{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 1757,
  "global_step": 35140,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "grad_norm": 23.659637451171875,
      "learning_rate": 9.97723392145703e-07,
      "loss": 3.8346,
      "step": 1757
    },
    {
      "epoch": 0.1,
      "eval_nli-pairs_loss": 2.8535287380218506,
      "eval_nli-pairs_runtime": 23.1384,
      "eval_nli-pairs_samples_per_second": 294.229,
      "eval_nli-pairs_steps_per_second": 18.411,
      "step": 1757
    },
    {
      "epoch": 0.1,
      "eval_scitail-pairs-pos_loss": 2.323117256164551,
      "eval_scitail-pairs-pos_runtime": 5.1803,
      "eval_scitail-pairs-pos_samples_per_second": 251.722,
      "eval_scitail-pairs-pos_steps_per_second": 15.829,
      "step": 1757
    },
    {
      "epoch": 0.1,
      "eval_qnli-contrastive_loss": 3.09728741645813,
      "eval_qnli-contrastive_runtime": 15.5151,
      "eval_qnli-contrastive_samples_per_second": 352.109,
      "eval_qnli-contrastive_steps_per_second": 22.043,
      "step": 1757
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.776155471801758,
      "learning_rate": 1.9965850882185546e-06,
      "loss": 1.8532,
      "step": 3514
    },
    {
      "epoch": 0.2,
      "eval_nli-pairs_loss": 1.3508331775665283,
      "eval_nli-pairs_runtime": 22.8642,
      "eval_nli-pairs_samples_per_second": 297.758,
      "eval_nli-pairs_steps_per_second": 18.632,
      "step": 3514
    },
    {
      "epoch": 0.2,
      "eval_scitail-pairs-pos_loss": 0.9754649996757507,
      "eval_scitail-pairs-pos_runtime": 5.1924,
      "eval_scitail-pairs-pos_samples_per_second": 251.136,
      "eval_scitail-pairs-pos_steps_per_second": 15.792,
      "step": 3514
    },
    {
      "epoch": 0.2,
      "eval_qnli-contrastive_loss": 2.0602548122406006,
      "eval_qnli-contrastive_runtime": 15.6036,
      "eval_qnli-contrastive_samples_per_second": 350.111,
      "eval_qnli-contrastive_steps_per_second": 21.918,
      "step": 3514
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6704504489898682,
      "learning_rate": 2.99601593625498e-06,
      "loss": 1.2185,
      "step": 5271
    },
    {
      "epoch": 0.3,
      "eval_nli-pairs_loss": 0.9380640983581543,
      "eval_nli-pairs_runtime": 23.1102,
      "eval_nli-pairs_samples_per_second": 294.589,
      "eval_nli-pairs_steps_per_second": 18.433,
      "step": 5271
    },
    {
      "epoch": 0.3,
      "eval_scitail-pairs-pos_loss": 0.7407301664352417,
      "eval_scitail-pairs-pos_runtime": 5.2512,
      "eval_scitail-pairs-pos_samples_per_second": 248.322,
      "eval_scitail-pairs-pos_steps_per_second": 15.615,
      "step": 5271
    },
    {
      "epoch": 0.3,
      "eval_qnli-contrastive_loss": 1.2534083127975464,
      "eval_qnli-contrastive_runtime": 15.6475,
      "eval_qnli-contrastive_samples_per_second": 349.129,
      "eval_qnli-contrastive_steps_per_second": 21.857,
      "step": 5271
    },
    {
      "epoch": 0.4,
      "grad_norm": 26.072860717773438,
      "learning_rate": 3.99601593625498e-06,
      "loss": 0.9584,
      "step": 7028
    },
    {
      "epoch": 0.4,
      "eval_nli-pairs_loss": 0.749484658241272,
      "eval_nli-pairs_runtime": 23.0514,
      "eval_nli-pairs_samples_per_second": 295.34,
      "eval_nli-pairs_steps_per_second": 18.48,
      "step": 7028
    },
    {
      "epoch": 0.4,
      "eval_scitail-pairs-pos_loss": 0.661561131477356,
      "eval_scitail-pairs-pos_runtime": 5.2207,
      "eval_scitail-pairs-pos_samples_per_second": 249.774,
      "eval_scitail-pairs-pos_steps_per_second": 15.707,
      "step": 7028
    },
    {
      "epoch": 0.4,
      "eval_qnli-contrastive_loss": 0.5139556527137756,
      "eval_qnli-contrastive_runtime": 15.681,
      "eval_qnli-contrastive_samples_per_second": 348.384,
      "eval_qnli-contrastive_steps_per_second": 21.81,
      "step": 7028
    },
    {
      "epoch": 0.5,
      "grad_norm": 24.09697914123535,
      "learning_rate": 4.995446784291406e-06,
      "loss": 0.8157,
      "step": 8785
    },
    {
      "epoch": 0.5,
      "eval_nli-pairs_loss": 0.6549726724624634,
      "eval_nli-pairs_runtime": 23.2274,
      "eval_nli-pairs_samples_per_second": 293.102,
      "eval_nli-pairs_steps_per_second": 18.34,
      "step": 8785
    },
    {
      "epoch": 0.5,
      "eval_scitail-pairs-pos_loss": 0.6056841611862183,
      "eval_scitail-pairs-pos_runtime": 5.2473,
      "eval_scitail-pairs-pos_samples_per_second": 248.508,
      "eval_scitail-pairs-pos_steps_per_second": 15.627,
      "step": 8785
    },
    {
      "epoch": 0.5,
      "eval_qnli-contrastive_loss": 0.3295331299304962,
      "eval_qnli-contrastive_runtime": 15.7204,
      "eval_qnli-contrastive_samples_per_second": 347.511,
      "eval_qnli-contrastive_steps_per_second": 21.755,
      "step": 8785
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.664803504943848,
      "learning_rate": 5.994877632327832e-06,
      "loss": 0.6698,
      "step": 10542
    },
    {
      "epoch": 0.6,
      "eval_nli-pairs_loss": 0.5809468626976013,
      "eval_nli-pairs_runtime": 22.9525,
      "eval_nli-pairs_samples_per_second": 296.612,
      "eval_nli-pairs_steps_per_second": 18.56,
      "step": 10542
    },
    {
      "epoch": 0.6,
      "eval_scitail-pairs-pos_loss": 0.5820835828781128,
      "eval_scitail-pairs-pos_runtime": 5.1829,
      "eval_scitail-pairs-pos_samples_per_second": 251.599,
      "eval_scitail-pairs-pos_steps_per_second": 15.821,
      "step": 10542
    },
    {
      "epoch": 0.6,
      "eval_qnli-contrastive_loss": 0.24226614832878113,
      "eval_qnli-contrastive_runtime": 15.6321,
      "eval_qnli-contrastive_samples_per_second": 349.473,
      "eval_qnli-contrastive_steps_per_second": 21.878,
      "step": 10542
    },
    {
      "epoch": 0.7,
      "grad_norm": 66.77753448486328,
      "learning_rate": 6.994877632327832e-06,
      "loss": 0.6497,
      "step": 12299
    },
    {
      "epoch": 0.7,
      "eval_nli-pairs_loss": 0.5178281664848328,
      "eval_nli-pairs_runtime": 23.0673,
      "eval_nli-pairs_samples_per_second": 295.136,
      "eval_nli-pairs_steps_per_second": 18.468,
      "step": 12299
    },
    {
      "epoch": 0.7,
      "eval_scitail-pairs-pos_loss": 0.504002571105957,
      "eval_scitail-pairs-pos_runtime": 5.1845,
      "eval_scitail-pairs-pos_samples_per_second": 251.52,
      "eval_scitail-pairs-pos_steps_per_second": 15.816,
      "step": 12299
    },
    {
      "epoch": 0.7,
      "eval_qnli-contrastive_loss": 0.24089547991752625,
      "eval_qnli-contrastive_runtime": 15.5228,
      "eval_qnli-contrastive_samples_per_second": 351.933,
      "eval_qnli-contrastive_steps_per_second": 22.032,
      "step": 12299
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6044542193412781,
      "learning_rate": 7.994308480364257e-06,
      "loss": 0.5737,
      "step": 14056
    },
    {
      "epoch": 0.8,
      "eval_nli-pairs_loss": 0.5019380450248718,
      "eval_nli-pairs_runtime": 23.0659,
      "eval_nli-pairs_samples_per_second": 295.154,
      "eval_nli-pairs_steps_per_second": 18.469,
      "step": 14056
    },
    {
      "epoch": 0.8,
      "eval_scitail-pairs-pos_loss": 0.49418017268180847,
      "eval_scitail-pairs-pos_runtime": 5.2457,
      "eval_scitail-pairs-pos_samples_per_second": 248.585,
      "eval_scitail-pairs-pos_steps_per_second": 15.632,
      "step": 14056
    },
    {
      "epoch": 0.8,
      "eval_qnli-contrastive_loss": 0.14995019137859344,
      "eval_qnli-contrastive_runtime": 15.7177,
      "eval_qnli-contrastive_samples_per_second": 347.57,
      "eval_qnli-contrastive_steps_per_second": 21.759,
      "step": 14056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4454790949821472,
      "learning_rate": 8.993739328400684e-06,
      "loss": 0.5896,
      "step": 15813
    },
    {
      "epoch": 0.9,
      "eval_nli-pairs_loss": 0.4803747236728668,
      "eval_nli-pairs_runtime": 23.0746,
      "eval_nli-pairs_samples_per_second": 295.043,
      "eval_nli-pairs_steps_per_second": 18.462,
      "step": 15813
    },
    {
      "epoch": 0.9,
      "eval_scitail-pairs-pos_loss": 0.47568026185035706,
      "eval_scitail-pairs-pos_runtime": 5.2076,
      "eval_scitail-pairs-pos_samples_per_second": 250.402,
      "eval_scitail-pairs-pos_steps_per_second": 15.746,
      "step": 15813
    },
    {
      "epoch": 0.9,
      "eval_qnli-contrastive_loss": 0.14648529887199402,
      "eval_qnli-contrastive_runtime": 15.5997,
      "eval_qnli-contrastive_samples_per_second": 350.199,
      "eval_qnli-contrastive_steps_per_second": 21.924,
      "step": 15813
    },
    {
      "epoch": 1.0,
      "grad_norm": 196.14842224121094,
      "learning_rate": 9.993739328400683e-06,
      "loss": 0.5174,
      "step": 17570
    },
    {
      "epoch": 1.0,
      "eval_nli-pairs_loss": 0.4586646258831024,
      "eval_nli-pairs_runtime": 22.8967,
      "eval_nli-pairs_samples_per_second": 297.336,
      "eval_nli-pairs_steps_per_second": 18.605,
      "step": 17570
    },
    {
      "epoch": 1.0,
      "eval_scitail-pairs-pos_loss": 0.5253121256828308,
      "eval_scitail-pairs-pos_runtime": 5.1603,
      "eval_scitail-pairs-pos_samples_per_second": 252.699,
      "eval_scitail-pairs-pos_steps_per_second": 15.891,
      "step": 17570
    },
    {
      "epoch": 1.0,
      "eval_qnli-contrastive_loss": 0.0533733032643795,
      "eval_qnli-contrastive_runtime": 15.5083,
      "eval_qnli-contrastive_samples_per_second": 352.263,
      "eval_qnli-contrastive_steps_per_second": 22.053,
      "step": 17570
    },
    {
      "epoch": 1.1,
      "grad_norm": 18.839372634887695,
      "learning_rate": 9.75831232890717e-06,
      "loss": 0.5059,
      "step": 19327
    },
    {
      "epoch": 1.1,
      "eval_nli-pairs_loss": 0.45871272683143616,
      "eval_nli-pairs_runtime": 22.8984,
      "eval_nli-pairs_samples_per_second": 297.313,
      "eval_nli-pairs_steps_per_second": 18.604,
      "step": 19327
    },
    {
      "epoch": 1.1,
      "eval_scitail-pairs-pos_loss": 0.5492986440658569,
      "eval_scitail-pairs-pos_runtime": 5.1782,
      "eval_scitail-pairs-pos_samples_per_second": 251.824,
      "eval_scitail-pairs-pos_steps_per_second": 15.836,
      "step": 19327
    },
    {
      "epoch": 1.1,
      "eval_qnli-contrastive_loss": 0.027841920033097267,
      "eval_qnli-contrastive_runtime": 15.522,
      "eval_qnli-contrastive_samples_per_second": 351.952,
      "eval_qnli-contrastive_steps_per_second": 22.033,
      "step": 19327
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.800241947174072,
      "learning_rate": 9.051905444616243e-06,
      "loss": 0.4654,
      "step": 21084
    },
    {
      "epoch": 1.2,
      "eval_nli-pairs_loss": 0.44151321053504944,
      "eval_nli-pairs_runtime": 23.1311,
      "eval_nli-pairs_samples_per_second": 294.323,
      "eval_nli-pairs_steps_per_second": 18.417,
      "step": 21084
    },
    {
      "epoch": 1.2,
      "eval_scitail-pairs-pos_loss": 0.4850437045097351,
      "eval_scitail-pairs-pos_runtime": 5.2939,
      "eval_scitail-pairs-pos_samples_per_second": 246.321,
      "eval_scitail-pairs-pos_steps_per_second": 15.49,
      "step": 21084
    },
    {
      "epoch": 1.2,
      "eval_qnli-contrastive_loss": 0.05170624330639839,
      "eval_qnli-contrastive_runtime": 15.7737,
      "eval_qnli-contrastive_samples_per_second": 346.336,
      "eval_qnli-contrastive_steps_per_second": 21.682,
      "step": 21084
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.41899746656417847,
      "learning_rate": 7.948320938272786e-06,
      "loss": 0.4224,
      "step": 22841
    },
    {
      "epoch": 1.3,
      "eval_nli-pairs_loss": 0.39569494128227234,
      "eval_nli-pairs_runtime": 23.2638,
      "eval_nli-pairs_samples_per_second": 292.643,
      "eval_nli-pairs_steps_per_second": 18.312,
      "step": 22841
    },
    {
      "epoch": 1.3,
      "eval_scitail-pairs-pos_loss": 0.42922988533973694,
      "eval_scitail-pairs-pos_runtime": 5.2769,
      "eval_scitail-pairs-pos_samples_per_second": 247.114,
      "eval_scitail-pairs-pos_steps_per_second": 15.539,
      "step": 22841
    },
    {
      "epoch": 1.3,
      "eval_qnli-contrastive_loss": 0.0938122496008873,
      "eval_qnli-contrastive_runtime": 15.6681,
      "eval_qnli-contrastive_samples_per_second": 348.67,
      "eval_qnli-contrastive_steps_per_second": 21.828,
      "step": 22841
    },
    {
      "epoch": 1.4,
      "grad_norm": 3.0029168128967285,
      "learning_rate": 6.556983832253587e-06,
      "loss": 0.4125,
      "step": 24598
    },
    {
      "epoch": 1.4,
      "eval_nli-pairs_loss": 0.3794442415237427,
      "eval_nli-pairs_runtime": 23.2107,
      "eval_nli-pairs_samples_per_second": 293.313,
      "eval_nli-pairs_steps_per_second": 18.354,
      "step": 24598
    },
    {
      "epoch": 1.4,
      "eval_scitail-pairs-pos_loss": 0.4623956084251404,
      "eval_scitail-pairs-pos_runtime": 5.2884,
      "eval_scitail-pairs-pos_samples_per_second": 246.577,
      "eval_scitail-pairs-pos_steps_per_second": 15.506,
      "step": 24598
    },
    {
      "epoch": 1.4,
      "eval_qnli-contrastive_loss": 0.0838843286037445,
      "eval_qnli-contrastive_runtime": 15.7017,
      "eval_qnli-contrastive_samples_per_second": 347.924,
      "eval_qnli-contrastive_steps_per_second": 21.781,
      "step": 24598
    },
    {
      "epoch": 1.5,
      "grad_norm": 10.91913890838623,
      "learning_rate": 5.012516292320938e-06,
      "loss": 0.4072,
      "step": 26355
    },
    {
      "epoch": 1.5,
      "eval_nli-pairs_loss": 0.3877629041671753,
      "eval_nli-pairs_runtime": 23.1072,
      "eval_nli-pairs_samples_per_second": 294.627,
      "eval_nli-pairs_steps_per_second": 18.436,
      "step": 26355
    },
    {
      "epoch": 1.5,
      "eval_scitail-pairs-pos_loss": 0.4480924606323242,
      "eval_scitail-pairs-pos_runtime": 5.2741,
      "eval_scitail-pairs-pos_samples_per_second": 247.244,
      "eval_scitail-pairs-pos_steps_per_second": 15.548,
      "step": 26355
    },
    {
      "epoch": 1.5,
      "eval_qnli-contrastive_loss": 0.06811495870351791,
      "eval_qnli-contrastive_runtime": 15.7641,
      "eval_qnli-contrastive_samples_per_second": 346.546,
      "eval_qnli-contrastive_steps_per_second": 21.695,
      "step": 26355
    },
    {
      "epoch": 1.6,
      "grad_norm": 3.676146984100342,
      "learning_rate": 3.4668235704897813e-06,
      "loss": 0.3572,
      "step": 28112
    },
    {
      "epoch": 1.6,
      "eval_nli-pairs_loss": 0.3715905547142029,
      "eval_nli-pairs_runtime": 23.1744,
      "eval_nli-pairs_samples_per_second": 293.773,
      "eval_nli-pairs_steps_per_second": 18.382,
      "step": 28112
    },
    {
      "epoch": 1.6,
      "eval_scitail-pairs-pos_loss": 0.49534013867378235,
      "eval_scitail-pairs-pos_runtime": 5.2856,
      "eval_scitail-pairs-pos_samples_per_second": 246.708,
      "eval_scitail-pairs-pos_steps_per_second": 15.514,
      "step": 28112
    },
    {
      "epoch": 1.6,
      "eval_qnli-contrastive_loss": 0.06735851615667343,
      "eval_qnli-contrastive_runtime": 15.7308,
      "eval_qnli-contrastive_samples_per_second": 347.281,
      "eval_qnli-contrastive_steps_per_second": 21.741,
      "step": 28112
    },
    {
      "epoch": 1.7,
      "grad_norm": 229.6580047607422,
      "learning_rate": 2.072658211127134e-06,
      "loss": 0.371,
      "step": 29869
    },
    {
      "epoch": 1.7,
      "eval_nli-pairs_loss": 0.36217835545539856,
      "eval_nli-pairs_runtime": 23.1495,
      "eval_nli-pairs_samples_per_second": 294.089,
      "eval_nli-pairs_steps_per_second": 18.402,
      "step": 29869
    },
    {
      "epoch": 1.7,
      "eval_scitail-pairs-pos_loss": 0.47673526406288147,
      "eval_scitail-pairs-pos_runtime": 5.2158,
      "eval_scitail-pairs-pos_samples_per_second": 250.008,
      "eval_scitail-pairs-pos_steps_per_second": 15.721,
      "step": 29869
    },
    {
      "epoch": 1.7,
      "eval_qnli-contrastive_loss": 0.06000087782740593,
      "eval_qnli-contrastive_runtime": 15.6328,
      "eval_qnli-contrastive_samples_per_second": 349.458,
      "eval_qnli-contrastive_steps_per_second": 21.877,
      "step": 29869
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6022229194641113,
      "learning_rate": 9.638670801112644e-07,
      "loss": 0.3332,
      "step": 31626
    },
    {
      "epoch": 1.8,
      "eval_nli-pairs_loss": 0.3600439131259918,
      "eval_nli-pairs_runtime": 23.0874,
      "eval_nli-pairs_samples_per_second": 294.879,
      "eval_nli-pairs_steps_per_second": 18.452,
      "step": 31626
    },
    {
      "epoch": 1.8,
      "eval_scitail-pairs-pos_loss": 0.465911865234375,
      "eval_scitail-pairs-pos_runtime": 5.3369,
      "eval_scitail-pairs-pos_samples_per_second": 244.338,
      "eval_scitail-pairs-pos_steps_per_second": 15.365,
      "step": 31626
    },
    {
      "epoch": 1.8,
      "eval_qnli-contrastive_loss": 0.05613844096660614,
      "eval_qnli-contrastive_runtime": 15.7089,
      "eval_qnli-contrastive_samples_per_second": 347.764,
      "eval_qnli-contrastive_steps_per_second": 21.771,
      "step": 31626
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.23106251657009125,
      "learning_rate": 2.4943593464921476e-07,
      "loss": 0.3695,
      "step": 33383
    },
    {
      "epoch": 1.9,
      "eval_nli-pairs_loss": 0.35667526721954346,
      "eval_nli-pairs_runtime": 23.1588,
      "eval_nli-pairs_samples_per_second": 293.971,
      "eval_nli-pairs_steps_per_second": 18.395,
      "step": 33383
    },
    {
      "epoch": 1.9,
      "eval_scitail-pairs-pos_loss": 0.4603894352912903,
      "eval_scitail-pairs-pos_runtime": 5.248,
      "eval_scitail-pairs-pos_samples_per_second": 248.476,
      "eval_scitail-pairs-pos_steps_per_second": 15.625,
      "step": 33383
    },
    {
      "epoch": 1.9,
      "eval_qnli-contrastive_loss": 0.06141861155629158,
      "eval_qnli-contrastive_runtime": 15.6709,
      "eval_qnli-contrastive_samples_per_second": 348.608,
      "eval_qnli-contrastive_steps_per_second": 21.824,
      "step": 33383
    },
    {
      "epoch": 2.0,
      "grad_norm": Infinity,
      "learning_rate": 2.5896487759191624e-11,
      "loss": 0.3315,
      "step": 35140
    },
    {
      "epoch": 2.0,
      "eval_nli-pairs_loss": 0.3597075045108795,
      "eval_nli-pairs_runtime": 23.1058,
      "eval_nli-pairs_samples_per_second": 294.645,
      "eval_nli-pairs_steps_per_second": 18.437,
      "step": 35140
    },
    {
      "epoch": 2.0,
      "eval_scitail-pairs-pos_loss": 0.47120198607444763,
      "eval_scitail-pairs-pos_runtime": 5.2532,
      "eval_scitail-pairs-pos_samples_per_second": 248.23,
      "eval_scitail-pairs-pos_steps_per_second": 15.61,
      "step": 35140
    },
    {
      "epoch": 2.0,
      "eval_qnli-contrastive_loss": 0.05398999899625778,
      "eval_qnli-contrastive_runtime": 15.7099,
      "eval_qnli-contrastive_samples_per_second": 347.743,
      "eval_qnli-contrastive_steps_per_second": 21.77,
      "step": 35140
    }
  ],
  "logging_steps": 1757,
  "max_steps": 35140,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 17570,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 30,
  "trial_name": null,
  "trial_params": null
}