{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 327,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0061162079510703364,
      "grad_norm": 0.8125,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 1.96957528591156,
      "step": 2
    },
    {
      "epoch": 0.012232415902140673,
      "grad_norm": 0.65234375,
      "learning_rate": 7.058823529411766e-06,
      "loss": 2.0347838401794434,
      "step": 4
    },
    {
      "epoch": 0.01834862385321101,
      "grad_norm": 0.578125,
      "learning_rate": 1.1764705882352942e-05,
      "loss": 1.9940425157546997,
      "step": 6
    },
    {
      "epoch": 0.024464831804281346,
      "grad_norm": 0.640625,
      "learning_rate": 1.647058823529412e-05,
      "loss": 2.06846284866333,
      "step": 8
    },
    {
      "epoch": 0.03058103975535168,
      "grad_norm": 0.80859375,
      "learning_rate": 2.1176470588235296e-05,
      "loss": 1.9072656631469727,
      "step": 10
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 0.62109375,
      "learning_rate": 2.5882352941176475e-05,
      "loss": 1.8469384908676147,
      "step": 12
    },
    {
      "epoch": 0.04281345565749235,
      "grad_norm": 0.67578125,
      "learning_rate": 3.0588235294117644e-05,
      "loss": 1.822761058807373,
      "step": 14
    },
    {
      "epoch": 0.04892966360856269,
      "grad_norm": 0.59765625,
      "learning_rate": 3.529411764705883e-05,
      "loss": 1.7572168111801147,
      "step": 16
    },
    {
      "epoch": 0.05504587155963303,
      "grad_norm": 0.45703125,
      "learning_rate": 4e-05,
      "loss": 1.7028628587722778,
      "step": 18
    },
    {
      "epoch": 0.06116207951070336,
      "grad_norm": 0.6796875,
      "learning_rate": 3.999630287622123e-05,
      "loss": 1.8212647438049316,
      "step": 20
    },
    {
      "epoch": 0.0672782874617737,
      "grad_norm": 0.40234375,
      "learning_rate": 3.9985213023632064e-05,
      "loss": 1.6563327312469482,
      "step": 22
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 0.4921875,
      "learning_rate": 3.996673499785002e-05,
      "loss": 1.6510090827941895,
      "step": 24
    },
    {
      "epoch": 0.07951070336391437,
      "grad_norm": 0.7109375,
      "learning_rate": 3.99408763894916e-05,
      "loss": 1.5338315963745117,
      "step": 26
    },
    {
      "epoch": 0.0856269113149847,
      "grad_norm": 0.404296875,
      "learning_rate": 3.9907647821054114e-05,
      "loss": 1.451381802558899,
      "step": 28
    },
    {
      "epoch": 0.09174311926605505,
      "grad_norm": 0.30859375,
      "learning_rate": 3.986706294255207e-05,
      "loss": 1.5903527736663818,
      "step": 30
    },
    {
      "epoch": 0.09785932721712538,
      "grad_norm": 0.5078125,
      "learning_rate": 3.981913842590985e-05,
      "loss": 1.5307936668395996,
      "step": 32
    },
    {
      "epoch": 0.10397553516819572,
      "grad_norm": 0.314453125,
      "learning_rate": 3.9763893958113005e-05,
      "loss": 1.413094401359558,
      "step": 34
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.625,
      "learning_rate": 3.970135223312103e-05,
      "loss": 1.4268063306808472,
      "step": 36
    },
    {
      "epoch": 0.1162079510703364,
      "grad_norm": 0.291015625,
      "learning_rate": 3.96315389425449e-05,
      "loss": 1.4923566579818726,
      "step": 38
    },
    {
      "epoch": 0.12232415902140673,
      "grad_norm": 0.298828125,
      "learning_rate": 3.9554482765093134e-05,
      "loss": 1.574055790901184,
      "step": 40
    },
    {
      "epoch": 0.12844036697247707,
      "grad_norm": 0.474609375,
      "learning_rate": 3.947021535479083e-05,
      "loss": 1.415814995765686,
      "step": 42
    },
    {
      "epoch": 0.1345565749235474,
      "grad_norm": 0.37109375,
      "learning_rate": 3.937877132797649e-05,
      "loss": 1.4422385692596436,
      "step": 44
    },
    {
      "epoch": 0.14067278287461774,
      "grad_norm": 0.375,
      "learning_rate": 3.928018824908187e-05,
      "loss": 1.4337655305862427,
      "step": 46
    },
    {
      "epoch": 0.14678899082568808,
      "grad_norm": 0.478515625,
      "learning_rate": 3.917450661520088e-05,
      "loss": 1.5849357843399048,
      "step": 48
    },
    {
      "epoch": 0.1529051987767584,
      "grad_norm": 0.578125,
      "learning_rate": 3.906176983945371e-05,
      "loss": 1.5441187620162964,
      "step": 50
    },
    {
      "epoch": 0.15902140672782875,
      "grad_norm": 0.310546875,
      "learning_rate": 3.894202423315306e-05,
      "loss": 1.4055734872817993,
      "step": 52
    },
    {
      "epoch": 0.1651376146788991,
      "grad_norm": 0.36328125,
      "learning_rate": 3.8815318986779875e-05,
      "loss": 1.543579339981079,
      "step": 54
    },
    {
      "epoch": 0.1712538226299694,
      "grad_norm": 0.328125,
      "learning_rate": 3.868170614977628e-05,
      "loss": 1.4346128702163696,
      "step": 56
    },
    {
      "epoch": 0.17737003058103976,
      "grad_norm": 0.310546875,
      "learning_rate": 3.854124060916415e-05,
      "loss": 1.5318936109542847,
      "step": 58
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 0.4140625,
      "learning_rate": 3.839398006699797e-05,
      "loss": 1.3749366998672485,
      "step": 60
    },
    {
      "epoch": 0.18960244648318042,
      "grad_norm": 0.47265625,
      "learning_rate": 3.823998501666131e-05,
      "loss": 1.436793565750122,
      "step": 62
    },
    {
      "epoch": 0.19571865443425077,
      "grad_norm": 0.40625,
      "learning_rate": 3.8079318718016666e-05,
      "loss": 1.3264566659927368,
      "step": 64
    },
    {
      "epoch": 0.2018348623853211,
      "grad_norm": 1.34375,
      "learning_rate": 3.7912047171418815e-05,
      "loss": 1.4578938484191895,
      "step": 66
    },
    {
      "epoch": 0.20795107033639143,
      "grad_norm": 0.490234375,
      "learning_rate": 3.773823909060248e-05,
      "loss": 1.3888816833496094,
      "step": 68
    },
    {
      "epoch": 0.21406727828746178,
      "grad_norm": 0.33984375,
      "learning_rate": 3.755796587445528e-05,
      "loss": 1.4028871059417725,
      "step": 70
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.287109375,
      "learning_rate": 3.7371301577687666e-05,
      "loss": 1.3480572700500488,
      "step": 72
    },
    {
      "epoch": 0.22629969418960244,
      "grad_norm": 1.4453125,
      "learning_rate": 3.717832288041188e-05,
      "loss": 1.4603177309036255,
      "step": 74
    },
    {
      "epoch": 0.2324159021406728,
      "grad_norm": 0.3984375,
      "learning_rate": 3.697910905664243e-05,
      "loss": 1.4285566806793213,
      "step": 76
    },
    {
      "epoch": 0.23853211009174313,
      "grad_norm": 0.353515625,
      "learning_rate": 3.6773741941730975e-05,
      "loss": 1.462870478630066,
      "step": 78
    },
    {
      "epoch": 0.24464831804281345,
      "grad_norm": 0.578125,
      "learning_rate": 3.656230589874905e-05,
      "loss": 1.3728106021881104,
      "step": 80
    },
    {
      "epoch": 0.25076452599388377,
      "grad_norm": 0.35546875,
      "learning_rate": 3.6344887783832474e-05,
      "loss": 1.3789210319519043,
      "step": 82
    },
    {
      "epoch": 0.25688073394495414,
      "grad_norm": 0.318359375,
      "learning_rate": 3.6121576910501517e-05,
      "loss": 1.438659429550171,
      "step": 84
    },
    {
      "epoch": 0.26299694189602446,
      "grad_norm": 0.6796875,
      "learning_rate": 3.589246501297172e-05,
      "loss": 1.4267497062683105,
      "step": 86
    },
    {
      "epoch": 0.2691131498470948,
      "grad_norm": 0.265625,
      "learning_rate": 3.565764620847024e-05,
      "loss": 1.297946572303772,
      "step": 88
    },
    {
      "epoch": 0.27522935779816515,
      "grad_norm": 0.50390625,
      "learning_rate": 3.541721695857328e-05,
      "loss": 1.3422534465789795,
      "step": 90
    },
    {
      "epoch": 0.28134556574923547,
      "grad_norm": 0.302734375,
      "learning_rate": 3.5171276029580485e-05,
      "loss": 1.3573137521743774,
      "step": 92
    },
    {
      "epoch": 0.2874617737003058,
      "grad_norm": 0.58203125,
      "learning_rate": 3.4919924451942625e-05,
      "loss": 1.4575246572494507,
      "step": 94
    },
    {
      "epoch": 0.29357798165137616,
      "grad_norm": 0.376953125,
      "learning_rate": 3.466326547875907e-05,
      "loss": 1.3683665990829468,
      "step": 96
    },
    {
      "epoch": 0.2996941896024465,
      "grad_norm": 0.244140625,
      "learning_rate": 3.440140454336236e-05,
      "loss": 1.2624497413635254,
      "step": 98
    },
    {
      "epoch": 0.3058103975535168,
      "grad_norm": 0.369140625,
      "learning_rate": 3.413444921600705e-05,
      "loss": 1.4250640869140625,
      "step": 100
    },
    {
      "epoch": 0.3119266055045872,
      "grad_norm": 0.69921875,
      "learning_rate": 3.3862509159680775e-05,
      "loss": 1.4060899019241333,
      "step": 102
    },
    {
      "epoch": 0.3180428134556575,
      "grad_norm": 0.416015625,
      "learning_rate": 3.3585696085055594e-05,
      "loss": 1.44656503200531,
      "step": 104
    },
    {
      "epoch": 0.3241590214067278,
      "grad_norm": 0.63671875,
      "learning_rate": 3.330412370459823e-05,
      "loss": 1.5084904432296753,
      "step": 106
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.408203125,
      "learning_rate": 3.301790768585793e-05,
      "loss": 1.4001638889312744,
      "step": 108
    },
    {
      "epoch": 0.3363914373088685,
      "grad_norm": 0.337890625,
      "learning_rate": 3.272716560395123e-05,
      "loss": 1.4370101690292358,
      "step": 110
    },
    {
      "epoch": 0.3425076452599388,
      "grad_norm": 0.61328125,
      "learning_rate": 3.243201689326306e-05,
      "loss": 1.4086581468582153,
      "step": 112
    },
    {
      "epoch": 0.3486238532110092,
      "grad_norm": 0.48828125,
      "learning_rate": 3.213258279838416e-05,
      "loss": 1.389402151107788,
      "step": 114
    },
    {
      "epoch": 0.3547400611620795,
      "grad_norm": 0.3515625,
      "learning_rate": 3.182898632430479e-05,
      "loss": 1.371553897857666,
      "step": 116
    },
    {
      "epoch": 0.36085626911314983,
      "grad_norm": 0.392578125,
      "learning_rate": 3.1521352185885326e-05,
      "loss": 1.4507074356079102,
      "step": 118
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 0.703125,
      "learning_rate": 3.120980675662449e-05,
      "loss": 1.3349779844284058,
      "step": 120
    },
    {
      "epoch": 0.3730886850152905,
      "grad_norm": 0.357421875,
      "learning_rate": 3.0894478016746106e-05,
      "loss": 1.3503204584121704,
      "step": 122
    },
    {
      "epoch": 0.37920489296636084,
      "grad_norm": 0.435546875,
      "learning_rate": 3.0575495500626015e-05,
      "loss": 1.4363230466842651,
      "step": 124
    },
    {
      "epoch": 0.3853211009174312,
      "grad_norm": 0.25390625,
      "learning_rate": 3.025299024358036e-05,
      "loss": 1.3254035711288452,
      "step": 126
    },
    {
      "epoch": 0.39143730886850153,
      "grad_norm": 0.357421875,
      "learning_rate": 2.9927094728037422e-05,
      "loss": 1.410750389099121,
      "step": 128
    },
    {
      "epoch": 0.39755351681957185,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9597942829114976e-05,
      "loss": 1.397456169128418,
      "step": 130
    },
    {
      "epoch": 0.4036697247706422,
      "grad_norm": 0.5,
      "learning_rate": 2.926566975962551e-05,
      "loss": 1.3679834604263306,
      "step": 132
    },
    {
      "epoch": 0.40978593272171254,
      "grad_norm": 0.380859375,
      "learning_rate": 2.8930412014531924e-05,
      "loss": 1.4191218614578247,
      "step": 134
    },
    {
      "epoch": 0.41590214067278286,
      "grad_norm": 0.41796875,
      "learning_rate": 2.859230731487661e-05,
      "loss": 1.2690709829330444,
      "step": 136
    },
    {
      "epoch": 0.42201834862385323,
      "grad_norm": 0.439453125,
      "learning_rate": 2.8251494551206767e-05,
      "loss": 1.3012070655822754,
      "step": 138
    },
    {
      "epoch": 0.42813455657492355,
      "grad_norm": 0.3515625,
      "learning_rate": 2.7908113726519356e-05,
      "loss": 1.2863593101501465,
      "step": 140
    },
    {
      "epoch": 0.43425076452599387,
      "grad_norm": 0.498046875,
      "learning_rate": 2.7562305898749054e-05,
      "loss": 1.442170262336731,
      "step": 142
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.3984375,
      "learning_rate": 2.7214213122822864e-05,
      "loss": 1.3614513874053955,
      "step": 144
    },
    {
      "epoch": 0.44648318042813456,
      "grad_norm": 0.28125,
      "learning_rate": 2.6863978392305118e-05,
      "loss": 1.2583218812942505,
      "step": 146
    },
    {
      "epoch": 0.4525993883792049,
      "grad_norm": 0.349609375,
      "learning_rate": 2.651174558065697e-05,
      "loss": 1.4321554899215698,
      "step": 148
    },
    {
      "epoch": 0.45871559633027525,
      "grad_norm": 0.341796875,
      "learning_rate": 2.6157659382134384e-05,
      "loss": 1.414106845855713,
      "step": 150
    },
    {
      "epoch": 0.4648318042813456,
      "grad_norm": 0.57421875,
      "learning_rate": 2.5801865252348935e-05,
      "loss": 1.4366655349731445,
      "step": 152
    },
    {
      "epoch": 0.4709480122324159,
      "grad_norm": 0.392578125,
      "learning_rate": 2.5444509348515912e-05,
      "loss": 1.3970258235931396,
      "step": 154
    },
    {
      "epoch": 0.47706422018348627,
      "grad_norm": 3.078125,
      "learning_rate": 2.508573846941417e-05,
      "loss": 1.4000701904296875,
      "step": 156
    },
    {
      "epoch": 0.4831804281345566,
      "grad_norm": 0.431640625,
      "learning_rate": 2.472569999508238e-05,
      "loss": 1.4267001152038574,
      "step": 158
    },
    {
      "epoch": 0.4892966360856269,
      "grad_norm": 0.33203125,
      "learning_rate": 2.4364541826276638e-05,
      "loss": 1.4192265272140503,
      "step": 160
    },
    {
      "epoch": 0.4954128440366973,
      "grad_norm": 0.283203125,
      "learning_rate": 2.4002412323714026e-05,
      "loss": 1.314334511756897,
      "step": 162
    },
    {
      "epoch": 0.5015290519877675,
      "grad_norm": 0.56640625,
      "learning_rate": 2.363946024712732e-05,
      "loss": 1.355038046836853,
      "step": 164
    },
    {
      "epoch": 0.5076452599388379,
      "grad_norm": 0.6484375,
      "learning_rate": 2.3275834694155716e-05,
      "loss": 1.2806464433670044,
      "step": 166
    },
    {
      "epoch": 0.5137614678899083,
      "grad_norm": 0.6015625,
      "learning_rate": 2.2911685039096834e-05,
      "loss": 1.3307536840438843,
      "step": 168
    },
    {
      "epoch": 0.5198776758409785,
      "grad_norm": 0.333984375,
      "learning_rate": 2.2547160871544973e-05,
      "loss": 1.3525010347366333,
      "step": 170
    },
    {
      "epoch": 0.5259938837920489,
      "grad_norm": 0.404296875,
      "learning_rate": 2.2182411934941004e-05,
      "loss": 1.3771812915802002,
      "step": 172
    },
    {
      "epoch": 0.5321100917431193,
      "grad_norm": 0.337890625,
      "learning_rate": 2.1817588065059008e-05,
      "loss": 1.342699408531189,
      "step": 174
    },
    {
      "epoch": 0.5382262996941896,
      "grad_norm": 0.52734375,
      "learning_rate": 2.145283912845504e-05,
      "loss": 1.3631731271743774,
      "step": 176
    },
    {
      "epoch": 0.5443425076452599,
      "grad_norm": 0.4375,
      "learning_rate": 2.1088314960903172e-05,
      "loss": 1.3468397855758667,
      "step": 178
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.341796875,
      "learning_rate": 2.072416530584429e-05,
      "loss": 1.3079278469085693,
      "step": 180
    },
    {
      "epoch": 0.5565749235474006,
      "grad_norm": 0.478515625,
      "learning_rate": 2.0360539752872688e-05,
      "loss": 1.3927726745605469,
      "step": 182
    },
    {
      "epoch": 0.5626911314984709,
      "grad_norm": 0.71875,
      "learning_rate": 1.9997587676285976e-05,
      "loss": 1.3745498657226562,
      "step": 184
    },
    {
      "epoch": 0.5688073394495413,
      "grad_norm": 0.259765625,
      "learning_rate": 1.9635458173723365e-05,
      "loss": 1.3934822082519531,
      "step": 186
    },
    {
      "epoch": 0.5749235474006116,
      "grad_norm": 0.3203125,
      "learning_rate": 1.9274300004917625e-05,
      "loss": 1.4668513536453247,
      "step": 188
    },
    {
      "epoch": 0.581039755351682,
      "grad_norm": 0.375,
      "learning_rate": 1.8914261530585842e-05,
      "loss": 1.2894922494888306,
      "step": 190
    },
    {
      "epoch": 0.5871559633027523,
      "grad_norm": 0.322265625,
      "learning_rate": 1.855549065148409e-05,
      "loss": 1.3256540298461914,
      "step": 192
    },
    {
      "epoch": 0.5932721712538226,
      "grad_norm": 0.92578125,
      "learning_rate": 1.8198134747651067e-05,
      "loss": 1.3802309036254883,
      "step": 194
    },
    {
      "epoch": 0.599388379204893,
      "grad_norm": 0.33203125,
      "learning_rate": 1.7842340617865625e-05,
      "loss": 1.3032485246658325,
      "step": 196
    },
    {
      "epoch": 0.6055045871559633,
      "grad_norm": 0.54296875,
      "learning_rate": 1.748825441934303e-05,
      "loss": 1.236058235168457,
      "step": 198
    },
    {
      "epoch": 0.6116207951070336,
      "grad_norm": 0.376953125,
      "learning_rate": 1.713602160769489e-05,
      "loss": 1.4403380155563354,
      "step": 200
    },
    {
      "epoch": 0.617737003058104,
      "grad_norm": 0.359375,
      "learning_rate": 1.6785786877177145e-05,
      "loss": 1.3596861362457275,
      "step": 202
    },
    {
      "epoch": 0.6238532110091743,
      "grad_norm": 0.408203125,
      "learning_rate": 1.643769410125095e-05,
      "loss": 1.3777844905853271,
      "step": 204
    },
    {
      "epoch": 0.6299694189602446,
      "grad_norm": 0.4609375,
      "learning_rate": 1.609188627348065e-05,
      "loss": 1.3549814224243164,
      "step": 206
    },
    {
      "epoch": 0.636085626911315,
      "grad_norm": 0.60546875,
      "learning_rate": 1.574850544879324e-05,
      "loss": 1.2403396368026733,
      "step": 208
    },
    {
      "epoch": 0.6422018348623854,
      "grad_norm": 0.421875,
      "learning_rate": 1.5407692685123392e-05,
      "loss": 1.3685152530670166,
      "step": 210
    },
    {
      "epoch": 0.6483180428134556,
      "grad_norm": 0.298828125,
      "learning_rate": 1.5069587985468078e-05,
      "loss": 1.3501070737838745,
      "step": 212
    },
    {
      "epoch": 0.654434250764526,
      "grad_norm": 0.625,
      "learning_rate": 1.4734330240374504e-05,
      "loss": 1.292812466621399,
      "step": 214
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.3125,
      "learning_rate": 1.4402057170885026e-05,
      "loss": 1.287298560142517,
      "step": 216
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.27734375,
      "learning_rate": 1.4072905271962585e-05,
      "loss": 1.3475459814071655,
      "step": 218
    },
    {
      "epoch": 0.672782874617737,
      "grad_norm": 0.376953125,
      "learning_rate": 1.3747009756419657e-05,
      "loss": 1.2376188039779663,
      "step": 220
    },
    {
      "epoch": 0.6788990825688074,
      "grad_norm": 0.47265625,
      "learning_rate": 1.3424504499373994e-05,
      "loss": 1.315477728843689,
      "step": 222
    },
    {
      "epoch": 0.6850152905198776,
      "grad_norm": 0.365234375,
      "learning_rate": 1.31055219832539e-05,
      "loss": 1.364182949066162,
      "step": 224
    },
    {
      "epoch": 0.691131498470948,
      "grad_norm": 0.326171875,
      "learning_rate": 1.2790193243375521e-05,
      "loss": 1.3197802305221558,
      "step": 226
    },
    {
      "epoch": 0.6972477064220184,
      "grad_norm": 0.26171875,
      "learning_rate": 1.2478647814114683e-05,
      "loss": 1.2909021377563477,
      "step": 228
    },
    {
      "epoch": 0.7033639143730887,
      "grad_norm": 0.3984375,
      "learning_rate": 1.2171013675695222e-05,
      "loss": 1.2902398109436035,
      "step": 230
    },
    {
      "epoch": 0.709480122324159,
      "grad_norm": 0.8515625,
      "learning_rate": 1.1867417201615848e-05,
      "loss": 1.3986810445785522,
      "step": 232
    },
    {
      "epoch": 0.7155963302752294,
      "grad_norm": 0.29296875,
      "learning_rate": 1.1567983106736946e-05,
      "loss": 1.2758734226226807,
      "step": 234
    },
    {
      "epoch": 0.7217125382262997,
      "grad_norm": 0.37109375,
      "learning_rate": 1.1272834396048777e-05,
      "loss": 1.317265272140503,
      "step": 236
    },
    {
      "epoch": 0.72782874617737,
      "grad_norm": 3.296875,
      "learning_rate": 1.0982092314142068e-05,
      "loss": 1.4039820432662964,
      "step": 238
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 0.40234375,
      "learning_rate": 1.0695876295401771e-05,
      "loss": 1.330233097076416,
      "step": 240
    },
    {
      "epoch": 0.7400611620795107,
      "grad_norm": 1.7890625,
      "learning_rate": 1.041430391494441e-05,
      "loss": 1.3207037448883057,
      "step": 242
    },
    {
      "epoch": 0.746177370030581,
      "grad_norm": 0.2451171875,
      "learning_rate": 1.013749084031923e-05,
      "loss": 1.3986611366271973,
      "step": 244
    },
    {
      "epoch": 0.7522935779816514,
      "grad_norm": 0.42578125,
      "learning_rate": 9.86555078399295e-06,
      "loss": 1.3146181106567383,
      "step": 246
    },
    {
      "epoch": 0.7584097859327217,
      "grad_norm": 0.310546875,
      "learning_rate": 9.598595456637642e-06,
      "loss": 1.3460875749588013,
      "step": 248
    },
    {
      "epoch": 0.764525993883792,
      "grad_norm": 0.267578125,
      "learning_rate": 9.336734521240934e-06,
      "loss": 1.2672019004821777,
      "step": 250
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.291015625,
      "learning_rate": 9.080075548057383e-06,
      "loss": 1.3179923295974731,
      "step": 252
    },
    {
      "epoch": 0.7767584097859327,
      "grad_norm": 0.703125,
      "learning_rate": 8.828723970419519e-06,
      "loss": 1.4186333417892456,
      "step": 254
    },
    {
      "epoch": 0.7828746177370031,
      "grad_norm": 0.390625,
      "learning_rate": 8.582783041426728e-06,
      "loss": 1.2970129251480103,
      "step": 256
    },
    {
      "epoch": 0.7889908256880734,
      "grad_norm": 0.21875,
      "learning_rate": 8.342353791529765e-06,
      "loss": 1.3273588418960571,
      "step": 258
    },
    {
      "epoch": 0.7951070336391437,
      "grad_norm": 0.37109375,
      "learning_rate": 8.107534987028286e-06,
      "loss": 1.3187005519866943,
      "step": 260
    },
    {
      "epoch": 0.8012232415902141,
      "grad_norm": 0.345703125,
      "learning_rate": 7.878423089498492e-06,
      "loss": 1.288523554801941,
      "step": 262
    },
    {
      "epoch": 0.8073394495412844,
      "grad_norm": 0.6484375,
      "learning_rate": 7.655112216167533e-06,
      "loss": 1.3323516845703125,
      "step": 264
    },
    {
      "epoch": 0.8134556574923547,
      "grad_norm": 0.345703125,
      "learning_rate": 7.437694101250949e-06,
      "loss": 1.3667104244232178,
      "step": 266
    },
    {
      "epoch": 0.8195718654434251,
      "grad_norm": 0.294921875,
      "learning_rate": 7.226258058269031e-06,
      "loss": 1.354952096939087,
      "step": 268
    },
    {
      "epoch": 0.8256880733944955,
      "grad_norm": 0.42578125,
      "learning_rate": 7.020890943357573e-06,
      "loss": 1.2679226398468018,
      "step": 270
    },
    {
      "epoch": 0.8318042813455657,
      "grad_norm": 0.447265625,
      "learning_rate": 6.821677119588124e-06,
      "loss": 1.3207221031188965,
      "step": 272
    },
    {
      "epoch": 0.8379204892966361,
      "grad_norm": 0.62109375,
      "learning_rate": 6.628698422312341e-06,
      "loss": 1.4085071086883545,
      "step": 274
    },
    {
      "epoch": 0.8440366972477065,
      "grad_norm": 0.474609375,
      "learning_rate": 6.442034125544726e-06,
      "loss": 1.3547308444976807,
      "step": 276
    },
    {
      "epoch": 0.8501529051987767,
      "grad_norm": 0.31640625,
      "learning_rate": 6.261760909397523e-06,
      "loss": 1.346556305885315,
      "step": 278
    },
    {
      "epoch": 0.8562691131498471,
      "grad_norm": 0.46875,
      "learning_rate": 6.087952828581187e-06,
      "loss": 1.3009809255599976,
      "step": 280
    },
    {
      "epoch": 0.8623853211009175,
      "grad_norm": 0.8671875,
      "learning_rate": 5.920681281983339e-06,
      "loss": 1.3994947671890259,
      "step": 282
    },
    {
      "epoch": 0.8685015290519877,
      "grad_norm": 0.23046875,
      "learning_rate": 5.760014983338687e-06,
      "loss": 1.291093111038208,
      "step": 284
    },
    {
      "epoch": 0.8746177370030581,
      "grad_norm": 0.83984375,
      "learning_rate": 5.606019933002025e-06,
      "loss": 1.2565807104110718,
      "step": 286
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.326171875,
      "learning_rate": 5.458759390835851e-06,
      "loss": 1.2513775825500488,
      "step": 288
    },
    {
      "epoch": 0.8868501529051988,
      "grad_norm": 0.49609375,
      "learning_rate": 5.318293850223726e-06,
      "loss": 1.3214612007141113,
      "step": 290
    },
    {
      "epoch": 0.8929663608562691,
      "grad_norm": 0.5859375,
      "learning_rate": 5.184681013220134e-06,
      "loss": 1.4418059587478638,
      "step": 292
    },
    {
      "epoch": 0.8990825688073395,
      "grad_norm": 0.29296875,
      "learning_rate": 5.057975766846941e-06,
      "loss": 1.3208688497543335,
      "step": 294
    },
    {
      "epoch": 0.9051987767584098,
      "grad_norm": 0.66796875,
      "learning_rate": 4.938230160546292e-06,
      "loss": 1.4112577438354492,
      "step": 296
    },
    {
      "epoch": 0.9113149847094801,
      "grad_norm": 0.322265625,
      "learning_rate": 4.825493384799122e-06,
      "loss": 1.356927514076233,
      "step": 298
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 0.357421875,
      "learning_rate": 4.719811750918136e-06,
      "loss": 1.3079042434692383,
      "step": 300
    },
    {
      "epoch": 0.9235474006116208,
      "grad_norm": 0.40234375,
      "learning_rate": 4.621228672023517e-06,
      "loss": 1.2854888439178467,
      "step": 302
    },
    {
      "epoch": 0.9296636085626911,
      "grad_norm": 0.2890625,
      "learning_rate": 4.529784645209172e-06,
      "loss": 1.2977776527404785,
      "step": 304
    },
    {
      "epoch": 0.9357798165137615,
      "grad_norm": 0.30859375,
      "learning_rate": 4.4455172349068696e-06,
      "loss": 1.2907675504684448,
      "step": 306
    },
    {
      "epoch": 0.9418960244648318,
      "grad_norm": 0.53515625,
      "learning_rate": 4.3684610574551e-06,
      "loss": 1.2298390865325928,
      "step": 308
    },
    {
      "epoch": 0.9480122324159022,
      "grad_norm": 0.33984375,
      "learning_rate": 4.298647766878974e-06,
      "loss": 1.3057259321212769,
      "step": 310
    },
    {
      "epoch": 0.9541284403669725,
      "grad_norm": 0.3125,
      "learning_rate": 4.236106041887007e-06,
      "loss": 1.3313639163970947,
      "step": 312
    },
    {
      "epoch": 0.9602446483180428,
      "grad_norm": 0.66015625,
      "learning_rate": 4.180861574090156e-06,
      "loss": 1.3252257108688354,
      "step": 314
    },
    {
      "epoch": 0.9663608562691132,
      "grad_norm": 0.6171875,
      "learning_rate": 4.1329370574479296e-06,
      "loss": 1.258653163909912,
      "step": 316
    },
    {
      "epoch": 0.9724770642201835,
      "grad_norm": 0.6328125,
      "learning_rate": 4.0923521789458876e-06,
      "loss": 1.3827545642852783,
      "step": 318
    },
    {
      "epoch": 0.9785932721712538,
      "grad_norm": 1.2578125,
      "learning_rate": 4.059123610508407e-06,
      "loss": 1.2586948871612549,
      "step": 320
    },
    {
      "epoch": 0.9847094801223242,
      "grad_norm": 0.333984375,
      "learning_rate": 4.033265002149985e-06,
      "loss": 1.1728885173797607,
      "step": 322
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.37109375,
      "learning_rate": 4.014786976367939e-06,
      "loss": 1.230262041091919,
      "step": 324
    },
    {
      "epoch": 0.9969418960244648,
      "grad_norm": 0.78125,
      "learning_rate": 4.00369712377877e-06,
      "loss": 1.3188743591308594,
      "step": 326
    },
    {
      "epoch": 1.0,
      "step": 327,
      "total_flos": 8.183535699912294e+17,
      "train_loss": 1.4031060467437138,
      "train_runtime": 3980.916,
      "train_samples_per_second": 2.629,
      "train_steps_per_second": 0.082
    }
  ],
  "logging_steps": 2,
  "max_steps": 327,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 9999999,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.183535699912294e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}