07,
      "loss": 11.4386,
      "step": 22157
    },
    {
      "epoch": 3.8746229508196723,
      "grad_norm": 0.6831972756656869,
      "learning_rate": 5.107134480978104e-07,
      "loss": 11.2349,
      "step": 22158
    },
    {
      "epoch": 3.8747978142076502,
      "grad_norm": 0.6788735081808464,
      "learning_rate": 5.092850946838534e-07,
      "loss": 11.2214,
      "step": 22159
    },
    {
      "epoch": 3.8749726775956286,
      "grad_norm": 0.6872622490602631,
      "learning_rate": 5.078587363641973e-07,
      "loss": 11.5314,
      "step": 22160
    },
    {
      "epoch": 3.8751475409836065,
      "grad_norm": 0.7024905271752905,
      "learning_rate": 5.064343731674637e-07,
      "loss": 11.2567,
      "step": 22161
    },
    {
      "epoch": 3.875322404371585,
      "grad_norm": 0.6796481674196792,
      "learning_rate": 5.050120051222073e-07,
      "loss": 11.4714,
      "step": 22162
    },
    {
      "epoch": 3.8754972677595627,
      "grad_norm": 0.681978369367002,
      "learning_rate": 5.035916322569501e-07,
      "loss": 11.4286,
      "step": 22163
    },
    {
      "epoch": 3.875672131147541,
      "grad_norm": 0.6242436714051884,
      "learning_rate": 5.021732546001689e-07,
      "loss": 11.4963,
      "step": 22164
    },
    {
      "epoch": 3.875846994535519,
      "grad_norm": 0.6349786242036068,
      "learning_rate": 5.007568721803191e-07,
      "loss": 11.5209,
      "step": 22165
    },
    {
      "epoch": 3.8760218579234973,
      "grad_norm": 0.6678420369315741,
      "learning_rate": 4.993424850257889e-07,
      "loss": 11.4054,
      "step": 22166
    },
    {
      "epoch": 3.8761967213114756,
      "grad_norm": 0.6697176894957374,
      "learning_rate": 4.979300931649556e-07,
      "loss": 11.3957,
      "step": 22167
    },
    {
      "epoch": 3.8763715846994535,
      "grad_norm": 0.673565510177419,
      "learning_rate": 4.96519696626141e-07,
      "loss": 11.4412,
      "step": 22168
    },
    {
      "epoch": 3.8765464480874314,
      "grad_norm": 0.6682713645269934,
      "learning_rate": 4.951112954376002e-07,
      "loss": 11.4358,
      "step": 22169
    },
    {
      "epoch": 3.8767213114754098,
      "grad_norm": 0.702165816915655,
      "learning_rate": 4.93704889627622e-07,
      "loss": 11.3009,
      "step": 22170
    },
    {
      "epoch": 3.876896174863388,
      "grad_norm": 0.6644601409101248,
      "learning_rate": 4.923004792243724e-07,
      "loss": 11.3664,
      "step": 22171
    },
    {
      "epoch": 3.877071038251366,
      "grad_norm": 0.6415008001565574,
      "learning_rate": 4.90898064256029e-07,
      "loss": 11.424,
      "step": 22172
    },
    {
      "epoch": 3.8772459016393443,
      "grad_norm": 0.626450202385798,
      "learning_rate": 4.894976447507138e-07,
      "loss": 11.553,
      "step": 22173
    },
    {
      "epoch": 3.8774207650273222,
      "grad_norm": 0.684076032937199,
      "learning_rate": 4.880992207365043e-07,
      "loss": 11.4682,
      "step": 22174
    },
    {
      "epoch": 3.8775956284153006,
      "grad_norm": 0.6359313574095206,
      "learning_rate": 4.867027922414557e-07,
      "loss": 11.5976,
      "step": 22175
    },
    {
      "epoch": 3.877770491803279,
      "grad_norm": 0.6192109281040455,
      "learning_rate": 4.85308359293557e-07,
      "loss": 11.4073,
      "step": 22176
    },
    {
      "epoch": 3.877945355191257,
      "grad_norm": 0.682400720572294,
      "learning_rate": 4.839159219207746e-07,
      "loss": 11.6021,
      "step": 22177
    },
    {
      "epoch": 3.8781202185792347,
      "grad_norm": 0.7860309141032353,
      "learning_rate": 4.825254801510415e-07,
      "loss": 11.3102,
      "step": 22178
    },
    {
      "epoch": 3.878295081967213,
      "grad_norm": 0.8579200142139435,
      "learning_rate": 4.811370340122134e-07,
      "loss": 11.2174,
      "step": 22179
    },
    {
      "epoch": 3.8784699453551914,
      "grad_norm": 0.7149349280393278,
      "learning_rate": 4.797505835321681e-07,
      "loss": 11.4044,
      "step": 22180
    },
    {
      "epoch": 3.8786448087431693,
      "grad_norm": 0.7641000384238547,
      "learning_rate": 4.783661287386831e-07,
      "loss": 11.2021,
      "step": 22181
    },
    {
      "epoch": 3.8788196721311476,
      "grad_norm": 0.6781172358930921,
      "learning_rate": 4.769836696595365e-07,
      "loss": 11.6837,
      "step": 22182
    },
    {
      "epoch": 3.8789945355191255,
      "grad_norm": 0.6930647994007944,
      "learning_rate": 4.7560320632242805e-07,
      "loss": 11.3837,
      "step": 22183
    },
    {
      "epoch": 3.879169398907104,
      "grad_norm": 0.6409862907227999,
      "learning_rate": 4.7422473875505804e-07,
      "loss": 11.3032,
      "step": 22184
    },
    {
      "epoch": 3.8793442622950822,
      "grad_norm": 0.6849543321388921,
      "learning_rate": 4.72848266985082e-07,
      "loss": 11.4639,
      "step": 22185
    },
    {
      "epoch": 3.87951912568306,
      "grad_norm": 0.6615538297656143,
      "learning_rate": 4.7147379104007796e-07,
      "loss": 11.5185,
      "step": 22186
    },
    {
      "epoch": 3.879693989071038,
      "grad_norm": 0.6881758359191796,
      "learning_rate": 4.7010131094761265e-07,
      "loss": 11.4067,
      "step": 22187
    },
    {
      "epoch": 3.8798688524590164,
      "grad_norm": 0.6975402879149911,
      "learning_rate": 4.6873082673521975e-07,
      "loss": 11.4489,
      "step": 22188
    },
    {
      "epoch": 3.8800437158469947,
      "grad_norm": 0.6762839438010615,
      "learning_rate": 4.6736233843036604e-07,
      "loss": 11.5871,
      "step": 22189
    },
    {
      "epoch": 3.8802185792349726,
      "grad_norm": 0.7056828895394366,
      "learning_rate": 4.659958460605074e-07,
      "loss": 11.0985,
      "step": 22190
    },
    {
      "epoch": 3.880393442622951,
      "grad_norm": 0.6845553045708537,
      "learning_rate": 4.646313496530441e-07,
      "loss": 11.3348,
      "step": 22191
    },
    {
      "epoch": 3.880568306010929,
      "grad_norm": 0.6780832846395589,
      "learning_rate": 4.632688492353321e-07,
      "loss": 11.3722,
      "step": 22192
    },
    {
      "epoch": 3.880743169398907,
      "grad_norm": 0.6488017571841375,
      "learning_rate": 4.619083448346939e-07,
      "loss": 11.1996,
      "step": 22193
    },
    {
      "epoch": 3.8809180327868855,
      "grad_norm": 0.7142184065558188,
      "learning_rate": 4.6054983647842996e-07,
      "loss": 11.6591,
      "step": 22194
    },
    {
      "epoch": 3.8810928961748634,
      "grad_norm": 0.6756458386619534,
      "learning_rate": 4.5919332419376293e-07,
      "loss": 11.4599,
      "step": 22195
    },
    {
      "epoch": 3.8812677595628413,
      "grad_norm": 0.6400909080338353,
      "learning_rate": 4.5783880800789323e-07,
      "loss": 11.3035,
      "step": 22196
    },
    {
      "epoch": 3.8814426229508197,
      "grad_norm": 0.6838164125976941,
      "learning_rate": 4.564862879479881e-07,
      "loss": 11.4045,
      "step": 22197
    },
    {
      "epoch": 3.881617486338798,
      "grad_norm": 0.6165412463488587,
      "learning_rate": 4.551357640411813e-07,
      "loss": 11.2348,
      "step": 22198
    },
    {
      "epoch": 3.881792349726776,
      "grad_norm": 0.6485767380388875,
      "learning_rate": 4.537872363145401e-07,
      "loss": 11.2802,
      "step": 22199
    },
    {
      "epoch": 3.8819672131147542,
      "grad_norm": 0.7044447816779422,
      "learning_rate": 4.5244070479510957e-07,
      "loss": 11.3283,
      "step": 22200
    },
    {
      "epoch": 3.882142076502732,
      "grad_norm": 0.613727463827586,
      "learning_rate": 4.5109616950990143e-07,
      "loss": 11.5158,
      "step": 22201
    },
    {
      "epoch": 3.8823169398907105,
      "grad_norm": 0.6553066160987435,
      "learning_rate": 4.4975363048586075e-07,
      "loss": 11.6103,
      "step": 22202
    },
    {
      "epoch": 3.8824918032786884,
      "grad_norm": 0.7220592420422508,
      "learning_rate": 4.4841308774993264e-07,
      "loss": 11.5923,
      "step": 22203
    },
    {
      "epoch": 3.8826666666666667,
      "grad_norm": 0.7105568698192866,
      "learning_rate": 4.470745413289734e-07,
      "loss": 11.4671,
      "step": 22204
    },
    {
      "epoch": 3.8828415300546446,
      "grad_norm": 0.6483022081553683,
      "learning_rate": 4.457379912498394e-07,
      "loss": 11.3901,
      "step": 22205
    },
    {
      "epoch": 3.883016393442623,
      "grad_norm": 0.6526601849369131,
      "learning_rate": 4.4440343753933135e-07,
      "loss": 11.4697,
      "step": 22206
    },
    {
      "epoch": 3.8831912568306013,
      "grad_norm": 0.667762545271228,
      "learning_rate": 4.4307088022421674e-07,
      "loss": 11.3561,
      "step": 22207
    },
    {
      "epoch": 3.883366120218579,
      "grad_norm": 0.6263482499165302,
      "learning_rate": 4.417403193311964e-07,
      "loss": 11.3538,
      "step": 22208
    },
    {
      "epoch": 3.8835409836065575,
      "grad_norm": 0.6793110849887293,
      "learning_rate": 4.404117548869824e-07,
      "loss": 11.3367,
      "step": 22209
    },
    {
      "epoch": 3.8837158469945354,
      "grad_norm": 0.6185973796560804,
      "learning_rate": 4.390851869181867e-07,
      "loss": 11.1618,
      "step": 22210
    },
    {
      "epoch": 3.8838907103825138,
      "grad_norm": 0.734274158394552,
      "learning_rate": 4.3776061545143243e-07,
      "loss": 11.4565,
      "step": 22211
    },
    {
      "epoch": 3.8840655737704917,
      "grad_norm": 0.6321435060666544,
      "learning_rate": 4.364380405132762e-07,
      "loss": 11.1803,
      "step": 22212
    },
    {
      "epoch": 3.88424043715847,
      "grad_norm": 0.7769043925160022,
      "learning_rate": 4.3511746213024116e-07,
      "loss": 11.2623,
      "step": 22213
    },
    {
      "epoch": 3.884415300546448,
      "grad_norm": 0.6614370448699066,
      "learning_rate": 4.337988803287951e-07,
      "loss": 11.3072,
      "step": 22214
    },
    {
      "epoch": 3.8845901639344262,
      "grad_norm": 0.6427040977572442,
      "learning_rate": 4.324822951353946e-07,
      "loss": 11.481,
      "step": 22215
    },
    {
      "epoch": 3.8847650273224046,
      "grad_norm": 0.651789161095092,
      "learning_rate": 4.311677065764408e-07,
      "loss": 11.534,
      "step": 22216
    },
    {
      "epoch": 3.8849398907103825,
      "grad_norm": 0.7653578338271074,
      "learning_rate": 4.2985511467827923e-07,
      "loss": 11.2668,
      "step": 22217
    },
    {
      "epoch": 3.8851147540983604,
      "grad_norm": 0.639042612748267,
      "learning_rate": 4.285445194672555e-07,
      "loss": 11.5768,
      "step": 22218
    },
    {
      "epoch": 3.8852896174863387,
      "grad_norm": 0.6256835158776975,
      "learning_rate": 4.272359209696264e-07,
      "loss": 11.51,
      "step": 22219
    },
    {
      "epoch": 3.885464480874317,
      "grad_norm": 0.634643660988628,
      "learning_rate": 4.259293192116598e-07,
      "loss": 11.434,
      "step": 22220
    },
    {
      "epoch": 3.885639344262295,
      "grad_norm": 0.6829518058532971,
      "learning_rate": 4.2462471421953474e-07,
      "loss": 11.3093,
      "step": 22221
    },
    {
      "epoch": 3.8858142076502733,
      "grad_norm": 0.6270312409794895,
      "learning_rate": 4.2332210601941926e-07,
      "loss": 11.5303,
      "step": 22222
    },
    {
      "epoch": 3.885989071038251,
      "grad_norm": 0.7404245708397776,
      "learning_rate": 4.220214946374368e-07,
      "loss": 11.0614,
      "step": 22223
    },
    {
      "epoch": 3.8861639344262295,
      "grad_norm": 0.6960740176711419,
      "learning_rate": 4.2072288009966654e-07,
      "loss": 11.5394,
      "step": 22224
    },
    {
      "epoch": 3.886338797814208,
      "grad_norm": 0.6526451954974383,
      "learning_rate": 4.1942626243215434e-07,
      "loss": 11.2785,
      "step": 22225
    },
    {
      "epoch": 3.8865136612021858,
      "grad_norm": 0.779592335530782,
      "learning_rate": 4.1813164166090156e-07,
      "loss": 11.5715,
      "step": 22226
    },
    {
      "epoch": 3.8866885245901637,
      "grad_norm": 0.7031529696733404,
      "learning_rate": 4.168390178118653e-07,
      "loss": 11.4917,
      "step": 22227
    },
    {
      "epoch": 3.886863387978142,
      "grad_norm": 0.6670761477908899,
      "learning_rate": 4.155483909109581e-07,
      "loss": 11.5961,
      "step": 22228
    },
    {
      "epoch": 3.8870382513661204,
      "grad_norm": 0.7143008011573753,
      "learning_rate": 4.142597609840815e-07,
      "loss": 11.4541,
      "step": 22229
    },
    {
      "epoch": 3.8872131147540983,
      "grad_norm": 0.6351083599886632,
      "learning_rate": 4.1297312805705925e-07,
      "loss": 11.5953,
      "step": 22230
    },
    {
      "epoch": 3.8873879781420766,
      "grad_norm": 0.74567204428237,
      "learning_rate": 4.11688492155704e-07,
      "loss": 11.4231,
      "step": 22231
    },
    {
      "epoch": 3.8875628415300545,
      "grad_norm": 0.6882362444158759,
      "learning_rate": 4.10405853305762e-07,
      "loss": 11.0521,
      "step": 22232
    },
    {
      "epoch": 3.887737704918033,
      "grad_norm": 0.7127030474368812,
      "learning_rate": 4.091252115329569e-07,
      "loss": 11.1924,
      "step": 22233
    },
    {
      "epoch": 3.887912568306011,
      "grad_norm": 0.6994247437635791,
      "learning_rate": 4.078465668629905e-07,
      "loss": 11.2649,
      "step": 22234
    },
    {
      "epoch": 3.888087431693989,
      "grad_norm": 0.6709160996201925,
      "learning_rate": 4.065699193214867e-07,
      "loss": 11.2175,
      "step": 22235
    },
    {
      "epoch": 3.888262295081967,
      "grad_norm": 0.7692000515727233,
      "learning_rate": 4.052952689340472e-07,
      "loss": 11.2266,
      "step": 22236
    },
    {
      "epoch": 3.8884371584699453,
      "grad_norm": 0.6605820567717074,
      "learning_rate": 4.0402261572621836e-07,
      "loss": 11.4638,
      "step": 22237
    },
    {
      "epoch": 3.8886120218579237,
      "grad_norm": 0.6855377331354158,
      "learning_rate": 4.027519597235463e-07,
      "loss": 11.4508,
      "step": 22238
    },
    {
      "epoch": 3.8887868852459015,
      "grad_norm": 0.6879647631812212,
      "learning_rate": 4.014833009515107e-07,
      "loss": 11.1649,
      "step": 22239
    },
    {
      "epoch": 3.88896174863388,
      "grad_norm": 0.6574166976994603,
      "learning_rate": 4.0021663943553556e-07,
      "loss": 11.3098,
      "step": 22240
    },
    {
      "epoch": 3.889136612021858,
      "grad_norm": 0.6302260840911835,
      "learning_rate": 3.989519752010229e-07,
      "loss": 11.5324,
      "step": 22241
    },
    {
      "epoch": 3.889311475409836,
      "grad_norm": 0.7099820857608753,
      "learning_rate": 3.976893082733413e-07,
      "loss": 11.2894,
      "step": 22242
    },
    {
      "epoch": 3.8894863387978145,
      "grad_norm": 0.6833256013055157,
      "learning_rate": 3.9642863867781487e-07,
      "loss": 11.3316,
      "step": 22243
    },
    {
      "epoch": 3.8896612021857924,
      "grad_norm": 0.6168767595945467,
      "learning_rate": 3.9516996643971237e-07,
      "loss": 11.2518,
      "step": 22244
    },
    {
      "epoch": 3.8898360655737703,
      "grad_norm": 0.6456778659343408,
      "learning_rate": 3.9391329158428024e-07,
      "loss": 11.4893,
      "step": 22245
    },
    {
      "epoch": 3.8900109289617486,
      "grad_norm": 0.690095415768641,
      "learning_rate": 3.9265861413670945e-07,
      "loss": 11.6516,
      "step": 22246
    },
    {
      "epoch": 3.890185792349727,
      "grad_norm": 0.6122321953210355,
      "learning_rate": 3.9140593412217985e-07,
      "loss": 11.1954,
      "step": 22247
    },
    {
      "epoch": 3.890360655737705,
      "grad_norm": 0.6330631604460268,
      "learning_rate": 3.901552515657936e-07,
      "loss": 11.5366,
      "step": 22248
    },
    {
      "epoch": 3.890535519125683,
      "grad_norm": 0.6888265611479838,
      "learning_rate": 3.8890656649264166e-07,
      "loss": 11.4433,
      "step": 22249
    },
    {
      "epoch": 3.890710382513661,
      "grad_norm": 0.6639170298706945,
      "learning_rate": 3.876598789277597e-07,
      "loss": 11.4717,
      "step": 22250
    },
    {
      "epoch": 3.8908852459016394,
      "grad_norm": 0.7656462150291412,
      "learning_rate": 3.864151888961387e-07,
      "loss": 11.4189,
      "step": 22251
    },
    {
      "epoch": 3.8910601092896178,
      "grad_norm": 0.708968228549619,
      "learning_rate": 3.851724964227477e-07,
      "loss": 11.1588,
      "step": 22252
    },
    {
      "epoch": 3.8912349726775957,
      "grad_norm": 0.695886114477738,
      "learning_rate": 3.839318015325111e-07,
      "loss": 11.3311,
      "step": 22253
    },
    {
      "epoch": 3.8914098360655736,
      "grad_norm": 0.7191283418314706,
      "learning_rate": 3.82693104250309e-07,
      "loss": 11.4736,
      "step": 22254
    },
    {
      "epoch": 3.891584699453552,
      "grad_norm": 0.793561379270248,
      "learning_rate": 3.814564046009661e-07,
      "loss": 11.3845,
      "step": 22255
    },
    {
      "epoch": 3.8917595628415302,
      "grad_norm": 0.6971266924894866,
      "learning_rate": 3.802217026092847e-07,
      "loss": 11.3857,
      "step": 22256
    },
    {
      "epoch": 3.891934426229508,
      "grad_norm": 0.6930925122655199,
      "learning_rate": 3.7898899830004496e-07,
      "loss": 11.3228,
      "step": 22257
    },
    {
      "epoch": 3.8921092896174865,
      "grad_norm": 0.6709150161102535,
      "learning_rate": 3.777582916979494e-07,
      "loss": 11.6851,
      "step": 22258
    },
    {
      "epoch": 3.8922841530054644,
      "grad_norm": 0.6526906419410352,
      "learning_rate": 3.7652958282767826e-07,
      "loss": 11.375,
      "step": 22259
    },
    {
      "epoch": 3.8924590163934427,
      "grad_norm": 0.673137994855007,
      "learning_rate": 3.7530287171387843e-07,
      "loss": 11.513,
      "step": 22260
    },
    {
      "epoch": 3.8926338797814206,
      "grad_norm": 0.620091412011204,
      "learning_rate": 3.740781583811304e-07,
      "loss": 11.3435,
      "step": 22261
    },
    {
      "epoch": 3.892808743169399,
      "grad_norm": 0.6499414749995565,
      "learning_rate": 3.728554428540143e-07,
      "loss": 11.0457,
      "step": 22262
    },
    {
      "epoch": 3.892983606557377,
      "grad_norm": 0.6781166674973128,
      "learning_rate": 3.716347251570551e-07,
      "loss": 11.2073,
      "step": 22263
    },
    {
      "epoch": 3.893158469945355,
      "grad_norm": 0.7235370847382394,
      "learning_rate": 3.70416005314711e-07,
      "loss": 11.1549,
      "step": 22264
    },
    {
      "epoch": 3.8933333333333335,
      "grad_norm": 0.6828149987854916,
      "learning_rate": 3.6919928335141797e-07,
      "loss": 11.5489,
      "step": 22265
    },
    {
      "epoch": 3.8935081967213114,
      "grad_norm": 0.6550657124433807,
      "learning_rate": 3.6798455929158983e-07,
      "loss": 11.2109,
      "step": 22266
    },
    {
      "epoch": 3.8936830601092898,
      "grad_norm": 0.6460238871639367,
      "learning_rate": 3.6677183315959596e-07,
      "loss": 11.3243,
      "step": 22267
    },
    {
      "epoch": 3.8938579234972677,
      "grad_norm": 0.7158257163780568,
      "learning_rate": 3.6556110497972805e-07,
      "loss": 11.572,
      "step": 22268
    },
    {
      "epoch": 3.894032786885246,
      "grad_norm": 0.631211629727949,
      "learning_rate": 3.6435237477627784e-07,
      "loss": 11.5504,
      "step": 22269
    },
    {
      "epoch": 3.894207650273224,
      "grad_norm": 0.7273097016447747,
      "learning_rate": 3.6314564257348137e-07,
      "loss": 11.6761,
      "step": 22270
    },
    {
      "epoch": 3.8943825136612022,
      "grad_norm": 0.7164565929048706,
      "learning_rate": 3.619409083955527e-07,
      "loss": 11.4297,
      "step": 22271
    },
    {
      "epoch": 3.89455737704918,
      "grad_norm": 0.7032755289064014,
      "learning_rate": 3.607381722666281e-07,
      "loss": 11.3493,
      "step": 22272
    },
    {
      "epoch": 3.8947322404371585,
      "grad_norm": 0.7006042567910626,
      "learning_rate": 3.595374342108437e-07,
      "loss": 11.2294,
      "step": 22273
    },
    {
      "epoch": 3.894907103825137,
      "grad_norm": 0.6654073404027098,
      "learning_rate": 3.583386942522582e-07,
      "loss": 11.1131,
      "step": 22274
    },
    {
      "epoch": 3.8950819672131147,
      "grad_norm": 0.7014260478687478,
      "learning_rate": 3.5714195241492996e-07,
      "loss": 11.4043,
      "step": 22275
    },
    {
      "epoch": 3.8952568306010926,
      "grad_norm": 0.6851849715893366,
      "learning_rate": 3.55947208722851e-07,
      "loss": 11.2617,
      "step": 22276
    },
    {
      "epoch": 3.895431693989071,
      "grad_norm": 0.6016775005667121,
      "learning_rate": 3.5475446319997994e-07,
      "loss": 11.4561,
      "step": 22277
    },
    {
      "epoch": 3.8956065573770493,
      "grad_norm": 0.627653837438943,
      "learning_rate": 3.5356371587021987e-07,
      "loss": 11.2642,
      "step": 22278
    },
    {
      "epoch": 3.895781420765027,
      "grad_norm": 0.6902659704051252,
      "learning_rate": 3.523749667574738e-07,
      "loss": 11.2444,
      "step": 22279
    },
    {
      "epoch": 3.8959562841530055,
      "grad_norm": 0.7089236626101324,
      "learning_rate": 3.5118821588555616e-07,
      "loss": 11.1902,
      "step": 22280
    },
    {
      "epoch": 3.8961311475409834,
      "grad_norm": 0.6468906959819994,
      "learning_rate": 3.5000346327829225e-07,
      "loss": 11.3759,
      "step": 22281
    },
    {
      "epoch": 3.896306010928962,
      "grad_norm": 0.6965669018754335,
      "learning_rate": 3.4882070895941865e-07,
      "loss": 11.4174,
      "step": 22282
    },
    {
      "epoch": 3.89648087431694,
      "grad_norm": 0.6908062420554586,
      "learning_rate": 3.476399529526608e-07,
      "loss": 11.3105,
      "step": 22283
    },
    {
      "epoch": 3.896655737704918,
      "grad_norm": 0.7601307973358488,
      "learning_rate": 3.4646119528168873e-07,
      "loss": 11.4203,
      "step": 22284
    },
    {
      "epoch": 3.896830601092896,
      "grad_norm": 0.7122635519646328,
      "learning_rate": 3.4528443597015017e-07,
      "loss": 11.2589,
      "step": 22285
    },
    {
      "epoch": 3.8970054644808743,
      "grad_norm": 0.672808985110941,
      "learning_rate": 3.441096750416373e-07,
      "loss": 11.336,
      "step": 22286
    },
    {
      "epoch": 3.8971803278688526,
      "grad_norm": 0.6869245292401678,
      "learning_rate": 3.429369125197091e-07,
      "loss": 11.3089,
      "step": 22287
    },
    {
      "epoch": 3.8973551912568305,
      "grad_norm": 0.7415976547509923,
      "learning_rate": 3.417661484278911e-07,
      "loss": 11.0868,
      "step": 22288
    },
    {
      "epoch": 3.897530054644809,
      "grad_norm": 0.6941696770668374,
      "learning_rate": 3.405973827896425e-07,
      "loss": 11.3636,
      "step": 22289
    },
    {
      "epoch": 3.8977049180327867,
      "grad_norm": 0.750142584898772,
      "learning_rate": 3.3943061562841107e-07,
      "loss": 11.3958,
      "step": 22290
    },
    {
      "epoch": 3.897879781420765,
      "grad_norm": 0.6950133861304649,
      "learning_rate": 3.3826584696760035e-07,
      "loss": 11.488,
      "step": 22291
    },
    {
      "epoch": 3.8980546448087434,
      "grad_norm": 0.6994283029121396,
      "learning_rate": 3.371030768305583e-07,
      "loss": 11.3473,
      "step": 22292
    },
    {
      "epoch": 3.8982295081967213,
      "grad_norm": 0.6672528269493524,
      "learning_rate": 3.359423052405997e-07,
      "loss": 11.1867,
      "step": 22293
    },
    {
      "epoch": 3.898404371584699,
      "grad_norm": 0.6930388849256658,
      "learning_rate": 3.347835322210169e-07,
      "loss": 11.1378,
      "step": 22294
    },
    {
      "epoch": 3.8985792349726776,
      "grad_norm": 0.7679257370287702,
      "learning_rate": 3.3362675779502475e-07,
      "loss": 11.1742,
      "step": 22295
    },
    {
      "epoch": 3.898754098360656,
      "grad_norm": 0.7918448774672174,
      "learning_rate": 3.3247198198583793e-07,
      "loss": 11.3772,
      "step": 22296
    },
    {
      "epoch": 3.898928961748634,
      "grad_norm": 0.664905685055803,
      "learning_rate": 3.3131920481661584e-07,
      "loss": 11.5289,
      "step": 22297
    },
    {
      "epoch": 3.899103825136612,
      "grad_norm": 0.6535279607413051,
      "learning_rate": 3.30168426310451e-07,
      "loss": 11.566,
      "step": 22298
    },
    {
      "epoch": 3.89927868852459,
      "grad_norm": 0.6610333616969004,
      "learning_rate": 3.2901964649044717e-07,
      "loss": 11.4414,
      "step": 22299
    },
    {
      "epoch": 3.8994535519125684,
      "grad_norm": 0.6460011102364671,
      "learning_rate": 3.278728653796304e-07,
      "loss": 11.27,
      "step": 22300
    },
    {
      "epoch": 3.8996284153005467,
      "grad_norm": 0.6569171000430771,
      "learning_rate": 3.2672808300099335e-07,
      "loss": 11.5827,
      "step": 22301
    },
    {
      "epoch": 3.8998032786885246,
      "grad_norm": 0.6487977500402962,
      "learning_rate": 3.255852993774844e-07,
      "loss": 11.5594,
      "step": 22302
    },
    {
      "epoch": 3.8999781420765025,
      "grad_norm": 0.6700202460211585,
      "learning_rate": 3.2444451453204075e-07,
      "loss": 11.4531,
      "step": 22303
    },
    {
      "epoch": 3.900153005464481,
      "grad_norm": 0.7073693518182756,
      "learning_rate": 3.233057284875329e-07,
      "loss": 11.3043,
      "step": 22304
    },
    {
      "epoch": 3.900327868852459,
      "grad_norm": 0.7095715037515684,
      "learning_rate": 3.221689412667872e-07,
      "loss": 11.2905,
      "step": 22305
    },
    {
      "epoch": 3.900502732240437,
      "grad_norm": 0.6777905404629665,
      "learning_rate": 3.2103415289260754e-07,
      "loss": 11.3491,
      "step": 22306
    },
    {
      "epoch": 3.9006775956284154,
      "grad_norm": 0.665934767542708,
      "learning_rate": 3.199013633877423e-07,
      "loss": 11.652,
      "step": 22307
    },
    {
      "epoch": 3.9008524590163933,
      "grad_norm": 0.6828107596744761,
      "learning_rate": 3.1877057277491794e-07,
      "loss": 11.511,
      "step": 22308
    },
    {
      "epoch": 3.9010273224043717,
      "grad_norm": 0.7043973943878218,
      "learning_rate": 3.176417810768051e-07,
      "loss": 11.5213,
      "step": 22309
    },
    {
      "epoch": 3.9012021857923496,
      "grad_norm": 0.6351810172146772,
      "learning_rate": 3.165149883160412e-07,
      "loss": 11.3296,
      "step": 22310
    },
    {
      "epoch": 3.901377049180328,
      "grad_norm": 0.6834511658962493,
      "learning_rate": 3.153901945152193e-07,
      "loss": 11.4623,
      "step": 22311
    },
    {
      "epoch": 3.901551912568306,
      "grad_norm": 0.7254583229249035,
      "learning_rate": 3.142673996968992e-07,
      "loss": 11.3255,
      "step": 22312
    },
    {
      "epoch": 3.901726775956284,
      "grad_norm": 0.6041345388581472,
      "learning_rate": 3.131466038835851e-07,
      "loss": 11.5374,
      "step": 22313
    },
    {
      "epoch": 3.9019016393442625,
      "grad_norm": 0.6891890192183181,
      "learning_rate": 3.1202780709775894e-07,
      "loss": 11.2121,
      "step": 22314
    },
    {
      "epoch": 3.9020765027322404,
      "grad_norm": 0.7151794567903631,
      "learning_rate": 3.1091100936186945e-07,
      "loss": 11.3645,
      "step": 22315
    },
    {
      "epoch": 3.9022513661202187,
      "grad_norm": 0.6763553333700305,
      "learning_rate": 3.0979621069828766e-07,
      "loss": 11.5803,
      "step": 22316
    },
    {
      "epoch": 3.9024262295081966,
      "grad_norm": 0.6533791383866989,
      "learning_rate": 3.0868341112938457e-07,
      "loss": 11.5891,
      "step": 22317
    },
    {
      "epoch": 3.902601092896175,
      "grad_norm": 0.7822703700344612,
      "learning_rate": 3.075726106774757e-07,
      "loss": 11.3859,
      "step": 22318
    },
    {
      "epoch": 3.902775956284153,
      "grad_norm": 0.7975241043235978,
      "learning_rate": 3.0646380936483197e-07,
      "loss": 11.4371,
      "step": 22319
    },
    {
      "epoch": 3.902950819672131,
      "grad_norm": 0.6397721459019495,
      "learning_rate": 3.053570072136913e-07,
      "loss": 11.1804,
      "step": 22320
    },
    {
      "epoch": 3.903125683060109,
      "grad_norm": 0.7904782770463713,
      "learning_rate": 3.042522042462359e-07,
      "loss": 11.4534,
      "step": 22321
    },
    {
      "epoch": 3.9033005464480874,
      "grad_norm": 0.6746591747107116,
      "learning_rate": 3.031494004846369e-07,
      "loss": 11.3963,
      "step": 22322
    },
    {
      "epoch": 3.903475409836066,
      "grad_norm": 0.6728297337704702,
      "learning_rate": 3.020485959509989e-07,
      "loss": 11.5352,
      "step": 22323
    },
    {
      "epoch": 3.9036502732240437,
      "grad_norm": 0.6738146400382076,
      "learning_rate": 3.00949790667393e-07,
      "loss": 11.0943,
      "step": 22324
    },
    {
      "epoch": 3.903825136612022,
      "grad_norm": 0.6503642916317639,
      "learning_rate": 2.998529846558684e-07,
      "loss": 11.2993,
      "step": 22325
    },
    {
      "epoch": 3.904,
      "grad_norm": 0.6524613058829911,
      "learning_rate": 2.9875817793840743e-07,
      "loss": 11.4094,
      "step": 22326
    },
    {
      "epoch": 3.9041748633879783,
      "grad_norm": 0.7170848161953844,
      "learning_rate": 2.976653705369814e-07,
      "loss": 10.7353,
      "step": 22327
    },
    {
      "epoch": 3.904349726775956,
      "grad_norm": 0.6626878554171323,
      "learning_rate": 2.9657456247347283e-07,
      "loss": 11.4505,
      "step": 22328
    },
    {
      "epoch": 3.9045245901639345,
      "grad_norm": 0.7862165681025868,
      "learning_rate": 2.954857537697753e-07,
      "loss": 11.0442,
      "step": 22329
    },
    {
      "epoch": 3.9046994535519124,
      "grad_norm": 0.6761763231686767,
      "learning_rate": 2.9439894444772687e-07,
      "loss": 11.4034,
      "step": 22330
    },
    {
      "epoch": 3.9048743169398907,
      "grad_norm": 0.7388375524526043,
      "learning_rate": 2.9331413452912124e-07,
      "loss": 11.3469,
      "step": 22331
    },
    {
      "epoch": 3.905049180327869,
      "grad_norm": 0.6354549449393662,
      "learning_rate": 2.9223132403570773e-07,
      "loss": 11.5886,
      "step": 22332
    },
    {
      "epoch": 3.905224043715847,
      "grad_norm": 0.7154761436376116,
      "learning_rate": 2.9115051298919114e-07,
      "loss": 11.4184,
      "step": 22333
    },
    {
      "epoch": 3.905398907103825,
      "grad_norm": 0.6725580552394036,
      "learning_rate": 2.9007170141125417e-07,
      "loss": 11.1962,
      "step": 22334
    },
    {
      "epoch": 3.905573770491803,
      "grad_norm": 0.667776490266442,
      "learning_rate": 2.889948893235239e-07,
      "loss": 11.3872,
      "step": 22335
    },
    {
      "epoch": 3.9057486338797816,
      "grad_norm": 0.6965234492446444,
      "learning_rate": 2.8792007674761644e-07,
      "loss": 11.3865,
      "step": 22336
    },
    {
      "epoch": 3.9059234972677594,
      "grad_norm": 0.7132801546501903,
      "learning_rate": 2.8684726370504786e-07,
      "loss": 11.4123,
      "step": 22337
    },
    {
      "epoch": 3.906098360655738,
      "grad_norm": 0.6794839568823411,
      "learning_rate": 2.8577645021736765e-07,
      "loss": 11.3247,
      "step": 22338
    },
    {
      "epoch": 3.9062732240437157,
      "grad_norm": 0.7060795757396537,
      "learning_rate": 2.8470763630602524e-07,
      "loss": 11.449,
      "step": 22339
    },
    {
      "epoch": 3.906448087431694,
      "grad_norm": 0.6355299271285475,
      "learning_rate": 2.836408219924591e-07,
      "loss": 11.5797,
      "step": 22340
    },
    {
      "epoch": 3.9066229508196724,
      "grad_norm": 0.6779004211368562,
      "learning_rate": 2.825760072980632e-07,
      "loss": 11.4827,
      "step": 22341
    },
    {
      "epoch": 3.9067978142076503,
      "grad_norm": 0.6743547255312803,
      "learning_rate": 2.8151319224419824e-07,
      "loss": 11.4927,
      "step": 22342
    },
    {
      "epoch": 3.906972677595628,
      "grad_norm": 0.6713055091767669,
      "learning_rate": 2.804523768521583e-07,
      "loss": 11.3982,
      "step": 22343
    },
    {
      "epoch": 3.9071475409836065,
      "grad_norm": 0.6581745516787396,
      "learning_rate": 2.7939356114323747e-07,
      "loss": 11.4136,
      "step": 22344
    },
    {
      "epoch": 3.907322404371585,
      "grad_norm": 0.7335367743981951,
      "learning_rate": 2.7833674513865203e-07,
      "loss": 11.3681,
      "step": 22345
    },
    {
      "epoch": 3.9074972677595627,
      "grad_norm": 0.6811503694159978,
      "learning_rate": 2.772819288596074e-07,
      "loss": 11.4775,
      "step": 22346
    },
    {
      "epoch": 3.907672131147541,
      "grad_norm": 0.682279022122587,
      "learning_rate": 2.76229112327242e-07,
      "loss": 11.282,
      "step": 22347
    },
    {
      "epoch": 3.907846994535519,
      "grad_norm": 0.6578464212715147,
      "learning_rate": 2.751782955626836e-07,
      "loss": 11.2663,
      "step": 22348
    },
    {
      "epoch": 3.9080218579234973,
      "grad_norm": 0.7149713119496532,
      "learning_rate": 2.741294785869819e-07,
      "loss": 11.4362,
      "step": 22349
    },
    {
      "epoch": 3.9081967213114757,
      "grad_norm": 0.628223213923468,
      "learning_rate": 2.7308266142119785e-07,
      "loss": 11.6273,
      "step": 22350
    },
    {
      "epoch": 3.9083715846994536,
      "grad_norm": 0.6370371610876674,
      "learning_rate": 2.720378440863036e-07,
      "loss": 11.604,
      "step": 22351
    },
    {
      "epoch": 3.9085464480874315,
      "grad_norm": 0.6460321612198416,
      "learning_rate": 2.7099502660324907e-07,
      "loss": 11.2349,
      "step": 22352
    },
    {
      "epoch": 3.90872131147541,
      "grad_norm": 0.6472582098397405,
      "learning_rate": 2.6995420899296187e-07,
      "loss": 11.4628,
      "step": 22353
    },
    {
      "epoch": 3.908896174863388,
      "grad_norm": 0.6940231730962656,
      "learning_rate": 2.6891539127629205e-07,
      "loss": 11.257,
      "step": 22354
    },
    {
      "epoch": 3.909071038251366,
      "grad_norm": 0.7064407868528166,
      "learning_rate": 2.6787857347410075e-07,
      "loss": 11.0881,
      "step": 22355
    },
    {
      "epoch": 3.9092459016393444,
      "grad_norm": 0.7018723982510385,
      "learning_rate": 2.66843755607149e-07,
      "loss": 11.332,
      "step": 22356
    },
    {
      "epoch": 3.9094207650273223,
      "grad_norm": 0.6567724494823386,
      "learning_rate": 2.658109376961981e-07,
      "loss": 11.2567,
      "step": 22357
    },
    {
      "epoch": 3.9095956284153006,
      "grad_norm": 0.6745128349898629,
      "learning_rate": 2.647801197619648e-07,
      "loss": 11.4759,
      "step": 22358
    },
    {
      "epoch": 3.909770491803279,
      "grad_norm": 0.6559385830144011,
      "learning_rate": 2.637513018251325e-07,
      "loss": 11.3886,
      "step": 22359
    },
    {
      "epoch": 3.909945355191257,
      "grad_norm": 0.6809519232498393,
      "learning_rate": 2.627244839062959e-07,
      "loss": 11.1902,
      "step": 22360
    },
    {
      "epoch": 3.9101202185792348,
      "grad_norm": 0.7553800707615976,
      "learning_rate": 2.6169966602608287e-07,
      "loss": 11.2941,
      "step": 22361
    },
    {
      "epoch": 3.910295081967213,
      "grad_norm": 0.6629652657636222,
      "learning_rate": 2.606768482050215e-07,
      "loss": 10.9716,
      "step": 22362
    },
    {
      "epoch": 3.9104699453551914,
      "grad_norm": 0.6698185294708041,
      "learning_rate": 2.5965603046362865e-07,
      "loss": 11.0938,
      "step": 22363
    },
    {
      "epoch": 3.9106448087431693,
      "grad_norm": 0.6320871814873457,
      "learning_rate": 2.5863721282237687e-07,
      "loss": 11.626,
      "step": 22364
    },
    {
      "epoch": 3.9108196721311477,
      "grad_norm": 0.6644974172688662,
      "learning_rate": 2.5762039530169424e-07,
      "loss": 11.2927,
      "step": 22365
    },
    {
      "epoch": 3.9109945355191256,
      "grad_norm": 0.6120700090848901,
      "learning_rate": 2.566055779219645e-07,
      "loss": 11.6045,
      "step": 22366
    },
    {
      "epoch": 3.911169398907104,
      "grad_norm": 0.6313488636979233,
      "learning_rate": 2.555927607035602e-07,
      "loss": 11.5008,
      "step": 22367
    },
    {
      "epoch": 3.911344262295082,
      "grad_norm": 0.6584277935033327,
      "learning_rate": 2.545819436667651e-07,
      "loss": 11.5111,
      "step": 22368
    },
    {
      "epoch": 3.91151912568306,
      "grad_norm": 0.6536744361984123,
      "learning_rate": 2.5357312683187415e-07,
      "loss": 11.4437,
      "step": 22369
    },
    {
      "epoch": 3.911693989071038,
      "grad_norm": 0.6371987289064572,
      "learning_rate": 2.525663102190934e-07,
      "loss": 11.5723,
      "step": 22370
    },
    {
      "epoch": 3.9118688524590164,
      "grad_norm": 0.6237094602739637,
      "learning_rate": 2.515614938486177e-07,
      "loss": 11.5604,
      "step": 22371
    },
    {
      "epoch": 3.9120437158469947,
      "grad_norm": 0.6232232625221331,
      "learning_rate": 2.505586777406088e-07,
      "loss": 11.1611,
      "step": 22372
    },
    {
      "epoch": 3.9122185792349726,
      "grad_norm": 0.6630110616890823,
      "learning_rate": 2.495578619151728e-07,
      "loss": 11.16,
      "step": 22373
    },
    {
      "epoch": 3.912393442622951,
      "grad_norm": 0.5988776886943977,
      "learning_rate": 2.4855904639237147e-07,
      "loss": 11.5234,
      "step": 22374
    },
    {
      "epoch": 3.912568306010929,
      "grad_norm": 0.7101889670473277,
      "learning_rate": 2.475622311922443e-07,
      "loss": 11.3392,
      "step": 22375
    },
    {
      "epoch": 3.912743169398907,
      "grad_norm": 0.646852953902054,
      "learning_rate": 2.465674163347642e-07,
      "loss": 11.4368,
      "step": 22376
    },
    {
      "epoch": 3.912918032786885,
      "grad_norm": 0.6662363598343031,
      "learning_rate": 2.45574601839893e-07,
      "loss": 11.3498,
      "step": 22377
    },
    {
      "epoch": 3.9130928961748634,
      "grad_norm": 0.6211727283793983,
      "learning_rate": 2.445837877275481e-07,
      "loss": 11.4938,
      "step": 22378
    },
    {
      "epoch": 3.9132677595628413,
      "grad_norm": 0.6281873093593079,
      "learning_rate": 2.4359497401758024e-07,
      "loss": 11.5597,
      "step": 22379
    },
    {
      "epoch": 3.9134426229508197,
      "grad_norm": 0.590904725754895,
      "learning_rate": 2.4260816072981806e-07,
      "loss": 11.0417,
      "step": 22380
    },
    {
      "epoch": 3.913617486338798,
      "grad_norm": 0.6602530616560777,
      "learning_rate": 2.416233478840679e-07,
      "loss": 11.2202,
      "step": 22381
    },
    {
      "epoch": 3.913792349726776,
      "grad_norm": 0.7203112039978462,
      "learning_rate": 2.406405355000696e-07,
      "loss": 11.363,
      "step": 22382
    },
    {
      "epoch": 3.9139672131147543,
      "grad_norm": 0.650414518374207,
      "learning_rate": 2.3965972359751845e-07,
      "loss": 11.3041,
      "step": 22383
    },
    {
      "epoch": 3.914142076502732,
      "grad_norm": 0.6495678709076214,
      "learning_rate": 2.3868091219609865e-07,
      "loss": 10.9752,
      "step": 22384
    },
    {
      "epoch": 3.9143169398907105,
      "grad_norm": 0.6655924823287469,
      "learning_rate": 2.3770410131543908e-07,
      "loss": 11.5451,
      "step": 22385
    },
    {
      "epoch": 3.9144918032786884,
      "grad_norm": 0.6595741495164098,
      "learning_rate": 2.3672929097512396e-07,
      "loss": 11.2616,
      "step": 22386
    },
    {
      "epoch": 3.9146666666666667,
      "grad_norm": 0.6935882691508508,
      "learning_rate": 2.3575648119469328e-07,
      "loss": 11.6144,
      "step": 22387
    },
    {
      "epoch": 3.9148415300546446,
      "grad_norm": 0.6864965174816174,
      "learning_rate": 2.3478567199366475e-07,
      "loss": 11.1315,
      "step": 22388
    },
    {
      "epoch": 3.915016393442623,
      "grad_norm": 0.6716524191743157,
      "learning_rate": 2.3381686339150055e-07,
      "loss": 11.2097,
      "step": 22389
    },
    {
      "epoch": 3.9151912568306013,
      "grad_norm": 0.6835255971491956,
      "learning_rate": 2.328500554076407e-07,
      "loss": 11.6241,
      "step": 22390
    },
    {
      "epoch": 3.915366120218579,
      "grad_norm": 0.6677430622569491,
      "learning_rate": 2.3188524806144752e-07,
      "loss": 11.367,
      "step": 22391
    },
    {
      "epoch": 3.915540983606557,
      "grad_norm": 0.666330297350337,
      "learning_rate": 2.3092244137229435e-07,
      "loss": 11.4734,
      "step": 22392
    },
    {
      "epoch": 3.9157158469945355,
      "grad_norm": 0.6950498352544555,
      "learning_rate": 2.299616353594769e-07,
      "loss": 11.4899,
      "step": 22393
    },
    {
      "epoch": 3.915890710382514,
      "grad_norm": 0.7830741164490644,
      "learning_rate": 2.2900283004225754e-07,
      "loss": 11.1764,
      "step": 22394
    },
    {
      "epoch": 3.9160655737704917,
      "grad_norm": 0.6882217564459681,
      "learning_rate": 2.280460254398764e-07,
      "loss": 11.3294,
      "step": 22395
    },
    {
      "epoch": 3.91624043715847,
      "grad_norm": 0.7081021946532264,
      "learning_rate": 2.270912215715071e-07,
      "loss": 11.1707,
      "step": 22396
    },
    {
      "epoch": 3.916415300546448,
      "grad_norm": 0.7912134408728488,
      "learning_rate": 2.261384184563009e-07,
      "loss": 11.4957,
      "step": 22397
    },
    {
      "epoch": 3.9165901639344263,
      "grad_norm": 0.6030895362229274,
      "learning_rate": 2.251876161133648e-07,
      "loss": 11.1155,
      "step": 22398
    },
    {
      "epoch": 3.9167650273224046,
      "grad_norm": 0.7147577259817968,
      "learning_rate": 2.2423881456176132e-07,
      "loss": 11.2118,
      "step": 22399
    },
    {
      "epoch": 3.9169398907103825,
      "grad_norm": 0.733517467445307,
      "learning_rate": 2.2329201382053077e-07,
      "loss": 11.4012,
      "step": 22400
    },
    {
      "epoch": 3.9171147540983604,
      "grad_norm": 0.6963989112306165,
      "learning_rate": 2.2234721390864687e-07,
      "loss": 11.2945,
      "step": 22401
    },
    {
      "epoch": 3.9172896174863387,
      "grad_norm": 0.6873734565668874,
      "learning_rate": 2.2140441484506114e-07,
      "loss": 11.3515,
      "step": 22402
    },
    {
      "epoch": 3.917464480874317,
      "grad_norm": 0.6797908951568297,
      "learning_rate": 2.2046361664868064e-07,
      "loss": 11.512,
      "step": 22403
    },
    {
      "epoch": 3.917639344262295,
      "grad_norm": 0.6193203694011078,
      "learning_rate": 2.19524819338357e-07,
      "loss": 11.505,
      "step": 22404
    },
    {
      "epoch": 3.9178142076502733,
      "grad_norm": 0.8584465029655832,
      "learning_rate": 2.185880229329307e-07,
      "loss": 11.4105,
      "step": 22405
    },
    {
      "epoch": 3.9179890710382512,
      "grad_norm": 0.675539573785163,
      "learning_rate": 2.1765322745118666e-07,
      "loss": 11.564,
      "step": 22406
    },
    {
      "epoch": 3.9181639344262296,
      "grad_norm": 0.6790506436037012,
      "learning_rate": 2.1672043291186551e-07,
      "loss": 11.27,
      "step": 22407
    },
    {
      "epoch": 3.918338797814208,
      "grad_norm": 0.715233913521971,
      "learning_rate": 2.1578963933367446e-07,
      "loss": 11.5174,
      "step": 22408
    },
    {
      "epoch": 3.918513661202186,
      "grad_norm": 0.610678344165625,
      "learning_rate": 2.148608467352875e-07,
      "loss": 11.6139,
      "step": 22409
    },
    {
      "epoch": 3.9186885245901637,
      "grad_norm": 0.7232976655103264,
      "learning_rate": 2.1393405513531195e-07,
      "loss": 11.0328,
      "step": 22410
    },
    {
      "epoch": 3.918863387978142,
      "grad_norm": 0.691311879829175,
      "learning_rate": 2.1300926455234405e-07,
      "loss": 11.1225,
      "step": 22411
    },
    {
      "epoch": 3.9190382513661204,
      "grad_norm": 0.6644232725800548,
      "learning_rate": 2.120864750049245e-07,
      "loss": 11.3075,
      "step": 22412
    },
    {
      "epoch": 3.9192131147540983,
      "grad_norm": 0.6653753891641419,
      "learning_rate": 2.1116568651156076e-07,
      "loss": 11.4193,
      "step": 22413
    },
    {
      "epoch": 3.9193879781420766,
      "grad_norm": 0.6414763460027793,
      "learning_rate": 2.102468990907269e-07,
      "loss": 11.1324,
      "step": 22414
    },
    {
      "epoch": 3.9195628415300545,
      "grad_norm": 0.779783954026308,
      "learning_rate": 2.0933011276083048e-07,
      "loss": 11.4236,
      "step": 22415
    },
    {
      "epoch": 3.919737704918033,
      "grad_norm": 0.6906474734208169,
      "learning_rate": 2.0841532754025672e-07,
      "loss": 11.4769,
      "step": 22416
    },
    {
      "epoch": 3.919912568306011,
      "grad_norm": 0.7599250005802419,
      "learning_rate": 2.0750254344735766e-07,
      "loss": 11.5758,
      "step": 22417
    },
    {
      "epoch": 3.920087431693989,
      "grad_norm": 0.6640799192496413,
      "learning_rate": 2.0659176050044083e-07,
      "loss": 11.5864,
      "step": 22418
    },
    {
      "epoch": 3.920262295081967,
      "grad_norm": 0.7309414381691547,
      "learning_rate": 2.0568297871776942e-07,
      "loss": 11.3839,
      "step": 22419
    },
    {
      "epoch": 3.9204371584699453,
      "grad_norm": 0.6650234910674448,
      "learning_rate": 2.047761981175511e-07,
      "loss": 11.5293,
      "step": 22420
    },
    {
      "epoch": 3.9206120218579237,
      "grad_norm": 0.6934728149195988,
      "learning_rate": 2.0387141871799353e-07,
      "loss": 11.2917,
      "step": 22421
    },
    {
      "epoch": 3.9207868852459016,
      "grad_norm": 0.6256145510787265,
      "learning_rate": 2.0296864053721555e-07,
      "loss": 11.3549,
      "step": 22422
    },
    {
      "epoch": 3.92096174863388,
      "grad_norm": 0.6244712492350681,
      "learning_rate": 2.0206786359334707e-07,
      "loss": 11.4465,
      "step": 22423
    },
    {
      "epoch": 3.921136612021858,
      "grad_norm": 0.6944113134631347,
      "learning_rate": 2.0116908790442924e-07,
      "loss": 11.3554,
      "step": 22424
    },
    {
      "epoch": 3.921311475409836,
      "grad_norm": 0.696745365935947,
      "learning_rate": 2.0027231348849206e-07,
      "loss": 11.4331,
      "step": 22425
    },
    {
      "epoch": 3.921486338797814,
      "grad_norm": 0.6453173986238558,
      "learning_rate": 1.9937754036352118e-07,
      "loss": 11.3518,
      "step": 22426
    },
    {
      "epoch": 3.9216612021857924,
      "grad_norm": 0.68541745282696,
      "learning_rate": 1.984847685474689e-07,
      "loss": 11.3776,
      "step": 22427
    },
    {
      "epoch": 3.9218360655737703,
      "grad_norm": 0.642042387264642,
      "learning_rate": 1.9759399805822087e-07,
      "loss": 11.5305,
      "step": 22428
    },
    {
      "epoch": 3.9220109289617486,
      "grad_norm": 0.7467755374709739,
      "learning_rate": 1.9670522891364062e-07,
      "loss": 11.2255,
      "step": 22429
    },
    {
      "epoch": 3.922185792349727,
      "grad_norm": 0.7253561075532803,
      "learning_rate": 1.9581846113156943e-07,
      "loss": 11.2518,
      "step": 22430
    },
    {
      "epoch": 3.922360655737705,
      "grad_norm": 0.6287265780241507,
      "learning_rate": 1.9493369472977086e-07,
      "loss": 11.4671,
      "step": 22431
    },
    {
      "epoch": 3.922535519125683,
      "grad_norm": 0.7206541858393639,
      "learning_rate": 1.9405092972599737e-07,
      "loss": 11.4873,
      "step": 22432
    },
    {
      "epoch": 3.922710382513661,
      "grad_norm": 0.7326072611255157,
      "learning_rate": 1.931701661379348e-07,
      "loss": 11.4245,
      "step": 22433
    },
    {
      "epoch": 3.9228852459016395,
      "grad_norm": 0.7049461923442057,
      "learning_rate": 1.9229140398326906e-07,
      "loss": 11.3454,
      "step": 22434
    },
    {
      "epoch": 3.9230601092896173,
      "grad_norm": 0.6495329396725071,
      "learning_rate": 1.9141464327960822e-07,
      "loss": 11.542,
      "step": 22435
    },
    {
      "epoch": 3.9232349726775957,
      "grad_norm": 0.6189443928508042,
      "learning_rate": 1.9053988404453828e-07,
      "loss": 11.4282,
      "step": 22436
    },
    {
      "epoch": 3.9234098360655736,
      "grad_norm": 0.6642238285014429,
      "learning_rate": 1.8966712629558957e-07,
      "loss": 11.548,
      "step": 22437
    },
    {
      "epoch": 3.923584699453552,
      "grad_norm": 0.6693742259758386,
      "learning_rate": 1.887963700502815e-07,
      "loss": 11.0661,
      "step": 22438
    },
    {
      "epoch": 3.9237595628415303,
      "grad_norm": 0.7854212325227478,
      "learning_rate": 1.8792761532606673e-07,
      "loss": 11.2777,
      "step": 22439
    },
    {
      "epoch": 3.923934426229508,
      "grad_norm": 0.7071757397265037,
      "learning_rate": 1.8706086214036467e-07,
      "loss": 11.3676,
      "step": 22440
    },
    {
      "epoch": 3.924109289617486,
      "grad_norm": 0.6935196281804955,
      "learning_rate": 1.861961105105614e-07,
      "loss": 11.2692,
      "step": 22441
    },
    {
      "epoch": 3.9242841530054644,
      "grad_norm": 0.6352169847791079,
      "learning_rate": 1.8533336045398752e-07,
      "loss": 11.1565,
      "step": 22442
    },
    {
      "epoch": 3.9244590163934427,
      "grad_norm": 0.6679977657449148,
      "learning_rate": 1.8447261198795142e-07,
      "loss": 11.6428,
      "step": 22443
    },
    {
      "epoch": 3.9246338797814206,
      "grad_norm": 0.6486349948082655,
      "learning_rate": 1.836138651297281e-07,
      "loss": 11.3469,
      "step": 22444
    },
    {
      "epoch": 3.924808743169399,
      "grad_norm": 0.7099630592946221,
      "learning_rate": 1.82757119896515e-07,
      "loss": 11.527,
      "step": 22445
    },
    {
      "epoch": 3.924983606557377,
      "grad_norm": 0.6422673386113938,
      "learning_rate": 1.8190237630549834e-07,
      "loss": 11.3185,
      "step": 22446
    },
    {
      "epoch": 3.925158469945355,
      "grad_norm": 0.7057750363610122,
      "learning_rate": 1.8104963437381993e-07,
      "loss": 11.5859,
      "step": 22447
    },
    {
      "epoch": 3.9253333333333336,
      "grad_norm": 0.6493153808567145,
      "learning_rate": 1.8019889411857728e-07,
      "loss": 11.5026,
      "step": 22448
    },
    {
      "epoch": 3.9255081967213115,
      "grad_norm": 0.6558992354153169,
      "learning_rate": 1.7935015555683444e-07,
      "loss": 11.4362,
      "step": 22449
    },
    {
      "epoch": 3.9256830601092894,
      "grad_norm": 0.6789657012017538,
      "learning_rate": 1.7850341870562225e-07,
      "loss": 11.4903,
      "step": 22450
    },
    {
      "epoch": 3.9258579234972677,
      "grad_norm": 0.6485908589547336,
      "learning_rate": 1.7765868358189385e-07,
      "loss": 11.3114,
      "step": 22451
    },
    {
      "epoch": 3.926032786885246,
      "grad_norm": 0.7184614223877877,
      "learning_rate": 1.7681595020261342e-07,
      "loss": 11.5929,
      "step": 22452
    },
    {
      "epoch": 3.926207650273224,
      "grad_norm": 0.6304268830389076,
      "learning_rate": 1.7597521858466747e-07,
      "loss": 11.4595,
      "step": 22453
    },
    {
      "epoch": 3.9263825136612023,
      "grad_norm": 0.7158485657113126,
      "learning_rate": 1.7513648874492028e-07,
      "loss": 11.0197,
      "step": 22454
    },
    {
      "epoch": 3.92655737704918,
      "grad_norm": 0.7058768802649864,
      "learning_rate": 1.742997607001917e-07,
      "loss": 11.4729,
      "step": 22455
    },
    {
      "epoch": 3.9267322404371585,
      "grad_norm": 0.6487471132429062,
      "learning_rate": 1.7346503446725727e-07,
      "loss": 11.4637,
      "step": 22456
    },
    {
      "epoch": 3.926907103825137,
      "grad_norm": 0.7329894612969164,
      "learning_rate": 1.7263231006284796e-07,
      "loss": 11.4627,
      "step": 22457
    },
    {
      "epoch": 3.9270819672131148,
      "grad_norm": 0.6438329806701011,
      "learning_rate": 1.718015875036727e-07,
      "loss": 11.2352,
      "step": 22458
    },
    {
      "epoch": 3.9272568306010927,
      "grad_norm": 0.6981762048533232,
      "learning_rate": 1.709728668063959e-07,
      "loss": 11.103,
      "step": 22459
    },
    {
      "epoch": 3.927431693989071,
      "grad_norm": 0.7455474658737141,
      "learning_rate": 1.7014614798762652e-07,
      "loss": 11.1613,
      "step": 22460
    },
    {
      "epoch": 3.9276065573770493,
      "grad_norm": 0.6632239008579996,
      "learning_rate": 1.6932143106394017e-07,
      "loss": 11.2762,
      "step": 22461
    },
    {
      "epoch": 3.9277814207650272,
      "grad_norm": 0.650512978320554,
      "learning_rate": 1.6849871605187916e-07,
      "loss": 11.4351,
      "step": 22462
    },
    {
      "epoch": 3.9279562841530056,
      "grad_norm": 0.7052542134807211,
      "learning_rate": 1.6767800296795255e-07,
      "loss": 11.4054,
      "step": 22463
    },
    {
      "epoch": 3.9281311475409835,
      "grad_norm": 0.6237578382195864,
      "learning_rate": 1.6685929182860272e-07,
      "loss": 11.2766,
      "step": 22464
    },
    {
      "epoch": 3.928306010928962,
      "grad_norm": 0.7456835778848077,
      "learning_rate": 1.6604258265024985e-07,
      "loss": 11.2092,
      "step": 22465
    },
    {
      "epoch": 3.92848087431694,
      "grad_norm": 0.7180103918041955,
      "learning_rate": 1.6522787544926977e-07,
      "loss": 11.2584,
      "step": 22466
    },
    {
      "epoch": 3.928655737704918,
      "grad_norm": 0.7144838708700324,
      "learning_rate": 1.6441517024200492e-07,
      "loss": 11.2822,
      "step": 22467
    },
    {
      "epoch": 3.928830601092896,
      "grad_norm": 0.6495806760286263,
      "learning_rate": 1.636044670447534e-07,
      "loss": 11.323,
      "step": 22468
    },
    {
      "epoch": 3.9290054644808743,
      "grad_norm": 0.7163558748860201,
      "learning_rate": 1.6279576587378e-07,
      "loss": 11.4375,
      "step": 22469
    },
    {
      "epoch": 3.9291803278688526,
      "grad_norm": 0.6157535456651549,
      "learning_rate": 1.619890667452828e-07,
      "loss": 11.5723,
      "step": 22470
    },
    {
      "epoch": 3.9293551912568305,
      "grad_norm": 0.651362712703917,
      "learning_rate": 1.611843696754489e-07,
      "loss": 11.5009,
      "step": 22471
    },
    {
      "epoch": 3.929530054644809,
      "grad_norm": 0.6653675566426803,
      "learning_rate": 1.603816746804099e-07,
      "loss": 11.4036,
      "step": 22472
    },
    {
      "epoch": 3.9297049180327868,
      "grad_norm": 0.6609105016372661,
      "learning_rate": 1.5958098177626392e-07,
      "loss": 11.4915,
      "step": 22473
    },
    {
      "epoch": 3.929879781420765,
      "grad_norm": 0.676561343759443,
      "learning_rate": 1.5878229097907593e-07,
      "loss": 11.236,
      "step": 22474
    },
    {
      "epoch": 3.930054644808743,
      "grad_norm": 0.7857490513498898,
      "learning_rate": 1.5798560230485537e-07,
      "loss": 11.0662,
      "step": 22475
    },
    {
      "epoch": 3.9302295081967213,
      "grad_norm": 0.6045394485460099,
      "learning_rate": 1.5719091576957835e-07,
      "loss": 11.5919,
      "step": 22476
    },
    {
      "epoch": 3.9304043715846992,
      "grad_norm": 0.6796868738062164,
      "learning_rate": 1.5639823138918764e-07,
      "loss": 11.3518,
      "step": 22477
    },
    {
      "epoch": 3.9305792349726776,
      "grad_norm": 0.6237262080586945,
      "learning_rate": 1.5560754917955945e-07,
      "loss": 11.4522,
      "step": 22478
    },
    {
      "epoch": 3.930754098360656,
      "grad_norm": 0.695556014670135,
      "learning_rate": 1.5481886915656994e-07,
      "loss": 11.2242,
      "step": 22479
    },
    {
      "epoch": 3.930928961748634,
      "grad_norm": 0.6475178400173163,
      "learning_rate": 1.5403219133602876e-07,
      "loss": 11.5514,
      "step": 22480
    },
    {
      "epoch": 3.931103825136612,
      "grad_norm": 0.6381445442836027,
      "learning_rate": 1.53247515733701e-07,
      "loss": 11.2255,
      "step": 22481
    },
    {
      "epoch": 3.93127868852459,
      "grad_norm": 0.6645708627307901,
      "learning_rate": 1.5246484236534076e-07,
      "loss": 11.3716,
      "step": 22482
    },
    {
      "epoch": 3.9314535519125684,
      "grad_norm": 0.6113092566048256,
      "learning_rate": 1.5168417124662437e-07,
      "loss": 11.0126,
      "step": 22483
    },
    {
      "epoch": 3.9316284153005463,
      "grad_norm": 0.7005482489553634,
      "learning_rate": 1.509055023932171e-07,
      "loss": 11.4276,
      "step": 22484
    },
    {
      "epoch": 3.9318032786885246,
      "grad_norm": 0.6366024681442827,
      "learning_rate": 1.5012883582073976e-07,
      "loss": 11.5706,
      "step": 22485
    },
    {
      "epoch": 3.9319781420765025,
      "grad_norm": 0.7230879652107659,
      "learning_rate": 1.4935417154474662e-07,
      "loss": 11.276,
      "step": 22486
    },
    {
      "epoch": 3.932153005464481,
      "grad_norm": 0.6736647932172664,
      "learning_rate": 1.4858150958079188e-07,
      "loss": 11.4963,
      "step": 22487
    },
    {
      "epoch": 3.932327868852459,
      "grad_norm": 1.2074658494032884,
      "learning_rate": 1.4781084994436313e-07,
      "loss": 11.0637,
      "step": 22488
    },
    {
      "epoch": 3.932502732240437,
      "grad_norm": 0.6748496801277769,
      "learning_rate": 1.470421926509036e-07,
      "loss": 11.1646,
      "step": 22489
    },
    {
      "epoch": 3.9326775956284155,
      "grad_norm": 0.6333637499008907,
      "learning_rate": 1.4627553771584535e-07,
      "loss": 11.2123,
      "step": 22490
    },
    {
      "epoch": 3.9328524590163934,
      "grad_norm": 0.6249922367189861,
      "learning_rate": 1.455108851545539e-07,
      "loss": 11.7045,
      "step": 22491
    },
    {
      "epoch": 3.9330273224043717,
      "grad_norm": 0.7501063929163908,
      "learning_rate": 1.4474823498236146e-07,
      "loss": 11.4611,
      "step": 22492
    },
    {
      "epoch": 3.9332021857923496,
      "grad_norm": 0.690033773310146,
      "learning_rate": 1.4398758721455574e-07,
      "loss": 11.1504,
      "step": 22493
    },
    {
      "epoch": 3.933377049180328,
      "grad_norm": 0.6978628316049612,
      "learning_rate": 1.4322894186640236e-07,
      "loss": 11.5587,
      "step": 22494
    },
    {
      "epoch": 3.933551912568306,
      "grad_norm": 0.6599034508414116,
      "learning_rate": 1.424722989531113e-07,
      "loss": 11.2522,
      "step": 22495
    },
    {
      "epoch": 3.933726775956284,
      "grad_norm": 0.6243464038119256,
      "learning_rate": 1.4171765848984831e-07,
      "loss": 11.6332,
      "step": 22496
    },
    {
      "epoch": 3.9339016393442625,
      "grad_norm": 0.6976271578428674,
      "learning_rate": 1.4096502049175674e-07,
      "loss": 11.3793,
      "step": 22497
    },
    {
      "epoch": 3.9340765027322404,
      "grad_norm": 0.671485039564653,
      "learning_rate": 1.4021438497392457e-07,
      "loss": 11.3721,
      "step": 22498
    },
    {
      "epoch": 3.9342513661202183,
      "grad_norm": 0.6995118104053302,
      "learning_rate": 1.3946575195140642e-07,
      "loss": 11.4456,
      "step": 22499
    },
    {
      "epoch": 3.9344262295081966,
      "grad_norm": 0.6313576862591102,
      "learning_rate": 1.387191214392014e-07,
      "loss": 11.6027,
      "step": 22500
    },
    {
      "epoch": 3.934601092896175,
      "grad_norm": 0.7538798010347001,
      "learning_rate": 1.3797449345229752e-07,
      "loss": 11.3071,
      "step": 22501
    },
    {
      "epoch": 3.934775956284153,
      "grad_norm": 0.9181990757626116,
      "learning_rate": 1.3723186800562726e-07,
      "loss": 11.4263,
      "step": 22502
    },
    {
      "epoch": 3.9349508196721312,
      "grad_norm": 0.645824609377724,
      "learning_rate": 1.3649124511406764e-07,
      "loss": 11.3611,
      "step": 22503
    },
    {
      "epoch": 3.935125683060109,
      "grad_norm": 0.6734601919832514,
      "learning_rate": 1.3575262479248453e-07,
      "loss": 11.4203,
      "step": 22504
    },
    {
      "epoch": 3.9353005464480875,
      "grad_norm": 0.6806724361320413,
      "learning_rate": 1.3501600705568828e-07,
      "loss": 11.5164,
      "step": 22505
    },
    {
      "epoch": 3.935475409836066,
      "grad_norm": 0.6477107672316273,
      "learning_rate": 1.3428139191844492e-07,
      "loss": 11.3818,
      "step": 22506
    },
    {
      "epoch": 3.9356502732240437,
      "grad_norm": 0.642165656713554,
      "learning_rate": 1.3354877939548703e-07,
      "loss": 11.3405,
      "step": 22507
    },
    {
      "epoch": 3.9358251366120216,
      "grad_norm": 0.7038182962224989,
      "learning_rate": 1.3281816950150295e-07,
      "loss": 11.2191,
      "step": 22508
    },
    {
      "epoch": 3.936,
      "grad_norm": 0.6609625557811841,
      "learning_rate": 1.3208956225114756e-07,
      "loss": 11.3566,
      "step": 22509
    },
    {
      "epoch": 3.9361748633879783,
      "grad_norm": 0.7374829977323406,
      "learning_rate": 1.313629576590314e-07,
      "loss": 10.9511,
      "step": 22510
    },
    {
      "epoch": 3.936349726775956,
      "grad_norm": 0.5951941120152193,
      "learning_rate": 1.306383557397317e-07,
      "loss": 11.4414,
      "step": 22511
    },
    {
      "epoch": 3.9365245901639345,
      "grad_norm": 0.7175206530438353,
      "learning_rate": 1.299157565077702e-07,
      "loss": 11.5123,
      "step": 22512
    },
    {
      "epoch": 3.9366994535519124,
      "grad_norm": 0.6967967905198571,
      "learning_rate": 1.2919515997763532e-07,
      "loss": 11.5337,
      "step": 22513
    },
    {
      "epoch": 3.9368743169398908,
      "grad_norm": 0.7009390687559947,
      "learning_rate": 1.284765661637821e-07,
      "loss": 11.6098,
      "step": 22514
    },
    {
      "epoch": 3.937049180327869,
      "grad_norm": 0.6803651171975241,
      "learning_rate": 1.277599750806102e-07,
      "loss": 11.2024,
      "step": 22515
    },
    {
      "epoch": 3.937224043715847,
      "grad_norm": 0.6381131299610372,
      "learning_rate": 1.270453867425081e-07,
      "loss": 11.0253,
      "step": 22516
    },
    {
      "epoch": 3.937398907103825,
      "grad_norm": 0.6877912086141683,
      "learning_rate": 1.2633280116379763e-07,
      "loss": 11.2652,
      "step": 22517
    },
    {
      "epoch": 3.9375737704918032,
      "grad_norm": 0.7285307206070513,
      "learning_rate": 1.2562221835876743e-07,
      "loss": 10.8618,
      "step": 22518
    },
    {
      "epoch": 3.9377486338797816,
      "grad_norm": 0.6702697393027067,
      "learning_rate": 1.249136383416727e-07,
      "loss": 11.4652,
      "step": 22519
    },
    {
      "epoch": 3.9379234972677595,
      "grad_norm": 0.6086055296507513,
      "learning_rate": 1.242070611267021e-07,
      "loss": 11.4945,
      "step": 22520
    },
    {
      "epoch": 3.938098360655738,
      "grad_norm": 0.6515905122289277,
      "learning_rate": 1.2350248672804433e-07,
      "loss": 11.5058,
      "step": 22521
    },
    {
      "epoch": 3.9382732240437157,
      "grad_norm": 0.616580049620673,
      "learning_rate": 1.2279991515982138e-07,
      "loss": 11.3899,
      "step": 22522
    },
    {
      "epoch": 3.938448087431694,
      "grad_norm": 0.7049009240272271,
      "learning_rate": 1.2209934643613308e-07,
      "loss": 11.4374,
      "step": 22523
    },
    {
      "epoch": 3.9386229508196724,
      "grad_norm": 0.6730694830709607,
      "learning_rate": 1.2140078057101266e-07,
      "loss": 11.4388,
      "step": 22524
    },
    {
      "epoch": 3.9387978142076503,
      "grad_norm": 0.683927467209737,
      "learning_rate": 1.2070421757846007e-07,
      "loss": 11.4704,
      "step": 22525
    },
    {
      "epoch": 3.938972677595628,
      "grad_norm": 0.6795264203511227,
      "learning_rate": 1.2000965747247516e-07,
      "loss": 11.09,
      "step": 22526
    },
    {
      "epoch": 3.9391475409836065,
      "grad_norm": 0.7143304645163371,
      "learning_rate": 1.1931710026695797e-07,
      "loss": 11.3275,
      "step": 22527
    },
    {
      "epoch": 3.939322404371585,
      "grad_norm": 0.6920836379285172,
      "learning_rate": 1.1862654597581957e-07,
      "loss": 11.5525,
      "step": 22528
    },
    {
      "epoch": 3.9394972677595628,
      "grad_norm": 0.6522679516257818,
      "learning_rate": 1.1793799461288224e-07,
      "loss": 11.4762,
      "step": 22529
    },
    {
      "epoch": 3.939672131147541,
      "grad_norm": 0.7045299900069616,
      "learning_rate": 1.1725144619197937e-07,
      "loss": 11.4474,
      "step": 22530
    },
    {
      "epoch": 3.939846994535519,
      "grad_norm": 0.6698258206006242,
      "learning_rate": 1.165669007268555e-07,
      "loss": 11.2615,
      "step": 22531
    },
    {
      "epoch": 3.9400218579234974,
      "grad_norm": 0.6935902628862161,
      "learning_rate": 1.158843582312441e-07,
      "loss": 11.4492,
      "step": 22532
    },
    {
      "epoch": 3.9401967213114752,
      "grad_norm": 0.7650500444415612,
      "learning_rate": 1.1520381871884533e-07,
      "loss": 11.333,
      "step": 22533
    },
    {
      "epoch": 3.9403715846994536,
      "grad_norm": 0.6669731644784047,
      "learning_rate": 1.1452528220329273e-07,
      "loss": 11.1966,
      "step": 22534
    },
    {
      "epoch": 3.9405464480874315,
      "grad_norm": 0.7111906110422944,
      "learning_rate": 1.1384874869819761e-07,
      "loss": 11.235,
      "step": 22535
    },
    {
      "epoch": 3.94072131147541,
      "grad_norm": 0.6066995407052442,
      "learning_rate": 1.1317421821712693e-07,
      "loss": 11.2524,
      "step": 22536
    },
    {
      "epoch": 3.940896174863388,
      "grad_norm": 0.6509304538201129,
      "learning_rate": 1.1250169077360317e-07,
      "loss": 11.2982,
      "step": 22537
    },
    {
      "epoch": 3.941071038251366,
      "grad_norm": 0.6842523661213461,
      "learning_rate": 1.1183116638111558e-07,
      "loss": 11.5189,
      "step": 22538
    },
    {
      "epoch": 3.9412459016393444,
      "grad_norm": 0.6188058029333887,
      "learning_rate": 1.1116264505310891e-07,
      "loss": 11.4569,
      "step": 22539
    },
    {
      "epoch": 3.9414207650273223,
      "grad_norm": 0.6713649179922503,
      "learning_rate": 1.104961268029947e-07,
      "loss": 11.5777,
      "step": 22540
    },
    {
      "epoch": 3.9415956284153006,
      "grad_norm": 0.6968256052913118,
      "learning_rate": 1.0983161164412892e-07,
      "loss": 11.433,
      "step": 22541
    },
    {
      "epoch": 3.9417704918032785,
      "grad_norm": 0.6520087568481878,
      "learning_rate": 1.0916909958984533e-07,
      "loss": 11.3754,
      "step": 22542
    },
    {
      "epoch": 3.941945355191257,
      "grad_norm": 0.6628427238921545,
      "learning_rate": 1.0850859065343333e-07,
      "loss": 11.3891,
      "step": 22543
    },
    {
      "epoch": 3.942120218579235,
      "grad_norm": 0.6928109910591579,
      "learning_rate": 1.0785008484811566e-07,
      "loss": 11.2517,
      "step": 22544
    },
    {
      "epoch": 3.942295081967213,
      "grad_norm": 0.7107635240050634,
      "learning_rate": 1.0719358218712617e-07,
      "loss": 11.2548,
      "step": 22545
    },
    {
      "epoch": 3.9424699453551915,
      "grad_norm": 0.7382550544742804,
      "learning_rate": 1.0653908268360991e-07,
      "loss": 11.3084,
      "step": 22546
    },
    {
      "epoch": 3.9426448087431694,
      "grad_norm": 0.6603484123102085,
      "learning_rate": 1.0588658635070082e-07,
      "loss": 11.4311,
      "step": 22547
    },
    {
      "epoch": 3.9428196721311477,
      "grad_norm": 0.6758332144551012,
      "learning_rate": 1.0523609320147732e-07,
      "loss": 11.3366,
      "step": 22548
    },
    {
      "epoch": 3.9429945355191256,
      "grad_norm": 0.6632542087714229,
      "learning_rate": 1.0458760324899563e-07,
      "loss": 11.1625,
      "step": 22549
    },
    {
      "epoch": 3.943169398907104,
      "grad_norm": 0.6366181569264827,
      "learning_rate": 1.0394111650624539e-07,
      "loss": 11.2626,
      "step": 22550
    },
    {
      "epoch": 3.943344262295082,
      "grad_norm": 0.7205398345136551,
      "learning_rate": 1.0329663298619397e-07,
      "loss": 11.355,
      "step": 22551
    },
    {
      "epoch": 3.94351912568306,
      "grad_norm": 0.6624717982961525,
      "learning_rate": 1.0265415270177547e-07,
      "loss": 11.5321,
      "step": 22552
    },
    {
      "epoch": 3.943693989071038,
      "grad_norm": 0.6144949281684653,
      "learning_rate": 1.020136756658574e-07,
      "loss": 11.5838,
      "step": 22553
    },
    {
      "epoch": 3.9438688524590164,
      "grad_norm": 0.6809090348038217,
      "learning_rate": 1.0137520189129612e-07,
      "loss": 11.3221,
      "step": 22554
    },
    {
      "epoch": 3.9440437158469948,
      "grad_norm": 0.7358432580807663,
      "learning_rate": 1.0073873139090362e-07,
      "loss": 11.3452,
      "step": 22555
    },
    {
      "epoch": 3.9442185792349727,
      "grad_norm": 0.6718847850118473,
      "learning_rate": 1.0010426417741415e-07,
      "loss": 11.5283,
      "step": 22556
    },
    {
      "epoch": 3.9443934426229506,
      "grad_norm": 0.6511270392921777,
      "learning_rate": 9.947180026357305e-08,
      "loss": 11.3101,
      "step": 22557
    },
    {
      "epoch": 3.944568306010929,
      "grad_norm": 0.6367050264754648,
      "learning_rate": 9.884133966205911e-08,
      "loss": 11.6058,
      "step": 22558
    },
    {
      "epoch": 3.9447431693989072,
      "grad_norm": 0.6828995278202176,
      "learning_rate": 9.821288238550663e-08,
      "loss": 11.4139,
      "step": 22559
    },
    {
      "epoch": 3.944918032786885,
      "grad_norm": 0.6530421420381992,
      "learning_rate": 9.758642844653887e-08,
      "loss": 11.1463,
      "step": 22560
    },
    {
      "epoch": 3.9450928961748635,
      "grad_norm": 0.6812116186775645,
      "learning_rate": 9.696197785769024e-08,
      "loss": 11.5987,
      "step": 22561
    },
    {
      "epoch": 3.9452677595628414,
      "grad_norm": 0.6693038177302439,
      "learning_rate": 9.633953063149515e-08,
      "loss": 11.5949,
      "step": 22562
    },
    {
      "epoch": 3.9454426229508197,
      "grad_norm": 0.7040741225036563,
      "learning_rate": 9.571908678044362e-08,
      "loss": 11.358,
      "step": 22563
    },
    {
      "epoch": 3.945617486338798,
      "grad_norm": 1.2362498965789004,
      "learning_rate": 9.510064631697013e-08,
      "loss": 11.318,
      "step": 22564
    },
    {
      "epoch": 3.945792349726776,
      "grad_norm": 0.6937217882254195,
      "learning_rate": 9.44842092534759e-08,
      "loss": 11.559,
      "step": 22565
    },
    {
      "epoch": 3.945967213114754,
      "grad_norm": 0.6883883148702796,
      "learning_rate": 9.386977560232879e-08,
      "loss": 11.3627,
      "step": 22566
    },
    {
      "epoch": 3.946142076502732,
      "grad_norm": 0.6215244625690559,
      "learning_rate": 9.325734537584119e-08,
      "loss": 11.4215,
      "step": 22567
    },
    {
      "epoch": 3.9463169398907105,
      "grad_norm": 0.6840414111862467,
      "learning_rate": 9.264691858629215e-08,
      "loss": 11.272,
      "step": 22568
    },
    {
      "epoch": 3.9464918032786884,
      "grad_norm": 0.7035933460210688,
      "learning_rate": 9.203849524593855e-08,
      "loss": 11.5682,
      "step": 22569
    },
    {
      "epoch": 3.9466666666666668,
      "grad_norm": 0.6776959881499046,
      "learning_rate": 9.143207536695953e-08,
      "loss": 11.3298,
      "step": 22570
    },
    {
      "epoch": 3.9468415300546447,
      "grad_norm": 0.6864052105025368,
      "learning_rate": 9.082765896153423e-08,
      "loss": 11.5078,
      "step": 22571
    },
    {
      "epoch": 3.947016393442623,
      "grad_norm": 0.6764800189696792,
      "learning_rate": 9.022524604177518e-08,
      "loss": 11.5449,
      "step": 22572
    },
    {
      "epoch": 3.9471912568306013,
      "grad_norm": 0.6899481422262291,
      "learning_rate": 8.962483661976162e-08,
      "loss": 11.4283,
      "step": 22573
    },
    {
      "epoch": 3.9473661202185792,
      "grad_norm": 0.6934109621636503,
      "learning_rate": 8.902643070753946e-08,
      "loss": 11.5551,
      "step": 22574
    },
    {
      "epoch": 3.947540983606557,
      "grad_norm": 0.6091360830992557,
      "learning_rate": 8.843002831709912e-08,
      "loss": 11.4191,
      "step": 22575
    },
    {
      "epoch": 3.9477158469945355,
      "grad_norm": 0.667899763804758,
      "learning_rate": 8.783562946040879e-08,
      "loss": 11.3559,
      "step": 22576
    },
    {
      "epoch": 3.947890710382514,
      "grad_norm": 0.6072570595906205,
      "learning_rate": 8.724323414938118e-08,
      "loss": 11.619,
      "step": 22577
    },
    {
      "epoch": 3.9480655737704917,
      "grad_norm": 0.6421802424918382,
      "learning_rate": 8.665284239589567e-08,
      "loss": 11.4861,
      "step": 22578
    },
    {
      "epoch": 3.94824043715847,
      "grad_norm": 0.6989282512426216,
      "learning_rate": 8.606445421179831e-08,
      "loss": 11.5414,
      "step": 22579
    },
    {
      "epoch": 3.948415300546448,
      "grad_norm": 0.6163453444282139,
      "learning_rate": 8.547806960889082e-08,
      "loss": 11.2054,
      "step": 22580
    },
    {
      "epoch": 3.9485901639344263,
      "grad_norm": 0.7226693346462867,
      "learning_rate": 8.489368859891933e-08,
      "loss": 11.5185,
      "step": 22581
    },
    {
      "epoch": 3.9487650273224046,
      "grad_norm": 0.7095992146879613,
      "learning_rate": 8.43113111936189e-08,
      "loss": 11.2701,
      "step": 22582
    },
    {
      "epoch": 3.9489398907103825,
      "grad_norm": 0.7021470354400763,
      "learning_rate": 8.373093740464688e-08,
      "loss": 11.1392,
      "step": 22583
    },
    {
      "epoch": 3.9491147540983604,
      "grad_norm": 0.7314619073138535,
      "learning_rate": 8.31525672436606e-08,
      "loss": 11.2084,
      "step": 22584
    },
    {
      "epoch": 3.9492896174863388,
      "grad_norm": 0.7121134025210322,
      "learning_rate": 8.257620072223971e-08,
      "loss": 11.5439,
      "step": 22585
    },
    {
      "epoch": 3.949464480874317,
      "grad_norm": 0.9339438487629151,
      "learning_rate": 8.20018378519638e-08,
      "loss": 11.3632,
      "step": 22586
    },
    {
      "epoch": 3.949639344262295,
      "grad_norm": 0.6436955109091678,
      "learning_rate": 8.142947864433481e-08,
      "loss": 11.4725,
      "step": 22587
    },
    {
      "epoch": 3.9498142076502734,
      "grad_norm": 0.6795358782606206,
      "learning_rate": 8.085912311083244e-08,
      "loss": 11.392,
      "step": 22588
    },
    {
      "epoch": 3.9499890710382513,
      "grad_norm": 0.7308404880937609,
      "learning_rate": 8.029077126290307e-08,
      "loss": 11.0484,
      "step": 22589
    },
    {
      "epoch": 3.9501639344262296,
      "grad_norm": 0.6670487779050347,
      "learning_rate": 7.972442311193762e-08,
      "loss": 11.2683,
      "step": 22590
    },
    {
      "epoch": 3.9503387978142075,
      "grad_norm": 0.5735021157992766,
      "learning_rate": 7.916007866928255e-08,
      "loss": 11.1844,
      "step": 22591
    },
    {
      "epoch": 3.950513661202186,
      "grad_norm": 0.6295679394789773,
      "learning_rate": 7.859773794627323e-08,
      "loss": 11.3811,
      "step": 22592
    },
    {
      "epoch": 3.9506885245901637,
      "grad_norm": 0.6468511598628126,
      "learning_rate": 7.803740095417844e-08,
      "loss": 11.5155,
      "step": 22593
    },
    {
      "epoch": 3.950863387978142,
      "grad_norm": 0.6486324982887857,
      "learning_rate": 7.747906770423363e-08,
      "loss": 11.5019,
      "step": 22594
    },
    {
      "epoch": 3.9510382513661204,
      "grad_norm": 0.6115168267381722,
      "learning_rate": 7.692273820764095e-08,
      "loss": 11.4541,
      "step": 22595
    },
    {
      "epoch": 3.9512131147540983,
      "grad_norm": 0.6139374996595687,
      "learning_rate": 7.636841247554705e-08,
      "loss": 11.3769,
      "step": 22596
    },
    {
      "epoch": 3.9513879781420767,
      "grad_norm": 0.6526452019085025,
      "learning_rate": 7.581609051906524e-08,
      "loss": 11.5115,
      "step": 22597
    },
    {
      "epoch": 3.9515628415300545,
      "grad_norm": 0.6791473348990807,
      "learning_rate": 7.526577234928666e-08,
      "loss": 11.4104,
      "step": 22598
    },
    {
      "epoch": 3.951737704918033,
      "grad_norm": 0.7258809131820391,
      "learning_rate": 7.471745797723584e-08,
      "loss": 11.3228,
      "step": 22599
    },
    {
      "epoch": 3.951912568306011,
      "grad_norm": 0.6344691376646181,
      "learning_rate": 7.417114741391506e-08,
      "loss": 11.5904,
      "step": 22600
    },
    {
      "epoch": 3.952087431693989,
      "grad_norm": 0.6652714751897211,
      "learning_rate": 7.362684067027115e-08,
      "loss": 11.632,
      "step": 22601
    },
    {
      "epoch": 3.952262295081967,
      "grad_norm": 0.7082657867782888,
      "learning_rate": 7.308453775721758e-08,
      "loss": 11.5801,
      "step": 22602
    },
    {
      "epoch": 3.9524371584699454,
      "grad_norm": 0.7252488109538836,
      "learning_rate": 7.254423868564564e-08,
      "loss": 11.2578,
      "step": 22603
    },
    {
      "epoch": 3.9526120218579237,
      "grad_norm": 0.6788882095230786,
      "learning_rate": 7.200594346638001e-08,
      "loss": 11.3269,
      "step": 22604
    },
    {
      "epoch": 3.9527868852459016,
      "grad_norm": 0.6856534422404912,
      "learning_rate": 7.146965211020096e-08,
      "loss": 11.1,
      "step": 22605
    },
    {
      "epoch": 3.95296174863388,
      "grad_norm": 0.6618099420678527,
      "learning_rate": 7.093536462788875e-08,
      "loss": 11.4671,
      "step": 22606
    },
    {
      "epoch": 3.953136612021858,
      "grad_norm": 0.625781532651194,
      "learning_rate": 7.040308103014592e-08,
      "loss": 11.2553,
      "step": 22607
    },
    {
      "epoch": 3.953311475409836,
      "grad_norm": 0.6762853188184098,
      "learning_rate": 6.987280132763064e-08,
      "loss": 11.3942,
      "step": 22608
    },
    {
      "epoch": 3.953486338797814,
      "grad_norm": 0.7053903105244107,
      "learning_rate": 6.934452553101212e-08,
      "loss": 11.4878,
      "step": 22609
    },
    {
      "epoch": 3.9536612021857924,
      "grad_norm": 0.7527593775421905,
      "learning_rate": 6.88182536508486e-08,
      "loss": 11.473,
      "step": 22610
    },
    {
      "epoch": 3.9538360655737703,
      "grad_norm": 0.6551954853522169,
      "learning_rate": 6.829398569770939e-08,
      "loss": 11.3488,
      "step": 22611
    },
    {
      "epoch": 3.9540109289617487,
      "grad_norm": 0.7442896183926374,
      "learning_rate": 6.777172168210833e-08,
      "loss": 11.2184,
      "step": 22612
    },
    {
      "epoch": 3.954185792349727,
      "grad_norm": 0.9649158632036675,
      "learning_rate": 6.725146161451478e-08,
      "loss": 11.4699,
      "step": 22613
    },
    {
      "epoch": 3.954360655737705,
      "grad_norm": 0.7711298122812066,
      "learning_rate": 6.673320550537598e-08,
      "loss": 11.4048,
      "step": 22614
    },
    {
      "epoch": 3.954535519125683,
      "grad_norm": 0.6440236185538808,
      "learning_rate": 6.62169533650614e-08,
      "loss": 11.35,
      "step": 22615
    },
    {
      "epoch": 3.954710382513661,
      "grad_norm": 0.6722511605390623,
      "learning_rate": 6.570270520392941e-08,
      "loss": 11.4975,
      "step": 22616
    },
    {
      "epoch": 3.9548852459016395,
      "grad_norm": 0.6584996350506518,
      "learning_rate": 6.519046103230508e-08,
      "loss": 11.5174,
      "step": 22617
    },
    {
      "epoch": 3.9550601092896174,
      "grad_norm": 0.6658546876815671,
      "learning_rate": 6.468022086044689e-08,
      "loss": 11.6604,
      "step": 22618
    },
    {
      "epoch": 3.9552349726775957,
      "grad_norm": 0.6869405467656652,
      "learning_rate": 6.41719846985911e-08,
      "loss": 11.3754,
      "step": 22619
    },
    {
      "epoch": 3.9554098360655736,
      "grad_norm": 0.6367707361098169,
      "learning_rate": 6.366575255694062e-08,
      "loss": 11.2381,
      "step": 22620
    },
    {
      "epoch": 3.955584699453552,
      "grad_norm": 0.6152980466522573,
      "learning_rate": 6.316152444562074e-08,
      "loss": 11.3728,
      "step": 22621
    },
    {
      "epoch": 3.9557595628415303,
      "grad_norm": 0.6858061659954898,
      "learning_rate": 6.265930037476775e-08,
      "loss": 11.4244,
      "step": 22622
    },
    {
      "epoch": 3.955934426229508,
      "grad_norm": 0.6976864566613635,
      "learning_rate": 6.21590803544403e-08,
      "loss": 11.4044,
      "step": 22623
    },
    {
      "epoch": 3.956109289617486,
      "grad_norm": 0.7170191801430159,
      "learning_rate": 6.166086439468588e-08,
      "loss": 11.3059,
      "step": 22624
    },
    {
      "epoch": 3.9562841530054644,
      "grad_norm": 0.6099742649102717,
      "learning_rate": 6.116465250546322e-08,
      "loss": 11.0701,
      "step": 22625
    },
    {
      "epoch": 3.9564590163934428,
      "grad_norm": 0.6324352330473789,
      "learning_rate": 6.06704446967532e-08,
      "loss": 11.3954,
      "step": 22626
    },
    {
      "epoch": 3.9566338797814207,
      "grad_norm": 0.7522681976283068,
      "learning_rate": 6.017824097844793e-08,
      "loss": 11.2433,
      "step": 22627
    },
    {
      "epoch": 3.956808743169399,
      "grad_norm": 0.7396611262248499,
      "learning_rate": 5.968804136043948e-08,
      "loss": 11.2566,
      "step": 22628
    },
    {
      "epoch": 3.956983606557377,
      "grad_norm": 0.6844541379624935,
      "learning_rate": 5.919984585253113e-08,
      "loss": 11.7013,
      "step": 22629
    },
    {
      "epoch": 3.9571584699453553,
      "grad_norm": 0.6463338941597035,
      "learning_rate": 5.8713654464515046e-08,
      "loss": 11.2561,
      "step": 22630
    },
    {
      "epoch": 3.9573333333333336,
      "grad_norm": 0.7072573026223492,
      "learning_rate": 5.822946720616118e-08,
      "loss": 11.273,
      "step": 22631
    },
    {
      "epoch": 3.9575081967213115,
      "grad_norm": 0.8198354517675867,
      "learning_rate": 5.774728408717289e-08,
      "loss": 11.2289,
      "step": 22632
    },
    {
      "epoch": 3.9576830601092894,
      "grad_norm": 0.6952693920234445,
      "learning_rate": 5.726710511720912e-08,
      "loss": 11.1098,
      "step": 22633
    },
    {
      "epoch": 3.9578579234972677,
      "grad_norm": 0.6551312722695083,
      "learning_rate": 5.6788930305906595e-08,
      "loss": 11.1169,
      "step": 22634
    },
    {
      "epoch": 3.958032786885246,
      "grad_norm": 0.7020664501887802,
      "learning_rate": 5.631275966284655e-08,
      "loss": 11.5307,
      "step": 22635
    },
    {
      "epoch": 3.958207650273224,
      "grad_norm": 0.6478351636963768,
      "learning_rate": 5.5838593197588e-08,
      "loss": 11.2788,
      "step": 22636
    },
    {
      "epoch": 3.9583825136612023,
      "grad_norm": 0.6461004375008577,
      "learning_rate": 5.536643091963445e-08,
      "loss": 11.4449,
      "step": 22637
    },
    {
      "epoch": 3.95855737704918,
      "grad_norm": 0.698343721718206,
      "learning_rate": 5.4896272838445004e-08,
      "loss": 11.4533,
      "step": 22638
    },
    {
      "epoch": 3.9587322404371585,
      "grad_norm": 0.6761236192523956,
      "learning_rate": 5.4428118963467665e-08,
      "loss": 11.4484,
      "step": 22639
    },
    {
      "epoch": 3.958907103825137,
      "grad_norm": 0.6465165126363499,
      "learning_rate": 5.3961969304072715e-08,
      "loss": 11.3944,
      "step": 22640
    },
    {
      "epoch": 3.959081967213115,
      "grad_norm": 0.6315501180254033,
      "learning_rate": 5.3497823869619326e-08,
      "loss": 11.2505,
      "step": 22641
    },
    {
      "epoch": 3.9592568306010927,
      "grad_norm": 0.6569521279647375,
      "learning_rate": 5.303568266941117e-08,
      "loss": 11.1074,
      "step": 22642
    },
    {
      "epoch": 3.959431693989071,
      "grad_norm": 0.6574781837801765,
      "learning_rate": 5.257554571271861e-08,
      "loss": 11.6977,
      "step": 22643
    },
    {
      "epoch": 3.9596065573770494,
      "grad_norm": 0.7529367175012641,
      "learning_rate": 5.21174130087565e-08,
      "loss": 10.9183,
      "step": 22644
    },
    {
      "epoch": 3.9597814207650273,
      "grad_norm": 0.6536243734070728,
      "learning_rate": 5.166128456672858e-08,
      "loss": 11.4975,
      "step": 22645
    },
    {
      "epoch": 3.9599562841530056,
      "grad_norm": 0.7085958146137489,
      "learning_rate": 5.1207160395772003e-08,
      "loss": 11.2394,
      "step": 22646
    },
    {
      "epoch": 3.9601311475409835,
      "grad_norm": 0.6801733288250521,
      "learning_rate": 5.075504050499058e-08,
      "loss": 11.487,
      "step": 22647
    },
    {
      "epoch": 3.960306010928962,
      "grad_norm": 0.6961233884992505,
      "learning_rate": 5.030492490345484e-08,
      "loss": 11.4879,
      "step": 22648
    },
    {
      "epoch": 3.9604808743169397,
      "grad_norm": 0.6809502184746479,
      "learning_rate": 4.98568136001909e-08,
      "loss": 11.5957,
      "step": 22649
    },
    {
      "epoch": 3.960655737704918,
      "grad_norm": 0.6376508305208067,
      "learning_rate": 4.9410706604191556e-08,
      "loss": 11.5649,
      "step": 22650
    },
    {
      "epoch": 3.960830601092896,
      "grad_norm": 0.6422887240420355,
      "learning_rate": 4.8966603924394114e-08,
      "loss": 11.4146,
      "step": 22651
    },
    {
      "epoch": 3.9610054644808743,
      "grad_norm": 0.645462875927056,
      "learning_rate": 4.852450556970256e-08,
      "loss": 11.4989,
      "step": 22652
    },
    {
      "epoch": 3.9611803278688527,
      "grad_norm": 0.731545800415375,
      "learning_rate": 4.8084411548987574e-08,
      "loss": 11.1777,
      "step": 22653
    },
    {
      "epoch": 3.9613551912568306,
      "grad_norm": 0.655176804570541,
      "learning_rate": 4.764632187107543e-08,
      "loss": 11.3449,
      "step": 22654
    },
    {
      "epoch": 3.961530054644809,
      "grad_norm": 0.6648990178194288,
      "learning_rate": 4.7210236544747984e-08,
      "loss": 11.1079,
      "step": 22655
    },
    {
      "epoch": 3.961704918032787,
      "grad_norm": 0.7018362007903648,
      "learning_rate": 4.677615557874271e-08,
      "loss": 11.6166,
      "step": 22656
    },
    {
      "epoch": 3.961879781420765,
      "grad_norm": 0.6679943634403268,
      "learning_rate": 4.634407898178594e-08,
      "loss": 11.0721,
      "step": 22657
    },
    {
      "epoch": 3.962054644808743,
      "grad_norm": 0.8273438256470088,
      "learning_rate": 4.5914006762515224e-08,
      "loss": 11.4172,
      "step": 22658
    },
    {
      "epoch": 3.9622295081967214,
      "grad_norm": 0.6879101672898776,
      "learning_rate": 4.5485938929579194e-08,
      "loss": 11.5266,
      "step": 22659
    },
    {
      "epoch": 3.9624043715846993,
      "grad_norm": 0.6673922053814901,
      "learning_rate": 4.505987549153767e-08,
      "loss": 11.3848,
      "step": 22660
    },
    {
      "epoch": 3.9625792349726776,
      "grad_norm": 0.6512128419696749,
      "learning_rate": 4.4635816456961574e-08,
      "loss": 11.629,
      "step": 22661
    },
    {
      "epoch": 3.962754098360656,
      "grad_norm": 0.6970348243547829,
      "learning_rate": 4.421376183433301e-08,
      "loss": 11.4464,
      "step": 22662
    },
    {
      "epoch": 3.962928961748634,
      "grad_norm": 0.6077351753154495,
      "learning_rate": 4.379371163212298e-08,
      "loss": 11.5215,
      "step": 22663
    },
    {
      "epoch": 3.9631038251366117,
      "grad_norm": 0.6525956314772843,
      "learning_rate": 4.3375665858758073e-08,
      "loss": 11.478,
      "step": 22664
    },
    {
      "epoch": 3.96327868852459,
      "grad_norm": 0.6621172278988166,
      "learning_rate": 4.295962452262048e-08,
      "loss": 11.3396,
      "step": 22665
    },
    {
      "epoch": 3.9634535519125684,
      "grad_norm": 0.6719546883696965,
      "learning_rate": 4.254558763203687e-08,
      "loss": 11.2534,
      "step": 22666
    },
    {
      "epoch": 3.9636284153005463,
      "grad_norm": 0.6646726681348152,
      "learning_rate": 4.213355519533391e-08,
      "loss": 11.3428,
      "step": 22667
    },
    {
      "epoch": 3.9638032786885247,
      "grad_norm": 0.6416917132314481,
      "learning_rate": 4.172352722077166e-08,
      "loss": 11.5537,
      "step": 22668
    },
    {
      "epoch": 3.9639781420765026,
      "grad_norm": 0.6399053841650046,
      "learning_rate": 4.131550371655468e-08,
      "loss": 11.4564,
      "step": 22669
    },
    {
      "epoch": 3.964153005464481,
      "grad_norm": 0.6737883523554219,
      "learning_rate": 4.090948469087641e-08,
      "loss": 11.2026,
      "step": 22670
    },
    {
      "epoch": 3.9643278688524592,
      "grad_norm": 0.6796912286268251,
      "learning_rate": 4.050547015188588e-08,
      "loss": 11.2805,
      "step": 22671
    },
    {
      "epoch": 3.964502732240437,
      "grad_norm": 0.662742813854586,
      "learning_rate": 4.0103460107665524e-08,
      "loss": 11.3491,
      "step": 22672
    },
    {
      "epoch": 3.964677595628415,
      "grad_norm": 0.6699626994815034,
      "learning_rate": 3.970345456629776e-08,
      "loss": 11.5248,
      "step": 22673
    },
    {
      "epoch": 3.9648524590163934,
      "grad_norm": 0.7048997294140177,
      "learning_rate": 3.93054535357873e-08,
      "loss": 11.3395,
      "step": 22674
    },
    {
      "epoch": 3.9650273224043717,
      "grad_norm": 0.6893979911561214,
      "learning_rate": 3.8909457024116634e-08,
      "loss": 11.4534,
      "step": 22675
    },
    {
      "epoch": 3.9652021857923496,
      "grad_norm": 0.6487259150275367,
      "learning_rate": 3.851546503924608e-08,
      "loss": 11.3556,
      "step": 22676
    },
    {
      "epoch": 3.965377049180328,
      "grad_norm": 0.6582874702016313,
      "learning_rate": 3.8123477589047106e-08,
      "loss": 11.4702,
      "step": 22677
    },
    {
      "epoch": 3.965551912568306,
      "grad_norm": 0.7132409767239308,
      "learning_rate": 3.7733494681413405e-08,
      "loss": 11.3995,
      "step": 22678
    },
    {
      "epoch": 3.965726775956284,
      "grad_norm": 0.6733375888453254,
      "learning_rate": 3.7345516324138744e-08,
      "loss": 11.4703,
      "step": 22679
    },
    {
      "epoch": 3.9659016393442625,
      "grad_norm": 0.6811889158057184,
      "learning_rate": 3.695954252500577e-08,
      "loss": 11.6223,
      "step": 22680
    },
    {
      "epoch": 3.9660765027322404,
      "grad_norm": 0.603208946361701,
      "learning_rate": 3.657557329177497e-08,
      "loss": 11.6555,
      "step": 22681
    },
    {
      "epoch": 3.9662513661202183,
      "grad_norm": 0.6654101111239248,
      "learning_rate": 3.619360863211796e-08,
      "loss": 11.4837,
      "step": 22682
    },
    {
      "epoch": 3.9664262295081967,
      "grad_norm": 0.6869761171403663,
      "learning_rate": 3.5813648553717494e-08,
      "loss": 11.4622,
      "step": 22683
    },
    {
      "epoch": 3.966601092896175,
      "grad_norm": 0.7541525445456814,
      "learning_rate": 3.54356930641786e-08,
      "loss": 11.3295,
      "step": 22684
    },
    {
      "epoch": 3.966775956284153,
      "grad_norm": 0.6736183346851371,
      "learning_rate": 3.50597421710841e-08,
      "loss": 11.3413,
      "step": 22685
    },
    {
      "epoch": 3.9669508196721313,
      "grad_norm": 0.7083620362859288,
      "learning_rate": 3.4685795881983506e-08,
      "loss": 11.1964,
      "step": 22686
    },
    {
      "epoch": 3.967125683060109,
      "grad_norm": 0.6084382527311782,
      "learning_rate": 3.431385420435973e-08,
      "loss": 11.4656,
      "step": 22687
    },
    {
      "epoch": 3.9673005464480875,
      "grad_norm": 0.6204709869325564,
      "learning_rate": 3.394391714567347e-08,
      "loss": 11.3679,
      "step": 22688
    },
    {
      "epoch": 3.967475409836066,
      "grad_norm": 0.6549990058790447,
      "learning_rate": 3.3575984713363204e-08,
      "loss": 11.561,
      "step": 22689
    },
    {
      "epoch": 3.9676502732240437,
      "grad_norm": 0.6760640503340284,
      "learning_rate": 3.321005691477863e-08,
      "loss": 11.379,
      "step": 22690
    },
    {
      "epoch": 3.9678251366120216,
      "grad_norm": 0.6111840728049629,
      "learning_rate": 3.284613375726941e-08,
      "loss": 11.5754,
      "step": 22691
    },
    {
      "epoch": 3.968,
      "grad_norm": 0.6791594855913015,
      "learning_rate": 3.2484215248140824e-08,
      "loss": 11.2964,
      "step": 22692
    },
    {
      "epoch": 3.9681748633879783,
      "grad_norm": 0.6710288094893128,
      "learning_rate": 3.212430139464262e-08,
      "loss": 11.3978,
      "step": 22693
    },
    {
      "epoch": 3.968349726775956,
      "grad_norm": 0.6363378224310385,
      "learning_rate": 3.176639220399125e-08,
      "loss": 11.4815,
      "step": 22694
    },
    {
      "epoch": 3.9685245901639346,
      "grad_norm": 0.6856516088926804,
      "learning_rate": 3.141048768336985e-08,
      "loss": 11.5362,
      "step": 22695
    },
    {
      "epoch": 3.9686994535519124,
      "grad_norm": 0.6618783508818119,
      "learning_rate": 3.1056587839917165e-08,
      "loss": 11.3611,
      "step": 22696
    },
    {
      "epoch": 3.968874316939891,
      "grad_norm": 0.6771999284024536,
      "learning_rate": 3.0704692680716404e-08,
      "loss": 11.4366,
      "step": 22697
    },
    {
      "epoch": 3.9690491803278687,
      "grad_norm": 0.6402665971725869,
      "learning_rate": 3.0354802212839705e-08,
      "loss": 11.2574,
      "step": 22698
    },
    {
      "epoch": 3.969224043715847,
      "grad_norm": 0.7694072927589877,
      "learning_rate": 3.000691644329257e-08,
      "loss": 11.2498,
      "step": 22699
    },
    {
      "epoch": 3.969398907103825,
      "grad_norm": 0.6819535542489976,
      "learning_rate": 2.966103537905829e-08,
      "loss": 11.3645,
      "step": 22700
    },
    {
      "epoch": 3.9695737704918033,
      "grad_norm": 0.6153178957447443,
      "learning_rate": 2.9317159027064666e-08,
      "loss": 11.4677,
      "step": 22701
    },
    {
      "epoch": 3.9697486338797816,
      "grad_norm": 0.7247279414563349,
      "learning_rate": 2.8975287394206184e-08,
      "loss": 11.5092,
      "step": 22702
    },
    {
      "epoch": 3.9699234972677595,
      "grad_norm": 0.6796423806039962,
      "learning_rate": 2.8635420487355126e-08,
      "loss": 11.2565,
      "step": 22703
    },
    {
      "epoch": 3.970098360655738,
      "grad_norm": 0.7028774081541215,
      "learning_rate": 2.829755831331715e-08,
      "loss": 11.5041,
      "step": 22704
    },
    {
      "epoch": 3.9702732240437157,
      "grad_norm": 0.6963404546500237,
      "learning_rate": 2.7961700878853526e-08,
      "loss": 11.1373,
      "step": 22705
    },
    {
      "epoch": 3.970448087431694,
      "grad_norm": 0.6413432963352449,
      "learning_rate": 2.7627848190725504e-08,
      "loss": 11.2961,
      "step": 22706
    },
    {
      "epoch": 3.970622950819672,
      "grad_norm": 0.7073188348447952,
      "learning_rate": 2.729600025561663e-08,
      "loss": 11.0464,
      "step": 22707
    },
    {
      "epoch": 3.9707978142076503,
      "grad_norm": 0.6625737917279955,
      "learning_rate": 2.6966157080166032e-08,
      "loss": 11.4617,
      "step": 22708
    },
    {
      "epoch": 3.970972677595628,
      "grad_norm": 0.6205526439488422,
      "learning_rate": 2.6638318671012852e-08,
      "loss": 11.539,
      "step": 22709
    },
    {
      "epoch": 3.9711475409836066,
      "grad_norm": 0.7351072226066351,
      "learning_rate": 2.6312485034718504e-08,
      "loss": 11.2427,
      "step": 22710
    },
    {
      "epoch": 3.971322404371585,
      "grad_norm": 0.6243351327501022,
      "learning_rate": 2.59886561778111e-08,
      "loss": 11.3406,
      "step": 22711
    },
    {
      "epoch": 3.971497267759563,
      "grad_norm": 0.6930976430362683,
      "learning_rate": 2.5666832106796545e-08,
      "loss": 11.3355,
      "step": 22712
    },
    {
      "epoch": 3.971672131147541,
      "grad_norm": 0.7006751205561881,
      "learning_rate": 2.534701282812524e-08,
      "loss": 11.5057,
      "step": 22713
    },
    {
      "epoch": 3.971846994535519,
      "grad_norm": 0.6495223684906966,
      "learning_rate": 2.5029198348203163e-08,
      "loss": 11.1644,
      "step": 22714
    },
    {
      "epoch": 3.9720218579234974,
      "grad_norm": 0.6616368053302555,
      "learning_rate": 2.4713388673414105e-08,
      "loss": 11.4577,
      "step": 22715
    },
    {
      "epoch": 3.9721967213114753,
      "grad_norm": 0.6947431939782485,
      "learning_rate": 2.4399583810086336e-08,
      "loss": 11.4374,
      "step": 22716
    },
    {
      "epoch": 3.9723715846994536,
      "grad_norm": 0.6534290442613322,
      "learning_rate": 2.4087783764503712e-08,
      "loss": 11.5812,
      "step": 22717
    },
    {
      "epoch": 3.9725464480874315,
      "grad_norm": 0.7061690575429476,
      "learning_rate": 2.3777988542939e-08,
      "loss": 11.3863,
      "step": 22718
    },
    {
      "epoch": 3.97272131147541,
      "grad_norm": 0.7320337913501055,
      "learning_rate": 2.347019815158724e-08,
      "loss": 11.1337,
      "step": 22719
    },
    {
      "epoch": 3.972896174863388,
      "grad_norm": 0.612516709821532,
      "learning_rate": 2.316441259663238e-08,
      "loss": 11.5988,
      "step": 22720
    },
    {
      "epoch": 3.973071038251366,
      "grad_norm": 0.6028509461054515,
      "learning_rate": 2.286063188419174e-08,
      "loss": 11.5613,
      "step": 22721
    },
    {
      "epoch": 3.973245901639344,
      "grad_norm": 0.6859821307963988,
      "learning_rate": 2.2558856020371553e-08,
      "loss": 11.2596,
      "step": 22722
    },
    {
      "epoch": 3.9734207650273223,
      "grad_norm": 0.6183843728308986,
      "learning_rate": 2.2259085011211432e-08,
      "loss": 11.6798,
      "step": 22723
    },
    {
      "epoch": 3.9735956284153007,
      "grad_norm": 0.6450530381346626,
      "learning_rate": 2.1961318862728785e-08,
      "loss": 11.3205,
      "step": 22724
    },
    {
      "epoch": 3.9737704918032786,
      "grad_norm": 0.5889277800006111,
      "learning_rate": 2.1665557580896613e-08,
      "loss": 11.4021,
      "step": 22725
    },
    {
      "epoch": 3.973945355191257,
      "grad_norm": 0.6764668556718736,
      "learning_rate": 2.1371801171654605e-08,
      "loss": 11.4734,
      "step": 22726
    },
    {
      "epoch": 3.974120218579235,
      "grad_norm": 0.6539051237458867,
      "learning_rate": 2.108004964086474e-08,
      "loss": 11.3893,
      "step": 22727
    },
    {
      "epoch": 3.974295081967213,
      "grad_norm": 0.6817463157666717,
      "learning_rate": 2.0790302994411204e-08,
      "loss": 11.1622,
      "step": 22728
    },
    {
      "epoch": 3.9744699453551915,
      "grad_norm": 0.7752048524394048,
      "learning_rate": 2.0502561238078254e-08,
      "loss": 11.2418,
      "step": 22729
    },
    {
      "epoch": 3.9746448087431694,
      "grad_norm": 0.7530162680287563,
      "learning_rate": 2.0216824377650157e-08,
      "loss": 11.4082,
      "step": 22730
    },
    {
      "epoch": 3.9748196721311473,
      "grad_norm": 0.6753338719862891,
      "learning_rate": 1.9933092418844558e-08,
      "loss": 11.0933,
      "step": 22731
    },
    {
      "epoch": 3.9749945355191256,
      "grad_norm": 0.6271840079183097,
      "learning_rate": 1.9651365367368002e-08,
      "loss": 11.5155,
      "step": 22732
    },
    {
      "epoch": 3.975169398907104,
      "grad_norm": 0.6438141136839678,
      "learning_rate": 1.9371643228871527e-08,
      "loss": 11.5709,
      "step": 22733
    },
    {
      "epoch": 3.975344262295082,
      "grad_norm": 0.6622842724587449,
      "learning_rate": 1.9093926008939554e-08,
      "loss": 11.195,
      "step": 22734
    },
    {
      "epoch": 3.97551912568306,
      "grad_norm": 0.6403131092667701,
      "learning_rate": 1.8818213713156507e-08,
      "loss": 11.5382,
      "step": 22735
    },
    {
      "epoch": 3.975693989071038,
      "grad_norm": 0.720625552085523,
      "learning_rate": 1.8544506347051295e-08,
      "loss": 11.2748,
      "step": 22736
    },
    {
      "epoch": 3.9758688524590164,
      "grad_norm": 0.6401557928793236,
      "learning_rate": 1.827280391611952e-08,
      "loss": 11.1315,
      "step": 22737
    },
    {
      "epoch": 3.976043715846995,
      "grad_norm": 0.6546688973291722,
      "learning_rate": 1.8003106425801275e-08,
      "loss": 10.9609,
      "step": 22738
    },
    {
      "epoch": 3.9762185792349727,
      "grad_norm": 0.6520279184860727,
      "learning_rate": 1.773541388150335e-08,
      "loss": 11.2848,
      "step": 22739
    },
    {
      "epoch": 3.9763934426229506,
      "grad_norm": 0.6415620133422613,
      "learning_rate": 1.7469726288588118e-08,
      "loss": 11.2872,
      "step": 22740
    },
    {
      "epoch": 3.976568306010929,
      "grad_norm": 0.6892769280471069,
      "learning_rate": 1.7206043652406854e-08,
      "loss": 11.2886,
      "step": 22741
    },
    {
      "epoch": 3.9767431693989073,
      "grad_norm": 0.6389421769013678,
      "learning_rate": 1.6944365978222022e-08,
      "loss": 11.1656,
      "step": 22742
    },
    {
      "epoch": 3.976918032786885,
      "grad_norm": 0.6711779777518236,
      "learning_rate": 1.6684693271296072e-08,
      "loss": 11.5248,
      "step": 22743
    },
    {
      "epoch": 3.9770928961748635,
      "grad_norm": 0.6418325008771739,
      "learning_rate": 1.642702553682485e-08,
      "loss": 11.4386,
      "step": 22744
    },
    {
      "epoch": 3.9772677595628414,
      "grad_norm": 0.6519831578247122,
      "learning_rate": 1.617136277998199e-08,
      "loss": 11.4905,
      "step": 22745
    },
    {
      "epoch": 3.9774426229508197,
      "grad_norm": 0.6706637643099795,
      "learning_rate": 1.591770500589673e-08,
      "loss": 11.4338,
      "step": 22746
    },
    {
      "epoch": 3.977617486338798,
      "grad_norm": 0.6619108712899536,
      "learning_rate": 1.5666052219653894e-08,
      "loss": 11.5234,
      "step": 22747
    },
    {
      "epoch": 3.977792349726776,
      "grad_norm": 0.6466984697669518,
      "learning_rate": 1.5416404426293885e-08,
      "loss": 11.1895,
      "step": 22748
    },
    {
      "epoch": 3.977967213114754,
      "grad_norm": 0.6235332186773374,
      "learning_rate": 1.5168761630834916e-08,
      "loss": 11.4449,
      "step": 22749
    },
    {
      "epoch": 3.978142076502732,
      "grad_norm": 0.7701262136048352,
      "learning_rate": 1.4923123838228583e-08,
      "loss": 11.3055,
      "step": 22750
    },
    {
      "epoch": 3.9783169398907106,
      "grad_norm": 0.6458907196697019,
      "learning_rate": 1.4679491053415373e-08,
      "loss": 11.3222,
      "step": 22751
    },
    {
      "epoch": 3.9784918032786885,
      "grad_norm": 0.6180503640825193,
      "learning_rate": 1.4437863281269171e-08,
      "loss": 11.2421,
      "step": 22752
    },
    {
      "epoch": 3.978666666666667,
      "grad_norm": 0.581639791876392,
      "learning_rate": 1.419824052664165e-08,
      "loss": 11.3702,
      "step": 22753
    },
    {
      "epoch": 3.9788415300546447,
      "grad_norm": 0.6772238767719535,
      "learning_rate": 1.3960622794340072e-08,
      "loss": 11.334,
      "step": 22754
    },
    {
      "epoch": 3.979016393442623,
      "grad_norm": 0.6802538869253958,
      "learning_rate": 1.3725010089116198e-08,
      "loss": 11.1467,
      "step": 22755
    },
    {
      "epoch": 3.979191256830601,
      "grad_norm": 0.6557482984792957,
      "learning_rate": 1.3491402415710675e-08,
      "loss": 11.6034,
      "step": 22756
    },
    {
      "epoch": 3.9793661202185793,
      "grad_norm": 0.7298965507040411,
      "learning_rate": 1.3259799778808645e-08,
      "loss": 11.241,
      "step": 22757
    },
    {
      "epoch": 3.979540983606557,
      "grad_norm": 0.6758788693143537,
      "learning_rate": 1.3030202183039742e-08,
      "loss": 11.1535,
      "step": 22758
    },
    {
      "epoch": 3.9797158469945355,
      "grad_norm": 0.6414041286738466,
      "learning_rate": 1.2802609633011387e-08,
      "loss": 11.4078,
      "step": 22759
    },
    {
      "epoch": 3.979890710382514,
      "grad_norm": 0.7136287984641596,
      "learning_rate": 1.25770221332977e-08,
      "loss": 11.3403,
      "step": 22760
    },
    {
      "epoch": 3.9800655737704917,
      "grad_norm": 0.7554223787645373,
      "learning_rate": 1.2353439688406188e-08,
      "loss": 11.525,
      "step": 22761
    },
    {
      "epoch": 3.98024043715847,
      "grad_norm": 0.7121328904984415,
      "learning_rate": 1.2131862302844355e-08,
      "loss": 11.4325,
      "step": 22762
    },
    {
      "epoch": 3.980415300546448,
      "grad_norm": 0.6749758603470037,
      "learning_rate": 1.1912289981030888e-08,
      "loss": 11.3572,
      "step": 22763
    },
    {
      "epoch": 3.9805901639344263,
      "grad_norm": 0.6913386920932453,
      "learning_rate": 1.1694722727384477e-08,
      "loss": 11.3263,
      "step": 22764
    },
    {
      "epoch": 3.9807650273224042,
      "grad_norm": 0.6692226336310987,
      "learning_rate": 1.1479160546268298e-08,
      "loss": 11.563,
      "step": 22765
    },
    {
      "epoch": 3.9809398907103826,
      "grad_norm": 0.7190259792963483,
      "learning_rate": 1.1265603441990013e-08,
      "loss": 11.3458,
      "step": 22766
    },
    {
      "epoch": 3.9811147540983605,
      "grad_norm": 0.616581490875168,
      "learning_rate": 1.105405141884619e-08,
      "loss": 11.4037,
      "step": 22767
    },
    {
      "epoch": 3.981289617486339,
      "grad_norm": 0.671404695785203,
      "learning_rate": 1.0844504481077878e-08,
      "loss": 11.0138,
      "step": 22768
    },
    {
      "epoch": 3.981464480874317,
      "grad_norm": 0.7612924953744271,
      "learning_rate": 1.0636962632881719e-08,
      "loss": 11.4395,
      "step": 22769
    },
    {
      "epoch": 3.981639344262295,
      "grad_norm": 0.6948945434283168,
      "learning_rate": 1.0431425878421053e-08,
      "loss": 11.516,
      "step": 22770
    },
    {
      "epoch": 3.9818142076502734,
      "grad_norm": 0.6446436074548861,
      "learning_rate": 1.0227894221825907e-08,
      "loss": 11.2389,
      "step": 22771
    },
    {
      "epoch": 3.9819890710382513,
      "grad_norm": 0.6899748306808133,
      "learning_rate": 1.0026367667159697e-08,
      "loss": 11.6289,
      "step": 22772
    },
    {
      "epoch": 3.9821639344262296,
      "grad_norm": 0.6008203506999583,
      "learning_rate": 9.82684621847474e-09,
      "loss": 11.5027,
      "step": 22773
    },
    {
      "epoch": 3.9823387978142075,
      "grad_norm": 0.6803327606776063,
      "learning_rate": 9.629329879778937e-09,
      "loss": 11.5133,
      "step": 22774
    },
    {
      "epoch": 3.982513661202186,
      "grad_norm": 0.6764936019585142,
      "learning_rate": 9.433818655024685e-09,
      "loss": 11.5357,
      "step": 22775
    },
    {
      "epoch": 3.9826885245901638,
      "grad_norm": 0.6778435273198177,
      "learning_rate": 9.24031254813107e-09,
      "loss": 11.6756,
      "step": 22776
    },
    {
      "epoch": 3.982863387978142,
      "grad_norm": 0.7132804296388547,
      "learning_rate": 9.048811562983873e-09,
      "loss": 11.5207,
      "step": 22777
    },
    {
      "epoch": 3.9830382513661204,
      "grad_norm": 0.6181754232567492,
      "learning_rate": 8.859315703413362e-09,
      "loss": 11.1969,
      "step": 22778
    },
    {
      "epoch": 3.9832131147540983,
      "grad_norm": 0.6850783228043112,
      "learning_rate": 8.671824973238707e-09,
      "loss": 11.4488,
      "step": 22779
    },
    {
      "epoch": 3.9833879781420762,
      "grad_norm": 0.6736689015996812,
      "learning_rate": 8.486339376201358e-09,
      "loss": 11.4193,
      "step": 22780
    },
    {
      "epoch": 3.9835628415300546,
      "grad_norm": 0.7150395138158544,
      "learning_rate": 8.302858916020562e-09,
      "loss": 11.4186,
      "step": 22781
    },
    {
      "epoch": 3.983737704918033,
      "grad_norm": 0.6948371785616146,
      "learning_rate": 8.121383596393362e-09,
      "loss": 11.5784,
      "step": 22782
    },
    {
      "epoch": 3.983912568306011,
      "grad_norm": 0.640279892223364,
      "learning_rate": 7.941913420939084e-09,
      "loss": 11.1294,
      "step": 22783
    },
    {
      "epoch": 3.984087431693989,
      "grad_norm": 0.6848773814716924,
      "learning_rate": 7.764448393277058e-09,
      "loss": 11.5424,
      "step": 22784
    },
    {
      "epoch": 3.984262295081967,
      "grad_norm": 0.659244483579623,
      "learning_rate": 7.58898851693779e-09,
      "loss": 11.5086,
      "step": 22785
    },
    {
      "epoch": 3.9844371584699454,
      "grad_norm": 0.6463421581386738,
      "learning_rate": 7.415533795473995e-09,
      "loss": 11.5524,
      "step": 22786
    },
    {
      "epoch": 3.9846120218579237,
      "grad_norm": 0.6742649347156322,
      "learning_rate": 7.244084232338466e-09,
      "loss": 11.5214,
      "step": 22787
    },
    {
      "epoch": 3.9847868852459016,
      "grad_norm": 0.649051918042355,
      "learning_rate": 7.074639830972895e-09,
      "loss": 11.5364,
      "step": 22788
    },
    {
      "epoch": 3.9849617486338795,
      "grad_norm": 0.630503376205264,
      "learning_rate": 6.907200594785667e-09,
      "loss": 11.5294,
      "step": 22789
    },
    {
      "epoch": 3.985136612021858,
      "grad_norm": 0.657589070214603,
      "learning_rate": 6.741766527118554e-09,
      "loss": 11.2259,
      "step": 22790
    },
    {
      "epoch": 3.985311475409836,
      "grad_norm": 0.6915842961441149,
      "learning_rate": 6.578337631313325e-09,
      "loss": 10.9295,
      "step": 22791
    },
    {
      "epoch": 3.985486338797814,
      "grad_norm": 0.6903995718080916,
      "learning_rate": 6.416913910622935e-09,
      "loss": 11.6645,
      "step": 22792
    },
    {
      "epoch": 3.9856612021857925,
      "grad_norm": 0.6804147401444919,
      "learning_rate": 6.257495368289235e-09,
      "loss": 11.3755,
      "step": 22793
    },
    {
      "epoch": 3.9858360655737703,
      "grad_norm": 0.6694748628802878,
      "learning_rate": 6.100082007520769e-09,
      "loss": 11.4856,
      "step": 22794
    },
    {
      "epoch": 3.9860109289617487,
      "grad_norm": 0.6641834295781592,
      "learning_rate": 5.944673831470571e-09,
      "loss": 11.4688,
      "step": 22795
    },
    {
      "epoch": 3.986185792349727,
      "grad_norm": 0.6273528298574212,
      "learning_rate": 5.7912708432472654e-09,
      "loss": 11.402,
      "step": 22796
    },
    {
      "epoch": 3.986360655737705,
      "grad_norm": 0.6801729827104028,
      "learning_rate": 5.6398730459372714e-09,
      "loss": 11.5125,
      "step": 22797
    },
    {
      "epoch": 3.986535519125683,
      "grad_norm": 0.7162835988278989,
      "learning_rate": 5.490480442560397e-09,
      "loss": 11.4985,
      "step": 22798
    },
    {
      "epoch": 3.986710382513661,
      "grad_norm": 0.7293741227717059,
      "learning_rate": 5.3430930361253465e-09,
      "loss": 11.5423,
      "step": 22799
    },
    {
      "epoch": 3.9868852459016395,
      "grad_norm": 0.6452310444755694,
      "learning_rate": 5.197710829596414e-09,
      "loss": 11.3675,
      "step": 22800
    },
    {
      "epoch": 3.9870601092896174,
      "grad_norm": 0.6524320474613617,
      "learning_rate": 5.05433382586018e-09,
      "loss": 11.3269,
      "step": 22801
    },
    {
      "epoch": 3.9872349726775957,
      "grad_norm": 0.7122413244633778,
      "learning_rate": 4.9129620278254295e-09,
      "loss": 11.6512,
      "step": 22802
    },
    {
      "epoch": 3.9874098360655736,
      "grad_norm": 0.6734917604632741,
      "learning_rate": 4.773595438301026e-09,
      "loss": 11.4293,
      "step": 22803
    },
    {
      "epoch": 3.987584699453552,
      "grad_norm": 0.7089882536293683,
      "learning_rate": 4.636234060095834e-09,
      "loss": 11.0511,
      "step": 22804
    },
    {
      "epoch": 3.9877595628415303,
      "grad_norm": 0.6575309416658173,
      "learning_rate": 4.500877895963207e-09,
      "loss": 11.6424,
      "step": 22805
    },
    {
      "epoch": 3.9879344262295082,
      "grad_norm": 0.6583871321325614,
      "learning_rate": 4.3675269486009865e-09,
      "loss": 11.472,
      "step": 22806
    },
    {
      "epoch": 3.988109289617486,
      "grad_norm": 0.6265724361636235,
      "learning_rate": 4.236181220707014e-09,
      "loss": 11.5026,
      "step": 22807
    },
    {
      "epoch": 3.9882841530054645,
      "grad_norm": 0.6648106577477387,
      "learning_rate": 4.1068407149125185e-09,
      "loss": 11.3915,
      "step": 22808
    },
    {
      "epoch": 3.988459016393443,
      "grad_norm": 0.6209615310660325,
      "learning_rate": 3.9795054337932184e-09,
      "loss": 11.374,
      "step": 22809
    },
    {
      "epoch": 3.9886338797814207,
      "grad_norm": 0.6359063916833962,
      "learning_rate": 3.854175379913727e-09,
      "loss": 11.6851,
      "step": 22810
    },
    {
      "epoch": 3.988808743169399,
      "grad_norm": 0.6677595144898092,
      "learning_rate": 3.73085055578315e-09,
      "loss": 11.1693,
      "step": 22811
    },
    {
      "epoch": 3.988983606557377,
      "grad_norm": 0.6531454121183928,
      "learning_rate": 3.6095309638883856e-09,
      "loss": 11.5367,
      "step": 22812
    },
    {
      "epoch": 3.9891584699453553,
      "grad_norm": 0.7346032853478711,
      "learning_rate": 3.490216606649721e-09,
      "loss": 11.3385,
      "step": 22813
    },
    {
      "epoch": 3.989333333333333,
      "grad_norm": 0.7022589368228603,
      "learning_rate": 3.3729074864541355e-09,
      "loss": 11.2217,
      "step": 22814
    },
    {
      "epoch": 3.9895081967213115,
      "grad_norm": 0.6217047919712911,
      "learning_rate": 3.257603605666404e-09,
      "loss": 11.462,
      "step": 22815
    },
    {
      "epoch": 3.9896830601092894,
      "grad_norm": 0.9539706910691761,
      "learning_rate": 3.1443049665957903e-09,
      "loss": 11.3403,
      "step": 22816
    },
    {
      "epoch": 3.9898579234972678,
      "grad_norm": 0.6644116329558342,
      "learning_rate": 3.0330115715071494e-09,
      "loss": 11.3362,
      "step": 22817
    },
    {
      "epoch": 3.990032786885246,
      "grad_norm": 0.6961048789439079,
      "learning_rate": 2.9237234226431322e-09,
      "loss": 11.356,
      "step": 22818
    },
    {
      "epoch": 3.990207650273224,
      "grad_norm": 0.6508575442530864,
      "learning_rate": 2.8164405221797753e-09,
      "loss": 11.4439,
      "step": 22819
    },
    {
      "epoch": 3.9903825136612023,
      "grad_norm": 0.6950832396229895,
      "learning_rate": 2.711162872293116e-09,
      "loss": 11.0935,
      "step": 22820
    },
    {
      "epoch": 3.9905573770491802,
      "grad_norm": 0.6513366778597759,
      "learning_rate": 2.6078904750703736e-09,
      "loss": 11.2167,
      "step": 22821
    },
    {
      "epoch": 3.9907322404371586,
      "grad_norm": 0.6848137017247592,
      "learning_rate": 2.5066233325876653e-09,
      "loss": 11.4585,
      "step": 22822
    },
    {
      "epoch": 3.9909071038251365,
      "grad_norm": 0.6502333651317911,
      "learning_rate": 2.4073614468878015e-09,
      "loss": 11.5294,
      "step": 22823
    },
    {
      "epoch": 3.991081967213115,
      "grad_norm": 0.7051879922910426,
      "learning_rate": 2.310104819958081e-09,
      "loss": 11.3459,
      "step": 22824
    },
    {
      "epoch": 3.9912568306010927,
      "grad_norm": 0.6818716707921545,
      "learning_rate": 2.2148534537302925e-09,
      "loss": 11.5479,
      "step": 22825
    },
    {
      "epoch": 3.991431693989071,
      "grad_norm": 0.7296444542268798,
      "learning_rate": 2.121607350136223e-09,
      "loss": 11.2666,
      "step": 22826
    },
    {
      "epoch": 3.9916065573770494,
      "grad_norm": 0.6474046860333569,
      "learning_rate": 2.0303665110410484e-09,
      "loss": 11.5462,
      "step": 22827
    },
    {
      "epoch": 3.9917814207650273,
      "grad_norm": 0.7023437983062784,
      "learning_rate": 1.941130938265534e-09,
      "loss": 11.5012,
      "step": 22828
    },
    {
      "epoch": 3.9919562841530056,
      "grad_norm": 0.692167238346074,
      "learning_rate": 1.853900633608241e-09,
      "loss": 11.5541,
      "step": 22829
    },
    {
      "epoch": 3.9921311475409835,
      "grad_norm": 0.627779032186257,
      "learning_rate": 1.7686755988122195e-09,
      "loss": 11.6598,
      "step": 22830
    },
    {
      "epoch": 3.992306010928962,
      "grad_norm": 0.684582260025964,
      "learning_rate": 1.685455835587213e-09,
      "loss": 11.1241,
      "step": 22831
    },
    {
      "epoch": 3.9924808743169398,
      "grad_norm": 0.6830948106092758,
      "learning_rate": 1.6042413456096583e-09,
      "loss": 11.4774,
      "step": 22832
    },
    {
      "epoch": 3.992655737704918,
      "grad_norm": 0.6842563483004267,
      "learning_rate": 1.5250321305004811e-09,
      "loss": 11.4344,
      "step": 22833
    },
    {
      "epoch": 3.992830601092896,
      "grad_norm": 0.6211993952864256,
      "learning_rate": 1.4478281918473002e-09,
      "loss": 11.4593,
      "step": 22834
    },
    {
      "epoch": 3.9930054644808743,
      "grad_norm": 0.6738860075041696,
      "learning_rate": 1.372629531204428e-09,
      "loss": 11.1779,
      "step": 22835
    },
    {
      "epoch": 3.9931803278688527,
      "grad_norm": 0.6553551524277402,
      "learning_rate": 1.2994361500706652e-09,
      "loss": 11.5034,
      "step": 22836
    },
    {
      "epoch": 3.9933551912568306,
      "grad_norm": 0.6408563468608286,
      "learning_rate": 1.228248049922609e-09,
      "loss": 11.5978,
      "step": 22837
    },
    {
      "epoch": 3.9935300546448085,
      "grad_norm": 0.6768280355147853,
      "learning_rate": 1.1590652321813444e-09,
      "loss": 11.2608,
      "step": 22838
    },
    {
      "epoch": 3.993704918032787,
      "grad_norm": 0.7308175093228921,
      "learning_rate": 1.0918876982457527e-09,
      "loss": 11.49,
      "step": 22839
    },
    {
      "epoch": 3.993879781420765,
      "grad_norm": 0.6975521715545159,
      "learning_rate": 1.0267154494481013e-09,
      "loss": 11.1519,
      "step": 22840
    },
    {
      "epoch": 3.994054644808743,
      "grad_norm": 0.6228587130120002,
      "learning_rate": 9.635484871095556e-10,
      "loss": 11.2949,
      "step": 22841
    },
    {
      "epoch": 3.9942295081967214,
      "grad_norm": 0.6696373987707476,
      "learning_rate": 9.023868124846679e-10,
      "loss": 11.55,
      "step": 22842
    },
    {
      "epoch": 3.9944043715846993,
      "grad_norm": 0.6245955352264669,
      "learning_rate": 8.432304268057856e-10,
      "loss": 11.5388,
      "step": 22843
    },
    {
      "epoch": 3.9945792349726776,
      "grad_norm": 0.6965472200237337,
      "learning_rate": 7.86079331249745e-10,
      "loss": 11.1854,
      "step": 22844
    },
    {
      "epoch": 3.994754098360656,
      "grad_norm": 0.7068324598366698,
      "learning_rate": 7.309335269822804e-10,
      "loss": 11.3492,
      "step": 22845
    },
    {
      "epoch": 3.994928961748634,
      "grad_norm": 0.6247273208776846,
      "learning_rate": 6.777930150914103e-10,
      "loss": 11.5654,
      "step": 22846
    },
    {
      "epoch": 3.9951038251366118,
      "grad_norm": 0.6769818735982115,
      "learning_rate": 6.266577966429488e-10,
      "loss": 11.3763,
      "step": 22847
    },
    {
      "epoch": 3.99527868852459,
      "grad_norm": 0.6946409698459607,
      "learning_rate": 5.775278726805055e-10,
      "loss": 11.3813,
      "step": 22848
    },
    {
      "epoch": 3.9954535519125685,
      "grad_norm": 0.7422509829499969,
      "learning_rate": 5.304032441699746e-10,
      "loss": 10.9096,
      "step": 22849
    },
    {
      "epoch": 3.9956284153005464,
      "grad_norm": 0.6957320778403383,
      "learning_rate": 4.852839120661479e-10,
      "loss": 11.2378,
      "step": 22850
    },
    {
      "epoch": 3.9958032786885247,
      "grad_norm": 0.7165659163676024,
      "learning_rate": 4.4216987726830584e-10,
      "loss": 11.2621,
      "step": 22851
    },
    {
      "epoch": 3.9959781420765026,
      "grad_norm": 0.6667212021612149,
      "learning_rate": 4.010611406535247e-10,
      "loss": 11.174,
      "step": 22852
    },
    {
      "epoch": 3.996153005464481,
      "grad_norm": 0.7199055511522345,
      "learning_rate": 3.6195770303226737e-10,
      "loss": 11.3048,
      "step": 22853
    },
    {
      "epoch": 3.9963278688524593,
      "grad_norm": 0.6869189117444142,
      "learning_rate": 3.2485956518168994e-10,
      "loss": 11.6488,
      "step": 22854
    },
    {
      "epoch": 3.996502732240437,
      "grad_norm": 0.6649796889841451,
      "learning_rate": 2.897667278678462e-10,
      "loss": 11.5899,
      "step": 22855
    },
    {
      "epoch": 3.996677595628415,
      "grad_norm": 0.6348568745549076,
      "learning_rate": 2.566791917790745e-10,
      "loss": 11.4482,
      "step": 22856
    },
    {
      "epoch": 3.9968524590163934,
      "grad_norm": 0.6359752099734777,
      "learning_rate": 2.255969575926109e-10,
      "loss": 11.4261,
      "step": 22857
    },
    {
      "epoch": 3.9970273224043718,
      "grad_norm": 0.623787935273858,
      "learning_rate": 1.965200259079758e-10,
      "loss": 11.3335,
      "step": 22858
    },
    {
      "epoch": 3.9972021857923496,
      "grad_norm": 0.7003721295994969,
      "learning_rate": 1.6944839733579187e-10,
      "loss": 11.5164,
      "step": 22859
    },
    {
      "epoch": 3.997377049180328,
      "grad_norm": 0.6858269707400423,
      "learning_rate": 1.4438207239786395e-10,
      "loss": 11.3003,
      "step": 22860
    },
    {
      "epoch": 3.997551912568306,
      "grad_norm": 0.642542436628355,
      "learning_rate": 1.2132105161599682e-10,
      "loss": 11.3299,
      "step": 22861
    },
    {
      "epoch": 3.9977267759562842,
      "grad_norm": 0.6492048373871087,
      "learning_rate": 1.002653354231775e-10,
      "loss": 11.2773,
      "step": 22862
    },
    {
      "epoch": 3.9979016393442626,
      "grad_norm": 0.7141181360810043,
      "learning_rate": 8.121492427459742e-11,
      "loss": 11.5087,
      "step": 22863
    },
    {
      "epoch": 3.9980765027322405,
      "grad_norm": 0.6789789330442131,
      "learning_rate": 6.416981853663017e-11,
      "loss": 11.1214,
      "step": 22864
    },
    {
      "epoch": 3.9982513661202184,
      "grad_norm": 0.6712231809290276,
      "learning_rate": 4.913001854234267e-11,
      "loss": 11.4711,
      "step": 22865
    },
    {
      "epoch": 3.9984262295081967,
      "grad_norm": 0.7180822277240849,
      "learning_rate": 3.609552461369958e-11,
      "loss": 11.2591,
      "step": 22866
    },
    {
      "epoch": 3.998601092896175,
      "grad_norm": 0.6422562586474928,
      "learning_rate": 2.506633699494998e-11,
      "loss": 11.6013,
      "step": 22867
    },
    {
      "epoch": 3.998775956284153,
      "grad_norm": 0.6881408798653273,
      "learning_rate": 1.6042455919240696e-11,
      "loss": 11.3828,
      "step": 22868
    },
    {
      "epoch": 3.9989508196721313,
      "grad_norm": 0.673408667151373,
      "learning_rate": 9.023881564207415e-12,
      "loss": 11.3157,
      "step": 22869
    },
    {
      "epoch": 3.999125683060109,
      "grad_norm": 0.8489602849863843,
      "learning_rate": 4.010614063076901e-12,
      "loss": 11.1751,
      "step": 22870
    },
    {
      "epoch": 3.9993005464480875,
      "grad_norm": 0.6714336959990947,
      "learning_rate": 1.0026535157692252e-12,
      "loss": 11.2946,
      "step": 22871
    },
    {
      "epoch": 3.9994754098360654,
      "grad_norm": 0.6802835190671789,
      "learning_rate": 0.0,
      "loss": 11.3049,
      "step": 22872
    },
    {
      "epoch": 3.9994754098360654,
      "step": 22872,
      "total_flos": 2281443255246848.0,
      "train_loss": 11.9832280325531,
      "train_runtime": 162912.3586,
      "train_samples_per_second": 17.973,
      "train_steps_per_second": 0.14
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 22872,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2281443255246848.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}