checkpoint-1913/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.30003136762860727,
  "eval_steps": 479,
  "global_step": 1913,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0075282308657465494,
      "grad_norm": 87.12931823730469,
      "learning_rate": 3.147218736930155e-07,
      "loss": 12.3074,
      "step": 48
    },
    {
      "epoch": 0.015056461731493099,
      "grad_norm": 64.15751647949219,
      "learning_rate": 6.660393140945211e-07,
      "loss": 15.7221,
      "step": 96
    },
    {
      "epoch": 0.02258469259723965,
      "grad_norm": 81.39651489257812,
      "learning_rate": 1.0173567544960265e-06,
      "loss": 10.8027,
      "step": 144
    },
    {
      "epoch": 0.030112923462986198,
      "grad_norm": 22.048904418945312,
      "learning_rate": 1.3686741948975323e-06,
      "loss": 8.9559,
      "step": 192
    },
    {
      "epoch": 0.037641154328732745,
      "grad_norm": 11.542724609375,
      "learning_rate": 1.7126725219573398e-06,
      "loss": 8.8511,
      "step": 240
    },
    {
      "epoch": 0.0451693851944793,
      "grad_norm": 10.625059127807617,
      "learning_rate": 2.063989962358846e-06,
      "loss": 9.3478,
      "step": 288
    },
    {
      "epoch": 0.05269761606022585,
      "grad_norm": 14.21434211730957,
      "learning_rate": 2.415307402760351e-06,
      "loss": 8.8892,
      "step": 336
    },
    {
      "epoch": 0.060225846925972396,
      "grad_norm": 13.216053009033203,
      "learning_rate": 2.7666248431618565e-06,
      "loss": 8.3008,
      "step": 384
    },
    {
      "epoch": 0.06775407779171895,
      "grad_norm": 22.503334045410156,
      "learning_rate": 3.117942283563362e-06,
      "loss": 7.3455,
      "step": 432
    },
    {
      "epoch": 0.07512547051442911,
      "eval_nli-pairs_loss": 6.591032028198242,
      "eval_nli-pairs_runtime": 4.3469,
      "eval_nli-pairs_samples_per_second": 46.01,
      "eval_nli-pairs_steps_per_second": 1.15,
      "eval_sts-test_pearson_cosine": 0.39488461174644296,
      "eval_sts-test_pearson_dot": 0.15593446481859455,
      "eval_sts-test_pearson_euclidean": 0.39975070029693277,
      "eval_sts-test_pearson_manhattan": 0.4314268556737928,
      "eval_sts-test_pearson_max": 0.4314268556737928,
      "eval_sts-test_spearman_cosine": 0.3997824055251076,
      "eval_sts-test_spearman_dot": 0.14324216739430146,
      "eval_sts-test_spearman_euclidean": 0.40262274612650517,
      "eval_sts-test_spearman_manhattan": 0.42925492969387746,
      "eval_sts-test_spearman_max": 0.42925492969387746,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_vitaminc-pairs_loss": 6.247874736785889,
      "eval_vitaminc-pairs_runtime": 1.4382,
      "eval_vitaminc-pairs_samples_per_second": 115.426,
      "eval_vitaminc-pairs_steps_per_second": 2.781,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_sts-label_loss": 3.371708869934082,
      "eval_sts-label_runtime": 0.3925,
      "eval_sts-label_samples_per_second": 509.603,
      "eval_sts-label_steps_per_second": 12.74,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_qnli-contrastive_loss": 3.5311310291290283,
      "eval_qnli-contrastive_runtime": 0.2814,
      "eval_qnli-contrastive_samples_per_second": 710.798,
      "eval_qnli-contrastive_steps_per_second": 17.77,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_scitail-pairs-qa_loss": 5.4017333984375,
      "eval_scitail-pairs-qa_runtime": 1.1087,
      "eval_scitail-pairs-qa_samples_per_second": 180.387,
      "eval_scitail-pairs-qa_steps_per_second": 4.51,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_scitail-pairs-pos_loss": 4.29502534866333,
      "eval_scitail-pairs-pos_runtime": 2.4269,
      "eval_scitail-pairs-pos_samples_per_second": 82.41,
      "eval_scitail-pairs-pos_steps_per_second": 2.06,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_xsum-pairs_loss": 3.540722370147705,
      "eval_xsum-pairs_runtime": 1.0447,
      "eval_xsum-pairs_samples_per_second": 191.444,
      "eval_xsum-pairs_steps_per_second": 4.786,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_compression-pairs_loss": 2.8060033321380615,
      "eval_compression-pairs_runtime": 0.2399,
      "eval_compression-pairs_samples_per_second": 833.621,
      "eval_compression-pairs_steps_per_second": 20.841,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_sciq_pairs_loss": 10.621454238891602,
      "eval_sciq_pairs_runtime": 9.0638,
      "eval_sciq_pairs_samples_per_second": 22.066,
      "eval_sciq_pairs_steps_per_second": 0.552,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_qasc_pairs_loss": 7.7197771072387695,
      "eval_qasc_pairs_runtime": 1.2078,
      "eval_qasc_pairs_samples_per_second": 165.595,
      "eval_qasc_pairs_steps_per_second": 4.14,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_openbookqa_pairs_loss": 7.620975494384766,
      "eval_openbookqa_pairs_runtime": 1.053,
      "eval_openbookqa_pairs_samples_per_second": 189.941,
      "eval_openbookqa_pairs_steps_per_second": 4.749,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_msmarco_pairs_loss": 8.353594779968262,
      "eval_msmarco_pairs_runtime": 2.5338,
      "eval_msmarco_pairs_samples_per_second": 78.932,
      "eval_msmarco_pairs_steps_per_second": 1.973,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_nq_pairs_loss": 7.995354652404785,
      "eval_nq_pairs_runtime": 5.7107,
      "eval_nq_pairs_samples_per_second": 35.022,
      "eval_nq_pairs_steps_per_second": 0.876,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_trivia_pairs_loss": 8.177907943725586,
      "eval_trivia_pairs_runtime": 9.1824,
      "eval_trivia_pairs_samples_per_second": 21.781,
      "eval_trivia_pairs_steps_per_second": 0.545,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_quora_pairs_loss": 1.1983369588851929,
      "eval_quora_pairs_runtime": 0.659,
      "eval_quora_pairs_samples_per_second": 303.48,
      "eval_quora_pairs_steps_per_second": 7.587,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_gooaq_pairs_loss": 7.4573974609375,
      "eval_gooaq_pairs_runtime": 1.5917,
      "eval_gooaq_pairs_samples_per_second": 125.65,
      "eval_gooaq_pairs_steps_per_second": 3.141,
      "step": 479
    },
    {
      "epoch": 0.07512547051442911,
      "eval_mrpc_pairs_loss": 2.1152825355529785,
      "eval_mrpc_pairs_runtime": 0.241,
      "eval_mrpc_pairs_samples_per_second": 829.751,
      "eval_mrpc_pairs_steps_per_second": 20.744,
      "step": 479
    },
    {
      "epoch": 0.07528230865746549,
      "grad_norm": 17.753456115722656,
      "learning_rate": 3.469259723964868e-06,
      "loss": 8.0369,
      "step": 480
    },
    {
      "epoch": 0.08281053952321205,
      "grad_norm": 12.012594223022461,
      "learning_rate": 3.820577164366374e-06,
      "loss": 6.2732,
      "step": 528
    },
    {
      "epoch": 0.0903387703889586,
      "grad_norm": 34.92698287963867,
      "learning_rate": 4.1718946047678796e-06,
      "loss": 7.8529,
      "step": 576
    },
    {
      "epoch": 0.09786700125470514,
      "grad_norm": 23.564632415771484,
      "learning_rate": 4.523212045169385e-06,
      "loss": 5.8643,
      "step": 624
    },
    {
      "epoch": 0.1053952321204517,
      "grad_norm": 22.126293182373047,
      "learning_rate": 4.874529485570891e-06,
      "loss": 6.3179,
      "step": 672
    },
    {
      "epoch": 0.11292346298619825,
      "grad_norm": 25.067686080932617,
      "learning_rate": 5.225846925972396e-06,
      "loss": 6.1175,
      "step": 720
    },
    {
      "epoch": 0.12045169385194479,
      "grad_norm": 29.170730590820312,
      "learning_rate": 5.577164366373902e-06,
      "loss": 5.2392,
      "step": 768
    },
    {
      "epoch": 0.12797992471769135,
      "grad_norm": 29.377540588378906,
      "learning_rate": 5.928481806775407e-06,
      "loss": 5.8324,
      "step": 816
    },
    {
      "epoch": 0.1355081555834379,
      "grad_norm": 33.512088775634766,
      "learning_rate": 6.279799247176913e-06,
      "loss": 5.1523,
      "step": 864
    },
    {
      "epoch": 0.14303638644918445,
      "grad_norm": 32.54931640625,
      "learning_rate": 6.6311166875784185e-06,
      "loss": 6.0303,
      "step": 912
    },
    {
      "epoch": 0.15025094102885822,
      "eval_nli-pairs_loss": 4.317643165588379,
      "eval_nli-pairs_runtime": 4.0158,
      "eval_nli-pairs_samples_per_second": 49.803,
      "eval_nli-pairs_steps_per_second": 1.245,
      "eval_sts-test_pearson_cosine": 0.6751726661173544,
      "eval_sts-test_pearson_dot": 0.5308173325280101,
      "eval_sts-test_pearson_euclidean": 0.6660519042507951,
      "eval_sts-test_pearson_manhattan": 0.6723411683739887,
      "eval_sts-test_pearson_max": 0.6751726661173544,
      "eval_sts-test_spearman_cosine": 0.6458966208807124,
      "eval_sts-test_spearman_dot": 0.5040208096497271,
      "eval_sts-test_spearman_euclidean": 0.6447205374312966,
      "eval_sts-test_spearman_manhattan": 0.6498560301461127,
      "eval_sts-test_spearman_max": 0.6498560301461127,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_vitaminc-pairs_loss": 6.306981563568115,
      "eval_vitaminc-pairs_runtime": 1.4737,
      "eval_vitaminc-pairs_samples_per_second": 112.643,
      "eval_vitaminc-pairs_steps_per_second": 2.714,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_sts-label_loss": 3.7677345275878906,
      "eval_sts-label_runtime": 0.4136,
      "eval_sts-label_samples_per_second": 483.59,
      "eval_sts-label_steps_per_second": 12.09,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_qnli-contrastive_loss": 2.8633975982666016,
      "eval_qnli-contrastive_runtime": 0.2816,
      "eval_qnli-contrastive_samples_per_second": 710.123,
      "eval_qnli-contrastive_steps_per_second": 17.753,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_scitail-pairs-qa_loss": 1.4301409721374512,
      "eval_scitail-pairs-qa_runtime": 1.0526,
      "eval_scitail-pairs-qa_samples_per_second": 190.003,
      "eval_scitail-pairs-qa_steps_per_second": 4.75,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_scitail-pairs-pos_loss": 2.468087673187256,
      "eval_scitail-pairs-pos_runtime": 2.3275,
      "eval_scitail-pairs-pos_samples_per_second": 85.928,
      "eval_scitail-pairs-pos_steps_per_second": 2.148,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_xsum-pairs_loss": 2.153658151626587,
      "eval_xsum-pairs_runtime": 1.0409,
      "eval_xsum-pairs_samples_per_second": 192.151,
      "eval_xsum-pairs_steps_per_second": 4.804,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_compression-pairs_loss": 1.6288033723831177,
      "eval_compression-pairs_runtime": 0.2383,
      "eval_compression-pairs_samples_per_second": 839.253,
      "eval_compression-pairs_steps_per_second": 20.981,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_sciq_pairs_loss": 9.78779411315918,
      "eval_sciq_pairs_runtime": 9.0233,
      "eval_sciq_pairs_samples_per_second": 22.165,
      "eval_sciq_pairs_steps_per_second": 0.554,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_qasc_pairs_loss": 3.7814972400665283,
      "eval_qasc_pairs_runtime": 1.2108,
      "eval_qasc_pairs_samples_per_second": 165.179,
      "eval_qasc_pairs_steps_per_second": 4.129,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_openbookqa_pairs_loss": 5.005772590637207,
      "eval_openbookqa_pairs_runtime": 1.0415,
      "eval_openbookqa_pairs_samples_per_second": 192.037,
      "eval_openbookqa_pairs_steps_per_second": 4.801,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_msmarco_pairs_loss": 4.574879169464111,
      "eval_msmarco_pairs_runtime": 2.527,
      "eval_msmarco_pairs_samples_per_second": 79.146,
      "eval_msmarco_pairs_steps_per_second": 1.979,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_nq_pairs_loss": 5.281248569488525,
      "eval_nq_pairs_runtime": 5.6503,
      "eval_nq_pairs_samples_per_second": 35.397,
      "eval_nq_pairs_steps_per_second": 0.885,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_trivia_pairs_loss": 4.913428783416748,
      "eval_trivia_pairs_runtime": 9.0564,
      "eval_trivia_pairs_samples_per_second": 22.084,
      "eval_trivia_pairs_steps_per_second": 0.552,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_quora_pairs_loss": 0.9212128520011902,
      "eval_quora_pairs_runtime": 0.6293,
      "eval_quora_pairs_samples_per_second": 317.839,
      "eval_quora_pairs_steps_per_second": 7.946,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_gooaq_pairs_loss": 3.932173490524292,
      "eval_gooaq_pairs_runtime": 1.6066,
      "eval_gooaq_pairs_samples_per_second": 124.483,
      "eval_gooaq_pairs_steps_per_second": 3.112,
      "step": 958
    },
    {
      "epoch": 0.15025094102885822,
      "eval_mrpc_pairs_loss": 1.0853501558303833,
      "eval_mrpc_pairs_runtime": 0.2402,
      "eval_mrpc_pairs_samples_per_second": 832.56,
      "eval_mrpc_pairs_steps_per_second": 20.814,
      "step": 958
    },
    {
      "epoch": 0.15056461731493098,
      "grad_norm": 28.37123680114746,
      "learning_rate": 6.982434127979924e-06,
      "loss": 5.7748,
      "step": 960
    },
    {
      "epoch": 0.15809284818067754,
      "grad_norm": 26.708221435546875,
      "learning_rate": 7.33375156838143e-06,
      "loss": 4.8728,
      "step": 1008
    },
    {
      "epoch": 0.1656210790464241,
      "grad_norm": 26.786447525024414,
      "learning_rate": 7.685069008782934e-06,
      "loss": 4.7375,
      "step": 1056
    },
    {
      "epoch": 0.17314930991217065,
      "grad_norm": 26.215879440307617,
      "learning_rate": 8.03638644918444e-06,
      "loss": 4.6766,
      "step": 1104
    },
    {
      "epoch": 0.1806775407779172,
      "grad_norm": 35.618831634521484,
      "learning_rate": 8.387703889585947e-06,
      "loss": 4.3209,
      "step": 1152
    },
    {
      "epoch": 0.18820577164366373,
      "grad_norm": 37.166072845458984,
      "learning_rate": 8.739021329987453e-06,
      "loss": 3.7761,
      "step": 1200
    },
    {
      "epoch": 0.19573400250941028,
      "grad_norm": 35.78367233276367,
      "learning_rate": 9.090338770388957e-06,
      "loss": 4.2161,
      "step": 1248
    },
    {
      "epoch": 0.20326223337515684,
      "grad_norm": 36.299678802490234,
      "learning_rate": 9.441656210790464e-06,
      "loss": 4.9089,
      "step": 1296
    },
    {
      "epoch": 0.2107904642409034,
      "grad_norm": 18.610933303833008,
      "learning_rate": 9.792973651191968e-06,
      "loss": 4.3406,
      "step": 1344
    },
    {
      "epoch": 0.21831869510664995,
      "grad_norm": 9.592538833618164,
      "learning_rate": 1.0144291091593475e-05,
      "loss": 3.5664,
      "step": 1392
    },
    {
      "epoch": 0.22537641154328733,
      "eval_nli-pairs_loss": 3.2245519161224365,
      "eval_nli-pairs_runtime": 4.0436,
      "eval_nli-pairs_samples_per_second": 49.461,
      "eval_nli-pairs_steps_per_second": 1.237,
      "eval_sts-test_pearson_cosine": 0.7002978854888552,
      "eval_sts-test_pearson_dot": 0.5685392445320393,
      "eval_sts-test_pearson_euclidean": 0.6963744527231541,
      "eval_sts-test_pearson_manhattan": 0.7050517306003169,
      "eval_sts-test_pearson_max": 0.7050517306003169,
      "eval_sts-test_spearman_cosine": 0.6718756239728468,
      "eval_sts-test_spearman_dot": 0.5416448961602434,
      "eval_sts-test_spearman_euclidean": 0.6742379556154348,
      "eval_sts-test_spearman_manhattan": 0.6824201536078427,
      "eval_sts-test_spearman_max": 0.6824201536078427,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_vitaminc-pairs_loss": 6.715206623077393,
      "eval_vitaminc-pairs_runtime": 1.4251,
      "eval_vitaminc-pairs_samples_per_second": 116.481,
      "eval_vitaminc-pairs_steps_per_second": 2.807,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_sts-label_loss": 4.016364097595215,
      "eval_sts-label_runtime": 0.4049,
      "eval_sts-label_samples_per_second": 493.95,
      "eval_sts-label_steps_per_second": 12.349,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_qnli-contrastive_loss": 1.999517560005188,
      "eval_qnli-contrastive_runtime": 0.2804,
      "eval_qnli-contrastive_samples_per_second": 713.282,
      "eval_qnli-contrastive_steps_per_second": 17.832,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_scitail-pairs-qa_loss": 1.0403239727020264,
      "eval_scitail-pairs-qa_runtime": 1.0483,
      "eval_scitail-pairs-qa_samples_per_second": 190.793,
      "eval_scitail-pairs-qa_steps_per_second": 4.77,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_scitail-pairs-pos_loss": 1.9232473373413086,
      "eval_scitail-pairs-pos_runtime": 2.3447,
      "eval_scitail-pairs-pos_samples_per_second": 85.298,
      "eval_scitail-pairs-pos_steps_per_second": 2.132,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_xsum-pairs_loss": 1.6821197271347046,
      "eval_xsum-pairs_runtime": 1.0422,
      "eval_xsum-pairs_samples_per_second": 191.901,
      "eval_xsum-pairs_steps_per_second": 4.798,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_compression-pairs_loss": 1.1713249683380127,
      "eval_compression-pairs_runtime": 0.2392,
      "eval_compression-pairs_samples_per_second": 836.05,
      "eval_compression-pairs_steps_per_second": 20.901,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_sciq_pairs_loss": 9.443825721740723,
      "eval_sciq_pairs_runtime": 8.9916,
      "eval_sciq_pairs_samples_per_second": 22.243,
      "eval_sciq_pairs_steps_per_second": 0.556,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_qasc_pairs_loss": 2.9044029712677,
      "eval_qasc_pairs_runtime": 1.2182,
      "eval_qasc_pairs_samples_per_second": 164.182,
      "eval_qasc_pairs_steps_per_second": 4.105,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_openbookqa_pairs_loss": 4.360418796539307,
      "eval_openbookqa_pairs_runtime": 1.0522,
      "eval_openbookqa_pairs_samples_per_second": 190.077,
      "eval_openbookqa_pairs_steps_per_second": 4.752,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_msmarco_pairs_loss": 3.516049861907959,
      "eval_msmarco_pairs_runtime": 2.5595,
      "eval_msmarco_pairs_samples_per_second": 78.139,
      "eval_msmarco_pairs_steps_per_second": 1.953,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_nq_pairs_loss": 4.016308784484863,
      "eval_nq_pairs_runtime": 5.6561,
      "eval_nq_pairs_samples_per_second": 35.36,
      "eval_nq_pairs_steps_per_second": 0.884,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_trivia_pairs_loss": 3.781872272491455,
      "eval_trivia_pairs_runtime": 9.0801,
      "eval_trivia_pairs_samples_per_second": 22.026,
      "eval_trivia_pairs_steps_per_second": 0.551,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_quora_pairs_loss": 0.8747495412826538,
      "eval_quora_pairs_runtime": 0.6229,
      "eval_quora_pairs_samples_per_second": 321.076,
      "eval_quora_pairs_steps_per_second": 8.027,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_gooaq_pairs_loss": 3.0769765377044678,
      "eval_gooaq_pairs_runtime": 1.552,
      "eval_gooaq_pairs_samples_per_second": 128.863,
      "eval_gooaq_pairs_steps_per_second": 3.222,
      "step": 1437
    },
    {
      "epoch": 0.22537641154328733,
      "eval_mrpc_pairs_loss": 0.7370794415473938,
      "eval_mrpc_pairs_runtime": 0.2401,
      "eval_mrpc_pairs_samples_per_second": 832.867,
      "eval_mrpc_pairs_steps_per_second": 20.822,
      "step": 1437
    },
    {
      "epoch": 0.2258469259723965,
      "grad_norm": 29.73522186279297,
      "learning_rate": 1.049560853199498e-05,
      "loss": 4.7194,
      "step": 1440
    },
    {
      "epoch": 0.23337515683814303,
      "grad_norm": 30.467117309570312,
      "learning_rate": 1.0846925972396486e-05,
      "loss": 3.6345,
      "step": 1488
    },
    {
      "epoch": 0.24090338770388958,
      "grad_norm": 24.454021453857422,
      "learning_rate": 1.1198243412797992e-05,
      "loss": 3.5947,
      "step": 1536
    },
    {
      "epoch": 0.24843161856963614,
      "grad_norm": 20.165475845336914,
      "learning_rate": 1.1549560853199497e-05,
      "loss": 4.0526,
      "step": 1584
    },
    {
      "epoch": 0.2559598494353827,
      "grad_norm": 34.79319381713867,
      "learning_rate": 1.1900878293601003e-05,
      "loss": 3.7962,
      "step": 1632
    },
    {
      "epoch": 0.26348808030112925,
      "grad_norm": 127.97925567626953,
      "learning_rate": 1.2252195734002508e-05,
      "loss": 4.1927,
      "step": 1680
    },
    {
      "epoch": 0.2710163111668758,
      "grad_norm": 27.80243682861328,
      "learning_rate": 1.2603513174404014e-05,
      "loss": 3.6351,
      "step": 1728
    },
    {
      "epoch": 0.27854454203262236,
      "grad_norm": 31.81105613708496,
      "learning_rate": 1.295483061480552e-05,
      "loss": 3.4256,
      "step": 1776
    },
    {
      "epoch": 0.2860727728983689,
      "grad_norm": 32.932865142822266,
      "learning_rate": 1.3306148055207025e-05,
      "loss": 3.3175,
      "step": 1824
    },
    {
      "epoch": 0.2936010037641154,
      "grad_norm": 31.197385787963867,
      "learning_rate": 1.365746549560853e-05,
      "loss": 3.4984,
      "step": 1872
    }
  ],
  "logging_steps": 48,
  "max_steps": 19128,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1913,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 42,
  "trial_name": null,
  "trial_params": null
}