{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.19745502413339183,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013163668275559457,
      "grad_norm": 0.931600163339589,
      "learning_rate": 2e-05,
      "loss": 1.8673,
      "step": 1
    },
    {
      "epoch": 0.0026327336551118913,
      "grad_norm": 0.9936256803709301,
      "learning_rate": 1.999777729859618e-05,
      "loss": 1.8378,
      "step": 2
    },
    {
      "epoch": 0.003949100482667837,
      "grad_norm": 12.214270881503785,
      "learning_rate": 1.9991110182465032e-05,
      "loss": 2.1367,
      "step": 3
    },
    {
      "epoch": 0.005265467310223783,
      "grad_norm": 9.99406173901099,
      "learning_rate": 1.9980001615408228e-05,
      "loss": 2.1295,
      "step": 4
    },
    {
      "epoch": 0.006581834137779728,
      "grad_norm": 6.308728160435039,
      "learning_rate": 1.9964456535631287e-05,
      "loss": 2.056,
      "step": 5
    },
    {
      "epoch": 0.007898200965335674,
      "grad_norm": 4.202724213863458,
      "learning_rate": 1.9944481853548335e-05,
      "loss": 1.9969,
      "step": 6
    },
    {
      "epoch": 0.009214567792891619,
      "grad_norm": 1.9193300273303153,
      "learning_rate": 1.9920086448710162e-05,
      "loss": 1.9613,
      "step": 7
    },
    {
      "epoch": 0.010530934620447565,
      "grad_norm": 1.7647132314008822,
      "learning_rate": 1.9891281165856876e-05,
      "loss": 1.9269,
      "step": 8
    },
    {
      "epoch": 0.01184730144800351,
      "grad_norm": 1.99122961918462,
      "learning_rate": 1.9858078810097004e-05,
      "loss": 1.952,
      "step": 9
    },
    {
      "epoch": 0.013163668275559455,
      "grad_norm": 1.5519401230988923,
      "learning_rate": 1.98204941412151e-05,
      "loss": 1.9451,
      "step": 10
    },
    {
      "epoch": 0.014480035103115402,
      "grad_norm": 1.4544159858418224,
      "learning_rate": 1.9778543867110428e-05,
      "loss": 1.9486,
      "step": 11
    },
    {
      "epoch": 0.015796401930671347,
      "grad_norm": 1.1908276377905163,
      "learning_rate": 1.9732246636369605e-05,
      "loss": 1.9453,
      "step": 12
    },
    {
      "epoch": 0.017112768758227294,
      "grad_norm": 0.8231910135435802,
      "learning_rate": 1.968162302997659e-05,
      "loss": 1.9395,
      "step": 13
    },
    {
      "epoch": 0.018429135585783237,
      "grad_norm": 0.8453846325846259,
      "learning_rate": 1.962669555216358e-05,
      "loss": 1.9239,
      "step": 14
    },
    {
      "epoch": 0.019745502413339184,
      "grad_norm": 0.7899440432391708,
      "learning_rate": 1.9567488620406984e-05,
      "loss": 1.9503,
      "step": 15
    },
    {
      "epoch": 0.02106186924089513,
      "grad_norm": 0.7268699188153452,
      "learning_rate": 1.9504028554572865e-05,
      "loss": 1.9268,
      "step": 16
    },
    {
      "epoch": 0.022378236068451074,
      "grad_norm": 0.7787479899312675,
      "learning_rate": 1.943634356521671e-05,
      "loss": 1.9445,
      "step": 17
    },
    {
      "epoch": 0.02369460289600702,
      "grad_norm": 0.7860314318592968,
      "learning_rate": 1.9364463741042694e-05,
      "loss": 1.9026,
      "step": 18
    },
    {
      "epoch": 0.025010969723562967,
      "grad_norm": 0.8227905867692881,
      "learning_rate": 1.928842103552803e-05,
      "loss": 1.9123,
      "step": 19
    },
    {
      "epoch": 0.02632733655111891,
      "grad_norm": 0.6393575173296501,
      "learning_rate": 1.920824925271838e-05,
      "loss": 1.93,
      "step": 20
    },
    {
      "epoch": 0.027643703378674857,
      "grad_norm": 0.6958779848029143,
      "learning_rate": 1.9123984032200586e-05,
      "loss": 1.9082,
      "step": 21
    },
    {
      "epoch": 0.028960070206230804,
      "grad_norm": 0.8953696131145977,
      "learning_rate": 1.9035662833259433e-05,
      "loss": 1.928,
      "step": 22
    },
    {
      "epoch": 0.030276437033786747,
      "grad_norm": 0.7902542715972343,
      "learning_rate": 1.8943324918225495e-05,
      "loss": 1.9268,
      "step": 23
    },
    {
      "epoch": 0.031592803861342694,
      "grad_norm": 0.6093957180643569,
      "learning_rate": 1.8847011335021447e-05,
      "loss": 1.9127,
      "step": 24
    },
    {
      "epoch": 0.03290917068889864,
      "grad_norm": 0.7877047193824307,
      "learning_rate": 1.874676489891461e-05,
      "loss": 1.917,
      "step": 25
    },
    {
      "epoch": 0.03422553751645459,
      "grad_norm": 0.7553118597870196,
      "learning_rate": 1.8642630173483832e-05,
      "loss": 1.9174,
      "step": 26
    },
    {
      "epoch": 0.03554190434401053,
      "grad_norm": 0.5838893766961445,
      "learning_rate": 1.85346534508092e-05,
      "loss": 1.9217,
      "step": 27
    },
    {
      "epoch": 0.036858271171566474,
      "grad_norm": 0.5986570027695501,
      "learning_rate": 1.8422882730893323e-05,
      "loss": 1.9408,
      "step": 28
    },
    {
      "epoch": 0.03817463799912242,
      "grad_norm": 0.6984361842278407,
      "learning_rate": 1.8307367700323412e-05,
      "loss": 1.9373,
      "step": 29
    },
    {
      "epoch": 0.03949100482667837,
      "grad_norm": 0.6696007840263624,
      "learning_rate": 1.8188159710183595e-05,
      "loss": 1.9074,
      "step": 30
    },
    {
      "epoch": 0.040807371654234315,
      "grad_norm": 0.5310740462756095,
      "learning_rate": 1.8065311753227272e-05,
      "loss": 1.9526,
      "step": 31
    },
    {
      "epoch": 0.04212373848179026,
      "grad_norm": 0.644521653059876,
      "learning_rate": 1.7938878440319722e-05,
      "loss": 1.944,
      "step": 32
    },
    {
      "epoch": 0.0434401053093462,
      "grad_norm": 0.5932362527829019,
      "learning_rate": 1.7808915976161364e-05,
      "loss": 1.9472,
      "step": 33
    },
    {
      "epoch": 0.04475647213690215,
      "grad_norm": 0.5184979812374547,
      "learning_rate": 1.7675482134302503e-05,
      "loss": 1.9276,
      "step": 34
    },
    {
      "epoch": 0.046072838964458095,
      "grad_norm": 0.5314745174137672,
      "learning_rate": 1.753863623146066e-05,
      "loss": 1.9321,
      "step": 35
    },
    {
      "epoch": 0.04738920579201404,
      "grad_norm": 0.6965024427748673,
      "learning_rate": 1.7398439101151908e-05,
      "loss": 1.9186,
      "step": 36
    },
    {
      "epoch": 0.04870557261956999,
      "grad_norm": 0.55694205873126,
      "learning_rate": 1.7254953066647915e-05,
      "loss": 1.9234,
      "step": 37
    },
    {
      "epoch": 0.050021939447125935,
      "grad_norm": 0.5383974099703375,
      "learning_rate": 1.710824191327075e-05,
      "loss": 1.9566,
      "step": 38
    },
    {
      "epoch": 0.051338306274681875,
      "grad_norm": 0.6962830632192784,
      "learning_rate": 1.695837086003772e-05,
      "loss": 1.9326,
      "step": 39
    },
    {
      "epoch": 0.05265467310223782,
      "grad_norm": 0.6107010838431223,
      "learning_rate": 1.680540653066891e-05,
      "loss": 1.9311,
      "step": 40
    },
    {
      "epoch": 0.05397103992979377,
      "grad_norm": 0.5328038996998845,
      "learning_rate": 1.6649416923970248e-05,
      "loss": 1.9317,
      "step": 41
    },
    {
      "epoch": 0.055287406757349715,
      "grad_norm": 0.7445349666429579,
      "learning_rate": 1.649047138360529e-05,
      "loss": 1.9239,
      "step": 42
    },
    {
      "epoch": 0.05660377358490566,
      "grad_norm": 0.6188167997089075,
      "learning_rate": 1.632864056726917e-05,
      "loss": 1.9252,
      "step": 43
    },
    {
      "epoch": 0.05792014041246161,
      "grad_norm": 0.5162155927209101,
      "learning_rate": 1.6163996415278423e-05,
      "loss": 1.9512,
      "step": 44
    },
    {
      "epoch": 0.05923650724001755,
      "grad_norm": 0.6562188783091463,
      "learning_rate": 1.5996612118590604e-05,
      "loss": 1.9316,
      "step": 45
    },
    {
      "epoch": 0.060552874067573495,
      "grad_norm": 0.6054958981957034,
      "learning_rate": 1.5826562086267956e-05,
      "loss": 1.951,
      "step": 46
    },
    {
      "epoch": 0.06186924089512944,
      "grad_norm": 0.5009734563458916,
      "learning_rate": 1.565392191239959e-05,
      "loss": 1.9384,
      "step": 47
    },
    {
      "epoch": 0.06318560772268539,
      "grad_norm": 0.6149774784197523,
      "learning_rate": 1.5478768342496872e-05,
      "loss": 1.9291,
      "step": 48
    },
    {
      "epoch": 0.06450197455024133,
      "grad_norm": 0.5188055532647454,
      "learning_rate": 1.5301179239376936e-05,
      "loss": 1.9437,
      "step": 49
    },
    {
      "epoch": 0.06581834137779728,
      "grad_norm": 0.499242607392427,
      "learning_rate": 1.512123354854955e-05,
      "loss": 1.9233,
      "step": 50
    },
    {
      "epoch": 0.06713470820535322,
      "grad_norm": 0.5826064003666623,
      "learning_rate": 1.4939011263122635e-05,
      "loss": 1.924,
      "step": 51
    },
    {
      "epoch": 0.06845107503290918,
      "grad_norm": 0.4564059505380193,
      "learning_rate": 1.4754593388242117e-05,
      "loss": 1.9338,
      "step": 52
    },
    {
      "epoch": 0.06976744186046512,
      "grad_norm": 0.5034181044200677,
      "learning_rate": 1.4568061905081874e-05,
      "loss": 1.9251,
      "step": 53
    },
    {
      "epoch": 0.07108380868802106,
      "grad_norm": 0.4756818827200354,
      "learning_rate": 1.4379499734399797e-05,
      "loss": 1.9379,
      "step": 54
    },
    {
      "epoch": 0.07240017551557701,
      "grad_norm": 0.4700723456475412,
      "learning_rate": 1.4188990699676186e-05,
      "loss": 1.9349,
      "step": 55
    },
    {
      "epoch": 0.07371654234313295,
      "grad_norm": 0.47287697716336763,
      "learning_rate": 1.3996619489850822e-05,
      "loss": 1.9479,
      "step": 56
    },
    {
      "epoch": 0.0750329091706889,
      "grad_norm": 0.4382295270975528,
      "learning_rate": 1.3802471621675337e-05,
      "loss": 1.9312,
      "step": 57
    },
    {
      "epoch": 0.07634927599824484,
      "grad_norm": 0.46368313988666504,
      "learning_rate": 1.3606633401697557e-05,
      "loss": 1.9541,
      "step": 58
    },
    {
      "epoch": 0.0776656428258008,
      "grad_norm": 0.39742937175675913,
      "learning_rate": 1.340919188789477e-05,
      "loss": 1.9348,
      "step": 59
    },
    {
      "epoch": 0.07898200965335674,
      "grad_norm": 0.47766461658362885,
      "learning_rate": 1.3210234850972966e-05,
      "loss": 1.9534,
      "step": 60
    },
    {
      "epoch": 0.08029837648091268,
      "grad_norm": 0.38294937489900616,
      "learning_rate": 1.300985073534919e-05,
      "loss": 1.9527,
      "step": 61
    },
    {
      "epoch": 0.08161474330846863,
      "grad_norm": 0.45571007516851947,
      "learning_rate": 1.280812861983446e-05,
      "loss": 1.9327,
      "step": 62
    },
    {
      "epoch": 0.08293111013602457,
      "grad_norm": 0.406436332894551,
      "learning_rate": 1.2605158178034656e-05,
      "loss": 1.938,
      "step": 63
    },
    {
      "epoch": 0.08424747696358052,
      "grad_norm": 0.4137618221547989,
      "learning_rate": 1.2401029638486952e-05,
      "loss": 1.9159,
      "step": 64
    },
    {
      "epoch": 0.08556384379113646,
      "grad_norm": 0.4236609580457974,
      "learning_rate": 1.219583374454963e-05,
      "loss": 1.9536,
      "step": 65
    },
    {
      "epoch": 0.0868802106186924,
      "grad_norm": 0.3860737159839257,
      "learning_rate": 1.1989661714063e-05,
      "loss": 1.9707,
      "step": 66
    },
    {
      "epoch": 0.08819657744624836,
      "grad_norm": 0.3975564935817121,
      "learning_rate": 1.1782605198799371e-05,
      "loss": 1.9587,
      "step": 67
    },
    {
      "epoch": 0.0895129442738043,
      "grad_norm": 0.416261230219638,
      "learning_rate": 1.157475624372018e-05,
      "loss": 1.9312,
      "step": 68
    },
    {
      "epoch": 0.09082931110136025,
      "grad_norm": 0.3942676727918781,
      "learning_rate": 1.1366207246058269e-05,
      "loss": 1.9422,
      "step": 69
    },
    {
      "epoch": 0.09214567792891619,
      "grad_norm": 0.399225417996126,
      "learning_rate": 1.1157050914243614e-05,
      "loss": 1.9551,
      "step": 70
    },
    {
      "epoch": 0.09346204475647214,
      "grad_norm": 0.3618119017863118,
      "learning_rate": 1.0947380226690686e-05,
      "loss": 1.9506,
      "step": 71
    },
    {
      "epoch": 0.09477841158402808,
      "grad_norm": 0.397289020381986,
      "learning_rate": 1.0737288390465792e-05,
      "loss": 1.941,
      "step": 72
    },
    {
      "epoch": 0.09609477841158402,
      "grad_norm": 0.36337291366264995,
      "learning_rate": 1.0526868799852797e-05,
      "loss": 1.9642,
      "step": 73
    },
    {
      "epoch": 0.09741114523913998,
      "grad_norm": 0.3959002246907848,
      "learning_rate": 1.031621499483559e-05,
      "loss": 1.9576,
      "step": 74
    },
    {
      "epoch": 0.09872751206669592,
      "grad_norm": 0.3648014496582883,
      "learning_rate": 1.0105420619515798e-05,
      "loss": 1.9488,
      "step": 75
    },
    {
      "epoch": 0.10004387889425187,
      "grad_norm": 0.3842921579639764,
      "learning_rate": 9.894579380484206e-06,
      "loss": 1.965,
      "step": 76
    },
    {
      "epoch": 0.10136024572180781,
      "grad_norm": 0.345316153249247,
      "learning_rate": 9.683785005164412e-06,
      "loss": 1.9703,
      "step": 77
    },
    {
      "epoch": 0.10267661254936375,
      "grad_norm": 0.3588992907840815,
      "learning_rate": 9.473131200147205e-06,
      "loss": 1.977,
      "step": 78
    },
    {
      "epoch": 0.1039929793769197,
      "grad_norm": 0.36828092159317594,
      "learning_rate": 9.262711609534211e-06,
      "loss": 1.9627,
      "step": 79
    },
    {
      "epoch": 0.10530934620447564,
      "grad_norm": 0.3295111270806655,
      "learning_rate": 9.052619773309318e-06,
      "loss": 1.9545,
      "step": 80
    },
    {
      "epoch": 0.1066257130320316,
      "grad_norm": 0.3582482696698915,
      "learning_rate": 8.842949085756389e-06,
      "loss": 1.9502,
      "step": 81
    },
    {
      "epoch": 0.10794207985958754,
      "grad_norm": 0.3362612425719,
      "learning_rate": 8.633792753941733e-06,
      "loss": 1.9606,
      "step": 82
    },
    {
      "epoch": 0.10925844668714349,
      "grad_norm": 0.36579399718337663,
      "learning_rate": 8.425243756279824e-06,
      "loss": 1.9396,
      "step": 83
    },
    {
      "epoch": 0.11057481351469943,
      "grad_norm": 0.32294433921421,
      "learning_rate": 8.217394801200632e-06,
      "loss": 1.9614,
      "step": 84
    },
    {
      "epoch": 0.11189118034225537,
      "grad_norm": 0.313559152207834,
      "learning_rate": 8.010338285937006e-06,
      "loss": 1.9499,
      "step": 85
    },
    {
      "epoch": 0.11320754716981132,
      "grad_norm": 0.34274699619695836,
      "learning_rate": 7.804166255450372e-06,
      "loss": 1.9555,
      "step": 86
    },
    {
      "epoch": 0.11452391399736726,
      "grad_norm": 0.2928430423415806,
      "learning_rate": 7.598970361513052e-06,
      "loss": 1.9598,
      "step": 87
    },
    {
      "epoch": 0.11584028082492322,
      "grad_norm": 0.31563625091713465,
      "learning_rate": 7.394841821965345e-06,
      "loss": 1.9382,
      "step": 88
    },
    {
      "epoch": 0.11715664765247916,
      "grad_norm": 0.30567443374983616,
      "learning_rate": 7.191871380165538e-06,
      "loss": 1.9628,
      "step": 89
    },
    {
      "epoch": 0.1184730144800351,
      "grad_norm": 0.2927442103293411,
      "learning_rate": 6.990149264650814e-06,
      "loss": 1.9548,
      "step": 90
    },
    {
      "epoch": 0.11978938130759105,
      "grad_norm": 0.3151237913315471,
      "learning_rate": 6.789765149027039e-06,
      "loss": 1.9615,
      "step": 91
    },
    {
      "epoch": 0.12110574813514699,
      "grad_norm": 0.27431543131441216,
      "learning_rate": 6.590808112105232e-06,
      "loss": 1.9786,
      "step": 92
    },
    {
      "epoch": 0.12242211496270294,
      "grad_norm": 0.30094963212447634,
      "learning_rate": 6.3933665983024465e-06,
      "loss": 1.9635,
      "step": 93
    },
    {
      "epoch": 0.12373848179025888,
      "grad_norm": 0.33423840592917065,
      "learning_rate": 6.197528378324664e-06,
      "loss": 1.9408,
      "step": 94
    },
    {
      "epoch": 0.12505484861781482,
      "grad_norm": 0.2840583303611351,
      "learning_rate": 6.003380510149179e-06,
      "loss": 1.9691,
      "step": 95
    },
    {
      "epoch": 0.12637121544537078,
      "grad_norm": 0.292965101070784,
      "learning_rate": 5.8110093003238175e-06,
      "loss": 1.992,
      "step": 96
    },
    {
      "epoch": 0.12768758227292673,
      "grad_norm": 0.3178551545098196,
      "learning_rate": 5.620500265600206e-06,
      "loss": 1.9649,
      "step": 97
    },
    {
      "epoch": 0.12900394910048266,
      "grad_norm": 0.279172248724129,
      "learning_rate": 5.431938094918132e-06,
      "loss": 1.9762,
      "step": 98
    },
    {
      "epoch": 0.1303203159280386,
      "grad_norm": 0.2716366040979277,
      "learning_rate": 5.245406611757882e-06,
      "loss": 1.9749,
      "step": 99
    },
    {
      "epoch": 0.13163668275559456,
      "grad_norm": 0.2908349648827661,
      "learning_rate": 5.060988736877366e-06,
      "loss": 1.9564,
      "step": 100
    },
    {
      "epoch": 0.13295304958315052,
      "grad_norm": 0.2758723656083605,
      "learning_rate": 4.878766451450451e-06,
      "loss": 1.9636,
      "step": 101
    },
    {
      "epoch": 0.13426941641070644,
      "grad_norm": 0.2729402567820944,
      "learning_rate": 4.698820760623064e-06,
      "loss": 1.9581,
      "step": 102
    },
    {
      "epoch": 0.1355857832382624,
      "grad_norm": 0.27528121441853837,
      "learning_rate": 4.5212316575031325e-06,
      "loss": 1.9711,
      "step": 103
    },
    {
      "epoch": 0.13690215006581835,
      "grad_norm": 0.2555830734398191,
      "learning_rate": 4.346078087600411e-06,
      "loss": 1.9754,
      "step": 104
    },
    {
      "epoch": 0.13821851689337428,
      "grad_norm": 0.25960054597615906,
      "learning_rate": 4.173437913732048e-06,
      "loss": 1.9727,
      "step": 105
    },
    {
      "epoch": 0.13953488372093023,
      "grad_norm": 0.270030769275592,
      "learning_rate": 4.003387881409397e-06,
      "loss": 1.9771,
      "step": 106
    },
    {
      "epoch": 0.14085125054848618,
      "grad_norm": 0.2543552652913148,
      "learning_rate": 3.836003584721577e-06,
      "loss": 1.9764,
      "step": 107
    },
    {
      "epoch": 0.1421676173760421,
      "grad_norm": 0.2389845165836323,
      "learning_rate": 3.6713594327308343e-06,
      "loss": 1.9618,
      "step": 108
    },
    {
      "epoch": 0.14348398420359806,
      "grad_norm": 0.2512665163994992,
      "learning_rate": 3.509528616394716e-06,
      "loss": 1.9798,
      "step": 109
    },
    {
      "epoch": 0.14480035103115402,
      "grad_norm": 0.2471454297198911,
      "learning_rate": 3.3505830760297543e-06,
      "loss": 1.9758,
      "step": 110
    },
    {
      "epoch": 0.14611671785870997,
      "grad_norm": 0.2392047849987833,
      "learning_rate": 3.1945934693310897e-06,
      "loss": 1.9824,
      "step": 111
    },
    {
      "epoch": 0.1474330846862659,
      "grad_norm": 0.2380319536211479,
      "learning_rate": 3.0416291399622834e-06,
      "loss": 2.0079,
      "step": 112
    },
    {
      "epoch": 0.14874945151382185,
      "grad_norm": 0.24255490283267345,
      "learning_rate": 2.891758086729253e-06,
      "loss": 1.9604,
      "step": 113
    },
    {
      "epoch": 0.1500658183413778,
      "grad_norm": 0.23420260464590595,
      "learning_rate": 2.7450469333520856e-06,
      "loss": 1.9664,
      "step": 114
    },
    {
      "epoch": 0.15138218516893373,
      "grad_norm": 0.2251146463863628,
      "learning_rate": 2.6015608988480956e-06,
      "loss": 1.9709,
      "step": 115
    },
    {
      "epoch": 0.15269855199648968,
      "grad_norm": 0.22744967366950328,
      "learning_rate": 2.4613637685393433e-06,
      "loss": 1.9805,
      "step": 116
    },
    {
      "epoch": 0.15401491882404564,
      "grad_norm": 0.23486834327861175,
      "learning_rate": 2.324517865697501e-06,
      "loss": 1.9544,
      "step": 117
    },
    {
      "epoch": 0.1553312856516016,
      "grad_norm": 0.22043299483360715,
      "learning_rate": 2.19108402383864e-06,
      "loss": 1.9676,
      "step": 118
    },
    {
      "epoch": 0.15664765247915752,
      "grad_norm": 0.21096049925544214,
      "learning_rate": 2.06112155968028e-06,
      "loss": 1.9869,
      "step": 119
    },
    {
      "epoch": 0.15796401930671347,
      "grad_norm": 0.2137583184332987,
      "learning_rate": 1.9346882467727323e-06,
      "loss": 1.9919,
      "step": 120
    },
    {
      "epoch": 0.15928038613426942,
      "grad_norm": 0.22901063931129173,
      "learning_rate": 1.811840289816409e-06,
      "loss": 1.9848,
      "step": 121
    },
    {
      "epoch": 0.16059675296182535,
      "grad_norm": 0.2138281786000561,
      "learning_rate": 1.6926322996765899e-06,
      "loss": 1.986,
      "step": 122
    },
    {
      "epoch": 0.1619131197893813,
      "grad_norm": 0.20893232312607882,
      "learning_rate": 1.5771172691066793e-06,
      "loss": 1.99,
      "step": 123
    },
    {
      "epoch": 0.16322948661693726,
      "grad_norm": 0.20800045900048958,
      "learning_rate": 1.4653465491908003e-06,
      "loss": 2.0134,
      "step": 124
    },
    {
      "epoch": 0.1645458534444932,
      "grad_norm": 0.2206249225627763,
      "learning_rate": 1.3573698265161683e-06,
      "loss": 1.9693,
      "step": 125
    },
    {
      "epoch": 0.16586222027204914,
      "grad_norm": 0.21306519468205834,
      "learning_rate": 1.2532351010853916e-06,
      "loss": 1.9745,
      "step": 126
    },
    {
      "epoch": 0.1671785870996051,
      "grad_norm": 0.2151512836714885,
      "learning_rate": 1.152988664978556e-06,
      "loss": 1.9823,
      "step": 127
    },
    {
      "epoch": 0.16849495392716105,
      "grad_norm": 0.20684164113519568,
      "learning_rate": 1.0566750817745076e-06,
      "loss": 1.9894,
      "step": 128
    },
    {
      "epoch": 0.16981132075471697,
      "grad_norm": 0.19636899537018415,
      "learning_rate": 9.6433716674057e-07,
      "loss": 1.9816,
      "step": 129
    },
    {
      "epoch": 0.17112768758227292,
      "grad_norm": 0.21276425239679414,
      "learning_rate": 8.760159677994174e-07,
      "loss": 1.9862,
      "step": 130
    },
    {
      "epoch": 0.17244405440982888,
      "grad_norm": 0.20614677544311763,
      "learning_rate": 7.91750747281621e-07,
      "loss": 1.9774,
      "step": 131
    },
    {
      "epoch": 0.1737604212373848,
      "grad_norm": 0.20263348493522518,
      "learning_rate": 7.115789644719728e-07,
      "loss": 1.998,
      "step": 132
    },
    {
      "epoch": 0.17507678806494076,
      "grad_norm": 0.19442116908921966,
      "learning_rate": 6.355362589573078e-07,
      "loss": 1.9794,
      "step": 133
    },
    {
      "epoch": 0.1763931548924967,
      "grad_norm": 0.20075119036422134,
      "learning_rate": 5.636564347832907e-07,
      "loss": 1.985,
      "step": 134
    },
    {
      "epoch": 0.17770952172005267,
      "grad_norm": 0.18890178476442018,
      "learning_rate": 4.95971445427137e-07,
      "loss": 1.9861,
      "step": 135
    },
    {
      "epoch": 0.1790258885476086,
      "grad_norm": 0.1903954909200786,
      "learning_rate": 4.3251137959302023e-07,
      "loss": 1.9737,
      "step": 136
    },
    {
      "epoch": 0.18034225537516455,
      "grad_norm": 0.1952295599113416,
      "learning_rate": 3.733044478364234e-07,
      "loss": 1.9702,
      "step": 137
    },
    {
      "epoch": 0.1816586222027205,
      "grad_norm": 0.18425663881349275,
      "learning_rate": 3.1837697002341293e-07,
      "loss": 1.9821,
      "step": 138
    },
    {
      "epoch": 0.18297498903027642,
      "grad_norm": 0.18931999814704614,
      "learning_rate": 2.677533636303964e-07,
      "loss": 1.9723,
      "step": 139
    },
    {
      "epoch": 0.18429135585783238,
      "grad_norm": 0.19474153802676,
      "learning_rate": 2.214561328895748e-07,
      "loss": 1.977,
      "step": 140
    },
    {
      "epoch": 0.18560772268538833,
      "grad_norm": 0.19522045285097447,
      "learning_rate": 1.7950585878489856e-07,
      "loss": 1.9819,
      "step": 141
    },
    {
      "epoch": 0.18692408951294429,
      "grad_norm": 0.19059002126978597,
      "learning_rate": 1.419211899029971e-07,
      "loss": 1.9736,
      "step": 142
    },
    {
      "epoch": 0.1882404563405002,
      "grad_norm": 0.1860427387438694,
      "learning_rate": 1.0871883414312778e-07,
      "loss": 1.9837,
      "step": 143
    },
    {
      "epoch": 0.18955682316805617,
      "grad_norm": 0.18275671614821876,
      "learning_rate": 7.99135512898408e-08,
      "loss": 1.9759,
      "step": 144
    },
    {
      "epoch": 0.19087318999561212,
      "grad_norm": 0.1868194796514628,
      "learning_rate": 5.55181464516652e-08,
      "loss": 1.9779,
      "step": 145
    },
    {
      "epoch": 0.19218955682316805,
      "grad_norm": 0.1877086127216691,
      "learning_rate": 3.554346436871581e-08,
      "loss": 1.973,
      "step": 146
    },
    {
      "epoch": 0.193505923650724,
      "grad_norm": 0.1867320178137634,
      "learning_rate": 1.9998384591773945e-08,
      "loss": 1.9817,
      "step": 147
    },
    {
      "epoch": 0.19482229047827995,
      "grad_norm": 0.18572276222725903,
      "learning_rate": 8.889817534969425e-09,
      "loss": 1.9894,
      "step": 148
    },
    {
      "epoch": 0.1961386573058359,
      "grad_norm": 0.1871653043945299,
      "learning_rate": 2.222701403818972e-09,
      "loss": 1.9783,
      "step": 149
    },
    {
      "epoch": 0.19745502413339183,
      "grad_norm": 0.1978514970425985,
      "learning_rate": 0.0,
      "loss": 1.9906,
      "step": 150
    },
    {
      "epoch": 0.19745502413339183,
      "step": 150,
      "total_flos": 349677715193856.0,
      "train_loss": 1.957608106136322,
      "train_runtime": 5207.1563,
      "train_samples_per_second": 58.074,
      "train_steps_per_second": 0.029
    }
  ],
  "logging_steps": 1,
  "max_steps": 150,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 349677715193856.0,
  "train_batch_size": 42,
  "trial_name": null,
  "trial_params": null
}