trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 100.0,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6288470372071658,
      "eval_f1": 0.43084004602991943,
      "eval_loss": 1.2868107557296753,
      "eval_precision": 0.3991471215351812,
      "eval_recall": 0.468,
      "eval_runtime": 1.4977,
      "eval_samples_per_second": 36.054,
      "eval_steps_per_second": 0.668,
      "step": 20
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7672255397335783,
      "eval_f1": 0.7671957671957673,
      "eval_loss": 0.8454405665397644,
      "eval_precision": 0.7391102873030584,
      "eval_recall": 0.7975,
      "eval_runtime": 2.3492,
      "eval_samples_per_second": 22.986,
      "eval_steps_per_second": 0.426,
      "step": 40
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7731970601745521,
      "eval_f1": 0.8189655172413792,
      "eval_loss": 0.6648961305618286,
      "eval_precision": 0.7858455882352942,
      "eval_recall": 0.855,
      "eval_runtime": 2.4309,
      "eval_samples_per_second": 22.214,
      "eval_steps_per_second": 0.411,
      "step": 60
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.8240698208543867,
      "eval_f1": 0.8516630250060694,
      "eval_loss": 0.6003230810165405,
      "eval_precision": 0.827748938178386,
      "eval_recall": 0.877,
      "eval_runtime": 1.946,
      "eval_samples_per_second": 27.749,
      "eval_steps_per_second": 0.514,
      "step": 80
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.8000689021589343,
      "eval_f1": 0.8560497369679579,
      "eval_loss": 0.5940394401550293,
      "eval_precision": 0.8203483043079743,
      "eval_recall": 0.895,
      "eval_runtime": 1.5738,
      "eval_samples_per_second": 34.312,
      "eval_steps_per_second": 0.635,
      "step": 100
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.8124712907671107,
      "eval_f1": 0.8579682233991335,
      "eval_loss": 0.6028273701667786,
      "eval_precision": 0.8272980501392758,
      "eval_recall": 0.891,
      "eval_runtime": 2.3361,
      "eval_samples_per_second": 23.116,
      "eval_steps_per_second": 0.428,
      "step": 120
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.8300413412953606,
      "eval_f1": 0.8839373163565133,
      "eval_loss": 0.5925479531288147,
      "eval_precision": 0.8661228406909789,
      "eval_recall": 0.9025,
      "eval_runtime": 1.549,
      "eval_samples_per_second": 34.86,
      "eval_steps_per_second": 0.646,
      "step": 140
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.8147680293982544,
      "eval_f1": 0.864694471387003,
      "eval_loss": 0.6319019198417664,
      "eval_precision": 0.8394538606403014,
      "eval_recall": 0.8915,
      "eval_runtime": 1.7343,
      "eval_samples_per_second": 31.136,
      "eval_steps_per_second": 0.577,
      "step": 160
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.8295819935691319,
      "eval_f1": 0.8697776691913022,
      "eval_loss": 0.660470187664032,
      "eval_precision": 0.8504538939321548,
      "eval_recall": 0.89,
      "eval_runtime": 1.7198,
      "eval_samples_per_second": 31.398,
      "eval_steps_per_second": 0.581,
      "step": 180
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.8275149288011024,
      "eval_f1": 0.8871046228710462,
      "eval_loss": 0.6778170466423035,
      "eval_precision": 0.8639810426540284,
      "eval_recall": 0.9115,
      "eval_runtime": 1.5366,
      "eval_samples_per_second": 35.142,
      "eval_steps_per_second": 0.651,
      "step": 200
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.815686724850712,
      "eval_f1": 0.8839615668883962,
      "eval_loss": 0.7490019798278809,
      "eval_precision": 0.8712967459932006,
      "eval_recall": 0.897,
      "eval_runtime": 2.3087,
      "eval_samples_per_second": 23.39,
      "eval_steps_per_second": 0.433,
      "step": 220
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.8222324299494718,
      "eval_f1": 0.8948667324777887,
      "eval_loss": 0.7746959924697876,
      "eval_precision": 0.8835282651072125,
      "eval_recall": 0.9065,
      "eval_runtime": 1.9661,
      "eval_samples_per_second": 27.465,
      "eval_steps_per_second": 0.509,
      "step": 240
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.8058107487367937,
      "eval_f1": 0.8912883435582821,
      "eval_loss": 0.8429032564163208,
      "eval_precision": 0.8751807228915662,
      "eval_recall": 0.908,
      "eval_runtime": 1.9643,
      "eval_samples_per_second": 27.491,
      "eval_steps_per_second": 0.509,
      "step": 260
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.8129306384933395,
      "eval_f1": 0.8899308983218163,
      "eval_loss": 0.8374262452125549,
      "eval_precision": 0.8786549707602339,
      "eval_recall": 0.9015,
      "eval_runtime": 1.7097,
      "eval_samples_per_second": 31.584,
      "eval_steps_per_second": 0.585,
      "step": 280
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 0.8112080845199816,
      "eval_f1": 0.8747830399206545,
      "eval_loss": 0.9091736078262329,
      "eval_precision": 0.867683226758485,
      "eval_recall": 0.882,
      "eval_runtime": 2.3751,
      "eval_samples_per_second": 22.736,
      "eval_steps_per_second": 0.421,
      "step": 300
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.8288929719797887,
      "eval_f1": 0.8963474827245804,
      "eval_loss": 0.8785933256149292,
      "eval_precision": 0.884990253411306,
      "eval_recall": 0.908,
      "eval_runtime": 2.0542,
      "eval_samples_per_second": 26.288,
      "eval_steps_per_second": 0.487,
      "step": 320
    },
    {
      "epoch": 34.0,
      "eval_accuracy": 0.8307303628847037,
      "eval_f1": 0.8931750741839762,
      "eval_loss": 0.91584312915802,
      "eval_precision": 0.8835616438356164,
      "eval_recall": 0.903,
      "eval_runtime": 2.4582,
      "eval_samples_per_second": 21.967,
      "eval_steps_per_second": 0.407,
      "step": 340
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.8259072117593018,
      "eval_f1": 0.8948534843634572,
      "eval_loss": 0.9160082340240479,
      "eval_precision": 0.8816108685104318,
      "eval_recall": 0.9085,
      "eval_runtime": 2.3705,
      "eval_samples_per_second": 22.78,
      "eval_steps_per_second": 0.422,
      "step": 360
    },
    {
      "epoch": 38.0,
      "eval_accuracy": 0.8171796049609554,
      "eval_f1": 0.8942850134903116,
      "eval_loss": 0.9378513693809509,
      "eval_precision": 0.8777082330284064,
      "eval_recall": 0.9115,
      "eval_runtime": 2.2021,
      "eval_samples_per_second": 24.522,
      "eval_steps_per_second": 0.454,
      "step": 380
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.8079926504363804,
      "eval_f1": 0.8961134197017844,
      "eval_loss": 0.9751215577125549,
      "eval_precision": 0.8766140602582496,
      "eval_recall": 0.9165,
      "eval_runtime": 1.9472,
      "eval_samples_per_second": 27.733,
      "eval_steps_per_second": 0.514,
      "step": 400
    },
    {
      "epoch": 42.0,
      "eval_accuracy": 0.8140790078089113,
      "eval_f1": 0.8976067110782137,
      "eval_loss": 0.922682523727417,
      "eval_precision": 0.8860204578665368,
      "eval_recall": 0.9095,
      "eval_runtime": 1.9496,
      "eval_samples_per_second": 27.699,
      "eval_steps_per_second": 0.513,
      "step": 420
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.8154570509875976,
      "eval_f1": 0.8868017795353437,
      "eval_loss": 1.0090957880020142,
      "eval_precision": 0.8768328445747801,
      "eval_recall": 0.897,
      "eval_runtime": 1.9314,
      "eval_samples_per_second": 27.96,
      "eval_steps_per_second": 0.518,
      "step": 440
    },
    {
      "epoch": 46.0,
      "eval_accuracy": 0.8182131373449701,
      "eval_f1": 0.8956933034602937,
      "eval_loss": 0.9963611960411072,
      "eval_precision": 0.8919186911254338,
      "eval_recall": 0.8995,
      "eval_runtime": 2.4248,
      "eval_samples_per_second": 22.27,
      "eval_steps_per_second": 0.412,
      "step": 460
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.8203950390445567,
      "eval_f1": 0.896329365079365,
      "eval_loss": 1.0006074905395508,
      "eval_precision": 0.8892716535433071,
      "eval_recall": 0.9035,
      "eval_runtime": 2.4269,
      "eval_samples_per_second": 22.25,
      "eval_steps_per_second": 0.412,
      "step": 480
    },
    {
      "epoch": 50.0,
      "learning_rate": 5e-06,
      "loss": 0.235,
      "step": 500
    },
    {
      "epoch": 50.0,
      "eval_accuracy": 0.8178686265502986,
      "eval_f1": 0.8960199004975123,
      "eval_loss": 1.0283308029174805,
      "eval_precision": 0.8915841584158416,
      "eval_recall": 0.9005,
      "eval_runtime": 2.2977,
      "eval_samples_per_second": 23.502,
      "eval_steps_per_second": 0.435,
      "step": 500
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.827859439595774,
      "eval_f1": 0.8951014349332014,
      "eval_loss": 0.9926251769065857,
      "eval_precision": 0.885896180215475,
      "eval_recall": 0.9045,
      "eval_runtime": 1.6933,
      "eval_samples_per_second": 31.891,
      "eval_steps_per_second": 0.591,
      "step": 520
    },
    {
      "epoch": 54.0,
      "eval_accuracy": 0.8228066146072577,
      "eval_f1": 0.8948148148148148,
      "eval_loss": 1.0112966299057007,
      "eval_precision": 0.8839024390243903,
      "eval_recall": 0.906,
      "eval_runtime": 2.4054,
      "eval_samples_per_second": 22.449,
      "eval_steps_per_second": 0.416,
      "step": 540
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.8290078089113458,
      "eval_f1": 0.8993819530284302,
      "eval_loss": 1.004191279411316,
      "eval_precision": 0.8894865525672372,
      "eval_recall": 0.9095,
      "eval_runtime": 1.6522,
      "eval_samples_per_second": 32.684,
      "eval_steps_per_second": 0.605,
      "step": 560
    },
    {
      "epoch": 58.0,
      "eval_accuracy": 0.8225769407441433,
      "eval_f1": 0.8922392486406326,
      "eval_loss": 1.0357481241226196,
      "eval_precision": 0.8822091886608016,
      "eval_recall": 0.9025,
      "eval_runtime": 2.1102,
      "eval_samples_per_second": 25.59,
      "eval_steps_per_second": 0.474,
      "step": 580
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.8178686265502986,
      "eval_f1": 0.901213171577123,
      "eval_loss": 1.0394996404647827,
      "eval_precision": 0.8925944090240314,
      "eval_recall": 0.91,
      "eval_runtime": 2.4208,
      "eval_samples_per_second": 22.307,
      "eval_steps_per_second": 0.413,
      "step": 600
    },
    {
      "epoch": 62.0,
      "eval_accuracy": 0.8201653651814423,
      "eval_f1": 0.8971962616822431,
      "eval_loss": 1.004025936126709,
      "eval_precision": 0.882865440464666,
      "eval_recall": 0.912,
      "eval_runtime": 2.4098,
      "eval_samples_per_second": 22.408,
      "eval_steps_per_second": 0.415,
      "step": 620
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.8195911805236564,
      "eval_f1": 0.8941929133858268,
      "eval_loss": 1.029054880142212,
      "eval_precision": 0.8803294573643411,
      "eval_recall": 0.9085,
      "eval_runtime": 2.4017,
      "eval_samples_per_second": 22.484,
      "eval_steps_per_second": 0.416,
      "step": 640
    },
    {
      "epoch": 66.0,
      "eval_accuracy": 0.8279742765273312,
      "eval_f1": 0.8993055555555556,
      "eval_loss": 1.075648546218872,
      "eval_precision": 0.8922244094488189,
      "eval_recall": 0.9065,
      "eval_runtime": 2.4111,
      "eval_samples_per_second": 22.396,
      "eval_steps_per_second": 0.415,
      "step": 660
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.8234956361966008,
      "eval_f1": 0.8966716343765524,
      "eval_loss": 1.1056932210922241,
      "eval_precision": 0.8909180651530109,
      "eval_recall": 0.9025,
      "eval_runtime": 2.4045,
      "eval_samples_per_second": 22.458,
      "eval_steps_per_second": 0.416,
      "step": 680
    },
    {
      "epoch": 70.0,
      "eval_accuracy": 0.8268259072117593,
      "eval_f1": 0.9007407407407408,
      "eval_loss": 1.1429905891418457,
      "eval_precision": 0.8897560975609756,
      "eval_recall": 0.912,
      "eval_runtime": 2.3943,
      "eval_samples_per_second": 22.553,
      "eval_steps_per_second": 0.418,
      "step": 700
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.8221175930179145,
      "eval_f1": 0.9012012748222604,
      "eval_loss": 1.0474272966384888,
      "eval_precision": 0.8840788840788841,
      "eval_recall": 0.919,
      "eval_runtime": 2.6325,
      "eval_samples_per_second": 20.512,
      "eval_steps_per_second": 0.38,
      "step": 720
    },
    {
      "epoch": 74.0,
      "eval_accuracy": 0.8191318327974276,
      "eval_f1": 0.9079694053787319,
      "eval_loss": 1.1182180643081665,
      "eval_precision": 0.8962493911349245,
      "eval_recall": 0.92,
      "eval_runtime": 2.4698,
      "eval_samples_per_second": 21.864,
      "eval_steps_per_second": 0.405,
      "step": 740
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.8267110702802021,
      "eval_f1": 0.9056324110671936,
      "eval_loss": 1.1421239376068115,
      "eval_precision": 0.89501953125,
      "eval_recall": 0.9165,
      "eval_runtime": 2.0412,
      "eval_samples_per_second": 26.455,
      "eval_steps_per_second": 0.49,
      "step": 760
    },
    {
      "epoch": 78.0,
      "eval_accuracy": 0.825447864033073,
      "eval_f1": 0.9112103174603173,
      "eval_loss": 1.1723241806030273,
      "eval_precision": 0.9040354330708661,
      "eval_recall": 0.9185,
      "eval_runtime": 2.0338,
      "eval_samples_per_second": 26.551,
      "eval_steps_per_second": 0.492,
      "step": 780
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.8314193844740468,
      "eval_f1": 0.9091358024691357,
      "eval_loss": 1.0977429151535034,
      "eval_precision": 0.8980487804878049,
      "eval_recall": 0.9205,
      "eval_runtime": 2.708,
      "eval_samples_per_second": 19.941,
      "eval_steps_per_second": 0.369,
      "step": 800
    },
    {
      "epoch": 82.0,
      "eval_accuracy": 0.826596233348645,
      "eval_f1": 0.9020771513353116,
      "eval_loss": 1.1165635585784912,
      "eval_precision": 0.8923679060665362,
      "eval_recall": 0.912,
      "eval_runtime": 2.3964,
      "eval_samples_per_second": 22.533,
      "eval_steps_per_second": 0.417,
      "step": 820
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.83210840606339,
      "eval_f1": 0.9058561897702001,
      "eval_loss": 1.1296281814575195,
      "eval_precision": 0.8954567659990229,
      "eval_recall": 0.9165,
      "eval_runtime": 1.5143,
      "eval_samples_per_second": 35.659,
      "eval_steps_per_second": 0.66,
      "step": 840
    },
    {
      "epoch": 86.0,
      "eval_accuracy": 0.8313045475424896,
      "eval_f1": 0.9070687098368759,
      "eval_loss": 1.122943639755249,
      "eval_precision": 0.896871945259042,
      "eval_recall": 0.9175,
      "eval_runtime": 2.2961,
      "eval_samples_per_second": 23.518,
      "eval_steps_per_second": 0.436,
      "step": 860
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.8325677537896188,
      "eval_f1": 0.9080573405832921,
      "eval_loss": 1.1123418807983398,
      "eval_precision": 0.8978494623655914,
      "eval_recall": 0.9185,
      "eval_runtime": 2.1658,
      "eval_samples_per_second": 24.933,
      "eval_steps_per_second": 0.462,
      "step": 880
    },
    {
      "epoch": 90.0,
      "eval_accuracy": 0.8325677537896188,
      "eval_f1": 0.9037843185753152,
      "eval_loss": 1.1032230854034424,
      "eval_precision": 0.8942731277533039,
      "eval_recall": 0.9135,
      "eval_runtime": 1.6625,
      "eval_samples_per_second": 32.481,
      "eval_steps_per_second": 0.601,
      "step": 900
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.83578318787322,
      "eval_f1": 0.9086407526615499,
      "eval_loss": 1.0933949947357178,
      "eval_precision": 0.8999509563511525,
      "eval_recall": 0.9175,
      "eval_runtime": 2.2261,
      "eval_samples_per_second": 24.258,
      "eval_steps_per_second": 0.449,
      "step": 920
    },
    {
      "epoch": 94.0,
      "eval_accuracy": 0.8360128617363344,
      "eval_f1": 0.9090909090909091,
      "eval_loss": 1.0976922512054443,
      "eval_precision": 0.9008345606283751,
      "eval_recall": 0.9175,
      "eval_runtime": 2.4337,
      "eval_samples_per_second": 22.189,
      "eval_steps_per_second": 0.411,
      "step": 940
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.833945796968305,
      "eval_f1": 0.9066205533596838,
      "eval_loss": 1.1002885103225708,
      "eval_precision": 0.89599609375,
      "eval_recall": 0.9175,
      "eval_runtime": 2.3183,
      "eval_samples_per_second": 23.293,
      "eval_steps_per_second": 0.431,
      "step": 960
    },
    {
      "epoch": 98.0,
      "eval_accuracy": 0.8329122645842904,
      "eval_f1": 0.9057164068299927,
      "eval_loss": 1.1073625087738037,
      "eval_precision": 0.8966193042626164,
      "eval_recall": 0.915,
      "eval_runtime": 2.4241,
      "eval_samples_per_second": 22.276,
      "eval_steps_per_second": 0.413,
      "step": 980
    },
    {
      "epoch": 100.0,
      "learning_rate": 0.0,
      "loss": 0.0088,
      "step": 1000
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.8317638952687184,
      "eval_f1": 0.9058561897702001,
      "eval_loss": 1.1073323488235474,
      "eval_precision": 0.8954567659990229,
      "eval_recall": 0.9165,
      "eval_runtime": 1.5045,
      "eval_samples_per_second": 35.892,
      "eval_steps_per_second": 0.665,
      "step": 1000
    },
    {
      "epoch": 100.0,
      "step": 1000,
      "total_flos": 4006801297113088.0,
      "train_loss": 0.12190062952041626,
      "train_runtime": 303.1848,
      "train_samples_per_second": 52.773,
      "train_steps_per_second": 3.298
    }
  ],
  "max_steps": 1000,
  "num_train_epochs": 100,
  "total_flos": 4006801297113088.0,
  "trial_name": null,
  "trial_params": null
}