diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,3980 +1,628 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 4.712550607287449,
   "eval_steps": 100,
-  "global_step": 494,
+  "global_step": 75,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0020242914979757085,
-      "grad_norm": 0.19618171453475952,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 0.6744,
-      "mean_token_accuracy": 0.8156972527503967,
+      "epoch": 0.06477732793522267,
+      "grad_norm": 0.08484393358230591,
+      "learning_rate": 2.5e-08,
+      "loss": 0.693,
+      "mean_token_accuracy": 0.8232172057032585,
       "step": 1
     },
     {
-      "epoch": 0.004048582995951417,
-      "grad_norm": 0.20514528453350067,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 0.6586,
-      "mean_token_accuracy": 0.8400664329528809,
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.08104647696018219,
+      "learning_rate": 5e-08,
+      "loss": 0.6361,
+      "mean_token_accuracy": 0.8352038823068142,
       "step": 2
     },
     {
-      "epoch": 0.006072874493927126,
-      "grad_norm": 0.1573631763458252,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 0.576,
-      "mean_token_accuracy": 0.8486753702163696,
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.08257265388965607,
+      "learning_rate": 7.5e-08,
+      "loss": 0.669,
+      "mean_token_accuracy": 0.8248961418867111,
       "step": 3
     },
     {
-      "epoch": 0.008097165991902834,
-      "grad_norm": 0.1505558341741562,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 0.5729,
-      "mean_token_accuracy": 0.860769510269165,
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.08169891685247421,
+      "learning_rate": 1e-07,
+      "loss": 0.654,
+      "mean_token_accuracy": 0.8317753672599792,
       "step": 4
     },
     {
-      "epoch": 0.010121457489878543,
-      "grad_norm": 0.1700662225484848,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.6392,
-      "mean_token_accuracy": 0.8422141671180725,
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.07086937129497528,
+      "learning_rate": 1.25e-07,
+      "loss": 0.6029,
+      "mean_token_accuracy": 0.8417873457074165,
       "step": 5
     },
     {
-      "epoch": 0.012145748987854251,
-      "grad_norm": 0.18006764352321625,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 0.6843,
-      "mean_token_accuracy": 0.8051391839981079,
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.08497548848390579,
+      "learning_rate": 1.5e-07,
+      "loss": 0.6561,
+      "mean_token_accuracy": 0.8321865610778332,
       "step": 6
     },
     {
-      "epoch": 0.01417004048582996,
-      "grad_norm": 0.14205701649188995,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 0.608,
-      "mean_token_accuracy": 0.8361465334892273,
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.07375866174697876,
+      "learning_rate": 1.75e-07,
+      "loss": 0.6029,
+      "mean_token_accuracy": 0.8472057469189167,
       "step": 7
     },
     {
-      "epoch": 0.016194331983805668,
-      "grad_norm": 0.14348368346691132,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 0.6215,
-      "mean_token_accuracy": 0.8492392897605896,
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.08089514076709747,
+      "learning_rate": 2e-07,
+      "loss": 0.6429,
+      "mean_token_accuracy": 0.8303790092468262,
       "step": 8
     },
     {
-      "epoch": 0.018218623481781375,
-      "grad_norm": 0.18333880603313446,
-      "learning_rate": 3.6000000000000003e-06,
-      "loss": 0.6696,
-      "mean_token_accuracy": 0.8336831331253052,
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.07564779371023178,
+      "learning_rate": 1.9989008914857112e-07,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.8341928347945213,
       "step": 9
     },
     {
-      "epoch": 0.020242914979757085,
-      "grad_norm": 0.13913995027542114,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 0.6407,
-      "mean_token_accuracy": 0.8342278003692627,
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.08072903752326965,
+      "learning_rate": 1.995605982021898e-07,
+      "loss": 0.6476,
+      "mean_token_accuracy": 0.8313158191740513,
       "step": 10
     },
     {
-      "epoch": 0.022267206477732792,
-      "grad_norm": 0.14520516991615295,
-      "learning_rate": 4.4e-06,
-      "loss": 0.6821,
-      "mean_token_accuracy": 0.829963207244873,
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.08074145019054413,
+      "learning_rate": 1.9901225145346506e-07,
+      "loss": 0.6519,
+      "mean_token_accuracy": 0.8337250761687756,
       "step": 11
     },
     {
-      "epoch": 0.024291497975708502,
-      "grad_norm": 0.15494030714035034,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 0.6534,
-      "mean_token_accuracy": 0.8419155478477478,
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.08061974495649338,
+      "learning_rate": 1.9824625428755758e-07,
+      "loss": 0.651,
+      "mean_token_accuracy": 0.8378880433738232,
       "step": 12
     },
     {
-      "epoch": 0.02631578947368421,
-      "grad_norm": 0.1409451961517334,
-      "learning_rate": 5.2e-06,
-      "loss": 0.5572,
-      "mean_token_accuracy": 0.8476144075393677,
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.0789395272731781,
+      "learning_rate": 1.9726429053248129e-07,
+      "loss": 0.6443,
+      "mean_token_accuracy": 0.8334936276078224,
       "step": 13
     },
     {
-      "epoch": 0.02834008097165992,
-      "grad_norm": 0.11936034262180328,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 0.5364,
-      "mean_token_accuracy": 0.8614328503608704,
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.07814885675907135,
+      "learning_rate": 1.96068518757684e-07,
+      "loss": 0.6565,
+      "mean_token_accuracy": 0.8271612599492073,
       "step": 14
     },
     {
-      "epoch": 0.030364372469635626,
-      "grad_norm": 0.15133579075336456,
-      "learning_rate": 6e-06,
-      "loss": 0.6552,
-      "mean_token_accuracy": 0.8225256204605103,
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.07767179608345032,
+      "learning_rate": 1.946615675290434e-07,
+      "loss": 0.6348,
+      "mean_token_accuracy": 0.8349605351686478,
       "step": 15
     },
     {
-      "epoch": 0.032388663967611336,
-      "grad_norm": 0.18947570025920868,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 0.6948,
-      "mean_token_accuracy": 0.8155401349067688,
+      "epoch": 1.0,
+      "grad_norm": 0.07767179608345032,
+      "learning_rate": 1.9304652963070865e-07,
+      "loss": 0.6711,
+      "mean_token_accuracy": 0.8187758156231472,
       "step": 16
     },
     {
-      "epoch": 0.03441295546558704,
-      "grad_norm": 0.12663717567920685,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 0.6184,
-      "mean_token_accuracy": 0.8341503143310547,
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.1275881677865982,
+      "learning_rate": 1.9122695526648966e-07,
+      "loss": 0.6142,
+      "mean_token_accuracy": 0.838625643402338,
       "step": 17
     },
     {
-      "epoch": 0.03643724696356275,
-      "grad_norm": 0.1532672792673111,
-      "learning_rate": 7.2000000000000005e-06,
-      "loss": 0.6429,
-      "mean_token_accuracy": 0.8517258167266846,
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.07696238905191422,
+      "learning_rate": 1.8920684425573862e-07,
+      "loss": 0.6289,
+      "mean_token_accuracy": 0.835904911160469,
       "step": 18
     },
     {
-      "epoch": 0.038461538461538464,
-      "grad_norm": 0.17713797092437744,
-      "learning_rate": 7.600000000000001e-06,
-      "loss": 0.6789,
-      "mean_token_accuracy": 0.8132250308990479,
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.08134917914867401,
+      "learning_rate": 1.8699063724087903e-07,
+      "loss": 0.6511,
+      "mean_token_accuracy": 0.830647598952055,
       "step": 19
     },
     {
-      "epoch": 0.04048582995951417,
-      "grad_norm": 0.12441661208868027,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.6009,
-      "mean_token_accuracy": 0.8490859866142273,
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.08185581117868423,
+      "learning_rate": 1.8458320592590972e-07,
+      "loss": 0.6395,
+      "mean_token_accuracy": 0.8346012011170387,
       "step": 20
     },
     {
-      "epoch": 0.04251012145748988,
-      "grad_norm": 0.1268952190876007,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 0.6118,
-      "mean_token_accuracy": 0.8372530341148376,
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.07670310139656067,
+      "learning_rate": 1.8198984236734245e-07,
+      "loss": 0.5885,
+      "mean_token_accuracy": 0.8427848629653454,
       "step": 21
     },
     {
-      "epoch": 0.044534412955465584,
-      "grad_norm": 0.17228662967681885,
-      "learning_rate": 8.8e-06,
-      "loss": 0.6998,
-      "mean_token_accuracy": 0.8195767402648926,
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.0753348097205162,
+      "learning_rate": 1.792162473411129e-07,
+      "loss": 0.6299,
+      "mean_token_accuracy": 0.834791149944067,
       "step": 22
     },
     {
-      "epoch": 0.0465587044534413,
-      "grad_norm": 0.17604027688503265,
-      "learning_rate": 9.200000000000002e-06,
-      "loss": 0.6755,
-      "mean_token_accuracy": 0.8294089436531067,
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.07955452054738998,
+      "learning_rate": 1.7626851781103817e-07,
+      "loss": 0.6499,
+      "mean_token_accuracy": 0.8349028266966343,
       "step": 23
     },
     {
-      "epoch": 0.048582995951417005,
-      "grad_norm": 0.14454977214336395,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 0.7248,
-      "mean_token_accuracy": 0.8143091201782227,
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.07731788605451584,
+      "learning_rate": 1.731531335263669e-07,
+      "loss": 0.6252,
+      "mean_token_accuracy": 0.8352472670376301,
       "step": 24
     },
     {
-      "epoch": 0.05060728744939271,
-      "grad_norm": 0.17172342538833618,
-      "learning_rate": 1e-05,
-      "loss": 0.5982,
-      "mean_token_accuracy": 0.8516228795051575,
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.07537000626325607,
+      "learning_rate": 1.6987694277788415e-07,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.8319784663617611,
       "step": 25
     },
     {
-      "epoch": 0.05263157894736842,
-      "grad_norm": 0.19159506261348724,
-      "learning_rate": 1.04e-05,
-      "loss": 0.7188,
-      "mean_token_accuracy": 0.8197706341743469,
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.08128884434700012,
+      "learning_rate": 1.6644714734388215e-07,
+      "loss": 0.6501,
+      "mean_token_accuracy": 0.8315212316811085,
       "step": 26
     },
     {
-      "epoch": 0.05465587044534413,
-      "grad_norm": 0.12568704783916473,
-      "learning_rate": 1.0800000000000002e-05,
-      "loss": 0.4287,
-      "mean_token_accuracy": 0.8862559199333191,
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.08110049366950989,
+      "learning_rate": 1.628712866590885e-07,
+      "loss": 0.6727,
+      "mean_token_accuracy": 0.8280021287500858,
       "step": 27
     },
     {
-      "epoch": 0.05668016194331984,
-      "grad_norm": 0.13636384904384613,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 0.5182,
-      "mean_token_accuracy": 0.8564677834510803,
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.08107449114322662,
+      "learning_rate": 1.5915722124135225e-07,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.8243995904922485,
       "step": 28
     },
     {
-      "epoch": 0.058704453441295545,
-      "grad_norm": 0.15473420917987823,
-      "learning_rate": 1.16e-05,
-      "loss": 0.6351,
-      "mean_token_accuracy": 0.8287112712860107,
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.08128422498703003,
+      "learning_rate": 1.5531311541251992e-07,
+      "loss": 0.6592,
+      "mean_token_accuracy": 0.8349018841981888,
       "step": 29
     },
     {
-      "epoch": 0.06072874493927125,
-      "grad_norm": 0.14979153871536255,
-      "learning_rate": 1.2e-05,
-      "loss": 0.6087,
-      "mean_token_accuracy": 0.8364104628562927,
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.08368204534053802,
+      "learning_rate": 1.5134741935148418e-07,
+      "loss": 0.6931,
+      "mean_token_accuracy": 0.8207659162580967,
       "step": 30
     },
     {
-      "epoch": 0.06275303643724696,
-      "grad_norm": 0.1627970039844513,
-      "learning_rate": 1.2400000000000002e-05,
-      "loss": 0.6458,
-      "mean_token_accuracy": 0.8359329700469971,
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.08243861049413681,
+      "learning_rate": 1.4726885051885652e-07,
+      "loss": 0.677,
+      "mean_token_accuracy": 0.829660214483738,
       "step": 31
     },
     {
-      "epoch": 0.06477732793522267,
-      "grad_norm": 0.17766571044921875,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 0.6847,
-      "mean_token_accuracy": 0.8317025303840637,
+      "epoch": 2.0,
+      "grad_norm": 0.08243861049413681,
+      "learning_rate": 1.4308637449409705e-07,
+      "loss": 0.6144,
+      "mean_token_accuracy": 0.8387703725269863,
       "step": 32
     },
     {
-      "epoch": 0.06680161943319839,
-      "grad_norm": 0.16728782653808594,
-      "learning_rate": 1.3200000000000002e-05,
-      "loss": 0.5836,
-      "mean_token_accuracy": 0.8538188934326172,
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.1279604583978653,
+      "learning_rate": 1.3880918526722496e-07,
+      "loss": 0.6332,
+      "mean_token_accuracy": 0.8366866521537304,
       "step": 33
     },
     {
-      "epoch": 0.06882591093117409,
-      "grad_norm": 0.21175438165664673,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 0.6896,
-      "mean_token_accuracy": 0.8219363689422607,
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.08055976778268814,
+      "learning_rate": 1.344466850284333e-07,
+      "loss": 0.6709,
+      "mean_token_accuracy": 0.8283472806215286,
       "step": 34
     },
     {
-      "epoch": 0.0708502024291498,
-      "grad_norm": 0.1283571571111679,
-      "learning_rate": 1.4e-05,
-      "loss": 0.5793,
-      "mean_token_accuracy": 0.8478589653968811,
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.08526450395584106,
+      "learning_rate": 1.3000846350003406e-07,
+      "loss": 0.6429,
+      "mean_token_accuracy": 0.8317065984010696,
       "step": 35
     },
     {
-      "epoch": 0.0728744939271255,
-      "grad_norm": 0.15766221284866333,
-      "learning_rate": 1.4400000000000001e-05,
-      "loss": 0.5341,
-      "mean_token_accuracy": 0.8500925302505493,
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.07659583538770676,
+      "learning_rate": 1.2550427685616765e-07,
+      "loss": 0.6523,
+      "mean_token_accuracy": 0.8286443240940571,
       "step": 36
     },
     {
-      "epoch": 0.07489878542510121,
-      "grad_norm": 0.18162991106510162,
-      "learning_rate": 1.48e-05,
-      "loss": 0.7043,
-      "mean_token_accuracy": 0.8179640769958496,
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.07898923009634018,
+      "learning_rate": 1.2094402627661445e-07,
+      "loss": 0.6219,
+      "mean_token_accuracy": 0.8340383470058441,
       "step": 37
     },
     {
-      "epoch": 0.07692307692307693,
-      "grad_norm": 0.1570591777563095,
-      "learning_rate": 1.5200000000000002e-05,
-      "loss": 0.597,
-      "mean_token_accuracy": 0.8484577536582947,
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.07619204372167587,
+      "learning_rate": 1.1633773618185301e-07,
+      "loss": 0.6139,
+      "mean_token_accuracy": 0.8424343690276146,
       "step": 38
     },
     {
-      "epoch": 0.07894736842105263,
-      "grad_norm": 0.21057575941085815,
-      "learning_rate": 1.5600000000000003e-05,
-      "loss": 0.7738,
-      "mean_token_accuracy": 0.7961832284927368,
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.07402335107326508,
+      "learning_rate": 1.1169553219720827e-07,
+      "loss": 0.6798,
+      "mean_token_accuracy": 0.8285421878099442,
       "step": 39
     },
     {
-      "epoch": 0.08097165991902834,
-      "grad_norm": 0.21168996393680573,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.7407,
-      "mean_token_accuracy": 0.808038055896759,
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.08590105175971985,
+      "learning_rate": 1.0702761889452929e-07,
+      "loss": 0.6632,
+      "mean_token_accuracy": 0.8267885185778141,
       "step": 40
     },
     {
-      "epoch": 0.08299595141700405,
-      "grad_norm": 0.13393346965312958,
-      "learning_rate": 1.64e-05,
-      "loss": 0.5588,
-      "mean_token_accuracy": 0.8508504033088684,
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.08060938864946365,
+      "learning_rate": 1.0234425736032605e-07,
+      "loss": 0.6725,
+      "mean_token_accuracy": 0.8298077434301376,
       "step": 41
     },
     {
-      "epoch": 0.08502024291497975,
-      "grad_norm": 0.13981293141841888,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 0.6503,
-      "mean_token_accuracy": 0.8287777304649353,
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.08330174535512924,
+      "learning_rate": 9.765574263967395e-08,
+      "loss": 0.6404,
+      "mean_token_accuracy": 0.8322437591850758,
       "step": 42
     },
     {
-      "epoch": 0.08704453441295547,
-      "grad_norm": 0.17055265605449677,
-      "learning_rate": 1.72e-05,
-      "loss": 0.7479,
-      "mean_token_accuracy": 0.7955465316772461,
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.0802123174071312,
+      "learning_rate": 9.297238110547073e-08,
+      "loss": 0.6333,
+      "mean_token_accuracy": 0.8358821533620358,
       "step": 43
     },
     {
-      "epoch": 0.08906882591093117,
-      "grad_norm": 0.1803690642118454,
-      "learning_rate": 1.76e-05,
-      "loss": 0.6342,
-      "mean_token_accuracy": 0.8441314697265625,
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.07823917269706726,
+      "learning_rate": 8.830446780279175e-08,
+      "loss": 0.633,
+      "mean_token_accuracy": 0.8377866670489311,
       "step": 44
     },
     {
-      "epoch": 0.09109311740890688,
-      "grad_norm": 0.15695004165172577,
-      "learning_rate": 1.8e-05,
-      "loss": 0.6291,
-      "mean_token_accuracy": 0.8388278484344482,
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.07679135352373123,
+      "learning_rate": 8.366226381814697e-08,
+      "loss": 0.6263,
+      "mean_token_accuracy": 0.8360908254981041,
       "step": 45
     },
     {
-      "epoch": 0.0931174089068826,
-      "grad_norm": 0.15088340640068054,
-      "learning_rate": 1.8400000000000003e-05,
-      "loss": 0.5862,
-      "mean_token_accuracy": 0.845678985118866,
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.07800301164388657,
+      "learning_rate": 7.905597372338558e-08,
+      "loss": 0.6464,
+      "mean_token_accuracy": 0.8296581134200096,
       "step": 46
     },
     {
-      "epoch": 0.0951417004048583,
-      "grad_norm": 0.19936536252498627,
-      "learning_rate": 1.88e-05,
-      "loss": 0.683,
-      "mean_token_accuracy": 0.8353626132011414,
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.07923697680234909,
+      "learning_rate": 7.449572314383236e-08,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.8269040808081627,
       "step": 47
     },
     {
-      "epoch": 0.09716599190283401,
-      "grad_norm": 0.18100468814373016,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 0.6099,
-      "mean_token_accuracy": 0.8441349864006042,
+      "epoch": 3.0,
+      "grad_norm": 0.096576027572155,
+      "learning_rate": 6.999153649996594e-08,
+      "loss": 0.6081,
+      "mean_token_accuracy": 0.8382441401481628,
       "step": 48
     },
     {
-      "epoch": 0.09919028340080972,
-      "grad_norm": 0.17626872658729553,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 0.6067,
-      "mean_token_accuracy": 0.8398520350456238,
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.10343178361654282,
+      "learning_rate": 6.555331497156671e-08,
+      "loss": 0.6464,
+      "mean_token_accuracy": 0.8325350135564804,
       "step": 49
     },
     {
-      "epoch": 0.10121457489878542,
-      "grad_norm": 0.16862498223781586,
-      "learning_rate": 2e-05,
-      "loss": 0.6604,
-      "mean_token_accuracy": 0.8260050415992737,
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.07666892558336258,
+      "learning_rate": 6.119081473277501e-08,
+      "loss": 0.6379,
+      "mean_token_accuracy": 0.8326343894004822,
       "step": 50
     },
     {
-      "epoch": 0.10323886639676114,
-      "grad_norm": 0.21076682209968567,
-      "learning_rate": 1.9999749676283775e-05,
-      "loss": 0.7307,
-      "mean_token_accuracy": 0.830464243888855,
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.07846437394618988,
+      "learning_rate": 5.691362550590296e-08,
+      "loss": 0.6438,
+      "mean_token_accuracy": 0.8362045586109161,
       "step": 51
     },
     {
-      "epoch": 0.10526315789473684,
-      "grad_norm": 0.21457180380821228,
-      "learning_rate": 1.999899871766749e-05,
-      "loss": 0.7789,
-      "mean_token_accuracy": 0.8072487711906433,
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.07778104394674301,
+      "learning_rate": 5.2731149481143456e-08,
+      "loss": 0.6296,
+      "mean_token_accuracy": 0.8366052508354187,
       "step": 52
     },
     {
-      "epoch": 0.10728744939271255,
-      "grad_norm": 0.149997279047966,
-      "learning_rate": 1.9997747161747696e-05,
-      "loss": 0.5559,
-      "mean_token_accuracy": 0.8497348427772522,
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.07893098890781403,
+      "learning_rate": 4.8652580648515785e-08,
+      "loss": 0.6426,
+      "mean_token_accuracy": 0.83216942101717,
       "step": 53
     },
     {
-      "epoch": 0.10931174089068826,
-      "grad_norm": 0.1614534556865692,
-      "learning_rate": 1.999599507118322e-05,
-      "loss": 0.6496,
-      "mean_token_accuracy": 0.8317460417747498,
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.08473635464906693,
+      "learning_rate": 4.4686884587480054e-08,
+      "loss": 0.6913,
+      "mean_token_accuracy": 0.8237782865762711,
       "step": 54
     },
     {
-      "epoch": 0.11133603238866396,
-      "grad_norm": 0.3064078390598297,
-      "learning_rate": 1.999374253369202e-05,
-      "loss": 0.7523,
-      "mean_token_accuracy": 0.8246474266052246,
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.08035323023796082,
+      "learning_rate": 4.084277875864776e-08,
+      "loss": 0.635,
+      "mean_token_accuracy": 0.8352065198123455,
       "step": 55
     },
     {
-      "epoch": 0.11336032388663968,
-      "grad_norm": 0.14961189031600952,
-      "learning_rate": 1.999098966204682e-05,
-      "loss": 0.5924,
-      "mean_token_accuracy": 0.8457792401313782,
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.0813593938946724,
+      "learning_rate": 3.712871334091153e-08,
+      "loss": 0.6778,
+      "mean_token_accuracy": 0.8271778710186481,
       "step": 56
     },
     {
-      "epoch": 0.11538461538461539,
-      "grad_norm": 0.26088255643844604,
-      "learning_rate": 1.9987736594069417e-05,
-      "loss": 0.7341,
-      "mean_token_accuracy": 0.8132271766662598,
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.07975359261035919,
+      "learning_rate": 3.355285265611784e-08,
+      "loss": 0.632,
+      "mean_token_accuracy": 0.8359011709690094,
       "step": 57
     },
     {
-      "epoch": 0.11740890688259109,
-      "grad_norm": 0.20378543436527252,
-      "learning_rate": 1.9983983492623832e-05,
-      "loss": 0.6641,
-      "mean_token_accuracy": 0.8299359083175659,
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.07455204427242279,
+      "learning_rate": 3.0123057222115835e-08,
+      "loss": 0.6287,
+      "mean_token_accuracy": 0.8337721861898899,
       "step": 58
     },
     {
-      "epoch": 0.1194331983805668,
-      "grad_norm": 0.24202823638916016,
-      "learning_rate": 1.9979730545608128e-05,
-      "loss": 0.854,
-      "mean_token_accuracy": 0.7922971248626709,
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.07816082239151001,
+      "learning_rate": 2.6846866473633123e-08,
+      "loss": 0.6399,
+      "mean_token_accuracy": 0.835223838686943,
       "step": 59
     },
     {
-      "epoch": 0.1214574898785425,
-      "grad_norm": 0.17491643130779266,
-      "learning_rate": 1.9974977965945e-05,
-      "loss": 0.6383,
-      "mean_token_accuracy": 0.8353293538093567,
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.08226487785577774,
+      "learning_rate": 2.3731482188961815e-08,
+      "loss": 0.6772,
+      "mean_token_accuracy": 0.8268170021474361,
       "step": 60
     },
     {
-      "epoch": 0.12348178137651822,
-      "grad_norm": 0.192088782787323,
-      "learning_rate": 1.996972599157113e-05,
-      "loss": 0.677,
-      "mean_token_accuracy": 0.8339276909828186,
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.08300112187862396,
+      "learning_rate": 2.0783752658887067e-08,
+      "loss": 0.6465,
+      "mean_token_accuracy": 0.8339819870889187,
       "step": 61
     },
     {
-      "epoch": 0.12550607287449392,
-      "grad_norm": 0.2212095856666565,
-      "learning_rate": 1.9963974885425267e-05,
-      "loss": 0.663,
-      "mean_token_accuracy": 0.8254847526550293,
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.08047865331172943,
+      "learning_rate": 1.801015763265754e-08,
+      "loss": 0.6524,
+      "mean_token_accuracy": 0.8351892940700054,
       "step": 62
     },
     {
-      "epoch": 0.12753036437246965,
-      "grad_norm": 0.16962337493896484,
-      "learning_rate": 1.9957724935435065e-05,
-      "loss": 0.7721,
-      "mean_token_accuracy": 0.8089033365249634,
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.07589241862297058,
+      "learning_rate": 1.5416794074090255e-08,
+      "loss": 0.598,
+      "mean_token_accuracy": 0.8409505113959312,
       "step": 63
     },
     {
-      "epoch": 0.12955465587044535,
-      "grad_norm": 0.20441047847270966,
-      "learning_rate": 1.995097645450266e-05,
-      "loss": 0.645,
-      "mean_token_accuracy": 0.8362248539924622,
+      "epoch": 4.0,
+      "grad_norm": 0.07589241862297058,
+      "learning_rate": 1.3009362759120978e-08,
+      "loss": 0.6815,
+      "mean_token_accuracy": 0.8250831876482282,
       "step": 64
     },
     {
-      "epoch": 0.13157894736842105,
-      "grad_norm": 0.20698894560337067,
-      "learning_rate": 1.994372978048903e-05,
-      "loss": 0.7018,
-      "mean_token_accuracy": 0.8064516186714172,
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.12950177490711212,
+      "learning_rate": 1.079315574426135e-08,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.836986843496561,
       "step": 65
     },
     {
-      "epoch": 0.13360323886639677,
-      "grad_norm": 0.13746251165866852,
-      "learning_rate": 1.9935985276197033e-05,
-      "loss": 0.4596,
-      "mean_token_accuracy": 0.8803600072860718,
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.07754851877689362,
+      "learning_rate": 8.773044733510337e-09,
+      "loss": 0.6614,
+      "mean_token_accuracy": 0.829079158604145,
       "step": 66
     },
     {
-      "epoch": 0.13562753036437247,
-      "grad_norm": 0.21085698902606964,
-      "learning_rate": 1.9927743329353295e-05,
-      "loss": 0.6831,
-      "mean_token_accuracy": 0.8211008906364441,
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.07672492414712906,
+      "learning_rate": 6.953470369291348e-09,
+      "loss": 0.6493,
+      "mean_token_accuracy": 0.8342937044799328,
       "step": 67
     },
     {
-      "epoch": 0.13765182186234817,
-      "grad_norm": 0.20604638755321503,
-      "learning_rate": 1.9919004352588768e-05,
-      "loss": 0.701,
-      "mean_token_accuracy": 0.8291005492210388,
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.0766540914773941,
+      "learning_rate": 5.338432470956589e-09,
+      "loss": 0.6079,
+      "mean_token_accuracy": 0.8416761197149754,
       "step": 68
     },
     {
-      "epoch": 0.1396761133603239,
-      "grad_norm": 0.15121905505657196,
-      "learning_rate": 1.9909768783418086e-05,
-      "loss": 0.4735,
-      "mean_token_accuracy": 0.8771929740905762,
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.07487937808036804,
+      "learning_rate": 3.931481242315993e-09,
+      "loss": 0.6135,
+      "mean_token_accuracy": 0.839394424110651,
       "step": 69
     },
     {
-      "epoch": 0.1417004048582996,
-      "grad_norm": 0.13172437250614166,
-      "learning_rate": 1.9900037084217637e-05,
-      "loss": 0.4863,
-      "mean_token_accuracy": 0.8677502870559692,
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.07662676274776459,
+      "learning_rate": 2.7357094675186986e-09,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8233718760311604,
       "step": 70
     },
     {
-      "epoch": 0.1437246963562753,
-      "grad_norm": 0.17617544531822205,
-      "learning_rate": 1.9889809742202454e-05,
-      "loss": 0.5514,
-      "mean_token_accuracy": 0.8573604226112366,
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.08624370396137238,
+      "learning_rate": 1.7537457124423893e-09,
+      "loss": 0.649,
+      "mean_token_accuracy": 0.8329134620726109,
       "step": 71
     },
     {
-      "epoch": 0.145748987854251,
-      "grad_norm": 0.19886170327663422,
-      "learning_rate": 1.9879087269401782e-05,
-      "loss": 0.6655,
-      "mean_token_accuracy": 0.8351132273674011,
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.08158352971076965,
+      "learning_rate": 9.877485465349057e-10,
+      "loss": 0.646,
+      "mean_token_accuracy": 0.8340884000062943,
       "step": 72
     },
     {
-      "epoch": 0.14777327935222673,
-      "grad_norm": 0.205868661403656,
-      "learning_rate": 1.986787020263347e-05,
-      "loss": 0.695,
-      "mean_token_accuracy": 0.8311954736709595,
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.08172011375427246,
+      "learning_rate": 4.394017978101905e-10,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8312402181327343,
       "step": 73
     },
     {
-      "epoch": 0.14979757085020243,
-      "grad_norm": 0.1743887960910797,
-      "learning_rate": 1.9856159103477085e-05,
-      "loss": 0.6084,
-      "mean_token_accuracy": 0.8449342846870422,
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.07922390103340149,
+      "learning_rate": 1.0991085142886269e-10,
+      "loss": 0.666,
+      "mean_token_accuracy": 0.8276109620928764,
       "step": 74
     },
     {
-      "epoch": 0.15182186234817813,
-      "grad_norm": 0.22704970836639404,
-      "learning_rate": 1.98439545582458e-05,
-      "loss": 0.6138,
-      "mean_token_accuracy": 0.8425281047821045,
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.08142900466918945,
+      "learning_rate": 0.0,
+      "loss": 0.659,
+      "mean_token_accuracy": 0.8328934013843536,
       "step": 75
     },
     {
-      "epoch": 0.15384615384615385,
-      "grad_norm": 0.22189851105213165,
-      "learning_rate": 1.9831257177957045e-05,
-      "loss": 0.6893,
-      "mean_token_accuracy": 0.8370082378387451,
-      "step": 76
-    },
-    {
-      "epoch": 0.15587044534412955,
-      "grad_norm": 0.1812412440776825,
-      "learning_rate": 1.9818067598301894e-05,
-      "loss": 0.6025,
-      "mean_token_accuracy": 0.8505423665046692,
-      "step": 77
-    },
-    {
-      "epoch": 0.15789473684210525,
-      "grad_norm": 0.23385366797447205,
-      "learning_rate": 1.9804386479613268e-05,
-      "loss": 0.6333,
-      "mean_token_accuracy": 0.827102780342102,
-      "step": 78
-    },
-    {
-      "epoch": 0.15991902834008098,
-      "grad_norm": 0.2589596211910248,
-      "learning_rate": 1.9790214506832868e-05,
-      "loss": 0.6613,
-      "mean_token_accuracy": 0.8239401578903198,
-      "step": 79
-    },
-    {
-      "epoch": 0.16194331983805668,
-      "grad_norm": 0.24977940320968628,
-      "learning_rate": 1.9775552389476865e-05,
-      "loss": 0.6244,
-      "mean_token_accuracy": 0.8267026543617249,
-      "step": 80
-    },
-    {
-      "epoch": 0.16396761133603238,
-      "grad_norm": 0.22027316689491272,
-      "learning_rate": 1.97604008616004e-05,
-      "loss": 0.6324,
-      "mean_token_accuracy": 0.8370981812477112,
-      "step": 81
-    },
-    {
-      "epoch": 0.1659919028340081,
-      "grad_norm": 0.1776619702577591,
-      "learning_rate": 1.9744760681760832e-05,
-      "loss": 0.5265,
-      "mean_token_accuracy": 0.8645498156547546,
-      "step": 82
-    },
-    {
-      "epoch": 0.1680161943319838,
-      "grad_norm": 0.19093215465545654,
-      "learning_rate": 1.9728632632979746e-05,
-      "loss": 0.6313,
-      "mean_token_accuracy": 0.8432900309562683,
-      "step": 83
-    },
-    {
-      "epoch": 0.1700404858299595,
-      "grad_norm": 0.22855107486248016,
-      "learning_rate": 1.9712017522703764e-05,
-      "loss": 0.5718,
-      "mean_token_accuracy": 0.8341137170791626,
-      "step": 84
-    },
-    {
-      "epoch": 0.1720647773279352,
-      "grad_norm": 0.20743606984615326,
-      "learning_rate": 1.9694916182764113e-05,
-      "loss": 0.574,
-      "mean_token_accuracy": 0.8542357087135315,
-      "step": 85
-    },
-    {
-      "epoch": 0.17408906882591094,
-      "grad_norm": 0.32714810967445374,
-      "learning_rate": 1.967732946933499e-05,
-      "loss": 0.6813,
-      "mean_token_accuracy": 0.8177083134651184,
-      "step": 86
-    },
-    {
-      "epoch": 0.17611336032388664,
-      "grad_norm": 0.2417590320110321,
-      "learning_rate": 1.9659258262890683e-05,
-      "loss": 0.7423,
-      "mean_token_accuracy": 0.8204029202461243,
-      "step": 87
-    },
-    {
-      "epoch": 0.17813765182186234,
-      "grad_norm": 0.22506733238697052,
-      "learning_rate": 1.9640703468161508e-05,
-      "loss": 0.6379,
-      "mean_token_accuracy": 0.835094153881073,
-      "step": 88
-    },
-    {
-      "epoch": 0.18016194331983806,
-      "grad_norm": 0.2758257985115051,
-      "learning_rate": 1.9621666014088495e-05,
-      "loss": 0.6055,
-      "mean_token_accuracy": 0.8465368747711182,
-      "step": 89
-    },
-    {
-      "epoch": 0.18218623481781376,
-      "grad_norm": 0.26795482635498047,
-      "learning_rate": 1.9602146853776894e-05,
-      "loss": 0.6584,
-      "mean_token_accuracy": 0.8275015354156494,
-      "step": 90
-    },
-    {
-      "epoch": 0.18421052631578946,
-      "grad_norm": 0.22383807599544525,
-      "learning_rate": 1.9582146964448457e-05,
-      "loss": 0.5799,
-      "mean_token_accuracy": 0.842945396900177,
-      "step": 91
-    },
-    {
-      "epoch": 0.1862348178137652,
-      "grad_norm": 0.21481236815452576,
-      "learning_rate": 1.956166734739251e-05,
-      "loss": 0.5439,
-      "mean_token_accuracy": 0.8482651710510254,
-      "step": 92
-    },
-    {
-      "epoch": 0.1882591093117409,
-      "grad_norm": 0.19826941192150116,
-      "learning_rate": 1.954070902791582e-05,
-      "loss": 0.5595,
-      "mean_token_accuracy": 0.8426966071128845,
-      "step": 93
-    },
-    {
-      "epoch": 0.1902834008097166,
-      "grad_norm": 0.2748395800590515,
-      "learning_rate": 1.9519273055291266e-05,
-      "loss": 0.6142,
-      "mean_token_accuracy": 0.8315182328224182,
-      "step": 94
-    },
-    {
-      "epoch": 0.19230769230769232,
-      "grad_norm": 0.23335261642932892,
-      "learning_rate": 1.949736050270532e-05,
-      "loss": 0.5591,
-      "mean_token_accuracy": 0.8485158681869507,
-      "step": 95
-    },
-    {
-      "epoch": 0.19433198380566802,
-      "grad_norm": 0.26217031478881836,
-      "learning_rate": 1.9474972467204298e-05,
-      "loss": 0.6583,
-      "mean_token_accuracy": 0.8374301791191101,
-      "step": 96
-    },
-    {
-      "epoch": 0.19635627530364372,
-      "grad_norm": 0.2468375861644745,
-      "learning_rate": 1.945211006963945e-05,
-      "loss": 0.4778,
-      "mean_token_accuracy": 0.8702101111412048,
-      "step": 97
-    },
-    {
-      "epoch": 0.19838056680161945,
-      "grad_norm": 0.2425367534160614,
-      "learning_rate": 1.9428774454610845e-05,
-      "loss": 0.5753,
-      "mean_token_accuracy": 0.8493397235870361,
-      "step": 98
-    },
-    {
-      "epoch": 0.20040485829959515,
-      "grad_norm": 0.21901822090148926,
-      "learning_rate": 1.9404966790410047e-05,
-      "loss": 0.6074,
-      "mean_token_accuracy": 0.851259708404541,
-      "step": 99
-    },
-    {
-      "epoch": 0.20242914979757085,
-      "grad_norm": 0.17670230567455292,
-      "learning_rate": 1.938068826896166e-05,
-      "loss": 0.6112,
-      "mean_token_accuracy": 0.8372820615768433,
-      "step": 100
-    },
-    {
-      "epoch": 0.20445344129554655,
-      "grad_norm": 0.21904608607292175,
-      "learning_rate": 1.9355940105763622e-05,
-      "loss": 0.5904,
-      "mean_token_accuracy": 0.8459681272506714,
-      "step": 101
-    },
-    {
-      "epoch": 0.20647773279352227,
-      "grad_norm": 0.2911970317363739,
-      "learning_rate": 1.9330723539826373e-05,
-      "loss": 0.7404,
-      "mean_token_accuracy": 0.8156565427780151,
-      "step": 102
-    },
-    {
-      "epoch": 0.20850202429149797,
-      "grad_norm": 0.23587605357170105,
-      "learning_rate": 1.930503983361081e-05,
-      "loss": 0.5728,
-      "mean_token_accuracy": 0.8425158262252808,
-      "step": 103
-    },
-    {
-      "epoch": 0.21052631578947367,
-      "grad_norm": 0.22998309135437012,
-      "learning_rate": 1.9278890272965097e-05,
-      "loss": 0.6436,
-      "mean_token_accuracy": 0.836202085018158,
-      "step": 104
-    },
-    {
-      "epoch": 0.2125506072874494,
-      "grad_norm": 0.21080300211906433,
-      "learning_rate": 1.925227616706026e-05,
-      "loss": 0.5063,
-      "mean_token_accuracy": 0.868156909942627,
-      "step": 105
-    },
-    {
-      "epoch": 0.2145748987854251,
-      "grad_norm": 0.3365795910358429,
-      "learning_rate": 1.9225198848324687e-05,
-      "loss": 0.6685,
-      "mean_token_accuracy": 0.8274034857749939,
-      "step": 106
-    },
-    {
-      "epoch": 0.2165991902834008,
-      "grad_norm": 0.25566524267196655,
-      "learning_rate": 1.9197659672377388e-05,
-      "loss": 0.5392,
-      "mean_token_accuracy": 0.8387492299079895,
-      "step": 107
-    },
-    {
-      "epoch": 0.21862348178137653,
-      "grad_norm": 0.24027182161808014,
-      "learning_rate": 1.9169660017960135e-05,
-      "loss": 0.5435,
-      "mean_token_accuracy": 0.8624338507652283,
-      "step": 108
-    },
-    {
-      "epoch": 0.22064777327935223,
-      "grad_norm": 0.2666286826133728,
-      "learning_rate": 1.9141201286868435e-05,
-      "loss": 0.7356,
-      "mean_token_accuracy": 0.8124300241470337,
-      "step": 109
-    },
-    {
-      "epoch": 0.22267206477732793,
-      "grad_norm": 0.21396684646606445,
-      "learning_rate": 1.911228490388136e-05,
-      "loss": 0.4923,
-      "mean_token_accuracy": 0.8706277012825012,
-      "step": 110
-    },
-    {
-      "epoch": 0.22469635627530365,
-      "grad_norm": 0.23490038514137268,
-      "learning_rate": 1.908291231669019e-05,
-      "loss": 0.651,
-      "mean_token_accuracy": 0.8370558619499207,
-      "step": 111
-    },
-    {
-      "epoch": 0.22672064777327935,
-      "grad_norm": 0.2001212239265442,
-      "learning_rate": 1.905308499582597e-05,
-      "loss": 0.5137,
-      "mean_token_accuracy": 0.8653126955032349,
-      "step": 112
-    },
-    {
-      "epoch": 0.22874493927125505,
-      "grad_norm": 0.26428893208503723,
-      "learning_rate": 1.9022804434585854e-05,
-      "loss": 0.5852,
-      "mean_token_accuracy": 0.8668292760848999,
-      "step": 113
-    },
-    {
-      "epoch": 0.23076923076923078,
-      "grad_norm": 0.18154019117355347,
-      "learning_rate": 1.8992072148958368e-05,
-      "loss": 0.4395,
-      "mean_token_accuracy": 0.8839049935340881,
-      "step": 114
-    },
-    {
-      "epoch": 0.23279352226720648,
-      "grad_norm": 0.1800609678030014,
-      "learning_rate": 1.8960889677547506e-05,
-      "loss": 0.4731,
-      "mean_token_accuracy": 0.8780291080474854,
-      "step": 115
-    },
-    {
-      "epoch": 0.23481781376518218,
-      "grad_norm": 0.24386590719223022,
-      "learning_rate": 1.8929258581495688e-05,
-      "loss": 0.517,
-      "mean_token_accuracy": 0.8482933044433594,
-      "step": 116
-    },
-    {
-      "epoch": 0.23684210526315788,
-      "grad_norm": 0.2463385909795761,
-      "learning_rate": 1.8897180444405615e-05,
-      "loss": 0.6356,
-      "mean_token_accuracy": 0.8325533270835876,
-      "step": 117
-    },
-    {
-      "epoch": 0.2388663967611336,
-      "grad_norm": 0.2074287086725235,
-      "learning_rate": 1.8864656872260985e-05,
-      "loss": 0.5643,
-      "mean_token_accuracy": 0.8535178899765015,
-      "step": 118
-    },
-    {
-      "epoch": 0.2408906882591093,
-      "grad_norm": 0.2767307758331299,
-      "learning_rate": 1.8831689493346095e-05,
-      "loss": 0.5837,
-      "mean_token_accuracy": 0.8493902683258057,
-      "step": 119
-    },
-    {
-      "epoch": 0.242914979757085,
-      "grad_norm": 0.2797743082046509,
-      "learning_rate": 1.8798279958164295e-05,
-      "loss": 0.6587,
-      "mean_token_accuracy": 0.8430637121200562,
-      "step": 120
-    },
-    {
-      "epoch": 0.24493927125506074,
-      "grad_norm": 0.28018176555633545,
-      "learning_rate": 1.8764429939355394e-05,
-      "loss": 0.5805,
-      "mean_token_accuracy": 0.8473895788192749,
-      "step": 121
-    },
-    {
-      "epoch": 0.24696356275303644,
-      "grad_norm": 0.1992814540863037,
-      "learning_rate": 1.8730141131611882e-05,
-      "loss": 0.529,
-      "mean_token_accuracy": 0.8487154841423035,
-      "step": 122
-    },
-    {
-      "epoch": 0.24898785425101214,
-      "grad_norm": 0.3097265362739563,
-      "learning_rate": 1.8695415251594123e-05,
-      "loss": 0.6846,
-      "mean_token_accuracy": 0.8307022452354431,
-      "step": 123
-    },
-    {
-      "epoch": 0.25101214574898784,
-      "grad_norm": 0.31372424960136414,
-      "learning_rate": 1.866025403784439e-05,
-      "loss": 0.7371,
-      "mean_token_accuracy": 0.8107274770736694,
-      "step": 124
-    },
-    {
-      "epoch": 0.25303643724696356,
-      "grad_norm": 0.30829596519470215,
-      "learning_rate": 1.8624659250699807e-05,
-      "loss": 0.607,
-      "mean_token_accuracy": 0.859689474105835,
-      "step": 125
-    },
-    {
-      "epoch": 0.2550607287449393,
-      "grad_norm": 0.29508087038993835,
-      "learning_rate": 1.8588632672204264e-05,
-      "loss": 0.4883,
-      "mean_token_accuracy": 0.8674641251564026,
-      "step": 126
-    },
-    {
-      "epoch": 0.25708502024291496,
-      "grad_norm": 0.2224210500717163,
-      "learning_rate": 1.8552176106019156e-05,
-      "loss": 0.5458,
-      "mean_token_accuracy": 0.85467129945755,
-      "step": 127
-    },
-    {
-      "epoch": 0.2591093117408907,
-      "grad_norm": 0.2578910291194916,
-      "learning_rate": 1.8515291377333114e-05,
-      "loss": 0.5492,
-      "mean_token_accuracy": 0.8488425016403198,
-      "step": 128
-    },
-    {
-      "epoch": 0.2611336032388664,
-      "grad_norm": 0.2848791480064392,
-      "learning_rate": 1.847798033277061e-05,
-      "loss": 0.5902,
-      "mean_token_accuracy": 0.854683518409729,
-      "step": 129
-    },
-    {
-      "epoch": 0.2631578947368421,
-      "grad_norm": 0.2781189978122711,
-      "learning_rate": 1.8440244840299507e-05,
-      "loss": 0.6038,
-      "mean_token_accuracy": 0.8376891613006592,
-      "step": 130
-    },
-    {
-      "epoch": 0.2651821862348178,
-      "grad_norm": 0.23901361227035522,
-      "learning_rate": 1.8402086789137547e-05,
-      "loss": 0.6103,
-      "mean_token_accuracy": 0.8358832001686096,
-      "step": 131
-    },
-    {
-      "epoch": 0.26720647773279355,
-      "grad_norm": 0.23731403052806854,
-      "learning_rate": 1.8363508089657763e-05,
-      "loss": 0.5742,
-      "mean_token_accuracy": 0.8496110439300537,
-      "step": 132
-    },
-    {
-      "epoch": 0.2692307692307692,
-      "grad_norm": 0.36295345425605774,
-      "learning_rate": 1.8324510673292844e-05,
-      "loss": 0.6462,
-      "mean_token_accuracy": 0.8392658233642578,
-      "step": 133
-    },
-    {
-      "epoch": 0.27125506072874495,
-      "grad_norm": 0.23564256727695465,
-      "learning_rate": 1.8285096492438424e-05,
-      "loss": 0.5936,
-      "mean_token_accuracy": 0.8450990319252014,
-      "step": 134
-    },
-    {
-      "epoch": 0.2732793522267207,
-      "grad_norm": 0.21983715891838074,
-      "learning_rate": 1.8245267520355348e-05,
-      "loss": 0.5966,
-      "mean_token_accuracy": 0.8399209380149841,
-      "step": 135
-    },
-    {
-      "epoch": 0.27530364372469635,
-      "grad_norm": 0.2106492817401886,
-      "learning_rate": 1.8205025751070878e-05,
-      "loss": 0.5501,
-      "mean_token_accuracy": 0.8545739054679871,
-      "step": 136
-    },
-    {
-      "epoch": 0.2773279352226721,
-      "grad_norm": 0.21314162015914917,
-      "learning_rate": 1.8164373199278858e-05,
-      "loss": 0.5591,
-      "mean_token_accuracy": 0.8543897271156311,
-      "step": 137
-    },
-    {
-      "epoch": 0.2793522267206478,
-      "grad_norm": 0.21074095368385315,
-      "learning_rate": 1.812331190023886e-05,
-      "loss": 0.4741,
-      "mean_token_accuracy": 0.8711554408073425,
-      "step": 138
-    },
-    {
-      "epoch": 0.2813765182186235,
-      "grad_norm": 0.23884296417236328,
-      "learning_rate": 1.8081843909674277e-05,
-      "loss": 0.5084,
-      "mean_token_accuracy": 0.8729857802391052,
-      "step": 139
-    },
-    {
-      "epoch": 0.2834008097165992,
-      "grad_norm": 0.3194005787372589,
-      "learning_rate": 1.8039971303669407e-05,
-      "loss": 0.6815,
-      "mean_token_accuracy": 0.807113528251648,
-      "step": 140
-    },
-    {
-      "epoch": 0.2854251012145749,
-      "grad_norm": 0.2949461042881012,
-      "learning_rate": 1.799769617856552e-05,
-      "loss": 0.5174,
-      "mean_token_accuracy": 0.8580542206764221,
-      "step": 141
-    },
-    {
-      "epoch": 0.2874493927125506,
-      "grad_norm": 0.2602043151855469,
-      "learning_rate": 1.79550206508559e-05,
-      "loss": 0.6328,
-      "mean_token_accuracy": 0.8295890688896179,
-      "step": 142
-    },
-    {
-      "epoch": 0.2894736842105263,
-      "grad_norm": 0.27205556631088257,
-      "learning_rate": 1.7911946857079886e-05,
-      "loss": 0.5921,
-      "mean_token_accuracy": 0.8282608985900879,
-      "step": 143
-    },
-    {
-      "epoch": 0.291497975708502,
-      "grad_norm": 0.23467545211315155,
-      "learning_rate": 1.78684769537159e-05,
-      "loss": 0.554,
-      "mean_token_accuracy": 0.8527607321739197,
-      "step": 144
-    },
-    {
-      "epoch": 0.2935222672064777,
-      "grad_norm": 0.27535098791122437,
-      "learning_rate": 1.78246131170735e-05,
-      "loss": 0.4676,
-      "mean_token_accuracy": 0.871345043182373,
-      "step": 145
-    },
-    {
-      "epoch": 0.29554655870445345,
-      "grad_norm": 0.25174370408058167,
-      "learning_rate": 1.7780357543184396e-05,
-      "loss": 0.5593,
-      "mean_token_accuracy": 0.8446115255355835,
-      "step": 146
-    },
-    {
-      "epoch": 0.2975708502024291,
-      "grad_norm": 0.2598954439163208,
-      "learning_rate": 1.773571244769254e-05,
-      "loss": 0.5884,
-      "mean_token_accuracy": 0.8401237726211548,
-      "step": 147
-    },
-    {
-      "epoch": 0.29959514170040485,
-      "grad_norm": 0.40423285961151123,
-      "learning_rate": 1.769068006574317e-05,
-      "loss": 0.6519,
-      "mean_token_accuracy": 0.8430913090705872,
-      "step": 148
-    },
-    {
-      "epoch": 0.3016194331983806,
-      "grad_norm": 0.2684638500213623,
-      "learning_rate": 1.7645262651870926e-05,
-      "loss": 0.4679,
-      "mean_token_accuracy": 0.8538409471511841,
-      "step": 149
-    },
-    {
-      "epoch": 0.30364372469635625,
-      "grad_norm": 0.244306743144989,
-      "learning_rate": 1.7599462479886976e-05,
-      "loss": 0.4809,
-      "mean_token_accuracy": 0.8723497986793518,
-      "step": 150
-    },
-    {
-      "epoch": 0.305668016194332,
-      "grad_norm": 0.27933186292648315,
-      "learning_rate": 1.755328184276517e-05,
-      "loss": 0.6049,
-      "mean_token_accuracy": 0.8480749130249023,
-      "step": 151
-    },
-    {
-      "epoch": 0.3076923076923077,
-      "grad_norm": 0.2595280110836029,
-      "learning_rate": 1.7506723052527243e-05,
-      "loss": 0.5375,
-      "mean_token_accuracy": 0.8573270440101624,
-      "step": 152
-    },
-    {
-      "epoch": 0.3097165991902834,
-      "grad_norm": 0.36356401443481445,
-      "learning_rate": 1.7459788440127083e-05,
-      "loss": 0.6101,
-      "mean_token_accuracy": 0.8315719962120056,
-      "step": 153
-    },
-    {
-      "epoch": 0.3117408906882591,
-      "grad_norm": 0.28417396545410156,
-      "learning_rate": 1.7412480355334006e-05,
-      "loss": 0.6359,
-      "mean_token_accuracy": 0.8482972383499146,
-      "step": 154
-    },
-    {
-      "epoch": 0.31376518218623484,
-      "grad_norm": 0.3155061900615692,
-      "learning_rate": 1.7364801166615124e-05,
-      "loss": 0.6643,
-      "mean_token_accuracy": 0.8326530456542969,
-      "step": 155
-    },
-    {
-      "epoch": 0.3157894736842105,
-      "grad_norm": 0.21922272443771362,
-      "learning_rate": 1.7316753261016782e-05,
-      "loss": 0.5184,
-      "mean_token_accuracy": 0.8553861379623413,
-      "step": 156
-    },
-    {
-      "epoch": 0.31781376518218624,
-      "grad_norm": 0.3232700228691101,
-      "learning_rate": 1.7268339044045044e-05,
-      "loss": 0.5586,
-      "mean_token_accuracy": 0.8481873273849487,
-      "step": 157
-    },
-    {
-      "epoch": 0.31983805668016196,
-      "grad_norm": 0.2255302518606186,
-      "learning_rate": 1.7219560939545246e-05,
-      "loss": 0.5903,
-      "mean_token_accuracy": 0.8607842922210693,
-      "step": 158
-    },
-    {
-      "epoch": 0.32186234817813764,
-      "grad_norm": 0.17876094579696655,
-      "learning_rate": 1.7170421389580666e-05,
-      "loss": 0.4914,
-      "mean_token_accuracy": 0.8620192408561707,
-      "step": 159
-    },
-    {
-      "epoch": 0.32388663967611336,
-      "grad_norm": 0.22846977412700653,
-      "learning_rate": 1.712092285431026e-05,
-      "loss": 0.4533,
-      "mean_token_accuracy": 0.8775308728218079,
-      "step": 160
-    },
-    {
-      "epoch": 0.3259109311740891,
-      "grad_norm": 0.2580243647098541,
-      "learning_rate": 1.7071067811865477e-05,
-      "loss": 0.6124,
-      "mean_token_accuracy": 0.8417159914970398,
-      "step": 161
-    },
-    {
-      "epoch": 0.32793522267206476,
-      "grad_norm": 0.303861141204834,
-      "learning_rate": 1.702085875822623e-05,
-      "loss": 0.5966,
-      "mean_token_accuracy": 0.8357366919517517,
-      "step": 162
-    },
-    {
-      "epoch": 0.3299595141700405,
-      "grad_norm": 0.26614582538604736,
-      "learning_rate": 1.6970298207095887e-05,
-      "loss": 0.5795,
-      "mean_token_accuracy": 0.8428571224212646,
-      "step": 163
-    },
-    {
-      "epoch": 0.3319838056680162,
-      "grad_norm": 0.2822970151901245,
-      "learning_rate": 1.6919388689775463e-05,
-      "loss": 0.5743,
-      "mean_token_accuracy": 0.8430072069168091,
-      "step": 164
-    },
-    {
-      "epoch": 0.3340080971659919,
-      "grad_norm": 0.29884374141693115,
-      "learning_rate": 1.6868132755036875e-05,
-      "loss": 0.6735,
-      "mean_token_accuracy": 0.8227627873420715,
-      "step": 165
-    },
-    {
-      "epoch": 0.3360323886639676,
-      "grad_norm": 0.32726043462753296,
-      "learning_rate": 1.681653296899533e-05,
-      "loss": 0.6551,
-      "mean_token_accuracy": 0.8337730765342712,
-      "step": 166
-    },
-    {
-      "epoch": 0.33805668016194335,
-      "grad_norm": 0.16429030895233154,
-      "learning_rate": 1.676459191498087e-05,
-      "loss": 0.4281,
-      "mean_token_accuracy": 0.8818212151527405,
-      "step": 167
-    },
-    {
-      "epoch": 0.340080971659919,
-      "grad_norm": 0.2392423003911972,
-      "learning_rate": 1.6712312193409032e-05,
-      "loss": 0.5083,
-      "mean_token_accuracy": 0.8621307015419006,
-      "step": 168
-    },
-    {
-      "epoch": 0.34210526315789475,
-      "grad_norm": 0.20973503589630127,
-      "learning_rate": 1.6659696421650645e-05,
-      "loss": 0.454,
-      "mean_token_accuracy": 0.8688760995864868,
-      "step": 169
-    },
-    {
-      "epoch": 0.3441295546558704,
-      "grad_norm": 0.18245910108089447,
-      "learning_rate": 1.6606747233900816e-05,
-      "loss": 0.4816,
-      "mean_token_accuracy": 0.8563218116760254,
-      "step": 170
-    },
-    {
-      "epoch": 0.34615384615384615,
-      "grad_norm": 0.2780061662197113,
-      "learning_rate": 1.655346728104704e-05,
-      "loss": 0.5072,
-      "mean_token_accuracy": 0.8650442361831665,
-      "step": 171
-    },
-    {
-      "epoch": 0.3481781376518219,
-      "grad_norm": 0.319735050201416,
-      "learning_rate": 1.6499859230536468e-05,
-      "loss": 0.6437,
-      "mean_token_accuracy": 0.8298412561416626,
-      "step": 172
-    },
-    {
-      "epoch": 0.35020242914979755,
-      "grad_norm": 0.27722129225730896,
-      "learning_rate": 1.6445925766242392e-05,
-      "loss": 0.6204,
-      "mean_token_accuracy": 0.8339704871177673,
-      "step": 173
-    },
-    {
-      "epoch": 0.3522267206477733,
-      "grad_norm": 0.2680011987686157,
-      "learning_rate": 1.639166958832985e-05,
-      "loss": 0.5956,
-      "mean_token_accuracy": 0.8411633372306824,
-      "step": 174
-    },
-    {
-      "epoch": 0.354251012145749,
-      "grad_norm": 0.34449663758277893,
-      "learning_rate": 1.6337093413120463e-05,
-      "loss": 0.5733,
-      "mean_token_accuracy": 0.8423880338668823,
-      "step": 175
-    },
-    {
-      "epoch": 0.3562753036437247,
-      "grad_norm": 0.24189139902591705,
-      "learning_rate": 1.6282199972956425e-05,
-      "loss": 0.5352,
-      "mean_token_accuracy": 0.8527716398239136,
-      "step": 176
-    },
-    {
-      "epoch": 0.3582995951417004,
-      "grad_norm": 0.31007909774780273,
-      "learning_rate": 1.6226992016063726e-05,
-      "loss": 0.5431,
-      "mean_token_accuracy": 0.8473648428916931,
-      "step": 177
-    },
-    {
-      "epoch": 0.3603238866396761,
-      "grad_norm": 0.19545142352581024,
-      "learning_rate": 1.6171472306414554e-05,
-      "loss": 0.4527,
-      "mean_token_accuracy": 0.8697916865348816,
-      "step": 178
-    },
-    {
-      "epoch": 0.3623481781376518,
-      "grad_norm": 0.35753557085990906,
-      "learning_rate": 1.6115643623588915e-05,
-      "loss": 0.5145,
-      "mean_token_accuracy": 0.8581011295318604,
-      "step": 179
-    },
-    {
-      "epoch": 0.3643724696356275,
-      "grad_norm": 0.23397327959537506,
-      "learning_rate": 1.6059508762635482e-05,
-      "loss": 0.4717,
-      "mean_token_accuracy": 0.860744297504425,
-      "step": 180
-    },
-    {
-      "epoch": 0.36639676113360325,
-      "grad_norm": 0.29638534784317017,
-      "learning_rate": 1.6003070533931657e-05,
-      "loss": 0.4542,
-      "mean_token_accuracy": 0.8717948794364929,
-      "step": 181
-    },
-    {
-      "epoch": 0.3684210526315789,
-      "grad_norm": 0.23162566125392914,
-      "learning_rate": 1.594633176304287e-05,
-      "loss": 0.4617,
-      "mean_token_accuracy": 0.8729259371757507,
-      "step": 182
-    },
-    {
-      "epoch": 0.37044534412955465,
-      "grad_norm": 0.2598213851451874,
-      "learning_rate": 1.588929529058111e-05,
-      "loss": 0.5532,
-      "mean_token_accuracy": 0.8479591608047485,
-      "step": 183
-    },
-    {
-      "epoch": 0.3724696356275304,
-      "grad_norm": 0.3145219385623932,
-      "learning_rate": 1.5831963972062734e-05,
-      "loss": 0.5311,
-      "mean_token_accuracy": 0.8493150472640991,
-      "step": 184
-    },
-    {
-      "epoch": 0.37449392712550605,
-      "grad_norm": 0.26244890689849854,
-      "learning_rate": 1.5774340677765483e-05,
-      "loss": 0.545,
-      "mean_token_accuracy": 0.8599779605865479,
-      "step": 185
-    },
-    {
-      "epoch": 0.3765182186234818,
-      "grad_norm": 0.1403060108423233,
-      "learning_rate": 1.5716428292584788e-05,
-      "loss": 0.3864,
-      "mean_token_accuracy": 0.8956466317176819,
-      "step": 186
-    },
-    {
-      "epoch": 0.3785425101214575,
-      "grad_norm": 0.3069169223308563,
-      "learning_rate": 1.5658229715889345e-05,
-      "loss": 0.5963,
-      "mean_token_accuracy": 0.8490465879440308,
-      "step": 187
-    },
-    {
-      "epoch": 0.3805668016194332,
-      "grad_norm": 0.27172955870628357,
-      "learning_rate": 1.5599747861375957e-05,
-      "loss": 0.5844,
-      "mean_token_accuracy": 0.8412802219390869,
-      "step": 188
-    },
-    {
-      "epoch": 0.3825910931174089,
-      "grad_norm": 0.2943626046180725,
-      "learning_rate": 1.5540985656923648e-05,
-      "loss": 0.5273,
-      "mean_token_accuracy": 0.850491464138031,
-      "step": 189
-    },
-    {
-      "epoch": 0.38461538461538464,
-      "grad_norm": 0.30295711755752563,
-      "learning_rate": 1.54819460444471e-05,
-      "loss": 0.6003,
-      "mean_token_accuracy": 0.8302251100540161,
-      "step": 190
-    },
-    {
-      "epoch": 0.3866396761133603,
-      "grad_norm": 0.21159401535987854,
-      "learning_rate": 1.5422631979749354e-05,
-      "loss": 0.582,
-      "mean_token_accuracy": 0.8241075873374939,
-      "step": 191
-    },
-    {
-      "epoch": 0.38866396761133604,
-      "grad_norm": 0.2957298755645752,
-      "learning_rate": 1.5363046432373824e-05,
-      "loss": 0.5709,
-      "mean_token_accuracy": 0.8461936712265015,
-      "step": 192
-    },
-    {
-      "epoch": 0.39068825910931176,
-      "grad_norm": 0.3318915069103241,
-      "learning_rate": 1.5303192385455652e-05,
-      "loss": 0.5812,
-      "mean_token_accuracy": 0.8348472118377686,
-      "step": 193
-    },
-    {
-      "epoch": 0.39271255060728744,
-      "grad_norm": 0.18238049745559692,
-      "learning_rate": 1.5243072835572319e-05,
-      "loss": 0.3858,
-      "mean_token_accuracy": 0.8846918344497681,
-      "step": 194
-    },
-    {
-      "epoch": 0.39473684210526316,
-      "grad_norm": 0.22623980045318604,
-      "learning_rate": 1.5182690792593659e-05,
-      "loss": 0.4931,
-      "mean_token_accuracy": 0.8711094856262207,
-      "step": 195
-    },
-    {
-      "epoch": 0.3967611336032389,
-      "grad_norm": 0.30967429280281067,
-      "learning_rate": 1.5122049279531143e-05,
-      "loss": 0.4871,
-      "mean_token_accuracy": 0.8649900555610657,
-      "step": 196
-    },
-    {
-      "epoch": 0.39878542510121456,
-      "grad_norm": 0.21760259568691254,
-      "learning_rate": 1.5061151332386565e-05,
-      "loss": 0.4727,
-      "mean_token_accuracy": 0.8606798052787781,
-      "step": 197
-    },
-    {
-      "epoch": 0.4008097165991903,
-      "grad_norm": 0.26807570457458496,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 0.459,
-      "mean_token_accuracy": 0.8632004261016846,
-      "step": 198
-    },
-    {
-      "epoch": 0.402834008097166,
-      "grad_norm": 0.26478174328804016,
-      "learning_rate": 1.4938598343897215e-05,
-      "loss": 0.5557,
-      "mean_token_accuracy": 0.8421052694320679,
-      "step": 199
-    },
-    {
-      "epoch": 0.4048582995951417,
-      "grad_norm": 0.29175126552581787,
-      "learning_rate": 1.4876949438136348e-05,
-      "loss": 0.5912,
-      "mean_token_accuracy": 0.8419395685195923,
-      "step": 200
-    },
-    {
-      "epoch": 0.4068825910931174,
-      "grad_norm": 0.22912077605724335,
-      "learning_rate": 1.4815056369154039e-05,
-      "loss": 0.525,
-      "mean_token_accuracy": 0.8536022901535034,
-      "step": 201
-    },
-    {
-      "epoch": 0.4089068825910931,
-      "grad_norm": 0.25623202323913574,
-      "learning_rate": 1.47529222356109e-05,
-      "loss": 0.5464,
-      "mean_token_accuracy": 0.8433228135108948,
-      "step": 202
-    },
-    {
-      "epoch": 0.4109311740890688,
-      "grad_norm": 0.2668095529079437,
-      "learning_rate": 1.4690550148236371e-05,
-      "loss": 0.5017,
-      "mean_token_accuracy": 0.8548858761787415,
-      "step": 203
-    },
-    {
-      "epoch": 0.41295546558704455,
-      "grad_norm": 0.27633169293403625,
-      "learning_rate": 1.4627943229672992e-05,
-      "loss": 0.5546,
-      "mean_token_accuracy": 0.861094057559967,
-      "step": 204
-    },
-    {
-      "epoch": 0.4149797570850202,
-      "grad_norm": 0.2418694645166397,
-      "learning_rate": 1.4565104614320065e-05,
-      "loss": 0.5139,
-      "mean_token_accuracy": 0.8555907607078552,
-      "step": 205
-    },
-    {
-      "epoch": 0.41700404858299595,
-      "grad_norm": 0.2270972579717636,
-      "learning_rate": 1.4502037448176734e-05,
-      "loss": 0.4913,
-      "mean_token_accuracy": 0.8657086491584778,
-      "step": 206
-    },
-    {
-      "epoch": 0.4190283400809717,
-      "grad_norm": 0.22995734214782715,
-      "learning_rate": 1.4438744888684481e-05,
-      "loss": 0.4249,
-      "mean_token_accuracy": 0.8714148998260498,
-      "step": 207
-    },
-    {
-      "epoch": 0.42105263157894735,
-      "grad_norm": 0.25857847929000854,
-      "learning_rate": 1.4375230104569044e-05,
-      "loss": 0.4941,
-      "mean_token_accuracy": 0.8693322539329529,
-      "step": 208
-    },
-    {
-      "epoch": 0.4230769230769231,
-      "grad_norm": 0.26808950304985046,
-      "learning_rate": 1.4311496275681785e-05,
-      "loss": 0.5223,
-      "mean_token_accuracy": 0.8595696687698364,
-      "step": 209
-    },
-    {
-      "epoch": 0.4251012145748988,
-      "grad_norm": 0.22069592773914337,
-      "learning_rate": 1.424754659284048e-05,
-      "loss": 0.4852,
-      "mean_token_accuracy": 0.8667218685150146,
-      "step": 210
-    },
-    {
-      "epoch": 0.4271255060728745,
-      "grad_norm": 0.23519064486026764,
-      "learning_rate": 1.418338425766958e-05,
-      "loss": 0.4135,
-      "mean_token_accuracy": 0.8849557638168335,
-      "step": 211
-    },
-    {
-      "epoch": 0.4291497975708502,
-      "grad_norm": 0.18871621787548065,
-      "learning_rate": 1.4119012482439929e-05,
-      "loss": 0.4434,
-      "mean_token_accuracy": 0.875,
-      "step": 212
-    },
-    {
-      "epoch": 0.4311740890688259,
-      "grad_norm": 0.28710120916366577,
-      "learning_rate": 1.4054434489907916e-05,
-      "loss": 0.536,
-      "mean_token_accuracy": 0.8565318584442139,
-      "step": 213
-    },
-    {
-      "epoch": 0.4331983805668016,
-      "grad_norm": 0.2687305510044098,
-      "learning_rate": 1.3989653513154165e-05,
-      "loss": 0.5182,
-      "mean_token_accuracy": 0.8477508425712585,
-      "step": 214
-    },
-    {
-      "epoch": 0.4352226720647773,
-      "grad_norm": 0.20654776692390442,
-      "learning_rate": 1.3924672795421638e-05,
-      "loss": 0.4205,
-      "mean_token_accuracy": 0.883825957775116,
-      "step": 215
-    },
-    {
-      "epoch": 0.43724696356275305,
-      "grad_norm": 0.2604771554470062,
-      "learning_rate": 1.3859495589953289e-05,
-      "loss": 0.5191,
-      "mean_token_accuracy": 0.8543003797531128,
-      "step": 216
-    },
-    {
-      "epoch": 0.4392712550607287,
-      "grad_norm": 0.23390381038188934,
-      "learning_rate": 1.3794125159829173e-05,
-      "loss": 0.5133,
-      "mean_token_accuracy": 0.8571428656578064,
-      "step": 217
-    },
-    {
-      "epoch": 0.44129554655870445,
-      "grad_norm": 0.23897536098957062,
-      "learning_rate": 1.3728564777803089e-05,
-      "loss": 0.4926,
-      "mean_token_accuracy": 0.8516854047775269,
-      "step": 218
-    },
-    {
-      "epoch": 0.4433198380566802,
-      "grad_norm": 0.21264900267124176,
-      "learning_rate": 1.3662817726138729e-05,
-      "loss": 0.5032,
-      "mean_token_accuracy": 0.8576989769935608,
-      "step": 219
-    },
-    {
-      "epoch": 0.44534412955465585,
-      "grad_norm": 0.24573111534118652,
-      "learning_rate": 1.359688729644536e-05,
-      "loss": 0.4246,
-      "mean_token_accuracy": 0.8809523582458496,
-      "step": 220
-    },
-    {
-      "epoch": 0.4473684210526316,
-      "grad_norm": 0.23440654575824738,
-      "learning_rate": 1.3530776789513009e-05,
-      "loss": 0.4817,
-      "mean_token_accuracy": 0.868558406829834,
-      "step": 221
-    },
-    {
-      "epoch": 0.4493927125506073,
-      "grad_norm": 0.2577238976955414,
-      "learning_rate": 1.3464489515147239e-05,
-      "loss": 0.4677,
-      "mean_token_accuracy": 0.8709677457809448,
-      "step": 222
-    },
-    {
-      "epoch": 0.451417004048583,
-      "grad_norm": 0.244438037276268,
-      "learning_rate": 1.3398028792003413e-05,
-      "loss": 0.4473,
-      "mean_token_accuracy": 0.8778778910636902,
-      "step": 223
-    },
-    {
-      "epoch": 0.4534412955465587,
-      "grad_norm": 0.3506868779659271,
-      "learning_rate": 1.3331397947420578e-05,
-      "loss": 0.5558,
-      "mean_token_accuracy": 0.8539267182350159,
-      "step": 224
-    },
-    {
-      "epoch": 0.45546558704453444,
-      "grad_norm": 0.3051239252090454,
-      "learning_rate": 1.3264600317254854e-05,
-      "loss": 0.5066,
-      "mean_token_accuracy": 0.8567582368850708,
-      "step": 225
-    },
-    {
-      "epoch": 0.4574898785425101,
-      "grad_norm": 0.3017394542694092,
-      "learning_rate": 1.3197639245712454e-05,
-      "loss": 0.596,
-      "mean_token_accuracy": 0.8487140536308289,
-      "step": 226
-    },
-    {
-      "epoch": 0.45951417004048584,
-      "grad_norm": 0.3093607723712921,
-      "learning_rate": 1.3130518085182224e-05,
-      "loss": 0.6099,
-      "mean_token_accuracy": 0.8373684287071228,
-      "step": 227
-    },
-    {
-      "epoch": 0.46153846153846156,
-      "grad_norm": 0.19963714480400085,
-      "learning_rate": 1.3063240196067837e-05,
-      "loss": 0.3668,
-      "mean_token_accuracy": 0.8941338062286377,
-      "step": 228
-    },
-    {
-      "epoch": 0.46356275303643724,
-      "grad_norm": 0.2264312505722046,
-      "learning_rate": 1.2995808946619533e-05,
-      "loss": 0.4301,
-      "mean_token_accuracy": 0.8826290965080261,
-      "step": 229
-    },
-    {
-      "epoch": 0.46558704453441296,
-      "grad_norm": 0.1361990123987198,
-      "learning_rate": 1.2928227712765504e-05,
-      "loss": 0.341,
-      "mean_token_accuracy": 0.896472156047821,
-      "step": 230
-    },
-    {
-      "epoch": 0.4676113360323887,
-      "grad_norm": 0.2794758975505829,
-      "learning_rate": 1.2860499877942876e-05,
-      "loss": 0.4997,
-      "mean_token_accuracy": 0.8587601184844971,
-      "step": 231
-    },
-    {
-      "epoch": 0.46963562753036436,
-      "grad_norm": 0.2864310145378113,
-      "learning_rate": 1.2792628832928302e-05,
-      "loss": 0.4895,
-      "mean_token_accuracy": 0.8697621822357178,
-      "step": 232
-    },
-    {
-      "epoch": 0.4716599190283401,
-      "grad_norm": 0.2873767614364624,
-      "learning_rate": 1.2724617975668229e-05,
-      "loss": 0.5106,
-      "mean_token_accuracy": 0.8528000116348267,
-      "step": 233
-    },
-    {
-      "epoch": 0.47368421052631576,
-      "grad_norm": 0.3079386055469513,
-      "learning_rate": 1.2656470711108763e-05,
-      "loss": 0.6188,
-      "mean_token_accuracy": 0.8250824809074402,
-      "step": 234
-    },
-    {
-      "epoch": 0.4757085020242915,
-      "grad_norm": 0.27556318044662476,
-      "learning_rate": 1.2588190451025209e-05,
-      "loss": 0.4288,
-      "mean_token_accuracy": 0.8745723962783813,
-      "step": 235
-    },
-    {
-      "epoch": 0.4777327935222672,
-      "grad_norm": 0.20111826062202454,
-      "learning_rate": 1.2519780613851254e-05,
-      "loss": 0.4395,
-      "mean_token_accuracy": 0.8747698068618774,
-      "step": 236
-    },
-    {
-      "epoch": 0.4797570850202429,
-      "grad_norm": 0.3482305407524109,
-      "learning_rate": 1.2451244624507831e-05,
-      "loss": 0.5573,
-      "mean_token_accuracy": 0.8556756973266602,
-      "step": 237
-    },
-    {
-      "epoch": 0.4817813765182186,
-      "grad_norm": 0.3096829056739807,
-      "learning_rate": 1.238258591423165e-05,
-      "loss": 0.5096,
-      "mean_token_accuracy": 0.8562538623809814,
-      "step": 238
-    },
-    {
-      "epoch": 0.48380566801619435,
-      "grad_norm": 0.26353561878204346,
-      "learning_rate": 1.2313807920403419e-05,
-      "loss": 0.421,
-      "mean_token_accuracy": 0.8860892653465271,
-      "step": 239
-    },
-    {
-      "epoch": 0.48582995951417,
-      "grad_norm": 0.26074233651161194,
-      "learning_rate": 1.2244914086375726e-05,
-      "loss": 0.4924,
-      "mean_token_accuracy": 0.8551951050758362,
-      "step": 240
-    },
-    {
-      "epoch": 0.48785425101214575,
-      "grad_norm": 0.2902613580226898,
-      "learning_rate": 1.2175907861300698e-05,
-      "loss": 0.5623,
-      "mean_token_accuracy": 0.8518710732460022,
-      "step": 241
-    },
-    {
-      "epoch": 0.4898785425101215,
-      "grad_norm": 0.31945541501045227,
-      "learning_rate": 1.2106792699957264e-05,
-      "loss": 0.5792,
-      "mean_token_accuracy": 0.8478597402572632,
-      "step": 242
-    },
-    {
-      "epoch": 0.49190283400809715,
-      "grad_norm": 0.24584610760211945,
-      "learning_rate": 1.2037572062578238e-05,
-      "loss": 0.4751,
-      "mean_token_accuracy": 0.8572905659675598,
-      "step": 243
-    },
-    {
-      "epoch": 0.4939271255060729,
-      "grad_norm": 0.28070396184921265,
-      "learning_rate": 1.1968249414677055e-05,
-      "loss": 0.5414,
-      "mean_token_accuracy": 0.8388301134109497,
-      "step": 244
-    },
-    {
-      "epoch": 0.4959514170040486,
-      "grad_norm": 0.3328787386417389,
-      "learning_rate": 1.1898828226874284e-05,
-      "loss": 0.5668,
-      "mean_token_accuracy": 0.8313993215560913,
-      "step": 245
-    },
-    {
-      "epoch": 0.4979757085020243,
-      "grad_norm": 0.2542007863521576,
-      "learning_rate": 1.1829311974723868e-05,
-      "loss": 0.5974,
-      "mean_token_accuracy": 0.8257485032081604,
-      "step": 246
-    },
-    {
-      "epoch": 0.5,
-      "grad_norm": 0.22894838452339172,
-      "learning_rate": 1.1759704138539121e-05,
-      "loss": 0.381,
-      "mean_token_accuracy": 0.8852607607841492,
-      "step": 247
-    },
-    {
-      "epoch": 0.5020242914979757,
-      "grad_norm": 0.37377893924713135,
-      "learning_rate": 1.1690008203218493e-05,
-      "loss": 0.519,
-      "mean_token_accuracy": 0.8542805314064026,
-      "step": 248
-    },
-    {
-      "epoch": 0.5040485829959515,
-      "grad_norm": 0.2569988965988159,
-      "learning_rate": 1.1620227658071088e-05,
-      "loss": 0.4751,
-      "mean_token_accuracy": 0.8624535202980042,
-      "step": 249
-    },
-    {
-      "epoch": 0.5060728744939271,
-      "grad_norm": 0.27775225043296814,
-      "learning_rate": 1.155036599664198e-05,
-      "loss": 0.4182,
-      "mean_token_accuracy": 0.8656914830207825,
-      "step": 250
-    },
-    {
-      "epoch": 0.5080971659919028,
-      "grad_norm": 0.25449249148368835,
-      "learning_rate": 1.1480426716537316e-05,
-      "loss": 0.524,
-      "mean_token_accuracy": 0.8571428656578064,
-      "step": 251
-    },
-    {
-      "epoch": 0.5101214574898786,
-      "grad_norm": 0.28567492961883545,
-      "learning_rate": 1.1410413319249193e-05,
-      "loss": 0.5068,
-      "mean_token_accuracy": 0.8494461178779602,
-      "step": 252
-    },
-    {
-      "epoch": 0.5121457489878543,
-      "grad_norm": 0.2155705690383911,
-      "learning_rate": 1.1340329309980379e-05,
-      "loss": 0.3154,
-      "mean_token_accuracy": 0.901196300983429,
-      "step": 253
-    },
-    {
-      "epoch": 0.5141700404858299,
-      "grad_norm": 0.2748875617980957,
-      "learning_rate": 1.1270178197468788e-05,
-      "loss": 0.4494,
-      "mean_token_accuracy": 0.8714776635169983,
-      "step": 254
-    },
-    {
-      "epoch": 0.5161943319838057,
-      "grad_norm": 0.3891545236110687,
-      "learning_rate": 1.119996349381187e-05,
-      "loss": 0.6305,
-      "mean_token_accuracy": 0.8360042572021484,
-      "step": 255
-    },
-    {
-      "epoch": 0.5182186234817814,
-      "grad_norm": 0.24013206362724304,
-      "learning_rate": 1.112968871429073e-05,
-      "loss": 0.4703,
-      "mean_token_accuracy": 0.8685857057571411,
-      "step": 256
-    },
-    {
-      "epoch": 0.520242914979757,
-      "grad_norm": 0.3602214753627777,
-      "learning_rate": 1.1059357377194161e-05,
-      "loss": 0.548,
-      "mean_token_accuracy": 0.847953200340271,
-      "step": 257
-    },
-    {
-      "epoch": 0.5222672064777328,
-      "grad_norm": 0.26108884811401367,
-      "learning_rate": 1.09889730036425e-05,
-      "loss": 0.4013,
-      "mean_token_accuracy": 0.8851963877677917,
-      "step": 258
-    },
-    {
-      "epoch": 0.5242914979757085,
-      "grad_norm": 0.28321635723114014,
-      "learning_rate": 1.0918539117411334e-05,
-      "loss": 0.4847,
-      "mean_token_accuracy": 0.8601503968238831,
-      "step": 259
-    },
-    {
-      "epoch": 0.5263157894736842,
-      "grad_norm": 0.30514952540397644,
-      "learning_rate": 1.0848059244755093e-05,
-      "loss": 0.5191,
-      "mean_token_accuracy": 0.8522663712501526,
-      "step": 260
-    },
-    {
-      "epoch": 0.52834008097166,
-      "grad_norm": 0.3193584978580475,
-      "learning_rate": 1.0777536914230509e-05,
-      "loss": 0.5076,
-      "mean_token_accuracy": 0.8504902124404907,
-      "step": 261
-    },
-    {
-      "epoch": 0.5303643724696356,
-      "grad_norm": 0.34578680992126465,
-      "learning_rate": 1.0706975656519946e-05,
-      "loss": 0.6462,
-      "mean_token_accuracy": 0.8127555251121521,
-      "step": 262
-    },
-    {
-      "epoch": 0.5323886639676113,
-      "grad_norm": 0.22260263562202454,
-      "learning_rate": 1.0636379004254665e-05,
-      "loss": 0.382,
-      "mean_token_accuracy": 0.8847006559371948,
-      "step": 263
-    },
-    {
-      "epoch": 0.5344129554655871,
-      "grad_norm": 0.3325004577636719,
-      "learning_rate": 1.0565750491837925e-05,
-      "loss": 0.5118,
-      "mean_token_accuracy": 0.8544698357582092,
-      "step": 264
-    },
-    {
-      "epoch": 0.5364372469635628,
-      "grad_norm": 0.21064940094947815,
-      "learning_rate": 1.049509365526807e-05,
-      "loss": 0.4833,
-      "mean_token_accuracy": 0.8545101881027222,
-      "step": 265
-    },
-    {
-      "epoch": 0.5384615384615384,
-      "grad_norm": 0.2569122612476349,
-      "learning_rate": 1.0424412031961485e-05,
-      "loss": 0.4462,
-      "mean_token_accuracy": 0.8711549639701843,
-      "step": 266
-    },
-    {
-      "epoch": 0.5404858299595142,
-      "grad_norm": 0.2512376010417938,
-      "learning_rate": 1.0353709160575488e-05,
-      "loss": 0.4553,
-      "mean_token_accuracy": 0.8676716685295105,
-      "step": 267
-    },
-    {
-      "epoch": 0.5425101214574899,
-      "grad_norm": 0.20290617644786835,
-      "learning_rate": 1.0282988580831183e-05,
-      "loss": 0.3591,
-      "mean_token_accuracy": 0.8925163149833679,
-      "step": 268
-    },
-    {
-      "epoch": 0.5445344129554656,
-      "grad_norm": 0.276850163936615,
-      "learning_rate": 1.0212253833336237e-05,
-      "loss": 0.5178,
-      "mean_token_accuracy": 0.8581696152687073,
-      "step": 269
-    },
-    {
-      "epoch": 0.5465587044534413,
-      "grad_norm": 0.27752986550331116,
-      "learning_rate": 1.0141508459407622e-05,
-      "loss": 0.499,
-      "mean_token_accuracy": 0.852477490901947,
-      "step": 270
-    },
-    {
-      "epoch": 0.548582995951417,
-      "grad_norm": 0.27458417415618896,
-      "learning_rate": 1.0070756000894321e-05,
-      "loss": 0.528,
-      "mean_token_accuracy": 0.8500254154205322,
-      "step": 271
-    },
-    {
-      "epoch": 0.5506072874493927,
-      "grad_norm": 0.25697359442710876,
-      "learning_rate": 1e-05,
-      "loss": 0.4372,
-      "mean_token_accuracy": 0.8874056935310364,
-      "step": 272
-    },
-    {
-      "epoch": 0.5526315789473685,
-      "grad_norm": 0.2661818563938141,
-      "learning_rate": 9.929243999105682e-06,
-      "loss": 0.4539,
-      "mean_token_accuracy": 0.8609970808029175,
-      "step": 273
-    },
-    {
-      "epoch": 0.5546558704453441,
-      "grad_norm": 0.20793814957141876,
-      "learning_rate": 9.858491540592383e-06,
-      "loss": 0.4136,
-      "mean_token_accuracy": 0.8824717402458191,
-      "step": 274
-    },
-    {
-      "epoch": 0.5566801619433198,
-      "grad_norm": 0.20243680477142334,
-      "learning_rate": 9.787746166663765e-06,
-      "loss": 0.4097,
-      "mean_token_accuracy": 0.8892011046409607,
-      "step": 275
-    },
-    {
-      "epoch": 0.5587044534412956,
-      "grad_norm": 0.25867629051208496,
-      "learning_rate": 9.71701141916882e-06,
-      "loss": 0.4539,
-      "mean_token_accuracy": 0.8806194067001343,
-      "step": 276
-    },
-    {
-      "epoch": 0.5607287449392713,
-      "grad_norm": 0.2549677789211273,
-      "learning_rate": 9.646290839424515e-06,
-      "loss": 0.3627,
-      "mean_token_accuracy": 0.896221399307251,
-      "step": 277
-    },
-    {
-      "epoch": 0.562753036437247,
-      "grad_norm": 0.27826279401779175,
-      "learning_rate": 9.57558796803852e-06,
-      "loss": 0.5258,
-      "mean_token_accuracy": 0.8454780578613281,
-      "step": 278
-    },
-    {
-      "epoch": 0.5647773279352226,
-      "grad_norm": 0.2794351875782013,
-      "learning_rate": 9.504906344731933e-06,
-      "loss": 0.5382,
-      "mean_token_accuracy": 0.8399999737739563,
-      "step": 279
-    },
-    {
-      "epoch": 0.5668016194331984,
-      "grad_norm": 0.2891370952129364,
-      "learning_rate": 9.434249508162076e-06,
-      "loss": 0.5144,
-      "mean_token_accuracy": 0.8665386438369751,
-      "step": 280
-    },
-    {
-      "epoch": 0.5688259109311741,
-      "grad_norm": 0.2668806314468384,
-      "learning_rate": 9.363620995745337e-06,
-      "loss": 0.476,
-      "mean_token_accuracy": 0.8648930788040161,
-      "step": 281
-    },
-    {
-      "epoch": 0.5708502024291497,
-      "grad_norm": 0.19786901772022247,
-      "learning_rate": 9.293024343480056e-06,
-      "loss": 0.4575,
-      "mean_token_accuracy": 0.8725023865699768,
-      "step": 282
-    },
-    {
-      "epoch": 0.5728744939271255,
-      "grad_norm": 0.2296639382839203,
-      "learning_rate": 9.222463085769495e-06,
-      "loss": 0.4558,
-      "mean_token_accuracy": 0.867986798286438,
-      "step": 283
-    },
-    {
-      "epoch": 0.5748987854251012,
-      "grad_norm": 0.2851167619228363,
-      "learning_rate": 9.151940755244912e-06,
-      "loss": 0.4893,
-      "mean_token_accuracy": 0.8656179904937744,
-      "step": 284
-    },
-    {
-      "epoch": 0.5769230769230769,
-      "grad_norm": 0.28831976652145386,
-      "learning_rate": 9.081460882588668e-06,
-      "loss": 0.4983,
-      "mean_token_accuracy": 0.8594695329666138,
-      "step": 285
-    },
-    {
-      "epoch": 0.5789473684210527,
-      "grad_norm": 0.24200013279914856,
-      "learning_rate": 9.011026996357504e-06,
-      "loss": 0.4087,
-      "mean_token_accuracy": 0.8737270832061768,
-      "step": 286
-    },
-    {
-      "epoch": 0.5809716599190283,
-      "grad_norm": 0.2352951318025589,
-      "learning_rate": 8.94064262280584e-06,
-      "loss": 0.3563,
-      "mean_token_accuracy": 0.8963627219200134,
-      "step": 287
-    },
-    {
-      "epoch": 0.582995951417004,
-      "grad_norm": 0.3065440058708191,
-      "learning_rate": 8.870311285709274e-06,
-      "loss": 0.5417,
-      "mean_token_accuracy": 0.8533872365951538,
-      "step": 288
-    },
-    {
-      "epoch": 0.5850202429149798,
-      "grad_norm": 0.3517458140850067,
-      "learning_rate": 8.80003650618813e-06,
-      "loss": 0.4485,
-      "mean_token_accuracy": 0.874932587146759,
-      "step": 289
-    },
-    {
-      "epoch": 0.5870445344129555,
-      "grad_norm": 0.2901054620742798,
-      "learning_rate": 8.729821802531213e-06,
-      "loss": 0.5214,
-      "mean_token_accuracy": 0.8495886921882629,
-      "step": 290
-    },
-    {
-      "epoch": 0.5890688259109311,
-      "grad_norm": 0.24167115986347198,
-      "learning_rate": 8.659670690019626e-06,
-      "loss": 0.4589,
-      "mean_token_accuracy": 0.8680351972579956,
-      "step": 291
-    },
-    {
-      "epoch": 0.5910931174089069,
-      "grad_norm": 0.2503730058670044,
-      "learning_rate": 8.58958668075081e-06,
-      "loss": 0.4722,
-      "mean_token_accuracy": 0.8686291575431824,
-      "step": 292
-    },
-    {
-      "epoch": 0.5931174089068826,
-      "grad_norm": 0.2157752513885498,
-      "learning_rate": 8.519573283462688e-06,
-      "loss": 0.4448,
-      "mean_token_accuracy": 0.8718662858009338,
-      "step": 293
-    },
-    {
-      "epoch": 0.5951417004048583,
-      "grad_norm": 0.26292556524276733,
-      "learning_rate": 8.449634003358022e-06,
-      "loss": 0.4789,
-      "mean_token_accuracy": 0.8660488724708557,
-      "step": 294
-    },
-    {
-      "epoch": 0.597165991902834,
-      "grad_norm": 0.2639862895011902,
-      "learning_rate": 8.379772341928916e-06,
-      "loss": 0.4656,
-      "mean_token_accuracy": 0.871257483959198,
-      "step": 295
-    },
-    {
-      "epoch": 0.5991902834008097,
-      "grad_norm": 0.19293735921382904,
-      "learning_rate": 8.309991796781512e-06,
-      "loss": 0.394,
-      "mean_token_accuracy": 0.8828034996986389,
-      "step": 296
-    },
-    {
-      "epoch": 0.6012145748987854,
-      "grad_norm": 0.28275519609451294,
-      "learning_rate": 8.24029586146088e-06,
-      "loss": 0.5222,
-      "mean_token_accuracy": 0.8616324663162231,
-      "step": 297
-    },
-    {
-      "epoch": 0.6032388663967612,
-      "grad_norm": 0.31240344047546387,
-      "learning_rate": 8.170688025276134e-06,
-      "loss": 0.5318,
-      "mean_token_accuracy": 0.8440040946006775,
-      "step": 298
-    },
-    {
-      "epoch": 0.6052631578947368,
-      "grad_norm": 0.2823457419872284,
-      "learning_rate": 8.101171773125716e-06,
-      "loss": 0.497,
-      "mean_token_accuracy": 0.862606942653656,
-      "step": 299
-    },
-    {
-      "epoch": 0.6072874493927125,
-      "grad_norm": 0.2254837453365326,
-      "learning_rate": 8.031750585322948e-06,
-      "loss": 0.4309,
-      "mean_token_accuracy": 0.8666666746139526,
-      "step": 300
-    },
-    {
-      "epoch": 0.6093117408906883,
-      "grad_norm": 0.2284829467535019,
-      "learning_rate": 7.962427937421763e-06,
-      "loss": 0.3243,
-      "mean_token_accuracy": 0.9150023460388184,
-      "step": 301
-    },
-    {
-      "epoch": 0.611336032388664,
-      "grad_norm": 0.1687772572040558,
-      "learning_rate": 7.89320730004274e-06,
-      "loss": 0.3102,
-      "mean_token_accuracy": 0.9114139676094055,
-      "step": 302
-    },
-    {
-      "epoch": 0.6133603238866396,
-      "grad_norm": 0.2668203115463257,
-      "learning_rate": 7.824092138699307e-06,
-      "loss": 0.4753,
-      "mean_token_accuracy": 0.8627451062202454,
-      "step": 303
-    },
-    {
-      "epoch": 0.6153846153846154,
-      "grad_norm": 0.2429952621459961,
-      "learning_rate": 7.755085913624274e-06,
-      "loss": 0.5117,
-      "mean_token_accuracy": 0.8498145937919617,
-      "step": 304
-    },
-    {
-      "epoch": 0.6174089068825911,
-      "grad_norm": 0.2407144159078598,
-      "learning_rate": 7.686192079596586e-06,
-      "loss": 0.4016,
-      "mean_token_accuracy": 0.8928571343421936,
-      "step": 305
-    },
-    {
-      "epoch": 0.6194331983805668,
-      "grad_norm": 0.27329036593437195,
-      "learning_rate": 7.617414085768352e-06,
-      "loss": 0.4611,
-      "mean_token_accuracy": 0.8771044015884399,
-      "step": 306
-    },
-    {
-      "epoch": 0.6214574898785425,
-      "grad_norm": 0.2208373248577118,
-      "learning_rate": 7.548755375492173e-06,
-      "loss": 0.418,
-      "mean_token_accuracy": 0.8668587803840637,
-      "step": 307
-    },
-    {
-      "epoch": 0.6234817813765182,
-      "grad_norm": 0.2131100744009018,
-      "learning_rate": 7.480219386148751e-06,
-      "loss": 0.409,
-      "mean_token_accuracy": 0.8864306807518005,
-      "step": 308
-    },
-    {
-      "epoch": 0.6255060728744939,
-      "grad_norm": 0.27837830781936646,
-      "learning_rate": 7.411809548974792e-06,
-      "loss": 0.4889,
-      "mean_token_accuracy": 0.8648374080657959,
-      "step": 309
-    },
-    {
-      "epoch": 0.6275303643724697,
-      "grad_norm": 0.2820056974887848,
-      "learning_rate": 7.343529288891239e-06,
-      "loss": 0.5201,
-      "mean_token_accuracy": 0.8595438003540039,
-      "step": 310
-    },
-    {
-      "epoch": 0.6295546558704453,
-      "grad_norm": 0.2284225970506668,
-      "learning_rate": 7.275382024331773e-06,
-      "loss": 0.3983,
-      "mean_token_accuracy": 0.8888888955116272,
-      "step": 311
-    },
-    {
-      "epoch": 0.631578947368421,
-      "grad_norm": 0.2649460434913635,
-      "learning_rate": 7.2073711670717e-06,
-      "loss": 0.5232,
-      "mean_token_accuracy": 0.8596938848495483,
-      "step": 312
-    },
-    {
-      "epoch": 0.6336032388663968,
-      "grad_norm": 0.4928232431411743,
-      "learning_rate": 7.13950012205713e-06,
-      "loss": 0.4908,
-      "mean_token_accuracy": 0.8723404407501221,
-      "step": 313
-    },
-    {
-      "epoch": 0.6356275303643725,
-      "grad_norm": 0.26822856068611145,
-      "learning_rate": 7.071772287234497e-06,
-      "loss": 0.4883,
-      "mean_token_accuracy": 0.8623949289321899,
-      "step": 314
-    },
-    {
-      "epoch": 0.6376518218623481,
-      "grad_norm": 0.21105557680130005,
-      "learning_rate": 7.004191053380469e-06,
-      "loss": 0.4825,
-      "mean_token_accuracy": 0.8558024764060974,
-      "step": 315
-    },
-    {
-      "epoch": 0.6396761133603239,
-      "grad_norm": 0.29273661971092224,
-      "learning_rate": 6.936759803932167e-06,
-      "loss": 0.5179,
-      "mean_token_accuracy": 0.8536452651023865,
-      "step": 316
-    },
-    {
-      "epoch": 0.6417004048582996,
-      "grad_norm": 0.21913200616836548,
-      "learning_rate": 6.869481914817779e-06,
-      "loss": 0.4951,
-      "mean_token_accuracy": 0.8600091338157654,
-      "step": 317
-    },
-    {
-      "epoch": 0.6437246963562753,
-      "grad_norm": 0.26065701246261597,
-      "learning_rate": 6.802360754287548e-06,
-      "loss": 0.4621,
-      "mean_token_accuracy": 0.8590487241744995,
-      "step": 318
-    },
-    {
-      "epoch": 0.645748987854251,
-      "grad_norm": 0.2516014873981476,
-      "learning_rate": 6.735399682745145e-06,
-      "loss": 0.4134,
-      "mean_token_accuracy": 0.8689809441566467,
-      "step": 319
-    },
-    {
-      "epoch": 0.6477732793522267,
-      "grad_norm": 0.3060797154903412,
-      "learning_rate": 6.668602052579425e-06,
-      "loss": 0.5212,
-      "mean_token_accuracy": 0.8627451062202454,
-      "step": 320
-    },
-    {
-      "epoch": 0.6497975708502024,
-      "grad_norm": 0.307005375623703,
-      "learning_rate": 6.601971207996592e-06,
-      "loss": 0.5049,
-      "mean_token_accuracy": 0.8486055731773376,
-      "step": 321
-    },
-    {
-      "epoch": 0.6518218623481782,
-      "grad_norm": 0.2623290717601776,
-      "learning_rate": 6.535510484852767e-06,
-      "loss": 0.4424,
-      "mean_token_accuracy": 0.8855451345443726,
-      "step": 322
-    },
-    {
-      "epoch": 0.6538461538461539,
-      "grad_norm": 0.31143125891685486,
-      "learning_rate": 6.469223210486992e-06,
-      "loss": 0.459,
-      "mean_token_accuracy": 0.8782935738563538,
-      "step": 323
-    },
-    {
-      "epoch": 0.6558704453441295,
-      "grad_norm": 0.2084709107875824,
-      "learning_rate": 6.403112703554643e-06,
-      "loss": 0.4153,
-      "mean_token_accuracy": 0.8768005967140198,
-      "step": 324
-    },
-    {
-      "epoch": 0.6578947368421053,
-      "grad_norm": 0.26807963848114014,
-      "learning_rate": 6.337182273861273e-06,
-      "loss": 0.483,
-      "mean_token_accuracy": 0.8573106527328491,
-      "step": 325
-    },
-    {
-      "epoch": 0.659919028340081,
-      "grad_norm": 0.2525019645690918,
-      "learning_rate": 6.2714352221969155e-06,
-      "loss": 0.381,
-      "mean_token_accuracy": 0.8902173638343811,
-      "step": 326
-    },
-    {
-      "epoch": 0.6619433198380567,
-      "grad_norm": 0.2668350040912628,
-      "learning_rate": 6.205874840170833e-06,
-      "loss": 0.4758,
-      "mean_token_accuracy": 0.8671524524688721,
-      "step": 327
-    },
-    {
-      "epoch": 0.6639676113360324,
-      "grad_norm": 0.30968189239501953,
-      "learning_rate": 6.140504410046712e-06,
-      "loss": 0.4616,
-      "mean_token_accuracy": 0.8822922110557556,
-      "step": 328
-    },
-    {
-      "epoch": 0.6659919028340081,
-      "grad_norm": 0.22733403742313385,
-      "learning_rate": 6.075327204578363e-06,
-      "loss": 0.4326,
-      "mean_token_accuracy": 0.8846761584281921,
-      "step": 329
-    },
-    {
-      "epoch": 0.6680161943319838,
-      "grad_norm": 0.3149683475494385,
-      "learning_rate": 6.010346486845837e-06,
-      "loss": 0.5278,
-      "mean_token_accuracy": 0.8606253266334534,
-      "step": 330
-    },
-    {
-      "epoch": 0.6700404858299596,
-      "grad_norm": 0.24868245422840118,
-      "learning_rate": 5.945565510092086e-06,
-      "loss": 0.3365,
-      "mean_token_accuracy": 0.8998779058456421,
-      "step": 331
-    },
-    {
-      "epoch": 0.6720647773279352,
-      "grad_norm": 0.36468973755836487,
-      "learning_rate": 5.880987517560075e-06,
-      "loss": 0.4912,
-      "mean_token_accuracy": 0.8575851321220398,
-      "step": 332
-    },
-    {
-      "epoch": 0.6740890688259109,
-      "grad_norm": 0.29678285121917725,
-      "learning_rate": 5.81661574233042e-06,
-      "loss": 0.4878,
-      "mean_token_accuracy": 0.8686288595199585,
-      "step": 333
-    },
-    {
-      "epoch": 0.6761133603238867,
-      "grad_norm": 0.23850063979625702,
-      "learning_rate": 5.752453407159521e-06,
-      "loss": 0.471,
-      "mean_token_accuracy": 0.8644067645072937,
-      "step": 334
-    },
-    {
-      "epoch": 0.6781376518218624,
-      "grad_norm": 0.33608052134513855,
-      "learning_rate": 5.688503724318217e-06,
-      "loss": 0.5153,
-      "mean_token_accuracy": 0.863723635673523,
-      "step": 335
-    },
-    {
-      "epoch": 0.680161943319838,
-      "grad_norm": 0.322111576795578,
-      "learning_rate": 5.6247698954309616e-06,
-      "loss": 0.5171,
-      "mean_token_accuracy": 0.8543990254402161,
-      "step": 336
-    },
-    {
-      "epoch": 0.6821862348178138,
-      "grad_norm": 0.2473558485507965,
-      "learning_rate": 5.561255111315525e-06,
-      "loss": 0.3823,
-      "mean_token_accuracy": 0.8928987383842468,
-      "step": 337
-    },
-    {
-      "epoch": 0.6842105263157895,
-      "grad_norm": 0.2070508897304535,
-      "learning_rate": 5.497962551823266e-06,
-      "loss": 0.4532,
-      "mean_token_accuracy": 0.8769230842590332,
-      "step": 338
-    },
-    {
-      "epoch": 0.6862348178137652,
-      "grad_norm": 0.27380725741386414,
-      "learning_rate": 5.434895385679937e-06,
-      "loss": 0.473,
-      "mean_token_accuracy": 0.8681750893592834,
-      "step": 339
-    },
-    {
-      "epoch": 0.6882591093117408,
-      "grad_norm": 0.26557326316833496,
-      "learning_rate": 5.3720567703270135e-06,
-      "loss": 0.4951,
-      "mean_token_accuracy": 0.8644970655441284,
-      "step": 340
-    },
-    {
-      "epoch": 0.6902834008097166,
-      "grad_norm": 0.27739718556404114,
-      "learning_rate": 5.3094498517636324e-06,
-      "loss": 0.4504,
-      "mean_token_accuracy": 0.8617021441459656,
-      "step": 341
-    },
-    {
-      "epoch": 0.6923076923076923,
-      "grad_norm": 0.25480371713638306,
-      "learning_rate": 5.247077764389099e-06,
-      "loss": 0.4402,
-      "mean_token_accuracy": 0.8725548386573792,
-      "step": 342
-    },
-    {
-      "epoch": 0.694331983805668,
-      "grad_norm": 0.23700827360153198,
-      "learning_rate": 5.18494363084596e-06,
-      "loss": 0.4703,
-      "mean_token_accuracy": 0.8650107979774475,
-      "step": 343
-    },
-    {
-      "epoch": 0.6963562753036437,
-      "grad_norm": 0.2832289934158325,
-      "learning_rate": 5.1230505618636575e-06,
-      "loss": 0.5179,
-      "mean_token_accuracy": 0.8526434302330017,
-      "step": 344
-    },
-    {
-      "epoch": 0.6983805668016194,
-      "grad_norm": 0.28387221693992615,
-      "learning_rate": 5.061401656102791e-06,
-      "loss": 0.4843,
-      "mean_token_accuracy": 0.8752436637878418,
-      "step": 345
-    },
-    {
-      "epoch": 0.7004048582995951,
-      "grad_norm": 0.2733575701713562,
-      "learning_rate": 5.000000000000003e-06,
-      "loss": 0.4283,
-      "mean_token_accuracy": 0.8881770372390747,
-      "step": 346
-    },
-    {
-      "epoch": 0.7024291497975709,
-      "grad_norm": 0.17921440303325653,
-      "learning_rate": 4.938848667613436e-06,
-      "loss": 0.374,
-      "mean_token_accuracy": 0.8849824070930481,
-      "step": 347
-    },
-    {
-      "epoch": 0.7044534412955465,
-      "grad_norm": 0.29820355772972107,
-      "learning_rate": 4.8779507204688595e-06,
-      "loss": 0.6646,
-      "mean_token_accuracy": 0.8174545168876648,
-      "step": 348
-    },
-    {
-      "epoch": 0.7064777327935222,
-      "grad_norm": 0.23080453276634216,
-      "learning_rate": 4.817309207406347e-06,
-      "loss": 0.3988,
-      "mean_token_accuracy": 0.8915223479270935,
-      "step": 349
-    },
-    {
-      "epoch": 0.708502024291498,
-      "grad_norm": 0.23018287122249603,
-      "learning_rate": 4.756927164427685e-06,
-      "loss": 0.3978,
-      "mean_token_accuracy": 0.890070915222168,
-      "step": 350
-    },
-    {
-      "epoch": 0.7105263157894737,
-      "grad_norm": 0.1862955093383789,
-      "learning_rate": 4.696807614544352e-06,
-      "loss": 0.3969,
-      "mean_token_accuracy": 0.8772727251052856,
-      "step": 351
-    },
-    {
-      "epoch": 0.7125506072874493,
-      "grad_norm": 0.3891066610813141,
-      "learning_rate": 4.636953567626176e-06,
-      "loss": 0.4872,
-      "mean_token_accuracy": 0.8715929985046387,
-      "step": 352
-    },
-    {
-      "epoch": 0.7145748987854251,
-      "grad_norm": 0.32266443967819214,
-      "learning_rate": 4.57736802025065e-06,
-      "loss": 0.4629,
-      "mean_token_accuracy": 0.8703703880310059,
-      "step": 353
-    },
-    {
-      "epoch": 0.7165991902834008,
-      "grad_norm": 0.19085142016410828,
-      "learning_rate": 4.518053955552903e-06,
-      "loss": 0.3958,
-      "mean_token_accuracy": 0.8828560709953308,
-      "step": 354
-    },
-    {
-      "epoch": 0.7186234817813765,
-      "grad_norm": 0.2669807970523834,
-      "learning_rate": 4.459014343076356e-06,
-      "loss": 0.4376,
-      "mean_token_accuracy": 0.8822800517082214,
-      "step": 355
-    },
-    {
-      "epoch": 0.7206477732793523,
-      "grad_norm": 0.2982715964317322,
-      "learning_rate": 4.400252138624047e-06,
-      "loss": 0.391,
-      "mean_token_accuracy": 0.8809523582458496,
-      "step": 356
-    },
-    {
-      "epoch": 0.7226720647773279,
-      "grad_norm": 0.2584725618362427,
-      "learning_rate": 4.341770284110655e-06,
-      "loss": 0.4552,
-      "mean_token_accuracy": 0.8733572363853455,
-      "step": 357
-    },
-    {
-      "epoch": 0.7246963562753036,
-      "grad_norm": 0.18682384490966797,
-      "learning_rate": 4.283571707415214e-06,
-      "loss": 0.4377,
-      "mean_token_accuracy": 0.8692899346351624,
-      "step": 358
-    },
-    {
-      "epoch": 0.7267206477732794,
-      "grad_norm": 0.2730652391910553,
-      "learning_rate": 4.2256593222345185e-06,
-      "loss": 0.3772,
-      "mean_token_accuracy": 0.8911022543907166,
-      "step": 359
-    },
-    {
-      "epoch": 0.728744939271255,
-      "grad_norm": 0.4370400905609131,
-      "learning_rate": 4.168036027937267e-06,
-      "loss": 0.4908,
-      "mean_token_accuracy": 0.8653010725975037,
-      "step": 360
-    },
-    {
-      "epoch": 0.7307692307692307,
-      "grad_norm": 0.21713419258594513,
-      "learning_rate": 4.1107047094188946e-06,
-      "loss": 0.4043,
-      "mean_token_accuracy": 0.8931367993354797,
-      "step": 361
-    },
-    {
-      "epoch": 0.7327935222672065,
-      "grad_norm": 0.36545848846435547,
-      "learning_rate": 4.053668236957135e-06,
-      "loss": 0.4879,
-      "mean_token_accuracy": 0.8663522005081177,
-      "step": 362
-    },
-    {
-      "epoch": 0.7348178137651822,
-      "grad_norm": 0.2589530050754547,
-      "learning_rate": 3.996929466068344e-06,
-      "loss": 0.4796,
-      "mean_token_accuracy": 0.8690228462219238,
-      "step": 363
-    },
-    {
-      "epoch": 0.7368421052631579,
-      "grad_norm": 0.24485215544700623,
-      "learning_rate": 3.940491237364519e-06,
-      "loss": 0.4668,
-      "mean_token_accuracy": 0.8754062652587891,
-      "step": 364
-    },
-    {
-      "epoch": 0.7388663967611336,
-      "grad_norm": 0.23230288922786713,
-      "learning_rate": 3.884356376411089e-06,
-      "loss": 0.4699,
-      "mean_token_accuracy": 0.8739306330680847,
-      "step": 365
-    },
-    {
-      "epoch": 0.7408906882591093,
-      "grad_norm": 0.31604430079460144,
-      "learning_rate": 3.828527693585451e-06,
-      "loss": 0.6245,
-      "mean_token_accuracy": 0.8259986639022827,
-      "step": 366
-    },
-    {
-      "epoch": 0.742914979757085,
-      "grad_norm": 0.268093466758728,
-      "learning_rate": 3.7730079839362755e-06,
-      "loss": 0.4716,
-      "mean_token_accuracy": 0.8759820461273193,
-      "step": 367
-    },
-    {
-      "epoch": 0.7449392712550608,
-      "grad_norm": 0.2674328684806824,
-      "learning_rate": 3.7178000270435765e-06,
-      "loss": 0.4649,
-      "mean_token_accuracy": 0.8691232800483704,
-      "step": 368
-    },
-    {
-      "epoch": 0.7469635627530364,
-      "grad_norm": 0.2278822362422943,
-      "learning_rate": 3.662906586879542e-06,
-      "loss": 0.3952,
-      "mean_token_accuracy": 0.8885647654533386,
-      "step": 369
-    },
-    {
-      "epoch": 0.7489878542510121,
-      "grad_norm": 0.2409820705652237,
-      "learning_rate": 3.6083304116701535e-06,
-      "loss": 0.4588,
-      "mean_token_accuracy": 0.8738207817077637,
-      "step": 370
-    },
-    {
-      "epoch": 0.7510121457489879,
-      "grad_norm": 0.2809779644012451,
-      "learning_rate": 3.5540742337576083e-06,
-      "loss": 0.4546,
-      "mean_token_accuracy": 0.8706353902816772,
-      "step": 371
-    },
-    {
-      "epoch": 0.7530364372469636,
-      "grad_norm": 0.2390257716178894,
-      "learning_rate": 3.5001407694635326e-06,
-      "loss": 0.4162,
-      "mean_token_accuracy": 0.889140248298645,
-      "step": 372
-    },
-    {
-      "epoch": 0.7550607287449392,
-      "grad_norm": 0.36595165729522705,
-      "learning_rate": 3.446532718952966e-06,
-      "loss": 0.5713,
-      "mean_token_accuracy": 0.8397144675254822,
-      "step": 373
-    },
-    {
-      "epoch": 0.757085020242915,
-      "grad_norm": 0.25939351320266724,
-      "learning_rate": 3.3932527660991877e-06,
-      "loss": 0.4312,
-      "mean_token_accuracy": 0.8785249590873718,
-      "step": 374
-    },
-    {
-      "epoch": 0.7591093117408907,
-      "grad_norm": 0.3776956796646118,
-      "learning_rate": 3.340303578349361e-06,
-      "loss": 0.5116,
-      "mean_token_accuracy": 0.8577494621276855,
-      "step": 375
-    },
-    {
-      "epoch": 0.7611336032388664,
-      "grad_norm": 0.24847571551799774,
-      "learning_rate": 3.2876878065909714e-06,
-      "loss": 0.4522,
-      "mean_token_accuracy": 0.8745308518409729,
-      "step": 376
-    },
-    {
-      "epoch": 0.7631578947368421,
-      "grad_norm": 0.19581596553325653,
-      "learning_rate": 3.2354080850191328e-06,
-      "loss": 0.3578,
-      "mean_token_accuracy": 0.8948983550071716,
-      "step": 377
-    },
-    {
-      "epoch": 0.7651821862348178,
-      "grad_norm": 0.2568861246109009,
-      "learning_rate": 3.1834670310046735e-06,
-      "loss": 0.4396,
-      "mean_token_accuracy": 0.8738559484481812,
-      "step": 378
-    },
-    {
-      "epoch": 0.7672064777327935,
-      "grad_norm": 0.2038806527853012,
-      "learning_rate": 3.1318672449631283e-06,
-      "loss": 0.3829,
-      "mean_token_accuracy": 0.8911939263343811,
-      "step": 379
-    },
-    {
-      "epoch": 0.7692307692307693,
-      "grad_norm": 0.3116026222705841,
-      "learning_rate": 3.0806113102245395e-06,
-      "loss": 0.4545,
-      "mean_token_accuracy": 0.8694765567779541,
-      "step": 380
-    },
-    {
-      "epoch": 0.771255060728745,
-      "grad_norm": 0.24429158866405487,
-      "learning_rate": 3.029701792904117e-06,
-      "loss": 0.4071,
-      "mean_token_accuracy": 0.8861004114151001,
-      "step": 381
-    },
-    {
-      "epoch": 0.7732793522267206,
-      "grad_norm": 0.2990184724330902,
-      "learning_rate": 2.979141241773775e-06,
-      "loss": 0.4215,
-      "mean_token_accuracy": 0.883500874042511,
-      "step": 382
-    },
-    {
-      "epoch": 0.7753036437246964,
-      "grad_norm": 0.3037009835243225,
-      "learning_rate": 2.9289321881345257e-06,
-      "loss": 0.3884,
-      "mean_token_accuracy": 0.8895916938781738,
-      "step": 383
-    },
-    {
-      "epoch": 0.7773279352226721,
-      "grad_norm": 0.27103233337402344,
-      "learning_rate": 2.879077145689746e-06,
-      "loss": 0.4003,
-      "mean_token_accuracy": 0.8842247724533081,
-      "step": 384
-    },
-    {
-      "epoch": 0.7793522267206477,
-      "grad_norm": 0.2830134928226471,
-      "learning_rate": 2.829578610419337e-06,
-      "loss": 0.524,
-      "mean_token_accuracy": 0.8542538285255432,
-      "step": 385
-    },
-    {
-      "epoch": 0.7813765182186235,
-      "grad_norm": 0.30209383368492126,
-      "learning_rate": 2.780439060454756e-06,
-      "loss": 0.5034,
-      "mean_token_accuracy": 0.8675914406776428,
-      "step": 386
-    },
-    {
-      "epoch": 0.7834008097165992,
-      "grad_norm": 0.22681166231632233,
-      "learning_rate": 2.7316609559549568e-06,
-      "loss": 0.4467,
-      "mean_token_accuracy": 0.8747357130050659,
-      "step": 387
-    },
-    {
-      "epoch": 0.7854251012145749,
-      "grad_norm": 0.22978028655052185,
-      "learning_rate": 2.683246738983217e-06,
-      "loss": 0.3554,
-      "mean_token_accuracy": 0.892988920211792,
-      "step": 388
-    },
-    {
-      "epoch": 0.7874493927125507,
-      "grad_norm": 0.278886616230011,
-      "learning_rate": 2.6351988333848787e-06,
-      "loss": 0.4305,
-      "mean_token_accuracy": 0.8729411959648132,
-      "step": 389
-    },
-    {
-      "epoch": 0.7894736842105263,
-      "grad_norm": 0.2988165318965912,
-      "learning_rate": 2.587519644666001e-06,
-      "loss": 0.4546,
-      "mean_token_accuracy": 0.8679119348526001,
-      "step": 390
-    },
-    {
-      "epoch": 0.791497975708502,
-      "grad_norm": 0.3190794885158539,
-      "learning_rate": 2.5402115598729182e-06,
-      "loss": 0.4893,
-      "mean_token_accuracy": 0.866288959980011,
-      "step": 391
-    },
-    {
-      "epoch": 0.7935222672064778,
-      "grad_norm": 0.30137163400650024,
-      "learning_rate": 2.493276947472756e-06,
-      "loss": 0.4787,
-      "mean_token_accuracy": 0.8605625629425049,
-      "step": 392
-    },
-    {
-      "epoch": 0.7955465587044535,
-      "grad_norm": 0.3619798719882965,
-      "learning_rate": 2.446718157234832e-06,
-      "loss": 0.5242,
-      "mean_token_accuracy": 0.8560822606086731,
-      "step": 393
-    },
-    {
-      "epoch": 0.7975708502024291,
-      "grad_norm": 0.3067499101161957,
-      "learning_rate": 2.4005375201130275e-06,
-      "loss": 0.4447,
-      "mean_token_accuracy": 0.8781384229660034,
-      "step": 394
-    },
-    {
-      "epoch": 0.7995951417004049,
-      "grad_norm": 0.31653648614883423,
-      "learning_rate": 2.354737348129077e-06,
-      "loss": 0.5013,
-      "mean_token_accuracy": 0.8644986152648926,
-      "step": 395
-    },
-    {
-      "epoch": 0.8016194331983806,
-      "grad_norm": 0.3089480996131897,
-      "learning_rate": 2.3093199342568316e-06,
-      "loss": 0.4508,
-      "mean_token_accuracy": 0.8853985071182251,
-      "step": 396
-    },
-    {
-      "epoch": 0.8036437246963563,
-      "grad_norm": 0.2246381789445877,
-      "learning_rate": 2.2642875523074613e-06,
-      "loss": 0.3853,
-      "mean_token_accuracy": 0.8819203972816467,
-      "step": 397
-    },
-    {
-      "epoch": 0.805668016194332,
-      "grad_norm": 0.29770413041114807,
-      "learning_rate": 2.2196424568156073e-06,
-      "loss": 0.5376,
-      "mean_token_accuracy": 0.8378896713256836,
-      "step": 398
-    },
-    {
-      "epoch": 0.8076923076923077,
-      "grad_norm": 0.333595335483551,
-      "learning_rate": 2.1753868829265046e-06,
-      "loss": 0.4522,
-      "mean_token_accuracy": 0.8549618124961853,
-      "step": 399
-    },
-    {
-      "epoch": 0.8097165991902834,
-      "grad_norm": 0.2043197900056839,
-      "learning_rate": 2.1315230462840985e-06,
-      "loss": 0.3687,
-      "mean_token_accuracy": 0.8900883197784424,
-      "step": 400
-    },
-    {
-      "epoch": 0.8117408906882592,
-      "grad_norm": 0.2769086956977844,
-      "learning_rate": 2.0880531429201146e-06,
-      "loss": 0.4606,
-      "mean_token_accuracy": 0.8631333708763123,
-      "step": 401
-    },
-    {
-      "epoch": 0.8137651821862348,
-      "grad_norm": 0.22918100655078888,
-      "learning_rate": 2.0449793491441026e-06,
-      "loss": 0.4826,
-      "mean_token_accuracy": 0.8662207126617432,
-      "step": 402
-    },
-    {
-      "epoch": 0.8157894736842105,
-      "grad_norm": 0.24323149025440216,
-      "learning_rate": 2.0023038214344827e-06,
-      "loss": 0.4381,
-      "mean_token_accuracy": 0.8717339634895325,
-      "step": 403
-    },
-    {
-      "epoch": 0.8178137651821862,
-      "grad_norm": 0.2817937135696411,
-      "learning_rate": 1.960028696330596e-06,
-      "loss": 0.4194,
-      "mean_token_accuracy": 0.8925619721412659,
-      "step": 404
-    },
-    {
-      "epoch": 0.819838056680162,
-      "grad_norm": 0.28915348649024963,
-      "learning_rate": 1.9181560903257234e-06,
-      "loss": 0.5171,
-      "mean_token_accuracy": 0.8500000238418579,
-      "step": 405
-    },
-    {
-      "epoch": 0.8218623481781376,
-      "grad_norm": 0.17370840907096863,
-      "learning_rate": 1.8766880997611424e-06,
-      "loss": 0.3169,
-      "mean_token_accuracy": 0.9196190237998962,
-      "step": 406
-    },
-    {
-      "epoch": 0.8238866396761133,
-      "grad_norm": 0.3060840368270874,
-      "learning_rate": 1.8356268007211442e-06,
-      "loss": 0.4136,
-      "mean_token_accuracy": 0.8882314562797546,
-      "step": 407
-    },
-    {
-      "epoch": 0.8259109311740891,
-      "grad_norm": 0.19518621265888214,
-      "learning_rate": 1.7949742489291256e-06,
-      "loss": 0.3566,
-      "mean_token_accuracy": 0.8933706283569336,
-      "step": 408
-    },
-    {
-      "epoch": 0.8279352226720648,
-      "grad_norm": 0.2959750294685364,
-      "learning_rate": 1.7547324796446553e-06,
-      "loss": 0.4379,
-      "mean_token_accuracy": 0.8780388832092285,
-      "step": 409
-    },
-    {
-      "epoch": 0.8299595141700404,
-      "grad_norm": 0.24308601021766663,
-      "learning_rate": 1.7149035075615795e-06,
-      "loss": 0.3944,
-      "mean_token_accuracy": 0.8858375549316406,
-      "step": 410
-    },
-    {
-      "epoch": 0.8319838056680162,
-      "grad_norm": 0.22043316066265106,
-      "learning_rate": 1.6754893267071593e-06,
-      "loss": 0.3242,
-      "mean_token_accuracy": 0.9051490426063538,
-      "step": 411
-    },
-    {
-      "epoch": 0.8340080971659919,
-      "grad_norm": 0.33916962146759033,
-      "learning_rate": 1.6364919103422394e-06,
-      "loss": 0.5231,
-      "mean_token_accuracy": 0.8668769598007202,
-      "step": 412
-    },
-    {
-      "epoch": 0.8360323886639676,
-      "grad_norm": 0.3752172887325287,
-      "learning_rate": 1.5979132108624572e-06,
-      "loss": 0.5339,
-      "mean_token_accuracy": 0.8474466800689697,
-      "step": 413
-    },
-    {
-      "epoch": 0.8380566801619433,
-      "grad_norm": 0.2694341242313385,
-      "learning_rate": 1.5597551597004968e-06,
-      "loss": 0.4139,
-      "mean_token_accuracy": 0.8806228637695312,
-      "step": 414
-    },
-    {
-      "epoch": 0.840080971659919,
-      "grad_norm": 0.2853461503982544,
-      "learning_rate": 1.522019667229393e-06,
-      "loss": 0.4129,
-      "mean_token_accuracy": 0.8857493996620178,
-      "step": 415
-    },
-    {
-      "epoch": 0.8421052631578947,
-      "grad_norm": 0.2771718502044678,
-      "learning_rate": 1.4847086226668871e-06,
-      "loss": 0.417,
-      "mean_token_accuracy": 0.8848413825035095,
-      "step": 416
-    },
-    {
-      "epoch": 0.8441295546558705,
-      "grad_norm": 0.28253501653671265,
-      "learning_rate": 1.4478238939808454e-06,
-      "loss": 0.4416,
-      "mean_token_accuracy": 0.8766881823539734,
-      "step": 417
-    },
-    {
-      "epoch": 0.8461538461538461,
-      "grad_norm": 0.2291014939546585,
-      "learning_rate": 1.4113673277957395e-06,
-      "loss": 0.4423,
-      "mean_token_accuracy": 0.8711934089660645,
-      "step": 418
-    },
-    {
-      "epoch": 0.8481781376518218,
-      "grad_norm": 0.2619485855102539,
-      "learning_rate": 1.3753407493001968e-06,
-      "loss": 0.374,
-      "mean_token_accuracy": 0.8914362788200378,
-      "step": 419
-    },
-    {
-      "epoch": 0.8502024291497976,
-      "grad_norm": 0.20695680379867554,
-      "learning_rate": 1.339745962155613e-06,
-      "loss": 0.3784,
-      "mean_token_accuracy": 0.8880422115325928,
-      "step": 420
-    },
-    {
-      "epoch": 0.8522267206477733,
-      "grad_norm": 0.24776305258274078,
-      "learning_rate": 1.3045847484058748e-06,
-      "loss": 0.3826,
-      "mean_token_accuracy": 0.889227032661438,
-      "step": 421
-    },
-    {
-      "epoch": 0.854251012145749,
-      "grad_norm": 0.2515615224838257,
-      "learning_rate": 1.2698588683881185e-06,
-      "loss": 0.4579,
-      "mean_token_accuracy": 0.8664596080780029,
-      "step": 422
-    },
-    {
-      "epoch": 0.8562753036437247,
-      "grad_norm": 0.301064133644104,
-      "learning_rate": 1.2355700606446119e-06,
-      "loss": 0.4299,
-      "mean_token_accuracy": 0.8724428415298462,
-      "step": 423
-    },
-    {
-      "epoch": 0.8582995951417004,
-      "grad_norm": 0.24376371502876282,
-      "learning_rate": 1.2017200418357077e-06,
-      "loss": 0.4607,
-      "mean_token_accuracy": 0.8685946464538574,
-      "step": 424
-    },
-    {
-      "epoch": 0.8603238866396761,
-      "grad_norm": 0.27435746788978577,
-      "learning_rate": 1.1683105066539068e-06,
-      "loss": 0.4675,
-      "mean_token_accuracy": 0.8655617237091064,
-      "step": 425
-    },
-    {
-      "epoch": 0.8623481781376519,
-      "grad_norm": 0.2541494071483612,
-      "learning_rate": 1.1353431277390125e-06,
-      "loss": 0.4656,
-      "mean_token_accuracy": 0.862048864364624,
-      "step": 426
-    },
-    {
-      "epoch": 0.8643724696356275,
-      "grad_norm": 0.27517831325531006,
-      "learning_rate": 1.1028195555943877e-06,
-      "loss": 0.4935,
-      "mean_token_accuracy": 0.8678010702133179,
-      "step": 427
-    },
-    {
-      "epoch": 0.8663967611336032,
-      "grad_norm": 0.22525008022785187,
-      "learning_rate": 1.0707414185043163e-06,
-      "loss": 0.3822,
-      "mean_token_accuracy": 0.8995452523231506,
-      "step": 428
-    },
-    {
-      "epoch": 0.868421052631579,
-      "grad_norm": 0.3820272386074066,
-      "learning_rate": 1.0391103224524957e-06,
-      "loss": 0.446,
-      "mean_token_accuracy": 0.8697771430015564,
-      "step": 429
-    },
-    {
-      "epoch": 0.8704453441295547,
-      "grad_norm": 0.164522185921669,
-      "learning_rate": 1.0079278510416313e-06,
-      "loss": 0.3243,
-      "mean_token_accuracy": 0.8974565863609314,
-      "step": 430
-    },
-    {
-      "epoch": 0.8724696356275303,
-      "grad_norm": 0.2887153923511505,
-      "learning_rate": 9.771955654141496e-07,
-      "loss": 0.457,
-      "mean_token_accuracy": 0.8749398589134216,
-      "step": 431
-    },
-    {
-      "epoch": 0.8744939271255061,
-      "grad_norm": 0.26364317536354065,
-      "learning_rate": 9.469150041740338e-07,
-      "loss": 0.4695,
-      "mean_token_accuracy": 0.8601856827735901,
-      "step": 432
-    },
-    {
-      "epoch": 0.8765182186234818,
-      "grad_norm": 0.23383867740631104,
-      "learning_rate": 9.170876833098119e-07,
-      "loss": 0.3645,
-      "mean_token_accuracy": 0.9048991203308105,
-      "step": 433
-    },
-    {
-      "epoch": 0.8785425101214575,
-      "grad_norm": 0.2173306941986084,
-      "learning_rate": 8.87715096118642e-07,
-      "loss": 0.458,
-      "mean_token_accuracy": 0.8632075190544128,
-      "step": 434
-    },
-    {
-      "epoch": 0.8805668016194332,
-      "grad_norm": 0.29172396659851074,
-      "learning_rate": 8.587987131315656e-07,
-      "loss": 0.4671,
-      "mean_token_accuracy": 0.8700696229934692,
-      "step": 435
-    },
-    {
-      "epoch": 0.8825910931174089,
-      "grad_norm": 0.23296132683753967,
-      "learning_rate": 8.303399820398672e-07,
-      "loss": 0.404,
-      "mean_token_accuracy": 0.8798627257347107,
-      "step": 436
-    },
-    {
-      "epoch": 0.8846153846153846,
-      "grad_norm": 0.25785768032073975,
-      "learning_rate": 8.023403276226127e-07,
-      "loss": 0.4228,
-      "mean_token_accuracy": 0.8814306855201721,
-      "step": 437
-    },
-    {
-      "epoch": 0.8866396761133604,
-      "grad_norm": 0.26605987548828125,
-      "learning_rate": 7.74801151675314e-07,
-      "loss": 0.4327,
-      "mean_token_accuracy": 0.8785097002983093,
-      "step": 438
-    },
-    {
-      "epoch": 0.888663967611336,
-      "grad_norm": 0.25588130950927734,
-      "learning_rate": 7.477238329397419e-07,
-      "loss": 0.4919,
-      "mean_token_accuracy": 0.8562639951705933,
-      "step": 439
-    },
-    {
-      "epoch": 0.8906882591093117,
-      "grad_norm": 0.2864658832550049,
-      "learning_rate": 7.211097270349065e-07,
-      "loss": 0.4974,
-      "mean_token_accuracy": 0.8614386916160583,
-      "step": 440
-    },
-    {
-      "epoch": 0.8927125506072875,
-      "grad_norm": 0.2550905644893646,
-      "learning_rate": 6.949601663891891e-07,
-      "loss": 0.4284,
-      "mean_token_accuracy": 0.8829268217086792,
-      "step": 441
-    },
-    {
-      "epoch": 0.8947368421052632,
-      "grad_norm": 0.19677165150642395,
-      "learning_rate": 6.692764601736268e-07,
-      "loss": 0.4149,
-      "mean_token_accuracy": 0.875516951084137,
-      "step": 442
-    },
-    {
-      "epoch": 0.8967611336032388,
-      "grad_norm": 0.3176872432231903,
-      "learning_rate": 6.440598942363796e-07,
-      "loss": 0.4427,
-      "mean_token_accuracy": 0.8617944717407227,
-      "step": 443
-    },
-    {
-      "epoch": 0.8987854251012146,
-      "grad_norm": 0.27307766675949097,
-      "learning_rate": 6.193117310383412e-07,
-      "loss": 0.4616,
-      "mean_token_accuracy": 0.8798370957374573,
-      "step": 444
-    },
-    {
-      "epoch": 0.9008097165991903,
-      "grad_norm": 0.28930559754371643,
-      "learning_rate": 5.950332095899547e-07,
-      "loss": 0.4958,
-      "mean_token_accuracy": 0.8664363622665405,
-      "step": 445
-    },
-    {
-      "epoch": 0.902834008097166,
-      "grad_norm": 0.24667935073375702,
-      "learning_rate": 5.71225545389158e-07,
-      "loss": 0.5217,
-      "mean_token_accuracy": 0.8562564849853516,
-      "step": 446
-    },
-    {
-      "epoch": 0.9048582995951417,
-      "grad_norm": 0.2966341972351074,
-      "learning_rate": 5.478899303605512e-07,
-      "loss": 0.5367,
-      "mean_token_accuracy": 0.8706467747688293,
-      "step": 447
-    },
-    {
-      "epoch": 0.9068825910931174,
-      "grad_norm": 0.21977975964546204,
-      "learning_rate": 5.250275327957033e-07,
-      "loss": 0.4222,
-      "mean_token_accuracy": 0.8754432797431946,
-      "step": 448
-    },
-    {
-      "epoch": 0.9089068825910931,
-      "grad_norm": 0.26447972655296326,
-      "learning_rate": 5.026394972946813e-07,
-      "loss": 0.4087,
-      "mean_token_accuracy": 0.8903620839118958,
-      "step": 449
-    },
-    {
-      "epoch": 0.9109311740890689,
-      "grad_norm": 0.2512384355068207,
-      "learning_rate": 4.807269447087348e-07,
-      "loss": 0.471,
-      "mean_token_accuracy": 0.8707289099693298,
-      "step": 450
-    },
-    {
-      "epoch": 0.9129554655870445,
-      "grad_norm": 0.2299729883670807,
-      "learning_rate": 4.592909720841843e-07,
-      "loss": 0.413,
-      "mean_token_accuracy": 0.8833592534065247,
-      "step": 451
-    },
-    {
-      "epoch": 0.9149797570850202,
-      "grad_norm": 0.2830863296985626,
-      "learning_rate": 4.3833265260749157e-07,
-      "loss": 0.4341,
-      "mean_token_accuracy": 0.896039605140686,
-      "step": 452
-    },
-    {
-      "epoch": 0.917004048582996,
-      "grad_norm": 0.2753170132637024,
-      "learning_rate": 4.178530355515409e-07,
-      "loss": 0.4285,
-      "mean_token_accuracy": 0.8766725063323975,
-      "step": 453
-    },
-    {
-      "epoch": 0.9190283400809717,
-      "grad_norm": 0.18439187109470367,
-      "learning_rate": 3.97853146223105e-07,
-      "loss": 0.3962,
-      "mean_token_accuracy": 0.8877434134483337,
-      "step": 454
-    },
-    {
-      "epoch": 0.9210526315789473,
-      "grad_norm": 0.3276051878929138,
-      "learning_rate": 3.783339859115065e-07,
-      "loss": 0.4733,
-      "mean_token_accuracy": 0.8768656849861145,
-      "step": 455
-    },
-    {
-      "epoch": 0.9230769230769231,
-      "grad_norm": 0.302237868309021,
-      "learning_rate": 3.5929653183849444e-07,
-      "loss": 0.4583,
-      "mean_token_accuracy": 0.8618030548095703,
-      "step": 456
-    },
-    {
-      "epoch": 0.9251012145748988,
-      "grad_norm": 0.2920514941215515,
-      "learning_rate": 3.4074173710931804e-07,
-      "loss": 0.4657,
-      "mean_token_accuracy": 0.8741217851638794,
-      "step": 457
-    },
-    {
-      "epoch": 0.9271255060728745,
-      "grad_norm": 0.14648732542991638,
-      "learning_rate": 3.226705306650113e-07,
-      "loss": 0.3451,
-      "mean_token_accuracy": 0.8986467719078064,
-      "step": 458
-    },
-    {
-      "epoch": 0.9291497975708503,
-      "grad_norm": 0.24604742228984833,
-      "learning_rate": 3.050838172358883e-07,
-      "loss": 0.3458,
-      "mean_token_accuracy": 0.9110772609710693,
-      "step": 459
-    },
-    {
-      "epoch": 0.9311740890688259,
-      "grad_norm": 0.24755385518074036,
-      "learning_rate": 2.879824772962381e-07,
-      "loss": 0.3859,
-      "mean_token_accuracy": 0.891435980796814,
-      "step": 460
-    },
-    {
-      "epoch": 0.9331983805668016,
-      "grad_norm": 0.2505938708782196,
-      "learning_rate": 2.7136736702025436e-07,
-      "loss": 0.4337,
-      "mean_token_accuracy": 0.8723084330558777,
-      "step": 461
-    },
-    {
-      "epoch": 0.9352226720647774,
-      "grad_norm": 0.2497698813676834,
-      "learning_rate": 2.552393182391677e-07,
-      "loss": 0.4469,
-      "mean_token_accuracy": 0.8665005564689636,
-      "step": 462
-    },
-    {
-      "epoch": 0.937246963562753,
-      "grad_norm": 0.3191662132740021,
-      "learning_rate": 2.395991383995999e-07,
-      "loss": 0.4424,
-      "mean_token_accuracy": 0.8791821599006653,
-      "step": 463
-    },
-    {
-      "epoch": 0.9392712550607287,
-      "grad_norm": 0.4106998145580292,
-      "learning_rate": 2.2444761052313857e-07,
-      "loss": 0.5896,
-      "mean_token_accuracy": 0.8302023410797119,
-      "step": 464
-    },
-    {
-      "epoch": 0.9412955465587044,
-      "grad_norm": 0.18561993539333344,
-      "learning_rate": 2.0978549316713615e-07,
-      "loss": 0.3504,
-      "mean_token_accuracy": 0.8924598097801208,
-      "step": 465
-    },
-    {
-      "epoch": 0.9433198380566802,
-      "grad_norm": 0.33718597888946533,
-      "learning_rate": 1.9561352038673264e-07,
-      "loss": 0.4549,
-      "mean_token_accuracy": 0.8738977313041687,
-      "step": 466
-    },
-    {
-      "epoch": 0.9453441295546559,
-      "grad_norm": 0.3026043474674225,
-      "learning_rate": 1.8193240169810943e-07,
-      "loss": 0.4668,
-      "mean_token_accuracy": 0.8618420958518982,
-      "step": 467
-    },
-    {
-      "epoch": 0.9473684210526315,
-      "grad_norm": 0.26501694321632385,
-      "learning_rate": 1.6874282204295765e-07,
-      "loss": 0.4307,
-      "mean_token_accuracy": 0.8807380795478821,
-      "step": 468
-    },
-    {
-      "epoch": 0.9493927125506073,
-      "grad_norm": 0.26896053552627563,
-      "learning_rate": 1.5604544175419901e-07,
-      "loss": 0.3877,
-      "mean_token_accuracy": 0.8920484185218811,
-      "step": 469
-    },
-    {
-      "epoch": 0.951417004048583,
-      "grad_norm": 0.2413828819990158,
-      "learning_rate": 1.4384089652291544e-07,
-      "loss": 0.3673,
-      "mean_token_accuracy": 0.8918406367301941,
-      "step": 470
-    },
-    {
-      "epoch": 0.9534412955465587,
-      "grad_norm": 0.24448032677173615,
-      "learning_rate": 1.3212979736653142e-07,
-      "loss": 0.3627,
-      "mean_token_accuracy": 0.8885869383811951,
-      "step": 471
-    },
-    {
-      "epoch": 0.9554655870445344,
-      "grad_norm": 0.29778435826301575,
-      "learning_rate": 1.209127305982205e-07,
-      "loss": 0.4382,
-      "mean_token_accuracy": 0.8771374225616455,
-      "step": 472
-    },
-    {
-      "epoch": 0.9574898785425101,
-      "grad_norm": 0.20374563336372375,
-      "learning_rate": 1.1019025779754666e-07,
-      "loss": 0.3854,
-      "mean_token_accuracy": 0.886918842792511,
-      "step": 473
-    },
-    {
-      "epoch": 0.9595141700404858,
-      "grad_norm": 0.33799806237220764,
-      "learning_rate": 9.996291578236228e-08,
-      "loss": 0.5051,
-      "mean_token_accuracy": 0.8672335147857666,
-      "step": 474
-    },
-    {
-      "epoch": 0.9615384615384616,
-      "grad_norm": 0.3109436333179474,
-      "learning_rate": 9.023121658191636e-08,
-      "loss": 0.4186,
-      "mean_token_accuracy": 0.8837453722953796,
-      "step": 475
-    },
-    {
-      "epoch": 0.9635627530364372,
-      "grad_norm": 0.27846312522888184,
-      "learning_rate": 8.099564741123167e-08,
-      "loss": 0.4549,
-      "mean_token_accuracy": 0.8804086446762085,
-      "step": 476
-    },
-    {
-      "epoch": 0.9655870445344129,
-      "grad_norm": 0.247327521443367,
-      "learning_rate": 7.225667064670761e-08,
-      "loss": 0.3735,
-      "mean_token_accuracy": 0.8910034894943237,
-      "step": 477
-    },
-    {
-      "epoch": 0.9676113360323887,
-      "grad_norm": 0.2059151977300644,
-      "learning_rate": 6.401472380297091e-08,
-      "loss": 0.3844,
-      "mean_token_accuracy": 0.8830559253692627,
-      "step": 478
-    },
-    {
-      "epoch": 0.9696356275303644,
-      "grad_norm": 0.26618608832359314,
-      "learning_rate": 5.6270219510975445e-08,
-      "loss": 0.4498,
-      "mean_token_accuracy": 0.8715971112251282,
-      "step": 479
-    },
-    {
-      "epoch": 0.97165991902834,
-      "grad_norm": 0.31504425406455994,
-      "learning_rate": 4.902354549733979e-08,
-      "loss": 0.4592,
-      "mean_token_accuracy": 0.8812423944473267,
-      "step": 480
-    },
-    {
-      "epoch": 0.9736842105263158,
-      "grad_norm": 0.24930904805660248,
-      "learning_rate": 4.227506456493835e-08,
-      "loss": 0.4226,
-      "mean_token_accuracy": 0.8748299479484558,
-      "step": 481
-    },
-    {
-      "epoch": 0.9757085020242915,
-      "grad_norm": 0.23372402787208557,
-      "learning_rate": 3.602511457473479e-08,
-      "loss": 0.3934,
-      "mean_token_accuracy": 0.8909899592399597,
-      "step": 482
-    },
-    {
-      "epoch": 0.9777327935222672,
-      "grad_norm": 0.25504592061042786,
-      "learning_rate": 3.027400842887218e-08,
-      "loss": 0.4117,
-      "mean_token_accuracy": 0.8802729249000549,
-      "step": 483
-    },
-    {
-      "epoch": 0.979757085020243,
-      "grad_norm": 0.2677006721496582,
-      "learning_rate": 2.5022034055003363e-08,
-      "loss": 0.3958,
-      "mean_token_accuracy": 0.8953003883361816,
-      "step": 484
-    },
-    {
-      "epoch": 0.9817813765182186,
-      "grad_norm": 0.32685738801956177,
-      "learning_rate": 2.0269454391874665e-08,
-      "loss": 0.4479,
-      "mean_token_accuracy": 0.8705402612686157,
-      "step": 485
-    },
-    {
-      "epoch": 0.9838056680161943,
-      "grad_norm": 0.267930269241333,
-      "learning_rate": 1.6016507376169776e-08,
-      "loss": 0.3837,
-      "mean_token_accuracy": 0.8849610686302185,
-      "step": 486
-    },
-    {
-      "epoch": 0.9858299595141701,
-      "grad_norm": 0.3171535134315491,
-      "learning_rate": 1.2263405930585947e-08,
-      "loss": 0.5383,
-      "mean_token_accuracy": 0.8503844141960144,
-      "step": 487
-    },
-    {
-      "epoch": 0.9878542510121457,
-      "grad_norm": 0.31351831555366516,
-      "learning_rate": 9.010337953185843e-09,
-      "loss": 0.4605,
-      "mean_token_accuracy": 0.8758657574653625,
-      "step": 488
-    },
-    {
-      "epoch": 0.9898785425101214,
-      "grad_norm": 0.2406328022480011,
-      "learning_rate": 6.257466307980631e-09,
-      "loss": 0.4072,
-      "mean_token_accuracy": 0.8733096122741699,
-      "step": 489
-    },
-    {
-      "epoch": 0.9919028340080972,
-      "grad_norm": 0.3346785604953766,
-      "learning_rate": 4.00492881678427e-09,
-      "loss": 0.4925,
-      "mean_token_accuracy": 0.8693274855613708,
-      "step": 490
-    },
-    {
-      "epoch": 0.9939271255060729,
-      "grad_norm": 0.26899194717407227,
-      "learning_rate": 2.2528382523057115e-09,
-      "loss": 0.4615,
-      "mean_token_accuracy": 0.8673709034919739,
-      "step": 491
-    },
-    {
-      "epoch": 0.9959514170040485,
-      "grad_norm": 0.19365371763706207,
-      "learning_rate": 1.0012823325111776e-09,
-      "loss": 0.3367,
-      "mean_token_accuracy": 0.8977272510528564,
-      "step": 492
-    },
-    {
-      "epoch": 0.9979757085020243,
-      "grad_norm": 0.2652401030063629,
-      "learning_rate": 2.503237162254646e-10,
-      "loss": 0.5036,
-      "mean_token_accuracy": 0.8553218841552734,
-      "step": 493
-    },
-    {
-      "epoch": 1.0,
-      "grad_norm": 0.3128429055213928,
-      "learning_rate": 0.0,
-      "loss": 0.485,
-      "mean_token_accuracy": 0.8615916967391968,
-      "step": 494
-    },
-    {
-      "epoch": 1.0,
-      "step": 494,
-      "total_flos": 1.3768541467574272e+16,
-      "train_loss": 0.5140652489444988,
-      "train_runtime": 284.5417,
-      "train_samples_per_second": 13.882,
-      "train_steps_per_second": 1.736
+      "epoch": 4.712550607287449,
+      "step": 75,
+      "total_flos": 3.216438415536947e+16,
+      "train_loss": 0.6463175455729167,
+      "train_runtime": 630.4846,
+      "train_samples_per_second": 15.663,
+      "train_steps_per_second": 0.119
     }
   ],
   "logging_steps": 1,
-  "max_steps": 494,
+  "max_steps": 75,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
-  "save_steps": 100,
+  "num_train_epochs": 5,
+  "save_steps": 1,
   "stateful_callbacks": {
     "TrainerControl": {
       "args": {
@@ -3987,7 +635,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.3768541467574272e+16,
+  "total_flos": 3.216438415536947e+16,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null