diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6683 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 88.78504672897196,
+  "eval_steps": 500,
+  "global_step": 9500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.09345794392523364,
+      "grad_norm": 11.088369369506836,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.7742,
+      "step": 10
+    },
+    {
+      "epoch": 0.18691588785046728,
+      "grad_norm": 11.834894180297852,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.6346,
+      "step": 20
+    },
+    {
+      "epoch": 0.2803738317757009,
+      "grad_norm": 7.324471473693848,
+      "learning_rate": 6e-06,
+      "loss": 1.3161,
+      "step": 30
+    },
+    {
+      "epoch": 0.37383177570093457,
+      "grad_norm": 2.9169681072235107,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6619,
+      "step": 40
+    },
+    {
+      "epoch": 0.4672897196261682,
+      "grad_norm": 2.7369236946105957,
+      "learning_rate": 1e-05,
+      "loss": 0.3784,
+      "step": 50
+    },
+    {
+      "epoch": 0.5607476635514018,
+      "grad_norm": 2.0095913410186768,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2843,
+      "step": 60
+    },
+    {
+      "epoch": 0.6542056074766355,
+      "grad_norm": 1.0399119853973389,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.2026,
+      "step": 70
+    },
+    {
+      "epoch": 0.7476635514018691,
+      "grad_norm": 1.0228025913238525,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.171,
+      "step": 80
+    },
+    {
+      "epoch": 0.8411214953271028,
+      "grad_norm": 1.065486192703247,
+      "learning_rate": 1.8e-05,
+      "loss": 0.143,
+      "step": 90
+    },
+    {
+      "epoch": 0.9345794392523364,
+      "grad_norm": 0.994318962097168,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 100
+    },
+    {
+      "epoch": 1.02803738317757,
+      "grad_norm": 1.089653730392456,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.1088,
+      "step": 110
+    },
+    {
+      "epoch": 1.1214953271028036,
+      "grad_norm": 0.7751863598823547,
+      "learning_rate": 2.4e-05,
+      "loss": 0.1128,
+      "step": 120
+    },
+    {
+      "epoch": 1.2149532710280373,
+      "grad_norm": 0.7448062896728516,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.0817,
+      "step": 130
+    },
+    {
+      "epoch": 1.308411214953271,
+      "grad_norm": 1.0464024543762207,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.0797,
+      "step": 140
+    },
+    {
+      "epoch": 1.4018691588785046,
+      "grad_norm": 0.945536196231842,
+      "learning_rate": 3e-05,
+      "loss": 0.0854,
+      "step": 150
+    },
+    {
+      "epoch": 1.4953271028037383,
+      "grad_norm": 0.4769207239151001,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0774,
+      "step": 160
+    },
+    {
+      "epoch": 1.588785046728972,
+      "grad_norm": 0.6780673265457153,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.0676,
+      "step": 170
+    },
+    {
+      "epoch": 1.6822429906542056,
+      "grad_norm": 0.4976266622543335,
+      "learning_rate": 3.6e-05,
+      "loss": 0.0605,
+      "step": 180
+    },
+    {
+      "epoch": 1.7757009345794392,
+      "grad_norm": 0.5387341380119324,
+      "learning_rate": 3.8e-05,
+      "loss": 0.0666,
+      "step": 190
+    },
+    {
+      "epoch": 1.8691588785046729,
+      "grad_norm": 1.1668950319290161,
+      "learning_rate": 4e-05,
+      "loss": 0.0668,
+      "step": 200
+    },
+    {
+      "epoch": 1.9626168224299065,
+      "grad_norm": 0.6982051730155945,
+      "learning_rate": 4.2e-05,
+      "loss": 0.0619,
+      "step": 210
+    },
+    {
+      "epoch": 2.05607476635514,
+      "grad_norm": 0.8520908355712891,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.0626,
+      "step": 220
+    },
+    {
+      "epoch": 2.149532710280374,
+      "grad_norm": 0.6880495548248291,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.0544,
+      "step": 230
+    },
+    {
+      "epoch": 2.2429906542056073,
+      "grad_norm": 1.1675457954406738,
+      "learning_rate": 4.8e-05,
+      "loss": 0.0597,
+      "step": 240
+    },
+    {
+      "epoch": 2.336448598130841,
+      "grad_norm": 0.6466752886772156,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "step": 250
+    },
+    {
+      "epoch": 2.4299065420560746,
+      "grad_norm": 0.8839955925941467,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.0574,
+      "step": 260
+    },
+    {
+      "epoch": 2.5233644859813085,
+      "grad_norm": 0.44215482473373413,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.0507,
+      "step": 270
+    },
+    {
+      "epoch": 2.616822429906542,
+      "grad_norm": 0.6018381118774414,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0602,
+      "step": 280
+    },
+    {
+      "epoch": 2.710280373831776,
+      "grad_norm": 0.5012953281402588,
+      "learning_rate": 5.8e-05,
+      "loss": 0.0475,
+      "step": 290
+    },
+    {
+      "epoch": 2.803738317757009,
+      "grad_norm": 0.5049079060554504,
+      "learning_rate": 6e-05,
+      "loss": 0.0543,
+      "step": 300
+    },
+    {
+      "epoch": 2.897196261682243,
+      "grad_norm": 0.6439036130905151,
+      "learning_rate": 6.2e-05,
+      "loss": 0.0481,
+      "step": 310
+    },
+    {
+      "epoch": 2.9906542056074765,
+      "grad_norm": 0.760713517665863,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.0483,
+      "step": 320
+    },
+    {
+      "epoch": 3.0841121495327104,
+      "grad_norm": 0.8276482224464417,
+      "learning_rate": 6.6e-05,
+      "loss": 0.0506,
+      "step": 330
+    },
+    {
+      "epoch": 3.177570093457944,
+      "grad_norm": 0.7145904898643494,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.039,
+      "step": 340
+    },
+    {
+      "epoch": 3.2710280373831777,
+      "grad_norm": 0.5840898752212524,
+      "learning_rate": 7e-05,
+      "loss": 0.0473,
+      "step": 350
+    },
+    {
+      "epoch": 3.364485981308411,
+      "grad_norm": 0.8430977463722229,
+      "learning_rate": 7.2e-05,
+      "loss": 0.0499,
+      "step": 360
+    },
+    {
+      "epoch": 3.457943925233645,
+      "grad_norm": 0.4334861636161804,
+      "learning_rate": 7.4e-05,
+      "loss": 0.0421,
+      "step": 370
+    },
+    {
+      "epoch": 3.5514018691588785,
+      "grad_norm": 0.40518927574157715,
+      "learning_rate": 7.6e-05,
+      "loss": 0.0467,
+      "step": 380
+    },
+    {
+      "epoch": 3.6448598130841123,
+      "grad_norm": 0.6229057908058167,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.055,
+      "step": 390
+    },
+    {
+      "epoch": 3.7383177570093458,
+      "grad_norm": 0.5108091831207275,
+      "learning_rate": 8e-05,
+      "loss": 0.0399,
+      "step": 400
+    },
+    {
+      "epoch": 3.831775700934579,
+      "grad_norm": 0.733472466468811,
+      "learning_rate": 8.2e-05,
+      "loss": 0.0441,
+      "step": 410
+    },
+    {
+      "epoch": 3.925233644859813,
+      "grad_norm": 0.8219009637832642,
+      "learning_rate": 8.4e-05,
+      "loss": 0.0491,
+      "step": 420
+    },
+    {
+      "epoch": 4.018691588785047,
+      "grad_norm": 0.779040515422821,
+      "learning_rate": 8.6e-05,
+      "loss": 0.0485,
+      "step": 430
+    },
+    {
+      "epoch": 4.11214953271028,
+      "grad_norm": 0.7235063910484314,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.0449,
+      "step": 440
+    },
+    {
+      "epoch": 4.205607476635514,
+      "grad_norm": 0.9473637342453003,
+      "learning_rate": 9e-05,
+      "loss": 0.0428,
+      "step": 450
+    },
+    {
+      "epoch": 4.299065420560748,
+      "grad_norm": 0.5876563787460327,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.0411,
+      "step": 460
+    },
+    {
+      "epoch": 4.392523364485982,
+      "grad_norm": 0.7518793344497681,
+      "learning_rate": 9.4e-05,
+      "loss": 0.047,
+      "step": 470
+    },
+    {
+      "epoch": 4.485981308411215,
+      "grad_norm": 0.47040578722953796,
+      "learning_rate": 9.6e-05,
+      "loss": 0.0387,
+      "step": 480
+    },
+    {
+      "epoch": 4.579439252336448,
+      "grad_norm": 0.7008688449859619,
+      "learning_rate": 9.8e-05,
+      "loss": 0.0391,
+      "step": 490
+    },
+    {
+      "epoch": 4.672897196261682,
+      "grad_norm": 0.32632163166999817,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 500
+    },
+    {
+      "epoch": 4.766355140186916,
+      "grad_norm": 1.0094679594039917,
+      "learning_rate": 9.999972660400536e-05,
+      "loss": 0.039,
+      "step": 510
+    },
+    {
+      "epoch": 4.859813084112149,
+      "grad_norm": 0.382621169090271,
+      "learning_rate": 9.999890641901125e-05,
+      "loss": 0.0303,
+      "step": 520
+    },
+    {
+      "epoch": 4.953271028037383,
+      "grad_norm": 0.2948877811431885,
+      "learning_rate": 9.999753945398704e-05,
+      "loss": 0.0362,
+      "step": 530
+    },
+    {
+      "epoch": 5.046728971962617,
+      "grad_norm": 0.5779514908790588,
+      "learning_rate": 9.99956257238817e-05,
+      "loss": 0.0373,
+      "step": 540
+    },
+    {
+      "epoch": 5.140186915887851,
+      "grad_norm": 0.414638489484787,
+      "learning_rate": 9.999316524962345e-05,
+      "loss": 0.0308,
+      "step": 550
+    },
+    {
+      "epoch": 5.233644859813084,
+      "grad_norm": 0.28259772062301636,
+      "learning_rate": 9.999015805811965e-05,
+      "loss": 0.0353,
+      "step": 560
+    },
+    {
+      "epoch": 5.327102803738318,
+      "grad_norm": 0.599785566329956,
+      "learning_rate": 9.998660418225645e-05,
+      "loss": 0.0448,
+      "step": 570
+    },
+    {
+      "epoch": 5.420560747663552,
+      "grad_norm": 0.3182764947414398,
+      "learning_rate": 9.998250366089848e-05,
+      "loss": 0.0408,
+      "step": 580
+    },
+    {
+      "epoch": 5.5140186915887845,
+      "grad_norm": 0.59663325548172,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.0292,
+      "step": 590
+    },
+    {
+      "epoch": 5.607476635514018,
+      "grad_norm": 0.6391479969024658,
+      "learning_rate": 9.997266286704631e-05,
+      "loss": 0.0344,
+      "step": 600
+    },
+    {
+      "epoch": 5.700934579439252,
+      "grad_norm": 0.6931540369987488,
+      "learning_rate": 9.996692270216947e-05,
+      "loss": 0.0412,
+      "step": 610
+    },
+    {
+      "epoch": 5.794392523364486,
+      "grad_norm": 0.4313046932220459,
+      "learning_rate": 9.996063610703137e-05,
+      "loss": 0.0376,
+      "step": 620
+    },
+    {
+      "epoch": 5.88785046728972,
+      "grad_norm": 0.42874786257743835,
+      "learning_rate": 9.995380315038119e-05,
+      "loss": 0.0314,
+      "step": 630
+    },
+    {
+      "epoch": 5.981308411214953,
+      "grad_norm": 0.24948689341545105,
+      "learning_rate": 9.994642390694308e-05,
+      "loss": 0.0277,
+      "step": 640
+    },
+    {
+      "epoch": 6.074766355140187,
+      "grad_norm": 0.3999479413032532,
+      "learning_rate": 9.993849845741524e-05,
+      "loss": 0.03,
+      "step": 650
+    },
+    {
+      "epoch": 6.168224299065421,
+      "grad_norm": 0.3755294382572174,
+      "learning_rate": 9.993002688846913e-05,
+      "loss": 0.0313,
+      "step": 660
+    },
+    {
+      "epoch": 6.261682242990654,
+      "grad_norm": 0.540371298789978,
+      "learning_rate": 9.992100929274846e-05,
+      "loss": 0.0239,
+      "step": 670
+    },
+    {
+      "epoch": 6.355140186915888,
+      "grad_norm": 0.37286096811294556,
+      "learning_rate": 9.991144576886823e-05,
+      "loss": 0.0375,
+      "step": 680
+    },
+    {
+      "epoch": 6.4485981308411215,
+      "grad_norm": 0.6131537556648254,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.0268,
+      "step": 690
+    },
+    {
+      "epoch": 6.542056074766355,
+      "grad_norm": 0.3708513379096985,
+      "learning_rate": 9.989068136093873e-05,
+      "loss": 0.0285,
+      "step": 700
+    },
+    {
+      "epoch": 6.635514018691588,
+      "grad_norm": 0.4660991430282593,
+      "learning_rate": 9.987948070396571e-05,
+      "loss": 0.0321,
+      "step": 710
+    },
+    {
+      "epoch": 6.728971962616822,
+      "grad_norm": 0.41945624351501465,
+      "learning_rate": 9.986773457298311e-05,
+      "loss": 0.0269,
+      "step": 720
+    },
+    {
+      "epoch": 6.822429906542056,
+      "grad_norm": 0.4975600242614746,
+      "learning_rate": 9.985544309644475e-05,
+      "loss": 0.0357,
+      "step": 730
+    },
+    {
+      "epoch": 6.91588785046729,
+      "grad_norm": 0.6201094388961792,
+      "learning_rate": 9.984260640876821e-05,
+      "loss": 0.0313,
+      "step": 740
+    },
+    {
+      "epoch": 7.009345794392523,
+      "grad_norm": 0.4306051433086395,
+      "learning_rate": 9.98292246503335e-05,
+      "loss": 0.0294,
+      "step": 750
+    },
+    {
+      "epoch": 7.102803738317757,
+      "grad_norm": 0.4569713771343231,
+      "learning_rate": 9.981529796748134e-05,
+      "loss": 0.029,
+      "step": 760
+    },
+    {
+      "epoch": 7.196261682242991,
+      "grad_norm": 0.6840571165084839,
+      "learning_rate": 9.980082651251175e-05,
+      "loss": 0.0289,
+      "step": 770
+    },
+    {
+      "epoch": 7.289719626168225,
+      "grad_norm": 0.36494606733322144,
+      "learning_rate": 9.97858104436822e-05,
+      "loss": 0.0314,
+      "step": 780
+    },
+    {
+      "epoch": 7.383177570093458,
+      "grad_norm": 0.7286158204078674,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.0248,
+      "step": 790
+    },
+    {
+      "epoch": 7.4766355140186915,
+      "grad_norm": 0.6263937950134277,
+      "learning_rate": 9.975414512725057e-05,
+      "loss": 0.0304,
+      "step": 800
+    },
+    {
+      "epoch": 7.570093457943925,
+      "grad_norm": 0.4872860312461853,
+      "learning_rate": 9.973749622593534e-05,
+      "loss": 0.0246,
+      "step": 810
+    },
+    {
+      "epoch": 7.663551401869158,
+      "grad_norm": 0.5235103368759155,
+      "learning_rate": 9.972030340333001e-05,
+      "loss": 0.0404,
+      "step": 820
+    },
+    {
+      "epoch": 7.757009345794392,
+      "grad_norm": 0.5448396801948547,
+      "learning_rate": 9.970256684745258e-05,
+      "loss": 0.0302,
+      "step": 830
+    },
+    {
+      "epoch": 7.850467289719626,
+      "grad_norm": 0.40461036562919617,
+      "learning_rate": 9.968428675226714e-05,
+      "loss": 0.0226,
+      "step": 840
+    },
+    {
+      "epoch": 7.94392523364486,
+      "grad_norm": 0.44990962743759155,
+      "learning_rate": 9.966546331768191e-05,
+      "loss": 0.0244,
+      "step": 850
+    },
+    {
+      "epoch": 8.037383177570094,
+      "grad_norm": 0.3754257261753082,
+      "learning_rate": 9.964609674954696e-05,
+      "loss": 0.0304,
+      "step": 860
+    },
+    {
+      "epoch": 8.130841121495328,
+      "grad_norm": 0.6377671957015991,
+      "learning_rate": 9.962618725965196e-05,
+      "loss": 0.0255,
+      "step": 870
+    },
+    {
+      "epoch": 8.22429906542056,
+      "grad_norm": 0.509066641330719,
+      "learning_rate": 9.96057350657239e-05,
+      "loss": 0.0286,
+      "step": 880
+    },
+    {
+      "epoch": 8.317757009345794,
+      "grad_norm": 0.39900103211402893,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.0258,
+      "step": 890
+    },
+    {
+      "epoch": 8.411214953271028,
+      "grad_norm": 0.4878309965133667,
+      "learning_rate": 9.956320346634876e-05,
+      "loss": 0.0276,
+      "step": 900
+    },
+    {
+      "epoch": 8.504672897196262,
+      "grad_norm": 0.4488677680492401,
+      "learning_rate": 9.954112452602045e-05,
+      "loss": 0.0286,
+      "step": 910
+    },
+    {
+      "epoch": 8.598130841121495,
+      "grad_norm": 0.6243172883987427,
+      "learning_rate": 9.95185038118915e-05,
+      "loss": 0.0318,
+      "step": 920
+    },
+    {
+      "epoch": 8.69158878504673,
+      "grad_norm": 0.34763044118881226,
+      "learning_rate": 9.949534157133844e-05,
+      "loss": 0.0403,
+      "step": 930
+    },
+    {
+      "epoch": 8.785046728971963,
+      "grad_norm": 0.45288386940956116,
+      "learning_rate": 9.94716380576598e-05,
+      "loss": 0.0217,
+      "step": 940
+    },
+    {
+      "epoch": 8.878504672897197,
+      "grad_norm": 0.479641854763031,
+      "learning_rate": 9.944739353007344e-05,
+      "loss": 0.0241,
+      "step": 950
+    },
+    {
+      "epoch": 8.97196261682243,
+      "grad_norm": 0.7133591175079346,
+      "learning_rate": 9.942260825371358e-05,
+      "loss": 0.0316,
+      "step": 960
+    },
+    {
+      "epoch": 9.065420560747663,
+      "grad_norm": 0.47209638357162476,
+      "learning_rate": 9.939728249962807e-05,
+      "loss": 0.0217,
+      "step": 970
+    },
+    {
+      "epoch": 9.158878504672897,
+      "grad_norm": 0.4069143235683441,
+      "learning_rate": 9.937141654477528e-05,
+      "loss": 0.0243,
+      "step": 980
+    },
+    {
+      "epoch": 9.25233644859813,
+      "grad_norm": 0.45809483528137207,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.0231,
+      "step": 990
+    },
+    {
+      "epoch": 9.345794392523365,
+      "grad_norm": 0.3990670144557953,
+      "learning_rate": 9.931806517013612e-05,
+      "loss": 0.0245,
+      "step": 1000
+    },
+    {
+      "epoch": 9.439252336448599,
+      "grad_norm": 0.35518234968185425,
+      "learning_rate": 9.929058033379181e-05,
+      "loss": 0.0338,
+      "step": 1010
+    },
+    {
+      "epoch": 9.532710280373832,
+      "grad_norm": 0.3511808514595032,
+      "learning_rate": 9.926255646355804e-05,
+      "loss": 0.0219,
+      "step": 1020
+    },
+    {
+      "epoch": 9.626168224299064,
+      "grad_norm": 0.36739039421081543,
+      "learning_rate": 9.923399386589933e-05,
+      "loss": 0.0244,
+      "step": 1030
+    },
+    {
+      "epoch": 9.719626168224298,
+      "grad_norm": 0.3416367471218109,
+      "learning_rate": 9.92048928531717e-05,
+      "loss": 0.0208,
+      "step": 1040
+    },
+    {
+      "epoch": 9.813084112149532,
+      "grad_norm": 0.3843317925930023,
+      "learning_rate": 9.917525374361912e-05,
+      "loss": 0.0222,
+      "step": 1050
+    },
+    {
+      "epoch": 9.906542056074766,
+      "grad_norm": 0.3594072163105011,
+      "learning_rate": 9.914507686137019e-05,
+      "loss": 0.0292,
+      "step": 1060
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.29179903864860535,
+      "learning_rate": 9.911436253643445e-05,
+      "loss": 0.0183,
+      "step": 1070
+    },
+    {
+      "epoch": 10.093457943925234,
+      "grad_norm": 0.24897705018520355,
+      "learning_rate": 9.90831111046988e-05,
+      "loss": 0.0236,
+      "step": 1080
+    },
+    {
+      "epoch": 10.186915887850468,
+      "grad_norm": 0.357282817363739,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.0224,
+      "step": 1090
+    },
+    {
+      "epoch": 10.280373831775702,
+      "grad_norm": 0.14735649526119232,
+      "learning_rate": 9.901899829374047e-05,
+      "loss": 0.0181,
+      "step": 1100
+    },
+    {
+      "epoch": 10.373831775700934,
+      "grad_norm": 0.30263710021972656,
+      "learning_rate": 9.89861376156452e-05,
+      "loss": 0.0195,
+      "step": 1110
+    },
+    {
+      "epoch": 10.467289719626168,
+      "grad_norm": 0.13147243857383728,
+      "learning_rate": 9.895274123299723e-05,
+      "loss": 0.0185,
+      "step": 1120
+    },
+    {
+      "epoch": 10.560747663551401,
+      "grad_norm": 0.20842011272907257,
+      "learning_rate": 9.891880951101407e-05,
+      "loss": 0.0194,
+      "step": 1130
+    },
+    {
+      "epoch": 10.654205607476635,
+      "grad_norm": 0.28125032782554626,
+      "learning_rate": 9.888434282076758e-05,
+      "loss": 0.0167,
+      "step": 1140
+    },
+    {
+      "epoch": 10.74766355140187,
+      "grad_norm": 0.25410014390945435,
+      "learning_rate": 9.884934153917997e-05,
+      "loss": 0.0221,
+      "step": 1150
+    },
+    {
+      "epoch": 10.841121495327103,
+      "grad_norm": 0.31990426778793335,
+      "learning_rate": 9.881380604901964e-05,
+      "loss": 0.0168,
+      "step": 1160
+    },
+    {
+      "epoch": 10.934579439252337,
+      "grad_norm": 0.1671873927116394,
+      "learning_rate": 9.877773673889701e-05,
+      "loss": 0.0213,
+      "step": 1170
+    },
+    {
+      "epoch": 11.02803738317757,
+      "grad_norm": 0.38101059198379517,
+      "learning_rate": 9.87411340032603e-05,
+      "loss": 0.0209,
+      "step": 1180
+    },
+    {
+      "epoch": 11.121495327102803,
+      "grad_norm": 0.24433757364749908,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.0217,
+      "step": 1190
+    },
+    {
+      "epoch": 11.214953271028037,
+      "grad_norm": 0.35972458124160767,
+      "learning_rate": 9.86663298624003e-05,
+      "loss": 0.0277,
+      "step": 1200
+    },
+    {
+      "epoch": 11.30841121495327,
+      "grad_norm": 0.19008149206638336,
+      "learning_rate": 9.862812927522309e-05,
+      "loss": 0.0279,
+      "step": 1210
+    },
+    {
+      "epoch": 11.401869158878505,
+      "grad_norm": 0.5246997475624084,
+      "learning_rate": 9.858939689861506e-05,
+      "loss": 0.0246,
+      "step": 1220
+    },
+    {
+      "epoch": 11.495327102803738,
+      "grad_norm": 0.31999820470809937,
+      "learning_rate": 9.855013315614725e-05,
+      "loss": 0.0194,
+      "step": 1230
+    },
+    {
+      "epoch": 11.588785046728972,
+      "grad_norm": 0.26507657766342163,
+      "learning_rate": 9.851033847720166e-05,
+      "loss": 0.0222,
+      "step": 1240
+    },
+    {
+      "epoch": 11.682242990654206,
+      "grad_norm": 0.2151166796684265,
+      "learning_rate": 9.847001329696653e-05,
+      "loss": 0.0169,
+      "step": 1250
+    },
+    {
+      "epoch": 11.77570093457944,
+      "grad_norm": 0.22967422008514404,
+      "learning_rate": 9.842915805643155e-05,
+      "loss": 0.0171,
+      "step": 1260
+    },
+    {
+      "epoch": 11.869158878504672,
+      "grad_norm": 0.32531777024269104,
+      "learning_rate": 9.838777320238312e-05,
+      "loss": 0.0173,
+      "step": 1270
+    },
+    {
+      "epoch": 11.962616822429906,
+      "grad_norm": 0.3821974992752075,
+      "learning_rate": 9.834585918739936e-05,
+      "loss": 0.0209,
+      "step": 1280
+    },
+    {
+      "epoch": 12.05607476635514,
+      "grad_norm": 0.295969158411026,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.0196,
+      "step": 1290
+    },
+    {
+      "epoch": 12.149532710280374,
+      "grad_norm": 0.43346959352493286,
+      "learning_rate": 9.826044551386744e-05,
+      "loss": 0.0192,
+      "step": 1300
+    },
+    {
+      "epoch": 12.242990654205608,
+      "grad_norm": 0.2353905439376831,
+      "learning_rate": 9.821694678938953e-05,
+      "loss": 0.0187,
+      "step": 1310
+    },
+    {
+      "epoch": 12.336448598130842,
+      "grad_norm": 0.2748644948005676,
+      "learning_rate": 9.817292077210659e-05,
+      "loss": 0.0199,
+      "step": 1320
+    },
+    {
+      "epoch": 12.429906542056075,
+      "grad_norm": 0.32517120242118835,
+      "learning_rate": 9.812836794348004e-05,
+      "loss": 0.0242,
+      "step": 1330
+    },
+    {
+      "epoch": 12.523364485981308,
+      "grad_norm": 0.4817813038825989,
+      "learning_rate": 9.808328879073251e-05,
+      "loss": 0.0268,
+      "step": 1340
+    },
+    {
+      "epoch": 12.616822429906541,
+      "grad_norm": 0.28131112456321716,
+      "learning_rate": 9.803768380684242e-05,
+      "loss": 0.0177,
+      "step": 1350
+    },
+    {
+      "epoch": 12.710280373831775,
+      "grad_norm": 0.23528891801834106,
+      "learning_rate": 9.799155349053851e-05,
+      "loss": 0.0266,
+      "step": 1360
+    },
+    {
+      "epoch": 12.80373831775701,
+      "grad_norm": 0.33092162013053894,
+      "learning_rate": 9.794489834629455e-05,
+      "loss": 0.0194,
+      "step": 1370
+    },
+    {
+      "epoch": 12.897196261682243,
+      "grad_norm": 0.5667449235916138,
+      "learning_rate": 9.789771888432375e-05,
+      "loss": 0.0202,
+      "step": 1380
+    },
+    {
+      "epoch": 12.990654205607477,
+      "grad_norm": 0.5390638113021851,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.0231,
+      "step": 1390
+    },
+    {
+      "epoch": 13.08411214953271,
+      "grad_norm": 0.28344106674194336,
+      "learning_rate": 9.780178907671789e-05,
+      "loss": 0.0176,
+      "step": 1400
+    },
+    {
+      "epoch": 13.177570093457945,
+      "grad_norm": 0.3115670084953308,
+      "learning_rate": 9.775303978015585e-05,
+      "loss": 0.0285,
+      "step": 1410
+    },
+    {
+      "epoch": 13.271028037383177,
+      "grad_norm": 0.2856522798538208,
+      "learning_rate": 9.77037682640015e-05,
+      "loss": 0.0219,
+      "step": 1420
+    },
+    {
+      "epoch": 13.36448598130841,
+      "grad_norm": 0.41443973779678345,
+      "learning_rate": 9.765397506708023e-05,
+      "loss": 0.0235,
+      "step": 1430
+    },
+    {
+      "epoch": 13.457943925233645,
+      "grad_norm": 0.2849663496017456,
+      "learning_rate": 9.760366073392246e-05,
+      "loss": 0.0167,
+      "step": 1440
+    },
+    {
+      "epoch": 13.551401869158878,
+      "grad_norm": 0.28459078073501587,
+      "learning_rate": 9.755282581475769e-05,
+      "loss": 0.0253,
+      "step": 1450
+    },
+    {
+      "epoch": 13.644859813084112,
+      "grad_norm": 0.25047361850738525,
+      "learning_rate": 9.750147086550844e-05,
+      "loss": 0.0186,
+      "step": 1460
+    },
+    {
+      "epoch": 13.738317757009346,
+      "grad_norm": 0.2597232162952423,
+      "learning_rate": 9.744959644778422e-05,
+      "loss": 0.0137,
+      "step": 1470
+    },
+    {
+      "epoch": 13.83177570093458,
+      "grad_norm": 0.14577718079090118,
+      "learning_rate": 9.739720312887535e-05,
+      "loss": 0.0194,
+      "step": 1480
+    },
+    {
+      "epoch": 13.925233644859812,
+      "grad_norm": 0.2940736413002014,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.0165,
+      "step": 1490
+    },
+    {
+      "epoch": 14.018691588785046,
+      "grad_norm": 0.33982032537460327,
+      "learning_rate": 9.729086208503174e-05,
+      "loss": 0.0229,
+      "step": 1500
+    },
+    {
+      "epoch": 14.11214953271028,
+      "grad_norm": 0.2669644355773926,
+      "learning_rate": 9.723691552302562e-05,
+      "loss": 0.0151,
+      "step": 1510
+    },
+    {
+      "epoch": 14.205607476635514,
+      "grad_norm": 0.20052188634872437,
+      "learning_rate": 9.718245238567939e-05,
+      "loss": 0.0169,
+      "step": 1520
+    },
+    {
+      "epoch": 14.299065420560748,
+      "grad_norm": 0.27926936745643616,
+      "learning_rate": 9.712747326859315e-05,
+      "loss": 0.0184,
+      "step": 1530
+    },
+    {
+      "epoch": 14.392523364485982,
+      "grad_norm": 0.38229790329933167,
+      "learning_rate": 9.707197877300974e-05,
+      "loss": 0.0142,
+      "step": 1540
+    },
+    {
+      "epoch": 14.485981308411215,
+      "grad_norm": 0.2728178799152374,
+      "learning_rate": 9.701596950580806e-05,
+      "loss": 0.0231,
+      "step": 1550
+    },
+    {
+      "epoch": 14.57943925233645,
+      "grad_norm": 0.17084470391273499,
+      "learning_rate": 9.695944607949649e-05,
+      "loss": 0.0124,
+      "step": 1560
+    },
+    {
+      "epoch": 14.672897196261681,
+      "grad_norm": 0.24558572471141815,
+      "learning_rate": 9.690240911220618e-05,
+      "loss": 0.0153,
+      "step": 1570
+    },
+    {
+      "epoch": 14.766355140186915,
+      "grad_norm": 0.1569851040840149,
+      "learning_rate": 9.684485922768422e-05,
+      "loss": 0.0179,
+      "step": 1580
+    },
+    {
+      "epoch": 14.85981308411215,
+      "grad_norm": 0.3535224199295044,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.02,
+      "step": 1590
+    },
+    {
+      "epoch": 14.953271028037383,
+      "grad_norm": 0.4264807105064392,
+      "learning_rate": 9.672822322997305e-05,
+      "loss": 0.0207,
+      "step": 1600
+    },
+    {
+      "epoch": 15.046728971962617,
+      "grad_norm": 0.3585994839668274,
+      "learning_rate": 9.66691383922964e-05,
+      "loss": 0.0196,
+      "step": 1610
+    },
+    {
+      "epoch": 15.14018691588785,
+      "grad_norm": 0.27081626653671265,
+      "learning_rate": 9.660954318839933e-05,
+      "loss": 0.018,
+      "step": 1620
+    },
+    {
+      "epoch": 15.233644859813085,
+      "grad_norm": 0.31903624534606934,
+      "learning_rate": 9.654943827000548e-05,
+      "loss": 0.0221,
+      "step": 1630
+    },
+    {
+      "epoch": 15.327102803738319,
+      "grad_norm": 0.18142905831336975,
+      "learning_rate": 9.648882429441257e-05,
+      "loss": 0.0169,
+      "step": 1640
+    },
+    {
+      "epoch": 15.42056074766355,
+      "grad_norm": 0.23492714762687683,
+      "learning_rate": 9.642770192448536e-05,
+      "loss": 0.0157,
+      "step": 1650
+    },
+    {
+      "epoch": 15.514018691588785,
+      "grad_norm": 0.33975672721862793,
+      "learning_rate": 9.636607182864827e-05,
+      "loss": 0.0249,
+      "step": 1660
+    },
+    {
+      "epoch": 15.607476635514018,
+      "grad_norm": 0.413004070520401,
+      "learning_rate": 9.630393468087818e-05,
+      "loss": 0.0156,
+      "step": 1670
+    },
+    {
+      "epoch": 15.700934579439252,
+      "grad_norm": 0.2392245978116989,
+      "learning_rate": 9.624129116069694e-05,
+      "loss": 0.0178,
+      "step": 1680
+    },
+    {
+      "epoch": 15.794392523364486,
+      "grad_norm": 0.2815963327884674,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.0282,
+      "step": 1690
+    },
+    {
+      "epoch": 15.88785046728972,
+      "grad_norm": 0.23372851312160492,
+      "learning_rate": 9.611448774886924e-05,
+      "loss": 0.0164,
+      "step": 1700
+    },
+    {
+      "epoch": 15.981308411214954,
+      "grad_norm": 0.42078474164009094,
+      "learning_rate": 9.605032924392457e-05,
+      "loss": 0.0155,
+      "step": 1710
+    },
+    {
+      "epoch": 16.074766355140188,
+      "grad_norm": 0.35717156529426575,
+      "learning_rate": 9.598566713995718e-05,
+      "loss": 0.0184,
+      "step": 1720
+    },
+    {
+      "epoch": 16.16822429906542,
+      "grad_norm": 0.2764424979686737,
+      "learning_rate": 9.59205021441015e-05,
+      "loss": 0.0147,
+      "step": 1730
+    },
+    {
+      "epoch": 16.261682242990656,
+      "grad_norm": 0.26016825437545776,
+      "learning_rate": 9.58548349689915e-05,
+      "loss": 0.0224,
+      "step": 1740
+    },
+    {
+      "epoch": 16.35514018691589,
+      "grad_norm": 0.27049052715301514,
+      "learning_rate": 9.578866633275288e-05,
+      "loss": 0.0146,
+      "step": 1750
+    },
+    {
+      "epoch": 16.44859813084112,
+      "grad_norm": 0.338381826877594,
+      "learning_rate": 9.572199695899522e-05,
+      "loss": 0.0165,
+      "step": 1760
+    },
+    {
+      "epoch": 16.542056074766354,
+      "grad_norm": 0.26160427927970886,
+      "learning_rate": 9.565482757680415e-05,
+      "loss": 0.0157,
+      "step": 1770
+    },
+    {
+      "epoch": 16.635514018691588,
+      "grad_norm": 0.2207760363817215,
+      "learning_rate": 9.558715892073323e-05,
+      "loss": 0.013,
+      "step": 1780
+    },
+    {
+      "epoch": 16.72897196261682,
+      "grad_norm": 0.37250402569770813,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.0184,
+      "step": 1790
+    },
+    {
+      "epoch": 16.822429906542055,
+      "grad_norm": 0.28439027070999146,
+      "learning_rate": 9.545032675245813e-05,
+      "loss": 0.021,
+      "step": 1800
+    },
+    {
+      "epoch": 16.91588785046729,
+      "grad_norm": 0.3891927897930145,
+      "learning_rate": 9.538116473662861e-05,
+      "loss": 0.0141,
+      "step": 1810
+    },
+    {
+      "epoch": 17.009345794392523,
+      "grad_norm": 0.2712792754173279,
+      "learning_rate": 9.531150643965223e-05,
+      "loss": 0.0108,
+      "step": 1820
+    },
+    {
+      "epoch": 17.102803738317757,
+      "grad_norm": 0.4965173006057739,
+      "learning_rate": 9.524135262330098e-05,
+      "loss": 0.0162,
+      "step": 1830
+    },
+    {
+      "epoch": 17.19626168224299,
+      "grad_norm": 0.360164999961853,
+      "learning_rate": 9.517070405476575e-05,
+      "loss": 0.0175,
+      "step": 1840
+    },
+    {
+      "epoch": 17.289719626168225,
+      "grad_norm": 0.3308715522289276,
+      "learning_rate": 9.509956150664796e-05,
+      "loss": 0.0144,
+      "step": 1850
+    },
+    {
+      "epoch": 17.38317757009346,
+      "grad_norm": 0.24046167731285095,
+      "learning_rate": 9.502792575695112e-05,
+      "loss": 0.0142,
+      "step": 1860
+    },
+    {
+      "epoch": 17.476635514018692,
+      "grad_norm": 0.32182639837265015,
+      "learning_rate": 9.49557975890723e-05,
+      "loss": 0.0124,
+      "step": 1870
+    },
+    {
+      "epoch": 17.570093457943926,
+      "grad_norm": 0.19691896438598633,
+      "learning_rate": 9.488317779179361e-05,
+      "loss": 0.0152,
+      "step": 1880
+    },
+    {
+      "epoch": 17.66355140186916,
+      "grad_norm": 0.248452827334404,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.0195,
+      "step": 1890
+    },
+    {
+      "epoch": 17.757009345794394,
+      "grad_norm": 0.37300196290016174,
+      "learning_rate": 9.473646649103818e-05,
+      "loss": 0.022,
+      "step": 1900
+    },
+    {
+      "epoch": 17.850467289719624,
+      "grad_norm": 0.3653255105018616,
+      "learning_rate": 9.46623765919727e-05,
+      "loss": 0.0155,
+      "step": 1910
+    },
+    {
+      "epoch": 17.94392523364486,
+      "grad_norm": 0.1721230000257492,
+      "learning_rate": 9.458779827231237e-05,
+      "loss": 0.0113,
+      "step": 1920
+    },
+    {
+      "epoch": 18.037383177570092,
+      "grad_norm": 0.3271198868751526,
+      "learning_rate": 9.451273234763371e-05,
+      "loss": 0.0155,
+      "step": 1930
+    },
+    {
+      "epoch": 18.130841121495326,
+      "grad_norm": 0.3506911098957062,
+      "learning_rate": 9.443717963884569e-05,
+      "loss": 0.0113,
+      "step": 1940
+    },
+    {
+      "epoch": 18.22429906542056,
+      "grad_norm": 0.1824873387813568,
+      "learning_rate": 9.43611409721806e-05,
+      "loss": 0.0168,
+      "step": 1950
+    },
+    {
+      "epoch": 18.317757009345794,
+      "grad_norm": 0.3345133066177368,
+      "learning_rate": 9.428461717918511e-05,
+      "loss": 0.0166,
+      "step": 1960
+    },
+    {
+      "epoch": 18.411214953271028,
+      "grad_norm": 0.3359324634075165,
+      "learning_rate": 9.420760909671118e-05,
+      "loss": 0.0165,
+      "step": 1970
+    },
+    {
+      "epoch": 18.50467289719626,
+      "grad_norm": 0.24985487759113312,
+      "learning_rate": 9.413011756690685e-05,
+      "loss": 0.0211,
+      "step": 1980
+    },
+    {
+      "epoch": 18.598130841121495,
+      "grad_norm": 0.2868458926677704,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.0142,
+      "step": 1990
+    },
+    {
+      "epoch": 18.69158878504673,
+      "grad_norm": 0.3124500513076782,
+      "learning_rate": 9.397368756032445e-05,
+      "loss": 0.018,
+      "step": 2000
+    },
+    {
+      "epoch": 18.785046728971963,
+      "grad_norm": 0.328535795211792,
+      "learning_rate": 9.389475079423988e-05,
+      "loss": 0.0155,
+      "step": 2010
+    },
+    {
+      "epoch": 18.878504672897197,
+      "grad_norm": 0.2067783921957016,
+      "learning_rate": 9.381533400219318e-05,
+      "loss": 0.0125,
+      "step": 2020
+    },
+    {
+      "epoch": 18.97196261682243,
+      "grad_norm": 0.27102118730545044,
+      "learning_rate": 9.373543805267368e-05,
+      "loss": 0.0148,
+      "step": 2030
+    },
+    {
+      "epoch": 19.065420560747665,
+      "grad_norm": 0.28738537430763245,
+      "learning_rate": 9.365506381941066e-05,
+      "loss": 0.0145,
+      "step": 2040
+    },
+    {
+      "epoch": 19.1588785046729,
+      "grad_norm": 0.2228069007396698,
+      "learning_rate": 9.357421218136386e-05,
+      "loss": 0.013,
+      "step": 2050
+    },
+    {
+      "epoch": 19.252336448598133,
+      "grad_norm": 0.26949775218963623,
+      "learning_rate": 9.349288402271388e-05,
+      "loss": 0.018,
+      "step": 2060
+    },
+    {
+      "epoch": 19.345794392523363,
+      "grad_norm": 0.1682933121919632,
+      "learning_rate": 9.341108023285238e-05,
+      "loss": 0.015,
+      "step": 2070
+    },
+    {
+      "epoch": 19.439252336448597,
+      "grad_norm": 0.36588406562805176,
+      "learning_rate": 9.332880170637252e-05,
+      "loss": 0.0167,
+      "step": 2080
+    },
+    {
+      "epoch": 19.53271028037383,
+      "grad_norm": 0.1443902850151062,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.0129,
+      "step": 2090
+    },
+    {
+      "epoch": 19.626168224299064,
+      "grad_norm": 0.1862812638282776,
+      "learning_rate": 9.316282404787871e-05,
+      "loss": 0.0133,
+      "step": 2100
+    },
+    {
+      "epoch": 19.7196261682243,
+      "grad_norm": 0.26972711086273193,
+      "learning_rate": 9.30791267309698e-05,
+      "loss": 0.0183,
+      "step": 2110
+    },
+    {
+      "epoch": 19.813084112149532,
+      "grad_norm": 0.4285616874694824,
+      "learning_rate": 9.299495830763286e-05,
+      "loss": 0.0222,
+      "step": 2120
+    },
+    {
+      "epoch": 19.906542056074766,
+      "grad_norm": 0.38283267617225647,
+      "learning_rate": 9.291031969832026e-05,
+      "loss": 0.0195,
+      "step": 2130
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.29968154430389404,
+      "learning_rate": 9.282521182862629e-05,
+      "loss": 0.0134,
+      "step": 2140
+    },
+    {
+      "epoch": 20.093457943925234,
+      "grad_norm": 0.20329944789409637,
+      "learning_rate": 9.273963562927695e-05,
+      "loss": 0.0123,
+      "step": 2150
+    },
+    {
+      "epoch": 20.186915887850468,
+      "grad_norm": 0.26469314098358154,
+      "learning_rate": 9.265359203611987e-05,
+      "loss": 0.0223,
+      "step": 2160
+    },
+    {
+      "epoch": 20.2803738317757,
+      "grad_norm": 0.19388605654239655,
+      "learning_rate": 9.256708199011401e-05,
+      "loss": 0.0106,
+      "step": 2170
+    },
+    {
+      "epoch": 20.373831775700936,
+      "grad_norm": 0.2838001549243927,
+      "learning_rate": 9.248010643731935e-05,
+      "loss": 0.0093,
+      "step": 2180
+    },
+    {
+      "epoch": 20.46728971962617,
+      "grad_norm": 0.27465811371803284,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.0124,
+      "step": 2190
+    },
+    {
+      "epoch": 20.560747663551403,
+      "grad_norm": 0.17780600488185883,
+      "learning_rate": 9.230476262104677e-05,
+      "loss": 0.0141,
+      "step": 2200
+    },
+    {
+      "epoch": 20.654205607476637,
+      "grad_norm": 0.3275097608566284,
+      "learning_rate": 9.221639627510076e-05,
+      "loss": 0.0173,
+      "step": 2210
+    },
+    {
+      "epoch": 20.747663551401867,
+      "grad_norm": 0.42499256134033203,
+      "learning_rate": 9.212756825740873e-05,
+      "loss": 0.0154,
+      "step": 2220
+    },
+    {
+      "epoch": 20.8411214953271,
+      "grad_norm": 0.28516319394111633,
+      "learning_rate": 9.20382795393797e-05,
+      "loss": 0.0177,
+      "step": 2230
+    },
+    {
+      "epoch": 20.934579439252335,
+      "grad_norm": 0.2422408163547516,
+      "learning_rate": 9.194853109746074e-05,
+      "loss": 0.0133,
+      "step": 2240
+    },
+    {
+      "epoch": 21.02803738317757,
+      "grad_norm": 0.35381415486335754,
+      "learning_rate": 9.185832391312644e-05,
+      "loss": 0.0142,
+      "step": 2250
+    },
+    {
+      "epoch": 21.121495327102803,
+      "grad_norm": 0.11969628930091858,
+      "learning_rate": 9.176765897286813e-05,
+      "loss": 0.0211,
+      "step": 2260
+    },
+    {
+      "epoch": 21.214953271028037,
+      "grad_norm": 0.35882630944252014,
+      "learning_rate": 9.167653726818305e-05,
+      "loss": 0.019,
+      "step": 2270
+    },
+    {
+      "epoch": 21.30841121495327,
+      "grad_norm": 0.3682778477668762,
+      "learning_rate": 9.158495979556358e-05,
+      "loss": 0.016,
+      "step": 2280
+    },
+    {
+      "epoch": 21.401869158878505,
+      "grad_norm": 0.3526158630847931,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.017,
+      "step": 2290
+    },
+    {
+      "epoch": 21.49532710280374,
+      "grad_norm": 0.32683178782463074,
+      "learning_rate": 9.140044155740101e-05,
+      "loss": 0.015,
+      "step": 2300
+    },
+    {
+      "epoch": 21.588785046728972,
+      "grad_norm": 0.3018406629562378,
+      "learning_rate": 9.130750280971978e-05,
+      "loss": 0.0131,
+      "step": 2310
+    },
+    {
+      "epoch": 21.682242990654206,
+      "grad_norm": 0.438232958316803,
+      "learning_rate": 9.121411232980588e-05,
+      "loss": 0.0183,
+      "step": 2320
+    },
+    {
+      "epoch": 21.77570093457944,
+      "grad_norm": 0.26934388279914856,
+      "learning_rate": 9.112027113896262e-05,
+      "loss": 0.0135,
+      "step": 2330
+    },
+    {
+      "epoch": 21.869158878504674,
+      "grad_norm": 0.2123032808303833,
+      "learning_rate": 9.102598026342222e-05,
+      "loss": 0.013,
+      "step": 2340
+    },
+    {
+      "epoch": 21.962616822429908,
+      "grad_norm": 0.35099825263023376,
+      "learning_rate": 9.093124073433463e-05,
+      "loss": 0.0165,
+      "step": 2350
+    },
+    {
+      "epoch": 22.05607476635514,
+      "grad_norm": 0.3659217357635498,
+      "learning_rate": 9.083605358775612e-05,
+      "loss": 0.0117,
+      "step": 2360
+    },
+    {
+      "epoch": 22.149532710280372,
+      "grad_norm": 0.2593810260295868,
+      "learning_rate": 9.074041986463808e-05,
+      "loss": 0.0156,
+      "step": 2370
+    },
+    {
+      "epoch": 22.242990654205606,
+      "grad_norm": 0.20466811954975128,
+      "learning_rate": 9.064434061081562e-05,
+      "loss": 0.0138,
+      "step": 2380
+    },
+    {
+      "epoch": 22.33644859813084,
+      "grad_norm": 0.2669541537761688,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.0187,
+      "step": 2390
+    },
+    {
+      "epoch": 22.429906542056074,
+      "grad_norm": 0.22927866876125336,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 0.0125,
+      "step": 2400
+    },
+    {
+      "epoch": 22.523364485981308,
+      "grad_norm": 0.28933751583099365,
+      "learning_rate": 9.035344019648702e-05,
+      "loss": 0.0173,
+      "step": 2410
+    },
+    {
+      "epoch": 22.61682242990654,
+      "grad_norm": 0.2809857130050659,
+      "learning_rate": 9.025558937546988e-05,
+      "loss": 0.0143,
+      "step": 2420
+    },
+    {
+      "epoch": 22.710280373831775,
+      "grad_norm": 0.11940523982048035,
+      "learning_rate": 9.015729832577681e-05,
+      "loss": 0.011,
+      "step": 2430
+    },
+    {
+      "epoch": 22.80373831775701,
+      "grad_norm": 0.20271120965480804,
+      "learning_rate": 9.005856812230304e-05,
+      "loss": 0.0127,
+      "step": 2440
+    },
+    {
+      "epoch": 22.897196261682243,
+      "grad_norm": 0.23386332392692566,
+      "learning_rate": 8.995939984474624e-05,
+      "loss": 0.0168,
+      "step": 2450
+    },
+    {
+      "epoch": 22.990654205607477,
+      "grad_norm": 0.13469015061855316,
+      "learning_rate": 8.98597945775948e-05,
+      "loss": 0.0142,
+      "step": 2460
+    },
+    {
+      "epoch": 23.08411214953271,
+      "grad_norm": 0.2923325300216675,
+      "learning_rate": 8.975975341011596e-05,
+      "loss": 0.0128,
+      "step": 2470
+    },
+    {
+      "epoch": 23.177570093457945,
+      "grad_norm": 0.2187262773513794,
+      "learning_rate": 8.965927743634391e-05,
+      "loss": 0.0147,
+      "step": 2480
+    },
+    {
+      "epoch": 23.27102803738318,
+      "grad_norm": 0.2251782864332199,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.018,
+      "step": 2490
+    },
+    {
+      "epoch": 23.364485981308412,
+      "grad_norm": 0.2662113904953003,
+      "learning_rate": 8.945702546981969e-05,
+      "loss": 0.0171,
+      "step": 2500
+    },
+    {
+      "epoch": 23.457943925233646,
+      "grad_norm": 0.29429367184638977,
+      "learning_rate": 8.935525168886262e-05,
+      "loss": 0.0137,
+      "step": 2510
+    },
+    {
+      "epoch": 23.55140186915888,
+      "grad_norm": 0.19048871099948883,
+      "learning_rate": 8.92530475251784e-05,
+      "loss": 0.0103,
+      "step": 2520
+    },
+    {
+      "epoch": 23.64485981308411,
+      "grad_norm": 0.20753830671310425,
+      "learning_rate": 8.91504140964553e-05,
+      "loss": 0.012,
+      "step": 2530
+    },
+    {
+      "epoch": 23.738317757009344,
+      "grad_norm": 0.3121615946292877,
+      "learning_rate": 8.90473525250761e-05,
+      "loss": 0.0202,
+      "step": 2540
+    },
+    {
+      "epoch": 23.83177570093458,
+      "grad_norm": 0.3104843497276306,
+      "learning_rate": 8.894386393810563e-05,
+      "loss": 0.011,
+      "step": 2550
+    },
+    {
+      "epoch": 23.925233644859812,
+      "grad_norm": 0.29644063115119934,
+      "learning_rate": 8.883994946727849e-05,
+      "loss": 0.0192,
+      "step": 2560
+    },
+    {
+      "epoch": 24.018691588785046,
+      "grad_norm": 0.32935768365859985,
+      "learning_rate": 8.873561024898668e-05,
+      "loss": 0.0163,
+      "step": 2570
+    },
+    {
+      "epoch": 24.11214953271028,
+      "grad_norm": 0.19332408905029297,
+      "learning_rate": 8.863084742426719e-05,
+      "loss": 0.0163,
+      "step": 2580
+    },
+    {
+      "epoch": 24.205607476635514,
+      "grad_norm": 0.2649053931236267,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.0127,
+      "step": 2590
+    },
+    {
+      "epoch": 24.299065420560748,
+      "grad_norm": 0.3077411949634552,
+      "learning_rate": 8.842005554284296e-05,
+      "loss": 0.0182,
+      "step": 2600
+    },
+    {
+      "epoch": 24.39252336448598,
+      "grad_norm": 0.16626809537410736,
+      "learning_rate": 8.831402879132446e-05,
+      "loss": 0.0138,
+      "step": 2610
+    },
+    {
+      "epoch": 24.485981308411215,
+      "grad_norm": 0.3031405210494995,
+      "learning_rate": 8.820758304372557e-05,
+      "loss": 0.0139,
+      "step": 2620
+    },
+    {
+      "epoch": 24.57943925233645,
+      "grad_norm": 0.6083452105522156,
+      "learning_rate": 8.810071946411989e-05,
+      "loss": 0.0193,
+      "step": 2630
+    },
+    {
+      "epoch": 24.672897196261683,
+      "grad_norm": 0.08449734002351761,
+      "learning_rate": 8.799343922115044e-05,
+      "loss": 0.0171,
+      "step": 2640
+    },
+    {
+      "epoch": 24.766355140186917,
+      "grad_norm": 0.20243260264396667,
+      "learning_rate": 8.788574348801675e-05,
+      "loss": 0.0155,
+      "step": 2650
+    },
+    {
+      "epoch": 24.85981308411215,
+      "grad_norm": 0.29259979724884033,
+      "learning_rate": 8.77776334424621e-05,
+      "loss": 0.0131,
+      "step": 2660
+    },
+    {
+      "epoch": 24.953271028037385,
+      "grad_norm": 0.2411697506904602,
+      "learning_rate": 8.766911026676064e-05,
+      "loss": 0.0186,
+      "step": 2670
+    },
+    {
+      "epoch": 25.046728971962615,
+      "grad_norm": 0.33529865741729736,
+      "learning_rate": 8.756017514770443e-05,
+      "loss": 0.0117,
+      "step": 2680
+    },
+    {
+      "epoch": 25.14018691588785,
+      "grad_norm": 0.23808050155639648,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.011,
+      "step": 2690
+    },
+    {
+      "epoch": 25.233644859813083,
+      "grad_norm": 0.1847400665283203,
+      "learning_rate": 8.73410738492077e-05,
+      "loss": 0.0128,
+      "step": 2700
+    },
+    {
+      "epoch": 25.327102803738317,
+      "grad_norm": 0.13571485877037048,
+      "learning_rate": 8.723091006582389e-05,
+      "loss": 0.0103,
+      "step": 2710
+    },
+    {
+      "epoch": 25.42056074766355,
+      "grad_norm": 0.20057013630867004,
+      "learning_rate": 8.71203391311725e-05,
+      "loss": 0.0144,
+      "step": 2720
+    },
+    {
+      "epoch": 25.514018691588785,
+      "grad_norm": 0.18877534568309784,
+      "learning_rate": 8.700936225443959e-05,
+      "loss": 0.0087,
+      "step": 2730
+    },
+    {
+      "epoch": 25.60747663551402,
+      "grad_norm": 0.34305885434150696,
+      "learning_rate": 8.689798064925049e-05,
+      "loss": 0.0111,
+      "step": 2740
+    },
+    {
+      "epoch": 25.700934579439252,
+      "grad_norm": 0.30774959921836853,
+      "learning_rate": 8.678619553365659e-05,
+      "loss": 0.0118,
+      "step": 2750
+    },
+    {
+      "epoch": 25.794392523364486,
+      "grad_norm": 0.20990929007530212,
+      "learning_rate": 8.6674008130122e-05,
+      "loss": 0.0123,
+      "step": 2760
+    },
+    {
+      "epoch": 25.88785046728972,
+      "grad_norm": 0.14915867149829865,
+      "learning_rate": 8.656141966551019e-05,
+      "loss": 0.0104,
+      "step": 2770
+    },
+    {
+      "epoch": 25.981308411214954,
+      "grad_norm": 0.21087510883808136,
+      "learning_rate": 8.644843137107059e-05,
+      "loss": 0.0103,
+      "step": 2780
+    },
+    {
+      "epoch": 26.074766355140188,
+      "grad_norm": 0.3087444007396698,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.0146,
+      "step": 2790
+    },
+    {
+      "epoch": 26.16822429906542,
+      "grad_norm": 0.15579691529273987,
+      "learning_rate": 8.622126023955446e-05,
+      "loss": 0.0134,
+      "step": 2800
+    },
+    {
+      "epoch": 26.261682242990656,
+      "grad_norm": 0.29436925053596497,
+      "learning_rate": 8.610707988678503e-05,
+      "loss": 0.0124,
+      "step": 2810
+    },
+    {
+      "epoch": 26.35514018691589,
+      "grad_norm": 0.3273058533668518,
+      "learning_rate": 8.599250467277483e-05,
+      "loss": 0.0101,
+      "step": 2820
+    },
+    {
+      "epoch": 26.44859813084112,
+      "grad_norm": 0.28451430797576904,
+      "learning_rate": 8.587753585050004e-05,
+      "loss": 0.0195,
+      "step": 2830
+    },
+    {
+      "epoch": 26.542056074766354,
+      "grad_norm": 0.255038857460022,
+      "learning_rate": 8.576217467724128e-05,
+      "loss": 0.0142,
+      "step": 2840
+    },
+    {
+      "epoch": 26.635514018691588,
+      "grad_norm": 0.18231236934661865,
+      "learning_rate": 8.564642241456986e-05,
+      "loss": 0.0135,
+      "step": 2850
+    },
+    {
+      "epoch": 26.72897196261682,
+      "grad_norm": 0.28548383712768555,
+      "learning_rate": 8.553028032833397e-05,
+      "loss": 0.0142,
+      "step": 2860
+    },
+    {
+      "epoch": 26.822429906542055,
+      "grad_norm": 0.21996966004371643,
+      "learning_rate": 8.541374968864487e-05,
+      "loss": 0.0153,
+      "step": 2870
+    },
+    {
+      "epoch": 26.91588785046729,
+      "grad_norm": 0.18419910967350006,
+      "learning_rate": 8.529683176986295e-05,
+      "loss": 0.0114,
+      "step": 2880
+    },
+    {
+      "epoch": 27.009345794392523,
+      "grad_norm": 0.2264631688594818,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.0136,
+      "step": 2890
+    },
+    {
+      "epoch": 27.102803738317757,
+      "grad_norm": 0.28802967071533203,
+      "learning_rate": 8.506183921362443e-05,
+      "loss": 0.0147,
+      "step": 2900
+    },
+    {
+      "epoch": 27.19626168224299,
+      "grad_norm": 0.3631379306316376,
+      "learning_rate": 8.494376714600878e-05,
+      "loss": 0.0144,
+      "step": 2910
+    },
+    {
+      "epoch": 27.289719626168225,
+      "grad_norm": 0.24260710179805756,
+      "learning_rate": 8.482531293895412e-05,
+      "loss": 0.0148,
+      "step": 2920
+    },
+    {
+      "epoch": 27.38317757009346,
+      "grad_norm": 0.1798403412103653,
+      "learning_rate": 8.470647788785665e-05,
+      "loss": 0.0147,
+      "step": 2930
+    },
+    {
+      "epoch": 27.476635514018692,
+      "grad_norm": 0.34532681107521057,
+      "learning_rate": 8.458726329227747e-05,
+      "loss": 0.017,
+      "step": 2940
+    },
+    {
+      "epoch": 27.570093457943926,
+      "grad_norm": 0.13224120438098907,
+      "learning_rate": 8.44676704559283e-05,
+      "loss": 0.0231,
+      "step": 2950
+    },
+    {
+      "epoch": 27.66355140186916,
+      "grad_norm": 0.3429749011993408,
+      "learning_rate": 8.434770068665723e-05,
+      "loss": 0.0164,
+      "step": 2960
+    },
+    {
+      "epoch": 27.757009345794394,
+      "grad_norm": 0.24342355132102966,
+      "learning_rate": 8.422735529643444e-05,
+      "loss": 0.0138,
+      "step": 2970
+    },
+    {
+      "epoch": 27.850467289719624,
+      "grad_norm": 0.21670503914356232,
+      "learning_rate": 8.410663560133784e-05,
+      "loss": 0.0119,
+      "step": 2980
+    },
+    {
+      "epoch": 27.94392523364486,
+      "grad_norm": 0.21148911118507385,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.0139,
+      "step": 2990
+    },
+    {
+      "epoch": 28.037383177570092,
+      "grad_norm": 0.24043697118759155,
+      "learning_rate": 8.386407858128706e-05,
+      "loss": 0.0215,
+      "step": 3000
+    },
+    {
+      "epoch": 28.130841121495326,
+      "grad_norm": 0.25758662819862366,
+      "learning_rate": 8.37422439088976e-05,
+      "loss": 0.0136,
+      "step": 3010
+    },
+    {
+      "epoch": 28.22429906542056,
+      "grad_norm": 0.1965651512145996,
+      "learning_rate": 8.362004023673474e-05,
+      "loss": 0.0139,
+      "step": 3020
+    },
+    {
+      "epoch": 28.317757009345794,
+      "grad_norm": 0.1873561143875122,
+      "learning_rate": 8.349746890119826e-05,
+      "loss": 0.014,
+      "step": 3030
+    },
+    {
+      "epoch": 28.411214953271028,
+      "grad_norm": 0.3306177258491516,
+      "learning_rate": 8.337453124270863e-05,
+      "loss": 0.0149,
+      "step": 3040
+    },
+    {
+      "epoch": 28.50467289719626,
+      "grad_norm": 0.5167064666748047,
+      "learning_rate": 8.32512286056924e-05,
+      "loss": 0.0183,
+      "step": 3050
+    },
+    {
+      "epoch": 28.598130841121495,
+      "grad_norm": 0.28625836968421936,
+      "learning_rate": 8.31275623385675e-05,
+      "loss": 0.0129,
+      "step": 3060
+    },
+    {
+      "epoch": 28.69158878504673,
+      "grad_norm": 0.17390304803848267,
+      "learning_rate": 8.300353379372834e-05,
+      "loss": 0.0115,
+      "step": 3070
+    },
+    {
+      "epoch": 28.785046728971963,
+      "grad_norm": 0.2651899456977844,
+      "learning_rate": 8.287914432753123e-05,
+      "loss": 0.0245,
+      "step": 3080
+    },
+    {
+      "epoch": 28.878504672897197,
+      "grad_norm": 0.16162163019180298,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.0152,
+      "step": 3090
+    },
+    {
+      "epoch": 28.97196261682243,
+      "grad_norm": 0.10550884157419205,
+      "learning_rate": 8.262928807620843e-05,
+      "loss": 0.0145,
+      "step": 3100
+    },
+    {
+      "epoch": 29.065420560747665,
+      "grad_norm": 0.17049351334571838,
+      "learning_rate": 8.250382402347065e-05,
+      "loss": 0.0118,
+      "step": 3110
+    },
+    {
+      "epoch": 29.1588785046729,
+      "grad_norm": 0.3678361475467682,
+      "learning_rate": 8.237800451412095e-05,
+      "loss": 0.0141,
+      "step": 3120
+    },
+    {
+      "epoch": 29.252336448598133,
+      "grad_norm": 0.3102375268936157,
+      "learning_rate": 8.225183092410128e-05,
+      "loss": 0.0133,
+      "step": 3130
+    },
+    {
+      "epoch": 29.345794392523363,
+      "grad_norm": 0.18469548225402832,
+      "learning_rate": 8.212530463322583e-05,
+      "loss": 0.0114,
+      "step": 3140
+    },
+    {
+      "epoch": 29.439252336448597,
+      "grad_norm": 0.17053820192813873,
+      "learning_rate": 8.199842702516583e-05,
+      "loss": 0.0157,
+      "step": 3150
+    },
+    {
+      "epoch": 29.53271028037383,
+      "grad_norm": 0.22760725021362305,
+      "learning_rate": 8.18711994874345e-05,
+      "loss": 0.0103,
+      "step": 3160
+    },
+    {
+      "epoch": 29.626168224299064,
+      "grad_norm": 0.212839275598526,
+      "learning_rate": 8.174362341137177e-05,
+      "loss": 0.014,
+      "step": 3170
+    },
+    {
+      "epoch": 29.7196261682243,
+      "grad_norm": 0.16992239654064178,
+      "learning_rate": 8.161570019212921e-05,
+      "loss": 0.0113,
+      "step": 3180
+    },
+    {
+      "epoch": 29.813084112149532,
+      "grad_norm": 0.2040039300918579,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.0097,
+      "step": 3190
+    },
+    {
+      "epoch": 29.906542056074766,
+      "grad_norm": 0.23768551647663116,
+      "learning_rate": 8.135881792367686e-05,
+      "loss": 0.0135,
+      "step": 3200
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.27631258964538574,
+      "learning_rate": 8.12298616836904e-05,
+      "loss": 0.0111,
+      "step": 3210
+    },
+    {
+      "epoch": 30.093457943925234,
+      "grad_norm": 0.13083235919475555,
+      "learning_rate": 8.110056391894005e-05,
+      "loss": 0.0115,
+      "step": 3220
+    },
+    {
+      "epoch": 30.186915887850468,
+      "grad_norm": 0.21918922662734985,
+      "learning_rate": 8.097092604340542e-05,
+      "loss": 0.0112,
+      "step": 3230
+    },
+    {
+      "epoch": 30.2803738317757,
+      "grad_norm": 0.20092561841011047,
+      "learning_rate": 8.084094947478556e-05,
+      "loss": 0.0109,
+      "step": 3240
+    },
+    {
+      "epoch": 30.373831775700936,
+      "grad_norm": 0.09447859972715378,
+      "learning_rate": 8.07106356344834e-05,
+      "loss": 0.0127,
+      "step": 3250
+    },
+    {
+      "epoch": 30.46728971962617,
+      "grad_norm": 0.29426026344299316,
+      "learning_rate": 8.057998594759022e-05,
+      "loss": 0.0104,
+      "step": 3260
+    },
+    {
+      "epoch": 30.560747663551403,
+      "grad_norm": 0.2886331379413605,
+      "learning_rate": 8.044900184287007e-05,
+      "loss": 0.0133,
+      "step": 3270
+    },
+    {
+      "epoch": 30.654205607476637,
+      "grad_norm": 0.23450273275375366,
+      "learning_rate": 8.031768475274413e-05,
+      "loss": 0.0113,
+      "step": 3280
+    },
+    {
+      "epoch": 30.747663551401867,
+      "grad_norm": 0.2236676663160324,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.0144,
+      "step": 3290
+    },
+    {
+      "epoch": 30.8411214953271,
+      "grad_norm": 0.15358710289001465,
+      "learning_rate": 8.005405736415126e-05,
+      "loss": 0.0169,
+      "step": 3300
+    },
+    {
+      "epoch": 30.934579439252335,
+      "grad_norm": 0.20338520407676697,
+      "learning_rate": 7.992174994867123e-05,
+      "loss": 0.0095,
+      "step": 3310
+    },
+    {
+      "epoch": 31.02803738317757,
+      "grad_norm": 0.2024795562028885,
+      "learning_rate": 7.978911531372765e-05,
+      "loss": 0.0086,
+      "step": 3320
+    },
+    {
+      "epoch": 31.121495327102803,
+      "grad_norm": 0.22821472585201263,
+      "learning_rate": 7.965615490979163e-05,
+      "loss": 0.0132,
+      "step": 3330
+    },
+    {
+      "epoch": 31.214953271028037,
+      "grad_norm": 0.1402302086353302,
+      "learning_rate": 7.952287019089685e-05,
+      "loss": 0.0119,
+      "step": 3340
+    },
+    {
+      "epoch": 31.30841121495327,
+      "grad_norm": 0.10818405449390411,
+      "learning_rate": 7.938926261462366e-05,
+      "loss": 0.0115,
+      "step": 3350
+    },
+    {
+      "epoch": 31.401869158878505,
+      "grad_norm": 0.11518066376447678,
+      "learning_rate": 7.925533364208309e-05,
+      "loss": 0.013,
+      "step": 3360
+    },
+    {
+      "epoch": 31.49532710280374,
+      "grad_norm": 0.2154664248228073,
+      "learning_rate": 7.912108473790092e-05,
+      "loss": 0.0176,
+      "step": 3370
+    },
+    {
+      "epoch": 31.588785046728972,
+      "grad_norm": 0.1289885938167572,
+      "learning_rate": 7.898651737020166e-05,
+      "loss": 0.0099,
+      "step": 3380
+    },
+    {
+      "epoch": 31.682242990654206,
+      "grad_norm": 0.17317914962768555,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.0113,
+      "step": 3390
+    },
+    {
+      "epoch": 31.77570093457944,
+      "grad_norm": 0.26972246170043945,
+      "learning_rate": 7.871643313414718e-05,
+      "loss": 0.0115,
+      "step": 3400
+    },
+    {
+      "epoch": 31.869158878504674,
+      "grad_norm": 0.21863438189029694,
+      "learning_rate": 7.858091921938988e-05,
+      "loss": 0.0107,
+      "step": 3410
+    },
+    {
+      "epoch": 31.962616822429908,
+      "grad_norm": 0.3031831979751587,
+      "learning_rate": 7.844509274827907e-05,
+      "loss": 0.0117,
+      "step": 3420
+    },
+    {
+      "epoch": 32.05607476635514,
+      "grad_norm": 0.20491284132003784,
+      "learning_rate": 7.830895520619128e-05,
+      "loss": 0.0101,
+      "step": 3430
+    },
+    {
+      "epoch": 32.149532710280376,
+      "grad_norm": 0.27927613258361816,
+      "learning_rate": 7.817250808190483e-05,
+      "loss": 0.0176,
+      "step": 3440
+    },
+    {
+      "epoch": 32.242990654205606,
+      "grad_norm": 0.15980954468250275,
+      "learning_rate": 7.803575286758364e-05,
+      "loss": 0.013,
+      "step": 3450
+    },
+    {
+      "epoch": 32.33644859813084,
+      "grad_norm": 0.23138810694217682,
+      "learning_rate": 7.789869105876083e-05,
+      "loss": 0.0128,
+      "step": 3460
+    },
+    {
+      "epoch": 32.429906542056074,
+      "grad_norm": 0.2822995185852051,
+      "learning_rate": 7.776132415432234e-05,
+      "loss": 0.0119,
+      "step": 3470
+    },
+    {
+      "epoch": 32.52336448598131,
+      "grad_norm": 0.23915642499923706,
+      "learning_rate": 7.762365365649067e-05,
+      "loss": 0.0097,
+      "step": 3480
+    },
+    {
+      "epoch": 32.61682242990654,
+      "grad_norm": 0.2594238817691803,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.0154,
+      "step": 3490
+    },
+    {
+      "epoch": 32.71028037383178,
+      "grad_norm": 0.22654129564762115,
+      "learning_rate": 7.734740790612136e-05,
+      "loss": 0.0104,
+      "step": 3500
+    },
+    {
+      "epoch": 32.80373831775701,
+      "grad_norm": 0.2319777011871338,
+      "learning_rate": 7.720883567456298e-05,
+      "loss": 0.0155,
+      "step": 3510
+    },
+    {
+      "epoch": 32.89719626168224,
+      "grad_norm": 0.1848135143518448,
+      "learning_rate": 7.70699658915369e-05,
+      "loss": 0.0151,
+      "step": 3520
+    },
+    {
+      "epoch": 32.99065420560748,
+      "grad_norm": 0.1986227184534073,
+      "learning_rate": 7.693080007570084e-05,
+      "loss": 0.0086,
+      "step": 3530
+    },
+    {
+      "epoch": 33.08411214953271,
+      "grad_norm": 0.21213096380233765,
+      "learning_rate": 7.679133974894983e-05,
+      "loss": 0.0155,
+      "step": 3540
+    },
+    {
+      "epoch": 33.177570093457945,
+      "grad_norm": 0.33364537358283997,
+      "learning_rate": 7.66515864363997e-05,
+      "loss": 0.011,
+      "step": 3550
+    },
+    {
+      "epoch": 33.271028037383175,
+      "grad_norm": 0.1819266825914383,
+      "learning_rate": 7.651154166637025e-05,
+      "loss": 0.0123,
+      "step": 3560
+    },
+    {
+      "epoch": 33.36448598130841,
+      "grad_norm": 0.2073090821504593,
+      "learning_rate": 7.637120697036866e-05,
+      "loss": 0.0183,
+      "step": 3570
+    },
+    {
+      "epoch": 33.45794392523364,
+      "grad_norm": 0.2921043038368225,
+      "learning_rate": 7.623058388307269e-05,
+      "loss": 0.0156,
+      "step": 3580
+    },
+    {
+      "epoch": 33.55140186915888,
+      "grad_norm": 0.3672184348106384,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.0111,
+      "step": 3590
+    },
+    {
+      "epoch": 33.64485981308411,
+      "grad_norm": 0.3048618733882904,
+      "learning_rate": 7.594847868906076e-05,
+      "loss": 0.0126,
+      "step": 3600
+    },
+    {
+      "epoch": 33.73831775700935,
+      "grad_norm": 0.1969730406999588,
+      "learning_rate": 7.580699966740201e-05,
+      "loss": 0.0117,
+      "step": 3610
+    },
+    {
+      "epoch": 33.83177570093458,
+      "grad_norm": 0.1948702484369278,
+      "learning_rate": 7.566523842452958e-05,
+      "loss": 0.0098,
+      "step": 3620
+    },
+    {
+      "epoch": 33.925233644859816,
+      "grad_norm": 0.1653144657611847,
+      "learning_rate": 7.552319651072164e-05,
+      "loss": 0.0094,
+      "step": 3630
+    },
+    {
+      "epoch": 34.018691588785046,
+      "grad_norm": 0.2479052096605301,
+      "learning_rate": 7.538087547932585e-05,
+      "loss": 0.008,
+      "step": 3640
+    },
+    {
+      "epoch": 34.11214953271028,
+      "grad_norm": 0.07916168868541718,
+      "learning_rate": 7.52382768867422e-05,
+      "loss": 0.0113,
+      "step": 3650
+    },
+    {
+      "epoch": 34.205607476635514,
+      "grad_norm": 0.174769327044487,
+      "learning_rate": 7.509540229240601e-05,
+      "loss": 0.0093,
+      "step": 3660
+    },
+    {
+      "epoch": 34.299065420560744,
+      "grad_norm": 0.23629306256771088,
+      "learning_rate": 7.495225325877103e-05,
+      "loss": 0.0121,
+      "step": 3670
+    },
+    {
+      "epoch": 34.39252336448598,
+      "grad_norm": 0.2444513440132141,
+      "learning_rate": 7.480883135129211e-05,
+      "loss": 0.0088,
+      "step": 3680
+    },
+    {
+      "epoch": 34.48598130841121,
+      "grad_norm": 0.10653328895568848,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.0105,
+      "step": 3690
+    },
+    {
+      "epoch": 34.57943925233645,
+      "grad_norm": 0.22507068514823914,
+      "learning_rate": 7.452117519152542e-05,
+      "loss": 0.0113,
+      "step": 3700
+    },
+    {
+      "epoch": 34.67289719626168,
+      "grad_norm": 0.24139828979969025,
+      "learning_rate": 7.437694408499933e-05,
+      "loss": 0.0135,
+      "step": 3710
+    },
+    {
+      "epoch": 34.76635514018692,
+      "grad_norm": 0.22388507425785065,
+      "learning_rate": 7.423244639611826e-05,
+      "loss": 0.0124,
+      "step": 3720
+    },
+    {
+      "epoch": 34.85981308411215,
+      "grad_norm": 0.16689877212047577,
+      "learning_rate": 7.408768370508576e-05,
+      "loss": 0.0106,
+      "step": 3730
+    },
+    {
+      "epoch": 34.953271028037385,
+      "grad_norm": 0.37698590755462646,
+      "learning_rate": 7.394265759500348e-05,
+      "loss": 0.0151,
+      "step": 3740
+    },
+    {
+      "epoch": 35.046728971962615,
+      "grad_norm": 0.24599884450435638,
+      "learning_rate": 7.379736965185368e-05,
+      "loss": 0.011,
+      "step": 3750
+    },
+    {
+      "epoch": 35.14018691588785,
+      "grad_norm": 0.21924257278442383,
+      "learning_rate": 7.365182146448205e-05,
+      "loss": 0.008,
+      "step": 3760
+    },
+    {
+      "epoch": 35.23364485981308,
+      "grad_norm": 0.2563648521900177,
+      "learning_rate": 7.350601462458024e-05,
+      "loss": 0.0101,
+      "step": 3770
+    },
+    {
+      "epoch": 35.32710280373832,
+      "grad_norm": 0.18041463196277618,
+      "learning_rate": 7.335995072666848e-05,
+      "loss": 0.0098,
+      "step": 3780
+    },
+    {
+      "epoch": 35.42056074766355,
+      "grad_norm": 0.1949167102575302,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.0158,
+      "step": 3790
+    },
+    {
+      "epoch": 35.51401869158879,
+      "grad_norm": 0.21856240928173065,
+      "learning_rate": 7.30670581489344e-05,
+      "loss": 0.0175,
+      "step": 3800
+    },
+    {
+      "epoch": 35.60747663551402,
+      "grad_norm": 0.1561160832643509,
+      "learning_rate": 7.292023267213835e-05,
+      "loss": 0.0139,
+      "step": 3810
+    },
+    {
+      "epoch": 35.70093457943925,
+      "grad_norm": 0.26308417320251465,
+      "learning_rate": 7.277315654334997e-05,
+      "loss": 0.0121,
+      "step": 3820
+    },
+    {
+      "epoch": 35.794392523364486,
+      "grad_norm": 0.08401590585708618,
+      "learning_rate": 7.262583137097018e-05,
+      "loss": 0.0089,
+      "step": 3830
+    },
+    {
+      "epoch": 35.88785046728972,
+      "grad_norm": 0.21631336212158203,
+      "learning_rate": 7.247825876612353e-05,
+      "loss": 0.0128,
+      "step": 3840
+    },
+    {
+      "epoch": 35.981308411214954,
+      "grad_norm": 0.12208600342273712,
+      "learning_rate": 7.233044034264034e-05,
+      "loss": 0.0085,
+      "step": 3850
+    },
+    {
+      "epoch": 36.074766355140184,
+      "grad_norm": 0.29350805282592773,
+      "learning_rate": 7.218237771703921e-05,
+      "loss": 0.0134,
+      "step": 3860
+    },
+    {
+      "epoch": 36.16822429906542,
+      "grad_norm": 0.14051783084869385,
+      "learning_rate": 7.203407250850928e-05,
+      "loss": 0.0129,
+      "step": 3870
+    },
+    {
+      "epoch": 36.26168224299065,
+      "grad_norm": 0.17099681496620178,
+      "learning_rate": 7.188552633889259e-05,
+      "loss": 0.0114,
+      "step": 3880
+    },
+    {
+      "epoch": 36.35514018691589,
+      "grad_norm": 0.22999656200408936,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.0131,
+      "step": 3890
+    },
+    {
+      "epoch": 36.44859813084112,
+      "grad_norm": 0.29541051387786865,
+      "learning_rate": 7.158771761692464e-05,
+      "loss": 0.0114,
+      "step": 3900
+    },
+    {
+      "epoch": 36.54205607476636,
+      "grad_norm": 0.20923186838626862,
+      "learning_rate": 7.143845832136188e-05,
+      "loss": 0.0094,
+      "step": 3910
+    },
+    {
+      "epoch": 36.63551401869159,
+      "grad_norm": 0.21942150592803955,
+      "learning_rate": 7.128896457825364e-05,
+      "loss": 0.0105,
+      "step": 3920
+    },
+    {
+      "epoch": 36.728971962616825,
+      "grad_norm": 0.17471730709075928,
+      "learning_rate": 7.113923802243957e-05,
+      "loss": 0.0119,
+      "step": 3930
+    },
+    {
+      "epoch": 36.822429906542055,
+      "grad_norm": 0.1934959888458252,
+      "learning_rate": 7.09892802913053e-05,
+      "loss": 0.014,
+      "step": 3940
+    },
+    {
+      "epoch": 36.91588785046729,
+      "grad_norm": 0.26659610867500305,
+      "learning_rate": 7.083909302476453e-05,
+      "loss": 0.0155,
+      "step": 3950
+    },
+    {
+      "epoch": 37.00934579439252,
+      "grad_norm": 0.2176859825849533,
+      "learning_rate": 7.068867786524116e-05,
+      "loss": 0.0175,
+      "step": 3960
+    },
+    {
+      "epoch": 37.10280373831776,
+      "grad_norm": 0.17019227147102356,
+      "learning_rate": 7.053803645765128e-05,
+      "loss": 0.0116,
+      "step": 3970
+    },
+    {
+      "epoch": 37.19626168224299,
+      "grad_norm": 0.20591872930526733,
+      "learning_rate": 7.038717044938519e-05,
+      "loss": 0.0096,
+      "step": 3980
+    },
+    {
+      "epoch": 37.28971962616822,
+      "grad_norm": 0.1310962587594986,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.0093,
+      "step": 3990
+    },
+    {
+      "epoch": 37.38317757009346,
+      "grad_norm": 0.17182035744190216,
+      "learning_rate": 7.008477123264848e-05,
+      "loss": 0.0077,
+      "step": 4000
+    },
+    {
+      "epoch": 37.47663551401869,
+      "grad_norm": 0.2555645704269409,
+      "learning_rate": 6.993324133116726e-05,
+      "loss": 0.0097,
+      "step": 4010
+    },
+    {
+      "epoch": 37.570093457943926,
+      "grad_norm": 0.1435556560754776,
+      "learning_rate": 6.978149344295242e-05,
+      "loss": 0.0077,
+      "step": 4020
+    },
+    {
+      "epoch": 37.66355140186916,
+      "grad_norm": 0.13911636173725128,
+      "learning_rate": 6.962952922749457e-05,
+      "loss": 0.01,
+      "step": 4030
+    },
+    {
+      "epoch": 37.757009345794394,
+      "grad_norm": 0.2193230539560318,
+      "learning_rate": 6.947735034665002e-05,
+      "loss": 0.0133,
+      "step": 4040
+    },
+    {
+      "epoch": 37.850467289719624,
+      "grad_norm": 0.15190434455871582,
+      "learning_rate": 6.932495846462261e-05,
+      "loss": 0.0135,
+      "step": 4050
+    },
+    {
+      "epoch": 37.94392523364486,
+      "grad_norm": 0.14792603254318237,
+      "learning_rate": 6.917235524794558e-05,
+      "loss": 0.0132,
+      "step": 4060
+    },
+    {
+      "epoch": 38.03738317757009,
+      "grad_norm": 0.14730827510356903,
+      "learning_rate": 6.901954236546323e-05,
+      "loss": 0.0164,
+      "step": 4070
+    },
+    {
+      "epoch": 38.13084112149533,
+      "grad_norm": 0.2186291366815567,
+      "learning_rate": 6.886652148831279e-05,
+      "loss": 0.0102,
+      "step": 4080
+    },
+    {
+      "epoch": 38.22429906542056,
+      "grad_norm": 0.1827789694070816,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.0097,
+      "step": 4090
+    },
+    {
+      "epoch": 38.3177570093458,
+      "grad_norm": 0.18084844946861267,
+      "learning_rate": 6.855986244591104e-05,
+      "loss": 0.0101,
+      "step": 4100
+    },
+    {
+      "epoch": 38.41121495327103,
+      "grad_norm": 0.21352370083332062,
+      "learning_rate": 6.840622763423391e-05,
+      "loss": 0.0092,
+      "step": 4110
+    },
+    {
+      "epoch": 38.504672897196265,
+      "grad_norm": 0.09848731011152267,
+      "learning_rate": 6.825239153500029e-05,
+      "loss": 0.0098,
+      "step": 4120
+    },
+    {
+      "epoch": 38.598130841121495,
+      "grad_norm": 0.15104982256889343,
+      "learning_rate": 6.809835583053715e-05,
+      "loss": 0.0081,
+      "step": 4130
+    },
+    {
+      "epoch": 38.691588785046726,
+      "grad_norm": 0.18223416805267334,
+      "learning_rate": 6.794412220535426e-05,
+      "loss": 0.0111,
+      "step": 4140
+    },
+    {
+      "epoch": 38.78504672897196,
+      "grad_norm": 0.231476828455925,
+      "learning_rate": 6.778969234612584e-05,
+      "loss": 0.0092,
+      "step": 4150
+    },
+    {
+      "epoch": 38.87850467289719,
+      "grad_norm": 0.14167046546936035,
+      "learning_rate": 6.763506794167208e-05,
+      "loss": 0.0073,
+      "step": 4160
+    },
+    {
+      "epoch": 38.97196261682243,
+      "grad_norm": 0.12900766730308533,
+      "learning_rate": 6.748025068294067e-05,
+      "loss": 0.0076,
+      "step": 4170
+    },
+    {
+      "epoch": 39.06542056074766,
+      "grad_norm": 0.16524341702461243,
+      "learning_rate": 6.732524226298841e-05,
+      "loss": 0.0091,
+      "step": 4180
+    },
+    {
+      "epoch": 39.1588785046729,
+      "grad_norm": 0.17994999885559082,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.0099,
+      "step": 4190
+    },
+    {
+      "epoch": 39.25233644859813,
+      "grad_norm": 0.2520671784877777,
+      "learning_rate": 6.701465872208216e-05,
+      "loss": 0.0082,
+      "step": 4200
+    },
+    {
+      "epoch": 39.345794392523366,
+      "grad_norm": 0.2442283183336258,
+      "learning_rate": 6.685908699762002e-05,
+      "loss": 0.0098,
+      "step": 4210
+    },
+    {
+      "epoch": 39.4392523364486,
+      "grad_norm": 0.27867448329925537,
+      "learning_rate": 6.670333090488356e-05,
+      "loss": 0.011,
+      "step": 4220
+    },
+    {
+      "epoch": 39.532710280373834,
+      "grad_norm": 0.1637038290500641,
+      "learning_rate": 6.654739214719641e-05,
+      "loss": 0.0089,
+      "step": 4230
+    },
+    {
+      "epoch": 39.626168224299064,
+      "grad_norm": 0.20169733464717865,
+      "learning_rate": 6.639127242987988e-05,
+      "loss": 0.0084,
+      "step": 4240
+    },
+    {
+      "epoch": 39.7196261682243,
+      "grad_norm": 0.23685987293720245,
+      "learning_rate": 6.623497346023418e-05,
+      "loss": 0.0121,
+      "step": 4250
+    },
+    {
+      "epoch": 39.81308411214953,
+      "grad_norm": 0.09718881547451019,
+      "learning_rate": 6.607849694751977e-05,
+      "loss": 0.0068,
+      "step": 4260
+    },
+    {
+      "epoch": 39.90654205607477,
+      "grad_norm": 0.13366201519966125,
+      "learning_rate": 6.592184460293877e-05,
+      "loss": 0.009,
+      "step": 4270
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 0.23680470883846283,
+      "learning_rate": 6.576501813961609e-05,
+      "loss": 0.0103,
+      "step": 4280
+    },
+    {
+      "epoch": 40.09345794392523,
+      "grad_norm": 0.33618593215942383,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.0175,
+      "step": 4290
+    },
+    {
+      "epoch": 40.18691588785047,
+      "grad_norm": 0.18473008275032043,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 0.0083,
+      "step": 4300
+    },
+    {
+      "epoch": 40.2803738317757,
+      "grad_norm": 0.2896006405353546,
+      "learning_rate": 6.529351119689688e-05,
+      "loss": 0.0115,
+      "step": 4310
+    },
+    {
+      "epoch": 40.373831775700936,
+      "grad_norm": 0.3565766513347626,
+      "learning_rate": 6.513600542765817e-05,
+      "loss": 0.0085,
+      "step": 4320
+    },
+    {
+      "epoch": 40.467289719626166,
+      "grad_norm": 0.1844252496957779,
+      "learning_rate": 6.497833413348909e-05,
+      "loss": 0.0107,
+      "step": 4330
+    },
+    {
+      "epoch": 40.5607476635514,
+      "grad_norm": 0.19039011001586914,
+      "learning_rate": 6.48204990386577e-05,
+      "loss": 0.0068,
+      "step": 4340
+    },
+    {
+      "epoch": 40.654205607476634,
+      "grad_norm": 0.1492714285850525,
+      "learning_rate": 6.466250186922325e-05,
+      "loss": 0.0086,
+      "step": 4350
+    },
+    {
+      "epoch": 40.74766355140187,
+      "grad_norm": 0.1916590929031372,
+      "learning_rate": 6.450434435301751e-05,
+      "loss": 0.0144,
+      "step": 4360
+    },
+    {
+      "epoch": 40.8411214953271,
+      "grad_norm": 0.08804311603307724,
+      "learning_rate": 6.43460282196257e-05,
+      "loss": 0.0073,
+      "step": 4370
+    },
+    {
+      "epoch": 40.93457943925234,
+      "grad_norm": 0.18781927227973938,
+      "learning_rate": 6.418755520036775e-05,
+      "loss": 0.0096,
+      "step": 4380
+    },
+    {
+      "epoch": 41.02803738317757,
+      "grad_norm": 0.12092869728803635,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.0135,
+      "step": 4390
+    },
+    {
+      "epoch": 41.12149532710281,
+      "grad_norm": 0.08973784744739532,
+      "learning_rate": 6.387014543809223e-05,
+      "loss": 0.007,
+      "step": 4400
+    },
+    {
+      "epoch": 41.21495327102804,
+      "grad_norm": 0.2680470049381256,
+      "learning_rate": 6.371121216621698e-05,
+      "loss": 0.008,
+      "step": 4410
+    },
+    {
+      "epoch": 41.308411214953274,
+      "grad_norm": 0.30315253138542175,
+      "learning_rate": 6.355212895072223e-05,
+      "loss": 0.0087,
+      "step": 4420
+    },
+    {
+      "epoch": 41.401869158878505,
+      "grad_norm": 0.13689424097537994,
+      "learning_rate": 6.339289753131649e-05,
+      "loss": 0.0075,
+      "step": 4430
+    },
+    {
+      "epoch": 41.495327102803735,
+      "grad_norm": 0.16243036091327667,
+      "learning_rate": 6.323351964932908e-05,
+      "loss": 0.0092,
+      "step": 4440
+    },
+    {
+      "epoch": 41.58878504672897,
+      "grad_norm": 0.13021892309188843,
+      "learning_rate": 6.307399704769099e-05,
+      "loss": 0.0095,
+      "step": 4450
+    },
+    {
+      "epoch": 41.6822429906542,
+      "grad_norm": 0.27589356899261475,
+      "learning_rate": 6.291433147091583e-05,
+      "loss": 0.0108,
+      "step": 4460
+    },
+    {
+      "epoch": 41.77570093457944,
+      "grad_norm": 0.12132186442613602,
+      "learning_rate": 6.275452466508077e-05,
+      "loss": 0.0085,
+      "step": 4470
+    },
+    {
+      "epoch": 41.86915887850467,
+      "grad_norm": 0.1552329659461975,
+      "learning_rate": 6.259457837780742e-05,
+      "loss": 0.0064,
+      "step": 4480
+    },
+    {
+      "epoch": 41.96261682242991,
+      "grad_norm": 0.20105503499507904,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.0069,
+      "step": 4490
+    },
+    {
+      "epoch": 42.05607476635514,
+      "grad_norm": 0.24328036606311798,
+      "learning_rate": 6.227427435703997e-05,
+      "loss": 0.015,
+      "step": 4500
+    },
+    {
+      "epoch": 42.149532710280376,
+      "grad_norm": 0.27159082889556885,
+      "learning_rate": 6.211392012633932e-05,
+      "loss": 0.0084,
+      "step": 4510
+    },
+    {
+      "epoch": 42.242990654205606,
+      "grad_norm": 0.21687833964824677,
+      "learning_rate": 6.195343341974899e-05,
+      "loss": 0.0127,
+      "step": 4520
+    },
+    {
+      "epoch": 42.33644859813084,
+      "grad_norm": 0.22451163828372955,
+      "learning_rate": 6.179281599232591e-05,
+      "loss": 0.0112,
+      "step": 4530
+    },
+    {
+      "epoch": 42.429906542056074,
+      "grad_norm": 0.10428455471992493,
+      "learning_rate": 6.163206960055651e-05,
+      "loss": 0.0065,
+      "step": 4540
+    },
+    {
+      "epoch": 42.52336448598131,
+      "grad_norm": 0.2592661678791046,
+      "learning_rate": 6.147119600233758e-05,
+      "loss": 0.0138,
+      "step": 4550
+    },
+    {
+      "epoch": 42.61682242990654,
+      "grad_norm": 0.21353816986083984,
+      "learning_rate": 6.131019695695702e-05,
+      "loss": 0.0103,
+      "step": 4560
+    },
+    {
+      "epoch": 42.71028037383178,
+      "grad_norm": 0.12959344685077667,
+      "learning_rate": 6.11490742250746e-05,
+      "loss": 0.009,
+      "step": 4570
+    },
+    {
+      "epoch": 42.80373831775701,
+      "grad_norm": 0.17373885214328766,
+      "learning_rate": 6.0987829568702656e-05,
+      "loss": 0.0092,
+      "step": 4580
+    },
+    {
+      "epoch": 42.89719626168224,
+      "grad_norm": 0.11210908740758896,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.0079,
+      "step": 4590
+    },
+    {
+      "epoch": 42.99065420560748,
+      "grad_norm": 0.15213441848754883,
+      "learning_rate": 6.066498153718735e-05,
+      "loss": 0.0068,
+      "step": 4600
+    },
+    {
+      "epoch": 43.08411214953271,
+      "grad_norm": 0.20361490547657013,
+      "learning_rate": 6.05033816926583e-05,
+      "loss": 0.0149,
+      "step": 4610
+    },
+    {
+      "epoch": 43.177570093457945,
+      "grad_norm": 0.12980744242668152,
+      "learning_rate": 6.034166698482984e-05,
+      "loss": 0.012,
+      "step": 4620
+    },
+    {
+      "epoch": 43.271028037383175,
+      "grad_norm": 0.1619279980659485,
+      "learning_rate": 6.017983918218812e-05,
+      "loss": 0.0075,
+      "step": 4630
+    },
+    {
+      "epoch": 43.36448598130841,
+      "grad_norm": 0.10983559489250183,
+      "learning_rate": 6.001790005445607e-05,
+      "loss": 0.0084,
+      "step": 4640
+    },
+    {
+      "epoch": 43.45794392523364,
+      "grad_norm": 0.18888936936855316,
+      "learning_rate": 5.985585137257401e-05,
+      "loss": 0.0167,
+      "step": 4650
+    },
+    {
+      "epoch": 43.55140186915888,
+      "grad_norm": 0.3127266466617584,
+      "learning_rate": 5.969369490868042e-05,
+      "loss": 0.014,
+      "step": 4660
+    },
+    {
+      "epoch": 43.64485981308411,
+      "grad_norm": 0.21398553252220154,
+      "learning_rate": 5.953143243609235e-05,
+      "loss": 0.0109,
+      "step": 4670
+    },
+    {
+      "epoch": 43.73831775700935,
+      "grad_norm": 0.1972840428352356,
+      "learning_rate": 5.9369065729286245e-05,
+      "loss": 0.0078,
+      "step": 4680
+    },
+    {
+      "epoch": 43.83177570093458,
+      "grad_norm": 0.17708909511566162,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.0136,
+      "step": 4690
+    },
+    {
+      "epoch": 43.925233644859816,
+      "grad_norm": 0.25454819202423096,
+      "learning_rate": 5.90440267166055e-05,
+      "loss": 0.0107,
+      "step": 4700
+    },
+    {
+      "epoch": 44.018691588785046,
+      "grad_norm": 0.1953362226486206,
+      "learning_rate": 5.888135796530544e-05,
+      "loss": 0.0113,
+      "step": 4710
+    },
+    {
+      "epoch": 44.11214953271028,
+      "grad_norm": 0.21734946966171265,
+      "learning_rate": 5.871859208889759e-05,
+      "loss": 0.007,
+      "step": 4720
+    },
+    {
+      "epoch": 44.205607476635514,
+      "grad_norm": 0.13528119027614594,
+      "learning_rate": 5.85557308673635e-05,
+      "loss": 0.0061,
+      "step": 4730
+    },
+    {
+      "epoch": 44.299065420560744,
+      "grad_norm": 0.09264456480741501,
+      "learning_rate": 5.8392776081727385e-05,
+      "loss": 0.0115,
+      "step": 4740
+    },
+    {
+      "epoch": 44.39252336448598,
+      "grad_norm": 0.09667211025953293,
+      "learning_rate": 5.8229729514036705e-05,
+      "loss": 0.0066,
+      "step": 4750
+    },
+    {
+      "epoch": 44.48598130841121,
+      "grad_norm": 0.18434172868728638,
+      "learning_rate": 5.8066592947342555e-05,
+      "loss": 0.0091,
+      "step": 4760
+    },
+    {
+      "epoch": 44.57943925233645,
+      "grad_norm": 0.2720284163951874,
+      "learning_rate": 5.7903368165680327e-05,
+      "loss": 0.0113,
+      "step": 4770
+    },
+    {
+      "epoch": 44.67289719626168,
+      "grad_norm": 0.23133063316345215,
+      "learning_rate": 5.7740056954050084e-05,
+      "loss": 0.0101,
+      "step": 4780
+    },
+    {
+      "epoch": 44.76635514018692,
+      "grad_norm": 0.28980597853660583,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.0128,
+      "step": 4790
+    },
+    {
+      "epoch": 44.85981308411215,
+      "grad_norm": 0.2679378390312195,
+      "learning_rate": 5.74131823855921e-05,
+      "loss": 0.0075,
+      "step": 4800
+    },
+    {
+      "epoch": 44.953271028037385,
+      "grad_norm": 0.16744691133499146,
+      "learning_rate": 5.72496226034123e-05,
+      "loss": 0.0081,
+      "step": 4810
+    },
+    {
+      "epoch": 45.046728971962615,
+      "grad_norm": 0.2577405869960785,
+      "learning_rate": 5.7085983540521216e-05,
+      "loss": 0.0068,
+      "step": 4820
+    },
+    {
+      "epoch": 45.14018691588785,
+      "grad_norm": 0.1690078228712082,
+      "learning_rate": 5.692226698644938e-05,
+      "loss": 0.0086,
+      "step": 4830
+    },
+    {
+      "epoch": 45.23364485981308,
+      "grad_norm": 0.11636698246002197,
+      "learning_rate": 5.675847473157485e-05,
+      "loss": 0.0092,
+      "step": 4840
+    },
+    {
+      "epoch": 45.32710280373832,
+      "grad_norm": 0.15227118134498596,
+      "learning_rate": 5.6594608567103456e-05,
+      "loss": 0.0075,
+      "step": 4850
+    },
+    {
+      "epoch": 45.42056074766355,
+      "grad_norm": 0.126779705286026,
+      "learning_rate": 5.6430670285049314e-05,
+      "loss": 0.0075,
+      "step": 4860
+    },
+    {
+      "epoch": 45.51401869158879,
+      "grad_norm": 0.23741380870342255,
+      "learning_rate": 5.6266661678215216e-05,
+      "loss": 0.0098,
+      "step": 4870
+    },
+    {
+      "epoch": 45.60747663551402,
+      "grad_norm": 0.12933692336082458,
+      "learning_rate": 5.6102584540173006e-05,
+      "loss": 0.0075,
+      "step": 4880
+    },
+    {
+      "epoch": 45.70093457943925,
+      "grad_norm": 0.1902732253074646,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.0087,
+      "step": 4890
+    },
+    {
+      "epoch": 45.794392523364486,
+      "grad_norm": 0.24616940319538116,
+      "learning_rate": 5.577423184847932e-05,
+      "loss": 0.0077,
+      "step": 4900
+    },
+    {
+      "epoch": 45.88785046728972,
+      "grad_norm": 0.16900432109832764,
+      "learning_rate": 5.560995988564023e-05,
+      "loss": 0.0097,
+      "step": 4910
+    },
+    {
+      "epoch": 45.981308411214954,
+      "grad_norm": 0.20487163960933685,
+      "learning_rate": 5.544562657317863e-05,
+      "loss": 0.0114,
+      "step": 4920
+    },
+    {
+      "epoch": 46.074766355140184,
+      "grad_norm": 0.25133049488067627,
+      "learning_rate": 5.52812337082173e-05,
+      "loss": 0.0088,
+      "step": 4930
+    },
+    {
+      "epoch": 46.16822429906542,
+      "grad_norm": 0.1800190508365631,
+      "learning_rate": 5.511678308853026e-05,
+      "loss": 0.0096,
+      "step": 4940
+    },
+    {
+      "epoch": 46.26168224299065,
+      "grad_norm": 0.2863096594810486,
+      "learning_rate": 5.495227651252315e-05,
+      "loss": 0.0076,
+      "step": 4950
+    },
+    {
+      "epoch": 46.35514018691589,
+      "grad_norm": 0.3172850012779236,
+      "learning_rate": 5.478771577921351e-05,
+      "loss": 0.0104,
+      "step": 4960
+    },
+    {
+      "epoch": 46.44859813084112,
+      "grad_norm": 0.13900935649871826,
+      "learning_rate": 5.462310268821118e-05,
+      "loss": 0.0077,
+      "step": 4970
+    },
+    {
+      "epoch": 46.54205607476636,
+      "grad_norm": 0.22267590463161469,
+      "learning_rate": 5.445843903969854e-05,
+      "loss": 0.0059,
+      "step": 4980
+    },
+    {
+      "epoch": 46.63551401869159,
+      "grad_norm": 0.16322097182273865,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.0104,
+      "step": 4990
+    },
+    {
+      "epoch": 46.728971962616825,
+      "grad_norm": 0.18399694561958313,
+      "learning_rate": 5.4128967273616625e-05,
+      "loss": 0.0106,
+      "step": 5000
+    },
+    {
+      "epoch": 46.822429906542055,
+      "grad_norm": 0.1907566636800766,
+      "learning_rate": 5.396416275909779e-05,
+      "loss": 0.0089,
+      "step": 5010
+    },
+    {
+      "epoch": 46.91588785046729,
+      "grad_norm": 0.2239440530538559,
+      "learning_rate": 5.379931489313016e-05,
+      "loss": 0.0086,
+      "step": 5020
+    },
+    {
+      "epoch": 47.00934579439252,
+      "grad_norm": 0.16346921026706696,
+      "learning_rate": 5.363442547846356e-05,
+      "loss": 0.0089,
+      "step": 5030
+    },
+    {
+      "epoch": 47.10280373831776,
+      "grad_norm": 0.14416854083538055,
+      "learning_rate": 5.3469496318302204e-05,
+      "loss": 0.0065,
+      "step": 5040
+    },
+    {
+      "epoch": 47.19626168224299,
+      "grad_norm": 0.18693040311336517,
+      "learning_rate": 5.330452921628497e-05,
+      "loss": 0.0077,
+      "step": 5050
+    },
+    {
+      "epoch": 47.28971962616822,
+      "grad_norm": 0.10827270895242691,
+      "learning_rate": 5.313952597646568e-05,
+      "loss": 0.0056,
+      "step": 5060
+    },
+    {
+      "epoch": 47.38317757009346,
+      "grad_norm": 0.15888318419456482,
+      "learning_rate": 5.297448840329329e-05,
+      "loss": 0.0087,
+      "step": 5070
+    },
+    {
+      "epoch": 47.47663551401869,
+      "grad_norm": 0.11162968724966049,
+      "learning_rate": 5.280941830159227e-05,
+      "loss": 0.0093,
+      "step": 5080
+    },
+    {
+      "epoch": 47.570093457943926,
+      "grad_norm": 0.11820860952138901,
+      "learning_rate": 5.264431747654284e-05,
+      "loss": 0.0108,
+      "step": 5090
+    },
+    {
+      "epoch": 47.66355140186916,
+      "grad_norm": 0.3276582956314087,
+      "learning_rate": 5.247918773366112e-05,
+      "loss": 0.0131,
+      "step": 5100
+    },
+    {
+      "epoch": 47.757009345794394,
+      "grad_norm": 0.14899931848049164,
+      "learning_rate": 5.231403087877955e-05,
+      "loss": 0.0074,
+      "step": 5110
+    },
+    {
+      "epoch": 47.850467289719624,
+      "grad_norm": 0.07620976120233536,
+      "learning_rate": 5.214884871802703e-05,
+      "loss": 0.009,
+      "step": 5120
+    },
+    {
+      "epoch": 47.94392523364486,
+      "grad_norm": 0.18864719569683075,
+      "learning_rate": 5.198364305780922e-05,
+      "loss": 0.0064,
+      "step": 5130
+    },
+    {
+      "epoch": 48.03738317757009,
+      "grad_norm": 0.13755668699741364,
+      "learning_rate": 5.1818415704788725e-05,
+      "loss": 0.0085,
+      "step": 5140
+    },
+    {
+      "epoch": 48.13084112149533,
+      "grad_norm": 0.22894635796546936,
+      "learning_rate": 5.165316846586541e-05,
+      "loss": 0.0113,
+      "step": 5150
+    },
+    {
+      "epoch": 48.22429906542056,
+      "grad_norm": 0.11449041217565536,
+      "learning_rate": 5.148790314815663e-05,
+      "loss": 0.0082,
+      "step": 5160
+    },
+    {
+      "epoch": 48.3177570093458,
+      "grad_norm": 0.1928773671388626,
+      "learning_rate": 5.132262155897739e-05,
+      "loss": 0.0084,
+      "step": 5170
+    },
+    {
+      "epoch": 48.41121495327103,
+      "grad_norm": 0.15431612730026245,
+      "learning_rate": 5.1157325505820694e-05,
+      "loss": 0.0096,
+      "step": 5180
+    },
+    {
+      "epoch": 48.504672897196265,
+      "grad_norm": 0.21660259366035461,
+      "learning_rate": 5.0992016796337686e-05,
+      "loss": 0.008,
+      "step": 5190
+    },
+    {
+      "epoch": 48.598130841121495,
+      "grad_norm": 0.11220486462116241,
+      "learning_rate": 5.0826697238317935e-05,
+      "loss": 0.0072,
+      "step": 5200
+    },
+    {
+      "epoch": 48.691588785046726,
+      "grad_norm": 0.15329080820083618,
+      "learning_rate": 5.066136863966963e-05,
+      "loss": 0.0088,
+      "step": 5210
+    },
+    {
+      "epoch": 48.78504672897196,
+      "grad_norm": 0.1777033805847168,
+      "learning_rate": 5.0496032808399815e-05,
+      "loss": 0.0052,
+      "step": 5220
+    },
+    {
+      "epoch": 48.87850467289719,
+      "grad_norm": 0.23826682567596436,
+      "learning_rate": 5.033069155259471e-05,
+      "loss": 0.0094,
+      "step": 5230
+    },
+    {
+      "epoch": 48.97196261682243,
+      "grad_norm": 0.13108570873737335,
+      "learning_rate": 5.016534668039976e-05,
+      "loss": 0.0062,
+      "step": 5240
+    },
+    {
+      "epoch": 49.06542056074766,
+      "grad_norm": 0.1880958378314972,
+      "learning_rate": 5e-05,
+      "loss": 0.0053,
+      "step": 5250
+    },
+    {
+      "epoch": 49.1588785046729,
+      "grad_norm": 0.11604796350002289,
+      "learning_rate": 4.9834653319600246e-05,
+      "loss": 0.0083,
+      "step": 5260
+    },
+    {
+      "epoch": 49.25233644859813,
+      "grad_norm": 0.10272664576768875,
+      "learning_rate": 4.96693084474053e-05,
+      "loss": 0.0081,
+      "step": 5270
+    },
+    {
+      "epoch": 49.345794392523366,
+      "grad_norm": 0.1421612948179245,
+      "learning_rate": 4.950396719160018e-05,
+      "loss": 0.005,
+      "step": 5280
+    },
+    {
+      "epoch": 49.4392523364486,
+      "grad_norm": 0.11580304056406021,
+      "learning_rate": 4.93386313603304e-05,
+      "loss": 0.0071,
+      "step": 5290
+    },
+    {
+      "epoch": 49.532710280373834,
+      "grad_norm": 0.0858597457408905,
+      "learning_rate": 4.917330276168208e-05,
+      "loss": 0.0107,
+      "step": 5300
+    },
+    {
+      "epoch": 49.626168224299064,
+      "grad_norm": 0.17755255103111267,
+      "learning_rate": 4.9007983203662326e-05,
+      "loss": 0.0188,
+      "step": 5310
+    },
+    {
+      "epoch": 49.7196261682243,
+      "grad_norm": 0.14021337032318115,
+      "learning_rate": 4.884267449417931e-05,
+      "loss": 0.0081,
+      "step": 5320
+    },
+    {
+      "epoch": 49.81308411214953,
+      "grad_norm": 0.10999365150928497,
+      "learning_rate": 4.867737844102261e-05,
+      "loss": 0.0103,
+      "step": 5330
+    },
+    {
+      "epoch": 49.90654205607477,
+      "grad_norm": 0.11541043221950531,
+      "learning_rate": 4.851209685184338e-05,
+      "loss": 0.0072,
+      "step": 5340
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 0.10045769810676575,
+      "learning_rate": 4.834683153413459e-05,
+      "loss": 0.0146,
+      "step": 5350
+    },
+    {
+      "epoch": 50.09345794392523,
+      "grad_norm": 0.1158151924610138,
+      "learning_rate": 4.818158429521129e-05,
+      "loss": 0.0078,
+      "step": 5360
+    },
+    {
+      "epoch": 50.18691588785047,
+      "grad_norm": 0.1430526077747345,
+      "learning_rate": 4.801635694219079e-05,
+      "loss": 0.0054,
+      "step": 5370
+    },
+    {
+      "epoch": 50.2803738317757,
+      "grad_norm": 0.09269680827856064,
+      "learning_rate": 4.785115128197298e-05,
+      "loss": 0.0097,
+      "step": 5380
+    },
+    {
+      "epoch": 50.373831775700936,
+      "grad_norm": 0.21225248277187347,
+      "learning_rate": 4.7685969121220456e-05,
+      "loss": 0.0098,
+      "step": 5390
+    },
+    {
+      "epoch": 50.467289719626166,
+      "grad_norm": 0.18716934323310852,
+      "learning_rate": 4.7520812266338885e-05,
+      "loss": 0.0083,
+      "step": 5400
+    },
+    {
+      "epoch": 50.5607476635514,
+      "grad_norm": 0.24767927825450897,
+      "learning_rate": 4.735568252345718e-05,
+      "loss": 0.0075,
+      "step": 5410
+    },
+    {
+      "epoch": 50.654205607476634,
+      "grad_norm": 0.14961543679237366,
+      "learning_rate": 4.7190581698407725e-05,
+      "loss": 0.0106,
+      "step": 5420
+    },
+    {
+      "epoch": 50.74766355140187,
+      "grad_norm": 0.16245384514331818,
+      "learning_rate": 4.702551159670672e-05,
+      "loss": 0.0062,
+      "step": 5430
+    },
+    {
+      "epoch": 50.8411214953271,
+      "grad_norm": 0.1727730631828308,
+      "learning_rate": 4.6860474023534335e-05,
+      "loss": 0.0086,
+      "step": 5440
+    },
+    {
+      "epoch": 50.93457943925234,
+      "grad_norm": 0.1440618336200714,
+      "learning_rate": 4.669547078371504e-05,
+      "loss": 0.009,
+      "step": 5450
+    },
+    {
+      "epoch": 51.02803738317757,
+      "grad_norm": 0.17170698940753937,
+      "learning_rate": 4.65305036816978e-05,
+      "loss": 0.0067,
+      "step": 5460
+    },
+    {
+      "epoch": 51.12149532710281,
+      "grad_norm": 0.3083714246749878,
+      "learning_rate": 4.6365574521536445e-05,
+      "loss": 0.0114,
+      "step": 5470
+    },
+    {
+      "epoch": 51.21495327102804,
+      "grad_norm": 0.1776435673236847,
+      "learning_rate": 4.620068510686985e-05,
+      "loss": 0.0054,
+      "step": 5480
+    },
+    {
+      "epoch": 51.308411214953274,
+      "grad_norm": 0.13433292508125305,
+      "learning_rate": 4.60358372409022e-05,
+      "loss": 0.0065,
+      "step": 5490
+    },
+    {
+      "epoch": 51.401869158878505,
+      "grad_norm": 0.1641828864812851,
+      "learning_rate": 4.5871032726383386e-05,
+      "loss": 0.0059,
+      "step": 5500
+    },
+    {
+      "epoch": 51.495327102803735,
+      "grad_norm": 0.13108348846435547,
+      "learning_rate": 4.570627336558915e-05,
+      "loss": 0.0069,
+      "step": 5510
+    },
+    {
+      "epoch": 51.58878504672897,
+      "grad_norm": 0.10418945550918579,
+      "learning_rate": 4.554156096030149e-05,
+      "loss": 0.007,
+      "step": 5520
+    },
+    {
+      "epoch": 51.6822429906542,
+      "grad_norm": 0.1613224595785141,
+      "learning_rate": 4.537689731178883e-05,
+      "loss": 0.0087,
+      "step": 5530
+    },
+    {
+      "epoch": 51.77570093457944,
+      "grad_norm": 0.10617467761039734,
+      "learning_rate": 4.5212284220786494e-05,
+      "loss": 0.0068,
+      "step": 5540
+    },
+    {
+      "epoch": 51.86915887850467,
+      "grad_norm": 0.18668167293071747,
+      "learning_rate": 4.504772348747687e-05,
+      "loss": 0.0071,
+      "step": 5550
+    },
+    {
+      "epoch": 51.96261682242991,
+      "grad_norm": 0.1861657351255417,
+      "learning_rate": 4.488321691146975e-05,
+      "loss": 0.0078,
+      "step": 5560
+    },
+    {
+      "epoch": 52.05607476635514,
+      "grad_norm": 0.16021041572093964,
+      "learning_rate": 4.471876629178273e-05,
+      "loss": 0.0084,
+      "step": 5570
+    },
+    {
+      "epoch": 52.149532710280376,
+      "grad_norm": 0.16291090846061707,
+      "learning_rate": 4.4554373426821374e-05,
+      "loss": 0.0053,
+      "step": 5580
+    },
+    {
+      "epoch": 52.242990654205606,
+      "grad_norm": 0.19529825448989868,
+      "learning_rate": 4.439004011435979e-05,
+      "loss": 0.0069,
+      "step": 5590
+    },
+    {
+      "epoch": 52.33644859813084,
+      "grad_norm": 0.10670498758554459,
+      "learning_rate": 4.4225768151520694e-05,
+      "loss": 0.0091,
+      "step": 5600
+    },
+    {
+      "epoch": 52.429906542056074,
+      "grad_norm": 0.14729143679141998,
+      "learning_rate": 4.406155933475599e-05,
+      "loss": 0.0073,
+      "step": 5610
+    },
+    {
+      "epoch": 52.52336448598131,
+      "grad_norm": 0.10457372665405273,
+      "learning_rate": 4.3897415459827e-05,
+      "loss": 0.0103,
+      "step": 5620
+    },
+    {
+      "epoch": 52.61682242990654,
+      "grad_norm": 0.09747841209173203,
+      "learning_rate": 4.373333832178478e-05,
+      "loss": 0.0044,
+      "step": 5630
+    },
+    {
+      "epoch": 52.71028037383178,
+      "grad_norm": 0.2818126082420349,
+      "learning_rate": 4.3569329714950704e-05,
+      "loss": 0.0111,
+      "step": 5640
+    },
+    {
+      "epoch": 52.80373831775701,
+      "grad_norm": 0.1272769421339035,
+      "learning_rate": 4.3405391432896555e-05,
+      "loss": 0.0069,
+      "step": 5650
+    },
+    {
+      "epoch": 52.89719626168224,
+      "grad_norm": 0.08397116512060165,
+      "learning_rate": 4.324152526842517e-05,
+      "loss": 0.007,
+      "step": 5660
+    },
+    {
+      "epoch": 52.99065420560748,
+      "grad_norm": 0.10440504550933838,
+      "learning_rate": 4.307773301355062e-05,
+      "loss": 0.0052,
+      "step": 5670
+    },
+    {
+      "epoch": 53.08411214953271,
+      "grad_norm": 0.16689011454582214,
+      "learning_rate": 4.291401645947879e-05,
+      "loss": 0.0111,
+      "step": 5680
+    },
+    {
+      "epoch": 53.177570093457945,
+      "grad_norm": 0.19110624492168427,
+      "learning_rate": 4.275037739658771e-05,
+      "loss": 0.0045,
+      "step": 5690
+    },
+    {
+      "epoch": 53.271028037383175,
+      "grad_norm": 0.2127784937620163,
+      "learning_rate": 4.2586817614407895e-05,
+      "loss": 0.0096,
+      "step": 5700
+    },
+    {
+      "epoch": 53.36448598130841,
+      "grad_norm": 0.15279392898082733,
+      "learning_rate": 4.2423338901602985e-05,
+      "loss": 0.0084,
+      "step": 5710
+    },
+    {
+      "epoch": 53.45794392523364,
+      "grad_norm": 0.14837054908275604,
+      "learning_rate": 4.2259943045949934e-05,
+      "loss": 0.0061,
+      "step": 5720
+    },
+    {
+      "epoch": 53.55140186915888,
+      "grad_norm": 0.15841516852378845,
+      "learning_rate": 4.209663183431969e-05,
+      "loss": 0.0076,
+      "step": 5730
+    },
+    {
+      "epoch": 53.64485981308411,
+      "grad_norm": 0.13824456930160522,
+      "learning_rate": 4.1933407052657456e-05,
+      "loss": 0.0051,
+      "step": 5740
+    },
+    {
+      "epoch": 53.73831775700935,
+      "grad_norm": 0.16600170731544495,
+      "learning_rate": 4.17702704859633e-05,
+      "loss": 0.0065,
+      "step": 5750
+    },
+    {
+      "epoch": 53.83177570093458,
+      "grad_norm": 0.0763145163655281,
+      "learning_rate": 4.160722391827262e-05,
+      "loss": 0.0055,
+      "step": 5760
+    },
+    {
+      "epoch": 53.925233644859816,
+      "grad_norm": 0.1596927046775818,
+      "learning_rate": 4.14442691326365e-05,
+      "loss": 0.0054,
+      "step": 5770
+    },
+    {
+      "epoch": 54.018691588785046,
+      "grad_norm": 0.07967574149370193,
+      "learning_rate": 4.1281407911102425e-05,
+      "loss": 0.0074,
+      "step": 5780
+    },
+    {
+      "epoch": 54.11214953271028,
+      "grad_norm": 0.12622042000293732,
+      "learning_rate": 4.111864203469457e-05,
+      "loss": 0.0071,
+      "step": 5790
+    },
+    {
+      "epoch": 54.205607476635514,
+      "grad_norm": 0.19472530484199524,
+      "learning_rate": 4.095597328339452e-05,
+      "loss": 0.0065,
+      "step": 5800
+    },
+    {
+      "epoch": 54.299065420560744,
+      "grad_norm": 0.17462767660617828,
+      "learning_rate": 4.079340343612165e-05,
+      "loss": 0.0116,
+      "step": 5810
+    },
+    {
+      "epoch": 54.39252336448598,
+      "grad_norm": 0.1130964532494545,
+      "learning_rate": 4.063093427071376e-05,
+      "loss": 0.0116,
+      "step": 5820
+    },
+    {
+      "epoch": 54.48598130841121,
+      "grad_norm": 0.17297130823135376,
+      "learning_rate": 4.046856756390767e-05,
+      "loss": 0.0073,
+      "step": 5830
+    },
+    {
+      "epoch": 54.57943925233645,
+      "grad_norm": 0.1260395348072052,
+      "learning_rate": 4.0306305091319595e-05,
+      "loss": 0.0066,
+      "step": 5840
+    },
+    {
+      "epoch": 54.67289719626168,
+      "grad_norm": 0.06664054840803146,
+      "learning_rate": 4.0144148627425993e-05,
+      "loss": 0.0064,
+      "step": 5850
+    },
+    {
+      "epoch": 54.76635514018692,
+      "grad_norm": 0.12369495630264282,
+      "learning_rate": 3.9982099945543945e-05,
+      "loss": 0.0072,
+      "step": 5860
+    },
+    {
+      "epoch": 54.85981308411215,
+      "grad_norm": 0.12784737348556519,
+      "learning_rate": 3.982016081781189e-05,
+      "loss": 0.0054,
+      "step": 5870
+    },
+    {
+      "epoch": 54.953271028037385,
+      "grad_norm": 0.1303785741329193,
+      "learning_rate": 3.965833301517017e-05,
+      "loss": 0.0084,
+      "step": 5880
+    },
+    {
+      "epoch": 55.046728971962615,
+      "grad_norm": 0.3063434660434723,
+      "learning_rate": 3.949661830734172e-05,
+      "loss": 0.0064,
+      "step": 5890
+    },
+    {
+      "epoch": 55.14018691588785,
+      "grad_norm": 0.16514930129051208,
+      "learning_rate": 3.933501846281267e-05,
+      "loss": 0.0072,
+      "step": 5900
+    },
+    {
+      "epoch": 55.23364485981308,
+      "grad_norm": 0.13724902272224426,
+      "learning_rate": 3.917353524881302e-05,
+      "loss": 0.0094,
+      "step": 5910
+    },
+    {
+      "epoch": 55.32710280373832,
+      "grad_norm": 0.2272414267063141,
+      "learning_rate": 3.901217043129735e-05,
+      "loss": 0.0095,
+      "step": 5920
+    },
+    {
+      "epoch": 55.42056074766355,
+      "grad_norm": 0.19817429780960083,
+      "learning_rate": 3.8850925774925425e-05,
+      "loss": 0.0087,
+      "step": 5930
+    },
+    {
+      "epoch": 55.51401869158879,
+      "grad_norm": 0.15857623517513275,
+      "learning_rate": 3.8689803043043e-05,
+      "loss": 0.0097,
+      "step": 5940
+    },
+    {
+      "epoch": 55.60747663551402,
+      "grad_norm": 0.10527165979146957,
+      "learning_rate": 3.852880399766243e-05,
+      "loss": 0.0059,
+      "step": 5950
+    },
+    {
+      "epoch": 55.70093457943925,
+      "grad_norm": 0.13487043976783752,
+      "learning_rate": 3.836793039944349e-05,
+      "loss": 0.0069,
+      "step": 5960
+    },
+    {
+      "epoch": 55.794392523364486,
+      "grad_norm": 0.12190820276737213,
+      "learning_rate": 3.820718400767409e-05,
+      "loss": 0.0061,
+      "step": 5970
+    },
+    {
+      "epoch": 55.88785046728972,
+      "grad_norm": 0.15790730714797974,
+      "learning_rate": 3.8046566580251e-05,
+      "loss": 0.0049,
+      "step": 5980
+    },
+    {
+      "epoch": 55.981308411214954,
+      "grad_norm": 0.25264978408813477,
+      "learning_rate": 3.788607987366069e-05,
+      "loss": 0.007,
+      "step": 5990
+    },
+    {
+      "epoch": 56.074766355140184,
+      "grad_norm": 0.23202531039714813,
+      "learning_rate": 3.772572564296005e-05,
+      "loss": 0.0098,
+      "step": 6000
+    },
+    {
+      "epoch": 56.16822429906542,
+      "grad_norm": 0.13732583820819855,
+      "learning_rate": 3.756550564175727e-05,
+      "loss": 0.007,
+      "step": 6010
+    },
+    {
+      "epoch": 56.26168224299065,
+      "grad_norm": 0.20250828564167023,
+      "learning_rate": 3.74054216221926e-05,
+      "loss": 0.0081,
+      "step": 6020
+    },
+    {
+      "epoch": 56.35514018691589,
+      "grad_norm": 0.08914332091808319,
+      "learning_rate": 3.7245475334919246e-05,
+      "loss": 0.0082,
+      "step": 6030
+    },
+    {
+      "epoch": 56.44859813084112,
+      "grad_norm": 0.21319356560707092,
+      "learning_rate": 3.7085668529084184e-05,
+      "loss": 0.0103,
+      "step": 6040
+    },
+    {
+      "epoch": 56.54205607476636,
+      "grad_norm": 0.13997061550617218,
+      "learning_rate": 3.6926002952309016e-05,
+      "loss": 0.0066,
+      "step": 6050
+    },
+    {
+      "epoch": 56.63551401869159,
+      "grad_norm": 0.14500273764133453,
+      "learning_rate": 3.676648035067093e-05,
+      "loss": 0.0062,
+      "step": 6060
+    },
+    {
+      "epoch": 56.728971962616825,
+      "grad_norm": 0.1436031460762024,
+      "learning_rate": 3.6607102468683526e-05,
+      "loss": 0.0061,
+      "step": 6070
+    },
+    {
+      "epoch": 56.822429906542055,
+      "grad_norm": 0.12058055400848389,
+      "learning_rate": 3.6447871049277796e-05,
+      "loss": 0.0055,
+      "step": 6080
+    },
+    {
+      "epoch": 56.91588785046729,
+      "grad_norm": 0.1439819186925888,
+      "learning_rate": 3.628878783378302e-05,
+      "loss": 0.0087,
+      "step": 6090
+    },
+    {
+      "epoch": 57.00934579439252,
+      "grad_norm": 0.09871205687522888,
+      "learning_rate": 3.612985456190778e-05,
+      "loss": 0.0087,
+      "step": 6100
+    },
+    {
+      "epoch": 57.10280373831776,
+      "grad_norm": 0.05546163022518158,
+      "learning_rate": 3.597107297172084e-05,
+      "loss": 0.0101,
+      "step": 6110
+    },
+    {
+      "epoch": 57.19626168224299,
+      "grad_norm": 0.05012049525976181,
+      "learning_rate": 3.581244479963225e-05,
+      "loss": 0.0063,
+      "step": 6120
+    },
+    {
+      "epoch": 57.28971962616822,
+      "grad_norm": 0.23360881209373474,
+      "learning_rate": 3.5653971780374295e-05,
+      "loss": 0.0103,
+      "step": 6130
+    },
+    {
+      "epoch": 57.38317757009346,
+      "grad_norm": 0.1165015920996666,
+      "learning_rate": 3.5495655646982505e-05,
+      "loss": 0.0061,
+      "step": 6140
+    },
+    {
+      "epoch": 57.47663551401869,
+      "grad_norm": 0.16637849807739258,
+      "learning_rate": 3.533749813077677e-05,
+      "loss": 0.0063,
+      "step": 6150
+    },
+    {
+      "epoch": 57.570093457943926,
+      "grad_norm": 0.2202228456735611,
+      "learning_rate": 3.517950096134232e-05,
+      "loss": 0.0084,
+      "step": 6160
+    },
+    {
+      "epoch": 57.66355140186916,
+      "grad_norm": 0.20240144431591034,
+      "learning_rate": 3.5021665866510925e-05,
+      "loss": 0.0073,
+      "step": 6170
+    },
+    {
+      "epoch": 57.757009345794394,
+      "grad_norm": 0.11815023422241211,
+      "learning_rate": 3.4863994572341843e-05,
+      "loss": 0.0067,
+      "step": 6180
+    },
+    {
+      "epoch": 57.850467289719624,
+      "grad_norm": 0.1485767364501953,
+      "learning_rate": 3.470648880310313e-05,
+      "loss": 0.005,
+      "step": 6190
+    },
+    {
+      "epoch": 57.94392523364486,
+      "grad_norm": 0.16621701419353485,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 0.0077,
+      "step": 6200
+    },
+    {
+      "epoch": 58.03738317757009,
+      "grad_norm": 0.14100301265716553,
+      "learning_rate": 3.439198072741921e-05,
+      "loss": 0.0067,
+      "step": 6210
+    },
+    {
+      "epoch": 58.13084112149533,
+      "grad_norm": 0.10529860854148865,
+      "learning_rate": 3.423498186038393e-05,
+      "loss": 0.011,
+      "step": 6220
+    },
+    {
+      "epoch": 58.22429906542056,
+      "grad_norm": 0.23530688881874084,
+      "learning_rate": 3.407815539706124e-05,
+      "loss": 0.0087,
+      "step": 6230
+    },
+    {
+      "epoch": 58.3177570093458,
+      "grad_norm": 0.06678273528814316,
+      "learning_rate": 3.392150305248024e-05,
+      "loss": 0.0055,
+      "step": 6240
+    },
+    {
+      "epoch": 58.41121495327103,
+      "grad_norm": 0.08229216933250427,
+      "learning_rate": 3.3765026539765834e-05,
+      "loss": 0.0064,
+      "step": 6250
+    },
+    {
+      "epoch": 58.504672897196265,
+      "grad_norm": 0.16697847843170166,
+      "learning_rate": 3.360872757012011e-05,
+      "loss": 0.0092,
+      "step": 6260
+    },
+    {
+      "epoch": 58.598130841121495,
+      "grad_norm": 0.20726190507411957,
+      "learning_rate": 3.3452607852803584e-05,
+      "loss": 0.0087,
+      "step": 6270
+    },
+    {
+      "epoch": 58.691588785046726,
+      "grad_norm": 0.16936185956001282,
+      "learning_rate": 3.329666909511645e-05,
+      "loss": 0.0066,
+      "step": 6280
+    },
+    {
+      "epoch": 58.78504672897196,
+      "grad_norm": 0.15490904450416565,
+      "learning_rate": 3.3140913002379995e-05,
+      "loss": 0.0058,
+      "step": 6290
+    },
+    {
+      "epoch": 58.87850467289719,
+      "grad_norm": 0.12930889427661896,
+      "learning_rate": 3.298534127791785e-05,
+      "loss": 0.0078,
+      "step": 6300
+    },
+    {
+      "epoch": 58.97196261682243,
+      "grad_norm": 0.10294876247644424,
+      "learning_rate": 3.282995562303754e-05,
+      "loss": 0.0047,
+      "step": 6310
+    },
+    {
+      "epoch": 59.06542056074766,
+      "grad_norm": 0.20665724575519562,
+      "learning_rate": 3.267475773701161e-05,
+      "loss": 0.0075,
+      "step": 6320
+    },
+    {
+      "epoch": 59.1588785046729,
+      "grad_norm": 0.32335957884788513,
+      "learning_rate": 3.251974931705933e-05,
+      "loss": 0.0083,
+      "step": 6330
+    },
+    {
+      "epoch": 59.25233644859813,
+      "grad_norm": 0.16983294486999512,
+      "learning_rate": 3.236493205832795e-05,
+      "loss": 0.007,
+      "step": 6340
+    },
+    {
+      "epoch": 59.345794392523366,
+      "grad_norm": 0.22015579044818878,
+      "learning_rate": 3.221030765387417e-05,
+      "loss": 0.0071,
+      "step": 6350
+    },
+    {
+      "epoch": 59.4392523364486,
+      "grad_norm": 0.18638084828853607,
+      "learning_rate": 3.205587779464576e-05,
+      "loss": 0.0055,
+      "step": 6360
+    },
+    {
+      "epoch": 59.532710280373834,
+      "grad_norm": 0.10701462626457214,
+      "learning_rate": 3.190164416946285e-05,
+      "loss": 0.0059,
+      "step": 6370
+    },
+    {
+      "epoch": 59.626168224299064,
+      "grad_norm": 0.15403886139392853,
+      "learning_rate": 3.1747608464999725e-05,
+      "loss": 0.0081,
+      "step": 6380
+    },
+    {
+      "epoch": 59.7196261682243,
+      "grad_norm": 0.1373993307352066,
+      "learning_rate": 3.1593772365766105e-05,
+      "loss": 0.0091,
+      "step": 6390
+    },
+    {
+      "epoch": 59.81308411214953,
+      "grad_norm": 0.08877486735582352,
+      "learning_rate": 3.144013755408895e-05,
+      "loss": 0.0071,
+      "step": 6400
+    },
+    {
+      "epoch": 59.90654205607477,
+      "grad_norm": 0.10183851420879364,
+      "learning_rate": 3.128670571009399e-05,
+      "loss": 0.0075,
+      "step": 6410
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 0.1058053970336914,
+      "learning_rate": 3.113347851168721e-05,
+      "loss": 0.0071,
+      "step": 6420
+    },
+    {
+      "epoch": 60.09345794392523,
+      "grad_norm": 0.16160264611244202,
+      "learning_rate": 3.098045763453678e-05,
+      "loss": 0.0064,
+      "step": 6430
+    },
+    {
+      "epoch": 60.18691588785047,
+      "grad_norm": 0.13019979000091553,
+      "learning_rate": 3.082764475205442e-05,
+      "loss": 0.0049,
+      "step": 6440
+    },
+    {
+      "epoch": 60.2803738317757,
+      "grad_norm": 0.0628226175904274,
+      "learning_rate": 3.0675041535377405e-05,
+      "loss": 0.0085,
+      "step": 6450
+    },
+    {
+      "epoch": 60.373831775700936,
+      "grad_norm": 0.10675778985023499,
+      "learning_rate": 3.052264965335e-05,
+      "loss": 0.0041,
+      "step": 6460
+    },
+    {
+      "epoch": 60.467289719626166,
+      "grad_norm": 0.1395292580127716,
+      "learning_rate": 3.0370470772505433e-05,
+      "loss": 0.006,
+      "step": 6470
+    },
+    {
+      "epoch": 60.5607476635514,
+      "grad_norm": 0.14375652372837067,
+      "learning_rate": 3.0218506557047598e-05,
+      "loss": 0.0042,
+      "step": 6480
+    },
+    {
+      "epoch": 60.654205607476634,
+      "grad_norm": 0.06682026386260986,
+      "learning_rate": 3.006675866883275e-05,
+      "loss": 0.0042,
+      "step": 6490
+    },
+    {
+      "epoch": 60.74766355140187,
+      "grad_norm": 0.15054212510585785,
+      "learning_rate": 2.991522876735154e-05,
+      "loss": 0.0051,
+      "step": 6500
+    },
+    {
+      "epoch": 60.8411214953271,
+      "grad_norm": 0.22600547969341278,
+      "learning_rate": 2.976391850971065e-05,
+      "loss": 0.0153,
+      "step": 6510
+    },
+    {
+      "epoch": 60.93457943925234,
+      "grad_norm": 0.2411378026008606,
+      "learning_rate": 2.9612829550614836e-05,
+      "loss": 0.0075,
+      "step": 6520
+    },
+    {
+      "epoch": 61.02803738317757,
+      "grad_norm": 0.21554498374462128,
+      "learning_rate": 2.9461963542348737e-05,
+      "loss": 0.0048,
+      "step": 6530
+    },
+    {
+      "epoch": 61.12149532710281,
+      "grad_norm": 0.23806604743003845,
+      "learning_rate": 2.931132213475884e-05,
+      "loss": 0.0093,
+      "step": 6540
+    },
+    {
+      "epoch": 61.21495327102804,
+      "grad_norm": 0.10214415937662125,
+      "learning_rate": 2.916090697523549e-05,
+      "loss": 0.0069,
+      "step": 6550
+    },
+    {
+      "epoch": 61.308411214953274,
+      "grad_norm": 0.0819956436753273,
+      "learning_rate": 2.9010719708694722e-05,
+      "loss": 0.0039,
+      "step": 6560
+    },
+    {
+      "epoch": 61.401869158878505,
+      "grad_norm": 0.09245266020298004,
+      "learning_rate": 2.8860761977560436e-05,
+      "loss": 0.0077,
+      "step": 6570
+    },
+    {
+      "epoch": 61.495327102803735,
+      "grad_norm": 0.08326391875743866,
+      "learning_rate": 2.8711035421746367e-05,
+      "loss": 0.0045,
+      "step": 6580
+    },
+    {
+      "epoch": 61.58878504672897,
+      "grad_norm": 0.1491008847951889,
+      "learning_rate": 2.8561541678638142e-05,
+      "loss": 0.0061,
+      "step": 6590
+    },
+    {
+      "epoch": 61.6822429906542,
+      "grad_norm": 0.30525025725364685,
+      "learning_rate": 2.8412282383075363e-05,
+      "loss": 0.0052,
+      "step": 6600
+    },
+    {
+      "epoch": 61.77570093457944,
+      "grad_norm": 0.09467942267656326,
+      "learning_rate": 2.8263259167333777e-05,
+      "loss": 0.0033,
+      "step": 6610
+    },
+    {
+      "epoch": 61.86915887850467,
+      "grad_norm": 0.1267421990633011,
+      "learning_rate": 2.811447366110741e-05,
+      "loss": 0.0079,
+      "step": 6620
+    },
+    {
+      "epoch": 61.96261682242991,
+      "grad_norm": 0.13594943284988403,
+      "learning_rate": 2.7965927491490705e-05,
+      "loss": 0.0049,
+      "step": 6630
+    },
+    {
+      "epoch": 62.05607476635514,
+      "grad_norm": 0.08311861008405685,
+      "learning_rate": 2.7817622282960815e-05,
+      "loss": 0.0039,
+      "step": 6640
+    },
+    {
+      "epoch": 62.149532710280376,
+      "grad_norm": 0.1365499198436737,
+      "learning_rate": 2.766955965735968e-05,
+      "loss": 0.0043,
+      "step": 6650
+    },
+    {
+      "epoch": 62.242990654205606,
+      "grad_norm": 0.2722339630126953,
+      "learning_rate": 2.7521741233876496e-05,
+      "loss": 0.007,
+      "step": 6660
+    },
+    {
+      "epoch": 62.33644859813084,
+      "grad_norm": 0.09474354982376099,
+      "learning_rate": 2.7374168629029813e-05,
+      "loss": 0.0058,
+      "step": 6670
+    },
+    {
+      "epoch": 62.429906542056074,
+      "grad_norm": 0.08654358983039856,
+      "learning_rate": 2.7226843456650037e-05,
+      "loss": 0.0045,
+      "step": 6680
+    },
+    {
+      "epoch": 62.52336448598131,
+      "grad_norm": 0.16651228070259094,
+      "learning_rate": 2.707976732786166e-05,
+      "loss": 0.0089,
+      "step": 6690
+    },
+    {
+      "epoch": 62.61682242990654,
+      "grad_norm": 0.10585249215364456,
+      "learning_rate": 2.693294185106562e-05,
+      "loss": 0.0082,
+      "step": 6700
+    },
+    {
+      "epoch": 62.71028037383178,
+      "grad_norm": 0.09274085611104965,
+      "learning_rate": 2.6786368631921836e-05,
+      "loss": 0.0051,
+      "step": 6710
+    },
+    {
+      "epoch": 62.80373831775701,
+      "grad_norm": 0.12623822689056396,
+      "learning_rate": 2.6640049273331515e-05,
+      "loss": 0.0045,
+      "step": 6720
+    },
+    {
+      "epoch": 62.89719626168224,
+      "grad_norm": 0.11692936718463898,
+      "learning_rate": 2.6493985375419778e-05,
+      "loss": 0.0093,
+      "step": 6730
+    },
+    {
+      "epoch": 62.99065420560748,
+      "grad_norm": 0.22904883325099945,
+      "learning_rate": 2.6348178535517966e-05,
+      "loss": 0.0121,
+      "step": 6740
+    },
+    {
+      "epoch": 63.08411214953271,
+      "grad_norm": 0.13337840139865875,
+      "learning_rate": 2.6202630348146324e-05,
+      "loss": 0.0049,
+      "step": 6750
+    },
+    {
+      "epoch": 63.177570093457945,
+      "grad_norm": 0.1077757328748703,
+      "learning_rate": 2.6057342404996522e-05,
+      "loss": 0.0093,
+      "step": 6760
+    },
+    {
+      "epoch": 63.271028037383175,
+      "grad_norm": 0.12814705073833466,
+      "learning_rate": 2.591231629491423e-05,
+      "loss": 0.009,
+      "step": 6770
+    },
+    {
+      "epoch": 63.36448598130841,
+      "grad_norm": 0.09796059876680374,
+      "learning_rate": 2.5767553603881767e-05,
+      "loss": 0.0067,
+      "step": 6780
+    },
+    {
+      "epoch": 63.45794392523364,
+      "grad_norm": 0.12925131618976593,
+      "learning_rate": 2.562305591500069e-05,
+      "loss": 0.005,
+      "step": 6790
+    },
+    {
+      "epoch": 63.55140186915888,
+      "grad_norm": 0.12728284299373627,
+      "learning_rate": 2.547882480847461e-05,
+      "loss": 0.0106,
+      "step": 6800
+    },
+    {
+      "epoch": 63.64485981308411,
+      "grad_norm": 0.16028623282909393,
+      "learning_rate": 2.5334861861591753e-05,
+      "loss": 0.0084,
+      "step": 6810
+    },
+    {
+      "epoch": 63.73831775700935,
+      "grad_norm": 0.13662315905094147,
+      "learning_rate": 2.5191168648707887e-05,
+      "loss": 0.0051,
+      "step": 6820
+    },
+    {
+      "epoch": 63.83177570093458,
+      "grad_norm": 0.15248225629329681,
+      "learning_rate": 2.5047746741228978e-05,
+      "loss": 0.0074,
+      "step": 6830
+    },
+    {
+      "epoch": 63.925233644859816,
+      "grad_norm": 0.07730412483215332,
+      "learning_rate": 2.490459770759398e-05,
+      "loss": 0.0046,
+      "step": 6840
+    },
+    {
+      "epoch": 64.01869158878505,
+      "grad_norm": 0.15421777963638306,
+      "learning_rate": 2.476172311325783e-05,
+      "loss": 0.006,
+      "step": 6850
+    },
+    {
+      "epoch": 64.11214953271028,
+      "grad_norm": 0.07014045119285583,
+      "learning_rate": 2.4619124520674146e-05,
+      "loss": 0.0043,
+      "step": 6860
+    },
+    {
+      "epoch": 64.20560747663552,
+      "grad_norm": 0.13341595232486725,
+      "learning_rate": 2.447680348927837e-05,
+      "loss": 0.0041,
+      "step": 6870
+    },
+    {
+      "epoch": 64.29906542056075,
+      "grad_norm": 0.3038124442100525,
+      "learning_rate": 2.433476157547044e-05,
+      "loss": 0.0105,
+      "step": 6880
+    },
+    {
+      "epoch": 64.39252336448598,
+      "grad_norm": 0.12947747111320496,
+      "learning_rate": 2.419300033259798e-05,
+      "loss": 0.004,
+      "step": 6890
+    },
+    {
+      "epoch": 64.48598130841121,
+      "grad_norm": 0.12829992175102234,
+      "learning_rate": 2.405152131093926e-05,
+      "loss": 0.0051,
+      "step": 6900
+    },
+    {
+      "epoch": 64.57943925233644,
+      "grad_norm": 0.10023665428161621,
+      "learning_rate": 2.3910326057686127e-05,
+      "loss": 0.0096,
+      "step": 6910
+    },
+    {
+      "epoch": 64.67289719626169,
+      "grad_norm": 0.19918078184127808,
+      "learning_rate": 2.3769416116927335e-05,
+      "loss": 0.0076,
+      "step": 6920
+    },
+    {
+      "epoch": 64.76635514018692,
+      "grad_norm": 0.1109958365559578,
+      "learning_rate": 2.362879302963135e-05,
+      "loss": 0.0048,
+      "step": 6930
+    },
+    {
+      "epoch": 64.85981308411215,
+      "grad_norm": 0.07922573387622833,
+      "learning_rate": 2.3488458333629777e-05,
+      "loss": 0.0047,
+      "step": 6940
+    },
+    {
+      "epoch": 64.95327102803738,
+      "grad_norm": 0.11461448669433594,
+      "learning_rate": 2.3348413563600325e-05,
+      "loss": 0.0057,
+      "step": 6950
+    },
+    {
+      "epoch": 65.04672897196262,
+      "grad_norm": 0.13653631508350372,
+      "learning_rate": 2.3208660251050158e-05,
+      "loss": 0.0056,
+      "step": 6960
+    },
+    {
+      "epoch": 65.14018691588785,
+      "grad_norm": 0.17430157959461212,
+      "learning_rate": 2.3069199924299174e-05,
+      "loss": 0.0062,
+      "step": 6970
+    },
+    {
+      "epoch": 65.23364485981308,
+      "grad_norm": 0.245283842086792,
+      "learning_rate": 2.29300341084631e-05,
+      "loss": 0.0054,
+      "step": 6980
+    },
+    {
+      "epoch": 65.32710280373831,
+      "grad_norm": 0.14271025359630585,
+      "learning_rate": 2.279116432543705e-05,
+      "loss": 0.0127,
+      "step": 6990
+    },
+    {
+      "epoch": 65.42056074766356,
+      "grad_norm": 0.12084395438432693,
+      "learning_rate": 2.2652592093878666e-05,
+      "loss": 0.0062,
+      "step": 7000
+    },
+    {
+      "epoch": 65.51401869158879,
+      "grad_norm": 0.12945696711540222,
+      "learning_rate": 2.251431892919171e-05,
+      "loss": 0.0071,
+      "step": 7010
+    },
+    {
+      "epoch": 65.60747663551402,
+      "grad_norm": 0.10073532909154892,
+      "learning_rate": 2.237634634350934e-05,
+      "loss": 0.0049,
+      "step": 7020
+    },
+    {
+      "epoch": 65.70093457943925,
+      "grad_norm": 0.0651387944817543,
+      "learning_rate": 2.2238675845677663e-05,
+      "loss": 0.0048,
+      "step": 7030
+    },
+    {
+      "epoch": 65.79439252336448,
+      "grad_norm": 0.19357287883758545,
+      "learning_rate": 2.2101308941239203e-05,
+      "loss": 0.0056,
+      "step": 7040
+    },
+    {
+      "epoch": 65.88785046728972,
+      "grad_norm": 0.0946953147649765,
+      "learning_rate": 2.196424713241637e-05,
+      "loss": 0.0058,
+      "step": 7050
+    },
+    {
+      "epoch": 65.98130841121495,
+      "grad_norm": 0.129776269197464,
+      "learning_rate": 2.182749191809518e-05,
+      "loss": 0.0051,
+      "step": 7060
+    },
+    {
+      "epoch": 66.07476635514018,
+      "grad_norm": 0.10389594733715057,
+      "learning_rate": 2.1691044793808734e-05,
+      "loss": 0.0079,
+      "step": 7070
+    },
+    {
+      "epoch": 66.16822429906541,
+      "grad_norm": 0.16596883535385132,
+      "learning_rate": 2.1554907251720945e-05,
+      "loss": 0.0078,
+      "step": 7080
+    },
+    {
+      "epoch": 66.26168224299066,
+      "grad_norm": 0.14865995943546295,
+      "learning_rate": 2.1419080780610123e-05,
+      "loss": 0.005,
+      "step": 7090
+    },
+    {
+      "epoch": 66.35514018691589,
+      "grad_norm": 0.16961994767189026,
+      "learning_rate": 2.128356686585282e-05,
+      "loss": 0.0072,
+      "step": 7100
+    },
+    {
+      "epoch": 66.44859813084112,
+      "grad_norm": 0.11203950643539429,
+      "learning_rate": 2.1148366989407496e-05,
+      "loss": 0.006,
+      "step": 7110
+    },
+    {
+      "epoch": 66.54205607476635,
+      "grad_norm": 0.14505212008953094,
+      "learning_rate": 2.1013482629798333e-05,
+      "loss": 0.0047,
+      "step": 7120
+    },
+    {
+      "epoch": 66.6355140186916,
+      "grad_norm": 0.08906478434801102,
+      "learning_rate": 2.0878915262099098e-05,
+      "loss": 0.0058,
+      "step": 7130
+    },
+    {
+      "epoch": 66.72897196261682,
+      "grad_norm": 0.23099195957183838,
+      "learning_rate": 2.0744666357916925e-05,
+      "loss": 0.0059,
+      "step": 7140
+    },
+    {
+      "epoch": 66.82242990654206,
+      "grad_norm": 0.10707362741231918,
+      "learning_rate": 2.061073738537635e-05,
+      "loss": 0.005,
+      "step": 7150
+    },
+    {
+      "epoch": 66.91588785046729,
+      "grad_norm": 0.055628702044487,
+      "learning_rate": 2.0477129809103147e-05,
+      "loss": 0.009,
+      "step": 7160
+    },
+    {
+      "epoch": 67.00934579439253,
+      "grad_norm": 0.15333709120750427,
+      "learning_rate": 2.0343845090208368e-05,
+      "loss": 0.0047,
+      "step": 7170
+    },
+    {
+      "epoch": 67.10280373831776,
+      "grad_norm": 0.130838081240654,
+      "learning_rate": 2.0210884686272368e-05,
+      "loss": 0.0052,
+      "step": 7180
+    },
+    {
+      "epoch": 67.19626168224299,
+      "grad_norm": 0.08488751947879791,
+      "learning_rate": 2.0078250051328784e-05,
+      "loss": 0.0045,
+      "step": 7190
+    },
+    {
+      "epoch": 67.28971962616822,
+      "grad_norm": 0.15652534365653992,
+      "learning_rate": 1.9945942635848748e-05,
+      "loss": 0.0073,
+      "step": 7200
+    },
+    {
+      "epoch": 67.38317757009345,
+      "grad_norm": 0.08217427134513855,
+      "learning_rate": 1.981396388672496e-05,
+      "loss": 0.0039,
+      "step": 7210
+    },
+    {
+      "epoch": 67.4766355140187,
+      "grad_norm": 0.11081071197986603,
+      "learning_rate": 1.9682315247255894e-05,
+      "loss": 0.004,
+      "step": 7220
+    },
+    {
+      "epoch": 67.57009345794393,
+      "grad_norm": 0.14157460629940033,
+      "learning_rate": 1.9550998157129946e-05,
+      "loss": 0.005,
+      "step": 7230
+    },
+    {
+      "epoch": 67.66355140186916,
+      "grad_norm": 0.13172607123851776,
+      "learning_rate": 1.942001405240979e-05,
+      "loss": 0.0049,
+      "step": 7240
+    },
+    {
+      "epoch": 67.75700934579439,
+      "grad_norm": 0.1288451850414276,
+      "learning_rate": 1.928936436551661e-05,
+      "loss": 0.0063,
+      "step": 7250
+    },
+    {
+      "epoch": 67.85046728971963,
+      "grad_norm": 0.061469387263059616,
+      "learning_rate": 1.9159050525214452e-05,
+      "loss": 0.0043,
+      "step": 7260
+    },
+    {
+      "epoch": 67.94392523364486,
+      "grad_norm": 0.19150938093662262,
+      "learning_rate": 1.9029073956594606e-05,
+      "loss": 0.0077,
+      "step": 7270
+    },
+    {
+      "epoch": 68.03738317757009,
+      "grad_norm": 0.11590924859046936,
+      "learning_rate": 1.8899436081059975e-05,
+      "loss": 0.0039,
+      "step": 7280
+    },
+    {
+      "epoch": 68.13084112149532,
+      "grad_norm": 0.18850106000900269,
+      "learning_rate": 1.877013831630961e-05,
+      "loss": 0.0045,
+      "step": 7290
+    },
+    {
+      "epoch": 68.22429906542057,
+      "grad_norm": 0.07713674753904343,
+      "learning_rate": 1.8641182076323148e-05,
+      "loss": 0.0108,
+      "step": 7300
+    },
+    {
+      "epoch": 68.3177570093458,
+      "grad_norm": 0.15120600163936615,
+      "learning_rate": 1.851256877134538e-05,
+      "loss": 0.0038,
+      "step": 7310
+    },
+    {
+      "epoch": 68.41121495327103,
+      "grad_norm": 0.16040542721748352,
+      "learning_rate": 1.838429980787081e-05,
+      "loss": 0.0036,
+      "step": 7320
+    },
+    {
+      "epoch": 68.50467289719626,
+      "grad_norm": 0.10460657626390457,
+      "learning_rate": 1.8256376588628238e-05,
+      "loss": 0.0045,
+      "step": 7330
+    },
+    {
+      "epoch": 68.59813084112149,
+      "grad_norm": 0.13424362242221832,
+      "learning_rate": 1.8128800512565513e-05,
+      "loss": 0.0055,
+      "step": 7340
+    },
+    {
+      "epoch": 68.69158878504673,
+      "grad_norm": 0.18600434064865112,
+      "learning_rate": 1.800157297483417e-05,
+      "loss": 0.0063,
+      "step": 7350
+    },
+    {
+      "epoch": 68.78504672897196,
+      "grad_norm": 0.08261007815599442,
+      "learning_rate": 1.787469536677419e-05,
+      "loss": 0.0075,
+      "step": 7360
+    },
+    {
+      "epoch": 68.8785046728972,
+      "grad_norm": 0.08533482998609543,
+      "learning_rate": 1.774816907589873e-05,
+      "loss": 0.0072,
+      "step": 7370
+    },
+    {
+      "epoch": 68.97196261682242,
+      "grad_norm": 0.06679965555667877,
+      "learning_rate": 1.7621995485879062e-05,
+      "loss": 0.0064,
+      "step": 7380
+    },
+    {
+      "epoch": 69.06542056074767,
+      "grad_norm": 0.0962221547961235,
+      "learning_rate": 1.749617597652934e-05,
+      "loss": 0.0048,
+      "step": 7390
+    },
+    {
+      "epoch": 69.1588785046729,
+      "grad_norm": 0.15369288623332977,
+      "learning_rate": 1.7370711923791567e-05,
+      "loss": 0.0043,
+      "step": 7400
+    },
+    {
+      "epoch": 69.25233644859813,
+      "grad_norm": 0.09784958511590958,
+      "learning_rate": 1.7245604699720535e-05,
+      "loss": 0.0045,
+      "step": 7410
+    },
+    {
+      "epoch": 69.34579439252336,
+      "grad_norm": 0.04342583566904068,
+      "learning_rate": 1.712085567246878e-05,
+      "loss": 0.0156,
+      "step": 7420
+    },
+    {
+      "epoch": 69.4392523364486,
+      "grad_norm": 0.09830091148614883,
+      "learning_rate": 1.699646620627168e-05,
+      "loss": 0.0092,
+      "step": 7430
+    },
+    {
+      "epoch": 69.53271028037383,
+      "grad_norm": 0.15314213931560516,
+      "learning_rate": 1.6872437661432517e-05,
+      "loss": 0.0048,
+      "step": 7440
+    },
+    {
+      "epoch": 69.62616822429906,
+      "grad_norm": 0.1179867535829544,
+      "learning_rate": 1.6748771394307585e-05,
+      "loss": 0.0045,
+      "step": 7450
+    },
+    {
+      "epoch": 69.7196261682243,
+      "grad_norm": 0.10799465328454971,
+      "learning_rate": 1.662546875729138e-05,
+      "loss": 0.0045,
+      "step": 7460
+    },
+    {
+      "epoch": 69.81308411214954,
+      "grad_norm": 0.07715349644422531,
+      "learning_rate": 1.6502531098801753e-05,
+      "loss": 0.0061,
+      "step": 7470
+    },
+    {
+      "epoch": 69.90654205607477,
+      "grad_norm": 0.2066328078508377,
+      "learning_rate": 1.637995976326527e-05,
+      "loss": 0.0121,
+      "step": 7480
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 0.10284323245286942,
+      "learning_rate": 1.62577560911024e-05,
+      "loss": 0.0056,
+      "step": 7490
+    },
+    {
+      "epoch": 70.09345794392523,
+      "grad_norm": 0.11130893975496292,
+      "learning_rate": 1.6135921418712956e-05,
+      "loss": 0.0058,
+      "step": 7500
+    },
+    {
+      "epoch": 70.18691588785046,
+      "grad_norm": 0.19642308354377747,
+      "learning_rate": 1.6014457078461353e-05,
+      "loss": 0.0038,
+      "step": 7510
+    },
+    {
+      "epoch": 70.2803738317757,
+      "grad_norm": 0.08608820289373398,
+      "learning_rate": 1.5893364398662176e-05,
+      "loss": 0.0082,
+      "step": 7520
+    },
+    {
+      "epoch": 70.37383177570094,
+      "grad_norm": 0.13513289391994476,
+      "learning_rate": 1.5772644703565565e-05,
+      "loss": 0.0106,
+      "step": 7530
+    },
+    {
+      "epoch": 70.46728971962617,
+      "grad_norm": 0.0810285359621048,
+      "learning_rate": 1.5652299313342773e-05,
+      "loss": 0.0059,
+      "step": 7540
+    },
+    {
+      "epoch": 70.5607476635514,
+      "grad_norm": 0.0977608859539032,
+      "learning_rate": 1.553232954407171e-05,
+      "loss": 0.0054,
+      "step": 7550
+    },
+    {
+      "epoch": 70.65420560747664,
+      "grad_norm": 0.0938141942024231,
+      "learning_rate": 1.5412736707722537e-05,
+      "loss": 0.0042,
+      "step": 7560
+    },
+    {
+      "epoch": 70.74766355140187,
+      "grad_norm": 0.06880947202444077,
+      "learning_rate": 1.5293522112143373e-05,
+      "loss": 0.0036,
+      "step": 7570
+    },
+    {
+      "epoch": 70.8411214953271,
+      "grad_norm": 0.1963202804327011,
+      "learning_rate": 1.517468706104589e-05,
+      "loss": 0.0055,
+      "step": 7580
+    },
+    {
+      "epoch": 70.93457943925233,
+      "grad_norm": 0.054219506680965424,
+      "learning_rate": 1.5056232853991209e-05,
+      "loss": 0.0069,
+      "step": 7590
+    },
+    {
+      "epoch": 71.02803738317758,
+      "grad_norm": 0.1340658813714981,
+      "learning_rate": 1.4938160786375572e-05,
+      "loss": 0.006,
+      "step": 7600
+    },
+    {
+      "epoch": 71.1214953271028,
+      "grad_norm": 0.1702129989862442,
+      "learning_rate": 1.4820472149416154e-05,
+      "loss": 0.0051,
+      "step": 7610
+    },
+    {
+      "epoch": 71.21495327102804,
+      "grad_norm": 0.09247569739818573,
+      "learning_rate": 1.470316823013707e-05,
+      "loss": 0.0029,
+      "step": 7620
+    },
+    {
+      "epoch": 71.30841121495327,
+      "grad_norm": 0.1598488688468933,
+      "learning_rate": 1.4586250311355132e-05,
+      "loss": 0.0109,
+      "step": 7630
+    },
+    {
+      "epoch": 71.40186915887851,
+      "grad_norm": 0.07022323459386826,
+      "learning_rate": 1.4469719671666043e-05,
+      "loss": 0.0038,
+      "step": 7640
+    },
+    {
+      "epoch": 71.49532710280374,
+      "grad_norm": 0.08927827328443527,
+      "learning_rate": 1.435357758543015e-05,
+      "loss": 0.0065,
+      "step": 7650
+    },
+    {
+      "epoch": 71.58878504672897,
+      "grad_norm": 0.06398759037256241,
+      "learning_rate": 1.4237825322758736e-05,
+      "loss": 0.0066,
+      "step": 7660
+    },
+    {
+      "epoch": 71.6822429906542,
+      "grad_norm": 0.06025317683815956,
+      "learning_rate": 1.412246414949997e-05,
+      "loss": 0.004,
+      "step": 7670
+    },
+    {
+      "epoch": 71.77570093457943,
+      "grad_norm": 0.05556010454893112,
+      "learning_rate": 1.4007495327225162e-05,
+      "loss": 0.0048,
+      "step": 7680
+    },
+    {
+      "epoch": 71.86915887850468,
+      "grad_norm": 0.06115364655852318,
+      "learning_rate": 1.389292011321498e-05,
+      "loss": 0.0051,
+      "step": 7690
+    },
+    {
+      "epoch": 71.96261682242991,
+      "grad_norm": 0.12382921576499939,
+      "learning_rate": 1.3778739760445552e-05,
+      "loss": 0.0046,
+      "step": 7700
+    },
+    {
+      "epoch": 72.05607476635514,
+      "grad_norm": 0.08170247077941895,
+      "learning_rate": 1.3664955517574968e-05,
+      "loss": 0.0039,
+      "step": 7710
+    },
+    {
+      "epoch": 72.14953271028037,
+      "grad_norm": 0.08965415507555008,
+      "learning_rate": 1.3551568628929434e-05,
+      "loss": 0.0074,
+      "step": 7720
+    },
+    {
+      "epoch": 72.24299065420561,
+      "grad_norm": 0.18327778577804565,
+      "learning_rate": 1.343858033448982e-05,
+      "loss": 0.0058,
+      "step": 7730
+    },
+    {
+      "epoch": 72.33644859813084,
+      "grad_norm": 0.2206907570362091,
+      "learning_rate": 1.3325991869878013e-05,
+      "loss": 0.0046,
+      "step": 7740
+    },
+    {
+      "epoch": 72.42990654205607,
+      "grad_norm": 0.14453735947608948,
+      "learning_rate": 1.3213804466343421e-05,
+      "loss": 0.0053,
+      "step": 7750
+    },
+    {
+      "epoch": 72.5233644859813,
+      "grad_norm": 0.0630757138133049,
+      "learning_rate": 1.3102019350749528e-05,
+      "loss": 0.004,
+      "step": 7760
+    },
+    {
+      "epoch": 72.61682242990655,
+      "grad_norm": 0.07396890968084335,
+      "learning_rate": 1.299063774556042e-05,
+      "loss": 0.0036,
+      "step": 7770
+    },
+    {
+      "epoch": 72.71028037383178,
+      "grad_norm": 0.07948428392410278,
+      "learning_rate": 1.2879660868827508e-05,
+      "loss": 0.0045,
+      "step": 7780
+    },
+    {
+      "epoch": 72.80373831775701,
+      "grad_norm": 0.13974665105342865,
+      "learning_rate": 1.2769089934176126e-05,
+      "loss": 0.0066,
+      "step": 7790
+    },
+    {
+      "epoch": 72.89719626168224,
+      "grad_norm": 0.0720183327794075,
+      "learning_rate": 1.2658926150792322e-05,
+      "loss": 0.0035,
+      "step": 7800
+    },
+    {
+      "epoch": 72.99065420560747,
+      "grad_norm": 0.0702391117811203,
+      "learning_rate": 1.2549170723409549e-05,
+      "loss": 0.0041,
+      "step": 7810
+    },
+    {
+      "epoch": 73.08411214953271,
+      "grad_norm": 0.10197412967681885,
+      "learning_rate": 1.243982485229559e-05,
+      "loss": 0.0045,
+      "step": 7820
+    },
+    {
+      "epoch": 73.17757009345794,
+      "grad_norm": 0.20599843561649323,
+      "learning_rate": 1.233088973323937e-05,
+      "loss": 0.0047,
+      "step": 7830
+    },
+    {
+      "epoch": 73.27102803738318,
+      "grad_norm": 0.0804935023188591,
+      "learning_rate": 1.2222366557537911e-05,
+      "loss": 0.0035,
+      "step": 7840
+    },
+    {
+      "epoch": 73.3644859813084,
+      "grad_norm": 0.12421528249979019,
+      "learning_rate": 1.2114256511983274e-05,
+      "loss": 0.004,
+      "step": 7850
+    },
+    {
+      "epoch": 73.45794392523365,
+      "grad_norm": 0.09544029086828232,
+      "learning_rate": 1.2006560778849578e-05,
+      "loss": 0.0041,
+      "step": 7860
+    },
+    {
+      "epoch": 73.55140186915888,
+      "grad_norm": 0.10466738790273666,
+      "learning_rate": 1.1899280535880119e-05,
+      "loss": 0.0064,
+      "step": 7870
+    },
+    {
+      "epoch": 73.64485981308411,
+      "grad_norm": 0.14154809713363647,
+      "learning_rate": 1.1792416956274444e-05,
+      "loss": 0.0062,
+      "step": 7880
+    },
+    {
+      "epoch": 73.73831775700934,
+      "grad_norm": 0.058023110032081604,
+      "learning_rate": 1.1685971208675539e-05,
+      "loss": 0.0051,
+      "step": 7890
+    },
+    {
+      "epoch": 73.83177570093459,
+      "grad_norm": 0.09609735012054443,
+      "learning_rate": 1.157994445715706e-05,
+      "loss": 0.0091,
+      "step": 7900
+    },
+    {
+      "epoch": 73.92523364485982,
+      "grad_norm": 0.09363383799791336,
+      "learning_rate": 1.1474337861210543e-05,
+      "loss": 0.0041,
+      "step": 7910
+    },
+    {
+      "epoch": 74.01869158878505,
+      "grad_norm": 0.14261376857757568,
+      "learning_rate": 1.1369152575732822e-05,
+      "loss": 0.0034,
+      "step": 7920
+    },
+    {
+      "epoch": 74.11214953271028,
+      "grad_norm": 0.06502056121826172,
+      "learning_rate": 1.1264389751013326e-05,
+      "loss": 0.0042,
+      "step": 7930
+    },
+    {
+      "epoch": 74.20560747663552,
+      "grad_norm": 0.08161542564630508,
+      "learning_rate": 1.1160050532721528e-05,
+      "loss": 0.0063,
+      "step": 7940
+    },
+    {
+      "epoch": 74.29906542056075,
+      "grad_norm": 0.05485191568732262,
+      "learning_rate": 1.1056136061894384e-05,
+      "loss": 0.0036,
+      "step": 7950
+    },
+    {
+      "epoch": 74.39252336448598,
+      "grad_norm": 0.14355270564556122,
+      "learning_rate": 1.095264747492391e-05,
+      "loss": 0.0047,
+      "step": 7960
+    },
+    {
+      "epoch": 74.48598130841121,
+      "grad_norm": 0.08420294523239136,
+      "learning_rate": 1.0849585903544706e-05,
+      "loss": 0.0053,
+      "step": 7970
+    },
+    {
+      "epoch": 74.57943925233644,
+      "grad_norm": 0.08172670006752014,
+      "learning_rate": 1.0746952474821614e-05,
+      "loss": 0.003,
+      "step": 7980
+    },
+    {
+      "epoch": 74.67289719626169,
+      "grad_norm": 0.07927118986845016,
+      "learning_rate": 1.0644748311137376e-05,
+      "loss": 0.0052,
+      "step": 7990
+    },
+    {
+      "epoch": 74.76635514018692,
+      "grad_norm": 0.10906045883893967,
+      "learning_rate": 1.0542974530180327e-05,
+      "loss": 0.0053,
+      "step": 8000
+    },
+    {
+      "epoch": 74.85981308411215,
+      "grad_norm": 0.10352658480405807,
+      "learning_rate": 1.0441632244932237e-05,
+      "loss": 0.0043,
+      "step": 8010
+    },
+    {
+      "epoch": 74.95327102803738,
+      "grad_norm": 0.13669471442699432,
+      "learning_rate": 1.0340722563656107e-05,
+      "loss": 0.0048,
+      "step": 8020
+    },
+    {
+      "epoch": 75.04672897196262,
+      "grad_norm": 0.12065304070711136,
+      "learning_rate": 1.0240246589884044e-05,
+      "loss": 0.0037,
+      "step": 8030
+    },
+    {
+      "epoch": 75.14018691588785,
+      "grad_norm": 0.11262913048267365,
+      "learning_rate": 1.0140205422405214e-05,
+      "loss": 0.0042,
+      "step": 8040
+    },
+    {
+      "epoch": 75.23364485981308,
+      "grad_norm": 0.06449510157108307,
+      "learning_rate": 1.0040600155253765e-05,
+      "loss": 0.0045,
+      "step": 8050
+    },
+    {
+      "epoch": 75.32710280373831,
+      "grad_norm": 0.15835675597190857,
+      "learning_rate": 9.941431877696955e-06,
+      "loss": 0.0061,
+      "step": 8060
+    },
+    {
+      "epoch": 75.42056074766356,
+      "grad_norm": 0.10386522859334946,
+      "learning_rate": 9.842701674223187e-06,
+      "loss": 0.004,
+      "step": 8070
+    },
+    {
+      "epoch": 75.51401869158879,
+      "grad_norm": 0.07288730889558792,
+      "learning_rate": 9.744410624530148e-06,
+      "loss": 0.0065,
+      "step": 8080
+    },
+    {
+      "epoch": 75.60747663551402,
+      "grad_norm": 0.05738939344882965,
+      "learning_rate": 9.646559803512994e-06,
+      "loss": 0.0034,
+      "step": 8090
+    },
+    {
+      "epoch": 75.70093457943925,
+      "grad_norm": 0.0958879292011261,
+      "learning_rate": 9.549150281252633e-06,
+      "loss": 0.0045,
+      "step": 8100
+    },
+    {
+      "epoch": 75.79439252336448,
+      "grad_norm": 0.0466642864048481,
+      "learning_rate": 9.452183123004e-06,
+      "loss": 0.0055,
+      "step": 8110
+    },
+    {
+      "epoch": 75.88785046728972,
+      "grad_norm": 0.1655125766992569,
+      "learning_rate": 9.355659389184396e-06,
+      "loss": 0.0076,
+      "step": 8120
+    },
+    {
+      "epoch": 75.98130841121495,
+      "grad_norm": 0.12360979616641998,
+      "learning_rate": 9.259580135361929e-06,
+      "loss": 0.0047,
+      "step": 8130
+    },
+    {
+      "epoch": 76.07476635514018,
+      "grad_norm": 0.06821216642856598,
+      "learning_rate": 9.163946412243896e-06,
+      "loss": 0.005,
+      "step": 8140
+    },
+    {
+      "epoch": 76.16822429906541,
+      "grad_norm": 0.04745886102318764,
+      "learning_rate": 9.068759265665384e-06,
+      "loss": 0.0034,
+      "step": 8150
+    },
+    {
+      "epoch": 76.26168224299066,
+      "grad_norm": 0.1106477677822113,
+      "learning_rate": 8.974019736577777e-06,
+      "loss": 0.0051,
+      "step": 8160
+    },
+    {
+      "epoch": 76.35514018691589,
+      "grad_norm": 0.06387385725975037,
+      "learning_rate": 8.879728861037384e-06,
+      "loss": 0.006,
+      "step": 8170
+    },
+    {
+      "epoch": 76.44859813084112,
+      "grad_norm": 0.08066016435623169,
+      "learning_rate": 8.785887670194138e-06,
+      "loss": 0.0037,
+      "step": 8180
+    },
+    {
+      "epoch": 76.54205607476635,
+      "grad_norm": 0.06591822952032089,
+      "learning_rate": 8.692497190280224e-06,
+      "loss": 0.0053,
+      "step": 8190
+    },
+    {
+      "epoch": 76.6355140186916,
+      "grad_norm": 0.1523038148880005,
+      "learning_rate": 8.599558442598998e-06,
+      "loss": 0.006,
+      "step": 8200
+    },
+    {
+      "epoch": 76.72897196261682,
+      "grad_norm": 0.08181269466876984,
+      "learning_rate": 8.507072443513702e-06,
+      "loss": 0.0046,
+      "step": 8210
+    },
+    {
+      "epoch": 76.82242990654206,
+      "grad_norm": 0.12298338115215302,
+      "learning_rate": 8.415040204436426e-06,
+      "loss": 0.0045,
+      "step": 8220
+    },
+    {
+      "epoch": 76.91588785046729,
+      "grad_norm": 0.28609699010849,
+      "learning_rate": 8.323462731816961e-06,
+      "loss": 0.0046,
+      "step": 8230
+    },
+    {
+      "epoch": 77.00934579439253,
+      "grad_norm": 0.11661451309919357,
+      "learning_rate": 8.232341027131885e-06,
+      "loss": 0.005,
+      "step": 8240
+    },
+    {
+      "epoch": 77.10280373831776,
+      "grad_norm": 0.13034674525260925,
+      "learning_rate": 8.141676086873572e-06,
+      "loss": 0.0043,
+      "step": 8250
+    },
+    {
+      "epoch": 77.19626168224299,
+      "grad_norm": 0.09487629681825638,
+      "learning_rate": 8.051468902539272e-06,
+      "loss": 0.0042,
+      "step": 8260
+    },
+    {
+      "epoch": 77.28971962616822,
+      "grad_norm": 0.12193930894136429,
+      "learning_rate": 7.96172046062032e-06,
+      "loss": 0.0047,
+      "step": 8270
+    },
+    {
+      "epoch": 77.38317757009345,
+      "grad_norm": 0.14566880464553833,
+      "learning_rate": 7.872431742591268e-06,
+      "loss": 0.0056,
+      "step": 8280
+    },
+    {
+      "epoch": 77.4766355140187,
+      "grad_norm": 0.09443622827529907,
+      "learning_rate": 7.783603724899257e-06,
+      "loss": 0.0066,
+      "step": 8290
+    },
+    {
+      "epoch": 77.57009345794393,
+      "grad_norm": 0.03891506791114807,
+      "learning_rate": 7.695237378953223e-06,
+      "loss": 0.0022,
+      "step": 8300
+    },
+    {
+      "epoch": 77.66355140186916,
+      "grad_norm": 0.05303707346320152,
+      "learning_rate": 7.607333671113409e-06,
+      "loss": 0.002,
+      "step": 8310
+    },
+    {
+      "epoch": 77.75700934579439,
+      "grad_norm": 0.1763453185558319,
+      "learning_rate": 7.519893562680663e-06,
+      "loss": 0.0049,
+      "step": 8320
+    },
+    {
+      "epoch": 77.85046728971963,
+      "grad_norm": 0.045349784195423126,
+      "learning_rate": 7.432918009885997e-06,
+      "loss": 0.0044,
+      "step": 8330
+    },
+    {
+      "epoch": 77.94392523364486,
+      "grad_norm": 0.12692664563655853,
+      "learning_rate": 7.3464079638801365e-06,
+      "loss": 0.0095,
+      "step": 8340
+    },
+    {
+      "epoch": 78.03738317757009,
+      "grad_norm": 0.07811149954795837,
+      "learning_rate": 7.260364370723044e-06,
+      "loss": 0.003,
+      "step": 8350
+    },
+    {
+      "epoch": 78.13084112149532,
+      "grad_norm": 0.15021850168704987,
+      "learning_rate": 7.174788171373731e-06,
+      "loss": 0.0069,
+      "step": 8360
+    },
+    {
+      "epoch": 78.22429906542057,
+      "grad_norm": 0.0827387198805809,
+      "learning_rate": 7.089680301679752e-06,
+      "loss": 0.0081,
+      "step": 8370
+    },
+    {
+      "epoch": 78.3177570093458,
+      "grad_norm": 0.08250360190868378,
+      "learning_rate": 7.005041692367154e-06,
+      "loss": 0.0054,
+      "step": 8380
+    },
+    {
+      "epoch": 78.41121495327103,
+      "grad_norm": 0.1127791777253151,
+      "learning_rate": 6.92087326903022e-06,
+      "loss": 0.0035,
+      "step": 8390
+    },
+    {
+      "epoch": 78.50467289719626,
+      "grad_norm": 0.05626427009701729,
+      "learning_rate": 6.837175952121306e-06,
+      "loss": 0.0032,
+      "step": 8400
+    },
+    {
+      "epoch": 78.59813084112149,
+      "grad_norm": 0.11432872712612152,
+      "learning_rate": 6.753950656940905e-06,
+      "loss": 0.003,
+      "step": 8410
+    },
+    {
+      "epoch": 78.69158878504673,
+      "grad_norm": 0.0607433021068573,
+      "learning_rate": 6.671198293627479e-06,
+      "loss": 0.0028,
+      "step": 8420
+    },
+    {
+      "epoch": 78.78504672897196,
+      "grad_norm": 0.05074009671807289,
+      "learning_rate": 6.588919767147639e-06,
+      "loss": 0.0047,
+      "step": 8430
+    },
+    {
+      "epoch": 78.8785046728972,
+      "grad_norm": 0.07059110701084137,
+      "learning_rate": 6.5071159772861436e-06,
+      "loss": 0.0024,
+      "step": 8440
+    },
+    {
+      "epoch": 78.97196261682242,
+      "grad_norm": 0.07970912009477615,
+      "learning_rate": 6.425787818636131e-06,
+      "loss": 0.003,
+      "step": 8450
+    },
+    {
+      "epoch": 79.06542056074767,
+      "grad_norm": 0.04442933201789856,
+      "learning_rate": 6.344936180589351e-06,
+      "loss": 0.0031,
+      "step": 8460
+    },
+    {
+      "epoch": 79.1588785046729,
+      "grad_norm": 0.0632733404636383,
+      "learning_rate": 6.264561947326331e-06,
+      "loss": 0.004,
+      "step": 8470
+    },
+    {
+      "epoch": 79.25233644859813,
+      "grad_norm": 0.2247694879770279,
+      "learning_rate": 6.184665997806832e-06,
+      "loss": 0.0062,
+      "step": 8480
+    },
+    {
+      "epoch": 79.34579439252336,
+      "grad_norm": 0.08065646141767502,
+      "learning_rate": 6.1052492057601275e-06,
+      "loss": 0.0039,
+      "step": 8490
+    },
+    {
+      "epoch": 79.4392523364486,
+      "grad_norm": 0.1812344342470169,
+      "learning_rate": 6.026312439675552e-06,
+      "loss": 0.0057,
+      "step": 8500
+    },
+    {
+      "epoch": 79.53271028037383,
+      "grad_norm": 0.1381814181804657,
+      "learning_rate": 5.947856562792925e-06,
+      "loss": 0.0101,
+      "step": 8510
+    },
+    {
+      "epoch": 79.62616822429906,
+      "grad_norm": 0.04312485083937645,
+      "learning_rate": 5.869882433093155e-06,
+      "loss": 0.0028,
+      "step": 8520
+    },
+    {
+      "epoch": 79.7196261682243,
+      "grad_norm": 0.24571475386619568,
+      "learning_rate": 5.79239090328883e-06,
+      "loss": 0.0066,
+      "step": 8530
+    },
+    {
+      "epoch": 79.81308411214954,
+      "grad_norm": 0.06329802423715591,
+      "learning_rate": 5.715382820814885e-06,
+      "loss": 0.003,
+      "step": 8540
+    },
+    {
+      "epoch": 79.90654205607477,
+      "grad_norm": 0.05647183954715729,
+      "learning_rate": 5.6388590278194096e-06,
+      "loss": 0.0031,
+      "step": 8550
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 0.0694732517004013,
+      "learning_rate": 5.562820361154314e-06,
+      "loss": 0.0038,
+      "step": 8560
+    },
+    {
+      "epoch": 80.09345794392523,
+      "grad_norm": 0.09058716893196106,
+      "learning_rate": 5.48726765236629e-06,
+      "loss": 0.0026,
+      "step": 8570
+    },
+    {
+      "epoch": 80.18691588785046,
+      "grad_norm": 0.12245356291532516,
+      "learning_rate": 5.412201727687644e-06,
+      "loss": 0.0044,
+      "step": 8580
+    },
+    {
+      "epoch": 80.2803738317757,
+      "grad_norm": 0.11849050968885422,
+      "learning_rate": 5.337623408027293e-06,
+      "loss": 0.0045,
+      "step": 8590
+    },
+    {
+      "epoch": 80.37383177570094,
+      "grad_norm": 0.13533665239810944,
+      "learning_rate": 5.263533508961827e-06,
+      "loss": 0.0037,
+      "step": 8600
+    },
+    {
+      "epoch": 80.46728971962617,
+      "grad_norm": 0.05687372013926506,
+      "learning_rate": 5.1899328407264855e-06,
+      "loss": 0.0029,
+      "step": 8610
+    },
+    {
+      "epoch": 80.5607476635514,
+      "grad_norm": 0.19712159037590027,
+      "learning_rate": 5.116822208206396e-06,
+      "loss": 0.003,
+      "step": 8620
+    },
+    {
+      "epoch": 80.65420560747664,
+      "grad_norm": 0.058728091418743134,
+      "learning_rate": 5.044202410927706e-06,
+      "loss": 0.0073,
+      "step": 8630
+    },
+    {
+      "epoch": 80.74766355140187,
+      "grad_norm": 0.05174853280186653,
+      "learning_rate": 4.972074243048897e-06,
+      "loss": 0.0025,
+      "step": 8640
+    },
+    {
+      "epoch": 80.8411214953271,
+      "grad_norm": 0.046942222863435745,
+      "learning_rate": 4.900438493352055e-06,
+      "loss": 0.0044,
+      "step": 8650
+    },
+    {
+      "epoch": 80.93457943925233,
+      "grad_norm": 0.06713097542524338,
+      "learning_rate": 4.829295945234258e-06,
+      "loss": 0.0025,
+      "step": 8660
+    },
+    {
+      "epoch": 81.02803738317758,
+      "grad_norm": 0.058530379086732864,
+      "learning_rate": 4.758647376699032e-06,
+      "loss": 0.0029,
+      "step": 8670
+    },
+    {
+      "epoch": 81.1214953271028,
+      "grad_norm": 0.0929771289229393,
+      "learning_rate": 4.688493560347773e-06,
+      "loss": 0.0034,
+      "step": 8680
+    },
+    {
+      "epoch": 81.21495327102804,
+      "grad_norm": 0.10967829823493958,
+      "learning_rate": 4.618835263371396e-06,
+      "loss": 0.0032,
+      "step": 8690
+    },
+    {
+      "epoch": 81.30841121495327,
+      "grad_norm": 0.05041574314236641,
+      "learning_rate": 4.549673247541875e-06,
+      "loss": 0.0039,
+      "step": 8700
+    },
+    {
+      "epoch": 81.40186915887851,
+      "grad_norm": 0.1108449250459671,
+      "learning_rate": 4.48100826920394e-06,
+      "loss": 0.0029,
+      "step": 8710
+    },
+    {
+      "epoch": 81.49532710280374,
+      "grad_norm": 0.09053438901901245,
+      "learning_rate": 4.412841079266777e-06,
+      "loss": 0.0044,
+      "step": 8720
+    },
+    {
+      "epoch": 81.58878504672897,
+      "grad_norm": 0.1006326973438263,
+      "learning_rate": 4.3451724231958644e-06,
+      "loss": 0.0033,
+      "step": 8730
+    },
+    {
+      "epoch": 81.6822429906542,
+      "grad_norm": 0.11440656334161758,
+      "learning_rate": 4.27800304100478e-06,
+      "loss": 0.0031,
+      "step": 8740
+    },
+    {
+      "epoch": 81.77570093457943,
+      "grad_norm": 0.048498962074518204,
+      "learning_rate": 4.2113336672471245e-06,
+      "loss": 0.0025,
+      "step": 8750
+    },
+    {
+      "epoch": 81.86915887850468,
+      "grad_norm": 0.08707248419523239,
+      "learning_rate": 4.145165031008508e-06,
+      "loss": 0.0041,
+      "step": 8760
+    },
+    {
+      "epoch": 81.96261682242991,
+      "grad_norm": 0.0621667355298996,
+      "learning_rate": 4.079497855898501e-06,
+      "loss": 0.0041,
+      "step": 8770
+    },
+    {
+      "epoch": 82.05607476635514,
+      "grad_norm": 0.12233584374189377,
+      "learning_rate": 4.01433286004283e-06,
+      "loss": 0.007,
+      "step": 8780
+    },
+    {
+      "epoch": 82.14953271028037,
+      "grad_norm": 0.14951051771640778,
+      "learning_rate": 3.949670756075447e-06,
+      "loss": 0.0045,
+      "step": 8790
+    },
+    {
+      "epoch": 82.24299065420561,
+      "grad_norm": 0.3771739900112152,
+      "learning_rate": 3.885512251130763e-06,
+      "loss": 0.0073,
+      "step": 8800
+    },
+    {
+      "epoch": 82.33644859813084,
+      "grad_norm": 0.04400696977972984,
+      "learning_rate": 3.821858046835913e-06,
+      "loss": 0.0045,
+      "step": 8810
+    },
+    {
+      "epoch": 82.42990654205607,
+      "grad_norm": 0.12610068917274475,
+      "learning_rate": 3.75870883930306e-06,
+      "loss": 0.0029,
+      "step": 8820
+    },
+    {
+      "epoch": 82.5233644859813,
+      "grad_norm": 0.26768192648887634,
+      "learning_rate": 3.696065319121833e-06,
+      "loss": 0.0068,
+      "step": 8830
+    },
+    {
+      "epoch": 82.61682242990655,
+      "grad_norm": 0.13012675940990448,
+      "learning_rate": 3.6339281713517303e-06,
+      "loss": 0.0049,
+      "step": 8840
+    },
+    {
+      "epoch": 82.71028037383178,
+      "grad_norm": 0.08191750943660736,
+      "learning_rate": 3.5722980755146517e-06,
+      "loss": 0.0068,
+      "step": 8850
+    },
+    {
+      "epoch": 82.80373831775701,
+      "grad_norm": 0.11502333730459213,
+      "learning_rate": 3.511175705587433e-06,
+      "loss": 0.0043,
+      "step": 8860
+    },
+    {
+      "epoch": 82.89719626168224,
+      "grad_norm": 0.14655309915542603,
+      "learning_rate": 3.4505617299945336e-06,
+      "loss": 0.0062,
+      "step": 8870
+    },
+    {
+      "epoch": 82.99065420560747,
+      "grad_norm": 0.1569114327430725,
+      "learning_rate": 3.390456811600673e-06,
+      "loss": 0.008,
+      "step": 8880
+    },
+    {
+      "epoch": 83.08411214953271,
+      "grad_norm": 0.08708454668521881,
+      "learning_rate": 3.3308616077036115e-06,
+      "loss": 0.0046,
+      "step": 8890
+    },
+    {
+      "epoch": 83.17757009345794,
+      "grad_norm": 0.05366307497024536,
+      "learning_rate": 3.271776770026963e-06,
+      "loss": 0.0022,
+      "step": 8900
+    },
+    {
+      "epoch": 83.27102803738318,
+      "grad_norm": 0.15384458005428314,
+      "learning_rate": 3.213202944713023e-06,
+      "loss": 0.0049,
+      "step": 8910
+    },
+    {
+      "epoch": 83.3644859813084,
+      "grad_norm": 0.06727658212184906,
+      "learning_rate": 3.155140772315773e-06,
+      "loss": 0.0031,
+      "step": 8920
+    },
+    {
+      "epoch": 83.45794392523365,
+      "grad_norm": 0.08927454799413681,
+      "learning_rate": 3.0975908877938277e-06,
+      "loss": 0.0033,
+      "step": 8930
+    },
+    {
+      "epoch": 83.55140186915888,
+      "grad_norm": 0.12398143857717514,
+      "learning_rate": 3.040553920503503e-06,
+      "loss": 0.0033,
+      "step": 8940
+    },
+    {
+      "epoch": 83.64485981308411,
+      "grad_norm": 0.16793964803218842,
+      "learning_rate": 2.9840304941919415e-06,
+      "loss": 0.0032,
+      "step": 8950
+    },
+    {
+      "epoch": 83.73831775700934,
+      "grad_norm": 0.1104469820857048,
+      "learning_rate": 2.928021226990263e-06,
+      "loss": 0.0032,
+      "step": 8960
+    },
+    {
+      "epoch": 83.83177570093459,
+      "grad_norm": 0.07358328253030777,
+      "learning_rate": 2.8725267314068495e-06,
+      "loss": 0.0028,
+      "step": 8970
+    },
+    {
+      "epoch": 83.92523364485982,
+      "grad_norm": 0.13550950586795807,
+      "learning_rate": 2.817547614320615e-06,
+      "loss": 0.0072,
+      "step": 8980
+    },
+    {
+      "epoch": 84.01869158878505,
+      "grad_norm": 0.09289432317018509,
+      "learning_rate": 2.7630844769743757e-06,
+      "loss": 0.0053,
+      "step": 8990
+    },
+    {
+      "epoch": 84.11214953271028,
+      "grad_norm": 0.0311606265604496,
+      "learning_rate": 2.7091379149682685e-06,
+      "loss": 0.0035,
+      "step": 9000
+    },
+    {
+      "epoch": 84.20560747663552,
+      "grad_norm": 0.07517659664154053,
+      "learning_rate": 2.6557085182532582e-06,
+      "loss": 0.0061,
+      "step": 9010
+    },
+    {
+      "epoch": 84.29906542056075,
+      "grad_norm": 0.07934467494487762,
+      "learning_rate": 2.602796871124663e-06,
+      "loss": 0.0055,
+      "step": 9020
+    },
+    {
+      "epoch": 84.39252336448598,
+      "grad_norm": 0.0909305214881897,
+      "learning_rate": 2.5504035522157854e-06,
+      "loss": 0.0034,
+      "step": 9030
+    },
+    {
+      "epoch": 84.48598130841121,
+      "grad_norm": 0.05545186623930931,
+      "learning_rate": 2.4985291344915674e-06,
+      "loss": 0.0032,
+      "step": 9040
+    },
+    {
+      "epoch": 84.57943925233644,
+      "grad_norm": 0.14734113216400146,
+      "learning_rate": 2.4471741852423237e-06,
+      "loss": 0.0045,
+      "step": 9050
+    },
+    {
+      "epoch": 84.67289719626169,
+      "grad_norm": 0.10017460584640503,
+      "learning_rate": 2.3963392660775575e-06,
+      "loss": 0.0028,
+      "step": 9060
+    },
+    {
+      "epoch": 84.76635514018692,
+      "grad_norm": 0.04614974930882454,
+      "learning_rate": 2.3460249329197824e-06,
+      "loss": 0.0031,
+      "step": 9070
+    },
+    {
+      "epoch": 84.85981308411215,
+      "grad_norm": 0.1374596208333969,
+      "learning_rate": 2.296231735998511e-06,
+      "loss": 0.0031,
+      "step": 9080
+    },
+    {
+      "epoch": 84.95327102803738,
+      "grad_norm": 0.07840343564748764,
+      "learning_rate": 2.2469602198441573e-06,
+      "loss": 0.0031,
+      "step": 9090
+    },
+    {
+      "epoch": 85.04672897196262,
+      "grad_norm": 0.04728247597813606,
+      "learning_rate": 2.1982109232821178e-06,
+      "loss": 0.0035,
+      "step": 9100
+    },
+    {
+      "epoch": 85.14018691588785,
+      "grad_norm": 0.05790605768561363,
+      "learning_rate": 2.149984379426906e-06,
+      "loss": 0.0028,
+      "step": 9110
+    },
+    {
+      "epoch": 85.23364485981308,
+      "grad_norm": 0.09767275303602219,
+      "learning_rate": 2.102281115676258e-06,
+      "loss": 0.0027,
+      "step": 9120
+    },
+    {
+      "epoch": 85.32710280373831,
+      "grad_norm": 0.05615156143903732,
+      "learning_rate": 2.0551016537054493e-06,
+      "loss": 0.0049,
+      "step": 9130
+    },
+    {
+      "epoch": 85.42056074766356,
+      "grad_norm": 0.0627375990152359,
+      "learning_rate": 2.008446509461498e-06,
+      "loss": 0.004,
+      "step": 9140
+    },
+    {
+      "epoch": 85.51401869158879,
+      "grad_norm": 0.06630418449640274,
+      "learning_rate": 1.962316193157593e-06,
+      "loss": 0.0032,
+      "step": 9150
+    },
+    {
+      "epoch": 85.60747663551402,
+      "grad_norm": 0.08775442093610764,
+      "learning_rate": 1.91671120926748e-06,
+      "loss": 0.0051,
+      "step": 9160
+    },
+    {
+      "epoch": 85.70093457943925,
+      "grad_norm": 0.061658404767513275,
+      "learning_rate": 1.8716320565199618e-06,
+      "loss": 0.0033,
+      "step": 9170
+    },
+    {
+      "epoch": 85.79439252336448,
+      "grad_norm": 0.054096709936857224,
+      "learning_rate": 1.8270792278934302e-06,
+      "loss": 0.0032,
+      "step": 9180
+    },
+    {
+      "epoch": 85.88785046728972,
+      "grad_norm": 0.06938756257295609,
+      "learning_rate": 1.7830532106104747e-06,
+      "loss": 0.0029,
+      "step": 9190
+    },
+    {
+      "epoch": 85.98130841121495,
+      "grad_norm": 0.13099592924118042,
+      "learning_rate": 1.7395544861325718e-06,
+      "loss": 0.0044,
+      "step": 9200
+    },
+    {
+      "epoch": 86.07476635514018,
+      "grad_norm": 0.08681978285312653,
+      "learning_rate": 1.696583530154794e-06,
+      "loss": 0.0051,
+      "step": 9210
+    },
+    {
+      "epoch": 86.16822429906541,
+      "grad_norm": 0.06871863454580307,
+      "learning_rate": 1.6541408126006463e-06,
+      "loss": 0.0023,
+      "step": 9220
+    },
+    {
+      "epoch": 86.26168224299066,
+      "grad_norm": 0.11858540028333664,
+      "learning_rate": 1.6122267976168781e-06,
+      "loss": 0.0054,
+      "step": 9230
+    },
+    {
+      "epoch": 86.35514018691589,
+      "grad_norm": 0.0753912627696991,
+      "learning_rate": 1.5708419435684462e-06,
+      "loss": 0.0045,
+      "step": 9240
+    },
+    {
+      "epoch": 86.44859813084112,
+      "grad_norm": 0.04186966270208359,
+      "learning_rate": 1.5299867030334814e-06,
+      "loss": 0.0058,
+      "step": 9250
+    },
+    {
+      "epoch": 86.54205607476635,
+      "grad_norm": 0.1056654155254364,
+      "learning_rate": 1.4896615227983468e-06,
+      "loss": 0.0027,
+      "step": 9260
+    },
+    {
+      "epoch": 86.6355140186916,
+      "grad_norm": 0.04532716050744057,
+      "learning_rate": 1.4498668438527597e-06,
+      "loss": 0.0031,
+      "step": 9270
+    },
+    {
+      "epoch": 86.72897196261682,
+      "grad_norm": 0.036859266459941864,
+      "learning_rate": 1.4106031013849496e-06,
+      "loss": 0.0063,
+      "step": 9280
+    },
+    {
+      "epoch": 86.82242990654206,
+      "grad_norm": 0.11480679363012314,
+      "learning_rate": 1.3718707247769135e-06,
+      "loss": 0.0029,
+      "step": 9290
+    },
+    {
+      "epoch": 86.91588785046729,
+      "grad_norm": 0.2239740788936615,
+      "learning_rate": 1.333670137599713e-06,
+      "loss": 0.0075,
+      "step": 9300
+    },
+    {
+      "epoch": 87.00934579439253,
+      "grad_norm": 0.14973869919776917,
+      "learning_rate": 1.2960017576088446e-06,
+      "loss": 0.0045,
+      "step": 9310
+    },
+    {
+      "epoch": 87.10280373831776,
+      "grad_norm": 0.10877890884876251,
+      "learning_rate": 1.2588659967397e-06,
+      "loss": 0.011,
+      "step": 9320
+    },
+    {
+      "epoch": 87.19626168224299,
+      "grad_norm": 0.08974131941795349,
+      "learning_rate": 1.222263261102985e-06,
+      "loss": 0.0037,
+      "step": 9330
+    },
+    {
+      "epoch": 87.28971962616822,
+      "grad_norm": 0.12366478145122528,
+      "learning_rate": 1.1861939509803687e-06,
+      "loss": 0.0044,
+      "step": 9340
+    },
+    {
+      "epoch": 87.38317757009345,
+      "grad_norm": 0.06801088899374008,
+      "learning_rate": 1.1506584608200367e-06,
+      "loss": 0.0023,
+      "step": 9350
+    },
+    {
+      "epoch": 87.4766355140187,
+      "grad_norm": 0.055119484663009644,
+      "learning_rate": 1.1156571792324211e-06,
+      "loss": 0.0024,
+      "step": 9360
+    },
+    {
+      "epoch": 87.57009345794393,
+      "grad_norm": 0.03374645486474037,
+      "learning_rate": 1.0811904889859336e-06,
+      "loss": 0.0082,
+      "step": 9370
+    },
+    {
+      "epoch": 87.66355140186916,
+      "grad_norm": 0.18118692934513092,
+      "learning_rate": 1.0472587670027678e-06,
+      "loss": 0.0026,
+      "step": 9380
+    },
+    {
+      "epoch": 87.75700934579439,
+      "grad_norm": 0.17593815922737122,
+      "learning_rate": 1.0138623843548078e-06,
+      "loss": 0.0115,
+      "step": 9390
+    },
+    {
+      "epoch": 87.85046728971963,
+      "grad_norm": 0.12268546968698502,
+      "learning_rate": 9.810017062595322e-07,
+      "loss": 0.003,
+      "step": 9400
+    },
+    {
+      "epoch": 87.94392523364486,
+      "grad_norm": 0.08846314996480942,
+      "learning_rate": 9.486770920760668e-07,
+      "loss": 0.0036,
+      "step": 9410
+    },
+    {
+      "epoch": 88.03738317757009,
+      "grad_norm": 0.09881670773029327,
+      "learning_rate": 9.168888953011989e-07,
+      "loss": 0.0026,
+      "step": 9420
+    },
+    {
+      "epoch": 88.13084112149532,
+      "grad_norm": 0.05135083198547363,
+      "learning_rate": 8.856374635655695e-07,
+      "loss": 0.0043,
+      "step": 9430
+    },
+    {
+      "epoch": 88.22429906542057,
+      "grad_norm": 0.04362175986170769,
+      "learning_rate": 8.549231386298151e-07,
+      "loss": 0.0037,
+      "step": 9440
+    },
+    {
+      "epoch": 88.3177570093458,
+      "grad_norm": 0.0345948152244091,
+      "learning_rate": 8.247462563808817e-07,
+      "loss": 0.0071,
+      "step": 9450
+    },
+    {
+      "epoch": 88.41121495327103,
+      "grad_norm": 0.09612035006284714,
+      "learning_rate": 7.951071468283167e-07,
+      "loss": 0.0021,
+      "step": 9460
+    },
+    {
+      "epoch": 88.50467289719626,
+      "grad_norm": 0.04552985727787018,
+      "learning_rate": 7.66006134100672e-07,
+      "loss": 0.003,
+      "step": 9470
+    },
+    {
+      "epoch": 88.59813084112149,
+      "grad_norm": 0.1042826697230339,
+      "learning_rate": 7.374435364419674e-07,
+      "loss": 0.0035,
+      "step": 9480
+    },
+    {
+      "epoch": 88.69158878504673,
+      "grad_norm": 0.15467211604118347,
+      "learning_rate": 7.094196662081831e-07,
+      "loss": 0.0034,
+      "step": 9490
+    },
+    {
+      "epoch": 88.78504672897196,
+      "grad_norm": 0.22603380680084229,
+      "learning_rate": 6.819348298638839e-07,
+      "loss": 0.0038,
+      "step": 9500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 94,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.69573051908096e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}