trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.874015748031496,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "eval_accuracy": 0.8157711500522223,
      "eval_f1": 0.5942329873125721,
      "eval_loss": 0.4848455488681793,
      "eval_precision": 0.5368903709879116,
      "eval_recall": 0.6652892561983471,
      "eval_runtime": 2.2498,
      "eval_samples_per_second": 41.337,
      "eval_steps_per_second": 1.333,
      "step": 20
    },
    {
      "epoch": 0.31,
      "eval_accuracy": 0.8389520714865962,
      "eval_f1": 0.7553421368547419,
      "eval_loss": 0.39182421565055847,
      "eval_precision": 0.705697622252131,
      "eval_recall": 0.8125,
      "eval_runtime": 2.9557,
      "eval_samples_per_second": 31.465,
      "eval_steps_per_second": 1.015,
      "step": 40
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.8849947777648833,
      "eval_f1": 0.8393364928909953,
      "eval_loss": 0.34172168374061584,
      "eval_precision": 0.7753940455341506,
      "eval_recall": 0.9147727272727273,
      "eval_runtime": 2.9806,
      "eval_samples_per_second": 31.202,
      "eval_steps_per_second": 1.007,
      "step": 60
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.8782929093652083,
      "eval_f1": 0.878610002347969,
      "eval_loss": 0.3457355201244354,
      "eval_precision": 0.8054240206629358,
      "eval_recall": 0.9664256198347108,
      "eval_runtime": 3.288,
      "eval_samples_per_second": 28.285,
      "eval_steps_per_second": 0.912,
      "step": 80
    },
    {
      "epoch": 0.79,
      "eval_accuracy": 0.9183880700940003,
      "eval_f1": 0.8993830090175606,
      "eval_loss": 0.26329365372657776,
      "eval_precision": 0.8318700614574188,
      "eval_recall": 0.9788223140495868,
      "eval_runtime": 3.262,
      "eval_samples_per_second": 28.51,
      "eval_steps_per_second": 0.92,
      "step": 100
    },
    {
      "epoch": 0.94,
      "eval_accuracy": 0.9218695601717535,
      "eval_f1": 0.908175497482618,
      "eval_loss": 0.27159208059310913,
      "eval_precision": 0.847427293064877,
      "eval_recall": 0.9783057851239669,
      "eval_runtime": 2.2237,
      "eval_samples_per_second": 41.822,
      "eval_steps_per_second": 1.349,
      "step": 120
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.931965881397238,
      "eval_f1": 0.8820210939416238,
      "eval_loss": 0.20828573405742645,
      "eval_precision": 0.8397944885567492,
      "eval_recall": 0.9287190082644629,
      "eval_runtime": 2.2685,
      "eval_samples_per_second": 40.996,
      "eval_steps_per_second": 1.322,
      "step": 140
    },
    {
      "epoch": 1.26,
      "eval_accuracy": 0.893669490541952,
      "eval_f1": 0.8734673859735164,
      "eval_loss": 0.33937641978263855,
      "eval_precision": 0.8314659197012139,
      "eval_recall": 0.9199380165289256,
      "eval_runtime": 3.2652,
      "eval_samples_per_second": 28.482,
      "eval_steps_per_second": 0.919,
      "step": 160
    },
    {
      "epoch": 1.42,
      "eval_accuracy": 0.883079958222119,
      "eval_f1": 0.8970068347867075,
      "eval_loss": 0.35709869861602783,
      "eval_precision": 0.8248807975726051,
      "eval_recall": 0.9829545454545454,
      "eval_runtime": 3.2871,
      "eval_samples_per_second": 28.292,
      "eval_steps_per_second": 0.913,
      "step": 180
    },
    {
      "epoch": 1.57,
      "eval_accuracy": 0.9112800278519206,
      "eval_f1": 0.8886283704572099,
      "eval_loss": 0.2868669033050537,
      "eval_precision": 0.8136539287247746,
      "eval_recall": 0.9788223140495868,
      "eval_runtime": 2.9919,
      "eval_samples_per_second": 31.084,
      "eval_steps_per_second": 1.003,
      "step": 200
    },
    {
      "epoch": 1.73,
      "eval_accuracy": 0.9393350353951492,
      "eval_f1": 0.9247259439707675,
      "eval_loss": 0.201907679438591,
      "eval_precision": 0.8750576302443522,
      "eval_recall": 0.9803719008264463,
      "eval_runtime": 2.2839,
      "eval_samples_per_second": 40.72,
      "eval_steps_per_second": 1.314,
      "step": 220
    },
    {
      "epoch": 1.89,
      "eval_accuracy": 0.9525937101079262,
      "eval_f1": 0.9215399610136452,
      "eval_loss": 0.1594531089067459,
      "eval_precision": 0.8722324723247232,
      "eval_recall": 0.9767561983471075,
      "eval_runtime": 2.9584,
      "eval_samples_per_second": 31.436,
      "eval_steps_per_second": 1.014,
      "step": 240
    },
    {
      "epoch": 2.05,
      "eval_accuracy": 0.9378263896947894,
      "eval_f1": 0.9166058837831268,
      "eval_loss": 0.22776451706886292,
      "eval_precision": 0.8658704639412035,
      "eval_recall": 0.9736570247933884,
      "eval_runtime": 3.2621,
      "eval_samples_per_second": 28.509,
      "eval_steps_per_second": 0.92,
      "step": 260
    },
    {
      "epoch": 2.2,
      "eval_accuracy": 0.9292677265869792,
      "eval_f1": 0.8771048002010556,
      "eval_loss": 0.23985013365745544,
      "eval_precision": 0.8541360744003916,
      "eval_recall": 0.9013429752066116,
      "eval_runtime": 2.2706,
      "eval_samples_per_second": 40.959,
      "eval_steps_per_second": 1.321,
      "step": 280
    },
    {
      "epoch": 2.36,
      "eval_accuracy": 0.9322850179876987,
      "eval_f1": 0.9097725568607848,
      "eval_loss": 0.2839757800102234,
      "eval_precision": 0.8813559322033898,
      "eval_recall": 0.9400826446280992,
      "eval_runtime": 3.2602,
      "eval_samples_per_second": 28.525,
      "eval_steps_per_second": 0.92,
      "step": 300
    },
    {
      "epoch": 2.52,
      "eval_accuracy": 0.9572356968782639,
      "eval_f1": 0.9368919930157147,
      "eval_loss": 0.15871362388134003,
      "eval_precision": 0.9059334298118669,
      "eval_recall": 0.9700413223140496,
      "eval_runtime": 2.2505,
      "eval_samples_per_second": 41.325,
      "eval_steps_per_second": 1.333,
      "step": 320
    },
    {
      "epoch": 2.68,
      "eval_accuracy": 0.9558431008471626,
      "eval_f1": 0.9427135678391959,
      "eval_loss": 0.16202741861343384,
      "eval_precision": 0.9178082191780822,
      "eval_recall": 0.96900826446281,
      "eval_runtime": 3.2378,
      "eval_samples_per_second": 28.723,
      "eval_steps_per_second": 0.927,
      "step": 340
    },
    {
      "epoch": 2.83,
      "eval_accuracy": 0.9299059997679007,
      "eval_f1": 0.9343610144206862,
      "eval_loss": 0.27064642310142517,
      "eval_precision": 0.9007670182166826,
      "eval_recall": 0.9705578512396694,
      "eval_runtime": 3.2832,
      "eval_samples_per_second": 28.326,
      "eval_steps_per_second": 0.914,
      "step": 360
    },
    {
      "epoch": 2.99,
      "eval_accuracy": 0.9477486364163862,
      "eval_f1": 0.9423980222496909,
      "eval_loss": 0.18907134234905243,
      "eval_precision": 0.9037458511142722,
      "eval_recall": 0.984504132231405,
      "eval_runtime": 3.2921,
      "eval_samples_per_second": 28.25,
      "eval_steps_per_second": 0.911,
      "step": 380
    },
    {
      "epoch": 3.15,
      "eval_accuracy": 0.9234072182894278,
      "eval_f1": 0.9175810155651952,
      "eval_loss": 0.26186808943748474,
      "eval_precision": 0.9067070095814422,
      "eval_recall": 0.9287190082644629,
      "eval_runtime": 3.296,
      "eval_samples_per_second": 28.216,
      "eval_steps_per_second": 0.91,
      "step": 400
    },
    {
      "epoch": 3.31,
      "eval_accuracy": 0.9601949634443542,
      "eval_f1": 0.9384654342871613,
      "eval_loss": 0.1693572849035263,
      "eval_precision": 0.9205166418281172,
      "eval_recall": 0.9571280991735537,
      "eval_runtime": 2.2486,
      "eval_samples_per_second": 41.36,
      "eval_steps_per_second": 1.334,
      "step": 420
    },
    {
      "epoch": 3.46,
      "eval_accuracy": 0.9532900081234769,
      "eval_f1": 0.9351432880844646,
      "eval_loss": 0.20067894458770752,
      "eval_precision": 0.910871694417238,
      "eval_recall": 0.9607438016528925,
      "eval_runtime": 3.2966,
      "eval_samples_per_second": 28.211,
      "eval_steps_per_second": 0.91,
      "step": 440
    },
    {
      "epoch": 3.62,
      "eval_accuracy": 0.9529998839503308,
      "eval_f1": 0.9408805031446541,
      "eval_loss": 0.2007509469985962,
      "eval_precision": 0.9171162334477685,
      "eval_recall": 0.9659090909090909,
      "eval_runtime": 2.9788,
      "eval_samples_per_second": 31.22,
      "eval_steps_per_second": 1.007,
      "step": 460
    },
    {
      "epoch": 3.78,
      "eval_accuracy": 0.9438609724962284,
      "eval_f1": 0.9448345035105317,
      "eval_loss": 0.2780458927154541,
      "eval_precision": 0.9181286549707602,
      "eval_recall": 0.9731404958677686,
      "eval_runtime": 2.2763,
      "eval_samples_per_second": 40.855,
      "eval_steps_per_second": 1.318,
      "step": 480
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.5e-05,
      "loss": 0.2747,
      "step": 500
    },
    {
      "epoch": 3.94,
      "eval_accuracy": 0.9434838110711384,
      "eval_f1": 0.9407256004087889,
      "eval_loss": 0.22288289666175842,
      "eval_precision": 0.9307381193124368,
      "eval_recall": 0.9509297520661157,
      "eval_runtime": 2.2152,
      "eval_samples_per_second": 41.983,
      "eval_steps_per_second": 1.354,
      "step": 500
    },
    {
      "epoch": 4.09,
      "eval_accuracy": 0.9413659046071718,
      "eval_f1": 0.9442211055276382,
      "eval_loss": 0.2373497188091278,
      "eval_precision": 0.9192759295499021,
      "eval_recall": 0.9705578512396694,
      "eval_runtime": 2.286,
      "eval_samples_per_second": 40.682,
      "eval_steps_per_second": 1.312,
      "step": 520
    },
    {
      "epoch": 4.25,
      "eval_accuracy": 0.9290066148311478,
      "eval_f1": 0.9232756393696718,
      "eval_loss": 0.32156404852867126,
      "eval_precision": 0.9235142118863049,
      "eval_recall": 0.9230371900826446,
      "eval_runtime": 3.2781,
      "eval_samples_per_second": 28.371,
      "eval_steps_per_second": 0.915,
      "step": 540
    },
    {
      "epoch": 4.41,
      "eval_accuracy": 0.9621387954044331,
      "eval_f1": 0.9501378100726634,
      "eval_loss": 0.1727248579263687,
      "eval_precision": 0.9226277372262773,
      "eval_recall": 0.9793388429752066,
      "eval_runtime": 3.2891,
      "eval_samples_per_second": 28.275,
      "eval_steps_per_second": 0.912,
      "step": 560
    },
    {
      "epoch": 4.57,
      "eval_accuracy": 0.9612974353023094,
      "eval_f1": 0.9563451776649746,
      "eval_loss": 0.2031358927488327,
      "eval_precision": 0.9401197604790419,
      "eval_recall": 0.9731404958677686,
      "eval_runtime": 2.9761,
      "eval_samples_per_second": 31.249,
      "eval_steps_per_second": 1.008,
      "step": 580
    },
    {
      "epoch": 4.72,
      "eval_accuracy": 0.9507079029824765,
      "eval_f1": 0.9494897959183674,
      "eval_loss": 0.24843396246433258,
      "eval_precision": 0.9380040322580645,
      "eval_recall": 0.9612603305785123,
      "eval_runtime": 2.242,
      "eval_samples_per_second": 41.481,
      "eval_steps_per_second": 1.338,
      "step": 600
    },
    {
      "epoch": 4.88,
      "eval_accuracy": 0.9590054543344552,
      "eval_f1": 0.9607390300230947,
      "eval_loss": 0.2189687043428421,
      "eval_precision": 0.9546149923508415,
      "eval_recall": 0.9669421487603306,
      "eval_runtime": 3.2795,
      "eval_samples_per_second": 28.358,
      "eval_steps_per_second": 0.915,
      "step": 620
    },
    {
      "epoch": 5.04,
      "eval_accuracy": 0.9571486596263201,
      "eval_f1": 0.9600409836065573,
      "eval_loss": 0.25752872228622437,
      "eval_precision": 0.9522357723577236,
      "eval_recall": 0.9679752066115702,
      "eval_runtime": 2.2527,
      "eval_samples_per_second": 41.284,
      "eval_steps_per_second": 1.332,
      "step": 640
    },
    {
      "epoch": 5.2,
      "eval_accuracy": 0.9608622490425902,
      "eval_f1": 0.9593869731800767,
      "eval_loss": 0.24718773365020752,
      "eval_precision": 0.9489641232945932,
      "eval_recall": 0.9700413223140496,
      "eval_runtime": 2.2796,
      "eval_samples_per_second": 40.796,
      "eval_steps_per_second": 1.316,
      "step": 660
    },
    {
      "epoch": 5.35,
      "eval_accuracy": 0.9522165486828362,
      "eval_f1": 0.9561671763506625,
      "eval_loss": 0.2798936367034912,
      "eval_precision": 0.9436619718309859,
      "eval_recall": 0.96900826446281,
      "eval_runtime": 3.2922,
      "eval_samples_per_second": 28.249,
      "eval_steps_per_second": 0.911,
      "step": 680
    },
    {
      "epoch": 5.51,
      "eval_accuracy": 0.9556400139259603,
      "eval_f1": 0.9535588086824837,
      "eval_loss": 0.26087334752082825,
      "eval_precision": 0.9323790720631787,
      "eval_recall": 0.9757231404958677,
      "eval_runtime": 2.2499,
      "eval_samples_per_second": 41.335,
      "eval_steps_per_second": 1.333,
      "step": 700
    },
    {
      "epoch": 5.67,
      "eval_accuracy": 0.9529418591157015,
      "eval_f1": 0.9502170028082715,
      "eval_loss": 0.23053081333637238,
      "eval_precision": 0.939424533064109,
      "eval_recall": 0.9612603305785123,
      "eval_runtime": 3.2663,
      "eval_samples_per_second": 28.473,
      "eval_steps_per_second": 0.918,
      "step": 720
    },
    {
      "epoch": 5.83,
      "eval_accuracy": 0.9524486480213531,
      "eval_f1": 0.954763709881223,
      "eval_loss": 0.2539260983467102,
      "eval_precision": 0.9346857991093518,
      "eval_recall": 0.9757231404958677,
      "eval_runtime": 3.298,
      "eval_samples_per_second": 28.199,
      "eval_steps_per_second": 0.91,
      "step": 740
    },
    {
      "epoch": 5.98,
      "eval_accuracy": 0.9543634675641175,
      "eval_f1": 0.9617654606107262,
      "eval_loss": 0.28202250599861145,
      "eval_precision": 0.9556348801631821,
      "eval_recall": 0.9679752066115702,
      "eval_runtime": 3.2935,
      "eval_samples_per_second": 28.237,
      "eval_steps_per_second": 0.911,
      "step": 760
    },
    {
      "epoch": 6.14,
      "eval_accuracy": 0.9599048392712081,
      "eval_f1": 0.9617065021845285,
      "eval_loss": 0.23321200907230377,
      "eval_precision": 0.9570332480818414,
      "eval_recall": 0.9664256198347108,
      "eval_runtime": 3.2967,
      "eval_samples_per_second": 28.21,
      "eval_steps_per_second": 0.91,
      "step": 780
    },
    {
      "epoch": 6.3,
      "eval_accuracy": 0.9590054543344552,
      "eval_f1": 0.961734693877551,
      "eval_loss": 0.25181254744529724,
      "eval_precision": 0.9501008064516129,
      "eval_recall": 0.9736570247933884,
      "eval_runtime": 2.9863,
      "eval_samples_per_second": 31.142,
      "eval_steps_per_second": 1.005,
      "step": 800
    },
    {
      "epoch": 6.46,
      "eval_accuracy": 0.95752582105141,
      "eval_f1": 0.9602258147292788,
      "eval_loss": 0.2890544533729553,
      "eval_precision": 0.954105048444671,
      "eval_recall": 0.9664256198347108,
      "eval_runtime": 2.2411,
      "eval_samples_per_second": 41.497,
      "eval_steps_per_second": 1.339,
      "step": 820
    },
    {
      "epoch": 6.61,
      "eval_accuracy": 0.9568875478704886,
      "eval_f1": 0.9569451563300871,
      "eval_loss": 0.29066193103790283,
      "eval_precision": 0.9496439471007121,
      "eval_recall": 0.9643595041322314,
      "eval_runtime": 2.2281,
      "eval_samples_per_second": 41.74,
      "eval_steps_per_second": 1.346,
      "step": 840
    },
    {
      "epoch": 6.77,
      "eval_accuracy": 0.9598178020192643,
      "eval_f1": 0.9613118114271074,
      "eval_loss": 0.26129651069641113,
      "eval_precision": 0.9537366548042705,
      "eval_recall": 0.96900826446281,
      "eval_runtime": 2.2516,
      "eval_samples_per_second": 41.305,
      "eval_steps_per_second": 1.332,
      "step": 860
    },
    {
      "epoch": 6.93,
      "eval_accuracy": 0.9592085412556575,
      "eval_f1": 0.96229802513465,
      "eval_loss": 0.25751829147338867,
      "eval_precision": 0.9556800815078961,
      "eval_recall": 0.96900826446281,
      "eval_runtime": 2.2645,
      "eval_samples_per_second": 41.068,
      "eval_steps_per_second": 1.325,
      "step": 880
    },
    {
      "epoch": 7.09,
      "eval_accuracy": 0.9583381687362191,
      "eval_f1": 0.9625832906201948,
      "eval_loss": 0.26336678862571716,
      "eval_precision": 0.9552390640895219,
      "eval_recall": 0.9700413223140496,
      "eval_runtime": 2.2487,
      "eval_samples_per_second": 41.357,
      "eval_steps_per_second": 1.334,
      "step": 900
    },
    {
      "epoch": 7.24,
      "eval_accuracy": 0.9581640942323314,
      "eval_f1": 0.9588550983899822,
      "eval_loss": 0.2639918923377991,
      "eval_precision": 0.9489124936772888,
      "eval_recall": 0.96900826446281,
      "eval_runtime": 2.996,
      "eval_samples_per_second": 31.041,
      "eval_steps_per_second": 1.001,
      "step": 920
    },
    {
      "epoch": 7.4,
      "eval_accuracy": 0.9608622490425902,
      "eval_f1": 0.9594284256187804,
      "eval_loss": 0.2650795876979828,
      "eval_precision": 0.9480584972264247,
      "eval_recall": 0.9710743801652892,
      "eval_runtime": 3.2719,
      "eval_samples_per_second": 28.424,
      "eval_steps_per_second": 0.917,
      "step": 940
    },
    {
      "epoch": 7.56,
      "eval_accuracy": 0.9626900313334107,
      "eval_f1": 0.9609992352791231,
      "eval_loss": 0.26358404755592346,
      "eval_precision": 0.9486663311524912,
      "eval_recall": 0.9736570247933884,
      "eval_runtime": 3.271,
      "eval_samples_per_second": 28.432,
      "eval_steps_per_second": 0.917,
      "step": 960
    },
    {
      "epoch": 7.72,
      "eval_accuracy": 0.9627190437507253,
      "eval_f1": 0.9622256253190403,
      "eval_loss": 0.2597278356552124,
      "eval_precision": 0.9510595358224017,
      "eval_recall": 0.9736570247933884,
      "eval_runtime": 2.9416,
      "eval_samples_per_second": 31.616,
      "eval_steps_per_second": 1.02,
      "step": 980
    },
    {
      "epoch": 7.87,
      "learning_rate": 0.0,
      "loss": 0.0555,
      "step": 1000
    },
    {
      "epoch": 7.87,
      "eval_accuracy": 0.9627190437507253,
      "eval_f1": 0.961489415965315,
      "eval_loss": 0.25795629620552063,
      "eval_precision": 0.9496221662468514,
      "eval_recall": 0.9736570247933884,
      "eval_runtime": 3.3,
      "eval_samples_per_second": 28.182,
      "eval_steps_per_second": 0.909,
      "step": 1000
    },
    {
      "epoch": 7.87,
      "step": 1000,
      "total_flos": 2101461798354944.0,
      "train_loss": 0.16514424514770507,
      "train_runtime": 1048.0167,
      "train_samples_per_second": 7.633,
      "train_steps_per_second": 0.954
    }
  ],
  "max_steps": 1000,
  "num_train_epochs": 8,
  "total_flos": 2101461798354944.0,
  "trial_name": null,
  "trial_params": null
}