{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 13.0,
  "eval_steps": 500,
  "global_step": 130,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "grad_norm": 6.213828556053401,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.8126,
      "step": 1
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.154333884226845,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.8022,
      "step": 2
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.1572939667727455,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.8098,
      "step": 3
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.294139579651079,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.7539,
      "step": 4
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.3706029848954,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.7616,
      "step": 5
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.942953821067203,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.7359,
      "step": 6
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1926733315845555,
      "learning_rate": 5.384615384615385e-06,
      "loss": 0.7421,
      "step": 7
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.15712060860087,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.7103,
      "step": 8
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.0824359711820675,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.7482,
      "step": 9
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.114466388393823,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.7496,
      "step": 10
    },
    {
      "epoch": 1.1,
      "grad_norm": 3.547073532082826,
      "learning_rate": 8.461538461538462e-06,
      "loss": 0.6863,
      "step": 11
    },
    {
      "epoch": 1.2,
      "grad_norm": 3.602494648595526,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.6814,
      "step": 12
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.4115418718917283,
      "learning_rate": 1e-05,
      "loss": 0.6625,
      "step": 13
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5462239429810118,
      "learning_rate": 9.998197638354428e-06,
      "loss": 0.6474,
      "step": 14
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5649561585920715,
      "learning_rate": 9.992791852820709e-06,
      "loss": 0.5914,
      "step": 15
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5948363173382194,
      "learning_rate": 9.983786540671052e-06,
      "loss": 0.5876,
      "step": 16
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4815406402305078,
      "learning_rate": 9.971188194237141e-06,
      "loss": 0.6309,
      "step": 17
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.938212388383564,
      "learning_rate": 9.955005896229543e-06,
      "loss": 0.578,
      "step": 18
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9734449748356621,
      "learning_rate": 9.935251313189564e-06,
      "loss": 0.5647,
      "step": 19
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9819259981094653,
      "learning_rate": 9.911938687078324e-06,
      "loss": 0.5816,
      "step": 20
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8134052660420134,
      "learning_rate": 9.885084825009085e-06,
      "loss": 0.5398,
      "step": 21
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.814985547066869,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.5356,
      "step": 22
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.7953179777799521,
      "learning_rate": 9.820833372667813e-06,
      "loss": 0.5513,
      "step": 23
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.728213802606368,
      "learning_rate": 9.783482104137127e-06,
      "loss": 0.5369,
      "step": 24
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.7872169456302136,
      "learning_rate": 9.742682209735727e-06,
      "loss": 0.5078,
      "step": 25
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8010203465993612,
      "learning_rate": 9.698463103929542e-06,
      "loss": 0.5162,
      "step": 26
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.769280088889877,
      "learning_rate": 9.650856666246693e-06,
      "loss": 0.4947,
      "step": 27
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5859341574412038,
      "learning_rate": 9.599897218294122e-06,
      "loss": 0.5124,
      "step": 28
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.5816734726467704,
      "learning_rate": 9.54562149901362e-06,
      "loss": 0.5056,
      "step": 29
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7300668644724154,
      "learning_rate": 9.488068638195072e-06,
      "loss": 0.4841,
      "step": 30
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7327992005605014,
      "learning_rate": 9.427280128266049e-06,
      "loss": 0.4745,
      "step": 31
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.5260448368656732,
      "learning_rate": 9.363299794378072e-06,
      "loss": 0.4274,
      "step": 32
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.5516276285696556,
      "learning_rate": 9.296173762811084e-06,
      "loss": 0.4348,
      "step": 33
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6279332374253952,
      "learning_rate": 9.225950427718974e-06,
      "loss": 0.4967,
      "step": 34
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.5980462942164727,
      "learning_rate": 9.152680416240059e-06,
      "loss": 0.4435,
      "step": 35
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.627070542516539,
      "learning_rate": 9.076416551997721e-06,
      "loss": 0.4252,
      "step": 36
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.5809234116540586,
      "learning_rate": 8.997213817017508e-06,
      "loss": 0.4449,
      "step": 37
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.49912451212113473,
      "learning_rate": 8.915129312088112e-06,
      "loss": 0.4284,
      "step": 38
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6488932611124092,
      "learning_rate": 8.83022221559489e-06,
      "loss": 0.4495,
      "step": 39
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.5306504057511611,
      "learning_rate": 8.742553740855507e-06,
      "loss": 0.4363,
      "step": 40
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.5444064457418794,
      "learning_rate": 8.652187091988516e-06,
      "loss": 0.3825,
      "step": 41
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.5674875556146136,
      "learning_rate": 8.559187418346703e-06,
      "loss": 0.3918,
      "step": 42
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.5047201758084203,
      "learning_rate": 8.463621767547998e-06,
      "loss": 0.3758,
      "step": 43
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.5580381915226214,
      "learning_rate": 8.36555903713785e-06,
      "loss": 0.3875,
      "step": 44
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.7217300320850437,
      "learning_rate": 8.265069924917925e-06,
      "loss": 0.3852,
      "step": 45
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.6373933677110715,
      "learning_rate": 8.162226877976886e-06,
      "loss": 0.3752,
      "step": 46
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.6003710979548027,
      "learning_rate": 8.057104040460062e-06,
      "loss": 0.373,
      "step": 47
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.6132290637712439,
      "learning_rate": 7.949777200115617e-06,
      "loss": 0.3877,
      "step": 48
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.5500210085624361,
      "learning_rate": 7.84032373365578e-06,
      "loss": 0.3562,
      "step": 49
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.5923411735348952,
      "learning_rate": 7.728822550972523e-06,
      "loss": 0.3677,
      "step": 50
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.5792762609842224,
      "learning_rate": 7.615354038247889e-06,
      "loss": 0.3288,
      "step": 51
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.46857443589189246,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.3185,
      "step": 52
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.5459107909557698,
      "learning_rate": 7.382843600106539e-06,
      "loss": 0.3277,
      "step": 53
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.5492651261355392,
      "learning_rate": 7.263969301848188e-06,
      "loss": 0.3039,
      "step": 54
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.5042252575848233,
      "learning_rate": 7.143462807015271e-06,
      "loss": 0.3181,
      "step": 55
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.47105700677899315,
      "learning_rate": 7.021410994121525e-06,
      "loss": 0.3004,
      "step": 56
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.5183825068120553,
      "learning_rate": 6.897901855769483e-06,
      "loss": 0.3137,
      "step": 57
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.5005555594653843,
      "learning_rate": 6.773024435212678e-06,
      "loss": 0.3158,
      "step": 58
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.5947259967810545,
      "learning_rate": 6.646868762160399e-06,
      "loss": 0.3144,
      "step": 59
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.5773654323680905,
      "learning_rate": 6.519525787871235e-06,
      "loss": 0.3093,
      "step": 60
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.45862803268566965,
      "learning_rate": 6.391087319582264e-06,
      "loss": 0.2596,
      "step": 61
    },
    {
      "epoch": 6.2,
      "grad_norm": 0.5209976192332931,
      "learning_rate": 6.261645954321109e-06,
      "loss": 0.2637,
      "step": 62
    },
    {
      "epoch": 6.3,
      "grad_norm": 0.5835536993737355,
      "learning_rate": 6.131295012148613e-06,
      "loss": 0.2774,
      "step": 63
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.5141309818804918,
      "learning_rate": 6.000128468880223e-06,
      "loss": 0.2611,
      "step": 64
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.49263811191429613,
      "learning_rate": 5.8682408883346535e-06,
      "loss": 0.2653,
      "step": 65
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.4517765356454247,
      "learning_rate": 5.735727354158581e-06,
      "loss": 0.2477,
      "step": 66
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.5166766655496908,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.2496,
      "step": 67
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.48269540363619245,
      "learning_rate": 5.469204947015897e-06,
      "loss": 0.2435,
      "step": 68
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.5202185650742834,
      "learning_rate": 5.335388221955012e-06,
      "loss": 0.2453,
      "step": 69
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.4430595781655239,
      "learning_rate": 5.201329700547077e-06,
      "loss": 0.2455,
      "step": 70
    },
    {
      "epoch": 7.1,
      "grad_norm": 0.6229686155001469,
      "learning_rate": 5.067126031566988e-06,
      "loss": 0.2067,
      "step": 71
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.51241045561568,
      "learning_rate": 4.932873968433014e-06,
      "loss": 0.2087,
      "step": 72
    },
    {
      "epoch": 7.3,
      "grad_norm": 0.6030097583105644,
      "learning_rate": 4.798670299452926e-06,
      "loss": 0.2142,
      "step": 73
    },
    {
      "epoch": 7.4,
      "grad_norm": 0.5327345394316408,
      "learning_rate": 4.664611778044988e-06,
      "loss": 0.2106,
      "step": 74
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.5126096948613228,
      "learning_rate": 4.530795052984104e-06,
      "loss": 0.2213,
      "step": 75
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.5286779949724281,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.2007,
      "step": 76
    },
    {
      "epoch": 7.7,
      "grad_norm": 0.47767513620509805,
      "learning_rate": 4.264272645841419e-06,
      "loss": 0.1936,
      "step": 77
    },
    {
      "epoch": 7.8,
      "grad_norm": 0.5741131317267973,
      "learning_rate": 4.131759111665349e-06,
      "loss": 0.195,
      "step": 78
    },
    {
      "epoch": 7.9,
      "grad_norm": 0.49795272487945746,
      "learning_rate": 3.999871531119779e-06,
      "loss": 0.1966,
      "step": 79
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.5400307593262416,
      "learning_rate": 3.86870498785139e-06,
      "loss": 0.1983,
      "step": 80
    },
    {
      "epoch": 8.1,
      "grad_norm": 0.5535818442441635,
      "learning_rate": 3.7383540456788915e-06,
      "loss": 0.1653,
      "step": 81
    },
    {
      "epoch": 8.2,
      "grad_norm": 0.4714379917172384,
      "learning_rate": 3.6089126804177373e-06,
      "loss": 0.1702,
      "step": 82
    },
    {
      "epoch": 8.3,
      "grad_norm": 0.5055257706755782,
      "learning_rate": 3.480474212128766e-06,
      "loss": 0.1619,
      "step": 83
    },
    {
      "epoch": 8.4,
      "grad_norm": 0.6847862308320251,
      "learning_rate": 3.3531312378396026e-06,
      "loss": 0.167,
      "step": 84
    },
    {
      "epoch": 8.5,
      "grad_norm": 0.48956881082152837,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.1712,
      "step": 85
    },
    {
      "epoch": 8.6,
      "grad_norm": 0.5294008019366655,
      "learning_rate": 3.1020981442305187e-06,
      "loss": 0.1451,
      "step": 86
    },
    {
      "epoch": 8.7,
      "grad_norm": 0.49270828281468143,
      "learning_rate": 2.978589005878476e-06,
      "loss": 0.1635,
      "step": 87
    },
    {
      "epoch": 8.8,
      "grad_norm": 0.5314326789397427,
      "learning_rate": 2.8565371929847286e-06,
      "loss": 0.1735,
      "step": 88
    },
    {
      "epoch": 8.9,
      "grad_norm": 0.4421832249096492,
      "learning_rate": 2.736030698151815e-06,
      "loss": 0.1541,
      "step": 89
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.44835619913007113,
      "learning_rate": 2.6171563998934605e-06,
      "loss": 0.1559,
      "step": 90
    },
    {
      "epoch": 9.1,
      "grad_norm": 0.5173257985928487,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.1521,
      "step": 91
    },
    {
      "epoch": 9.2,
      "grad_norm": 0.4396778178140979,
      "learning_rate": 2.384645961752113e-06,
      "loss": 0.1334,
      "step": 92
    },
    {
      "epoch": 9.3,
      "grad_norm": 0.46174209988956766,
      "learning_rate": 2.2711774490274767e-06,
      "loss": 0.1323,
      "step": 93
    },
    {
      "epoch": 9.4,
      "grad_norm": 0.5613702403140401,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.1368,
      "step": 94
    },
    {
      "epoch": 9.5,
      "grad_norm": 0.4633004031782555,
      "learning_rate": 2.050222799884387e-06,
      "loss": 0.1255,
      "step": 95
    },
    {
      "epoch": 9.6,
      "grad_norm": 0.4565672863354383,
      "learning_rate": 1.942895959539939e-06,
      "loss": 0.1314,
      "step": 96
    },
    {
      "epoch": 9.7,
      "grad_norm": 0.4923358761521844,
      "learning_rate": 1.8377731220231144e-06,
      "loss": 0.1299,
      "step": 97
    },
    {
      "epoch": 9.8,
      "grad_norm": 0.458004170448298,
      "learning_rate": 1.7349300750820758e-06,
      "loss": 0.1243,
      "step": 98
    },
    {
      "epoch": 9.9,
      "grad_norm": 0.46559037329342007,
      "learning_rate": 1.6344409628621482e-06,
      "loss": 0.1276,
      "step": 99
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.4380097095916762,
      "learning_rate": 1.5363782324520033e-06,
      "loss": 0.1286,
      "step": 100
    },
    {
      "epoch": 10.1,
      "grad_norm": 0.442216596442426,
      "learning_rate": 1.4408125816532981e-06,
      "loss": 0.119,
      "step": 101
    },
    {
      "epoch": 10.2,
      "grad_norm": 0.43183118447199603,
      "learning_rate": 1.347812908011485e-06,
      "loss": 0.1147,
      "step": 102
    },
    {
      "epoch": 10.3,
      "grad_norm": 0.4976656252992892,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.1112,
      "step": 103
    },
    {
      "epoch": 10.4,
      "grad_norm": 0.4459411745789807,
      "learning_rate": 1.1697777844051105e-06,
      "loss": 0.1173,
      "step": 104
    },
    {
      "epoch": 10.5,
      "grad_norm": 0.4259652626626087,
      "learning_rate": 1.0848706879118893e-06,
      "loss": 0.1109,
      "step": 105
    },
    {
      "epoch": 10.6,
      "grad_norm": 0.49172108505168904,
      "learning_rate": 1.0027861829824953e-06,
      "loss": 0.1089,
      "step": 106
    },
    {
      "epoch": 10.7,
      "grad_norm": 0.40859757667110025,
      "learning_rate": 9.235834480022788e-07,
      "loss": 0.1164,
      "step": 107
    },
    {
      "epoch": 10.8,
      "grad_norm": 0.4212026016262429,
      "learning_rate": 8.473195837599419e-07,
      "loss": 0.1136,
      "step": 108
    },
    {
      "epoch": 10.9,
      "grad_norm": 0.39894351993907856,
      "learning_rate": 7.740495722810271e-07,
      "loss": 0.1182,
      "step": 109
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.40209745536625574,
      "learning_rate": 7.03826237188916e-07,
      "loss": 0.1049,
      "step": 110
    },
    {
      "epoch": 11.1,
      "grad_norm": 0.4176269822151388,
      "learning_rate": 6.367002056219285e-07,
      "loss": 0.1047,
      "step": 111
    },
    {
      "epoch": 11.2,
      "grad_norm": 0.3770894497677813,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.0989,
      "step": 112
    },
    {
      "epoch": 11.3,
      "grad_norm": 0.37185322258031683,
      "learning_rate": 5.119313618049309e-07,
      "loss": 0.105,
      "step": 113
    },
    {
      "epoch": 11.4,
      "grad_norm": 0.3787089272773232,
      "learning_rate": 4.54378500986381e-07,
      "loss": 0.1031,
      "step": 114
    },
    {
      "epoch": 11.5,
      "grad_norm": 0.37053957439562274,
      "learning_rate": 4.001027817058789e-07,
      "loss": 0.0958,
      "step": 115
    },
    {
      "epoch": 11.6,
      "grad_norm": 0.4058964057092488,
      "learning_rate": 3.49143333753309e-07,
      "loss": 0.1055,
      "step": 116
    },
    {
      "epoch": 11.7,
      "grad_norm": 0.44760223355416123,
      "learning_rate": 3.015368960704584e-07,
      "loss": 0.1087,
      "step": 117
    },
    {
      "epoch": 11.8,
      "grad_norm": 0.39348297234500734,
      "learning_rate": 2.573177902642726e-07,
      "loss": 0.1034,
      "step": 118
    },
    {
      "epoch": 11.9,
      "grad_norm": 0.39098840867794804,
      "learning_rate": 2.1651789586287442e-07,
      "loss": 0.1061,
      "step": 119
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.35717784050826373,
      "learning_rate": 1.7916662733218848e-07,
      "loss": 0.1057,
      "step": 120
    },
    {
      "epoch": 12.1,
      "grad_norm": 0.34549284152384824,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.0984,
      "step": 121
    },
    {
      "epoch": 12.2,
      "grad_norm": 0.34888472975728496,
      "learning_rate": 1.1491517499091498e-07,
      "loss": 0.1036,
      "step": 122
    },
    {
      "epoch": 12.3,
      "grad_norm": 0.3516501694296099,
      "learning_rate": 8.80613129216762e-08,
      "loss": 0.0909,
      "step": 123
    },
    {
      "epoch": 12.4,
      "grad_norm": 0.3416301373346077,
      "learning_rate": 6.474868681043578e-08,
      "loss": 0.108,
      "step": 124
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.3276043064294318,
      "learning_rate": 4.499410377045765e-08,
      "loss": 0.0921,
      "step": 125
    },
    {
      "epoch": 12.6,
      "grad_norm": 0.410104272303672,
      "learning_rate": 2.8811805762860578e-08,
      "loss": 0.0972,
      "step": 126
    },
    {
      "epoch": 12.7,
      "grad_norm": 0.34136860185356355,
      "learning_rate": 1.6213459328950355e-08,
      "loss": 0.1032,
      "step": 127
    },
    {
      "epoch": 12.8,
      "grad_norm": 0.33954461079046544,
      "learning_rate": 7.2081471792911914e-09,
      "loss": 0.1072,
      "step": 128
    },
    {
      "epoch": 12.9,
      "grad_norm": 0.33742204552690996,
      "learning_rate": 1.8023616455731253e-09,
      "loss": 0.096,
      "step": 129
    },
    {
      "epoch": 13.0,
      "grad_norm": 0.35495441917915715,
      "learning_rate": 0.0,
      "loss": 0.1033,
      "step": 130
    },
    {
      "epoch": 13.0,
      "step": 130,
      "total_flos": 8.889376463808102e+16,
      "train_loss": 0.31648237000291163,
      "train_runtime": 2072.3636,
      "train_samples_per_second": 1.982,
      "train_steps_per_second": 0.063
    }
  ],
  "logging_steps": 1,
  "max_steps": 130,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 13,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.889376463808102e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}