{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.20015515903801395, "eval_steps": 60, "global_step": 258, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0007757951900698216, "grad_norm": NaN, "learning_rate": 0.0, "loss": 6.6871, "step": 1 }, { "epoch": 0.0015515903801396431, "grad_norm": 17.51189613342285, "learning_rate": 4.112808460634547e-08, "loss": 5.8546, "step": 2 }, { "epoch": 0.0023273855702094647, "grad_norm": 33.663917541503906, "learning_rate": 8.225616921269094e-08, "loss": 6.576, "step": 3 }, { "epoch": 0.0031031807602792862, "grad_norm": 23.573055267333984, "learning_rate": 1.2338425381903642e-07, "loss": 6.8301, "step": 4 }, { "epoch": 0.003878975950349108, "grad_norm": 9.426567077636719, "learning_rate": 1.6451233842538187e-07, "loss": 5.2308, "step": 5 }, { "epoch": 0.004654771140418929, "grad_norm": 19.919723510742188, "learning_rate": 2.0564042303172737e-07, "loss": 6.3588, "step": 6 }, { "epoch": 0.005430566330488751, "grad_norm": 20.105239868164062, "learning_rate": 2.4676850763807285e-07, "loss": 6.8417, "step": 7 }, { "epoch": 0.0062063615205585725, "grad_norm": Infinity, "learning_rate": 2.4676850763807285e-07, "loss": 10.3916, "step": 8 }, { "epoch": 0.0069821567106283944, "grad_norm": 10.925858497619629, "learning_rate": 2.878965922444183e-07, "loss": 5.5136, "step": 9 }, { "epoch": 0.007757951900698216, "grad_norm": 24.988182067871094, "learning_rate": 3.2902467685076374e-07, "loss": 7.4726, "step": 10 }, { "epoch": 0.008533747090768037, "grad_norm": 10.366790771484375, "learning_rate": 3.701527614571093e-07, "loss": 5.5414, "step": 11 }, { "epoch": 0.009309542280837859, "grad_norm": 10.754115104675293, "learning_rate": 4.1128084606345474e-07, "loss": 4.8507, "step": 12 }, { "epoch": 0.01008533747090768, "grad_norm": 11.047926902770996, "learning_rate": 4.524089306698002e-07, "loss": 4.9058, "step": 13 }, { "epoch": 0.010861132660977503, "grad_norm": 15.155884742736816, "learning_rate": 4.935370152761457e-07, "loss": 6.2301, "step": 14 }, { "epoch": 0.011636927851047323, "grad_norm": 32.3827018737793, "learning_rate": 5.346650998824911e-07, "loss": 7.9813, "step": 15 }, { "epoch": 0.012412723041117145, "grad_norm": 16.102903366088867, "learning_rate": 5.757931844888366e-07, "loss": 5.5269, "step": 16 }, { "epoch": 0.013188518231186967, "grad_norm": 53.7273063659668, "learning_rate": 6.169212690951821e-07, "loss": 10.0786, "step": 17 }, { "epoch": 0.013964313421256789, "grad_norm": 51.25851058959961, "learning_rate": 6.580493537015275e-07, "loss": 9.6075, "step": 18 }, { "epoch": 0.01474010861132661, "grad_norm": 10.667213439941406, "learning_rate": 6.99177438307873e-07, "loss": 4.671, "step": 19 }, { "epoch": 0.015515903801396431, "grad_norm": 24.299957275390625, "learning_rate": 7.403055229142186e-07, "loss": 6.945, "step": 20 }, { "epoch": 0.016291698991466253, "grad_norm": 16.173965454101562, "learning_rate": 7.81433607520564e-07, "loss": 6.233, "step": 21 }, { "epoch": 0.017067494181536073, "grad_norm": 10.136642456054688, "learning_rate": 8.225616921269095e-07, "loss": 5.5828, "step": 22 }, { "epoch": 0.017843289371605897, "grad_norm": 17.316375732421875, "learning_rate": 8.636897767332549e-07, "loss": 6.6991, "step": 23 }, { "epoch": 0.018619084561675717, "grad_norm": 12.166691780090332, "learning_rate": 9.048178613396004e-07, "loss": 5.9928, "step": 24 }, { "epoch": 0.019394879751745538, "grad_norm": 11.132185935974121, "learning_rate": 9.459459459459459e-07, "loss": 4.8443, "step": 25 }, { "epoch": 0.02017067494181536, "grad_norm": 57.284271240234375, "learning_rate": 9.870740305522914e-07, "loss": 11.0738, "step": 26 }, { "epoch": 0.02094647013188518, "grad_norm": 19.147293090820312, "learning_rate": 1.0282021151586369e-06, "loss": 6.4076, "step": 27 }, { "epoch": 0.021722265321955005, "grad_norm": 14.959942817687988, "learning_rate": 1.0693301997649822e-06, "loss": 6.2668, "step": 28 }, { "epoch": 0.022498060512024826, "grad_norm": 10.889761924743652, "learning_rate": 1.1104582843713277e-06, "loss": 5.092, "step": 29 }, { "epoch": 0.023273855702094646, "grad_norm": 13.842042922973633, "learning_rate": 1.1515863689776732e-06, "loss": 5.9955, "step": 30 }, { "epoch": 0.02404965089216447, "grad_norm": 19.306137084960938, "learning_rate": 1.1927144535840187e-06, "loss": 6.6071, "step": 31 }, { "epoch": 0.02482544608223429, "grad_norm": 10.03764533996582, "learning_rate": 1.2338425381903642e-06, "loss": 4.7984, "step": 32 }, { "epoch": 0.02560124127230411, "grad_norm": 15.16897964477539, "learning_rate": 1.2749706227967097e-06, "loss": 6.173, "step": 33 }, { "epoch": 0.026377036462373934, "grad_norm": 15.80406665802002, "learning_rate": 1.316098707403055e-06, "loss": 5.9604, "step": 34 }, { "epoch": 0.027152831652443754, "grad_norm": 17.38121795654297, "learning_rate": 1.3572267920094007e-06, "loss": 6.3435, "step": 35 }, { "epoch": 0.027928626842513578, "grad_norm": 12.920570373535156, "learning_rate": 1.398354876615746e-06, "loss": 6.0394, "step": 36 }, { "epoch": 0.028704422032583398, "grad_norm": 14.097166061401367, "learning_rate": 1.4394829612220915e-06, "loss": 6.0135, "step": 37 }, { "epoch": 0.02948021722265322, "grad_norm": 48.04243469238281, "learning_rate": 1.4806110458284372e-06, "loss": 8.0721, "step": 38 }, { "epoch": 0.030256012412723042, "grad_norm": 13.951966285705566, "learning_rate": 1.5217391304347825e-06, "loss": 6.2283, "step": 39 }, { "epoch": 0.031031807602792862, "grad_norm": 9.040204048156738, "learning_rate": 1.562867215041128e-06, "loss": 5.365, "step": 40 }, { "epoch": 0.03180760279286268, "grad_norm": 16.60372543334961, "learning_rate": 1.6039952996474733e-06, "loss": 6.0378, "step": 41 }, { "epoch": 0.032583397982932506, "grad_norm": 18.186500549316406, "learning_rate": 1.645123384253819e-06, "loss": 6.5136, "step": 42 }, { "epoch": 0.03335919317300233, "grad_norm": 10.157137870788574, "learning_rate": 1.6862514688601645e-06, "loss": 5.6955, "step": 43 }, { "epoch": 0.03413498836307215, "grad_norm": 12.988446235656738, "learning_rate": 1.7273795534665098e-06, "loss": 6.1769, "step": 44 }, { "epoch": 0.03491078355314197, "grad_norm": 17.56638526916504, "learning_rate": 1.7685076380728553e-06, "loss": 6.302, "step": 45 }, { "epoch": 0.035686578743211794, "grad_norm": 14.749937057495117, "learning_rate": 1.8096357226792008e-06, "loss": 6.0792, "step": 46 }, { "epoch": 0.03646237393328161, "grad_norm": 10.104000091552734, "learning_rate": 1.8507638072855463e-06, "loss": 5.4317, "step": 47 }, { "epoch": 0.037238169123351435, "grad_norm": 11.410161972045898, "learning_rate": 1.8918918918918918e-06, "loss": 5.7632, "step": 48 }, { "epoch": 0.03801396431342126, "grad_norm": 10.582472801208496, "learning_rate": 1.933019976498237e-06, "loss": 4.6767, "step": 49 }, { "epoch": 0.038789759503491075, "grad_norm": 14.601061820983887, "learning_rate": 1.9741480611045828e-06, "loss": 6.1871, "step": 50 }, { "epoch": 0.0395655546935609, "grad_norm": 10.493443489074707, "learning_rate": 2.015276145710928e-06, "loss": 5.572, "step": 51 }, { "epoch": 0.04034134988363072, "grad_norm": 24.4001522064209, "learning_rate": 2.0564042303172738e-06, "loss": 7.351, "step": 52 }, { "epoch": 0.041117145073700546, "grad_norm": 13.242673873901367, "learning_rate": 2.097532314923619e-06, "loss": 5.6907, "step": 53 }, { "epoch": 0.04189294026377036, "grad_norm": 11.323930740356445, "learning_rate": 2.1386603995299644e-06, "loss": 5.7986, "step": 54 }, { "epoch": 0.04266873545384019, "grad_norm": 13.408857345581055, "learning_rate": 2.17978848413631e-06, "loss": 5.9598, "step": 55 }, { "epoch": 0.04344453064391001, "grad_norm": 8.789331436157227, "learning_rate": 2.2209165687426554e-06, "loss": 6.024, "step": 56 }, { "epoch": 0.04422032583397983, "grad_norm": 10.100234985351562, "learning_rate": 2.2620446533490006e-06, "loss": 5.328, "step": 57 }, { "epoch": 0.04499612102404965, "grad_norm": 13.21589469909668, "learning_rate": 2.3031727379553464e-06, "loss": 5.5545, "step": 58 }, { "epoch": 0.045771916214119475, "grad_norm": 9.624663352966309, "learning_rate": 2.344300822561692e-06, "loss": 5.3813, "step": 59 }, { "epoch": 0.04654771140418929, "grad_norm": 5.808309555053711, "learning_rate": 2.3854289071680374e-06, "loss": 5.3023, "step": 60 }, { "epoch": 0.04654771140418929, "eval_Qnli-dev_cosine_accuracy": 0.59765625, "eval_Qnli-dev_cosine_accuracy_threshold": 0.9559981226921082, "eval_Qnli-dev_cosine_ap": 0.564511259877405, "eval_Qnli-dev_cosine_f1": 0.6291834002677376, "eval_Qnli-dev_cosine_f1_threshold": 0.7179018259048462, "eval_Qnli-dev_cosine_precision": 0.4598825831702544, "eval_Qnli-dev_cosine_recall": 0.9957627118644068, "eval_Qnli-dev_dot_accuracy": 0.55078125, "eval_Qnli-dev_dot_accuracy_threshold": 737.3067016601562, "eval_Qnli-dev_dot_ap": 0.4732733398818325, "eval_Qnli-dev_dot_f1": 0.629878869448183, "eval_Qnli-dev_dot_f1_threshold": 463.4846496582031, "eval_Qnli-dev_dot_precision": 0.46153846153846156, "eval_Qnli-dev_dot_recall": 0.9915254237288136, "eval_Qnli-dev_euclidean_accuracy": 0.6015625, "eval_Qnli-dev_euclidean_accuracy_threshold": 8.209783554077148, "eval_Qnli-dev_euclidean_ap": 0.5833709480494583, "eval_Qnli-dev_euclidean_f1": 0.6315789473684211, "eval_Qnli-dev_euclidean_f1_threshold": 16.22138214111328, "eval_Qnli-dev_euclidean_precision": 0.4633663366336634, "eval_Qnli-dev_euclidean_recall": 0.9915254237288136, "eval_Qnli-dev_manhattan_accuracy": 0.634765625, "eval_Qnli-dev_manhattan_accuracy_threshold": 146.8093719482422, "eval_Qnli-dev_manhattan_ap": 0.6173131758641012, "eval_Qnli-dev_manhattan_f1": 0.6291834002677376, "eval_Qnli-dev_manhattan_f1_threshold": 266.38592529296875, "eval_Qnli-dev_manhattan_precision": 0.4598825831702544, "eval_Qnli-dev_manhattan_recall": 0.9957627118644068, "eval_Qnli-dev_max_accuracy": 0.634765625, "eval_Qnli-dev_max_accuracy_threshold": 737.3067016601562, "eval_Qnli-dev_max_ap": 0.6173131758641012, "eval_Qnli-dev_max_f1": 0.6315789473684211, "eval_Qnli-dev_max_f1_threshold": 463.4846496582031, "eval_Qnli-dev_max_precision": 0.4633663366336634, "eval_Qnli-dev_max_recall": 0.9957627118644068, "eval_allNLI-dev_cosine_accuracy": 0.6640625, "eval_allNLI-dev_cosine_accuracy_threshold": 0.9944478273391724, "eval_allNLI-dev_cosine_ap": 0.37265957387885357, "eval_allNLI-dev_cosine_f1": 0.5104477611940298, "eval_allNLI-dev_cosine_f1_threshold": 0.8620990514755249, "eval_allNLI-dev_cosine_precision": 0.3440643863179074, "eval_allNLI-dev_cosine_recall": 0.9884393063583815, "eval_allNLI-dev_dot_accuracy": 0.66015625, "eval_allNLI-dev_dot_accuracy_threshold": 1061.320556640625, "eval_allNLI-dev_dot_ap": 0.329964611589565, "eval_allNLI-dev_dot_f1": 0.5088235294117648, "eval_allNLI-dev_dot_f1_threshold": 606.7950439453125, "eval_allNLI-dev_dot_precision": 0.34122287968441817, "eval_allNLI-dev_dot_recall": 1.0, "eval_allNLI-dev_euclidean_accuracy": 0.666015625, "eval_allNLI-dev_euclidean_accuracy_threshold": 5.134191513061523, "eval_allNLI-dev_euclidean_ap": 0.36839120834109285, "eval_allNLI-dev_euclidean_f1": 0.5113464447806354, "eval_allNLI-dev_euclidean_f1_threshold": 14.78046989440918, "eval_allNLI-dev_euclidean_precision": 0.3463114754098361, "eval_allNLI-dev_euclidean_recall": 0.976878612716763, "eval_allNLI-dev_manhattan_accuracy": 0.6640625, "eval_allNLI-dev_manhattan_accuracy_threshold": 54.993194580078125, "eval_allNLI-dev_manhattan_ap": 0.38115647907317707, "eval_allNLI-dev_manhattan_f1": 0.5111441307578009, "eval_allNLI-dev_manhattan_f1_threshold": 225.10543823242188, "eval_allNLI-dev_manhattan_precision": 0.344, "eval_allNLI-dev_manhattan_recall": 0.9942196531791907, "eval_allNLI-dev_max_accuracy": 0.666015625, "eval_allNLI-dev_max_accuracy_threshold": 1061.320556640625, "eval_allNLI-dev_max_ap": 0.38115647907317707, "eval_allNLI-dev_max_f1": 0.5113464447806354, "eval_allNLI-dev_max_f1_threshold": 606.7950439453125, "eval_allNLI-dev_max_precision": 0.3463114754098361, "eval_allNLI-dev_max_recall": 1.0, "eval_sequential_score": 0.6173131758641012, "eval_sts-test_pearson_cosine": 0.15527534331116627, "eval_sts-test_pearson_dot": 0.1562898413481788, "eval_sts-test_pearson_euclidean": 0.1565855982058341, "eval_sts-test_pearson_manhattan": 0.14924616505059007, "eval_sts-test_pearson_max": 0.1565855982058341, "eval_sts-test_spearman_cosine": 0.2096244403759917, "eval_sts-test_spearman_dot": 0.155353153646277, "eval_sts-test_spearman_euclidean": 0.1821313239351398, "eval_sts-test_spearman_manhattan": 0.174985308262377, "eval_sts-test_spearman_max": 0.2096244403759917, "eval_vitaminc-pairs_loss": 3.995089530944824, "eval_vitaminc-pairs_runtime": 5.4908, "eval_vitaminc-pairs_samples_per_second": 23.312, "eval_vitaminc-pairs_steps_per_second": 0.364, "step": 60 }, { "epoch": 0.04654771140418929, "eval_negation-triplets_loss": 5.3504109382629395, "eval_negation-triplets_runtime": 1.115, "eval_negation-triplets_samples_per_second": 114.8, "eval_negation-triplets_steps_per_second": 1.794, "step": 60 }, { "epoch": 0.04654771140418929, "eval_scitail-pairs-pos_loss": 2.7031445503234863, "eval_scitail-pairs-pos_runtime": 1.238, "eval_scitail-pairs-pos_samples_per_second": 103.395, "eval_scitail-pairs-pos_steps_per_second": 1.616, "step": 60 }, { "epoch": 0.04654771140418929, "eval_scitail-pairs-qa_loss": 3.3194987773895264, "eval_scitail-pairs-qa_runtime": 0.9091, "eval_scitail-pairs-qa_samples_per_second": 140.792, "eval_scitail-pairs-qa_steps_per_second": 2.2, "step": 60 }, { "epoch": 0.04654771140418929, "eval_xsum-pairs_loss": 5.565915584564209, "eval_xsum-pairs_runtime": 4.303, "eval_xsum-pairs_samples_per_second": 29.747, "eval_xsum-pairs_steps_per_second": 0.465, "step": 60 }, { "epoch": 0.04654771140418929, "eval_sciq_pairs_loss": 0.9480071067810059, "eval_sciq_pairs_runtime": 5.5715, "eval_sciq_pairs_samples_per_second": 22.974, "eval_sciq_pairs_steps_per_second": 0.359, "step": 60 }, { "epoch": 0.04654771140418929, "eval_qasc_pairs_loss": 6.708042144775391, "eval_qasc_pairs_runtime": 0.978, "eval_qasc_pairs_samples_per_second": 130.881, "eval_qasc_pairs_steps_per_second": 2.045, "step": 60 }, { "epoch": 0.04654771140418929, "eval_openbookqa_pairs_loss": 5.462226390838623, "eval_openbookqa_pairs_runtime": 0.8592, "eval_openbookqa_pairs_samples_per_second": 148.973, "eval_openbookqa_pairs_steps_per_second": 2.328, "step": 60 }, { "epoch": 0.04654771140418929, "eval_msmarco_pairs_loss": 8.80248737335205, "eval_msmarco_pairs_runtime": 1.9622, "eval_msmarco_pairs_samples_per_second": 65.234, "eval_msmarco_pairs_steps_per_second": 1.019, "step": 60 }, { "epoch": 0.04654771140418929, "eval_nq_pairs_loss": 7.891404628753662, "eval_nq_pairs_runtime": 4.4106, "eval_nq_pairs_samples_per_second": 29.021, "eval_nq_pairs_steps_per_second": 0.453, "step": 60 }, { "epoch": 0.04654771140418929, "eval_trivia_pairs_loss": 6.30972146987915, "eval_trivia_pairs_runtime": 4.5239, "eval_trivia_pairs_samples_per_second": 28.294, "eval_trivia_pairs_steps_per_second": 0.442, "step": 60 }, { "epoch": 0.04654771140418929, "eval_gooaq_pairs_loss": 6.851067066192627, "eval_gooaq_pairs_runtime": 1.2706, "eval_gooaq_pairs_samples_per_second": 100.74, "eval_gooaq_pairs_steps_per_second": 1.574, "step": 60 }, { "epoch": 0.04654771140418929, "eval_paws-pos_loss": 2.74210262298584, "eval_paws-pos_runtime": 1.0213, "eval_paws-pos_samples_per_second": 125.326, "eval_paws-pos_steps_per_second": 1.958, "step": 60 }, { "epoch": 0.04654771140418929, "eval_global_dataset_loss": 5.345330715179443, "eval_global_dataset_runtime": 21.1331, "eval_global_dataset_samples_per_second": 19.685, "eval_global_dataset_steps_per_second": 0.331, "step": 60 }, { "epoch": 0.047323506594259115, "grad_norm": 15.78637981414795, "learning_rate": 2.426556991774383e-06, "loss": 3.5705, "step": 61 }, { "epoch": 0.04809930178432894, "grad_norm": 7.229628562927246, "learning_rate": 2.4676850763807284e-06, "loss": 5.8784, "step": 62 }, { "epoch": 0.048875096974398756, "grad_norm": 35.124271392822266, "learning_rate": 2.5088131609870737e-06, "loss": 8.5966, "step": 63 }, { "epoch": 0.04965089216446858, "grad_norm": 10.43959903717041, "learning_rate": 2.5499412455934194e-06, "loss": 5.9225, "step": 64 }, { "epoch": 0.0504266873545384, "grad_norm": 9.716011047363281, "learning_rate": 2.5910693301997647e-06, "loss": 4.6667, "step": 65 }, { "epoch": 0.05120248254460822, "grad_norm": 12.058530807495117, "learning_rate": 2.63219741480611e-06, "loss": 6.12, "step": 66 }, { "epoch": 0.051978277734678044, "grad_norm": 9.680259704589844, "learning_rate": 2.6733254994124557e-06, "loss": 5.2713, "step": 67 }, { "epoch": 0.05275407292474787, "grad_norm": 20.48016929626465, "learning_rate": 2.7144535840188014e-06, "loss": 6.9483, "step": 68 }, { "epoch": 0.05352986811481769, "grad_norm": 15.96925163269043, "learning_rate": 2.7555816686251467e-06, "loss": 3.6892, "step": 69 }, { "epoch": 0.05430566330488751, "grad_norm": 25.622777938842773, "learning_rate": 2.796709753231492e-06, "loss": 7.5618, "step": 70 }, { "epoch": 0.05508145849495733, "grad_norm": 9.09017562866211, "learning_rate": 2.8378378378378377e-06, "loss": 5.8086, "step": 71 }, { "epoch": 0.055857253685027156, "grad_norm": 23.944202423095703, "learning_rate": 2.878965922444183e-06, "loss": 6.517, "step": 72 }, { "epoch": 0.05663304887509697, "grad_norm": 9.879794120788574, "learning_rate": 2.9200940070505282e-06, "loss": 5.5826, "step": 73 }, { "epoch": 0.057408844065166796, "grad_norm": 16.403295516967773, "learning_rate": 2.9612220916568744e-06, "loss": 6.5543, "step": 74 }, { "epoch": 0.05818463925523662, "grad_norm": 25.534475326538086, "learning_rate": 3.0023501762632197e-06, "loss": 6.5721, "step": 75 }, { "epoch": 0.05896043444530644, "grad_norm": 8.420788764953613, "learning_rate": 3.043478260869565e-06, "loss": 5.4454, "step": 76 }, { "epoch": 0.05973622963537626, "grad_norm": 20.24285888671875, "learning_rate": 3.0846063454759102e-06, "loss": 6.3208, "step": 77 }, { "epoch": 0.060512024825446084, "grad_norm": 15.204730987548828, "learning_rate": 3.125734430082256e-06, "loss": 6.5299, "step": 78 }, { "epoch": 0.0612878200155159, "grad_norm": 20.533952713012695, "learning_rate": 3.1668625146886012e-06, "loss": 7.1424, "step": 79 }, { "epoch": 0.062063615205585725, "grad_norm": 8.96998405456543, "learning_rate": 3.2079905992949465e-06, "loss": 5.1799, "step": 80 }, { "epoch": 0.06283941039565555, "grad_norm": NaN, "learning_rate": 3.2079905992949465e-06, "loss": 6.3358, "step": 81 }, { "epoch": 0.06361520558572537, "grad_norm": 28.576021194458008, "learning_rate": 3.2491186839012927e-06, "loss": 7.9256, "step": 82 }, { "epoch": 0.0643910007757952, "grad_norm": 8.81160831451416, "learning_rate": 3.290246768507638e-06, "loss": 5.4606, "step": 83 }, { "epoch": 0.06516679596586501, "grad_norm": 8.355425834655762, "learning_rate": 3.3313748531139832e-06, "loss": 5.6861, "step": 84 }, { "epoch": 0.06594259115593483, "grad_norm": 16.12509536743164, "learning_rate": 3.372502937720329e-06, "loss": 3.6595, "step": 85 }, { "epoch": 0.06671838634600466, "grad_norm": 7.375875473022461, "learning_rate": 3.4136310223266742e-06, "loss": 5.3626, "step": 86 }, { "epoch": 0.06749418153607448, "grad_norm": 7.526387691497803, "learning_rate": 3.4547591069330195e-06, "loss": 5.7214, "step": 87 }, { "epoch": 0.0682699767261443, "grad_norm": 16.23571014404297, "learning_rate": 3.495887191539365e-06, "loss": 3.7724, "step": 88 }, { "epoch": 0.06904577191621412, "grad_norm": 19.215839385986328, "learning_rate": 3.5370152761457105e-06, "loss": 6.8782, "step": 89 }, { "epoch": 0.06982156710628394, "grad_norm": 20.674381256103516, "learning_rate": 3.5781433607520563e-06, "loss": 7.1858, "step": 90 }, { "epoch": 0.07059736229635376, "grad_norm": 12.960572242736816, "learning_rate": 3.6192714453584015e-06, "loss": 6.1052, "step": 91 }, { "epoch": 0.07137315748642359, "grad_norm": 5.505962371826172, "learning_rate": 3.6603995299647473e-06, "loss": 5.6906, "step": 92 }, { "epoch": 0.0721489526764934, "grad_norm": 13.981863021850586, "learning_rate": 3.7015276145710925e-06, "loss": 6.3279, "step": 93 }, { "epoch": 0.07292474786656322, "grad_norm": 15.760379791259766, "learning_rate": 3.742655699177438e-06, "loss": 6.6561, "step": 94 }, { "epoch": 0.07370054305663305, "grad_norm": 4.809438705444336, "learning_rate": 3.7837837837837835e-06, "loss": 5.634, "step": 95 }, { "epoch": 0.07447633824670287, "grad_norm": 6.97947883605957, "learning_rate": 3.824911868390129e-06, "loss": 5.6729, "step": 96 }, { "epoch": 0.07525213343677269, "grad_norm": 6.912349224090576, "learning_rate": 3.866039952996474e-06, "loss": 5.1085, "step": 97 }, { "epoch": 0.07602792862684252, "grad_norm": 16.25594711303711, "learning_rate": 3.907168037602819e-06, "loss": 6.7441, "step": 98 }, { "epoch": 0.07680372381691233, "grad_norm": 12.987081527709961, "learning_rate": 3.9482961222091655e-06, "loss": 6.2151, "step": 99 }, { "epoch": 0.07757951900698215, "grad_norm": 16.0250244140625, "learning_rate": 3.989424206815511e-06, "loss": 6.6636, "step": 100 }, { "epoch": 0.07835531419705198, "grad_norm": 7.364856243133545, "learning_rate": 4.030552291421856e-06, "loss": 5.3276, "step": 101 }, { "epoch": 0.0791311093871218, "grad_norm": 7.465214252471924, "learning_rate": 4.071680376028202e-06, "loss": 5.1297, "step": 102 }, { "epoch": 0.07990690457719161, "grad_norm": 7.24612283706665, "learning_rate": 4.1128084606345476e-06, "loss": 5.0139, "step": 103 }, { "epoch": 0.08068269976726145, "grad_norm": 6.434986591339111, "learning_rate": 4.153936545240893e-06, "loss": 5.029, "step": 104 }, { "epoch": 0.08145849495733126, "grad_norm": 4.641157627105713, "learning_rate": 4.195064629847238e-06, "loss": 5.605, "step": 105 }, { "epoch": 0.08223429014740109, "grad_norm": 7.490912437438965, "learning_rate": 4.236192714453583e-06, "loss": 5.0171, "step": 106 }, { "epoch": 0.08301008533747091, "grad_norm": 17.2343692779541, "learning_rate": 4.277320799059929e-06, "loss": 3.6544, "step": 107 }, { "epoch": 0.08378588052754073, "grad_norm": 7.838309288024902, "learning_rate": 4.318448883666274e-06, "loss": 5.9355, "step": 108 }, { "epoch": 0.08456167571761056, "grad_norm": 8.586014747619629, "learning_rate": 4.35957696827262e-06, "loss": 5.7908, "step": 109 }, { "epoch": 0.08533747090768037, "grad_norm": 7.035303592681885, "learning_rate": 4.400705052878966e-06, "loss": 5.2447, "step": 110 }, { "epoch": 0.08611326609775019, "grad_norm": 15.303041458129883, "learning_rate": 4.441833137485311e-06, "loss": 6.5699, "step": 111 }, { "epoch": 0.08688906128782002, "grad_norm": 6.91060733795166, "learning_rate": 4.482961222091657e-06, "loss": 4.8604, "step": 112 }, { "epoch": 0.08766485647788984, "grad_norm": 4.178639888763428, "learning_rate": 4.524089306698001e-06, "loss": 5.6599, "step": 113 }, { "epoch": 0.08844065166795965, "grad_norm": 7.758280277252197, "learning_rate": 4.5652173913043474e-06, "loss": 4.9483, "step": 114 }, { "epoch": 0.08921644685802949, "grad_norm": 8.945616722106934, "learning_rate": 4.606345475910693e-06, "loss": 5.7634, "step": 115 }, { "epoch": 0.0899922420480993, "grad_norm": 6.94469690322876, "learning_rate": 4.647473560517038e-06, "loss": 5.0934, "step": 116 }, { "epoch": 0.09076803723816912, "grad_norm": 8.813899040222168, "learning_rate": 4.688601645123384e-06, "loss": 4.5253, "step": 117 }, { "epoch": 0.09154383242823895, "grad_norm": 10.021103858947754, "learning_rate": 4.7297297297297294e-06, "loss": 4.6447, "step": 118 }, { "epoch": 0.09231962761830877, "grad_norm": 4.420945644378662, "learning_rate": 4.770857814336075e-06, "loss": 5.5944, "step": 119 }, { "epoch": 0.09309542280837858, "grad_norm": 8.930946350097656, "learning_rate": 4.81198589894242e-06, "loss": 4.4379, "step": 120 }, { "epoch": 0.09309542280837858, "eval_Qnli-dev_cosine_accuracy": 0.611328125, "eval_Qnli-dev_cosine_accuracy_threshold": 0.9671535491943359, "eval_Qnli-dev_cosine_ap": 0.5751874558355262, "eval_Qnli-dev_cosine_f1": 0.6291834002677376, "eval_Qnli-dev_cosine_f1_threshold": 0.7754666805267334, "eval_Qnli-dev_cosine_precision": 0.4598825831702544, "eval_Qnli-dev_cosine_recall": 0.9957627118644068, "eval_Qnli-dev_dot_accuracy": 0.548828125, "eval_Qnli-dev_dot_accuracy_threshold": 812.9189453125, "eval_Qnli-dev_dot_ap": 0.47763633423847596, "eval_Qnli-dev_dot_f1": 0.6308724832214765, "eval_Qnli-dev_dot_f1_threshold": 487.3522033691406, "eval_Qnli-dev_dot_precision": 0.46168958742632615, "eval_Qnli-dev_dot_recall": 0.9957627118644068, "eval_Qnli-dev_euclidean_accuracy": 0.609375, "eval_Qnli-dev_euclidean_accuracy_threshold": 8.314947128295898, "eval_Qnli-dev_euclidean_ap": 0.5924255839030361, "eval_Qnli-dev_euclidean_f1": 0.6300268096514745, "eval_Qnli-dev_euclidean_f1_threshold": 16.918926239013672, "eval_Qnli-dev_euclidean_precision": 0.46078431372549017, "eval_Qnli-dev_euclidean_recall": 0.9957627118644068, "eval_Qnli-dev_manhattan_accuracy": 0.630859375, "eval_Qnli-dev_manhattan_accuracy_threshold": 138.335693359375, "eval_Qnli-dev_manhattan_ap": 0.6151670800573863, "eval_Qnli-dev_manhattan_f1": 0.6325966850828729, "eval_Qnli-dev_manhattan_f1_threshold": 208.749267578125, "eval_Qnli-dev_manhattan_precision": 0.4692622950819672, "eval_Qnli-dev_manhattan_recall": 0.9703389830508474, "eval_Qnli-dev_max_accuracy": 0.630859375, "eval_Qnli-dev_max_accuracy_threshold": 812.9189453125, "eval_Qnli-dev_max_ap": 0.6151670800573863, "eval_Qnli-dev_max_f1": 0.6325966850828729, "eval_Qnli-dev_max_f1_threshold": 487.3522033691406, "eval_Qnli-dev_max_precision": 0.4692622950819672, "eval_Qnli-dev_max_recall": 0.9957627118644068, "eval_allNLI-dev_cosine_accuracy": 0.673828125, "eval_allNLI-dev_cosine_accuracy_threshold": 0.9807866811752319, "eval_allNLI-dev_cosine_ap": 0.3981958701708742, "eval_allNLI-dev_cosine_f1": 0.5135951661631419, "eval_allNLI-dev_cosine_f1_threshold": 0.8885090351104736, "eval_allNLI-dev_cosine_precision": 0.3476482617586912, "eval_allNLI-dev_cosine_recall": 0.9826589595375722, "eval_allNLI-dev_dot_accuracy": 0.66015625, "eval_allNLI-dev_dot_accuracy_threshold": 1020.55224609375, "eval_allNLI-dev_dot_ap": 0.3253263107868747, "eval_allNLI-dev_dot_f1": 0.5089285714285714, "eval_allNLI-dev_dot_f1_threshold": 662.3919067382812, "eval_allNLI-dev_dot_precision": 0.342685370741483, "eval_allNLI-dev_dot_recall": 0.9884393063583815, "eval_allNLI-dev_euclidean_accuracy": 0.66796875, "eval_allNLI-dev_euclidean_accuracy_threshold": 5.130853652954102, "eval_allNLI-dev_euclidean_ap": 0.3883458149835009, "eval_allNLI-dev_euclidean_f1": 0.5128205128205128, "eval_allNLI-dev_euclidean_f1_threshold": 14.313749313354492, "eval_allNLI-dev_euclidean_precision": 0.3469387755102041, "eval_allNLI-dev_euclidean_recall": 0.9826589595375722, "eval_allNLI-dev_manhattan_accuracy": 0.66796875, "eval_allNLI-dev_manhattan_accuracy_threshold": 89.57350158691406, "eval_allNLI-dev_manhattan_ap": 0.40232496525900585, "eval_allNLI-dev_manhattan_f1": 0.5089820359281437, "eval_allNLI-dev_manhattan_f1_threshold": 217.869384765625, "eval_allNLI-dev_manhattan_precision": 0.3434343434343434, "eval_allNLI-dev_manhattan_recall": 0.9826589595375722, "eval_allNLI-dev_max_accuracy": 0.673828125, "eval_allNLI-dev_max_accuracy_threshold": 1020.55224609375, "eval_allNLI-dev_max_ap": 0.40232496525900585, "eval_allNLI-dev_max_f1": 0.5135951661631419, "eval_allNLI-dev_max_f1_threshold": 662.3919067382812, "eval_allNLI-dev_max_precision": 0.3476482617586912, "eval_allNLI-dev_max_recall": 0.9884393063583815, "eval_sequential_score": 0.6151670800573863, "eval_sts-test_pearson_cosine": 0.17530815823467685, "eval_sts-test_pearson_dot": 0.18707707355601674, "eval_sts-test_pearson_euclidean": 0.17782964327059717, "eval_sts-test_pearson_manhattan": 0.16938653812007057, "eval_sts-test_pearson_max": 0.18707707355601674, "eval_sts-test_spearman_cosine": 0.23193070661653478, "eval_sts-test_spearman_dot": 0.1855223507670731, "eval_sts-test_spearman_euclidean": 0.2037636361791284, "eval_sts-test_spearman_manhattan": 0.19557191778143276, "eval_sts-test_spearman_max": 0.23193070661653478, "eval_vitaminc-pairs_loss": 3.9059581756591797, "eval_vitaminc-pairs_runtime": 5.4911, "eval_vitaminc-pairs_samples_per_second": 23.311, "eval_vitaminc-pairs_steps_per_second": 0.364, "step": 120 }, { "epoch": 0.09309542280837858, "eval_negation-triplets_loss": 5.277580738067627, "eval_negation-triplets_runtime": 1.1046, "eval_negation-triplets_samples_per_second": 115.878, "eval_negation-triplets_steps_per_second": 1.811, "step": 120 }, { "epoch": 0.09309542280837858, "eval_scitail-pairs-pos_loss": 2.574610471725464, "eval_scitail-pairs-pos_runtime": 1.2106, "eval_scitail-pairs-pos_samples_per_second": 105.735, "eval_scitail-pairs-pos_steps_per_second": 1.652, "step": 120 }, { "epoch": 0.09309542280837858, "eval_scitail-pairs-qa_loss": 3.310026168823242, "eval_scitail-pairs-qa_runtime": 0.9039, "eval_scitail-pairs-qa_samples_per_second": 141.61, "eval_scitail-pairs-qa_steps_per_second": 2.213, "step": 120 }, { "epoch": 0.09309542280837858, "eval_xsum-pairs_loss": 5.120739936828613, "eval_xsum-pairs_runtime": 4.3092, "eval_xsum-pairs_samples_per_second": 29.704, "eval_xsum-pairs_steps_per_second": 0.464, "step": 120 }, { "epoch": 0.09309542280837858, "eval_sciq_pairs_loss": 0.878578245639801, "eval_sciq_pairs_runtime": 5.5099, "eval_sciq_pairs_samples_per_second": 23.231, "eval_sciq_pairs_steps_per_second": 0.363, "step": 120 }, { "epoch": 0.09309542280837858, "eval_qasc_pairs_loss": 5.170502185821533, "eval_qasc_pairs_runtime": 1.0003, "eval_qasc_pairs_samples_per_second": 127.96, "eval_qasc_pairs_steps_per_second": 1.999, "step": 120 }, { "epoch": 0.09309542280837858, "eval_openbookqa_pairs_loss": 5.240739822387695, "eval_openbookqa_pairs_runtime": 0.8534, "eval_openbookqa_pairs_samples_per_second": 149.988, "eval_openbookqa_pairs_steps_per_second": 2.344, "step": 120 }, { "epoch": 0.09309542280837858, "eval_msmarco_pairs_loss": 6.330340385437012, "eval_msmarco_pairs_runtime": 1.955, "eval_msmarco_pairs_samples_per_second": 65.475, "eval_msmarco_pairs_steps_per_second": 1.023, "step": 120 }, { "epoch": 0.09309542280837858, "eval_nq_pairs_loss": 5.914206504821777, "eval_nq_pairs_runtime": 4.4057, "eval_nq_pairs_samples_per_second": 29.053, "eval_nq_pairs_steps_per_second": 0.454, "step": 120 }, { "epoch": 0.09309542280837858, "eval_trivia_pairs_loss": 5.414165019989014, "eval_trivia_pairs_runtime": 4.5232, "eval_trivia_pairs_samples_per_second": 28.299, "eval_trivia_pairs_steps_per_second": 0.442, "step": 120 }, { "epoch": 0.09309542280837858, "eval_gooaq_pairs_loss": 5.54625129699707, "eval_gooaq_pairs_runtime": 1.2712, "eval_gooaq_pairs_samples_per_second": 100.693, "eval_gooaq_pairs_steps_per_second": 1.573, "step": 120 }, { "epoch": 0.09309542280837858, "eval_paws-pos_loss": 2.632805585861206, "eval_paws-pos_runtime": 1.0285, "eval_paws-pos_samples_per_second": 124.448, "eval_paws-pos_steps_per_second": 1.944, "step": 120 }, { "epoch": 0.09309542280837858, "eval_global_dataset_loss": 4.557917594909668, "eval_global_dataset_runtime": 21.1343, "eval_global_dataset_samples_per_second": 19.684, "eval_global_dataset_steps_per_second": 0.331, "step": 120 }, { "epoch": 0.09387121799844841, "grad_norm": 8.41954517364502, "learning_rate": 4.853113983548766e-06, "loss": 4.9112, "step": 121 }, { "epoch": 0.09464701318851823, "grad_norm": 11.133651733398438, "learning_rate": 4.894242068155111e-06, "loss": 4.6273, "step": 122 }, { "epoch": 0.09542280837858805, "grad_norm": 8.026440620422363, "learning_rate": 4.935370152761457e-06, "loss": 5.5262, "step": 123 }, { "epoch": 0.09619860356865788, "grad_norm": 8.465253829956055, "learning_rate": 4.976498237367803e-06, "loss": 4.837, "step": 124 }, { "epoch": 0.0969743987587277, "grad_norm": 16.9398193359375, "learning_rate": 5.017626321974147e-06, "loss": 6.3641, "step": 125 }, { "epoch": 0.09775019394879751, "grad_norm": 8.868680953979492, "learning_rate": 5.0587544065804934e-06, "loss": 4.6542, "step": 126 }, { "epoch": 0.09852598913886734, "grad_norm": 21.058149337768555, "learning_rate": 5.099882491186839e-06, "loss": 3.2895, "step": 127 }, { "epoch": 0.09930178432893716, "grad_norm": 10.110532760620117, "learning_rate": 5.141010575793184e-06, "loss": 5.5488, "step": 128 }, { "epoch": 0.10007757951900698, "grad_norm": 17.80249786376953, "learning_rate": 5.182138660399529e-06, "loss": 6.3668, "step": 129 }, { "epoch": 0.1008533747090768, "grad_norm": 11.01941204071045, "learning_rate": 5.2232667450058754e-06, "loss": 6.0549, "step": 130 }, { "epoch": 0.10162916989914662, "grad_norm": 9.745173454284668, "learning_rate": 5.26439482961222e-06, "loss": 4.7537, "step": 131 }, { "epoch": 0.10240496508921644, "grad_norm": 10.010377883911133, "learning_rate": 5.305522914218566e-06, "loss": 4.245, "step": 132 }, { "epoch": 0.10318076027928627, "grad_norm": 13.334067344665527, "learning_rate": 5.346650998824911e-06, "loss": 4.3526, "step": 133 }, { "epoch": 0.10395655546935609, "grad_norm": 23.037643432617188, "learning_rate": 5.387779083431257e-06, "loss": 2.9629, "step": 134 }, { "epoch": 0.10473235065942592, "grad_norm": 10.07296085357666, "learning_rate": 5.428907168037603e-06, "loss": 4.4471, "step": 135 }, { "epoch": 0.10550814584949574, "grad_norm": 9.801810264587402, "learning_rate": 5.470035252643947e-06, "loss": 4.5066, "step": 136 }, { "epoch": 0.10628394103956555, "grad_norm": 9.673730850219727, "learning_rate": 5.511163337250293e-06, "loss": 4.7698, "step": 137 }, { "epoch": 0.10705973622963538, "grad_norm": 15.595386505126953, "learning_rate": 5.5522914218566394e-06, "loss": 4.706, "step": 138 }, { "epoch": 0.1078355314197052, "grad_norm": 10.181653022766113, "learning_rate": 5.593419506462984e-06, "loss": 4.4883, "step": 139 }, { "epoch": 0.10861132660977502, "grad_norm": 16.41044807434082, "learning_rate": 5.63454759106933e-06, "loss": 6.0553, "step": 140 }, { "epoch": 0.10938712179984485, "grad_norm": 15.553648948669434, "learning_rate": 5.675675675675675e-06, "loss": 5.8958, "step": 141 }, { "epoch": 0.11016291698991466, "grad_norm": 11.734206199645996, "learning_rate": 5.716803760282021e-06, "loss": 4.1842, "step": 142 }, { "epoch": 0.11093871217998448, "grad_norm": 20.94911766052246, "learning_rate": 5.757931844888366e-06, "loss": 6.3887, "step": 143 }, { "epoch": 0.11171450737005431, "grad_norm": 17.768552780151367, "learning_rate": 5.799059929494712e-06, "loss": 4.0725, "step": 144 }, { "epoch": 0.11249030256012413, "grad_norm": 11.716679573059082, "learning_rate": 5.8401880141010565e-06, "loss": 4.6545, "step": 145 }, { "epoch": 0.11326609775019394, "grad_norm": 16.473339080810547, "learning_rate": 5.881316098707403e-06, "loss": 6.1092, "step": 146 }, { "epoch": 0.11404189294026378, "grad_norm": 11.649435997009277, "learning_rate": 5.922444183313749e-06, "loss": 3.7272, "step": 147 }, { "epoch": 0.11481768813033359, "grad_norm": 12.287504196166992, "learning_rate": 5.963572267920093e-06, "loss": 4.1651, "step": 148 }, { "epoch": 0.11559348332040341, "grad_norm": 12.262931823730469, "learning_rate": 6.004700352526439e-06, "loss": 3.8952, "step": 149 }, { "epoch": 0.11636927851047324, "grad_norm": 14.264664649963379, "learning_rate": 6.045828437132785e-06, "loss": 4.6196, "step": 150 }, { "epoch": 0.11714507370054306, "grad_norm": 12.257019996643066, "learning_rate": 6.08695652173913e-06, "loss": 3.7705, "step": 151 }, { "epoch": 0.11792086889061287, "grad_norm": 29.75714874267578, "learning_rate": 6.128084606345475e-06, "loss": 6.073, "step": 152 }, { "epoch": 0.1186966640806827, "grad_norm": 11.830326080322266, "learning_rate": 6.1692126909518205e-06, "loss": 3.7738, "step": 153 }, { "epoch": 0.11947245927075252, "grad_norm": 17.84813117980957, "learning_rate": 6.210340775558167e-06, "loss": 3.6523, "step": 154 }, { "epoch": 0.12024825446082234, "grad_norm": 18.622961044311523, "learning_rate": 6.251468860164512e-06, "loss": 5.4226, "step": 155 }, { "epoch": 0.12102404965089217, "grad_norm": 11.234856605529785, "learning_rate": 6.292596944770857e-06, "loss": 4.5508, "step": 156 }, { "epoch": 0.12179984484096198, "grad_norm": 12.604881286621094, "learning_rate": 6.3337250293772025e-06, "loss": 3.9043, "step": 157 }, { "epoch": 0.1225756400310318, "grad_norm": 13.025086402893066, "learning_rate": 6.374853113983549e-06, "loss": 3.66, "step": 158 }, { "epoch": 0.12335143522110163, "grad_norm": 36.854610443115234, "learning_rate": 6.415981198589893e-06, "loss": 6.0984, "step": 159 }, { "epoch": 0.12412723041117145, "grad_norm": 11.726686477661133, "learning_rate": 6.457109283196239e-06, "loss": 3.9498, "step": 160 }, { "epoch": 0.12490302560124127, "grad_norm": 11.277478218078613, "learning_rate": 6.498237367802585e-06, "loss": 4.4688, "step": 161 }, { "epoch": 0.1256788207913111, "grad_norm": 13.619148254394531, "learning_rate": 6.53936545240893e-06, "loss": 3.6255, "step": 162 }, { "epoch": 0.1264546159813809, "grad_norm": 13.29480266571045, "learning_rate": 6.580493537015276e-06, "loss": 3.658, "step": 163 }, { "epoch": 0.12723041117145073, "grad_norm": 12.314528465270996, "learning_rate": 6.621621621621621e-06, "loss": 3.4856, "step": 164 }, { "epoch": 0.12800620636152055, "grad_norm": 13.117006301879883, "learning_rate": 6.6627497062279665e-06, "loss": 5.3092, "step": 165 }, { "epoch": 0.1287820015515904, "grad_norm": 16.65713882446289, "learning_rate": 6.703877790834312e-06, "loss": 3.7321, "step": 166 }, { "epoch": 0.1295577967416602, "grad_norm": 13.305086135864258, "learning_rate": 6.745005875440658e-06, "loss": 3.2995, "step": 167 }, { "epoch": 0.13033359193173003, "grad_norm": 17.29372215270996, "learning_rate": 6.786133960047002e-06, "loss": 5.1161, "step": 168 }, { "epoch": 0.13110938712179984, "grad_norm": 12.293781280517578, "learning_rate": 6.8272620446533485e-06, "loss": 3.614, "step": 169 }, { "epoch": 0.13188518231186966, "grad_norm": 14.023563385009766, "learning_rate": 6.868390129259695e-06, "loss": 4.0901, "step": 170 }, { "epoch": 0.13266097750193948, "grad_norm": 12.515665054321289, "learning_rate": 6.909518213866039e-06, "loss": 3.4437, "step": 171 }, { "epoch": 0.13343677269200932, "grad_norm": 19.10329246520996, "learning_rate": 6.950646298472385e-06, "loss": 5.0212, "step": 172 }, { "epoch": 0.13421256788207914, "grad_norm": 22.987899780273438, "learning_rate": 6.99177438307873e-06, "loss": 1.3904, "step": 173 }, { "epoch": 0.13498836307214895, "grad_norm": 23.559511184692383, "learning_rate": 7.032902467685076e-06, "loss": 5.6536, "step": 174 }, { "epoch": 0.13576415826221877, "grad_norm": 15.472027778625488, "learning_rate": 7.074030552291421e-06, "loss": 5.0981, "step": 175 }, { "epoch": 0.1365399534522886, "grad_norm": 15.673625946044922, "learning_rate": 7.115158636897766e-06, "loss": 4.7528, "step": 176 }, { "epoch": 0.1373157486423584, "grad_norm": 31.196725845336914, "learning_rate": 7.1562867215041125e-06, "loss": 5.4556, "step": 177 }, { "epoch": 0.13809154383242825, "grad_norm": 13.222643852233887, "learning_rate": 7.197414806110458e-06, "loss": 2.8553, "step": 178 }, { "epoch": 0.13886733902249807, "grad_norm": 25.982681274414062, "learning_rate": 7.238542890716803e-06, "loss": 5.4703, "step": 179 }, { "epoch": 0.13964313421256788, "grad_norm": 20.191654205322266, "learning_rate": 7.279670975323148e-06, "loss": 4.7665, "step": 180 }, { "epoch": 0.13964313421256788, "eval_Qnli-dev_cosine_accuracy": 0.634765625, "eval_Qnli-dev_cosine_accuracy_threshold": 0.9040530323982239, "eval_Qnli-dev_cosine_ap": 0.6065022391968738, "eval_Qnli-dev_cosine_f1": 0.6353591160220995, "eval_Qnli-dev_cosine_f1_threshold": 0.7576093673706055, "eval_Qnli-dev_cosine_precision": 0.4713114754098361, "eval_Qnli-dev_cosine_recall": 0.9745762711864406, "eval_Qnli-dev_dot_accuracy": 0.576171875, "eval_Qnli-dev_dot_accuracy_threshold": 628.8133544921875, "eval_Qnli-dev_dot_ap": 0.526966347597625, "eval_Qnli-dev_dot_f1": 0.6307277628032345, "eval_Qnli-dev_dot_f1_threshold": 381.04156494140625, "eval_Qnli-dev_dot_precision": 0.4624505928853755, "eval_Qnli-dev_dot_recall": 0.9915254237288136, "eval_Qnli-dev_euclidean_accuracy": 0.623046875, "eval_Qnli-dev_euclidean_accuracy_threshold": 11.349706649780273, "eval_Qnli-dev_euclidean_ap": 0.6145366635531581, "eval_Qnli-dev_euclidean_f1": 0.6321525885558583, "eval_Qnli-dev_euclidean_f1_threshold": 18.34234619140625, "eval_Qnli-dev_euclidean_precision": 0.46586345381526106, "eval_Qnli-dev_euclidean_recall": 0.9830508474576272, "eval_Qnli-dev_manhattan_accuracy": 0.619140625, "eval_Qnli-dev_manhattan_accuracy_threshold": 173.58065795898438, "eval_Qnli-dev_manhattan_ap": 0.6265850112923517, "eval_Qnli-dev_manhattan_f1": 0.6324324324324324, "eval_Qnli-dev_manhattan_f1_threshold": 324.4728698730469, "eval_Qnli-dev_manhattan_precision": 0.4642857142857143, "eval_Qnli-dev_manhattan_recall": 0.9915254237288136, "eval_Qnli-dev_max_accuracy": 0.634765625, "eval_Qnli-dev_max_accuracy_threshold": 628.8133544921875, "eval_Qnli-dev_max_ap": 0.6265850112923517, "eval_Qnli-dev_max_f1": 0.6353591160220995, "eval_Qnli-dev_max_f1_threshold": 381.04156494140625, "eval_Qnli-dev_max_precision": 0.4713114754098361, "eval_Qnli-dev_max_recall": 0.9915254237288136, "eval_allNLI-dev_cosine_accuracy": 0.685546875, "eval_allNLI-dev_cosine_accuracy_threshold": 0.9723453521728516, "eval_allNLI-dev_cosine_ap": 0.43881901898113645, "eval_allNLI-dev_cosine_f1": 0.5192982456140351, "eval_allNLI-dev_cosine_f1_threshold": 0.8869714736938477, "eval_allNLI-dev_cosine_precision": 0.37279596977329976, "eval_allNLI-dev_cosine_recall": 0.8554913294797688, "eval_allNLI-dev_dot_accuracy": 0.66796875, "eval_allNLI-dev_dot_accuracy_threshold": 798.5744018554688, "eval_allNLI-dev_dot_ap": 0.3613142873320213, "eval_allNLI-dev_dot_f1": 0.5066273932253312, "eval_allNLI-dev_dot_f1_threshold": 449.14764404296875, "eval_allNLI-dev_dot_precision": 0.33992094861660077, "eval_allNLI-dev_dot_recall": 0.9942196531791907, "eval_allNLI-dev_euclidean_accuracy": 0.681640625, "eval_allNLI-dev_euclidean_accuracy_threshold": 6.218497276306152, "eval_allNLI-dev_euclidean_ap": 0.43303972275330854, "eval_allNLI-dev_euclidean_f1": 0.5192307692307693, "eval_allNLI-dev_euclidean_f1_threshold": 14.741231918334961, "eval_allNLI-dev_euclidean_precision": 0.35920177383592017, "eval_allNLI-dev_euclidean_recall": 0.9364161849710982, "eval_allNLI-dev_manhattan_accuracy": 0.6796875, "eval_allNLI-dev_manhattan_accuracy_threshold": 106.66935729980469, "eval_allNLI-dev_manhattan_ap": 0.4411371390279413, "eval_allNLI-dev_manhattan_f1": 0.514820592823713, "eval_allNLI-dev_manhattan_f1_threshold": 245.81549072265625, "eval_allNLI-dev_manhattan_precision": 0.3525641025641026, "eval_allNLI-dev_manhattan_recall": 0.953757225433526, "eval_allNLI-dev_max_accuracy": 0.685546875, "eval_allNLI-dev_max_accuracy_threshold": 798.5744018554688, "eval_allNLI-dev_max_ap": 0.4411371390279413, "eval_allNLI-dev_max_f1": 0.5192982456140351, "eval_allNLI-dev_max_f1_threshold": 449.14764404296875, "eval_allNLI-dev_max_precision": 0.37279596977329976, "eval_allNLI-dev_max_recall": 0.9942196531791907, "eval_sequential_score": 0.6265850112923517, "eval_sts-test_pearson_cosine": 0.21333982770993898, "eval_sts-test_pearson_dot": 0.21730500216309007, "eval_sts-test_pearson_euclidean": 0.23256570859273834, "eval_sts-test_pearson_manhattan": 0.2446910502330658, "eval_sts-test_pearson_max": 0.2446910502330658, "eval_sts-test_spearman_cosine": 0.29322398745215855, "eval_sts-test_spearman_dot": 0.22164131759237862, "eval_sts-test_spearman_euclidean": 0.2638306587716433, "eval_sts-test_spearman_manhattan": 0.27389212962153303, "eval_sts-test_spearman_max": 0.29322398745215855, "eval_vitaminc-pairs_loss": 3.76989483833313, "eval_vitaminc-pairs_runtime": 5.4587, "eval_vitaminc-pairs_samples_per_second": 23.449, "eval_vitaminc-pairs_steps_per_second": 0.366, "step": 180 }, { "epoch": 0.13964313421256788, "eval_negation-triplets_loss": 4.8690900802612305, "eval_negation-triplets_runtime": 1.1072, "eval_negation-triplets_samples_per_second": 115.606, "eval_negation-triplets_steps_per_second": 1.806, "step": 180 }, { "epoch": 0.13964313421256788, "eval_scitail-pairs-pos_loss": 1.604143738746643, "eval_scitail-pairs-pos_runtime": 1.2067, "eval_scitail-pairs-pos_samples_per_second": 106.076, "eval_scitail-pairs-pos_steps_per_second": 1.657, "step": 180 }, { "epoch": 0.13964313421256788, "eval_scitail-pairs-qa_loss": 1.654570460319519, "eval_scitail-pairs-qa_runtime": 0.9037, "eval_scitail-pairs-qa_samples_per_second": 141.636, "eval_scitail-pairs-qa_steps_per_second": 2.213, "step": 180 }, { "epoch": 0.13964313421256788, "eval_xsum-pairs_loss": 4.028111457824707, "eval_xsum-pairs_runtime": 4.3063, "eval_xsum-pairs_samples_per_second": 29.724, "eval_xsum-pairs_steps_per_second": 0.464, "step": 180 }, { "epoch": 0.13964313421256788, "eval_sciq_pairs_loss": 0.7544235587120056, "eval_sciq_pairs_runtime": 5.4956, "eval_sciq_pairs_samples_per_second": 23.291, "eval_sciq_pairs_steps_per_second": 0.364, "step": 180 }, { "epoch": 0.13964313421256788, "eval_qasc_pairs_loss": 4.187389850616455, "eval_qasc_pairs_runtime": 0.9748, "eval_qasc_pairs_samples_per_second": 131.305, "eval_qasc_pairs_steps_per_second": 2.052, "step": 180 }, { "epoch": 0.13964313421256788, "eval_openbookqa_pairs_loss": 5.340748310089111, "eval_openbookqa_pairs_runtime": 0.8705, "eval_openbookqa_pairs_samples_per_second": 147.041, "eval_openbookqa_pairs_steps_per_second": 2.298, "step": 180 }, { "epoch": 0.13964313421256788, "eval_msmarco_pairs_loss": 5.431215286254883, "eval_msmarco_pairs_runtime": 1.9525, "eval_msmarco_pairs_samples_per_second": 65.558, "eval_msmarco_pairs_steps_per_second": 1.024, "step": 180 }, { "epoch": 0.13964313421256788, "eval_nq_pairs_loss": 4.6815714836120605, "eval_nq_pairs_runtime": 4.4065, "eval_nq_pairs_samples_per_second": 29.048, "eval_nq_pairs_steps_per_second": 0.454, "step": 180 }, { "epoch": 0.13964313421256788, "eval_trivia_pairs_loss": 5.035283088684082, "eval_trivia_pairs_runtime": 4.5305, "eval_trivia_pairs_samples_per_second": 28.253, "eval_trivia_pairs_steps_per_second": 0.441, "step": 180 }, { "epoch": 0.13964313421256788, "eval_gooaq_pairs_loss": 4.48758602142334, "eval_gooaq_pairs_runtime": 1.2618, "eval_gooaq_pairs_samples_per_second": 101.442, "eval_gooaq_pairs_steps_per_second": 1.585, "step": 180 }, { "epoch": 0.13964313421256788, "eval_paws-pos_loss": 0.5738699436187744, "eval_paws-pos_runtime": 1.0233, "eval_paws-pos_samples_per_second": 125.089, "eval_paws-pos_steps_per_second": 1.955, "step": 180 }, { "epoch": 0.13964313421256788, "eval_global_dataset_loss": 3.1652355194091797, "eval_global_dataset_runtime": 21.133, "eval_global_dataset_samples_per_second": 19.685, "eval_global_dataset_steps_per_second": 0.331, "step": 180 }, { "epoch": 0.1404189294026377, "grad_norm": 15.412625312805176, "learning_rate": 7.3207990599294945e-06, "loss": 4.0213, "step": 181 }, { "epoch": 0.14119472459270752, "grad_norm": 17.41399383544922, "learning_rate": 7.361927144535839e-06, "loss": 4.6874, "step": 182 }, { "epoch": 0.14197051978277736, "grad_norm": 13.163263320922852, "learning_rate": 7.403055229142185e-06, "loss": 3.1823, "step": 183 }, { "epoch": 0.14274631497284718, "grad_norm": 14.50450325012207, "learning_rate": 7.444183313748531e-06, "loss": 3.1686, "step": 184 }, { "epoch": 0.143522110162917, "grad_norm": 20.749032974243164, "learning_rate": 7.485311398354876e-06, "loss": 2.8957, "step": 185 }, { "epoch": 0.1442979053529868, "grad_norm": 13.969277381896973, "learning_rate": 7.526439482961222e-06, "loss": 4.5781, "step": 186 }, { "epoch": 0.14507370054305663, "grad_norm": 15.575841903686523, "learning_rate": 7.567567567567567e-06, "loss": 3.7329, "step": 187 }, { "epoch": 0.14584949573312644, "grad_norm": 14.548502922058105, "learning_rate": 7.608695652173912e-06, "loss": 3.4419, "step": 188 }, { "epoch": 0.1466252909231963, "grad_norm": 24.58466339111328, "learning_rate": 7.649823736780258e-06, "loss": 5.6953, "step": 189 }, { "epoch": 0.1474010861132661, "grad_norm": 14.317368507385254, "learning_rate": 7.690951821386603e-06, "loss": 3.1869, "step": 190 }, { "epoch": 0.14817688130333592, "grad_norm": 16.537460327148438, "learning_rate": 7.732079905992948e-06, "loss": 3.6055, "step": 191 }, { "epoch": 0.14895267649340574, "grad_norm": 20.389019012451172, "learning_rate": 7.773207990599295e-06, "loss": 4.6231, "step": 192 }, { "epoch": 0.14972847168347556, "grad_norm": 17.62578773498535, "learning_rate": 7.814336075205639e-06, "loss": 4.2417, "step": 193 }, { "epoch": 0.15050426687354537, "grad_norm": 28.10770034790039, "learning_rate": 7.855464159811986e-06, "loss": 5.2779, "step": 194 }, { "epoch": 0.15128006206361522, "grad_norm": 19.735567092895508, "learning_rate": 7.896592244418331e-06, "loss": 4.3213, "step": 195 }, { "epoch": 0.15205585725368503, "grad_norm": 16.370105743408203, "learning_rate": 7.937720329024676e-06, "loss": 3.1158, "step": 196 }, { "epoch": 0.15283165244375485, "grad_norm": 21.58885383605957, "learning_rate": 7.978848413631022e-06, "loss": 2.27, "step": 197 }, { "epoch": 0.15360744763382467, "grad_norm": 14.355427742004395, "learning_rate": 8.019976498237367e-06, "loss": 3.5041, "step": 198 }, { "epoch": 0.15438324282389448, "grad_norm": 13.562695503234863, "learning_rate": 8.061104582843712e-06, "loss": 2.6007, "step": 199 }, { "epoch": 0.1551590380139643, "grad_norm": 14.875197410583496, "learning_rate": 8.102232667450058e-06, "loss": 2.4875, "step": 200 }, { "epoch": 0.15593483320403415, "grad_norm": 29.655797958374023, "learning_rate": 8.143360752056405e-06, "loss": 5.3046, "step": 201 }, { "epoch": 0.15671062839410396, "grad_norm": 21.172327041625977, "learning_rate": 8.184488836662748e-06, "loss": 3.0582, "step": 202 }, { "epoch": 0.15748642358417378, "grad_norm": 24.037527084350586, "learning_rate": 8.225616921269095e-06, "loss": 4.9347, "step": 203 }, { "epoch": 0.1582622187742436, "grad_norm": 18.551284790039062, "learning_rate": 8.26674500587544e-06, "loss": 2.855, "step": 204 }, { "epoch": 0.1590380139643134, "grad_norm": 18.76624870300293, "learning_rate": 8.307873090481786e-06, "loss": 1.7434, "step": 205 }, { "epoch": 0.15981380915438323, "grad_norm": 12.969701766967773, "learning_rate": 8.349001175088131e-06, "loss": 3.4045, "step": 206 }, { "epoch": 0.16058960434445307, "grad_norm": 15.40457820892334, "learning_rate": 8.390129259694476e-06, "loss": 3.4427, "step": 207 }, { "epoch": 0.1613653995345229, "grad_norm": 16.32777976989746, "learning_rate": 8.431257344300822e-06, "loss": 3.3483, "step": 208 }, { "epoch": 0.1621411947245927, "grad_norm": 17.573177337646484, "learning_rate": 8.472385428907167e-06, "loss": 1.5883, "step": 209 }, { "epoch": 0.16291698991466252, "grad_norm": 28.443233489990234, "learning_rate": 8.513513513513514e-06, "loss": 5.3066, "step": 210 }, { "epoch": 0.16369278510473234, "grad_norm": 13.738469123840332, "learning_rate": 8.554641598119857e-06, "loss": 0.6047, "step": 211 }, { "epoch": 0.16446858029480219, "grad_norm": 15.555298805236816, "learning_rate": 8.595769682726204e-06, "loss": 0.8018, "step": 212 }, { "epoch": 0.165244375484872, "grad_norm": 19.836238861083984, "learning_rate": 8.636897767332548e-06, "loss": 2.8775, "step": 213 }, { "epoch": 0.16602017067494182, "grad_norm": 10.346473693847656, "learning_rate": 8.678025851938895e-06, "loss": 0.5198, "step": 214 }, { "epoch": 0.16679596586501164, "grad_norm": 15.540273666381836, "learning_rate": 8.71915393654524e-06, "loss": 3.1591, "step": 215 }, { "epoch": 0.16757176105508145, "grad_norm": 23.148704528808594, "learning_rate": 8.760282021151586e-06, "loss": 2.7381, "step": 216 }, { "epoch": 0.16834755624515127, "grad_norm": 36.07377624511719, "learning_rate": 8.801410105757933e-06, "loss": 5.5722, "step": 217 }, { "epoch": 0.16912335143522111, "grad_norm": 31.731229782104492, "learning_rate": 8.842538190364276e-06, "loss": 3.998, "step": 218 }, { "epoch": 0.16989914662529093, "grad_norm": 19.629518508911133, "learning_rate": 8.883666274970621e-06, "loss": 2.2858, "step": 219 }, { "epoch": 0.17067494181536075, "grad_norm": 17.323598861694336, "learning_rate": 8.924794359576967e-06, "loss": 1.556, "step": 220 }, { "epoch": 0.17145073700543056, "grad_norm": 20.71698760986328, "learning_rate": 8.965922444183314e-06, "loss": 2.5352, "step": 221 }, { "epoch": 0.17222653219550038, "grad_norm": 16.524059295654297, "learning_rate": 9.007050528789659e-06, "loss": 3.1682, "step": 222 }, { "epoch": 0.1730023273855702, "grad_norm": 18.413827896118164, "learning_rate": 9.048178613396003e-06, "loss": 2.9073, "step": 223 }, { "epoch": 0.17377812257564004, "grad_norm": 18.246061325073242, "learning_rate": 9.08930669800235e-06, "loss": 2.547, "step": 224 }, { "epoch": 0.17455391776570986, "grad_norm": 20.940176010131836, "learning_rate": 9.130434782608695e-06, "loss": 4.1815, "step": 225 }, { "epoch": 0.17532971295577968, "grad_norm": 22.81971549987793, "learning_rate": 9.17156286721504e-06, "loss": 3.7504, "step": 226 }, { "epoch": 0.1761055081458495, "grad_norm": 29.75227165222168, "learning_rate": 9.212690951821385e-06, "loss": 5.033, "step": 227 }, { "epoch": 0.1768813033359193, "grad_norm": 27.30958366394043, "learning_rate": 9.253819036427732e-06, "loss": 5.2809, "step": 228 }, { "epoch": 0.17765709852598913, "grad_norm": 14.53124713897705, "learning_rate": 9.294947121034076e-06, "loss": 2.5598, "step": 229 }, { "epoch": 0.17843289371605897, "grad_norm": 10.330536842346191, "learning_rate": 9.336075205640421e-06, "loss": 0.4476, "step": 230 }, { "epoch": 0.1792086889061288, "grad_norm": 16.898452758789062, "learning_rate": 9.377203290246768e-06, "loss": 3.4592, "step": 231 }, { "epoch": 0.1799844840961986, "grad_norm": 13.675553321838379, "learning_rate": 9.418331374853114e-06, "loss": 2.9202, "step": 232 }, { "epoch": 0.18076027928626842, "grad_norm": 13.829231262207031, "learning_rate": 9.459459459459459e-06, "loss": 1.9092, "step": 233 }, { "epoch": 0.18153607447633824, "grad_norm": 19.693933486938477, "learning_rate": 9.500587544065804e-06, "loss": 1.9204, "step": 234 }, { "epoch": 0.18231186966640806, "grad_norm": 14.612395286560059, "learning_rate": 9.54171562867215e-06, "loss": 2.083, "step": 235 }, { "epoch": 0.1830876648564779, "grad_norm": 12.987626075744629, "learning_rate": 9.582843713278495e-06, "loss": 3.0433, "step": 236 }, { "epoch": 0.18386346004654772, "grad_norm": 13.259800910949707, "learning_rate": 9.62397179788484e-06, "loss": 1.5429, "step": 237 }, { "epoch": 0.18463925523661753, "grad_norm": 8.798643112182617, "learning_rate": 9.665099882491187e-06, "loss": 0.3347, "step": 238 }, { "epoch": 0.18541505042668735, "grad_norm": 13.148309707641602, "learning_rate": 9.706227967097532e-06, "loss": 1.8698, "step": 239 }, { "epoch": 0.18619084561675717, "grad_norm": 7.301493167877197, "learning_rate": 9.747356051703878e-06, "loss": 0.3031, "step": 240 }, { "epoch": 0.18619084561675717, "eval_Qnli-dev_cosine_accuracy": 0.6328125, "eval_Qnli-dev_cosine_accuracy_threshold": 0.8666848540306091, "eval_Qnli-dev_cosine_ap": 0.6314460649736229, "eval_Qnli-dev_cosine_f1": 0.6366366366366366, "eval_Qnli-dev_cosine_f1_threshold": 0.7626283168792725, "eval_Qnli-dev_cosine_precision": 0.4930232558139535, "eval_Qnli-dev_cosine_recall": 0.8983050847457628, "eval_Qnli-dev_dot_accuracy": 0.6015625, "eval_Qnli-dev_dot_accuracy_threshold": 507.15960693359375, "eval_Qnli-dev_dot_ap": 0.5696093250757446, "eval_Qnli-dev_dot_f1": 0.638680659670165, "eval_Qnli-dev_dot_f1_threshold": 389.59454345703125, "eval_Qnli-dev_dot_precision": 0.494199535962877, "eval_Qnli-dev_dot_recall": 0.902542372881356, "eval_Qnli-dev_euclidean_accuracy": 0.626953125, "eval_Qnli-dev_euclidean_accuracy_threshold": 11.648075103759766, "eval_Qnli-dev_euclidean_ap": 0.630384295725278, "eval_Qnli-dev_euclidean_f1": 0.6362098138747886, "eval_Qnli-dev_euclidean_f1_threshold": 14.59055233001709, "eval_Qnli-dev_euclidean_precision": 0.5295774647887324, "eval_Qnli-dev_euclidean_recall": 0.7966101694915254, "eval_Qnli-dev_manhattan_accuracy": 0.6328125, "eval_Qnli-dev_manhattan_accuracy_threshold": 210.99937438964844, "eval_Qnli-dev_manhattan_ap": 0.6305567474066167, "eval_Qnli-dev_manhattan_f1": 0.6392961876832844, "eval_Qnli-dev_manhattan_f1_threshold": 295.44036865234375, "eval_Qnli-dev_manhattan_precision": 0.48878923766816146, "eval_Qnli-dev_manhattan_recall": 0.923728813559322, "eval_Qnli-dev_max_accuracy": 0.6328125, "eval_Qnli-dev_max_accuracy_threshold": 507.15960693359375, "eval_Qnli-dev_max_ap": 0.6314460649736229, "eval_Qnli-dev_max_f1": 0.6392961876832844, "eval_Qnli-dev_max_f1_threshold": 389.59454345703125, "eval_Qnli-dev_max_precision": 0.5295774647887324, "eval_Qnli-dev_max_recall": 0.923728813559322, "eval_allNLI-dev_cosine_accuracy": 0.693359375, "eval_allNLI-dev_cosine_accuracy_threshold": 0.9488164782524109, "eval_allNLI-dev_cosine_ap": 0.5017454777328199, "eval_allNLI-dev_cosine_f1": 0.5440613026819924, "eval_allNLI-dev_cosine_f1_threshold": 0.8723046779632568, "eval_allNLI-dev_cosine_precision": 0.4068767908309456, "eval_allNLI-dev_cosine_recall": 0.8208092485549133, "eval_allNLI-dev_dot_accuracy": 0.673828125, "eval_allNLI-dev_dot_accuracy_threshold": 668.8308715820312, "eval_allNLI-dev_dot_ap": 0.3969704201342592, "eval_allNLI-dev_dot_f1": 0.5069124423963134, "eval_allNLI-dev_dot_f1_threshold": 449.8724365234375, "eval_allNLI-dev_dot_precision": 0.34518828451882844, "eval_allNLI-dev_dot_recall": 0.953757225433526, "eval_allNLI-dev_euclidean_accuracy": 0.6875, "eval_allNLI-dev_euclidean_accuracy_threshold": 7.274426460266113, "eval_allNLI-dev_euclidean_ap": 0.5011125896179964, "eval_allNLI-dev_euclidean_f1": 0.5429141716566867, "eval_allNLI-dev_euclidean_f1_threshold": 12.62590217590332, "eval_allNLI-dev_euclidean_precision": 0.4146341463414634, "eval_allNLI-dev_euclidean_recall": 0.7861271676300579, "eval_allNLI-dev_manhattan_accuracy": 0.689453125, "eval_allNLI-dev_manhattan_accuracy_threshold": 127.58004760742188, "eval_allNLI-dev_manhattan_ap": 0.5053834291774996, "eval_allNLI-dev_manhattan_f1": 0.5362035225048923, "eval_allNLI-dev_manhattan_f1_threshold": 221.26995849609375, "eval_allNLI-dev_manhattan_precision": 0.40532544378698226, "eval_allNLI-dev_manhattan_recall": 0.791907514450867, "eval_allNLI-dev_max_accuracy": 0.693359375, "eval_allNLI-dev_max_accuracy_threshold": 668.8308715820312, "eval_allNLI-dev_max_ap": 0.5053834291774996, "eval_allNLI-dev_max_f1": 0.5440613026819924, "eval_allNLI-dev_max_f1_threshold": 449.8724365234375, "eval_allNLI-dev_max_precision": 0.4146341463414634, "eval_allNLI-dev_max_recall": 0.953757225433526, "eval_sequential_score": 0.6314460649736229, "eval_sts-test_pearson_cosine": 0.4297034177530014, "eval_sts-test_pearson_dot": 0.3153020725919898, "eval_sts-test_pearson_euclidean": 0.4381297122634254, "eval_sts-test_pearson_manhattan": 0.4527271663445909, "eval_sts-test_pearson_max": 0.4527271663445909, "eval_sts-test_spearman_cosine": 0.49463792972220844, "eval_sts-test_spearman_dot": 0.31075978504991664, "eval_sts-test_spearman_euclidean": 0.455142390722399, "eval_sts-test_spearman_manhattan": 0.46803254405357775, "eval_sts-test_spearman_max": 0.49463792972220844, "eval_vitaminc-pairs_loss": 3.9216716289520264, "eval_vitaminc-pairs_runtime": 5.46, "eval_vitaminc-pairs_samples_per_second": 23.443, "eval_vitaminc-pairs_steps_per_second": 0.366, "step": 240 }, { "epoch": 0.18619084561675717, "eval_negation-triplets_loss": 3.8406004905700684, "eval_negation-triplets_runtime": 1.1041, "eval_negation-triplets_samples_per_second": 115.932, "eval_negation-triplets_steps_per_second": 1.811, "step": 240 }, { "epoch": 0.18619084561675717, "eval_scitail-pairs-pos_loss": 0.597940981388092, "eval_scitail-pairs-pos_runtime": 1.2771, "eval_scitail-pairs-pos_samples_per_second": 100.224, "eval_scitail-pairs-pos_steps_per_second": 1.566, "step": 240 }, { "epoch": 0.18619084561675717, "eval_scitail-pairs-qa_loss": 0.5431724786758423, "eval_scitail-pairs-qa_runtime": 0.9108, "eval_scitail-pairs-qa_samples_per_second": 140.54, "eval_scitail-pairs-qa_steps_per_second": 2.196, "step": 240 }, { "epoch": 0.18619084561675717, "eval_xsum-pairs_loss": 2.5905959606170654, "eval_xsum-pairs_runtime": 4.3069, "eval_xsum-pairs_samples_per_second": 29.72, "eval_xsum-pairs_steps_per_second": 0.464, "step": 240 }, { "epoch": 0.18619084561675717, "eval_sciq_pairs_loss": 0.44124865531921387, "eval_sciq_pairs_runtime": 5.5223, "eval_sciq_pairs_samples_per_second": 23.179, "eval_sciq_pairs_steps_per_second": 0.362, "step": 240 }, { "epoch": 0.18619084561675717, "eval_qasc_pairs_loss": 2.3103857040405273, "eval_qasc_pairs_runtime": 0.9809, "eval_qasc_pairs_samples_per_second": 130.488, "eval_qasc_pairs_steps_per_second": 2.039, "step": 240 }, { "epoch": 0.18619084561675717, "eval_openbookqa_pairs_loss": 3.97737979888916, "eval_openbookqa_pairs_runtime": 0.8546, "eval_openbookqa_pairs_samples_per_second": 149.783, "eval_openbookqa_pairs_steps_per_second": 2.34, "step": 240 }, { "epoch": 0.18619084561675717, "eval_msmarco_pairs_loss": 3.6929101943969727, "eval_msmarco_pairs_runtime": 1.9549, "eval_msmarco_pairs_samples_per_second": 65.476, "eval_msmarco_pairs_steps_per_second": 1.023, "step": 240 }, { "epoch": 0.18619084561675717, "eval_nq_pairs_loss": 3.8183703422546387, "eval_nq_pairs_runtime": 4.4066, "eval_nq_pairs_samples_per_second": 29.047, "eval_nq_pairs_steps_per_second": 0.454, "step": 240 }, { "epoch": 0.18619084561675717, "eval_trivia_pairs_loss": 3.726828098297119, "eval_trivia_pairs_runtime": 4.5284, "eval_trivia_pairs_samples_per_second": 28.266, "eval_trivia_pairs_steps_per_second": 0.442, "step": 240 }, { "epoch": 0.18619084561675717, "eval_gooaq_pairs_loss": 3.068204402923584, "eval_gooaq_pairs_runtime": 1.2681, "eval_gooaq_pairs_samples_per_second": 100.939, "eval_gooaq_pairs_steps_per_second": 1.577, "step": 240 }, { "epoch": 0.18619084561675717, "eval_paws-pos_loss": 0.11907516419887543, "eval_paws-pos_runtime": 1.0209, "eval_paws-pos_samples_per_second": 125.382, "eval_paws-pos_steps_per_second": 1.959, "step": 240 }, { "epoch": 0.18619084561675717, "eval_global_dataset_loss": 1.8651976585388184, "eval_global_dataset_runtime": 21.15, "eval_global_dataset_samples_per_second": 19.669, "eval_global_dataset_steps_per_second": 0.331, "step": 240 }, { "epoch": 0.186966640806827, "grad_norm": 17.317655563354492, "learning_rate": 9.788484136310221e-06, "loss": 3.2076, "step": 241 }, { "epoch": 0.18774243599689683, "grad_norm": 16.139144897460938, "learning_rate": 9.829612220916568e-06, "loss": 2.684, "step": 242 }, { "epoch": 0.18851823118696664, "grad_norm": 17.887540817260742, "learning_rate": 9.870740305522913e-06, "loss": 2.234, "step": 243 }, { "epoch": 0.18929402637703646, "grad_norm": 15.649737358093262, "learning_rate": 9.911868390129259e-06, "loss": 3.0141, "step": 244 }, { "epoch": 0.19006982156710628, "grad_norm": 6.042151927947998, "learning_rate": 9.952996474735606e-06, "loss": 0.271, "step": 245 }, { "epoch": 0.1908456167571761, "grad_norm": 16.89675521850586, "learning_rate": 9.99412455934195e-06, "loss": 1.878, "step": 246 }, { "epoch": 0.19162141194724594, "grad_norm": 21.4814395904541, "learning_rate": 1.0035252643948295e-05, "loss": 2.4108, "step": 247 }, { "epoch": 0.19239720713731576, "grad_norm": 24.15477752685547, "learning_rate": 1.007638072855464e-05, "loss": 3.6327, "step": 248 }, { "epoch": 0.19317300232738557, "grad_norm": 19.688291549682617, "learning_rate": 1.0117508813160987e-05, "loss": 2.9108, "step": 249 }, { "epoch": 0.1939487975174554, "grad_norm": 30.842660903930664, "learning_rate": 1.0158636897767332e-05, "loss": 4.425, "step": 250 }, { "epoch": 0.1947245927075252, "grad_norm": 17.901451110839844, "learning_rate": 1.0199764982373677e-05, "loss": 2.6128, "step": 251 }, { "epoch": 0.19550038789759502, "grad_norm": 19.920259475708008, "learning_rate": 1.0240893066980021e-05, "loss": 3.1872, "step": 252 }, { "epoch": 0.19627618308766487, "grad_norm": 17.547534942626953, "learning_rate": 1.0282021151586368e-05, "loss": 1.9839, "step": 253 }, { "epoch": 0.19705197827773469, "grad_norm": 30.619075775146484, "learning_rate": 1.0323149236192713e-05, "loss": 5.0317, "step": 254 }, { "epoch": 0.1978277734678045, "grad_norm": 16.690624237060547, "learning_rate": 1.0364277320799059e-05, "loss": 1.8456, "step": 255 }, { "epoch": 0.19860356865787432, "grad_norm": 17.084972381591797, "learning_rate": 1.0405405405405406e-05, "loss": 2.7799, "step": 256 }, { "epoch": 0.19937936384794414, "grad_norm": 16.932178497314453, "learning_rate": 1.0446533490011751e-05, "loss": 2.4114, "step": 257 }, { "epoch": 0.20015515903801395, "grad_norm": 14.226729393005371, "learning_rate": 1.0487661574618094e-05, "loss": 1.2619, "step": 258 } ], "logging_steps": 1, "max_steps": 2578, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 258, "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": 96, "trial_name": null, "trial_params": null }