{
  "best_metric": 1.4133847951889038,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9111617312072893,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004555808656036446,
      "grad_norm": 2.949545383453369,
      "learning_rate": 5e-06,
      "loss": 2.8394,
      "step": 1
    },
    {
      "epoch": 0.004555808656036446,
      "eval_loss": 3.0129241943359375,
      "eval_runtime": 31.3133,
      "eval_samples_per_second": 11.816,
      "eval_steps_per_second": 5.908,
      "step": 1
    },
    {
      "epoch": 0.009111617312072893,
      "grad_norm": 4.037952899932861,
      "learning_rate": 1e-05,
      "loss": 3.1101,
      "step": 2
    },
    {
      "epoch": 0.01366742596810934,
      "grad_norm": 3.4800024032592773,
      "learning_rate": 1.5e-05,
      "loss": 2.7619,
      "step": 3
    },
    {
      "epoch": 0.018223234624145785,
      "grad_norm": 3.447608470916748,
      "learning_rate": 2e-05,
      "loss": 2.839,
      "step": 4
    },
    {
      "epoch": 0.022779043280182234,
      "grad_norm": 2.9829483032226562,
      "learning_rate": 2.5e-05,
      "loss": 2.8275,
      "step": 5
    },
    {
      "epoch": 0.02733485193621868,
      "grad_norm": 2.593158006668091,
      "learning_rate": 3e-05,
      "loss": 2.6591,
      "step": 6
    },
    {
      "epoch": 0.03189066059225513,
      "grad_norm": 2.461625337600708,
      "learning_rate": 3.5e-05,
      "loss": 2.6385,
      "step": 7
    },
    {
      "epoch": 0.03644646924829157,
      "grad_norm": 2.4001970291137695,
      "learning_rate": 4e-05,
      "loss": 2.5091,
      "step": 8
    },
    {
      "epoch": 0.04100227790432802,
      "grad_norm": 2.275493621826172,
      "learning_rate": 4.5e-05,
      "loss": 2.4788,
      "step": 9
    },
    {
      "epoch": 0.04555808656036447,
      "grad_norm": 2.0763912200927734,
      "learning_rate": 5e-05,
      "loss": 2.216,
      "step": 10
    },
    {
      "epoch": 0.05011389521640091,
      "grad_norm": 2.512625217437744,
      "learning_rate": 5.500000000000001e-05,
      "loss": 2.2752,
      "step": 11
    },
    {
      "epoch": 0.05466970387243736,
      "grad_norm": 2.1182351112365723,
      "learning_rate": 6e-05,
      "loss": 2.1124,
      "step": 12
    },
    {
      "epoch": 0.05922551252847381,
      "grad_norm": 2.29195237159729,
      "learning_rate": 6.500000000000001e-05,
      "loss": 2.0274,
      "step": 13
    },
    {
      "epoch": 0.06378132118451026,
      "grad_norm": 2.02005672454834,
      "learning_rate": 7e-05,
      "loss": 2.0314,
      "step": 14
    },
    {
      "epoch": 0.0683371298405467,
      "grad_norm": 1.982064962387085,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.9968,
      "step": 15
    },
    {
      "epoch": 0.07289293849658314,
      "grad_norm": 1.8968816995620728,
      "learning_rate": 8e-05,
      "loss": 1.9468,
      "step": 16
    },
    {
      "epoch": 0.0774487471526196,
      "grad_norm": 2.0738155841827393,
      "learning_rate": 8.5e-05,
      "loss": 2.057,
      "step": 17
    },
    {
      "epoch": 0.08200455580865604,
      "grad_norm": 1.8767244815826416,
      "learning_rate": 9e-05,
      "loss": 2.0008,
      "step": 18
    },
    {
      "epoch": 0.08656036446469248,
      "grad_norm": 1.6860461235046387,
      "learning_rate": 9.5e-05,
      "loss": 1.8924,
      "step": 19
    },
    {
      "epoch": 0.09111617312072894,
      "grad_norm": 1.5266118049621582,
      "learning_rate": 0.0001,
      "loss": 1.9072,
      "step": 20
    },
    {
      "epoch": 0.09567198177676538,
      "grad_norm": 1.6042325496673584,
      "learning_rate": 9.999238475781957e-05,
      "loss": 1.8902,
      "step": 21
    },
    {
      "epoch": 0.10022779043280182,
      "grad_norm": 1.4315247535705566,
      "learning_rate": 9.99695413509548e-05,
      "loss": 1.8952,
      "step": 22
    },
    {
      "epoch": 0.10478359908883828,
      "grad_norm": 1.44978666305542,
      "learning_rate": 9.99314767377287e-05,
      "loss": 1.8573,
      "step": 23
    },
    {
      "epoch": 0.10933940774487472,
      "grad_norm": 1.4212685823440552,
      "learning_rate": 9.987820251299122e-05,
      "loss": 1.8122,
      "step": 24
    },
    {
      "epoch": 0.11389521640091116,
      "grad_norm": 1.4591600894927979,
      "learning_rate": 9.980973490458728e-05,
      "loss": 1.7836,
      "step": 25
    },
    {
      "epoch": 0.11845102505694761,
      "grad_norm": 1.4780867099761963,
      "learning_rate": 9.972609476841367e-05,
      "loss": 1.786,
      "step": 26
    },
    {
      "epoch": 0.12300683371298406,
      "grad_norm": 1.6087100505828857,
      "learning_rate": 9.962730758206611e-05,
      "loss": 1.7888,
      "step": 27
    },
    {
      "epoch": 0.1275626423690205,
      "grad_norm": 1.4750784635543823,
      "learning_rate": 9.951340343707852e-05,
      "loss": 1.7804,
      "step": 28
    },
    {
      "epoch": 0.13211845102505695,
      "grad_norm": 1.4991310834884644,
      "learning_rate": 9.938441702975689e-05,
      "loss": 1.7974,
      "step": 29
    },
    {
      "epoch": 0.1366742596810934,
      "grad_norm": 1.450607419013977,
      "learning_rate": 9.924038765061042e-05,
      "loss": 1.7722,
      "step": 30
    },
    {
      "epoch": 0.14123006833712984,
      "grad_norm": 1.3658661842346191,
      "learning_rate": 9.908135917238321e-05,
      "loss": 1.7674,
      "step": 31
    },
    {
      "epoch": 0.14578587699316628,
      "grad_norm": 1.373909831047058,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.7437,
      "step": 32
    },
    {
      "epoch": 0.15034168564920272,
      "grad_norm": 1.3886319398880005,
      "learning_rate": 9.871850323926177e-05,
      "loss": 1.7025,
      "step": 33
    },
    {
      "epoch": 0.1548974943052392,
      "grad_norm": 1.4907182455062866,
      "learning_rate": 9.851478631379982e-05,
      "loss": 1.7643,
      "step": 34
    },
    {
      "epoch": 0.15945330296127563,
      "grad_norm": 1.2421168088912964,
      "learning_rate": 9.829629131445342e-05,
      "loss": 1.6561,
      "step": 35
    },
    {
      "epoch": 0.16400911161731208,
      "grad_norm": 1.3806560039520264,
      "learning_rate": 9.806308479691595e-05,
      "loss": 1.7309,
      "step": 36
    },
    {
      "epoch": 0.16856492027334852,
      "grad_norm": 1.4004989862442017,
      "learning_rate": 9.781523779815179e-05,
      "loss": 1.713,
      "step": 37
    },
    {
      "epoch": 0.17312072892938496,
      "grad_norm": 1.4071886539459229,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.6977,
      "step": 38
    },
    {
      "epoch": 0.1776765375854214,
      "grad_norm": 1.3282196521759033,
      "learning_rate": 9.727592877996585e-05,
      "loss": 1.6091,
      "step": 39
    },
    {
      "epoch": 0.18223234624145787,
      "grad_norm": 1.3822507858276367,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.8031,
      "step": 40
    },
    {
      "epoch": 0.1867881548974943,
      "grad_norm": 1.4536041021347046,
      "learning_rate": 9.667902132486009e-05,
      "loss": 1.8066,
      "step": 41
    },
    {
      "epoch": 0.19134396355353075,
      "grad_norm": 1.260515809059143,
      "learning_rate": 9.635919272833938e-05,
      "loss": 1.6558,
      "step": 42
    },
    {
      "epoch": 0.1958997722095672,
      "grad_norm": 1.3222765922546387,
      "learning_rate": 9.602524267262203e-05,
      "loss": 1.7395,
      "step": 43
    },
    {
      "epoch": 0.20045558086560364,
      "grad_norm": 1.2623085975646973,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.6857,
      "step": 44
    },
    {
      "epoch": 0.20501138952164008,
      "grad_norm": 1.3057966232299805,
      "learning_rate": 9.53153893518325e-05,
      "loss": 1.6584,
      "step": 45
    },
    {
      "epoch": 0.20956719817767655,
      "grad_norm": 1.2543915510177612,
      "learning_rate": 9.493970231495835e-05,
      "loss": 1.6828,
      "step": 46
    },
    {
      "epoch": 0.214123006833713,
      "grad_norm": 1.345763921737671,
      "learning_rate": 9.45503262094184e-05,
      "loss": 1.6916,
      "step": 47
    },
    {
      "epoch": 0.21867881548974943,
      "grad_norm": 1.3236886262893677,
      "learning_rate": 9.414737964294636e-05,
      "loss": 1.6689,
      "step": 48
    },
    {
      "epoch": 0.22323462414578588,
      "grad_norm": 1.4331655502319336,
      "learning_rate": 9.373098535696979e-05,
      "loss": 1.5996,
      "step": 49
    },
    {
      "epoch": 0.22779043280182232,
      "grad_norm": 1.6575387716293335,
      "learning_rate": 9.330127018922194e-05,
      "loss": 1.5371,
      "step": 50
    },
    {
      "epoch": 0.22779043280182232,
      "eval_loss": 1.6352684497833252,
      "eval_runtime": 31.8367,
      "eval_samples_per_second": 11.622,
      "eval_steps_per_second": 5.811,
      "step": 50
    },
    {
      "epoch": 0.23234624145785876,
      "grad_norm": 1.1424968242645264,
      "learning_rate": 9.285836503510562e-05,
      "loss": 1.8293,
      "step": 51
    },
    {
      "epoch": 0.23690205011389523,
      "grad_norm": 1.1516770124435425,
      "learning_rate": 9.24024048078213e-05,
      "loss": 1.5923,
      "step": 52
    },
    {
      "epoch": 0.24145785876993167,
      "grad_norm": 1.227665662765503,
      "learning_rate": 9.193352839727121e-05,
      "loss": 1.7342,
      "step": 53
    },
    {
      "epoch": 0.2460136674259681,
      "grad_norm": 1.0690141916275024,
      "learning_rate": 9.145187862775209e-05,
      "loss": 1.6914,
      "step": 54
    },
    {
      "epoch": 0.2505694760820046,
      "grad_norm": 1.2659165859222412,
      "learning_rate": 9.09576022144496e-05,
      "loss": 1.6325,
      "step": 55
    },
    {
      "epoch": 0.255125284738041,
      "grad_norm": 1.2316527366638184,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5825,
      "step": 56
    },
    {
      "epoch": 0.25968109339407747,
      "grad_norm": 1.2181825637817383,
      "learning_rate": 8.993177550236464e-05,
      "loss": 1.615,
      "step": 57
    },
    {
      "epoch": 0.2642369020501139,
      "grad_norm": 1.1465332508087158,
      "learning_rate": 8.940053768033609e-05,
      "loss": 1.5696,
      "step": 58
    },
    {
      "epoch": 0.26879271070615035,
      "grad_norm": 1.1993345022201538,
      "learning_rate": 8.885729807284856e-05,
      "loss": 1.57,
      "step": 59
    },
    {
      "epoch": 0.2733485193621868,
      "grad_norm": 1.2644695043563843,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.6788,
      "step": 60
    },
    {
      "epoch": 0.27790432801822323,
      "grad_norm": 1.246644377708435,
      "learning_rate": 8.773547901113862e-05,
      "loss": 1.5384,
      "step": 61
    },
    {
      "epoch": 0.2824601366742597,
      "grad_norm": 1.2264738082885742,
      "learning_rate": 8.715724127386972e-05,
      "loss": 1.505,
      "step": 62
    },
    {
      "epoch": 0.2870159453302961,
      "grad_norm": 1.3458369970321655,
      "learning_rate": 8.656768508095853e-05,
      "loss": 1.4656,
      "step": 63
    },
    {
      "epoch": 0.29157175398633256,
      "grad_norm": 1.2337690591812134,
      "learning_rate": 8.596699001693255e-05,
      "loss": 1.6554,
      "step": 64
    },
    {
      "epoch": 0.296127562642369,
      "grad_norm": 1.191773533821106,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.5623,
      "step": 65
    },
    {
      "epoch": 0.30068337129840544,
      "grad_norm": 1.3478891849517822,
      "learning_rate": 8.473291852294987e-05,
      "loss": 1.5954,
      "step": 66
    },
    {
      "epoch": 0.3052391799544419,
      "grad_norm": 1.1061044931411743,
      "learning_rate": 8.409991800312493e-05,
      "loss": 1.5469,
      "step": 67
    },
    {
      "epoch": 0.3097949886104784,
      "grad_norm": 1.4086905717849731,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.5102,
      "step": 68
    },
    {
      "epoch": 0.3143507972665148,
      "grad_norm": 1.2382744550704956,
      "learning_rate": 8.280295144952536e-05,
      "loss": 1.5571,
      "step": 69
    },
    {
      "epoch": 0.31890660592255127,
      "grad_norm": 1.1993228197097778,
      "learning_rate": 8.213938048432697e-05,
      "loss": 1.4368,
      "step": 70
    },
    {
      "epoch": 0.3234624145785877,
      "grad_norm": 1.2619458436965942,
      "learning_rate": 8.146601955249188e-05,
      "loss": 1.5645,
      "step": 71
    },
    {
      "epoch": 0.32801822323462415,
      "grad_norm": 1.14224374294281,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.4839,
      "step": 72
    },
    {
      "epoch": 0.3325740318906606,
      "grad_norm": 1.2276979684829712,
      "learning_rate": 8.009075115760243e-05,
      "loss": 1.5884,
      "step": 73
    },
    {
      "epoch": 0.33712984054669703,
      "grad_norm": 1.203673005104065,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.6118,
      "step": 74
    },
    {
      "epoch": 0.3416856492027335,
      "grad_norm": 1.2912741899490356,
      "learning_rate": 7.86788218175523e-05,
      "loss": 1.506,
      "step": 75
    },
    {
      "epoch": 0.3462414578587699,
      "grad_norm": 1.325472116470337,
      "learning_rate": 7.795964517353735e-05,
      "loss": 1.5837,
      "step": 76
    },
    {
      "epoch": 0.35079726651480636,
      "grad_norm": 1.308457851409912,
      "learning_rate": 7.723195175075136e-05,
      "loss": 1.5956,
      "step": 77
    },
    {
      "epoch": 0.3553530751708428,
      "grad_norm": 1.253626823425293,
      "learning_rate": 7.649596321166024e-05,
      "loss": 1.6065,
      "step": 78
    },
    {
      "epoch": 0.35990888382687924,
      "grad_norm": 1.2846248149871826,
      "learning_rate": 7.575190374550272e-05,
      "loss": 1.5579,
      "step": 79
    },
    {
      "epoch": 0.36446469248291574,
      "grad_norm": 1.4598257541656494,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.5738,
      "step": 80
    },
    {
      "epoch": 0.3690205011389522,
      "grad_norm": 1.2243446111679077,
      "learning_rate": 7.424048101231686e-05,
      "loss": 1.501,
      "step": 81
    },
    {
      "epoch": 0.3735763097949886,
      "grad_norm": 1.2268356084823608,
      "learning_rate": 7.347357813929454e-05,
      "loss": 1.4573,
      "step": 82
    },
    {
      "epoch": 0.37813211845102507,
      "grad_norm": 1.2605137825012207,
      "learning_rate": 7.269952498697734e-05,
      "loss": 1.4354,
      "step": 83
    },
    {
      "epoch": 0.3826879271070615,
      "grad_norm": 1.2429718971252441,
      "learning_rate": 7.191855733945387e-05,
      "loss": 1.4893,
      "step": 84
    },
    {
      "epoch": 0.38724373576309795,
      "grad_norm": 1.2823604345321655,
      "learning_rate": 7.113091308703498e-05,
      "loss": 1.5339,
      "step": 85
    },
    {
      "epoch": 0.3917995444191344,
      "grad_norm": 1.2435634136199951,
      "learning_rate": 7.033683215379002e-05,
      "loss": 1.4933,
      "step": 86
    },
    {
      "epoch": 0.39635535307517084,
      "grad_norm": 1.3003066778182983,
      "learning_rate": 6.953655642446368e-05,
      "loss": 1.592,
      "step": 87
    },
    {
      "epoch": 0.4009111617312073,
      "grad_norm": 1.2032077312469482,
      "learning_rate": 6.873032967079561e-05,
      "loss": 1.5245,
      "step": 88
    },
    {
      "epoch": 0.4054669703872437,
      "grad_norm": 1.1708099842071533,
      "learning_rate": 6.7918397477265e-05,
      "loss": 1.4524,
      "step": 89
    },
    {
      "epoch": 0.41002277904328016,
      "grad_norm": 1.2689685821533203,
      "learning_rate": 6.710100716628344e-05,
      "loss": 1.5241,
      "step": 90
    },
    {
      "epoch": 0.4145785876993166,
      "grad_norm": 1.2181248664855957,
      "learning_rate": 6.627840772285784e-05,
      "loss": 1.529,
      "step": 91
    },
    {
      "epoch": 0.4191343963553531,
      "grad_norm": 1.2557731866836548,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.5894,
      "step": 92
    },
    {
      "epoch": 0.42369020501138954,
      "grad_norm": 1.2101337909698486,
      "learning_rate": 6.461858523613684e-05,
      "loss": 1.4469,
      "step": 93
    },
    {
      "epoch": 0.428246013667426,
      "grad_norm": 1.3716233968734741,
      "learning_rate": 6.378186779084995e-05,
      "loss": 1.4951,
      "step": 94
    },
    {
      "epoch": 0.4328018223234624,
      "grad_norm": 1.255835771560669,
      "learning_rate": 6.294095225512603e-05,
      "loss": 1.5405,
      "step": 95
    },
    {
      "epoch": 0.43735763097949887,
      "grad_norm": 1.3687196969985962,
      "learning_rate": 6.209609477998338e-05,
      "loss": 1.5678,
      "step": 96
    },
    {
      "epoch": 0.4419134396355353,
      "grad_norm": 1.275315761566162,
      "learning_rate": 6.124755271719325e-05,
      "loss": 1.5774,
      "step": 97
    },
    {
      "epoch": 0.44646924829157175,
      "grad_norm": 1.285813570022583,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 1.5615,
      "step": 98
    },
    {
      "epoch": 0.4510250569476082,
      "grad_norm": 1.4214696884155273,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 1.5977,
      "step": 99
    },
    {
      "epoch": 0.45558086560364464,
      "grad_norm": 1.5818448066711426,
      "learning_rate": 5.868240888334653e-05,
      "loss": 1.3598,
      "step": 100
    },
    {
      "epoch": 0.45558086560364464,
      "eval_loss": 1.5040299892425537,
      "eval_runtime": 31.7152,
      "eval_samples_per_second": 11.666,
      "eval_steps_per_second": 5.833,
      "step": 100
    },
    {
      "epoch": 0.4601366742596811,
      "grad_norm": 1.2834526300430298,
      "learning_rate": 5.782172325201155e-05,
      "loss": 1.6657,
      "step": 101
    },
    {
      "epoch": 0.4646924829157175,
      "grad_norm": 1.2173312902450562,
      "learning_rate": 5.695865504800327e-05,
      "loss": 1.4683,
      "step": 102
    },
    {
      "epoch": 0.46924829157175396,
      "grad_norm": 1.167249321937561,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 1.4075,
      "step": 103
    },
    {
      "epoch": 0.47380410022779046,
      "grad_norm": 1.164792537689209,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.4918,
      "step": 104
    },
    {
      "epoch": 0.4783599088838269,
      "grad_norm": 1.0876809358596802,
      "learning_rate": 5.435778713738292e-05,
      "loss": 1.5412,
      "step": 105
    },
    {
      "epoch": 0.48291571753986334,
      "grad_norm": 1.206875205039978,
      "learning_rate": 5.348782368720626e-05,
      "loss": 1.5219,
      "step": 106
    },
    {
      "epoch": 0.4874715261958998,
      "grad_norm": 1.2055302858352661,
      "learning_rate": 5.26167978121472e-05,
      "loss": 1.4695,
      "step": 107
    },
    {
      "epoch": 0.4920273348519362,
      "grad_norm": 1.2690155506134033,
      "learning_rate": 5.174497483512506e-05,
      "loss": 1.4632,
      "step": 108
    },
    {
      "epoch": 0.49658314350797267,
      "grad_norm": 1.1819018125534058,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 1.4095,
      "step": 109
    },
    {
      "epoch": 0.5011389521640092,
      "grad_norm": 1.2029247283935547,
      "learning_rate": 5e-05,
      "loss": 1.5005,
      "step": 110
    },
    {
      "epoch": 0.5056947608200456,
      "grad_norm": 1.1691571474075317,
      "learning_rate": 4.912737967813583e-05,
      "loss": 1.4709,
      "step": 111
    },
    {
      "epoch": 0.510250569476082,
      "grad_norm": 1.1688123941421509,
      "learning_rate": 4.825502516487497e-05,
      "loss": 1.4511,
      "step": 112
    },
    {
      "epoch": 0.5148063781321185,
      "grad_norm": 1.2863184213638306,
      "learning_rate": 4.738320218785281e-05,
      "loss": 1.5314,
      "step": 113
    },
    {
      "epoch": 0.5193621867881549,
      "grad_norm": 1.1890417337417603,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 1.4349,
      "step": 114
    },
    {
      "epoch": 0.5239179954441914,
      "grad_norm": 1.1162326335906982,
      "learning_rate": 4.564221286261709e-05,
      "loss": 1.4721,
      "step": 115
    },
    {
      "epoch": 0.5284738041002278,
      "grad_norm": 1.2410069704055786,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.469,
      "step": 116
    },
    {
      "epoch": 0.5330296127562643,
      "grad_norm": 1.2136036157608032,
      "learning_rate": 4.390653282974264e-05,
      "loss": 1.5315,
      "step": 117
    },
    {
      "epoch": 0.5375854214123007,
      "grad_norm": 1.136929988861084,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 1.4636,
      "step": 118
    },
    {
      "epoch": 0.5421412300683371,
      "grad_norm": 1.1477383375167847,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 1.5047,
      "step": 119
    },
    {
      "epoch": 0.5466970387243736,
      "grad_norm": 1.1968154907226562,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.391,
      "step": 120
    },
    {
      "epoch": 0.55125284738041,
      "grad_norm": 1.2228323221206665,
      "learning_rate": 4.045955023117276e-05,
      "loss": 1.3253,
      "step": 121
    },
    {
      "epoch": 0.5558086560364465,
      "grad_norm": 1.1717324256896973,
      "learning_rate": 3.960441545911204e-05,
      "loss": 1.3907,
      "step": 122
    },
    {
      "epoch": 0.5603644646924829,
      "grad_norm": 1.1672600507736206,
      "learning_rate": 3.875244728280676e-05,
      "loss": 1.523,
      "step": 123
    },
    {
      "epoch": 0.5649202733485194,
      "grad_norm": 1.1447895765304565,
      "learning_rate": 3.790390522001662e-05,
      "loss": 1.4402,
      "step": 124
    },
    {
      "epoch": 0.5694760820045558,
      "grad_norm": 1.1758711338043213,
      "learning_rate": 3.705904774487396e-05,
      "loss": 1.3618,
      "step": 125
    },
    {
      "epoch": 0.5740318906605922,
      "grad_norm": 1.2545517683029175,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 1.3818,
      "step": 126
    },
    {
      "epoch": 0.5785876993166287,
      "grad_norm": 1.3788303136825562,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 1.4612,
      "step": 127
    },
    {
      "epoch": 0.5831435079726651,
      "grad_norm": 1.2858103513717651,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.4394,
      "step": 128
    },
    {
      "epoch": 0.5876993166287016,
      "grad_norm": 1.2102996110916138,
      "learning_rate": 3.372159227714218e-05,
      "loss": 1.358,
      "step": 129
    },
    {
      "epoch": 0.592255125284738,
      "grad_norm": 1.297158122062683,
      "learning_rate": 3.289899283371657e-05,
      "loss": 1.4077,
      "step": 130
    },
    {
      "epoch": 0.5968109339407744,
      "grad_norm": 1.2475348711013794,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 1.4152,
      "step": 131
    },
    {
      "epoch": 0.6013667425968109,
      "grad_norm": 1.1932103633880615,
      "learning_rate": 3.12696703292044e-05,
      "loss": 1.4332,
      "step": 132
    },
    {
      "epoch": 0.6059225512528473,
      "grad_norm": 1.1948050260543823,
      "learning_rate": 3.046344357553632e-05,
      "loss": 1.3489,
      "step": 133
    },
    {
      "epoch": 0.6104783599088838,
      "grad_norm": 1.3081762790679932,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 1.3885,
      "step": 134
    },
    {
      "epoch": 0.6150341685649203,
      "grad_norm": 1.1615890264511108,
      "learning_rate": 2.886908691296504e-05,
      "loss": 1.4811,
      "step": 135
    },
    {
      "epoch": 0.6195899772209568,
      "grad_norm": 1.265378713607788,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 1.5078,
      "step": 136
    },
    {
      "epoch": 0.6241457858769932,
      "grad_norm": 1.1070181131362915,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 1.4311,
      "step": 137
    },
    {
      "epoch": 0.6287015945330297,
      "grad_norm": 1.2134034633636475,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 1.5122,
      "step": 138
    },
    {
      "epoch": 0.6332574031890661,
      "grad_norm": 1.108559012413025,
      "learning_rate": 2.575951898768315e-05,
      "loss": 1.4781,
      "step": 139
    },
    {
      "epoch": 0.6378132118451025,
      "grad_norm": 1.237922191619873,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.3595,
      "step": 140
    },
    {
      "epoch": 0.642369020501139,
      "grad_norm": 1.1259018182754517,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 1.4241,
      "step": 141
    },
    {
      "epoch": 0.6469248291571754,
      "grad_norm": 1.2384918928146362,
      "learning_rate": 2.350403678833976e-05,
      "loss": 1.49,
      "step": 142
    },
    {
      "epoch": 0.6514806378132119,
      "grad_norm": 1.2415729761123657,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 1.461,
      "step": 143
    },
    {
      "epoch": 0.6560364464692483,
      "grad_norm": 1.2439095973968506,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 1.5559,
      "step": 144
    },
    {
      "epoch": 0.6605922551252847,
      "grad_norm": 1.2855297327041626,
      "learning_rate": 2.132117818244771e-05,
      "loss": 1.4293,
      "step": 145
    },
    {
      "epoch": 0.6651480637813212,
      "grad_norm": 1.1867973804473877,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4501,
      "step": 146
    },
    {
      "epoch": 0.6697038724373576,
      "grad_norm": 1.2252352237701416,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 1.4223,
      "step": 147
    },
    {
      "epoch": 0.6742596810933941,
      "grad_norm": 1.245043158531189,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 1.4481,
      "step": 148
    },
    {
      "epoch": 0.6788154897494305,
      "grad_norm": 1.326700210571289,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 1.4731,
      "step": 149
    },
    {
      "epoch": 0.683371298405467,
      "grad_norm": 1.6518688201904297,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 1.4754,
      "step": 150
    },
    {
      "epoch": 0.683371298405467,
      "eval_loss": 1.4318032264709473,
      "eval_runtime": 31.8179,
      "eval_samples_per_second": 11.629,
      "eval_steps_per_second": 5.814,
      "step": 150
    },
    {
      "epoch": 0.6879271070615034,
      "grad_norm": 1.2807425260543823,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 1.6328,
      "step": 151
    },
    {
      "epoch": 0.6924829157175398,
      "grad_norm": 1.287100911140442,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 1.5518,
      "step": 152
    },
    {
      "epoch": 0.6970387243735763,
      "grad_norm": 1.1534583568572998,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 1.3375,
      "step": 153
    },
    {
      "epoch": 0.7015945330296127,
      "grad_norm": 1.143170714378357,
      "learning_rate": 1.526708147705013e-05,
      "loss": 1.4071,
      "step": 154
    },
    {
      "epoch": 0.7061503416856492,
      "grad_norm": 1.1452003717422485,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 1.3642,
      "step": 155
    },
    {
      "epoch": 0.7107061503416856,
      "grad_norm": 1.151031494140625,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 1.3288,
      "step": 156
    },
    {
      "epoch": 0.715261958997722,
      "grad_norm": 1.065831184387207,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 1.3793,
      "step": 157
    },
    {
      "epoch": 0.7198177676537585,
      "grad_norm": 1.0807403326034546,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 1.4891,
      "step": 158
    },
    {
      "epoch": 0.724373576309795,
      "grad_norm": 1.1514345407485962,
      "learning_rate": 1.22645209888614e-05,
      "loss": 1.4599,
      "step": 159
    },
    {
      "epoch": 0.7289293849658315,
      "grad_norm": 1.070698618888855,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.3903,
      "step": 160
    },
    {
      "epoch": 0.7334851936218679,
      "grad_norm": 1.1703367233276367,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 1.2956,
      "step": 161
    },
    {
      "epoch": 0.7380410022779044,
      "grad_norm": 1.0667177438735962,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 1.5136,
      "step": 162
    },
    {
      "epoch": 0.7425968109339408,
      "grad_norm": 1.1545445919036865,
      "learning_rate": 1.006822449763537e-05,
      "loss": 1.4203,
      "step": 163
    },
    {
      "epoch": 0.7471526195899773,
      "grad_norm": 1.2149230241775513,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4586,
      "step": 164
    },
    {
      "epoch": 0.7517084282460137,
      "grad_norm": 1.156860589981079,
      "learning_rate": 9.042397785550405e-06,
      "loss": 1.4383,
      "step": 165
    },
    {
      "epoch": 0.7562642369020501,
      "grad_norm": 1.16941499710083,
      "learning_rate": 8.548121372247918e-06,
      "loss": 1.3321,
      "step": 166
    },
    {
      "epoch": 0.7608200455580866,
      "grad_norm": 1.1057500839233398,
      "learning_rate": 8.066471602728803e-06,
      "loss": 1.3592,
      "step": 167
    },
    {
      "epoch": 0.765375854214123,
      "grad_norm": 1.1986337900161743,
      "learning_rate": 7.597595192178702e-06,
      "loss": 1.4408,
      "step": 168
    },
    {
      "epoch": 0.7699316628701595,
      "grad_norm": 1.1509459018707275,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 1.3854,
      "step": 169
    },
    {
      "epoch": 0.7744874715261959,
      "grad_norm": 1.0837211608886719,
      "learning_rate": 6.698729810778065e-06,
      "loss": 1.2917,
      "step": 170
    },
    {
      "epoch": 0.7790432801822323,
      "grad_norm": 1.191117286682129,
      "learning_rate": 6.269014643030213e-06,
      "loss": 1.4283,
      "step": 171
    },
    {
      "epoch": 0.7835990888382688,
      "grad_norm": 1.1785354614257812,
      "learning_rate": 5.852620357053651e-06,
      "loss": 1.3538,
      "step": 172
    },
    {
      "epoch": 0.7881548974943052,
      "grad_norm": 1.144243597984314,
      "learning_rate": 5.449673790581611e-06,
      "loss": 1.3326,
      "step": 173
    },
    {
      "epoch": 0.7927107061503417,
      "grad_norm": 1.1665964126586914,
      "learning_rate": 5.060297685041659e-06,
      "loss": 1.3421,
      "step": 174
    },
    {
      "epoch": 0.7972665148063781,
      "grad_norm": 1.104903221130371,
      "learning_rate": 4.684610648167503e-06,
      "loss": 1.3864,
      "step": 175
    },
    {
      "epoch": 0.8018223234624146,
      "grad_norm": 1.070692539215088,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.4014,
      "step": 176
    },
    {
      "epoch": 0.806378132118451,
      "grad_norm": 1.180831789970398,
      "learning_rate": 3.974757327377981e-06,
      "loss": 1.3249,
      "step": 177
    },
    {
      "epoch": 0.8109339407744874,
      "grad_norm": 1.1850495338439941,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 1.2902,
      "step": 178
    },
    {
      "epoch": 0.8154897494305239,
      "grad_norm": 1.3256181478500366,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 1.3133,
      "step": 179
    },
    {
      "epoch": 0.8200455580865603,
      "grad_norm": 1.2605777978897095,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.4665,
      "step": 180
    },
    {
      "epoch": 0.8246013667425968,
      "grad_norm": 1.2295830249786377,
      "learning_rate": 2.724071220034158e-06,
      "loss": 1.3893,
      "step": 181
    },
    {
      "epoch": 0.8291571753986332,
      "grad_norm": 1.1501970291137695,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2773,
      "step": 182
    },
    {
      "epoch": 0.8337129840546698,
      "grad_norm": 1.1711130142211914,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 1.3336,
      "step": 183
    },
    {
      "epoch": 0.8382687927107062,
      "grad_norm": 1.2128851413726807,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 1.4663,
      "step": 184
    },
    {
      "epoch": 0.8428246013667426,
      "grad_norm": 1.1585978269577026,
      "learning_rate": 1.70370868554659e-06,
      "loss": 1.3527,
      "step": 185
    },
    {
      "epoch": 0.8473804100227791,
      "grad_norm": 1.1824872493743896,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 1.3523,
      "step": 186
    },
    {
      "epoch": 0.8519362186788155,
      "grad_norm": 1.1193530559539795,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 1.3684,
      "step": 187
    },
    {
      "epoch": 0.856492027334852,
      "grad_norm": 1.1190437078475952,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 1.421,
      "step": 188
    },
    {
      "epoch": 0.8610478359908884,
      "grad_norm": 1.2521603107452393,
      "learning_rate": 9.186408276168013e-07,
      "loss": 1.4778,
      "step": 189
    },
    {
      "epoch": 0.8656036446469249,
      "grad_norm": 1.1620770692825317,
      "learning_rate": 7.596123493895991e-07,
      "loss": 1.3975,
      "step": 190
    },
    {
      "epoch": 0.8701594533029613,
      "grad_norm": 1.1811784505844116,
      "learning_rate": 6.15582970243117e-07,
      "loss": 1.3969,
      "step": 191
    },
    {
      "epoch": 0.8747152619589977,
      "grad_norm": 1.1842788457870483,
      "learning_rate": 4.865965629214819e-07,
      "loss": 1.4081,
      "step": 192
    },
    {
      "epoch": 0.8792710706150342,
      "grad_norm": 1.317945957183838,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 1.3582,
      "step": 193
    },
    {
      "epoch": 0.8838268792710706,
      "grad_norm": 1.251509666442871,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 1.4463,
      "step": 194
    },
    {
      "epoch": 0.8883826879271071,
      "grad_norm": 1.2192094326019287,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 1.3966,
      "step": 195
    },
    {
      "epoch": 0.8929384965831435,
      "grad_norm": 1.2007484436035156,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 1.4335,
      "step": 196
    },
    {
      "epoch": 0.89749430523918,
      "grad_norm": 1.181261420249939,
      "learning_rate": 6.852326227130834e-08,
      "loss": 1.4807,
      "step": 197
    },
    {
      "epoch": 0.9020501138952164,
      "grad_norm": 1.2215070724487305,
      "learning_rate": 3.04586490452119e-08,
      "loss": 1.4454,
      "step": 198
    },
    {
      "epoch": 0.9066059225512528,
      "grad_norm": 1.405391812324524,
      "learning_rate": 7.615242180436522e-09,
      "loss": 1.4751,
      "step": 199
    },
    {
      "epoch": 0.9111617312072893,
      "grad_norm": 1.6920068264007568,
      "learning_rate": 0.0,
      "loss": 1.4093,
      "step": 200
    },
    {
      "epoch": 0.9111617312072893,
      "eval_loss": 1.4133847951889038,
      "eval_runtime": 31.9435,
      "eval_samples_per_second": 11.583,
      "eval_steps_per_second": 5.791,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.122623108003922e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}