{
  "best_global_step": 60003,
  "best_metric": 0.8089721345660946,
  "best_model_checkpoint": "./doc_type_v2_primary_model_multilingual-e5-small/checkpoint-60003",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 60003,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.024998750062496876,
      "grad_norm": 17.081697463989258,
      "learning_rate": 4.9584187457293806e-05,
      "loss": 1.8537,
      "step": 500
    },
    {
      "epoch": 0.04999750012499375,
      "grad_norm": 16.341184616088867,
      "learning_rate": 4.9167541622918856e-05,
      "loss": 1.3289,
      "step": 1000
    },
    {
      "epoch": 0.07499625018749062,
      "grad_norm": 12.614828109741211,
      "learning_rate": 4.875089578854391e-05,
      "loss": 1.1698,
      "step": 1500
    },
    {
      "epoch": 0.0999950002499875,
      "grad_norm": 17.94846534729004,
      "learning_rate": 4.833424995416896e-05,
      "loss": 1.0996,
      "step": 2000
    },
    {
      "epoch": 0.12499375031248437,
      "grad_norm": 9.764547348022461,
      "learning_rate": 4.7917604119794014e-05,
      "loss": 1.0552,
      "step": 2500
    },
    {
      "epoch": 0.14999250037498124,
      "grad_norm": 5.973393440246582,
      "learning_rate": 4.7500958285419064e-05,
      "loss": 1.0462,
      "step": 3000
    },
    {
      "epoch": 0.17499125043747812,
      "grad_norm": 5.258781909942627,
      "learning_rate": 4.7084312451044115e-05,
      "loss": 1.0004,
      "step": 3500
    },
    {
      "epoch": 0.199990000499975,
      "grad_norm": 5.401681423187256,
      "learning_rate": 4.666766661666917e-05,
      "loss": 0.9812,
      "step": 4000
    },
    {
      "epoch": 0.2249887505624719,
      "grad_norm": 3.4015467166900635,
      "learning_rate": 4.625102078229422e-05,
      "loss": 0.9245,
      "step": 4500
    },
    {
      "epoch": 0.24998750062496874,
      "grad_norm": 11.498674392700195,
      "learning_rate": 4.583437494791927e-05,
      "loss": 0.9282,
      "step": 5000
    },
    {
      "epoch": 0.2749862506874656,
      "grad_norm": 6.841133117675781,
      "learning_rate": 4.541772911354433e-05,
      "loss": 0.9167,
      "step": 5500
    },
    {
      "epoch": 0.2999850007499625,
      "grad_norm": 5.397707939147949,
      "learning_rate": 4.500108327916937e-05,
      "loss": 0.8886,
      "step": 6000
    },
    {
      "epoch": 0.3249837508124594,
      "grad_norm": 7.148469924926758,
      "learning_rate": 4.458443744479443e-05,
      "loss": 0.8826,
      "step": 6500
    },
    {
      "epoch": 0.34998250087495625,
      "grad_norm": 3.2729530334472656,
      "learning_rate": 4.416779161041948e-05,
      "loss": 0.8443,
      "step": 7000
    },
    {
      "epoch": 0.3749812509374531,
      "grad_norm": 12.553752899169922,
      "learning_rate": 4.375114577604453e-05,
      "loss": 0.8374,
      "step": 7500
    },
    {
      "epoch": 0.39998000099995,
      "grad_norm": 9.571837425231934,
      "learning_rate": 4.333449994166959e-05,
      "loss": 0.8271,
      "step": 8000
    },
    {
      "epoch": 0.42497875106244687,
      "grad_norm": 11.265901565551758,
      "learning_rate": 4.291785410729464e-05,
      "loss": 0.8306,
      "step": 8500
    },
    {
      "epoch": 0.4499775011249438,
      "grad_norm": 18.747684478759766,
      "learning_rate": 4.250120827291969e-05,
      "loss": 0.8561,
      "step": 9000
    },
    {
      "epoch": 0.47497625118744063,
      "grad_norm": 7.2989726066589355,
      "learning_rate": 4.208456243854474e-05,
      "loss": 0.7851,
      "step": 9500
    },
    {
      "epoch": 0.4999750012499375,
      "grad_norm": 21.371959686279297,
      "learning_rate": 4.1667916604169796e-05,
      "loss": 0.7841,
      "step": 10000
    },
    {
      "epoch": 0.5249737513124344,
      "grad_norm": 19.508371353149414,
      "learning_rate": 4.1251270769794846e-05,
      "loss": 0.7678,
      "step": 10500
    },
    {
      "epoch": 0.5499725013749313,
      "grad_norm": 5.09838342666626,
      "learning_rate": 4.0834624935419896e-05,
      "loss": 0.7538,
      "step": 11000
    },
    {
      "epoch": 0.5749712514374281,
      "grad_norm": 6.288057804107666,
      "learning_rate": 4.041797910104495e-05,
      "loss": 0.735,
      "step": 11500
    },
    {
      "epoch": 0.599970001499925,
      "grad_norm": 2.406168222427368,
      "learning_rate": 4.000133326667e-05,
      "loss": 0.774,
      "step": 12000
    },
    {
      "epoch": 0.6249687515624219,
      "grad_norm": 11.135022163391113,
      "learning_rate": 3.9584687432295054e-05,
      "loss": 0.7368,
      "step": 12500
    },
    {
      "epoch": 0.6499675016249188,
      "grad_norm": 16.766277313232422,
      "learning_rate": 3.916804159792011e-05,
      "loss": 0.7435,
      "step": 13000
    },
    {
      "epoch": 0.6749662516874156,
      "grad_norm": 7.3794121742248535,
      "learning_rate": 3.8751395763545154e-05,
      "loss": 0.7035,
      "step": 13500
    },
    {
      "epoch": 0.6999650017499125,
      "grad_norm": 13.058135032653809,
      "learning_rate": 3.833474992917021e-05,
      "loss": 0.7552,
      "step": 14000
    },
    {
      "epoch": 0.7249637518124094,
      "grad_norm": 13.570932388305664,
      "learning_rate": 3.791810409479526e-05,
      "loss": 0.7443,
      "step": 14500
    },
    {
      "epoch": 0.7499625018749062,
      "grad_norm": 16.705114364624023,
      "learning_rate": 3.750145826042031e-05,
      "loss": 0.7461,
      "step": 15000
    },
    {
      "epoch": 0.7749612519374032,
      "grad_norm": 20.24770164489746,
      "learning_rate": 3.708481242604537e-05,
      "loss": 0.7352,
      "step": 15500
    },
    {
      "epoch": 0.7999600019999,
      "grad_norm": 10.8892183303833,
      "learning_rate": 3.666816659167042e-05,
      "loss": 0.6946,
      "step": 16000
    },
    {
      "epoch": 0.8249587520623969,
      "grad_norm": 24.564472198486328,
      "learning_rate": 3.625152075729547e-05,
      "loss": 0.6939,
      "step": 16500
    },
    {
      "epoch": 0.8499575021248937,
      "grad_norm": 14.484394073486328,
      "learning_rate": 3.583487492292053e-05,
      "loss": 0.7509,
      "step": 17000
    },
    {
      "epoch": 0.8749562521873906,
      "grad_norm": 11.327393531799316,
      "learning_rate": 3.541822908854558e-05,
      "loss": 0.6992,
      "step": 17500
    },
    {
      "epoch": 0.8999550022498876,
      "grad_norm": 12.824069023132324,
      "learning_rate": 3.500158325417063e-05,
      "loss": 0.7043,
      "step": 18000
    },
    {
      "epoch": 0.9249537523123844,
      "grad_norm": 1.3452341556549072,
      "learning_rate": 3.458493741979568e-05,
      "loss": 0.6977,
      "step": 18500
    },
    {
      "epoch": 0.9499525023748813,
      "grad_norm": 7.985979080200195,
      "learning_rate": 3.416829158542073e-05,
      "loss": 0.6952,
      "step": 19000
    },
    {
      "epoch": 0.9749512524373781,
      "grad_norm": 6.591372489929199,
      "learning_rate": 3.3751645751045785e-05,
      "loss": 0.708,
      "step": 19500
    },
    {
      "epoch": 0.999950002499875,
      "grad_norm": 4.785042762756348,
      "learning_rate": 3.3334999916670835e-05,
      "loss": 0.6695,
      "step": 20000
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.7876339482882986,
      "eval_loss": 0.6957715749740601,
      "eval_runtime": 12.0347,
      "eval_samples_per_second": 1662.524,
      "eval_steps_per_second": 207.815,
      "step": 20001
    },
    {
      "epoch": 1.024948752562372,
      "grad_norm": 15.502031326293945,
      "learning_rate": 3.2918354082295885e-05,
      "loss": 0.5363,
      "step": 20500
    },
    {
      "epoch": 1.0499475026248688,
      "grad_norm": 0.9488680362701416,
      "learning_rate": 3.2501708247920936e-05,
      "loss": 0.547,
      "step": 21000
    },
    {
      "epoch": 1.0749462526873657,
      "grad_norm": 4.085986614227295,
      "learning_rate": 3.208506241354599e-05,
      "loss": 0.5733,
      "step": 21500
    },
    {
      "epoch": 1.0999450027498625,
      "grad_norm": 15.25266170501709,
      "learning_rate": 3.166841657917104e-05,
      "loss": 0.5454,
      "step": 22000
    },
    {
      "epoch": 1.1249437528123594,
      "grad_norm": 11.815897941589355,
      "learning_rate": 3.125177074479609e-05,
      "loss": 0.5235,
      "step": 22500
    },
    {
      "epoch": 1.1499425028748562,
      "grad_norm": 17.311704635620117,
      "learning_rate": 3.083512491042115e-05,
      "loss": 0.5291,
      "step": 23000
    },
    {
      "epoch": 1.174941252937353,
      "grad_norm": 7.48703145980835,
      "learning_rate": 3.0418479076046197e-05,
      "loss": 0.5537,
      "step": 23500
    },
    {
      "epoch": 1.19994000299985,
      "grad_norm": 0.3721858263015747,
      "learning_rate": 3.000183324167125e-05,
      "loss": 0.555,
      "step": 24000
    },
    {
      "epoch": 1.2249387530623468,
      "grad_norm": 22.23200035095215,
      "learning_rate": 2.9585187407296305e-05,
      "loss": 0.5338,
      "step": 24500
    },
    {
      "epoch": 1.2499375031248436,
      "grad_norm": 2.753875255584717,
      "learning_rate": 2.9168541572921355e-05,
      "loss": 0.5615,
      "step": 25000
    },
    {
      "epoch": 1.2749362531873407,
      "grad_norm": 23.020252227783203,
      "learning_rate": 2.875189573854641e-05,
      "loss": 0.5155,
      "step": 25500
    },
    {
      "epoch": 1.2999350032498376,
      "grad_norm": 31.79548454284668,
      "learning_rate": 2.8335249904171456e-05,
      "loss": 0.5353,
      "step": 26000
    },
    {
      "epoch": 1.3249337533123344,
      "grad_norm": 0.2923097312450409,
      "learning_rate": 2.7918604069796513e-05,
      "loss": 0.5317,
      "step": 26500
    },
    {
      "epoch": 1.3499325033748313,
      "grad_norm": 9.347312927246094,
      "learning_rate": 2.7501958235421566e-05,
      "loss": 0.5429,
      "step": 27000
    },
    {
      "epoch": 1.3749312534373281,
      "grad_norm": 13.638419151306152,
      "learning_rate": 2.7085312401046613e-05,
      "loss": 0.5311,
      "step": 27500
    },
    {
      "epoch": 1.399930003499825,
      "grad_norm": 19.09702491760254,
      "learning_rate": 2.6668666566671667e-05,
      "loss": 0.5345,
      "step": 28000
    },
    {
      "epoch": 1.4249287535623218,
      "grad_norm": 0.6322915554046631,
      "learning_rate": 2.6252020732296717e-05,
      "loss": 0.5287,
      "step": 28500
    },
    {
      "epoch": 1.4499275036248187,
      "grad_norm": 19.159151077270508,
      "learning_rate": 2.583537489792177e-05,
      "loss": 0.5204,
      "step": 29000
    },
    {
      "epoch": 1.4749262536873156,
      "grad_norm": 0.7778434753417969,
      "learning_rate": 2.5418729063546824e-05,
      "loss": 0.5121,
      "step": 29500
    },
    {
      "epoch": 1.4999250037498126,
      "grad_norm": 20.512577056884766,
      "learning_rate": 2.5002083229171875e-05,
      "loss": 0.52,
      "step": 30000
    },
    {
      "epoch": 1.5249237538123093,
      "grad_norm": 8.87389087677002,
      "learning_rate": 2.458543739479693e-05,
      "loss": 0.5094,
      "step": 30500
    },
    {
      "epoch": 1.5499225038748063,
      "grad_norm": 21.17337989807129,
      "learning_rate": 2.416879156042198e-05,
      "loss": 0.5169,
      "step": 31000
    },
    {
      "epoch": 1.574921253937303,
      "grad_norm": 8.69658374786377,
      "learning_rate": 2.3752145726047032e-05,
      "loss": 0.5226,
      "step": 31500
    },
    {
      "epoch": 1.5999200039998,
      "grad_norm": 1.2267570495605469,
      "learning_rate": 2.3335499891672083e-05,
      "loss": 0.5281,
      "step": 32000
    },
    {
      "epoch": 1.624918754062297,
      "grad_norm": 14.757322311401367,
      "learning_rate": 2.2918854057297136e-05,
      "loss": 0.5246,
      "step": 32500
    },
    {
      "epoch": 1.6499175041247938,
      "grad_norm": 6.141539096832275,
      "learning_rate": 2.250220822292219e-05,
      "loss": 0.532,
      "step": 33000
    },
    {
      "epoch": 1.6749162541872906,
      "grad_norm": 15.90838623046875,
      "learning_rate": 2.208556238854724e-05,
      "loss": 0.5068,
      "step": 33500
    },
    {
      "epoch": 1.6999150042497875,
      "grad_norm": 3.071305751800537,
      "learning_rate": 2.166891655417229e-05,
      "loss": 0.4971,
      "step": 34000
    },
    {
      "epoch": 1.7249137543122843,
      "grad_norm": 5.962382793426514,
      "learning_rate": 2.1252270719797344e-05,
      "loss": 0.5122,
      "step": 34500
    },
    {
      "epoch": 1.7499125043747812,
      "grad_norm": 5.9214911460876465,
      "learning_rate": 2.0835624885422398e-05,
      "loss": 0.489,
      "step": 35000
    },
    {
      "epoch": 1.7749112544372783,
      "grad_norm": 8.897248268127441,
      "learning_rate": 2.0418979051047448e-05,
      "loss": 0.479,
      "step": 35500
    },
    {
      "epoch": 1.799910004499775,
      "grad_norm": 16.03746223449707,
      "learning_rate": 2.0002333216672502e-05,
      "loss": 0.4919,
      "step": 36000
    },
    {
      "epoch": 1.824908754562272,
      "grad_norm": 21.669597625732422,
      "learning_rate": 1.9585687382297552e-05,
      "loss": 0.4974,
      "step": 36500
    },
    {
      "epoch": 1.8499075046247686,
      "grad_norm": 3.668883800506592,
      "learning_rate": 1.9169041547922606e-05,
      "loss": 0.5045,
      "step": 37000
    },
    {
      "epoch": 1.8749062546872657,
      "grad_norm": 4.8963518142700195,
      "learning_rate": 1.8752395713547656e-05,
      "loss": 0.525,
      "step": 37500
    },
    {
      "epoch": 1.8999050047497625,
      "grad_norm": 19.771133422851562,
      "learning_rate": 1.833574987917271e-05,
      "loss": 0.4748,
      "step": 38000
    },
    {
      "epoch": 1.9249037548122594,
      "grad_norm": 20.69668960571289,
      "learning_rate": 1.791910404479776e-05,
      "loss": 0.4831,
      "step": 38500
    },
    {
      "epoch": 1.9499025048747562,
      "grad_norm": 3.1742944717407227,
      "learning_rate": 1.750245821042281e-05,
      "loss": 0.5091,
      "step": 39000
    },
    {
      "epoch": 1.974901254937253,
      "grad_norm": 0.3630174696445465,
      "learning_rate": 1.7085812376047867e-05,
      "loss": 0.4821,
      "step": 39500
    },
    {
      "epoch": 1.99990000499975,
      "grad_norm": 10.60681438446045,
      "learning_rate": 1.6669166541672918e-05,
      "loss": 0.4862,
      "step": 40000
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.7969511995029,
      "eval_loss": 0.7491226196289062,
      "eval_runtime": 12.147,
      "eval_samples_per_second": 1647.158,
      "eval_steps_per_second": 205.895,
      "step": 40002
    },
    {
      "epoch": 2.024898755062247,
      "grad_norm": 18.80621910095215,
      "learning_rate": 1.6252520707297968e-05,
      "loss": 0.357,
      "step": 40500
    },
    {
      "epoch": 2.049897505124744,
      "grad_norm": 3.8872764110565186,
      "learning_rate": 1.5835874872923022e-05,
      "loss": 0.333,
      "step": 41000
    },
    {
      "epoch": 2.0748962551872405,
      "grad_norm": 19.08934211730957,
      "learning_rate": 1.5419229038548072e-05,
      "loss": 0.374,
      "step": 41500
    },
    {
      "epoch": 2.0998950052497376,
      "grad_norm": 10.449114799499512,
      "learning_rate": 1.5002583204173126e-05,
      "loss": 0.3698,
      "step": 42000
    },
    {
      "epoch": 2.1248937553122342,
      "grad_norm": 6.660628318786621,
      "learning_rate": 1.4585937369798178e-05,
      "loss": 0.3759,
      "step": 42500
    },
    {
      "epoch": 2.1498925053747313,
      "grad_norm": 9.793807983398438,
      "learning_rate": 1.416929153542323e-05,
      "loss": 0.3543,
      "step": 43000
    },
    {
      "epoch": 2.174891255437228,
      "grad_norm": 20.215002059936523,
      "learning_rate": 1.375264570104828e-05,
      "loss": 0.3695,
      "step": 43500
    },
    {
      "epoch": 2.199890005499725,
      "grad_norm": 20.272212982177734,
      "learning_rate": 1.3335999866673335e-05,
      "loss": 0.3385,
      "step": 44000
    },
    {
      "epoch": 2.2248887555622217,
      "grad_norm": 12.721766471862793,
      "learning_rate": 1.2919354032298386e-05,
      "loss": 0.3583,
      "step": 44500
    },
    {
      "epoch": 2.2498875056247187,
      "grad_norm": 11.291624069213867,
      "learning_rate": 1.2502708197923438e-05,
      "loss": 0.3445,
      "step": 45000
    },
    {
      "epoch": 2.274886255687216,
      "grad_norm": 14.476861000061035,
      "learning_rate": 1.208606236354849e-05,
      "loss": 0.3575,
      "step": 45500
    },
    {
      "epoch": 2.2998850057497124,
      "grad_norm": 8.20272159576416,
      "learning_rate": 1.1669416529173542e-05,
      "loss": 0.3382,
      "step": 46000
    },
    {
      "epoch": 2.3248837558122095,
      "grad_norm": 2.5329763889312744,
      "learning_rate": 1.1252770694798594e-05,
      "loss": 0.3732,
      "step": 46500
    },
    {
      "epoch": 2.349882505874706,
      "grad_norm": 0.9955561757087708,
      "learning_rate": 1.0836124860423647e-05,
      "loss": 0.3454,
      "step": 47000
    },
    {
      "epoch": 2.3748812559372032,
      "grad_norm": 6.986231803894043,
      "learning_rate": 1.0419479026048697e-05,
      "loss": 0.3563,
      "step": 47500
    },
    {
      "epoch": 2.3998800059997,
      "grad_norm": 21.110620498657227,
      "learning_rate": 1.000283319167375e-05,
      "loss": 0.3302,
      "step": 48000
    },
    {
      "epoch": 2.424878756062197,
      "grad_norm": 0.08908458799123764,
      "learning_rate": 9.586187357298801e-06,
      "loss": 0.3421,
      "step": 48500
    },
    {
      "epoch": 2.4498775061246936,
      "grad_norm": 13.181462287902832,
      "learning_rate": 9.169541522923853e-06,
      "loss": 0.3119,
      "step": 49000
    },
    {
      "epoch": 2.4748762561871906,
      "grad_norm": 12.58914852142334,
      "learning_rate": 8.752895688548907e-06,
      "loss": 0.3578,
      "step": 49500
    },
    {
      "epoch": 2.4998750062496873,
      "grad_norm": 39.47843551635742,
      "learning_rate": 8.336249854173957e-06,
      "loss": 0.3584,
      "step": 50000
    },
    {
      "epoch": 2.5248737563121844,
      "grad_norm": 4.305168628692627,
      "learning_rate": 7.919604019799011e-06,
      "loss": 0.3142,
      "step": 50500
    },
    {
      "epoch": 2.5498725063746814,
      "grad_norm": 0.7413849830627441,
      "learning_rate": 7.502958185424062e-06,
      "loss": 0.3124,
      "step": 51000
    },
    {
      "epoch": 2.574871256437178,
      "grad_norm": 1.338671326637268,
      "learning_rate": 7.086312351049114e-06,
      "loss": 0.3262,
      "step": 51500
    },
    {
      "epoch": 2.599870006499675,
      "grad_norm": 26.348230361938477,
      "learning_rate": 6.669666516674167e-06,
      "loss": 0.3072,
      "step": 52000
    },
    {
      "epoch": 2.624868756562172,
      "grad_norm": 38.16984558105469,
      "learning_rate": 6.253020682299218e-06,
      "loss": 0.3274,
      "step": 52500
    },
    {
      "epoch": 2.649867506624669,
      "grad_norm": 13.00293254852295,
      "learning_rate": 5.83637484792427e-06,
      "loss": 0.3131,
      "step": 53000
    },
    {
      "epoch": 2.6748662566871655,
      "grad_norm": 3.519160270690918,
      "learning_rate": 5.419729013549323e-06,
      "loss": 0.3281,
      "step": 53500
    },
    {
      "epoch": 2.6998650067496626,
      "grad_norm": 15.743597984313965,
      "learning_rate": 5.003083179174375e-06,
      "loss": 0.3108,
      "step": 54000
    },
    {
      "epoch": 2.7248637568121596,
      "grad_norm": 20.438329696655273,
      "learning_rate": 4.586437344799427e-06,
      "loss": 0.3189,
      "step": 54500
    },
    {
      "epoch": 2.7498625068746563,
      "grad_norm": 45.14103317260742,
      "learning_rate": 4.169791510424479e-06,
      "loss": 0.3367,
      "step": 55000
    },
    {
      "epoch": 2.774861256937153,
      "grad_norm": 3.860975980758667,
      "learning_rate": 3.7531456760495313e-06,
      "loss": 0.2969,
      "step": 55500
    },
    {
      "epoch": 2.79986000699965,
      "grad_norm": 0.40173372626304626,
      "learning_rate": 3.3364998416745833e-06,
      "loss": 0.3332,
      "step": 56000
    },
    {
      "epoch": 2.824858757062147,
      "grad_norm": 2.1133482456207275,
      "learning_rate": 2.9198540072996353e-06,
      "loss": 0.3197,
      "step": 56500
    },
    {
      "epoch": 2.8498575071246437,
      "grad_norm": 25.709867477416992,
      "learning_rate": 2.5032081729246873e-06,
      "loss": 0.312,
      "step": 57000
    },
    {
      "epoch": 2.8748562571871408,
      "grad_norm": 22.588973999023438,
      "learning_rate": 2.0865623385497392e-06,
      "loss": 0.3275,
      "step": 57500
    },
    {
      "epoch": 2.8998550072496374,
      "grad_norm": 2.185502529144287,
      "learning_rate": 1.6699165041747914e-06,
      "loss": 0.2933,
      "step": 58000
    },
    {
      "epoch": 2.9248537573121345,
      "grad_norm": 12.381799697875977,
      "learning_rate": 1.2532706697998434e-06,
      "loss": 0.3123,
      "step": 58500
    },
    {
      "epoch": 2.949852507374631,
      "grad_norm": 0.39924994111061096,
      "learning_rate": 8.366248354248955e-07,
      "loss": 0.3045,
      "step": 59000
    },
    {
      "epoch": 2.974851257437128,
      "grad_norm": 16.00220489501953,
      "learning_rate": 4.199790010499475e-07,
      "loss": 0.2928,
      "step": 59500
    },
    {
      "epoch": 2.9998500074996253,
      "grad_norm": 2.6532301902770996,
      "learning_rate": 3.3331666749995837e-09,
      "loss": 0.3199,
      "step": 60000
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.8089721345660946,
      "eval_loss": 0.8624263405799866,
      "eval_runtime": 10.7103,
      "eval_samples_per_second": 1868.104,
      "eval_steps_per_second": 233.513,
      "step": 60003
    },
    {
      "epoch": 3.0,
      "step": 60003,
      "total_flos": 7908189620438016.0,
      "train_loss": 0.5708606184445773,
      "train_runtime": 1693.148,
      "train_samples_per_second": 283.503,
      "train_steps_per_second": 35.439
    }
  ],
  "logging_steps": 500,
  "max_steps": 60003,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 7908189620438016.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}