{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 50.0,
  "global_step": 320,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012618296529968454,
      "grad_norm": 431.9578552246094,
      "learning_rate": 6.25e-07,
      "loss": 5.363270282745361,
      "step": 1,
      "token_acc": 0.6227678571428571
    },
    {
      "epoch": 0.025236593059936908,
      "grad_norm": 424.51800537109375,
      "learning_rate": 1.25e-06,
      "loss": 5.4433393478393555,
      "step": 2,
      "token_acc": 0.5982142857142857
    },
    {
      "epoch": 0.03785488958990536,
      "grad_norm": 423.6788024902344,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 5.289504528045654,
      "step": 3,
      "token_acc": 0.6138392857142857
    },
    {
      "epoch": 0.050473186119873815,
      "grad_norm": 413.3780212402344,
      "learning_rate": 2.5e-06,
      "loss": 5.160651206970215,
      "step": 4,
      "token_acc": 0.6316964285714286
    },
    {
      "epoch": 0.06309148264984227,
      "grad_norm": 400.853515625,
      "learning_rate": 3.125e-06,
      "loss": 4.153102397918701,
      "step": 5,
      "token_acc": 0.6584821428571429
    },
    {
      "epoch": 0.07570977917981073,
      "grad_norm": 122.95317077636719,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 2.2181897163391113,
      "step": 6,
      "token_acc": 0.8013392857142857
    },
    {
      "epoch": 0.08832807570977919,
      "grad_norm": 265.0997009277344,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 1.9256503582000732,
      "step": 7,
      "token_acc": 0.734375
    },
    {
      "epoch": 0.10094637223974763,
      "grad_norm": 115.00222778320312,
      "learning_rate": 5e-06,
      "loss": 1.7824956178665161,
      "step": 8,
      "token_acc": 0.7857142857142857
    },
    {
      "epoch": 0.11356466876971609,
      "grad_norm": 123.10432434082031,
      "learning_rate": 5.625e-06,
      "loss": 1.8616050481796265,
      "step": 9,
      "token_acc": 0.7834821428571429
    },
    {
      "epoch": 0.12618296529968454,
      "grad_norm": 91.25232696533203,
      "learning_rate": 6.25e-06,
      "loss": 0.8788753747940063,
      "step": 10,
      "token_acc": 0.7879464285714286
    },
    {
      "epoch": 0.138801261829653,
      "grad_norm": 61.82997131347656,
      "learning_rate": 6.875e-06,
      "loss": 0.242479607462883,
      "step": 11,
      "token_acc": 0.8392857142857143
    },
    {
      "epoch": 0.15141955835962145,
      "grad_norm": 4.125986576080322,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.09646491706371307,
      "step": 12,
      "token_acc": 0.9419642857142857
    },
    {
      "epoch": 0.1640378548895899,
      "grad_norm": 43.74650192260742,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.3005182147026062,
      "step": 13,
      "token_acc": 0.9285714285714286
    },
    {
      "epoch": 0.17665615141955837,
      "grad_norm": 8.703579902648926,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.12107616662979126,
      "step": 14,
      "token_acc": 0.921875
    },
    {
      "epoch": 0.1892744479495268,
      "grad_norm": 8.164538383483887,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.12075388431549072,
      "step": 15,
      "token_acc": 0.9241071428571429
    },
    {
      "epoch": 0.20189274447949526,
      "grad_norm": 3.066166639328003,
      "learning_rate": 1e-05,
      "loss": 0.10026569664478302,
      "step": 16,
      "token_acc": 0.9397321428571429
    },
    {
      "epoch": 0.21451104100946372,
      "grad_norm": 7.341068744659424,
      "learning_rate": 1.0625e-05,
      "loss": 0.11513940989971161,
      "step": 17,
      "token_acc": 0.9263392857142857
    },
    {
      "epoch": 0.22712933753943218,
      "grad_norm": 5.823333740234375,
      "learning_rate": 1.125e-05,
      "loss": 0.11126351356506348,
      "step": 18,
      "token_acc": 0.9285714285714286
    },
    {
      "epoch": 0.23974763406940064,
      "grad_norm": 0.40053513646125793,
      "learning_rate": 1.1875e-05,
      "loss": 0.09615738689899445,
      "step": 19,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.25236593059936907,
      "grad_norm": 2.4639089107513428,
      "learning_rate": 1.25e-05,
      "loss": 0.09931515157222748,
      "step": 20,
      "token_acc": 0.9352678571428571
    },
    {
      "epoch": 0.26498422712933756,
      "grad_norm": 5.5744218826293945,
      "learning_rate": 1.3125e-05,
      "loss": 0.10292527079582214,
      "step": 21,
      "token_acc": 0.9308035714285714
    },
    {
      "epoch": 0.277602523659306,
      "grad_norm": 1.8323853015899658,
      "learning_rate": 1.375e-05,
      "loss": 0.09089958667755127,
      "step": 22,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.2902208201892745,
      "grad_norm": 3.054859161376953,
      "learning_rate": 1.4375e-05,
      "loss": 0.08756621927022934,
      "step": 23,
      "token_acc": 0.9508928571428571
    },
    {
      "epoch": 0.3028391167192429,
      "grad_norm": 3.557588815689087,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.08846282958984375,
      "step": 24,
      "token_acc": 0.9464285714285714
    },
    {
      "epoch": 0.31545741324921134,
      "grad_norm": 6.949894905090332,
      "learning_rate": 1.5625e-05,
      "loss": 0.12064121663570404,
      "step": 25,
      "token_acc": 0.9464285714285714
    },
    {
      "epoch": 0.3280757097791798,
      "grad_norm": 2.9046874046325684,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.10017058253288269,
      "step": 26,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.34069400630914826,
      "grad_norm": 1.3645375967025757,
      "learning_rate": 1.6875e-05,
      "loss": 0.08028960227966309,
      "step": 27,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 0.35331230283911674,
      "grad_norm": 1.0991082191467285,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.08341425657272339,
      "step": 28,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 0.3659305993690852,
      "grad_norm": 1.1110965013504028,
      "learning_rate": 1.8125e-05,
      "loss": 0.09392577409744263,
      "step": 29,
      "token_acc": 0.9397321428571429
    },
    {
      "epoch": 0.3785488958990536,
      "grad_norm": 2.394069194793701,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.09441205114126205,
      "step": 30,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.3911671924290221,
      "grad_norm": 1.75429368019104,
      "learning_rate": 1.9375e-05,
      "loss": 0.08634346723556519,
      "step": 31,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 0.4037854889589905,
      "grad_norm": 0.8835183382034302,
      "learning_rate": 2e-05,
      "loss": 0.07859288901090622,
      "step": 32,
      "token_acc": 0.9598214285714286
    },
    {
      "epoch": 0.416403785488959,
      "grad_norm": 4.872382164001465,
      "learning_rate": 1.99994050500015e-05,
      "loss": 0.09923085570335388,
      "step": 33,
      "token_acc": 0.9464285714285714
    },
    {
      "epoch": 0.42902208201892744,
      "grad_norm": 3.0629661083221436,
      "learning_rate": 1.999762027079909e-05,
      "loss": 0.09347166121006012,
      "step": 34,
      "token_acc": 0.9397321428571429
    },
    {
      "epoch": 0.4416403785488959,
      "grad_norm": 4.798449516296387,
      "learning_rate": 1.9994645874763657e-05,
      "loss": 0.10271646082401276,
      "step": 35,
      "token_acc": 0.953125
    },
    {
      "epoch": 0.45425867507886436,
      "grad_norm": 4.8639302253723145,
      "learning_rate": 1.999048221581858e-05,
      "loss": 0.11419010162353516,
      "step": 36,
      "token_acc": 0.9419642857142857
    },
    {
      "epoch": 0.4668769716088328,
      "grad_norm": 2.529944896697998,
      "learning_rate": 1.9985129789397633e-05,
      "loss": 0.10211747139692307,
      "step": 37,
      "token_acc": 0.9486607142857143
    },
    {
      "epoch": 0.4794952681388013,
      "grad_norm": 1.820770263671875,
      "learning_rate": 1.9978589232386036e-05,
      "loss": 0.09178828448057175,
      "step": 38,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.4921135646687697,
      "grad_norm": 4.190671920776367,
      "learning_rate": 1.9970861323044667e-05,
      "loss": 0.10167853534221649,
      "step": 39,
      "token_acc": 0.9241071428571429
    },
    {
      "epoch": 0.5047318611987381,
      "grad_norm": 3.732025623321533,
      "learning_rate": 1.9961946980917457e-05,
      "loss": 0.10093965381383896,
      "step": 40,
      "token_acc": 0.9397321428571429
    },
    {
      "epoch": 0.5173501577287066,
      "grad_norm": 0.2019195258617401,
      "learning_rate": 1.995184726672197e-05,
      "loss": 0.09062772989273071,
      "step": 41,
      "token_acc": 0.953125
    },
    {
      "epoch": 0.5299684542586751,
      "grad_norm": 3.979827404022217,
      "learning_rate": 1.9940563382223196e-05,
      "loss": 0.10666163265705109,
      "step": 42,
      "token_acc": 0.9241071428571429
    },
    {
      "epoch": 0.5425867507886435,
      "grad_norm": 1.1392254829406738,
      "learning_rate": 1.9928096670090552e-05,
      "loss": 0.08142286539077759,
      "step": 43,
      "token_acc": 0.9508928571428571
    },
    {
      "epoch": 0.555205047318612,
      "grad_norm": 1.829756259918213,
      "learning_rate": 1.9914448613738107e-05,
      "loss": 0.0897851288318634,
      "step": 44,
      "token_acc": 0.9508928571428571
    },
    {
      "epoch": 0.5678233438485805,
      "grad_norm": 2.5378029346466064,
      "learning_rate": 1.989962083714808e-05,
      "loss": 0.09194290637969971,
      "step": 45,
      "token_acc": 0.9486607142857143
    },
    {
      "epoch": 0.580441640378549,
      "grad_norm": 3.352098226547241,
      "learning_rate": 1.988361510467761e-05,
      "loss": 0.09649404138326645,
      "step": 46,
      "token_acc": 0.9464285714285714
    },
    {
      "epoch": 0.5930599369085173,
      "grad_norm": 4.520278453826904,
      "learning_rate": 1.9866433320848793e-05,
      "loss": 0.10063958168029785,
      "step": 47,
      "token_acc": 0.9352678571428571
    },
    {
      "epoch": 0.6056782334384858,
      "grad_norm": 2.0867037773132324,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 0.08037380874156952,
      "step": 48,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 0.6182965299684543,
      "grad_norm": 2.2182278633117676,
      "learning_rate": 1.9828549916653013e-05,
      "loss": 0.09005570411682129,
      "step": 49,
      "token_acc": 0.953125
    },
    {
      "epoch": 0.6309148264984227,
      "grad_norm": 5.569756031036377,
      "learning_rate": 1.9807852804032306e-05,
      "loss": 0.11957938224077225,
      "step": 50,
      "token_acc": 0.9352678571428571
    },
    {
      "epoch": 0.6435331230283912,
      "grad_norm": 2.665590524673462,
      "learning_rate": 1.9785988655009386e-05,
      "loss": 0.10452645272016525,
      "step": 51,
      "token_acc": 0.9397321428571429
    },
    {
      "epoch": 0.6561514195583596,
      "grad_norm": 2.575852155685425,
      "learning_rate": 1.9762960071199334e-05,
      "loss": 0.08253096789121628,
      "step": 52,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.668769716088328,
      "grad_norm": 0.6380723118782043,
      "learning_rate": 1.9738769792773338e-05,
      "loss": 0.07592251896858215,
      "step": 53,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 0.6813880126182965,
      "grad_norm": 2.6782314777374268,
      "learning_rate": 1.9713420698132614e-05,
      "loss": 0.08651553839445114,
      "step": 54,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.694006309148265,
      "grad_norm": 3.79976487159729,
      "learning_rate": 1.9686915803565934e-05,
      "loss": 0.10068698227405548,
      "step": 55,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.7066246056782335,
      "grad_norm": 2.8129332065582275,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 0.08456991612911224,
      "step": 56,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 0.7192429022082019,
      "grad_norm": 0.4787522256374359,
      "learning_rate": 1.963045136707763e-05,
      "loss": 0.08331582695245743,
      "step": 57,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 0.7318611987381703,
      "grad_norm": 2.618011951446533,
      "learning_rate": 1.960049854385929e-05,
      "loss": 0.08145452290773392,
      "step": 58,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.7444794952681388,
      "grad_norm": 3.104342222213745,
      "learning_rate": 1.956940335732209e-05,
      "loss": 0.09342701733112335,
      "step": 59,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.7570977917981072,
      "grad_norm": 0.8563794493675232,
      "learning_rate": 1.953716950748227e-05,
      "loss": 0.09253671765327454,
      "step": 60,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.7697160883280757,
      "grad_norm": 0.8682177066802979,
      "learning_rate": 1.9503800829845613e-05,
      "loss": 0.09381808340549469,
      "step": 61,
      "token_acc": 0.9464285714285714
    },
    {
      "epoch": 0.7823343848580442,
      "grad_norm": 0.4455837309360504,
      "learning_rate": 1.946930129495106e-05,
      "loss": 0.07521934062242508,
      "step": 62,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 0.7949526813880127,
      "grad_norm": 2.8189454078674316,
      "learning_rate": 1.9433675007898255e-05,
      "loss": 0.0848093032836914,
      "step": 63,
      "token_acc": 0.9598214285714286
    },
    {
      "epoch": 0.807570977917981,
      "grad_norm": 3.9658970832824707,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.08638809621334076,
      "step": 64,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.8201892744479495,
      "grad_norm": 0.4017783999443054,
      "learning_rate": 1.935905926757326e-05,
      "loss": 0.08558817207813263,
      "step": 65,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.832807570977918,
      "grad_norm": 0.6366514563560486,
      "learning_rate": 1.932007869282799e-05,
      "loss": 0.07810231298208237,
      "step": 66,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 0.8454258675078864,
      "grad_norm": 2.6642279624938965,
      "learning_rate": 1.9279989121921846e-05,
      "loss": 0.09261595457792282,
      "step": 67,
      "token_acc": 0.9486607142857143
    },
    {
      "epoch": 0.8580441640378549,
      "grad_norm": 1.3065534830093384,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.07471144944429398,
      "step": 68,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 0.8706624605678234,
      "grad_norm": 4.052402973175049,
      "learning_rate": 1.9196502204050925e-05,
      "loss": 0.10394419729709625,
      "step": 69,
      "token_acc": 0.9508928571428571
    },
    {
      "epoch": 0.8832807570977917,
      "grad_norm": 1.02911376953125,
      "learning_rate": 1.9153114791194475e-05,
      "loss": 0.09333102405071259,
      "step": 70,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.8958990536277602,
      "grad_norm": 2.5945627689361572,
      "learning_rate": 1.910863824921176e-05,
      "loss": 0.08339155465364456,
      "step": 71,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.9085173501577287,
      "grad_norm": 1.4363032579421997,
      "learning_rate": 1.9063077870366504e-05,
      "loss": 0.08213169872760773,
      "step": 72,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 0.9211356466876972,
      "grad_norm": 3.1371400356292725,
      "learning_rate": 1.901643907588816e-05,
      "loss": 0.10854453593492508,
      "step": 73,
      "token_acc": 0.9375
    },
    {
      "epoch": 0.9337539432176656,
      "grad_norm": 2.135662078857422,
      "learning_rate": 1.8968727415326885e-05,
      "loss": 0.08203621953725815,
      "step": 74,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 0.9463722397476341,
      "grad_norm": 0.19152504205703735,
      "learning_rate": 1.8919948565893144e-05,
      "loss": 0.08554915338754654,
      "step": 75,
      "token_acc": 0.9598214285714286
    },
    {
      "epoch": 0.9589905362776026,
      "grad_norm": 4.921126365661621,
      "learning_rate": 1.887010833178222e-05,
      "loss": 0.11141403019428253,
      "step": 76,
      "token_acc": 0.9263392857142857
    },
    {
      "epoch": 0.9716088328075709,
      "grad_norm": 3.546910524368286,
      "learning_rate": 1.881921264348355e-05,
      "loss": 0.09518605470657349,
      "step": 77,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.9842271293375394,
      "grad_norm": 1.8690577745437622,
      "learning_rate": 1.876726755707508e-05,
      "loss": 0.0894484743475914,
      "step": 78,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 0.9968454258675079,
      "grad_norm": 0.9913718700408936,
      "learning_rate": 1.8714279253502616e-05,
      "loss": 0.08974546194076538,
      "step": 79,
      "token_acc": 0.9486607142857143
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.460970401763916,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.0891992449760437,
      "step": 80,
      "token_acc": 0.9464285714285714
    },
    {
      "epoch": 1.0126182965299684,
      "grad_norm": 0.2746261656284332,
      "learning_rate": 1.860519833856079e-05,
      "loss": 0.0718148946762085,
      "step": 81,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.025236593059937,
      "grad_norm": 0.9162708520889282,
      "learning_rate": 1.854911870672947e-05,
      "loss": 0.06566942483186722,
      "step": 82,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.0378548895899053,
      "grad_norm": 1.560840368270874,
      "learning_rate": 1.849202181526579e-05,
      "loss": 0.08562697470188141,
      "step": 83,
      "token_acc": 0.953125
    },
    {
      "epoch": 1.0504731861198737,
      "grad_norm": 0.556181788444519,
      "learning_rate": 1.843391445812886e-05,
      "loss": 0.05147428810596466,
      "step": 84,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 1.0630914826498423,
      "grad_norm": 0.5080341696739197,
      "learning_rate": 1.837480354951308e-05,
      "loss": 0.053718820214271545,
      "step": 85,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 1.0757097791798107,
      "grad_norm": 1.0325117111206055,
      "learning_rate": 1.8314696123025456e-05,
      "loss": 0.07695874571800232,
      "step": 86,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.088328075709779,
      "grad_norm": 0.9591953754425049,
      "learning_rate": 1.8253599330848638e-05,
      "loss": 0.05609697848558426,
      "step": 87,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.1009463722397477,
      "grad_norm": 3.414123058319092,
      "learning_rate": 1.819152044288992e-05,
      "loss": 0.08527295291423798,
      "step": 88,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 1.113564668769716,
      "grad_norm": 4.552833557128906,
      "learning_rate": 1.8128466845916156e-05,
      "loss": 0.11498329043388367,
      "step": 89,
      "token_acc": 0.9419642857142857
    },
    {
      "epoch": 1.1261829652996846,
      "grad_norm": 0.9270873665809631,
      "learning_rate": 1.806444604267483e-05,
      "loss": 0.08343155682086945,
      "step": 90,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 1.138801261829653,
      "grad_norm": 1.5302633047103882,
      "learning_rate": 1.7999465651001297e-05,
      "loss": 0.07339704036712646,
      "step": 91,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.1514195583596214,
      "grad_norm": 0.6693511605262756,
      "learning_rate": 1.7933533402912354e-05,
      "loss": 0.05943416804075241,
      "step": 92,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 1.16403785488959,
      "grad_norm": 1.2709903717041016,
      "learning_rate": 1.786665714368617e-05,
      "loss": 0.06825555860996246,
      "step": 93,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 1.1766561514195584,
      "grad_norm": 3.1979079246520996,
      "learning_rate": 1.7798844830928818e-05,
      "loss": 0.06826388090848923,
      "step": 94,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.1892744479495267,
      "grad_norm": 1.0118235349655151,
      "learning_rate": 1.773010453362737e-05,
      "loss": 0.06091482192277908,
      "step": 95,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.2018927444794953,
      "grad_norm": 2.472378730773926,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.08082906156778336,
      "step": 96,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.2145110410094637,
      "grad_norm": 5.689515113830566,
      "learning_rate": 1.758987281247162e-05,
      "loss": 0.12219467014074326,
      "step": 97,
      "token_acc": 0.9508928571428571
    },
    {
      "epoch": 1.227129337539432,
      "grad_norm": 0.6733102202415466,
      "learning_rate": 1.7518398074789776e-05,
      "loss": 0.07546862959861755,
      "step": 98,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.2397476340694007,
      "grad_norm": 3.4482760429382324,
      "learning_rate": 1.7446028722923266e-05,
      "loss": 0.0797242522239685,
      "step": 99,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 1.252365930599369,
      "grad_norm": 1.9023994207382202,
      "learning_rate": 1.737277336810124e-05,
      "loss": 0.0745440348982811,
      "step": 100,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 1.2649842271293377,
      "grad_norm": 2.4060165882110596,
      "learning_rate": 1.7298640726978357e-05,
      "loss": 0.08716213703155518,
      "step": 101,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.277602523659306,
      "grad_norm": 0.644273579120636,
      "learning_rate": 1.7223639620597556e-05,
      "loss": 0.06642022728919983,
      "step": 102,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.2902208201892744,
      "grad_norm": 3.350196123123169,
      "learning_rate": 1.7147778973340466e-05,
      "loss": 0.08557718992233276,
      "step": 103,
      "token_acc": 0.9575892857142857
    },
    {
      "epoch": 1.3028391167192428,
      "grad_norm": 5.705350399017334,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.10517165064811707,
      "step": 104,
      "token_acc": 0.9441964285714286
    },
    {
      "epoch": 1.3154574132492114,
      "grad_norm": 3.6529672145843506,
      "learning_rate": 1.699351526403367e-05,
      "loss": 0.10772910714149475,
      "step": 105,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.3280757097791798,
      "grad_norm": 1.6319149732589722,
      "learning_rate": 1.6915130557822698e-05,
      "loss": 0.06114519387483597,
      "step": 106,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 1.3406940063091484,
      "grad_norm": 3.9356307983398438,
      "learning_rate": 1.6835923020228714e-05,
      "loss": 0.07817341387271881,
      "step": 107,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.3533123028391167,
      "grad_norm": 3.499027729034424,
      "learning_rate": 1.6755902076156606e-05,
      "loss": 0.08761963248252869,
      "step": 108,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.3659305993690851,
      "grad_norm": 2.311490535736084,
      "learning_rate": 1.6675077247298475e-05,
      "loss": 0.07218876481056213,
      "step": 109,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.3785488958990535,
      "grad_norm": 2.4191062450408936,
      "learning_rate": 1.659345815100069e-05,
      "loss": 0.0859629288315773,
      "step": 110,
      "token_acc": 0.9486607142857143
    },
    {
      "epoch": 1.391167192429022,
      "grad_norm": 1.2196522951126099,
      "learning_rate": 1.6511054499119493e-05,
      "loss": 0.07391684502363205,
      "step": 111,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.4037854889589905,
      "grad_norm": 1.7076737880706787,
      "learning_rate": 1.6427876096865394e-05,
      "loss": 0.060356542468070984,
      "step": 112,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 1.416403785488959,
      "grad_norm": 3.308213233947754,
      "learning_rate": 1.6343932841636455e-05,
      "loss": 0.09701358526945114,
      "step": 113,
      "token_acc": 0.953125
    },
    {
      "epoch": 1.4290220820189274,
      "grad_norm": 0.7297303080558777,
      "learning_rate": 1.6259234721840595e-05,
      "loss": 0.06975728273391724,
      "step": 114,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.4416403785488958,
      "grad_norm": 0.5889481902122498,
      "learning_rate": 1.6173791815707053e-05,
      "loss": 0.0819731056690216,
      "step": 115,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.4542586750788644,
      "grad_norm": 0.42490774393081665,
      "learning_rate": 1.608761429008721e-05,
      "loss": 0.07358215004205704,
      "step": 116,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.4668769716088328,
      "grad_norm": 0.7116528749465942,
      "learning_rate": 1.6000712399244813e-05,
      "loss": 0.07050474733114243,
      "step": 117,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 1.4794952681388014,
      "grad_norm": 0.9713101387023926,
      "learning_rate": 1.5913096483635827e-05,
      "loss": 0.06869879364967346,
      "step": 118,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.4921135646687698,
      "grad_norm": 0.8966596722602844,
      "learning_rate": 1.5824776968678024e-05,
      "loss": 0.06647786498069763,
      "step": 119,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.5047318611987381,
      "grad_norm": 1.447534203529358,
      "learning_rate": 1.573576436351046e-05,
      "loss": 0.07275252044200897,
      "step": 120,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.5173501577287065,
      "grad_norm": 2.7334752082824707,
      "learning_rate": 1.5646069259743007e-05,
      "loss": 0.07563081383705139,
      "step": 121,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.5299684542586751,
      "grad_norm": 0.4903987646102905,
      "learning_rate": 1.5555702330196024e-05,
      "loss": 0.05079253017902374,
      "step": 122,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 1.5425867507886435,
      "grad_norm": 2.213925838470459,
      "learning_rate": 1.5464674327630437e-05,
      "loss": 0.05332695692777634,
      "step": 123,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 1.555205047318612,
      "grad_norm": 1.2029969692230225,
      "learning_rate": 1.5372996083468242e-05,
      "loss": 0.06604963541030884,
      "step": 124,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.5678233438485805,
      "grad_norm": 1.1939482688903809,
      "learning_rate": 1.528067850650368e-05,
      "loss": 0.055782776325941086,
      "step": 125,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 1.5804416403785488,
      "grad_norm": 2.086958885192871,
      "learning_rate": 1.5187732581605217e-05,
      "loss": 0.07865685224533081,
      "step": 126,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.5930599369085172,
      "grad_norm": 0.9862927198410034,
      "learning_rate": 1.509416936840842e-05,
      "loss": 0.05956559255719185,
      "step": 127,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 1.6056782334384858,
      "grad_norm": 1.269698977470398,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.06963305175304413,
      "step": 128,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.6182965299684544,
      "grad_norm": 3.6143798828125,
      "learning_rate": 1.4905235681593079e-05,
      "loss": 0.08450886607170105,
      "step": 129,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.6309148264984228,
      "grad_norm": 0.8838794827461243,
      "learning_rate": 1.4809887689193878e-05,
      "loss": 0.06734046339988708,
      "step": 130,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.6435331230283912,
      "grad_norm": 0.6263232231140137,
      "learning_rate": 1.4713967368259981e-05,
      "loss": 0.0747184008359909,
      "step": 131,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.6561514195583595,
      "grad_norm": 0.47429347038269043,
      "learning_rate": 1.4617486132350343e-05,
      "loss": 0.07773523777723312,
      "step": 132,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.668769716088328,
      "grad_norm": 0.43498721718788147,
      "learning_rate": 1.45204554617672e-05,
      "loss": 0.06929275393486023,
      "step": 133,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.6813880126182965,
      "grad_norm": 1.1829330921173096,
      "learning_rate": 1.4422886902190014e-05,
      "loss": 0.0667298436164856,
      "step": 134,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.694006309148265,
      "grad_norm": 1.0130823850631714,
      "learning_rate": 1.4324792063301662e-05,
      "loss": 0.07056636363267899,
      "step": 135,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.7066246056782335,
      "grad_norm": 1.8610316514968872,
      "learning_rate": 1.4226182617406996e-05,
      "loss": 0.05995418131351471,
      "step": 136,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 1.7192429022082019,
      "grad_norm": 0.8720536828041077,
      "learning_rate": 1.4127070298043949e-05,
      "loss": 0.06970664858818054,
      "step": 137,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.7318611987381702,
      "grad_norm": 1.7892141342163086,
      "learning_rate": 1.4027466898587375e-05,
      "loss": 0.05199646204710007,
      "step": 138,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 1.7444794952681388,
      "grad_norm": 0.7550920844078064,
      "learning_rate": 1.3927384270845744e-05,
      "loss": 0.05898646265268326,
      "step": 139,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.7570977917981072,
      "grad_norm": 1.0391416549682617,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.07397307455539703,
      "step": 140,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.7697160883280758,
      "grad_norm": 1.2775495052337646,
      "learning_rate": 1.372582902144103e-05,
      "loss": 0.08825156092643738,
      "step": 141,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 1.7823343848580442,
      "grad_norm": 1.335414171218872,
      "learning_rate": 1.3624380382837017e-05,
      "loss": 0.06173168122768402,
      "step": 142,
      "token_acc": 0.9620535714285714
    },
    {
      "epoch": 1.7949526813880126,
      "grad_norm": 1.311166763305664,
      "learning_rate": 1.3522500479212337e-05,
      "loss": 0.06561106443405151,
      "step": 143,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.807570977917981,
      "grad_norm": 3.551767349243164,
      "learning_rate": 1.342020143325669e-05,
      "loss": 0.08286254853010178,
      "step": 144,
      "token_acc": 0.953125
    },
    {
      "epoch": 1.8201892744479495,
      "grad_norm": 4.463508605957031,
      "learning_rate": 1.3317495417533523e-05,
      "loss": 0.08482252806425095,
      "step": 145,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.8328075709779181,
      "grad_norm": 3.3428237438201904,
      "learning_rate": 1.3214394653031616e-05,
      "loss": 0.08433826267719269,
      "step": 146,
      "token_acc": 0.9598214285714286
    },
    {
      "epoch": 1.8454258675078865,
      "grad_norm": 2.9852447509765625,
      "learning_rate": 1.3110911407710909e-05,
      "loss": 0.09345546364784241,
      "step": 147,
      "token_acc": 0.953125
    },
    {
      "epoch": 1.8580441640378549,
      "grad_norm": 1.218334674835205,
      "learning_rate": 1.300705799504273e-05,
      "loss": 0.0507894903421402,
      "step": 148,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 1.8706624605678233,
      "grad_norm": 2.8220064640045166,
      "learning_rate": 1.2902846772544625e-05,
      "loss": 0.06830944865942001,
      "step": 149,
      "token_acc": 0.96875
    },
    {
      "epoch": 1.8832807570977916,
      "grad_norm": 4.01509428024292,
      "learning_rate": 1.2798290140309924e-05,
      "loss": 0.08686386048793793,
      "step": 150,
      "token_acc": 0.9508928571428571
    },
    {
      "epoch": 1.8958990536277602,
      "grad_norm": 3.739550828933716,
      "learning_rate": 1.2693400539532263e-05,
      "loss": 0.08545972406864166,
      "step": 151,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.9085173501577288,
      "grad_norm": 0.47104185819625854,
      "learning_rate": 1.2588190451025209e-05,
      "loss": 0.060449764132499695,
      "step": 152,
      "token_acc": 0.9709821428571429
    },
    {
      "epoch": 1.9211356466876972,
      "grad_norm": 2.637552499771118,
      "learning_rate": 1.2482672393737164e-05,
      "loss": 0.0789261907339096,
      "step": 153,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.9337539432176656,
      "grad_norm": 0.9178563952445984,
      "learning_rate": 1.2376858923261732e-05,
      "loss": 0.06671890616416931,
      "step": 154,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 1.946372239747634,
      "grad_norm": 1.6129941940307617,
      "learning_rate": 1.2270762630343734e-05,
      "loss": 0.06917385756969452,
      "step": 155,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 1.9589905362776026,
      "grad_norm": 1.5085026025772095,
      "learning_rate": 1.2164396139381029e-05,
      "loss": 0.08500629663467407,
      "step": 156,
      "token_acc": 0.9553571428571429
    },
    {
      "epoch": 1.971608832807571,
      "grad_norm": 1.3954477310180664,
      "learning_rate": 1.205777210692235e-05,
      "loss": 0.06161624193191528,
      "step": 157,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 1.9842271293375395,
      "grad_norm": 0.9118571281433105,
      "learning_rate": 1.1950903220161286e-05,
      "loss": 0.0707247406244278,
      "step": 158,
      "token_acc": 0.9665178571428571
    },
    {
      "epoch": 1.996845425867508,
      "grad_norm": 2.032381772994995,
      "learning_rate": 1.1843802195426634e-05,
      "loss": 0.04953468590974808,
      "step": 159,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.835229754447937,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.044536832720041275,
      "step": 160,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.0126182965299684,
      "grad_norm": 0.6722914576530457,
      "learning_rate": 1.162895473394589e-05,
      "loss": 0.03763537108898163,
      "step": 161,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.0252365930599368,
      "grad_norm": 0.5252609252929688,
      "learning_rate": 1.1521233861899168e-05,
      "loss": 0.03970091789960861,
      "step": 162,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.0378548895899056,
      "grad_norm": 0.5142501592636108,
      "learning_rate": 1.1413331978235677e-05,
      "loss": 0.03096415288746357,
      "step": 163,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.050473186119874,
      "grad_norm": 0.5961836576461792,
      "learning_rate": 1.130526192220052e-05,
      "loss": 0.028163593262434006,
      "step": 164,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.0630914826498423,
      "grad_norm": 1.2440811395645142,
      "learning_rate": 1.1197036553049626e-05,
      "loss": 0.06024373322725296,
      "step": 165,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 2.0757097791798107,
      "grad_norm": 0.5177900195121765,
      "learning_rate": 1.1088668748519646e-05,
      "loss": 0.022872207686305046,
      "step": 166,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 2.088328075709779,
      "grad_norm": 2.10178804397583,
      "learning_rate": 1.098017140329561e-05,
      "loss": 0.058632992208004,
      "step": 167,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.1009463722397475,
      "grad_norm": 1.7796574831008911,
      "learning_rate": 1.0871557427476585e-05,
      "loss": 0.026033399626612663,
      "step": 168,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.1135646687697163,
      "grad_norm": 0.8075276017189026,
      "learning_rate": 1.0762839745039526e-05,
      "loss": 0.017725663259625435,
      "step": 169,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.1261829652996846,
      "grad_norm": 1.7707884311676025,
      "learning_rate": 1.0654031292301432e-05,
      "loss": 0.049549028277397156,
      "step": 170,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.138801261829653,
      "grad_norm": 1.9634188413619995,
      "learning_rate": 1.0545145016380065e-05,
      "loss": 0.043986767530441284,
      "step": 171,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.1514195583596214,
      "grad_norm": 0.9482637047767639,
      "learning_rate": 1.0436193873653362e-05,
      "loss": 0.04435993731021881,
      "step": 172,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.1640378548895898,
      "grad_norm": 0.9664008021354675,
      "learning_rate": 1.0327190828217763e-05,
      "loss": 0.03148264437913895,
      "step": 173,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.176656151419558,
      "grad_norm": 0.9135804176330566,
      "learning_rate": 1.0218148850345613e-05,
      "loss": 0.024964461103081703,
      "step": 174,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.189274447949527,
      "grad_norm": 1.4149248600006104,
      "learning_rate": 1.0109080914941825e-05,
      "loss": 0.0473172701895237,
      "step": 175,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 2.2018927444794953,
      "grad_norm": 1.1210252046585083,
      "learning_rate": 1e-05,
      "loss": 0.05981011316180229,
      "step": 176,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.2145110410094637,
      "grad_norm": 0.8092201352119446,
      "learning_rate": 9.890919085058179e-06,
      "loss": 0.028452713042497635,
      "step": 177,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.227129337539432,
      "grad_norm": 1.2984387874603271,
      "learning_rate": 9.78185114965439e-06,
      "loss": 0.04357399046421051,
      "step": 178,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.2397476340694005,
      "grad_norm": 1.0969109535217285,
      "learning_rate": 9.67280917178224e-06,
      "loss": 0.04148834943771362,
      "step": 179,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.2523659305993693,
      "grad_norm": 0.9426726698875427,
      "learning_rate": 9.563806126346643e-06,
      "loss": 0.05436920374631882,
      "step": 180,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 2.2649842271293377,
      "grad_norm": 1.7621617317199707,
      "learning_rate": 9.454854983619936e-06,
      "loss": 0.03881138935685158,
      "step": 181,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.277602523659306,
      "grad_norm": 1.1210284233093262,
      "learning_rate": 9.34596870769857e-06,
      "loss": 0.03353865444660187,
      "step": 182,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.2902208201892744,
      "grad_norm": 1.9880902767181396,
      "learning_rate": 9.237160254960477e-06,
      "loss": 0.04133755341172218,
      "step": 183,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.302839116719243,
      "grad_norm": 1.0943409204483032,
      "learning_rate": 9.128442572523418e-06,
      "loss": 0.03071509674191475,
      "step": 184,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.315457413249211,
      "grad_norm": 2.113119125366211,
      "learning_rate": 9.019828596704394e-06,
      "loss": 0.03822978585958481,
      "step": 185,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.32807570977918,
      "grad_norm": 0.7326868176460266,
      "learning_rate": 8.911331251480357e-06,
      "loss": 0.03254137933254242,
      "step": 186,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 2.3406940063091484,
      "grad_norm": 0.802615761756897,
      "learning_rate": 8.802963446950378e-06,
      "loss": 0.030572062358260155,
      "step": 187,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.3533123028391167,
      "grad_norm": 2.769240617752075,
      "learning_rate": 8.694738077799487e-06,
      "loss": 0.05989953875541687,
      "step": 188,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.365930599369085,
      "grad_norm": 1.2988790273666382,
      "learning_rate": 8.586668021764328e-06,
      "loss": 0.03389206528663635,
      "step": 189,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.3785488958990535,
      "grad_norm": 1.8916139602661133,
      "learning_rate": 8.478766138100834e-06,
      "loss": 0.05102337524294853,
      "step": 190,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.3911671924290223,
      "grad_norm": 1.2074483633041382,
      "learning_rate": 8.371045266054114e-06,
      "loss": 0.04032492637634277,
      "step": 191,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.4037854889589907,
      "grad_norm": 0.7658787965774536,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.034190282225608826,
      "step": 192,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.416403785488959,
      "grad_norm": 1.7592679262161255,
      "learning_rate": 8.156197804573368e-06,
      "loss": 0.05537520349025726,
      "step": 193,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 2.4290220820189274,
      "grad_norm": 1.5274964570999146,
      "learning_rate": 8.04909677983872e-06,
      "loss": 0.022157451137900352,
      "step": 194,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.441640378548896,
      "grad_norm": 2.353959321975708,
      "learning_rate": 7.942227893077652e-06,
      "loss": 0.04223260655999184,
      "step": 195,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.454258675078864,
      "grad_norm": 0.7406657338142395,
      "learning_rate": 7.835603860618973e-06,
      "loss": 0.03615238517522812,
      "step": 196,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.4668769716088326,
      "grad_norm": 0.9043768644332886,
      "learning_rate": 7.72923736965627e-06,
      "loss": 0.06482378393411636,
      "step": 197,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 2.4794952681388014,
      "grad_norm": 2.2845795154571533,
      "learning_rate": 7.623141076738271e-06,
      "loss": 0.048506688326597214,
      "step": 198,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 2.4921135646687698,
      "grad_norm": 1.8952425718307495,
      "learning_rate": 7.5173276062628364e-06,
      "loss": 0.08035582304000854,
      "step": 199,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 2.504731861198738,
      "grad_norm": 0.7781740427017212,
      "learning_rate": 7.411809548974792e-06,
      "loss": 0.05676484853029251,
      "step": 200,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.5173501577287065,
      "grad_norm": 0.8096243739128113,
      "learning_rate": 7.306599460467741e-06,
      "loss": 0.03071010857820511,
      "step": 201,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.5299684542586753,
      "grad_norm": 1.261987566947937,
      "learning_rate": 7.201709859690081e-06,
      "loss": 0.03445509076118469,
      "step": 202,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.5425867507886437,
      "grad_norm": 1.141879916191101,
      "learning_rate": 7.097153227455379e-06,
      "loss": 0.035321786999702454,
      "step": 203,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.555205047318612,
      "grad_norm": 0.7148231863975525,
      "learning_rate": 6.992942004957271e-06,
      "loss": 0.033092375844717026,
      "step": 204,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.5678233438485805,
      "grad_norm": 1.4082436561584473,
      "learning_rate": 6.889088592289092e-06,
      "loss": 0.030955083668231964,
      "step": 205,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.580441640378549,
      "grad_norm": 0.6685624122619629,
      "learning_rate": 6.785605346968387e-06,
      "loss": 0.038570836186409,
      "step": 206,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.593059936908517,
      "grad_norm": 0.7129871249198914,
      "learning_rate": 6.682504582466482e-06,
      "loss": 0.03919385373592377,
      "step": 207,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.6056782334384856,
      "grad_norm": 0.6126706004142761,
      "learning_rate": 6.579798566743314e-06,
      "loss": 0.03597944229841232,
      "step": 208,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.6182965299684544,
      "grad_norm": 0.6538820266723633,
      "learning_rate": 6.4774995207876654e-06,
      "loss": 0.030326727777719498,
      "step": 209,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.630914826498423,
      "grad_norm": 1.1558011770248413,
      "learning_rate": 6.375619617162985e-06,
      "loss": 0.035182103514671326,
      "step": 210,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.643533123028391,
      "grad_norm": 0.8294020295143127,
      "learning_rate": 6.274170978558971e-06,
      "loss": 0.038366854190826416,
      "step": 211,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.6561514195583595,
      "grad_norm": 1.4801054000854492,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.03767174482345581,
      "step": 212,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.668769716088328,
      "grad_norm": 1.1237835884094238,
      "learning_rate": 6.072615729154261e-06,
      "loss": 0.03061925247311592,
      "step": 213,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 2.6813880126182967,
      "grad_norm": 0.7757349610328674,
      "learning_rate": 5.97253310141263e-06,
      "loss": 0.021479658782482147,
      "step": 214,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 2.694006309148265,
      "grad_norm": 0.7453773021697998,
      "learning_rate": 5.872929701956054e-06,
      "loss": 0.04876632243394852,
      "step": 215,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.7066246056782335,
      "grad_norm": 1.8686482906341553,
      "learning_rate": 5.773817382593008e-06,
      "loss": 0.04855792224407196,
      "step": 216,
      "token_acc": 0.9732142857142857
    },
    {
      "epoch": 2.719242902208202,
      "grad_norm": 1.142004370689392,
      "learning_rate": 5.675207936698337e-06,
      "loss": 0.04441647231578827,
      "step": 217,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 2.7318611987381702,
      "grad_norm": 1.2509138584136963,
      "learning_rate": 5.5771130978099896e-06,
      "loss": 0.03287196531891823,
      "step": 218,
      "token_acc": 0.9866071428571429
    },
    {
      "epoch": 2.7444794952681386,
      "grad_norm": 1.553643822669983,
      "learning_rate": 5.479544538232804e-06,
      "loss": 0.0430506095290184,
      "step": 219,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.757097791798107,
      "grad_norm": 0.924835741519928,
      "learning_rate": 5.382513867649663e-06,
      "loss": 0.027985451743006706,
      "step": 220,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 2.769716088328076,
      "grad_norm": 1.2851096391677856,
      "learning_rate": 5.286032631740023e-06,
      "loss": 0.02770395204424858,
      "step": 221,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.782334384858044,
      "grad_norm": 0.810985803604126,
      "learning_rate": 5.190112310806126e-06,
      "loss": 0.047174058854579926,
      "step": 222,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.7949526813880126,
      "grad_norm": 1.2850825786590576,
      "learning_rate": 5.094764318406921e-06,
      "loss": 0.0360386036336422,
      "step": 223,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.807570977917981,
      "grad_norm": 1.4199855327606201,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.044144198298454285,
      "step": 224,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.8201892744479498,
      "grad_norm": 0.5397220849990845,
      "learning_rate": 4.9058306315915826e-06,
      "loss": 0.01462258119136095,
      "step": 225,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 2.832807570977918,
      "grad_norm": 0.8922738432884216,
      "learning_rate": 4.812267418394784e-06,
      "loss": 0.053033940494060516,
      "step": 226,
      "token_acc": 0.9776785714285714
    },
    {
      "epoch": 2.8454258675078865,
      "grad_norm": 1.2049716711044312,
      "learning_rate": 4.7193214934963204e-06,
      "loss": 0.0357033908367157,
      "step": 227,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.858044164037855,
      "grad_norm": 1.461714506149292,
      "learning_rate": 4.627003916531761e-06,
      "loss": 0.046947937458753586,
      "step": 228,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.8706624605678233,
      "grad_norm": 0.5750783681869507,
      "learning_rate": 4.535325672369567e-06,
      "loss": 0.02531631849706173,
      "step": 229,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.8832807570977916,
      "grad_norm": 1.3903998136520386,
      "learning_rate": 4.444297669803981e-06,
      "loss": 0.027292776852846146,
      "step": 230,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.89589905362776,
      "grad_norm": 0.9362890720367432,
      "learning_rate": 4.353930740256997e-06,
      "loss": 0.05386661738157272,
      "step": 231,
      "token_acc": 0.9754464285714286
    },
    {
      "epoch": 2.908517350157729,
      "grad_norm": 1.5002843141555786,
      "learning_rate": 4.264235636489542e-06,
      "loss": 0.06073082610964775,
      "step": 232,
      "token_acc": 0.9799107142857143
    },
    {
      "epoch": 2.921135646687697,
      "grad_norm": 1.4437495470046997,
      "learning_rate": 4.17522303132198e-06,
      "loss": 0.038563214242458344,
      "step": 233,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.9337539432176656,
      "grad_norm": 1.0687687397003174,
      "learning_rate": 4.086903516364179e-06,
      "loss": 0.027859320864081383,
      "step": 234,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.946372239747634,
      "grad_norm": 0.667089581489563,
      "learning_rate": 3.999287600755192e-06,
      "loss": 0.03428979963064194,
      "step": 235,
      "token_acc": 0.9821428571428571
    },
    {
      "epoch": 2.958990536277603,
      "grad_norm": 0.9890245199203491,
      "learning_rate": 3.912385709912794e-06,
      "loss": 0.032398052513599396,
      "step": 236,
      "token_acc": 0.984375
    },
    {
      "epoch": 2.971608832807571,
      "grad_norm": 1.061720609664917,
      "learning_rate": 3.826208184292952e-06,
      "loss": 0.02387729287147522,
      "step": 237,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 2.9842271293375395,
      "grad_norm": 0.9489039778709412,
      "learning_rate": 3.7407652781594094e-06,
      "loss": 0.030074123293161392,
      "step": 238,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 2.996845425867508,
      "grad_norm": 0.9017006158828735,
      "learning_rate": 3.6560671583635467e-06,
      "loss": 0.02967180125415325,
      "step": 239,
      "token_acc": 0.984375
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.018955707550049,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 0.08006613701581955,
      "step": 240,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 3.0126182965299684,
      "grad_norm": 0.7195019125938416,
      "learning_rate": 3.4889455008805107e-06,
      "loss": 0.017341189086437225,
      "step": 241,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.0252365930599368,
      "grad_norm": 1.1500155925750732,
      "learning_rate": 3.4065418489993118e-06,
      "loss": 0.019664622843265533,
      "step": 242,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 3.0378548895899056,
      "grad_norm": 0.5775253176689148,
      "learning_rate": 3.324922752701528e-06,
      "loss": 0.016728028655052185,
      "step": 243,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 3.050473186119874,
      "grad_norm": 0.23594523966312408,
      "learning_rate": 3.2440979238433977e-06,
      "loss": 0.00863596424460411,
      "step": 244,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0630914826498423,
      "grad_norm": 0.6648944020271301,
      "learning_rate": 3.1640769797712865e-06,
      "loss": 0.008409049361944199,
      "step": 245,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.0757097791798107,
      "grad_norm": 0.5050109028816223,
      "learning_rate": 3.0848694421773075e-06,
      "loss": 0.016843078657984734,
      "step": 246,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.088328075709779,
      "grad_norm": 0.8173918724060059,
      "learning_rate": 3.0064847359663284e-06,
      "loss": 0.019274180755019188,
      "step": 247,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.1009463722397475,
      "grad_norm": 0.6629483699798584,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.013546002097427845,
      "step": 248,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.1135646687697163,
      "grad_norm": 0.43275824189186096,
      "learning_rate": 2.8522210266595386e-06,
      "loss": 0.019802484661340714,
      "step": 249,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 3.1261829652996846,
      "grad_norm": 0.3252338767051697,
      "learning_rate": 2.776360379402445e-06,
      "loss": 0.007681960240006447,
      "step": 250,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.138801261829653,
      "grad_norm": 0.8416299223899841,
      "learning_rate": 2.7013592730216464e-06,
      "loss": 0.03177838400006294,
      "step": 251,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 3.1514195583596214,
      "grad_norm": 0.5707000494003296,
      "learning_rate": 2.6272266318987606e-06,
      "loss": 0.006279752589762211,
      "step": 252,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1640378548895898,
      "grad_norm": 0.5500001311302185,
      "learning_rate": 2.5539712770767377e-06,
      "loss": 0.0063630277290940285,
      "step": 253,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.176656151419558,
      "grad_norm": 0.5682907700538635,
      "learning_rate": 2.4816019252102274e-06,
      "loss": 0.010782742872834206,
      "step": 254,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.189274447949527,
      "grad_norm": 0.4956236481666565,
      "learning_rate": 2.4101271875283818e-06,
      "loss": 0.01635258086025715,
      "step": 255,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.2018927444794953,
      "grad_norm": 0.4443606436252594,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.004104578867554665,
      "step": 256,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2145110410094637,
      "grad_norm": 1.1049095392227173,
      "learning_rate": 2.26989546637263e-06,
      "loss": 0.018124593421816826,
      "step": 257,
      "token_acc": 0.9888392857142857
    },
    {
      "epoch": 3.227129337539432,
      "grad_norm": 0.4260806143283844,
      "learning_rate": 2.201155169071184e-06,
      "loss": 0.00837795715779066,
      "step": 258,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.2397476340694005,
      "grad_norm": 0.807509183883667,
      "learning_rate": 2.1333428563138304e-06,
      "loss": 0.013505063951015472,
      "step": 259,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.2523659305993693,
      "grad_norm": 0.37788689136505127,
      "learning_rate": 2.0664665970876496e-06,
      "loss": 0.004273224622011185,
      "step": 260,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.2649842271293377,
      "grad_norm": 0.4900219142436981,
      "learning_rate": 2.0005343489987038e-06,
      "loss": 0.008484878577291965,
      "step": 261,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.277602523659306,
      "grad_norm": 0.7600153684616089,
      "learning_rate": 1.9355539573251737e-06,
      "loss": 0.021974962204694748,
      "step": 262,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.2902208201892744,
      "grad_norm": 0.6423557996749878,
      "learning_rate": 1.8715331540838488e-06,
      "loss": 0.011579688638448715,
      "step": 263,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.302839116719243,
      "grad_norm": 0.5795581340789795,
      "learning_rate": 1.808479557110081e-06,
      "loss": 0.005041406489908695,
      "step": 264,
      "token_acc": 1.0
    },
    {
      "epoch": 3.315457413249211,
      "grad_norm": 0.6128451824188232,
      "learning_rate": 1.7464006691513624e-06,
      "loss": 0.02049056440591812,
      "step": 265,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.32807570977918,
      "grad_norm": 0.4404939115047455,
      "learning_rate": 1.6853038769745466e-06,
      "loss": 0.004994688555598259,
      "step": 266,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3406940063091484,
      "grad_norm": 0.49892276525497437,
      "learning_rate": 1.6251964504869221e-06,
      "loss": 0.006245453841984272,
      "step": 267,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.3533123028391167,
      "grad_norm": 0.26739901304244995,
      "learning_rate": 1.566085541871145e-06,
      "loss": 0.004012120887637138,
      "step": 268,
      "token_acc": 1.0
    },
    {
      "epoch": 3.365930599369085,
      "grad_norm": 0.665448009967804,
      "learning_rate": 1.5079781847342122e-06,
      "loss": 0.013718968257308006,
      "step": 269,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.3785488958990535,
      "grad_norm": 0.550457239151001,
      "learning_rate": 1.4508812932705364e-06,
      "loss": 0.00480533204972744,
      "step": 270,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.3911671924290223,
      "grad_norm": 0.23645147681236267,
      "learning_rate": 1.3948016614392113e-06,
      "loss": 0.003103341441601515,
      "step": 271,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4037854889589907,
      "grad_norm": 0.7343894839286804,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.006245617754757404,
      "step": 272,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.416403785488959,
      "grad_norm": 1.0396212339401245,
      "learning_rate": 1.2857207464973876e-06,
      "loss": 0.021057892590761185,
      "step": 273,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 3.4290220820189274,
      "grad_norm": 0.7940521836280823,
      "learning_rate": 1.2327324429249232e-06,
      "loss": 0.0124064851552248,
      "step": 274,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.441640378548896,
      "grad_norm": 0.5233060717582703,
      "learning_rate": 1.1807873565164507e-06,
      "loss": 0.006917751859873533,
      "step": 275,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.454258675078864,
      "grad_norm": 0.6649695038795471,
      "learning_rate": 1.129891668217783e-06,
      "loss": 0.017966866493225098,
      "step": 276,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.4668769716088326,
      "grad_norm": 0.8949275612831116,
      "learning_rate": 1.0800514341068592e-06,
      "loss": 0.01210831105709076,
      "step": 277,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.4794952681388014,
      "grad_norm": 0.2179488092660904,
      "learning_rate": 1.0312725846731174e-06,
      "loss": 0.0016820882447063923,
      "step": 278,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4921135646687698,
      "grad_norm": 0.3757546544075012,
      "learning_rate": 9.835609241118404e-07,
      "loss": 0.002328360453248024,
      "step": 279,
      "token_acc": 1.0
    },
    {
      "epoch": 3.504731861198738,
      "grad_norm": 0.7376626133918762,
      "learning_rate": 9.369221296335007e-07,
      "loss": 0.005898178089410067,
      "step": 280,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.5173501577287065,
      "grad_norm": 0.5437489748001099,
      "learning_rate": 8.91361750788241e-07,
      "loss": 0.007657959125936031,
      "step": 281,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.5299684542586753,
      "grad_norm": 0.9899237155914307,
      "learning_rate": 8.468852088055291e-07,
      "loss": 0.016547206789255142,
      "step": 282,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.5425867507886437,
      "grad_norm": 0.793226420879364,
      "learning_rate": 8.034977959490775e-07,
      "loss": 0.010517780669033527,
      "step": 283,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.555205047318612,
      "grad_norm": 0.7523264288902283,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.009029757231473923,
      "step": 284,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.5678233438485805,
      "grad_norm": 0.25030848383903503,
      "learning_rate": 7.200108780781556e-07,
      "loss": 0.002805521246045828,
      "step": 285,
      "token_acc": 1.0
    },
    {
      "epoch": 3.580441640378549,
      "grad_norm": 0.7225795984268188,
      "learning_rate": 6.799213071720156e-07,
      "loss": 0.016245253384113312,
      "step": 286,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.593059936908517,
      "grad_norm": 0.7905085682868958,
      "learning_rate": 6.409407324267448e-07,
      "loss": 0.009255263954401016,
      "step": 287,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.6056782334384856,
      "grad_norm": 0.3408595323562622,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.0023827371187508106,
      "step": 288,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6182965299684544,
      "grad_norm": 0.6070659160614014,
      "learning_rate": 5.663249921017477e-07,
      "loss": 0.020646952092647552,
      "step": 289,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.630914826498423,
      "grad_norm": 0.5172240138053894,
      "learning_rate": 5.306987050489442e-07,
      "loss": 0.01163824275135994,
      "step": 290,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.643533123028391,
      "grad_norm": 0.6110843420028687,
      "learning_rate": 4.961991701543889e-07,
      "loss": 0.0061739785596728325,
      "step": 291,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.6561514195583595,
      "grad_norm": 1.2368091344833374,
      "learning_rate": 4.628304925177318e-07,
      "loss": 0.02270001359283924,
      "step": 292,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.668769716088328,
      "grad_norm": 1.2081069946289062,
      "learning_rate": 4.305966426779118e-07,
      "loss": 0.013647607527673244,
      "step": 293,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.6813880126182967,
      "grad_norm": 0.3540801405906677,
      "learning_rate": 3.99501456140714e-07,
      "loss": 0.0058923084288835526,
      "step": 294,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.694006309148265,
      "grad_norm": 0.48793691396713257,
      "learning_rate": 3.6954863292237297e-07,
      "loss": 0.01068500429391861,
      "step": 295,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.7066246056782335,
      "grad_norm": 1.2921711206436157,
      "learning_rate": 3.4074173710931804e-07,
      "loss": 0.026274289935827255,
      "step": 296,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.719242902208202,
      "grad_norm": 0.545202374458313,
      "learning_rate": 3.1308419643406915e-07,
      "loss": 0.0053621018305420876,
      "step": 297,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.7318611987381702,
      "grad_norm": 0.4888966381549835,
      "learning_rate": 2.865793018673857e-07,
      "loss": 0.007355076260864735,
      "step": 298,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.7444794952681386,
      "grad_norm": 0.7358994483947754,
      "learning_rate": 2.612302072266637e-07,
      "loss": 0.006454679183661938,
      "step": 299,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.757097791798107,
      "grad_norm": 1.312822937965393,
      "learning_rate": 2.370399288006664e-07,
      "loss": 0.010885766707360744,
      "step": 300,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.769716088328076,
      "grad_norm": 1.0236682891845703,
      "learning_rate": 2.140113449906167e-07,
      "loss": 0.018271073698997498,
      "step": 301,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 3.782334384858044,
      "grad_norm": 0.4519929587841034,
      "learning_rate": 1.921471959676957e-07,
      "loss": 0.008333021774888039,
      "step": 302,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.7949526813880126,
      "grad_norm": 1.1988343000411987,
      "learning_rate": 1.7145008334698898e-07,
      "loss": 0.01078954990953207,
      "step": 303,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.807570977917981,
      "grad_norm": 0.47362643480300903,
      "learning_rate": 1.519224698779198e-07,
      "loss": 0.004242708906531334,
      "step": 304,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.8201892744479498,
      "grad_norm": 1.592261791229248,
      "learning_rate": 1.3356667915121025e-07,
      "loss": 0.01740104705095291,
      "step": 305,
      "token_acc": 0.9910714285714286
    },
    {
      "epoch": 3.832807570977918,
      "grad_norm": 0.8455437421798706,
      "learning_rate": 1.1638489532239339e-07,
      "loss": 0.009887732565402985,
      "step": 306,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.8454258675078865,
      "grad_norm": 0.6133864521980286,
      "learning_rate": 1.0037916285192129e-07,
      "loss": 0.018338723108172417,
      "step": 307,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.858044164037855,
      "grad_norm": 0.6251723170280457,
      "learning_rate": 8.555138626189619e-08,
      "loss": 0.01048330869525671,
      "step": 308,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.8706624605678233,
      "grad_norm": 0.6142211556434631,
      "learning_rate": 7.19033299094496e-08,
      "loss": 0.005674176849424839,
      "step": 309,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.8832807570977916,
      "grad_norm": 0.8769956231117249,
      "learning_rate": 5.943661777680354e-08,
      "loss": 0.018422063440084457,
      "step": 310,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.89589905362776,
      "grad_norm": 0.9772692322731018,
      "learning_rate": 4.815273327803183e-08,
      "loss": 0.018329620361328125,
      "step": 311,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.908517350157729,
      "grad_norm": 0.7435838580131531,
      "learning_rate": 3.805301908254455e-08,
      "loss": 0.014172300696372986,
      "step": 312,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.921135646687697,
      "grad_norm": 0.6025273203849792,
      "learning_rate": 2.9138676955333676e-08,
      "loss": 0.006965091452002525,
      "step": 313,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 3.9337539432176656,
      "grad_norm": 0.5574731826782227,
      "learning_rate": 2.1410767613965212e-08,
      "loss": 0.00548956636339426,
      "step": 314,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.946372239747634,
      "grad_norm": 0.29842305183410645,
      "learning_rate": 1.487021060236904e-08,
      "loss": 0.005129786673933268,
      "step": 315,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.958990536277603,
      "grad_norm": 0.8428702354431152,
      "learning_rate": 9.517784181422018e-09,
      "loss": 0.011684278026223183,
      "step": 316,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.971608832807571,
      "grad_norm": 1.1606839895248413,
      "learning_rate": 5.354125236343155e-09,
      "loss": 0.014027003198862076,
      "step": 317,
      "token_acc": 0.9933035714285714
    },
    {
      "epoch": 3.9842271293375395,
      "grad_norm": 0.5172231197357178,
      "learning_rate": 2.379729200908676e-09,
      "loss": 0.010849256068468094,
      "step": 318,
      "token_acc": 0.9977678571428571
    },
    {
      "epoch": 3.996845425867508,
      "grad_norm": 0.9050202965736389,
      "learning_rate": 5.949499985025142e-10,
      "loss": 0.011914867907762527,
      "step": 319,
      "token_acc": 0.9955357142857143
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.7990683317184448,
      "learning_rate": 0.0,
      "loss": 0.02128974348306656,
      "step": 320,
      "token_acc": 0.9910714285714286
    }
  ],
  "logging_steps": 1,
  "max_steps": 320,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7682419089408.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}