{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.33531260915123995,
  "eval_steps": 500,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.7522330163029802,
      "learning_rate": 7.142857142857143e-06,
      "loss": 1.3831,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7473994784323043,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 1.3916,
      "step": 2
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.70544460641196,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 1.4248,
      "step": 3
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5834904606723813,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.4045,
      "step": 4
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8295213706761098,
      "learning_rate": 3.571428571428572e-05,
      "loss": 1.3918,
      "step": 5
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8133546220957368,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 1.3865,
      "step": 6
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7903446421142379,
      "learning_rate": 5e-05,
      "loss": 1.373,
      "step": 7
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6811759259942566,
      "learning_rate": 5.714285714285714e-05,
      "loss": 1.3682,
      "step": 8
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6292547023357415,
      "learning_rate": 6.428571428571429e-05,
      "loss": 1.3916,
      "step": 9
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5931255904191401,
      "learning_rate": 7.142857142857143e-05,
      "loss": 1.3833,
      "step": 10
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6447974541785289,
      "learning_rate": 7.857142857142858e-05,
      "loss": 1.4023,
      "step": 11
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5941331466143945,
      "learning_rate": 8.571428571428571e-05,
      "loss": 1.3423,
      "step": 12
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6603903337024011,
      "learning_rate": 9.285714285714286e-05,
      "loss": 1.3762,
      "step": 13
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5633045289970319,
      "learning_rate": 0.0001,
      "loss": 1.3843,
      "step": 14
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5277710229216985,
      "learning_rate": 9.999868398131282e-05,
      "loss": 1.3872,
      "step": 15
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5589957310018021,
      "learning_rate": 9.999473599452745e-05,
      "loss": 1.3452,
      "step": 16
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5000831179822082,
      "learning_rate": 9.99881562474689e-05,
      "loss": 1.3813,
      "step": 17
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5290519156012868,
      "learning_rate": 9.997894508649996e-05,
      "loss": 1.3762,
      "step": 18
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.519496248745277,
      "learning_rate": 9.996710299650301e-05,
      "loss": 1.3706,
      "step": 19
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47116309957086233,
      "learning_rate": 9.995263060085454e-05,
      "loss": 1.344,
      "step": 20
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4762611988791945,
      "learning_rate": 9.993552866139229e-05,
      "loss": 1.3513,
      "step": 21
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4669268035616973,
      "learning_rate": 9.99157980783751e-05,
      "loss": 1.3357,
      "step": 22
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46010212813136436,
      "learning_rate": 9.989343989043563e-05,
      "loss": 1.3391,
      "step": 23
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4608157209685965,
      "learning_rate": 9.98684552745256e-05,
      "loss": 1.3591,
      "step": 24
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4483237749946448,
      "learning_rate": 9.984084554585387e-05,
      "loss": 1.3433,
      "step": 25
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4491711395465447,
      "learning_rate": 9.98106121578172e-05,
      "loss": 1.3616,
      "step": 26
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42807282609895003,
      "learning_rate": 9.977775670192371e-05,
      "loss": 1.3342,
      "step": 27
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4236536126545817,
      "learning_rate": 9.97422809077092e-05,
      "loss": 1.3572,
      "step": 28
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4237881873453837,
      "learning_rate": 9.970418664264595e-05,
      "loss": 1.3394,
      "step": 29
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4363640797964494,
      "learning_rate": 9.96634759120446e-05,
      "loss": 1.3611,
      "step": 30
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3953629543521621,
      "learning_rate": 9.962015085894838e-05,
      "loss": 1.3208,
      "step": 31
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4092487417981463,
      "learning_rate": 9.957421376402051e-05,
      "loss": 1.3457,
      "step": 32
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4026279934538214,
      "learning_rate": 9.9525667045424e-05,
      "loss": 1.344,
      "step": 33
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.39575866758551026,
      "learning_rate": 9.947451325869439e-05,
      "loss": 1.3416,
      "step": 34
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.40044778357348837,
      "learning_rate": 9.942075509660527e-05,
      "loss": 1.3181,
      "step": 35
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.40607328352911176,
      "learning_rate": 9.936439538902644e-05,
      "loss": 1.321,
      "step": 36
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4080804869363711,
      "learning_rate": 9.930543710277509e-05,
      "loss": 1.3423,
      "step": 37
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.39208662642207626,
      "learning_rate": 9.924388334145943e-05,
      "loss": 1.3333,
      "step": 38
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4056602363153877,
      "learning_rate": 9.91797373453155e-05,
      "loss": 1.3521,
      "step": 39
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.40133586130843213,
      "learning_rate": 9.911300249103645e-05,
      "loss": 1.3284,
      "step": 40
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3839921181227527,
      "learning_rate": 9.904368229159493e-05,
      "loss": 1.334,
      "step": 41
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.37819843842451634,
      "learning_rate": 9.897178039605802e-05,
      "loss": 1.3201,
      "step": 42
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.38535218746253924,
      "learning_rate": 9.889730058939528e-05,
      "loss": 1.3418,
      "step": 43
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3816398827811328,
      "learning_rate": 9.88202467922794e-05,
      "loss": 1.292,
      "step": 44
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3872171757684603,
      "learning_rate": 9.874062306087982e-05,
      "loss": 1.3086,
      "step": 45
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40255942917819887,
      "learning_rate": 9.865843358664932e-05,
      "loss": 1.356,
      "step": 46
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.38408063484536253,
      "learning_rate": 9.857368269610324e-05,
      "loss": 1.3123,
      "step": 47
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3858660663689368,
      "learning_rate": 9.848637485059183e-05,
      "loss": 1.3228,
      "step": 48
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3733195606949192,
      "learning_rate": 9.839651464606529e-05,
      "loss": 1.3191,
      "step": 49
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3843895669282248,
      "learning_rate": 9.830410681283202e-05,
      "loss": 1.3137,
      "step": 50
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3839890456700812,
      "learning_rate": 9.82091562153094e-05,
      "loss": 1.3037,
      "step": 51
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3835789656095749,
      "learning_rate": 9.811166785176784e-05,
      "loss": 1.3281,
      "step": 52
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3742526616293116,
      "learning_rate": 9.801164685406771e-05,
      "loss": 1.3083,
      "step": 53
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3896917243786738,
      "learning_rate": 9.790909848738906e-05,
      "loss": 1.3142,
      "step": 54
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.37476225815080133,
      "learning_rate": 9.780402814995459e-05,
      "loss": 1.2991,
      "step": 55
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3715519529394886,
      "learning_rate": 9.769644137274538e-05,
      "loss": 1.2986,
      "step": 56
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3766009323599998,
      "learning_rate": 9.758634381920981e-05,
      "loss": 1.3054,
      "step": 57
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.38112908922611866,
      "learning_rate": 9.74737412849654e-05,
      "loss": 1.2891,
      "step": 58
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.37814911656520694,
      "learning_rate": 9.735863969749372e-05,
      "loss": 1.3025,
      "step": 59
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3825181116798945,
      "learning_rate": 9.724104511582838e-05,
      "loss": 1.3152,
      "step": 60
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37912087213705015,
      "learning_rate": 9.712096373023604e-05,
      "loss": 1.2844,
      "step": 61
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3776600267484215,
      "learning_rate": 9.69984018618906e-05,
      "loss": 1.3042,
      "step": 62
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3928544712245226,
      "learning_rate": 9.687336596254044e-05,
      "loss": 1.3259,
      "step": 63
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3652188942778873,
      "learning_rate": 9.674586261416873e-05,
      "loss": 1.2854,
      "step": 64
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3846903345562418,
      "learning_rate": 9.66158985286471e-05,
      "loss": 1.3091,
      "step": 65
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3872756941099026,
      "learning_rate": 9.648348054738208e-05,
      "loss": 1.2981,
      "step": 66
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38743186857828676,
      "learning_rate": 9.634861564095524e-05,
      "loss": 1.303,
      "step": 67
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39711005882387096,
      "learning_rate": 9.621131090875603e-05,
      "loss": 1.313,
      "step": 68
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3775435690594975,
      "learning_rate": 9.607157357860823e-05,
      "loss": 1.2766,
      "step": 69
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3875463212418074,
      "learning_rate": 9.59294110063893e-05,
      "loss": 1.3157,
      "step": 70
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3858636973285557,
      "learning_rate": 9.578483067564335e-05,
      "loss": 1.2834,
      "step": 71
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.37541703932578996,
      "learning_rate": 9.563784019718703e-05,
      "loss": 1.2957,
      "step": 72
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38113820507589535,
      "learning_rate": 9.548844730870901e-05,
      "loss": 1.2932,
      "step": 73
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38509417483302427,
      "learning_rate": 9.533665987436263e-05,
      "loss": 1.311,
      "step": 74
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36678005983919304,
      "learning_rate": 9.518248588435185e-05,
      "loss": 1.2852,
      "step": 75
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3875959550686585,
      "learning_rate": 9.502593345451078e-05,
      "loss": 1.2961,
      "step": 76
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3690516151813626,
      "learning_rate": 9.486701082587634e-05,
      "loss": 1.2837,
      "step": 77
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.37636909155427106,
      "learning_rate": 9.470572636425449e-05,
      "loss": 1.3242,
      "step": 78
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3723573615312267,
      "learning_rate": 9.454208855977986e-05,
      "loss": 1.2837,
      "step": 79
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3849981556406106,
      "learning_rate": 9.437610602646878e-05,
      "loss": 1.2944,
      "step": 80
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3778470741048999,
      "learning_rate": 9.420778750176588e-05,
      "loss": 1.2871,
      "step": 81
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.37645381040425563,
      "learning_rate": 9.403714184608411e-05,
      "loss": 1.3088,
      "step": 82
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.397847096161663,
      "learning_rate": 9.386417804233835e-05,
      "loss": 1.3196,
      "step": 83
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39589602529348583,
      "learning_rate": 9.368890519547251e-05,
      "loss": 1.2812,
      "step": 84
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3755562583506324,
      "learning_rate": 9.351133253198028e-05,
      "loss": 1.2759,
      "step": 85
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3757465002070439,
      "learning_rate": 9.333146939941938e-05,
      "loss": 1.2949,
      "step": 86
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38318152261822297,
      "learning_rate": 9.314932526591956e-05,
      "loss": 1.2795,
      "step": 87
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3756550651404947,
      "learning_rate": 9.296490971968415e-05,
      "loss": 1.2651,
      "step": 88
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38942044400837683,
      "learning_rate": 9.277823246848536e-05,
      "loss": 1.2903,
      "step": 89
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37348810668822124,
      "learning_rate": 9.258930333915324e-05,
      "loss": 1.2612,
      "step": 90
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3832983019734946,
      "learning_rate": 9.239813227705838e-05,
      "loss": 1.2805,
      "step": 91
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37762671923250346,
      "learning_rate": 9.220472934558837e-05,
      "loss": 1.3,
      "step": 92
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3791093875126196,
      "learning_rate": 9.20091047256181e-05,
      "loss": 1.2925,
      "step": 93
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.383463697649199,
      "learning_rate": 9.181126871497378e-05,
      "loss": 1.2815,
      "step": 94
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.379130797624887,
      "learning_rate": 9.16112317278909e-05,
      "loss": 1.2805,
      "step": 95
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3816039801864187,
      "learning_rate": 9.140900429446602e-05,
      "loss": 1.2825,
      "step": 96
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37629435581707915,
      "learning_rate": 9.120459706010233e-05,
      "loss": 1.2896,
      "step": 97
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38890563473703343,
      "learning_rate": 9.099802078494948e-05,
      "loss": 1.2734,
      "step": 98
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3782158636576143,
      "learning_rate": 9.078928634333698e-05,
      "loss": 1.2803,
      "step": 99
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38350433679936774,
      "learning_rate": 9.057840472320192e-05,
      "loss": 1.2639,
      "step": 100
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38940074349705917,
      "learning_rate": 9.036538702551036e-05,
      "loss": 1.2732,
      "step": 101
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39603306187884135,
      "learning_rate": 9.015024446367315e-05,
      "loss": 1.2976,
      "step": 102
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39112427722222776,
      "learning_rate": 8.993298836295556e-05,
      "loss": 1.2588,
      "step": 103
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3962255296124105,
      "learning_rate": 8.971363015988114e-05,
      "loss": 1.2915,
      "step": 104
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3858473392543198,
      "learning_rate": 8.949218140162966e-05,
      "loss": 1.2695,
      "step": 105
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3820634901255279,
      "learning_rate": 8.926865374542928e-05,
      "loss": 1.2876,
      "step": 106
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40421307595226363,
      "learning_rate": 8.904305895794292e-05,
      "loss": 1.2817,
      "step": 107
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3716456005592202,
      "learning_rate": 8.881540891464879e-05,
      "loss": 1.2761,
      "step": 108
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3815567099868862,
      "learning_rate": 8.858571559921538e-05,
      "loss": 1.2644,
      "step": 109
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3848695348441116,
      "learning_rate": 8.835399110287045e-05,
      "loss": 1.2622,
      "step": 110
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38017284820864067,
      "learning_rate": 8.812024762376477e-05,
      "loss": 1.2642,
      "step": 111
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3779202181266027,
      "learning_rate": 8.788449746632977e-05,
      "loss": 1.2766,
      "step": 112
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39015638614586473,
      "learning_rate": 8.764675304062993e-05,
      "loss": 1.2881,
      "step": 113
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39447312823941305,
      "learning_rate": 8.740702686170955e-05,
      "loss": 1.2776,
      "step": 114
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38606430769475053,
      "learning_rate": 8.716533154893389e-05,
      "loss": 1.283,
      "step": 115
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37710007527264683,
      "learning_rate": 8.692167982532486e-05,
      "loss": 1.2754,
      "step": 116
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3953761864482322,
      "learning_rate": 8.667608451689134e-05,
      "loss": 1.2891,
      "step": 117
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38041030780699725,
      "learning_rate": 8.642855855195394e-05,
      "loss": 1.2717,
      "step": 118
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37534026330530307,
      "learning_rate": 8.617911496046446e-05,
      "loss": 1.2861,
      "step": 119
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3642851567224028,
      "learning_rate": 8.592776687332002e-05,
      "loss": 1.2546,
      "step": 120
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37390620756705745,
      "learning_rate": 8.567452752167182e-05,
      "loss": 1.2583,
      "step": 121
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37970332189538375,
      "learning_rate": 8.54194102362286e-05,
      "loss": 1.2827,
      "step": 122
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3873078508435452,
      "learning_rate": 8.516242844655499e-05,
      "loss": 1.2708,
      "step": 123
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38706228040025265,
      "learning_rate": 8.490359568036446e-05,
      "loss": 1.269,
      "step": 124
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3745301218341843,
      "learning_rate": 8.464292556280733e-05,
      "loss": 1.2434,
      "step": 125
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38170858987166395,
      "learning_rate": 8.43804318157534e-05,
      "loss": 1.2456,
      "step": 126
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3735646632404992,
      "learning_rate": 8.411612825706977e-05,
      "loss": 1.2573,
      "step": 127
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4099110289018266,
      "learning_rate": 8.385002879989328e-05,
      "loss": 1.2915,
      "step": 128
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3877538310898807,
      "learning_rate": 8.35821474518983e-05,
      "loss": 1.2793,
      "step": 129
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3902454161481885,
      "learning_rate": 8.33124983145592e-05,
      "loss": 1.2683,
      "step": 130
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38878925776724954,
      "learning_rate": 8.304109558240815e-05,
      "loss": 1.2417,
      "step": 131
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39723877066251073,
      "learning_rate": 8.276795354228786e-05,
      "loss": 1.2751,
      "step": 132
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38395165663929565,
      "learning_rate": 8.249308657259943e-05,
      "loss": 1.2576,
      "step": 133
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39643029338267877,
      "learning_rate": 8.221650914254566e-05,
      "loss": 1.2742,
      "step": 134
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39444322211869104,
      "learning_rate": 8.193823581136918e-05,
      "loss": 1.2671,
      "step": 135
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3941951730886792,
      "learning_rate": 8.165828122758615e-05,
      "loss": 1.2612,
      "step": 136
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3877727475185371,
      "learning_rate": 8.137666012821514e-05,
      "loss": 1.2441,
      "step": 137
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40860172723134963,
      "learning_rate": 8.109338733800132e-05,
      "loss": 1.2839,
      "step": 138
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39465054288851986,
      "learning_rate": 8.08084777686361e-05,
      "loss": 1.2595,
      "step": 139
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38699274061736677,
      "learning_rate": 8.052194641797217e-05,
      "loss": 1.2539,
      "step": 140
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3901027157370492,
      "learning_rate": 8.023380836923405e-05,
      "loss": 1.269,
      "step": 141
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3818216948779056,
      "learning_rate": 7.994407879022396e-05,
      "loss": 1.2632,
      "step": 142
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.364402831939757,
      "learning_rate": 7.965277293252353e-05,
      "loss": 1.2329,
      "step": 143
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3815713950573408,
      "learning_rate": 7.935990613069086e-05,
      "loss": 1.2734,
      "step": 144
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.370245164697694,
      "learning_rate": 7.906549380145329e-05,
      "loss": 1.2412,
      "step": 145
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37768543350286454,
      "learning_rate": 7.876955144289592e-05,
      "loss": 1.2681,
      "step": 146
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37390287320878,
      "learning_rate": 7.847209463364573e-05,
      "loss": 1.2668,
      "step": 147
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37508076646147226,
      "learning_rate": 7.817313903205148e-05,
      "loss": 1.2566,
      "step": 148
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3775081062527898,
      "learning_rate": 7.78727003753595e-05,
      "loss": 1.2434,
      "step": 149
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3901559194494511,
      "learning_rate": 7.757079447888527e-05,
      "loss": 1.2351,
      "step": 150
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 447,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "total_flos": 2.222372777610969e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}