{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.989010989010989,
  "eval_steps": 500,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006593406593406593,
      "grad_norm": 4.499257564544678,
      "learning_rate": 0.0,
      "loss": 5.3313,
      "step": 1
    },
    {
      "epoch": 0.013186813186813187,
      "grad_norm": 3.7660536766052246,
      "learning_rate": 4e-05,
      "loss": 4.7758,
      "step": 2
    },
    {
      "epoch": 0.01978021978021978,
      "grad_norm": 4.873054504394531,
      "learning_rate": 8e-05,
      "loss": 5.0889,
      "step": 3
    },
    {
      "epoch": 0.026373626373626374,
      "grad_norm": 5.257667064666748,
      "learning_rate": 0.00012,
      "loss": 4.6723,
      "step": 4
    },
    {
      "epoch": 0.03296703296703297,
      "grad_norm": 5.000584125518799,
      "learning_rate": 0.00016,
      "loss": 4.7013,
      "step": 5
    },
    {
      "epoch": 0.03956043956043956,
      "grad_norm": 5.8762288093566895,
      "learning_rate": 0.0002,
      "loss": 4.1236,
      "step": 6
    },
    {
      "epoch": 0.046153846153846156,
      "grad_norm": 6.645964622497559,
      "learning_rate": 0.00019955654101995565,
      "loss": 2.4379,
      "step": 7
    },
    {
      "epoch": 0.05274725274725275,
      "grad_norm": 2.519747734069824,
      "learning_rate": 0.00019911308203991133,
      "loss": 1.7819,
      "step": 8
    },
    {
      "epoch": 0.05934065934065934,
      "grad_norm": 1.9227187633514404,
      "learning_rate": 0.00019866962305986697,
      "loss": 1.5306,
      "step": 9
    },
    {
      "epoch": 0.06593406593406594,
      "grad_norm": 1.9636014699935913,
      "learning_rate": 0.00019822616407982261,
      "loss": 1.45,
      "step": 10
    },
    {
      "epoch": 0.07252747252747253,
      "grad_norm": 1.5986920595169067,
      "learning_rate": 0.00019778270509977829,
      "loss": 1.2691,
      "step": 11
    },
    {
      "epoch": 0.07912087912087912,
      "grad_norm": 1.2248263359069824,
      "learning_rate": 0.00019733924611973393,
      "loss": 0.9071,
      "step": 12
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 0.8552874326705933,
      "learning_rate": 0.0001968957871396896,
      "loss": 1.0657,
      "step": 13
    },
    {
      "epoch": 0.09230769230769231,
      "grad_norm": 0.7169967293739319,
      "learning_rate": 0.00019645232815964525,
      "loss": 1.0386,
      "step": 14
    },
    {
      "epoch": 0.0989010989010989,
      "grad_norm": 0.9130010008811951,
      "learning_rate": 0.00019600886917960092,
      "loss": 0.9252,
      "step": 15
    },
    {
      "epoch": 0.1054945054945055,
      "grad_norm": 0.7639235258102417,
      "learning_rate": 0.00019556541019955653,
      "loss": 0.869,
      "step": 16
    },
    {
      "epoch": 0.11208791208791209,
      "grad_norm": 0.6696504950523376,
      "learning_rate": 0.0001951219512195122,
      "loss": 0.8536,
      "step": 17
    },
    {
      "epoch": 0.11868131868131868,
      "grad_norm": 1.052955150604248,
      "learning_rate": 0.00019467849223946785,
      "loss": 0.9968,
      "step": 18
    },
    {
      "epoch": 0.12527472527472527,
      "grad_norm": 0.7062643766403198,
      "learning_rate": 0.00019423503325942352,
      "loss": 1.0926,
      "step": 19
    },
    {
      "epoch": 0.13186813186813187,
      "grad_norm": 0.631159245967865,
      "learning_rate": 0.00019379157427937917,
      "loss": 1.0521,
      "step": 20
    },
    {
      "epoch": 0.13846153846153847,
      "grad_norm": 0.8975602388381958,
      "learning_rate": 0.00019334811529933484,
      "loss": 1.0703,
      "step": 21
    },
    {
      "epoch": 0.14505494505494507,
      "grad_norm": 0.6034598350524902,
      "learning_rate": 0.00019290465631929045,
      "loss": 0.7942,
      "step": 22
    },
    {
      "epoch": 0.15164835164835164,
      "grad_norm": 0.6604612469673157,
      "learning_rate": 0.00019246119733924613,
      "loss": 1.1393,
      "step": 23
    },
    {
      "epoch": 0.15824175824175823,
      "grad_norm": 0.5436087846755981,
      "learning_rate": 0.00019201773835920177,
      "loss": 0.8635,
      "step": 24
    },
    {
      "epoch": 0.16483516483516483,
      "grad_norm": 0.629786491394043,
      "learning_rate": 0.00019157427937915744,
      "loss": 0.9028,
      "step": 25
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 0.6100546717643738,
      "learning_rate": 0.00019113082039911309,
      "loss": 1.021,
      "step": 26
    },
    {
      "epoch": 0.17802197802197803,
      "grad_norm": 0.580397367477417,
      "learning_rate": 0.00019068736141906876,
      "loss": 0.9918,
      "step": 27
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 0.514330267906189,
      "learning_rate": 0.0001902439024390244,
      "loss": 0.9234,
      "step": 28
    },
    {
      "epoch": 0.1912087912087912,
      "grad_norm": 0.48265182971954346,
      "learning_rate": 0.00018980044345898005,
      "loss": 0.8785,
      "step": 29
    },
    {
      "epoch": 0.1978021978021978,
      "grad_norm": 0.5611594319343567,
      "learning_rate": 0.00018935698447893572,
      "loss": 0.8495,
      "step": 30
    },
    {
      "epoch": 0.2043956043956044,
      "grad_norm": 0.5207082629203796,
      "learning_rate": 0.00018891352549889136,
      "loss": 1.0655,
      "step": 31
    },
    {
      "epoch": 0.210989010989011,
      "grad_norm": 0.593443751335144,
      "learning_rate": 0.00018847006651884703,
      "loss": 1.0377,
      "step": 32
    },
    {
      "epoch": 0.2175824175824176,
      "grad_norm": 0.49693596363067627,
      "learning_rate": 0.00018802660753880268,
      "loss": 0.8228,
      "step": 33
    },
    {
      "epoch": 0.22417582417582418,
      "grad_norm": 0.6294080018997192,
      "learning_rate": 0.00018758314855875832,
      "loss": 1.123,
      "step": 34
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 0.5960468053817749,
      "learning_rate": 0.00018713968957871397,
      "loss": 0.8843,
      "step": 35
    },
    {
      "epoch": 0.23736263736263735,
      "grad_norm": 0.5886806845664978,
      "learning_rate": 0.00018669623059866964,
      "loss": 0.7077,
      "step": 36
    },
    {
      "epoch": 0.24395604395604395,
      "grad_norm": 0.7527482509613037,
      "learning_rate": 0.00018625277161862528,
      "loss": 1.1688,
      "step": 37
    },
    {
      "epoch": 0.25054945054945055,
      "grad_norm": 0.5052762031555176,
      "learning_rate": 0.00018580931263858095,
      "loss": 0.9452,
      "step": 38
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 0.6011185646057129,
      "learning_rate": 0.0001853658536585366,
      "loss": 0.9889,
      "step": 39
    },
    {
      "epoch": 0.26373626373626374,
      "grad_norm": 0.4893234074115753,
      "learning_rate": 0.00018492239467849224,
      "loss": 0.9016,
      "step": 40
    },
    {
      "epoch": 0.2703296703296703,
      "grad_norm": 0.5956540703773499,
      "learning_rate": 0.0001844789356984479,
      "loss": 0.8951,
      "step": 41
    },
    {
      "epoch": 0.27692307692307694,
      "grad_norm": 0.6853311657905579,
      "learning_rate": 0.00018403547671840356,
      "loss": 0.8943,
      "step": 42
    },
    {
      "epoch": 0.2835164835164835,
      "grad_norm": 0.514132559299469,
      "learning_rate": 0.0001835920177383592,
      "loss": 1.014,
      "step": 43
    },
    {
      "epoch": 0.29010989010989013,
      "grad_norm": 0.7246399521827698,
      "learning_rate": 0.00018314855875831487,
      "loss": 1.124,
      "step": 44
    },
    {
      "epoch": 0.2967032967032967,
      "grad_norm": 0.534878671169281,
      "learning_rate": 0.00018270509977827052,
      "loss": 0.6777,
      "step": 45
    },
    {
      "epoch": 0.3032967032967033,
      "grad_norm": 0.46303069591522217,
      "learning_rate": 0.00018226164079822616,
      "loss": 0.8077,
      "step": 46
    },
    {
      "epoch": 0.3098901098901099,
      "grad_norm": 0.5512669086456299,
      "learning_rate": 0.00018181818181818183,
      "loss": 0.9223,
      "step": 47
    },
    {
      "epoch": 0.31648351648351647,
      "grad_norm": 0.48670482635498047,
      "learning_rate": 0.00018137472283813748,
      "loss": 0.9329,
      "step": 48
    },
    {
      "epoch": 0.3230769230769231,
      "grad_norm": 0.6091970205307007,
      "learning_rate": 0.00018093126385809312,
      "loss": 1.1785,
      "step": 49
    },
    {
      "epoch": 0.32967032967032966,
      "grad_norm": 0.5351585745811462,
      "learning_rate": 0.0001804878048780488,
      "loss": 1.1025,
      "step": 50
    },
    {
      "epoch": 0.3362637362637363,
      "grad_norm": 0.5463150143623352,
      "learning_rate": 0.00018004434589800444,
      "loss": 0.8758,
      "step": 51
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 0.4296164810657501,
      "learning_rate": 0.00017960088691796008,
      "loss": 0.5897,
      "step": 52
    },
    {
      "epoch": 0.34945054945054943,
      "grad_norm": 0.7088434100151062,
      "learning_rate": 0.00017915742793791575,
      "loss": 0.8874,
      "step": 53
    },
    {
      "epoch": 0.35604395604395606,
      "grad_norm": 0.5368345975875854,
      "learning_rate": 0.0001787139689578714,
      "loss": 0.8522,
      "step": 54
    },
    {
      "epoch": 0.3626373626373626,
      "grad_norm": 0.5056143999099731,
      "learning_rate": 0.00017827050997782707,
      "loss": 0.9154,
      "step": 55
    },
    {
      "epoch": 0.36923076923076925,
      "grad_norm": 0.48936212062835693,
      "learning_rate": 0.00017782705099778271,
      "loss": 0.8478,
      "step": 56
    },
    {
      "epoch": 0.3758241758241758,
      "grad_norm": 0.6096042990684509,
      "learning_rate": 0.00017738359201773839,
      "loss": 1.0707,
      "step": 57
    },
    {
      "epoch": 0.3824175824175824,
      "grad_norm": 0.5620200634002686,
      "learning_rate": 0.000176940133037694,
      "loss": 0.8305,
      "step": 58
    },
    {
      "epoch": 0.389010989010989,
      "grad_norm": 0.5068883895874023,
      "learning_rate": 0.00017649667405764967,
      "loss": 0.7459,
      "step": 59
    },
    {
      "epoch": 0.3956043956043956,
      "grad_norm": 0.609451949596405,
      "learning_rate": 0.00017605321507760532,
      "loss": 0.7703,
      "step": 60
    },
    {
      "epoch": 0.4021978021978022,
      "grad_norm": 0.6350243091583252,
      "learning_rate": 0.000175609756097561,
      "loss": 1.0613,
      "step": 61
    },
    {
      "epoch": 0.4087912087912088,
      "grad_norm": 0.6099206805229187,
      "learning_rate": 0.00017516629711751663,
      "loss": 0.9054,
      "step": 62
    },
    {
      "epoch": 0.4153846153846154,
      "grad_norm": 0.53519207239151,
      "learning_rate": 0.0001747228381374723,
      "loss": 0.7721,
      "step": 63
    },
    {
      "epoch": 0.421978021978022,
      "grad_norm": 0.565395712852478,
      "learning_rate": 0.00017427937915742792,
      "loss": 0.7332,
      "step": 64
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.4922870993614197,
      "learning_rate": 0.0001738359201773836,
      "loss": 0.7914,
      "step": 65
    },
    {
      "epoch": 0.4351648351648352,
      "grad_norm": 0.5346812605857849,
      "learning_rate": 0.00017339246119733924,
      "loss": 0.8879,
      "step": 66
    },
    {
      "epoch": 0.44175824175824174,
      "grad_norm": 0.5556306838989258,
      "learning_rate": 0.0001729490022172949,
      "loss": 0.9712,
      "step": 67
    },
    {
      "epoch": 0.44835164835164837,
      "grad_norm": 0.5152814388275146,
      "learning_rate": 0.00017250554323725056,
      "loss": 0.8266,
      "step": 68
    },
    {
      "epoch": 0.45494505494505494,
      "grad_norm": 0.5951769351959229,
      "learning_rate": 0.00017206208425720623,
      "loss": 0.7834,
      "step": 69
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.5551192760467529,
      "learning_rate": 0.00017161862527716187,
      "loss": 0.9461,
      "step": 70
    },
    {
      "epoch": 0.46813186813186813,
      "grad_norm": 0.4997961223125458,
      "learning_rate": 0.00017117516629711752,
      "loss": 0.9438,
      "step": 71
    },
    {
      "epoch": 0.4747252747252747,
      "grad_norm": 0.693430483341217,
      "learning_rate": 0.0001707317073170732,
      "loss": 1.0394,
      "step": 72
    },
    {
      "epoch": 0.48131868131868133,
      "grad_norm": 0.5007423162460327,
      "learning_rate": 0.00017028824833702883,
      "loss": 0.6293,
      "step": 73
    },
    {
      "epoch": 0.4879120879120879,
      "grad_norm": 0.5647709965705872,
      "learning_rate": 0.0001698447893569845,
      "loss": 0.8842,
      "step": 74
    },
    {
      "epoch": 0.4945054945054945,
      "grad_norm": 0.6222750544548035,
      "learning_rate": 0.00016940133037694015,
      "loss": 1.1177,
      "step": 75
    },
    {
      "epoch": 0.5010989010989011,
      "grad_norm": 0.5330225825309753,
      "learning_rate": 0.00016895787139689582,
      "loss": 0.8427,
      "step": 76
    },
    {
      "epoch": 0.5076923076923077,
      "grad_norm": 0.6527089476585388,
      "learning_rate": 0.00016851441241685144,
      "loss": 0.955,
      "step": 77
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 0.705427348613739,
      "learning_rate": 0.0001680709534368071,
      "loss": 1.0531,
      "step": 78
    },
    {
      "epoch": 0.5208791208791209,
      "grad_norm": 0.5138844847679138,
      "learning_rate": 0.00016762749445676275,
      "loss": 0.9054,
      "step": 79
    },
    {
      "epoch": 0.5274725274725275,
      "grad_norm": 0.632585346698761,
      "learning_rate": 0.00016718403547671842,
      "loss": 0.9244,
      "step": 80
    },
    {
      "epoch": 0.5340659340659341,
      "grad_norm": 0.5227172374725342,
      "learning_rate": 0.00016674057649667407,
      "loss": 0.9188,
      "step": 81
    },
    {
      "epoch": 0.5406593406593406,
      "grad_norm": 0.5184674263000488,
      "learning_rate": 0.00016629711751662974,
      "loss": 0.8139,
      "step": 82
    },
    {
      "epoch": 0.5472527472527473,
      "grad_norm": 0.5389203429222107,
      "learning_rate": 0.00016585365853658536,
      "loss": 1.0064,
      "step": 83
    },
    {
      "epoch": 0.5538461538461539,
      "grad_norm": 0.5201700329780579,
      "learning_rate": 0.00016541019955654103,
      "loss": 0.9715,
      "step": 84
    },
    {
      "epoch": 0.5604395604395604,
      "grad_norm": 0.4657474160194397,
      "learning_rate": 0.00016496674057649667,
      "loss": 0.6887,
      "step": 85
    },
    {
      "epoch": 0.567032967032967,
      "grad_norm": 0.48901787400245667,
      "learning_rate": 0.00016452328159645234,
      "loss": 0.9894,
      "step": 86
    },
    {
      "epoch": 0.5736263736263736,
      "grad_norm": 0.5055707693099976,
      "learning_rate": 0.000164079822616408,
      "loss": 0.7828,
      "step": 87
    },
    {
      "epoch": 0.5802197802197803,
      "grad_norm": 0.5926926136016846,
      "learning_rate": 0.00016363636363636366,
      "loss": 0.8319,
      "step": 88
    },
    {
      "epoch": 0.5868131868131868,
      "grad_norm": 0.5132814049720764,
      "learning_rate": 0.0001631929046563193,
      "loss": 0.651,
      "step": 89
    },
    {
      "epoch": 0.5934065934065934,
      "grad_norm": 0.5646933317184448,
      "learning_rate": 0.00016274944567627495,
      "loss": 0.8097,
      "step": 90
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5312751531600952,
      "learning_rate": 0.0001623059866962306,
      "loss": 0.6701,
      "step": 91
    },
    {
      "epoch": 0.6065934065934065,
      "grad_norm": 0.8451480865478516,
      "learning_rate": 0.00016186252771618626,
      "loss": 0.8902,
      "step": 92
    },
    {
      "epoch": 0.6131868131868132,
      "grad_norm": 0.5425740480422974,
      "learning_rate": 0.0001614190687361419,
      "loss": 0.9255,
      "step": 93
    },
    {
      "epoch": 0.6197802197802198,
      "grad_norm": 0.7311372756958008,
      "learning_rate": 0.00016097560975609758,
      "loss": 1.0323,
      "step": 94
    },
    {
      "epoch": 0.6263736263736264,
      "grad_norm": 0.5717734098434448,
      "learning_rate": 0.00016053215077605322,
      "loss": 0.8789,
      "step": 95
    },
    {
      "epoch": 0.6329670329670329,
      "grad_norm": 0.6015205383300781,
      "learning_rate": 0.00016008869179600887,
      "loss": 0.9006,
      "step": 96
    },
    {
      "epoch": 0.6395604395604395,
      "grad_norm": 0.7425872683525085,
      "learning_rate": 0.00015964523281596454,
      "loss": 0.878,
      "step": 97
    },
    {
      "epoch": 0.6461538461538462,
      "grad_norm": 0.5961865186691284,
      "learning_rate": 0.00015920177383592018,
      "loss": 0.8861,
      "step": 98
    },
    {
      "epoch": 0.6527472527472528,
      "grad_norm": 0.5109598636627197,
      "learning_rate": 0.00015875831485587586,
      "loss": 0.6897,
      "step": 99
    },
    {
      "epoch": 0.6593406593406593,
      "grad_norm": 0.6388728618621826,
      "learning_rate": 0.0001583148558758315,
      "loss": 0.954,
      "step": 100
    },
    {
      "epoch": 0.6659340659340659,
      "grad_norm": 0.46394044160842896,
      "learning_rate": 0.00015787139689578714,
      "loss": 0.7564,
      "step": 101
    },
    {
      "epoch": 0.6725274725274726,
      "grad_norm": 0.5002279281616211,
      "learning_rate": 0.0001574279379157428,
      "loss": 0.8188,
      "step": 102
    },
    {
      "epoch": 0.6791208791208792,
      "grad_norm": 0.5562055110931396,
      "learning_rate": 0.00015698447893569846,
      "loss": 0.7793,
      "step": 103
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 0.5421922206878662,
      "learning_rate": 0.0001565410199556541,
      "loss": 0.7098,
      "step": 104
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 0.5745899677276611,
      "learning_rate": 0.00015609756097560978,
      "loss": 0.9227,
      "step": 105
    },
    {
      "epoch": 0.6989010989010989,
      "grad_norm": 0.5524622797966003,
      "learning_rate": 0.00015565410199556542,
      "loss": 0.8212,
      "step": 106
    },
    {
      "epoch": 0.7054945054945055,
      "grad_norm": 0.5849021077156067,
      "learning_rate": 0.00015521064301552106,
      "loss": 0.8768,
      "step": 107
    },
    {
      "epoch": 0.7120879120879121,
      "grad_norm": 0.5143813490867615,
      "learning_rate": 0.0001547671840354767,
      "loss": 0.8065,
      "step": 108
    },
    {
      "epoch": 0.7186813186813187,
      "grad_norm": 0.5505753755569458,
      "learning_rate": 0.00015432372505543238,
      "loss": 0.8493,
      "step": 109
    },
    {
      "epoch": 0.7252747252747253,
      "grad_norm": 0.5783767700195312,
      "learning_rate": 0.00015388026607538802,
      "loss": 0.9402,
      "step": 110
    },
    {
      "epoch": 0.7318681318681318,
      "grad_norm": 0.5197792053222656,
      "learning_rate": 0.0001534368070953437,
      "loss": 0.7935,
      "step": 111
    },
    {
      "epoch": 0.7384615384615385,
      "grad_norm": 0.6354763507843018,
      "learning_rate": 0.00015299334811529934,
      "loss": 0.9899,
      "step": 112
    },
    {
      "epoch": 0.7450549450549451,
      "grad_norm": 0.5677642226219177,
      "learning_rate": 0.00015254988913525498,
      "loss": 0.8933,
      "step": 113
    },
    {
      "epoch": 0.7516483516483516,
      "grad_norm": 0.5375794768333435,
      "learning_rate": 0.00015210643015521066,
      "loss": 0.772,
      "step": 114
    },
    {
      "epoch": 0.7582417582417582,
      "grad_norm": 0.5195283889770508,
      "learning_rate": 0.0001516629711751663,
      "loss": 0.9625,
      "step": 115
    },
    {
      "epoch": 0.7648351648351648,
      "grad_norm": 0.611674427986145,
      "learning_rate": 0.00015121951219512197,
      "loss": 0.7522,
      "step": 116
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 0.5556056499481201,
      "learning_rate": 0.00015077605321507762,
      "loss": 0.7993,
      "step": 117
    },
    {
      "epoch": 0.778021978021978,
      "grad_norm": 0.5370631814002991,
      "learning_rate": 0.0001503325942350333,
      "loss": 0.9481,
      "step": 118
    },
    {
      "epoch": 0.7846153846153846,
      "grad_norm": 0.6368606686592102,
      "learning_rate": 0.0001498891352549889,
      "loss": 0.9311,
      "step": 119
    },
    {
      "epoch": 0.7912087912087912,
      "grad_norm": 0.8189669251441956,
      "learning_rate": 0.00014944567627494458,
      "loss": 0.833,
      "step": 120
    },
    {
      "epoch": 0.7978021978021979,
      "grad_norm": 0.6427760124206543,
      "learning_rate": 0.00014900221729490022,
      "loss": 0.8768,
      "step": 121
    },
    {
      "epoch": 0.8043956043956044,
      "grad_norm": 0.5934387445449829,
      "learning_rate": 0.0001485587583148559,
      "loss": 0.9282,
      "step": 122
    },
    {
      "epoch": 0.810989010989011,
      "grad_norm": 0.5895951986312866,
      "learning_rate": 0.00014811529933481154,
      "loss": 0.8735,
      "step": 123
    },
    {
      "epoch": 0.8175824175824176,
      "grad_norm": 0.6617897152900696,
      "learning_rate": 0.0001476718403547672,
      "loss": 0.9052,
      "step": 124
    },
    {
      "epoch": 0.8241758241758241,
      "grad_norm": 0.529255211353302,
      "learning_rate": 0.00014722838137472282,
      "loss": 0.6584,
      "step": 125
    },
    {
      "epoch": 0.8307692307692308,
      "grad_norm": 0.728208601474762,
      "learning_rate": 0.0001467849223946785,
      "loss": 0.788,
      "step": 126
    },
    {
      "epoch": 0.8373626373626374,
      "grad_norm": 0.7506310939788818,
      "learning_rate": 0.00014634146341463414,
      "loss": 0.8579,
      "step": 127
    },
    {
      "epoch": 0.843956043956044,
      "grad_norm": 0.6317883133888245,
      "learning_rate": 0.0001458980044345898,
      "loss": 0.8235,
      "step": 128
    },
    {
      "epoch": 0.8505494505494505,
      "grad_norm": 0.5504070520401001,
      "learning_rate": 0.00014545454545454546,
      "loss": 0.8003,
      "step": 129
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.5831419825553894,
      "learning_rate": 0.00014501108647450113,
      "loss": 0.9854,
      "step": 130
    },
    {
      "epoch": 0.8637362637362638,
      "grad_norm": 0.5612383484840393,
      "learning_rate": 0.00014456762749445675,
      "loss": 0.8427,
      "step": 131
    },
    {
      "epoch": 0.8703296703296703,
      "grad_norm": 0.590522825717926,
      "learning_rate": 0.00014412416851441242,
      "loss": 0.8468,
      "step": 132
    },
    {
      "epoch": 0.8769230769230769,
      "grad_norm": 0.6120272278785706,
      "learning_rate": 0.00014368070953436806,
      "loss": 0.7596,
      "step": 133
    },
    {
      "epoch": 0.8835164835164835,
      "grad_norm": 0.5511749982833862,
      "learning_rate": 0.00014323725055432373,
      "loss": 0.8267,
      "step": 134
    },
    {
      "epoch": 0.8901098901098901,
      "grad_norm": 0.6696128249168396,
      "learning_rate": 0.00014279379157427938,
      "loss": 0.9554,
      "step": 135
    },
    {
      "epoch": 0.8967032967032967,
      "grad_norm": 0.5829819440841675,
      "learning_rate": 0.00014235033259423505,
      "loss": 1.0376,
      "step": 136
    },
    {
      "epoch": 0.9032967032967033,
      "grad_norm": 0.5929968953132629,
      "learning_rate": 0.0001419068736141907,
      "loss": 0.9404,
      "step": 137
    },
    {
      "epoch": 0.9098901098901099,
      "grad_norm": 0.5242624878883362,
      "learning_rate": 0.00014146341463414634,
      "loss": 0.6965,
      "step": 138
    },
    {
      "epoch": 0.9164835164835164,
      "grad_norm": 0.6169236302375793,
      "learning_rate": 0.000141019955654102,
      "loss": 0.7462,
      "step": 139
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 0.6347423195838928,
      "learning_rate": 0.00014057649667405765,
      "loss": 0.7877,
      "step": 140
    },
    {
      "epoch": 0.9296703296703297,
      "grad_norm": 0.5059258341789246,
      "learning_rate": 0.00014013303769401332,
      "loss": 0.7952,
      "step": 141
    },
    {
      "epoch": 0.9362637362637363,
      "grad_norm": 0.5628349781036377,
      "learning_rate": 0.00013968957871396897,
      "loss": 0.8828,
      "step": 142
    },
    {
      "epoch": 0.9428571428571428,
      "grad_norm": 0.6259740591049194,
      "learning_rate": 0.00013924611973392464,
      "loss": 0.8689,
      "step": 143
    },
    {
      "epoch": 0.9494505494505494,
      "grad_norm": 0.6700652241706848,
      "learning_rate": 0.00013880266075388026,
      "loss": 0.938,
      "step": 144
    },
    {
      "epoch": 0.9560439560439561,
      "grad_norm": 0.601922333240509,
      "learning_rate": 0.00013835920177383593,
      "loss": 1.1164,
      "step": 145
    },
    {
      "epoch": 0.9626373626373627,
      "grad_norm": 0.558859646320343,
      "learning_rate": 0.00013791574279379157,
      "loss": 0.7786,
      "step": 146
    },
    {
      "epoch": 0.9692307692307692,
      "grad_norm": 0.5523409843444824,
      "learning_rate": 0.00013747228381374724,
      "loss": 0.8864,
      "step": 147
    },
    {
      "epoch": 0.9758241758241758,
      "grad_norm": 0.5421144962310791,
      "learning_rate": 0.0001370288248337029,
      "loss": 0.8224,
      "step": 148
    },
    {
      "epoch": 0.9824175824175824,
      "grad_norm": 0.634665310382843,
      "learning_rate": 0.00013658536585365856,
      "loss": 1.0744,
      "step": 149
    },
    {
      "epoch": 0.989010989010989,
      "grad_norm": 0.5394982099533081,
      "learning_rate": 0.00013614190687361418,
      "loss": 0.7321,
      "step": 150
    }
  ],
  "logging_steps": 1,
  "max_steps": 456,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 15,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.566333305315328e+16,
  "train_batch_size": 22,
  "trial_name": null,
  "trial_params": null
}