diff --git "a/checkpoint-1806/trainer_state.json" "b/checkpoint-1806/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-1806/trainer_state.json" @@ -0,0 +1,18465 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.4010861132660977, + "eval_steps": 60, + "global_step": 1806, + "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 + }, + { + "epoch": 0.2009309542280838, + "grad_norm": 14.107755661010742, + "learning_rate": 1.052878965922444e-05, + "loss": 1.7154, + "step": 259 + }, + { + "epoch": 0.2017067494181536, + "grad_norm": 17.15689468383789, + "learning_rate": 1.0569917743830787e-05, + "loss": 2.9187, + "step": 260 + }, + { + "epoch": 0.20248254460822343, + "grad_norm": 13.765548706054688, + "learning_rate": 1.0611045828437132e-05, + "loss": 1.8434, + "step": 261 + }, + { + "epoch": 0.20325833979829325, + "grad_norm": 14.669478416442871, + "learning_rate": 1.0652173913043477e-05, + "loss": 1.7474, + "step": 262 + }, + { + "epoch": 0.20403413498836306, + "grad_norm": 30.17442512512207, + "learning_rate": 1.0693301997649823e-05, + "loss": 4.433, + "step": 263 + }, + { + "epoch": 0.20480993017843288, + "grad_norm": 16.97207260131836, + "learning_rate": 1.0734430082256168e-05, + "loss": 2.169, + "step": 264 + }, + { + "epoch": 0.20558572536850273, + "grad_norm": 25.517986297607422, + "learning_rate": 1.0775558166862513e-05, + "loss": 4.0286, + "step": 265 + }, + { + "epoch": 0.20636152055857254, + "grad_norm": 22.2929630279541, + "learning_rate": 1.0816686251468858e-05, + "loss": 3.6756, + "step": 266 + }, + { + "epoch": 0.20713731574864236, + "grad_norm": 15.684208869934082, + "learning_rate": 1.0857814336075205e-05, + "loss": 1.853, + "step": 267 + }, + { + "epoch": 0.20791311093871218, + "grad_norm": 19.631729125976562, + "learning_rate": 1.089894242068155e-05, + "loss": 2.9833, + "step": 268 + }, + { + "epoch": 0.208688906128782, + "grad_norm": 4.716369152069092, + "learning_rate": 1.0940070505287894e-05, + "loss": 0.3286, + "step": 269 + }, + { + "epoch": 0.20946470131885184, + "grad_norm": 16.292285919189453, + "learning_rate": 1.0981198589894241e-05, + "loss": 2.9449, + "step": 270 + }, + { + "epoch": 0.21024049650892165, + "grad_norm": 16.14474105834961, + "learning_rate": 1.1022326674500587e-05, + "loss": 2.3685, + "step": 271 + }, + { + "epoch": 0.21101629169899147, + "grad_norm": 14.462653160095215, + "learning_rate": 1.1063454759106932e-05, + "loss": 1.8971, + "step": 272 + }, + { + "epoch": 0.2117920868890613, + "grad_norm": 14.50413990020752, + "learning_rate": 1.1104582843713279e-05, + "loss": 2.2603, + "step": 273 + }, + { + "epoch": 0.2125678820791311, + "grad_norm": 24.562816619873047, + "learning_rate": 1.1145710928319624e-05, + "loss": 3.4866, + "step": 274 + }, + { + "epoch": 0.21334367726920092, + "grad_norm": 21.698535919189453, + "learning_rate": 1.1186839012925968e-05, + "loss": 3.3005, + "step": 275 + }, + { + "epoch": 0.21411947245927077, + "grad_norm": 12.849061965942383, + "learning_rate": 1.1227967097532313e-05, + "loss": 1.3361, + "step": 276 + }, + { + "epoch": 0.21489526764934058, + "grad_norm": 16.0836238861084, + "learning_rate": 1.126909518213866e-05, + "loss": 1.8639, + "step": 277 + }, + { + "epoch": 0.2156710628394104, + "grad_norm": 22.1470947265625, + "learning_rate": 1.1310223266745005e-05, + "loss": 3.9011, + "step": 278 + }, + { + "epoch": 0.21644685802948022, + "grad_norm": 26.26567268371582, + "learning_rate": 1.135135135135135e-05, + "loss": 3.8815, + "step": 279 + }, + { + "epoch": 0.21722265321955003, + "grad_norm": 14.440673828125, + "learning_rate": 1.1392479435957698e-05, + "loss": 2.766, + "step": 280 + }, + { + "epoch": 0.21799844840961985, + "grad_norm": 14.126577377319336, + "learning_rate": 1.1433607520564041e-05, + "loss": 2.545, + "step": 281 + }, + { + "epoch": 0.2187742435996897, + "grad_norm": 16.04927635192871, + "learning_rate": 1.1474735605170386e-05, + "loss": 1.2585, + "step": 282 + }, + { + "epoch": 0.2195500387897595, + "grad_norm": 14.855483055114746, + "learning_rate": 1.1515863689776732e-05, + "loss": 2.3415, + "step": 283 + }, + { + "epoch": 0.22032583397982933, + "grad_norm": 4.940772533416748, + "learning_rate": 1.1556991774383079e-05, + "loss": 0.1715, + "step": 284 + }, + { + "epoch": 0.22110162916989914, + "grad_norm": 16.687578201293945, + "learning_rate": 1.1598119858989424e-05, + "loss": 1.8196, + "step": 285 + }, + { + "epoch": 0.22187742435996896, + "grad_norm": 13.156393051147461, + "learning_rate": 1.1639247943595768e-05, + "loss": 1.3654, + "step": 286 + }, + { + "epoch": 0.22265321955003878, + "grad_norm": 21.225854873657227, + "learning_rate": 1.1680376028202113e-05, + "loss": 3.4925, + "step": 287 + }, + { + "epoch": 0.22342901474010862, + "grad_norm": 15.54011058807373, + "learning_rate": 1.172150411280846e-05, + "loss": 1.4732, + "step": 288 + }, + { + "epoch": 0.22420480993017844, + "grad_norm": 18.619935989379883, + "learning_rate": 1.1762632197414805e-05, + "loss": 2.6534, + "step": 289 + }, + { + "epoch": 0.22498060512024826, + "grad_norm": 15.872313499450684, + "learning_rate": 1.180376028202115e-05, + "loss": 1.3829, + "step": 290 + }, + { + "epoch": 0.22575640031031807, + "grad_norm": 14.481184005737305, + "learning_rate": 1.1844888366627497e-05, + "loss": 1.5105, + "step": 291 + }, + { + "epoch": 0.2265321955003879, + "grad_norm": 12.84127426147461, + "learning_rate": 1.1886016451233841e-05, + "loss": 1.2445, + "step": 292 + }, + { + "epoch": 0.2273079906904577, + "grad_norm": 14.835034370422363, + "learning_rate": 1.1927144535840186e-05, + "loss": 1.5034, + "step": 293 + }, + { + "epoch": 0.22808378588052755, + "grad_norm": 12.850936889648438, + "learning_rate": 1.1968272620446532e-05, + "loss": 1.2241, + "step": 294 + }, + { + "epoch": 0.22885958107059737, + "grad_norm": 21.77153778076172, + "learning_rate": 1.2009400705052879e-05, + "loss": 3.0699, + "step": 295 + }, + { + "epoch": 0.22963537626066718, + "grad_norm": 14.811423301696777, + "learning_rate": 1.2050528789659224e-05, + "loss": 1.534, + "step": 296 + }, + { + "epoch": 0.230411171450737, + "grad_norm": 25.660585403442383, + "learning_rate": 1.209165687426557e-05, + "loss": 3.7891, + "step": 297 + }, + { + "epoch": 0.23118696664080682, + "grad_norm": 13.855345726013184, + "learning_rate": 1.2132784958871913e-05, + "loss": 1.3415, + "step": 298 + }, + { + "epoch": 0.23196276183087663, + "grad_norm": 14.3121976852417, + "learning_rate": 1.217391304347826e-05, + "loss": 2.1582, + "step": 299 + }, + { + "epoch": 0.23273855702094648, + "grad_norm": 19.85712242126465, + "learning_rate": 1.2215041128084605e-05, + "loss": 2.9695, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_Qnli-dev_cosine_accuracy": 0.650390625, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8682992458343506, + "eval_Qnli-dev_cosine_ap": 0.6584959435780183, + "eval_Qnli-dev_cosine_f1": 0.6623586429725363, + "eval_Qnli-dev_cosine_f1_threshold": 0.7860881686210632, + "eval_Qnli-dev_cosine_precision": 0.5352480417754569, + "eval_Qnli-dev_cosine_recall": 0.8686440677966102, + "eval_Qnli-dev_dot_accuracy": 0.625, + "eval_Qnli-dev_dot_accuracy_threshold": 544.1656494140625, + "eval_Qnli-dev_dot_ap": 0.607359199645401, + "eval_Qnli-dev_dot_f1": 0.638700947225981, + "eval_Qnli-dev_dot_f1_threshold": 331.2305908203125, + "eval_Qnli-dev_dot_precision": 0.4691848906560636, + "eval_Qnli-dev_dot_recall": 1.0, + "eval_Qnli-dev_euclidean_accuracy": 0.650390625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.19542407989502, + "eval_Qnli-dev_euclidean_ap": 0.6566972613584862, + "eval_Qnli-dev_euclidean_f1": 0.6581059390048154, + "eval_Qnli-dev_euclidean_f1_threshold": 15.924192428588867, + "eval_Qnli-dev_euclidean_precision": 0.5297157622739018, + "eval_Qnli-dev_euclidean_recall": 0.8686440677966102, + "eval_Qnli-dev_manhattan_accuracy": 0.640625, + "eval_Qnli-dev_manhattan_accuracy_threshold": 220.612060546875, + "eval_Qnli-dev_manhattan_ap": 0.6535369258879113, + "eval_Qnli-dev_manhattan_f1": 0.6570048309178743, + "eval_Qnli-dev_manhattan_f1_threshold": 281.0206298828125, + "eval_Qnli-dev_manhattan_precision": 0.5298701298701298, + "eval_Qnli-dev_manhattan_recall": 0.864406779661017, + "eval_Qnli-dev_max_accuracy": 0.650390625, + "eval_Qnli-dev_max_accuracy_threshold": 544.1656494140625, + "eval_Qnli-dev_max_ap": 0.6584959435780183, + "eval_Qnli-dev_max_f1": 0.6623586429725363, + "eval_Qnli-dev_max_f1_threshold": 331.2305908203125, + "eval_Qnli-dev_max_precision": 0.5352480417754569, + "eval_Qnli-dev_max_recall": 1.0, + "eval_allNLI-dev_cosine_accuracy": 0.69921875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.90627121925354, + "eval_allNLI-dev_cosine_ap": 0.5474720738526634, + "eval_allNLI-dev_cosine_f1": 0.5641025641025642, + "eval_allNLI-dev_cosine_f1_threshold": 0.8720983266830444, + "eval_allNLI-dev_cosine_precision": 0.47265625, + "eval_allNLI-dev_cosine_recall": 0.6994219653179191, + "eval_allNLI-dev_dot_accuracy": 0.677734375, + "eval_allNLI-dev_dot_accuracy_threshold": 598.082275390625, + "eval_allNLI-dev_dot_ap": 0.4162474644502172, + "eval_allNLI-dev_dot_f1": 0.5182926829268293, + "eval_allNLI-dev_dot_f1_threshold": 402.790771484375, + "eval_allNLI-dev_dot_precision": 0.35196687370600416, + "eval_allNLI-dev_dot_recall": 0.9826589595375722, + "eval_allNLI-dev_euclidean_accuracy": 0.6953125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 8.477822303771973, + "eval_allNLI-dev_euclidean_ap": 0.5453999691725084, + "eval_allNLI-dev_euclidean_f1": 0.5651105651105651, + "eval_allNLI-dev_euclidean_f1_threshold": 11.948629379272461, + "eval_allNLI-dev_euclidean_precision": 0.49145299145299143, + "eval_allNLI-dev_euclidean_recall": 0.6647398843930635, + "eval_allNLI-dev_manhattan_accuracy": 0.6953125, + "eval_allNLI-dev_manhattan_accuracy_threshold": 179.72518920898438, + "eval_allNLI-dev_manhattan_ap": 0.5384026532538657, + "eval_allNLI-dev_manhattan_f1": 0.5780885780885782, + "eval_allNLI-dev_manhattan_f1_threshold": 221.75010681152344, + "eval_allNLI-dev_manhattan_precision": 0.484375, + "eval_allNLI-dev_manhattan_recall": 0.7167630057803468, + "eval_allNLI-dev_max_accuracy": 0.69921875, + "eval_allNLI-dev_max_accuracy_threshold": 598.082275390625, + "eval_allNLI-dev_max_ap": 0.5474720738526634, + "eval_allNLI-dev_max_f1": 0.5780885780885782, + "eval_allNLI-dev_max_f1_threshold": 402.790771484375, + "eval_allNLI-dev_max_precision": 0.49145299145299143, + "eval_allNLI-dev_max_recall": 0.9826589595375722, + "eval_sequential_score": 0.6584959435780183, + "eval_sts-test_pearson_cosine": 0.6824427260520828, + "eval_sts-test_pearson_dot": 0.542784014150645, + "eval_sts-test_pearson_euclidean": 0.7040347305945025, + "eval_sts-test_pearson_manhattan": 0.7042189189752545, + "eval_sts-test_pearson_max": 0.7042189189752545, + "eval_sts-test_spearman_cosine": 0.7332739863405849, + "eval_sts-test_spearman_dot": 0.5407445725673692, + "eval_sts-test_spearman_euclidean": 0.711579111833171, + "eval_sts-test_spearman_manhattan": 0.7123917947400389, + "eval_sts-test_spearman_max": 0.7332739863405849, + "eval_vitaminc-pairs_loss": 4.0183563232421875, + "eval_vitaminc-pairs_runtime": 5.4836, + "eval_vitaminc-pairs_samples_per_second": 23.342, + "eval_vitaminc-pairs_steps_per_second": 0.365, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_negation-triplets_loss": 2.8057239055633545, + "eval_negation-triplets_runtime": 1.1052, + "eval_negation-triplets_samples_per_second": 115.813, + "eval_negation-triplets_steps_per_second": 1.81, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_scitail-pairs-pos_loss": 0.30041298270225525, + "eval_scitail-pairs-pos_runtime": 1.2144, + "eval_scitail-pairs-pos_samples_per_second": 105.399, + "eval_scitail-pairs-pos_steps_per_second": 1.647, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_scitail-pairs-qa_loss": 0.23774191737174988, + "eval_scitail-pairs-qa_runtime": 0.9037, + "eval_scitail-pairs-qa_samples_per_second": 141.645, + "eval_scitail-pairs-qa_steps_per_second": 2.213, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_xsum-pairs_loss": 1.5138336420059204, + "eval_xsum-pairs_runtime": 4.3159, + "eval_xsum-pairs_samples_per_second": 29.658, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_sciq_pairs_loss": 0.25239843130111694, + "eval_sciq_pairs_runtime": 5.5233, + "eval_sciq_pairs_samples_per_second": 23.174, + "eval_sciq_pairs_steps_per_second": 0.362, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_qasc_pairs_loss": 1.7758697271347046, + "eval_qasc_pairs_runtime": 0.9777, + "eval_qasc_pairs_samples_per_second": 130.922, + "eval_qasc_pairs_steps_per_second": 2.046, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_openbookqa_pairs_loss": 3.0091030597686768, + "eval_openbookqa_pairs_runtime": 0.8558, + "eval_openbookqa_pairs_samples_per_second": 149.575, + "eval_openbookqa_pairs_steps_per_second": 2.337, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_msmarco_pairs_loss": 2.336099863052368, + "eval_msmarco_pairs_runtime": 1.9513, + "eval_msmarco_pairs_samples_per_second": 65.599, + "eval_msmarco_pairs_steps_per_second": 1.025, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_nq_pairs_loss": 2.841768264770508, + "eval_nq_pairs_runtime": 4.3953, + "eval_nq_pairs_samples_per_second": 29.122, + "eval_nq_pairs_steps_per_second": 0.455, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_trivia_pairs_loss": 2.525879144668579, + "eval_trivia_pairs_runtime": 4.5194, + "eval_trivia_pairs_samples_per_second": 28.322, + "eval_trivia_pairs_steps_per_second": 0.443, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_gooaq_pairs_loss": 1.8355695009231567, + "eval_gooaq_pairs_runtime": 1.2686, + "eval_gooaq_pairs_samples_per_second": 100.9, + "eval_gooaq_pairs_steps_per_second": 1.577, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_paws-pos_loss": 0.09221775829792023, + "eval_paws-pos_runtime": 1.0194, + "eval_paws-pos_samples_per_second": 125.569, + "eval_paws-pos_steps_per_second": 1.962, + "step": 300 + }, + { + "epoch": 0.23273855702094648, + "eval_global_dataset_loss": 1.3259693384170532, + "eval_global_dataset_runtime": 21.1299, + "eval_global_dataset_samples_per_second": 19.688, + "eval_global_dataset_steps_per_second": 0.331, + "step": 300 + }, + { + "epoch": 0.2335143522110163, + "grad_norm": 25.528133392333984, + "learning_rate": 1.225616921269095e-05, + "loss": 3.4647, + "step": 301 + }, + { + "epoch": 0.2342901474010861, + "grad_norm": 14.908467292785645, + "learning_rate": 1.2297297297297297e-05, + "loss": 1.4592, + "step": 302 + }, + { + "epoch": 0.23506594259115593, + "grad_norm": 5.650328159332275, + "learning_rate": 1.2338425381903641e-05, + "loss": 0.241, + "step": 303 + }, + { + "epoch": 0.23584173778122575, + "grad_norm": 23.929319381713867, + "learning_rate": 1.2379553466509986e-05, + "loss": 3.3644, + "step": 304 + }, + { + "epoch": 0.2366175329712956, + "grad_norm": 12.778107643127441, + "learning_rate": 1.2420681551116333e-05, + "loss": 1.2295, + "step": 305 + }, + { + "epoch": 0.2373933281613654, + "grad_norm": 5.3786702156066895, + "learning_rate": 1.2461809635722679e-05, + "loss": 0.3028, + "step": 306 + }, + { + "epoch": 0.23816912335143522, + "grad_norm": 13.764509201049805, + "learning_rate": 1.2502937720329024e-05, + "loss": 2.1296, + "step": 307 + }, + { + "epoch": 0.23894491854150504, + "grad_norm": 12.894301414489746, + "learning_rate": 1.2544065804935369e-05, + "loss": 0.9409, + "step": 308 + }, + { + "epoch": 0.23972071373157486, + "grad_norm": 13.425495147705078, + "learning_rate": 1.2585193889541714e-05, + "loss": 0.9241, + "step": 309 + }, + { + "epoch": 0.24049650892164467, + "grad_norm": 16.872089385986328, + "learning_rate": 1.262632197414806e-05, + "loss": 1.9657, + "step": 310 + }, + { + "epoch": 0.24127230411171452, + "grad_norm": 16.951143264770508, + "learning_rate": 1.2667450058754405e-05, + "loss": 1.6671, + "step": 311 + }, + { + "epoch": 0.24204809930178434, + "grad_norm": 25.25611686706543, + "learning_rate": 1.2708578143360752e-05, + "loss": 3.1371, + "step": 312 + }, + { + "epoch": 0.24282389449185415, + "grad_norm": 16.216075897216797, + "learning_rate": 1.2749706227967097e-05, + "loss": 2.0971, + "step": 313 + }, + { + "epoch": 0.24359968968192397, + "grad_norm": 16.46155548095703, + "learning_rate": 1.2790834312573443e-05, + "loss": 1.3007, + "step": 314 + }, + { + "epoch": 0.2443754848719938, + "grad_norm": 2.83866286277771, + "learning_rate": 1.2831962397179786e-05, + "loss": 0.1475, + "step": 315 + }, + { + "epoch": 0.2451512800620636, + "grad_norm": 14.218758583068848, + "learning_rate": 1.2873090481786133e-05, + "loss": 1.1716, + "step": 316 + }, + { + "epoch": 0.24592707525213345, + "grad_norm": 14.27675724029541, + "learning_rate": 1.2914218566392478e-05, + "loss": 1.0757, + "step": 317 + }, + { + "epoch": 0.24670287044220326, + "grad_norm": 17.716360092163086, + "learning_rate": 1.2955346650998824e-05, + "loss": 1.2618, + "step": 318 + }, + { + "epoch": 0.24747866563227308, + "grad_norm": 16.388242721557617, + "learning_rate": 1.299647473560517e-05, + "loss": 1.3991, + "step": 319 + }, + { + "epoch": 0.2482544608223429, + "grad_norm": 16.443925857543945, + "learning_rate": 1.3037602820211514e-05, + "loss": 2.0074, + "step": 320 + }, + { + "epoch": 0.24903025601241272, + "grad_norm": 24.945213317871094, + "learning_rate": 1.307873090481786e-05, + "loss": 2.9443, + "step": 321 + }, + { + "epoch": 0.24980605120248253, + "grad_norm": 23.642480850219727, + "learning_rate": 1.3119858989424205e-05, + "loss": 2.9287, + "step": 322 + }, + { + "epoch": 0.2505818463925524, + "grad_norm": 18.337188720703125, + "learning_rate": 1.3160987074030552e-05, + "loss": 2.096, + "step": 323 + }, + { + "epoch": 0.2513576415826222, + "grad_norm": 15.318852424621582, + "learning_rate": 1.3202115158636897e-05, + "loss": 1.5946, + "step": 324 + }, + { + "epoch": 0.252133436772692, + "grad_norm": 19.109052658081055, + "learning_rate": 1.3243243243243242e-05, + "loss": 2.4606, + "step": 325 + }, + { + "epoch": 0.2529092319627618, + "grad_norm": 13.111356735229492, + "learning_rate": 1.3284371327849586e-05, + "loss": 0.8688, + "step": 326 + }, + { + "epoch": 0.25368502715283164, + "grad_norm": 12.39558219909668, + "learning_rate": 1.3325499412455933e-05, + "loss": 1.7387, + "step": 327 + }, + { + "epoch": 0.25446082234290146, + "grad_norm": 14.992557525634766, + "learning_rate": 1.3366627497062278e-05, + "loss": 2.1972, + "step": 328 + }, + { + "epoch": 0.2552366175329713, + "grad_norm": 23.38945770263672, + "learning_rate": 1.3407755581668624e-05, + "loss": 2.5069, + "step": 329 + }, + { + "epoch": 0.2560124127230411, + "grad_norm": 12.7628173828125, + "learning_rate": 1.344888366627497e-05, + "loss": 0.887, + "step": 330 + }, + { + "epoch": 0.25678820791311097, + "grad_norm": 11.964229583740234, + "learning_rate": 1.3490011750881316e-05, + "loss": 0.7386, + "step": 331 + }, + { + "epoch": 0.2575640031031808, + "grad_norm": 12.059389114379883, + "learning_rate": 1.353113983548766e-05, + "loss": 0.6545, + "step": 332 + }, + { + "epoch": 0.2583397982932506, + "grad_norm": 14.090944290161133, + "learning_rate": 1.3572267920094005e-05, + "loss": 1.3643, + "step": 333 + }, + { + "epoch": 0.2591155934833204, + "grad_norm": 16.549198150634766, + "learning_rate": 1.3613396004700352e-05, + "loss": 1.776, + "step": 334 + }, + { + "epoch": 0.25989138867339023, + "grad_norm": 15.8597993850708, + "learning_rate": 1.3654524089306697e-05, + "loss": 1.4132, + "step": 335 + }, + { + "epoch": 0.26066718386346005, + "grad_norm": 12.522405624389648, + "learning_rate": 1.3695652173913042e-05, + "loss": 0.9845, + "step": 336 + }, + { + "epoch": 0.26144297905352987, + "grad_norm": 13.252232551574707, + "learning_rate": 1.373678025851939e-05, + "loss": 1.0809, + "step": 337 + }, + { + "epoch": 0.2622187742435997, + "grad_norm": 21.052268981933594, + "learning_rate": 1.3777908343125733e-05, + "loss": 2.4095, + "step": 338 + }, + { + "epoch": 0.2629945694336695, + "grad_norm": 10.559894561767578, + "learning_rate": 1.3819036427732078e-05, + "loss": 0.6106, + "step": 339 + }, + { + "epoch": 0.2637703646237393, + "grad_norm": 15.451229095458984, + "learning_rate": 1.3860164512338423e-05, + "loss": 2.0685, + "step": 340 + }, + { + "epoch": 0.26454615981380913, + "grad_norm": 13.602385520935059, + "learning_rate": 1.390129259694477e-05, + "loss": 1.0661, + "step": 341 + }, + { + "epoch": 0.26532195500387895, + "grad_norm": 2.945145845413208, + "learning_rate": 1.3942420681551116e-05, + "loss": 0.1213, + "step": 342 + }, + { + "epoch": 0.2660977501939488, + "grad_norm": 3.1986005306243896, + "learning_rate": 1.398354876615746e-05, + "loss": 0.1238, + "step": 343 + }, + { + "epoch": 0.26687354538401864, + "grad_norm": 15.844751358032227, + "learning_rate": 1.4024676850763806e-05, + "loss": 1.2497, + "step": 344 + }, + { + "epoch": 0.26764934057408846, + "grad_norm": 14.065977096557617, + "learning_rate": 1.4065804935370152e-05, + "loss": 1.1994, + "step": 345 + }, + { + "epoch": 0.2684251357641583, + "grad_norm": 26.08740234375, + "learning_rate": 1.4106933019976497e-05, + "loss": 3.1447, + "step": 346 + }, + { + "epoch": 0.2692009309542281, + "grad_norm": 15.031670570373535, + "learning_rate": 1.4148061104582842e-05, + "loss": 1.1243, + "step": 347 + }, + { + "epoch": 0.2699767261442979, + "grad_norm": 11.64657211303711, + "learning_rate": 1.4189189189189189e-05, + "loss": 0.6845, + "step": 348 + }, + { + "epoch": 0.2707525213343677, + "grad_norm": 1.9088811874389648, + "learning_rate": 1.4230317273795533e-05, + "loss": 0.0838, + "step": 349 + }, + { + "epoch": 0.27152831652443754, + "grad_norm": 14.127081871032715, + "learning_rate": 1.4271445358401878e-05, + "loss": 1.0022, + "step": 350 + }, + { + "epoch": 0.27230411171450736, + "grad_norm": 14.686738014221191, + "learning_rate": 1.4312573443008225e-05, + "loss": 1.0077, + "step": 351 + }, + { + "epoch": 0.2730799069045772, + "grad_norm": 14.152548789978027, + "learning_rate": 1.435370152761457e-05, + "loss": 1.1412, + "step": 352 + }, + { + "epoch": 0.273855702094647, + "grad_norm": 10.20125961303711, + "learning_rate": 1.4394829612220916e-05, + "loss": 0.4218, + "step": 353 + }, + { + "epoch": 0.2746314972847168, + "grad_norm": 7.937293529510498, + "learning_rate": 1.4435957696827261e-05, + "loss": 0.3458, + "step": 354 + }, + { + "epoch": 0.2754072924747867, + "grad_norm": 20.850278854370117, + "learning_rate": 1.4477085781433606e-05, + "loss": 2.3721, + "step": 355 + }, + { + "epoch": 0.2761830876648565, + "grad_norm": 12.219514846801758, + "learning_rate": 1.4518213866039951e-05, + "loss": 0.7093, + "step": 356 + }, + { + "epoch": 0.2769588828549263, + "grad_norm": 14.681598663330078, + "learning_rate": 1.4559341950646297e-05, + "loss": 1.5243, + "step": 357 + }, + { + "epoch": 0.27773467804499613, + "grad_norm": 14.27827262878418, + "learning_rate": 1.4600470035252644e-05, + "loss": 1.0516, + "step": 358 + }, + { + "epoch": 0.27851047323506595, + "grad_norm": 15.308382987976074, + "learning_rate": 1.4641598119858989e-05, + "loss": 2.0051, + "step": 359 + }, + { + "epoch": 0.27928626842513576, + "grad_norm": 14.46717357635498, + "learning_rate": 1.4682726204465333e-05, + "loss": 0.9223, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_Qnli-dev_cosine_accuracy": 0.662109375, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.834648609161377, + "eval_Qnli-dev_cosine_ap": 0.6826335916480684, + "eval_Qnli-dev_cosine_f1": 0.676056338028169, + "eval_Qnli-dev_cosine_f1_threshold": 0.7109700441360474, + "eval_Qnli-dev_cosine_precision": 0.5359801488833746, + "eval_Qnli-dev_cosine_recall": 0.9152542372881356, + "eval_Qnli-dev_dot_accuracy": 0.650390625, + "eval_Qnli-dev_dot_accuracy_threshold": 460.17987060546875, + "eval_Qnli-dev_dot_ap": 0.6380420859457262, + "eval_Qnli-dev_dot_f1": 0.648729446935725, + "eval_Qnli-dev_dot_f1_threshold": 329.4967041015625, + "eval_Qnli-dev_dot_precision": 0.5011547344110855, + "eval_Qnli-dev_dot_recall": 0.9194915254237288, + "eval_Qnli-dev_euclidean_accuracy": 0.662109375, + "eval_Qnli-dev_euclidean_accuracy_threshold": 13.164579391479492, + "eval_Qnli-dev_euclidean_ap": 0.6789758761369369, + "eval_Qnli-dev_euclidean_f1": 0.6689536878216124, + "eval_Qnli-dev_euclidean_f1_threshold": 16.087860107421875, + "eval_Qnli-dev_euclidean_precision": 0.5619596541786743, + "eval_Qnli-dev_euclidean_recall": 0.826271186440678, + "eval_Qnli-dev_manhattan_accuracy": 0.6640625, + "eval_Qnli-dev_manhattan_accuracy_threshold": 256.0917053222656, + "eval_Qnli-dev_manhattan_ap": 0.6811393297620214, + "eval_Qnli-dev_manhattan_f1": 0.6755852842809364, + "eval_Qnli-dev_manhattan_f1_threshold": 292.66973876953125, + "eval_Qnli-dev_manhattan_precision": 0.5580110497237569, + "eval_Qnli-dev_manhattan_recall": 0.8559322033898306, + "eval_Qnli-dev_max_accuracy": 0.6640625, + "eval_Qnli-dev_max_accuracy_threshold": 460.17987060546875, + "eval_Qnli-dev_max_ap": 0.6826335916480684, + "eval_Qnli-dev_max_f1": 0.676056338028169, + "eval_Qnli-dev_max_f1_threshold": 329.4967041015625, + "eval_Qnli-dev_max_precision": 0.5619596541786743, + "eval_Qnli-dev_max_recall": 0.9194915254237288, + "eval_allNLI-dev_cosine_accuracy": 0.703125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8892226219177246, + "eval_allNLI-dev_cosine_ap": 0.5625011599380847, + "eval_allNLI-dev_cosine_f1": 0.5734265734265734, + "eval_allNLI-dev_cosine_f1_threshold": 0.8265600800514221, + "eval_allNLI-dev_cosine_precision": 0.48046875, + "eval_allNLI-dev_cosine_recall": 0.7109826589595376, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 542.1672973632812, + "eval_allNLI-dev_dot_ap": 0.45850058587644604, + "eval_allNLI-dev_dot_f1": 0.533066132264529, + "eval_allNLI-dev_dot_f1_threshold": 398.1407165527344, + "eval_allNLI-dev_dot_precision": 0.40797546012269936, + "eval_allNLI-dev_dot_recall": 0.7687861271676301, + "eval_allNLI-dev_euclidean_accuracy": 0.69921875, + "eval_allNLI-dev_euclidean_accuracy_threshold": 8.344038963317871, + "eval_allNLI-dev_euclidean_ap": 0.5603702127176354, + "eval_allNLI-dev_euclidean_f1": 0.5787037037037037, + "eval_allNLI-dev_euclidean_f1_threshold": 13.60634994506836, + "eval_allNLI-dev_euclidean_precision": 0.4826254826254826, + "eval_allNLI-dev_euclidean_recall": 0.7225433526011561, + "eval_allNLI-dev_manhattan_accuracy": 0.69921875, + "eval_allNLI-dev_manhattan_accuracy_threshold": 198.1402587890625, + "eval_allNLI-dev_manhattan_ap": 0.5553994905174271, + "eval_allNLI-dev_manhattan_f1": 0.5852534562211982, + "eval_allNLI-dev_manhattan_f1_threshold": 244.5894775390625, + "eval_allNLI-dev_manhattan_precision": 0.48659003831417624, + "eval_allNLI-dev_manhattan_recall": 0.7341040462427746, + "eval_allNLI-dev_max_accuracy": 0.703125, + "eval_allNLI-dev_max_accuracy_threshold": 542.1672973632812, + "eval_allNLI-dev_max_ap": 0.5625011599380847, + "eval_allNLI-dev_max_f1": 0.5852534562211982, + "eval_allNLI-dev_max_f1_threshold": 398.1407165527344, + "eval_allNLI-dev_max_precision": 0.48659003831417624, + "eval_allNLI-dev_max_recall": 0.7687861271676301, + "eval_sequential_score": 0.6826335916480684, + "eval_sts-test_pearson_cosine": 0.7278874955358379, + "eval_sts-test_pearson_dot": 0.6256106866273135, + "eval_sts-test_pearson_euclidean": 0.7487377481079524, + "eval_sts-test_pearson_manhattan": 0.7484942867556642, + "eval_sts-test_pearson_max": 0.7487377481079524, + "eval_sts-test_spearman_cosine": 0.7684265049248817, + "eval_sts-test_spearman_dot": 0.6173193062081594, + "eval_sts-test_spearman_euclidean": 0.7532756521219937, + "eval_sts-test_spearman_manhattan": 0.7533450613284117, + "eval_sts-test_spearman_max": 0.7684265049248817, + "eval_vitaminc-pairs_loss": 4.337933540344238, + "eval_vitaminc-pairs_runtime": 5.4814, + "eval_vitaminc-pairs_samples_per_second": 23.352, + "eval_vitaminc-pairs_steps_per_second": 0.365, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_negation-triplets_loss": 2.4210715293884277, + "eval_negation-triplets_runtime": 1.1024, + "eval_negation-triplets_samples_per_second": 116.112, + "eval_negation-triplets_steps_per_second": 1.814, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_scitail-pairs-pos_loss": 0.22071462869644165, + "eval_scitail-pairs-pos_runtime": 1.2259, + "eval_scitail-pairs-pos_samples_per_second": 104.41, + "eval_scitail-pairs-pos_steps_per_second": 1.631, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_scitail-pairs-qa_loss": 0.06664571166038513, + "eval_scitail-pairs-qa_runtime": 0.9051, + "eval_scitail-pairs-qa_samples_per_second": 141.425, + "eval_scitail-pairs-qa_steps_per_second": 2.21, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_xsum-pairs_loss": 0.8802127838134766, + "eval_xsum-pairs_runtime": 4.3088, + "eval_xsum-pairs_samples_per_second": 29.707, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_sciq_pairs_loss": 0.1746787279844284, + "eval_sciq_pairs_runtime": 5.5117, + "eval_sciq_pairs_samples_per_second": 23.223, + "eval_sciq_pairs_steps_per_second": 0.363, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_qasc_pairs_loss": 1.290103554725647, + "eval_qasc_pairs_runtime": 0.9843, + "eval_qasc_pairs_samples_per_second": 130.044, + "eval_qasc_pairs_steps_per_second": 2.032, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_openbookqa_pairs_loss": 2.800795316696167, + "eval_openbookqa_pairs_runtime": 0.8569, + "eval_openbookqa_pairs_samples_per_second": 149.381, + "eval_openbookqa_pairs_steps_per_second": 2.334, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_msmarco_pairs_loss": 1.993630051612854, + "eval_msmarco_pairs_runtime": 1.957, + "eval_msmarco_pairs_samples_per_second": 65.406, + "eval_msmarco_pairs_steps_per_second": 1.022, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_nq_pairs_loss": 2.73992919921875, + "eval_nq_pairs_runtime": 4.4047, + "eval_nq_pairs_samples_per_second": 29.06, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_trivia_pairs_loss": 2.3568363189697266, + "eval_trivia_pairs_runtime": 4.5275, + "eval_trivia_pairs_samples_per_second": 28.272, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_gooaq_pairs_loss": 1.59003484249115, + "eval_gooaq_pairs_runtime": 1.2638, + "eval_gooaq_pairs_samples_per_second": 101.28, + "eval_gooaq_pairs_steps_per_second": 1.583, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_paws-pos_loss": 0.05985124409198761, + "eval_paws-pos_runtime": 1.0235, + "eval_paws-pos_samples_per_second": 125.063, + "eval_paws-pos_steps_per_second": 1.954, + "step": 360 + }, + { + "epoch": 0.27928626842513576, + "eval_global_dataset_loss": 1.1724058389663696, + "eval_global_dataset_runtime": 21.1371, + "eval_global_dataset_samples_per_second": 19.681, + "eval_global_dataset_steps_per_second": 0.331, + "step": 360 + }, + { + "epoch": 0.2800620636152056, + "grad_norm": 25.474246978759766, + "learning_rate": 1.4723854289071678e-05, + "loss": 2.7167, + "step": 361 + }, + { + "epoch": 0.2808378588052754, + "grad_norm": 1.4754058122634888, + "learning_rate": 1.4764982373678025e-05, + "loss": 0.0458, + "step": 362 + }, + { + "epoch": 0.2816136539953452, + "grad_norm": 2.33425235748291, + "learning_rate": 1.480611045828437e-05, + "loss": 0.1212, + "step": 363 + }, + { + "epoch": 0.28238944918541503, + "grad_norm": 15.722271919250488, + "learning_rate": 1.4847238542890715e-05, + "loss": 1.1448, + "step": 364 + }, + { + "epoch": 0.28316524437548485, + "grad_norm": 19.69548988342285, + "learning_rate": 1.4888366627497062e-05, + "loss": 1.4511, + "step": 365 + }, + { + "epoch": 0.2839410395655547, + "grad_norm": 24.48932647705078, + "learning_rate": 1.4929494712103406e-05, + "loss": 2.2643, + "step": 366 + }, + { + "epoch": 0.28471683475562454, + "grad_norm": 25.97068214416504, + "learning_rate": 1.4970622796709751e-05, + "loss": 2.4578, + "step": 367 + }, + { + "epoch": 0.28549262994569435, + "grad_norm": 13.350829124450684, + "learning_rate": 1.5011750881316097e-05, + "loss": 1.6065, + "step": 368 + }, + { + "epoch": 0.28626842513576417, + "grad_norm": 22.40682601928711, + "learning_rate": 1.5052878965922444e-05, + "loss": 2.4521, + "step": 369 + }, + { + "epoch": 0.287044220325834, + "grad_norm": 15.37533950805664, + "learning_rate": 1.5094007050528789e-05, + "loss": 1.7531, + "step": 370 + }, + { + "epoch": 0.2878200155159038, + "grad_norm": 22.3481388092041, + "learning_rate": 1.5135135135135134e-05, + "loss": 2.1787, + "step": 371 + }, + { + "epoch": 0.2885958107059736, + "grad_norm": 13.490084648132324, + "learning_rate": 1.5176263219741478e-05, + "loss": 1.0957, + "step": 372 + }, + { + "epoch": 0.28937160589604344, + "grad_norm": 2.181879997253418, + "learning_rate": 1.5217391304347825e-05, + "loss": 0.0586, + "step": 373 + }, + { + "epoch": 0.29014740108611325, + "grad_norm": 16.611886978149414, + "learning_rate": 1.525851938895417e-05, + "loss": 1.6929, + "step": 374 + }, + { + "epoch": 0.29092319627618307, + "grad_norm": 14.349035263061523, + "learning_rate": 1.5299647473560515e-05, + "loss": 1.0535, + "step": 375 + }, + { + "epoch": 0.2916989914662529, + "grad_norm": 17.074005126953125, + "learning_rate": 1.534077555816686e-05, + "loss": 1.9976, + "step": 376 + }, + { + "epoch": 0.2924747866563227, + "grad_norm": 20.094833374023438, + "learning_rate": 1.5381903642773206e-05, + "loss": 2.4483, + "step": 377 + }, + { + "epoch": 0.2932505818463926, + "grad_norm": 17.122468948364258, + "learning_rate": 1.542303172737955e-05, + "loss": 1.9593, + "step": 378 + }, + { + "epoch": 0.2940263770364624, + "grad_norm": 13.059803009033203, + "learning_rate": 1.5464159811985896e-05, + "loss": 0.8981, + "step": 379 + }, + { + "epoch": 0.2948021722265322, + "grad_norm": 12.673219680786133, + "learning_rate": 1.5505287896592242e-05, + "loss": 1.5796, + "step": 380 + }, + { + "epoch": 0.295577967416602, + "grad_norm": 12.000088691711426, + "learning_rate": 1.554641598119859e-05, + "loss": 0.8614, + "step": 381 + }, + { + "epoch": 0.29635376260667184, + "grad_norm": 13.178820610046387, + "learning_rate": 1.5587544065804936e-05, + "loss": 1.0546, + "step": 382 + }, + { + "epoch": 0.29712955779674166, + "grad_norm": 11.239422798156738, + "learning_rate": 1.5628672150411278e-05, + "loss": 0.729, + "step": 383 + }, + { + "epoch": 0.2979053529868115, + "grad_norm": 14.764410018920898, + "learning_rate": 1.5669800235017623e-05, + "loss": 1.0482, + "step": 384 + }, + { + "epoch": 0.2986811481768813, + "grad_norm": 12.975874900817871, + "learning_rate": 1.571092831962397e-05, + "loss": 1.1714, + "step": 385 + }, + { + "epoch": 0.2994569433669511, + "grad_norm": 19.684154510498047, + "learning_rate": 1.5752056404230317e-05, + "loss": 2.4894, + "step": 386 + }, + { + "epoch": 0.30023273855702093, + "grad_norm": 17.636709213256836, + "learning_rate": 1.5793184488836662e-05, + "loss": 2.3088, + "step": 387 + }, + { + "epoch": 0.30100853374709075, + "grad_norm": 14.460874557495117, + "learning_rate": 1.5834312573443007e-05, + "loss": 0.9935, + "step": 388 + }, + { + "epoch": 0.30178432893716056, + "grad_norm": 16.176727294921875, + "learning_rate": 1.5875440658049353e-05, + "loss": 1.939, + "step": 389 + }, + { + "epoch": 0.30256012412723043, + "grad_norm": 13.404376983642578, + "learning_rate": 1.5916568742655698e-05, + "loss": 0.9026, + "step": 390 + }, + { + "epoch": 0.30333591931730025, + "grad_norm": 16.802038192749023, + "learning_rate": 1.5957696827262043e-05, + "loss": 1.8123, + "step": 391 + }, + { + "epoch": 0.30411171450737007, + "grad_norm": 12.580886840820312, + "learning_rate": 1.599882491186839e-05, + "loss": 0.874, + "step": 392 + }, + { + "epoch": 0.3048875096974399, + "grad_norm": 18.432588577270508, + "learning_rate": 1.6039952996474734e-05, + "loss": 2.1435, + "step": 393 + }, + { + "epoch": 0.3056633048875097, + "grad_norm": 14.046128273010254, + "learning_rate": 1.608108108108108e-05, + "loss": 0.931, + "step": 394 + }, + { + "epoch": 0.3064391000775795, + "grad_norm": 19.87921714782715, + "learning_rate": 1.6122209165687425e-05, + "loss": 3.171, + "step": 395 + }, + { + "epoch": 0.30721489526764933, + "grad_norm": 19.801437377929688, + "learning_rate": 1.616333725029377e-05, + "loss": 2.1008, + "step": 396 + }, + { + "epoch": 0.30799069045771915, + "grad_norm": 1.4523849487304688, + "learning_rate": 1.6204465334900115e-05, + "loss": 0.0452, + "step": 397 + }, + { + "epoch": 0.30876648564778897, + "grad_norm": 12.92918586730957, + "learning_rate": 1.624559341950646e-05, + "loss": 1.9875, + "step": 398 + }, + { + "epoch": 0.3095422808378588, + "grad_norm": 8.270234107971191, + "learning_rate": 1.628672150411281e-05, + "loss": 0.3775, + "step": 399 + }, + { + "epoch": 0.3103180760279286, + "grad_norm": 18.898799896240234, + "learning_rate": 1.632784958871915e-05, + "loss": 2.6522, + "step": 400 + }, + { + "epoch": 0.3110938712179985, + "grad_norm": 21.519569396972656, + "learning_rate": 1.6368977673325496e-05, + "loss": 2.5176, + "step": 401 + }, + { + "epoch": 0.3118696664080683, + "grad_norm": 11.268598556518555, + "learning_rate": 1.641010575793184e-05, + "loss": 0.8288, + "step": 402 + }, + { + "epoch": 0.3126454615981381, + "grad_norm": 1.690314769744873, + "learning_rate": 1.645123384253819e-05, + "loss": 0.0501, + "step": 403 + }, + { + "epoch": 0.3134212567882079, + "grad_norm": 14.973150253295898, + "learning_rate": 1.6492361927144535e-05, + "loss": 0.9043, + "step": 404 + }, + { + "epoch": 0.31419705197827774, + "grad_norm": 13.083118438720703, + "learning_rate": 1.653349001175088e-05, + "loss": 0.9715, + "step": 405 + }, + { + "epoch": 0.31497284716834756, + "grad_norm": 11.925381660461426, + "learning_rate": 1.6574618096357226e-05, + "loss": 0.7641, + "step": 406 + }, + { + "epoch": 0.3157486423584174, + "grad_norm": 11.790251731872559, + "learning_rate": 1.661574618096357e-05, + "loss": 0.6805, + "step": 407 + }, + { + "epoch": 0.3165244375484872, + "grad_norm": 12.217658042907715, + "learning_rate": 1.6656874265569917e-05, + "loss": 0.834, + "step": 408 + }, + { + "epoch": 0.317300232738557, + "grad_norm": 22.908329010009766, + "learning_rate": 1.6698002350176262e-05, + "loss": 2.4305, + "step": 409 + }, + { + "epoch": 0.3180760279286268, + "grad_norm": 26.299768447875977, + "learning_rate": 1.6739130434782607e-05, + "loss": 3.5913, + "step": 410 + }, + { + "epoch": 0.31885182311869664, + "grad_norm": 11.436812400817871, + "learning_rate": 1.6780258519388953e-05, + "loss": 0.7332, + "step": 411 + }, + { + "epoch": 0.31962761830876646, + "grad_norm": 13.18742561340332, + "learning_rate": 1.6821386603995298e-05, + "loss": 1.2849, + "step": 412 + }, + { + "epoch": 0.32040341349883633, + "grad_norm": 13.520853042602539, + "learning_rate": 1.6862514688601643e-05, + "loss": 1.0282, + "step": 413 + }, + { + "epoch": 0.32117920868890615, + "grad_norm": 165.97146606445312, + "learning_rate": 1.690364277320799e-05, + "loss": 2.2408, + "step": 414 + }, + { + "epoch": 0.32195500387897596, + "grad_norm": 10.768506050109863, + "learning_rate": 1.6944770857814334e-05, + "loss": 0.6605, + "step": 415 + }, + { + "epoch": 0.3227307990690458, + "grad_norm": 1.6408238410949707, + "learning_rate": 1.6985898942420682e-05, + "loss": 0.0823, + "step": 416 + }, + { + "epoch": 0.3235065942591156, + "grad_norm": 12.204421997070312, + "learning_rate": 1.7027027027027028e-05, + "loss": 0.6969, + "step": 417 + }, + { + "epoch": 0.3242823894491854, + "grad_norm": 1.6341387033462524, + "learning_rate": 1.706815511163337e-05, + "loss": 0.0801, + "step": 418 + }, + { + "epoch": 0.32505818463925523, + "grad_norm": 12.545511245727539, + "learning_rate": 1.7109283196239715e-05, + "loss": 0.8736, + "step": 419 + }, + { + "epoch": 0.32583397982932505, + "grad_norm": 20.437318801879883, + "learning_rate": 1.7150411280846064e-05, + "loss": 1.9045, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_Qnli-dev_cosine_accuracy": 0.66796875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8542718887329102, + "eval_Qnli-dev_cosine_ap": 0.6819867518828648, + "eval_Qnli-dev_cosine_f1": 0.6744186046511628, + "eval_Qnli-dev_cosine_f1_threshold": 0.7597143650054932, + "eval_Qnli-dev_cosine_precision": 0.5546448087431693, + "eval_Qnli-dev_cosine_recall": 0.8601694915254238, + "eval_Qnli-dev_dot_accuracy": 0.63671875, + "eval_Qnli-dev_dot_accuracy_threshold": 480.2711486816406, + "eval_Qnli-dev_dot_ap": 0.6310400691102859, + "eval_Qnli-dev_dot_f1": 0.641200545702592, + "eval_Qnli-dev_dot_f1_threshold": 281.8914794921875, + "eval_Qnli-dev_dot_precision": 0.47283702213279677, + "eval_Qnli-dev_dot_recall": 0.9957627118644068, + "eval_Qnli-dev_euclidean_accuracy": 0.6640625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.65697193145752, + "eval_Qnli-dev_euclidean_ap": 0.6795312124891731, + "eval_Qnli-dev_euclidean_f1": 0.6771653543307087, + "eval_Qnli-dev_euclidean_f1_threshold": 16.999248504638672, + "eval_Qnli-dev_euclidean_precision": 0.5388471177944862, + "eval_Qnli-dev_euclidean_recall": 0.9110169491525424, + "eval_Qnli-dev_manhattan_accuracy": 0.66796875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 221.05398559570312, + "eval_Qnli-dev_manhattan_ap": 0.6781803545466516, + "eval_Qnli-dev_manhattan_f1": 0.6798679867986799, + "eval_Qnli-dev_manhattan_f1_threshold": 285.72674560546875, + "eval_Qnli-dev_manhattan_precision": 0.5567567567567567, + "eval_Qnli-dev_manhattan_recall": 0.8728813559322034, + "eval_Qnli-dev_max_accuracy": 0.66796875, + "eval_Qnli-dev_max_accuracy_threshold": 480.2711486816406, + "eval_Qnli-dev_max_ap": 0.6819867518828648, + "eval_Qnli-dev_max_f1": 0.6798679867986799, + "eval_Qnli-dev_max_f1_threshold": 285.72674560546875, + "eval_Qnli-dev_max_precision": 0.5567567567567567, + "eval_Qnli-dev_max_recall": 0.9957627118644068, + "eval_allNLI-dev_cosine_accuracy": 0.703125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9137765765190125, + "eval_allNLI-dev_cosine_ap": 0.5636332850461632, + "eval_allNLI-dev_cosine_f1": 0.5595854922279792, + "eval_allNLI-dev_cosine_f1_threshold": 0.8483971953392029, + "eval_allNLI-dev_cosine_precision": 0.5070422535211268, + "eval_allNLI-dev_cosine_recall": 0.6242774566473989, + "eval_allNLI-dev_dot_accuracy": 0.677734375, + "eval_allNLI-dev_dot_accuracy_threshold": 528.9793701171875, + "eval_allNLI-dev_dot_ap": 0.45724093637403346, + "eval_allNLI-dev_dot_f1": 0.5354609929078015, + "eval_allNLI-dev_dot_f1_threshold": 370.6198425292969, + "eval_allNLI-dev_dot_precision": 0.38618925831202044, + "eval_allNLI-dev_dot_recall": 0.8728323699421965, + "eval_allNLI-dev_euclidean_accuracy": 0.703125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.576183319091797, + "eval_allNLI-dev_euclidean_ap": 0.5590106513557265, + "eval_allNLI-dev_euclidean_f1": 0.5750000000000001, + "eval_allNLI-dev_euclidean_f1_threshold": 12.741243362426758, + "eval_allNLI-dev_euclidean_precision": 0.5066079295154186, + "eval_allNLI-dev_euclidean_recall": 0.6647398843930635, + "eval_allNLI-dev_manhattan_accuracy": 0.69921875, + "eval_allNLI-dev_manhattan_accuracy_threshold": 180.15008544921875, + "eval_allNLI-dev_manhattan_ap": 0.5567627961543863, + "eval_allNLI-dev_manhattan_f1": 0.5842105263157894, + "eval_allNLI-dev_manhattan_f1_threshold": 222.73663330078125, + "eval_allNLI-dev_manhattan_precision": 0.5362318840579711, + "eval_allNLI-dev_manhattan_recall": 0.6416184971098265, + "eval_allNLI-dev_max_accuracy": 0.703125, + "eval_allNLI-dev_max_accuracy_threshold": 528.9793701171875, + "eval_allNLI-dev_max_ap": 0.5636332850461632, + "eval_allNLI-dev_max_f1": 0.5842105263157894, + "eval_allNLI-dev_max_f1_threshold": 370.6198425292969, + "eval_allNLI-dev_max_precision": 0.5362318840579711, + "eval_allNLI-dev_max_recall": 0.8728323699421965, + "eval_sequential_score": 0.6819867518828648, + "eval_sts-test_pearson_cosine": 0.764240840847642, + "eval_sts-test_pearson_dot": 0.6960700855602846, + "eval_sts-test_pearson_euclidean": 0.783068375173582, + "eval_sts-test_pearson_manhattan": 0.7868630276962296, + "eval_sts-test_pearson_max": 0.7868630276962296, + "eval_sts-test_spearman_cosine": 0.7990335036358699, + "eval_sts-test_spearman_dot": 0.6838227237215536, + "eval_sts-test_spearman_euclidean": 0.7842250874851489, + "eval_sts-test_spearman_manhattan": 0.7874060966473723, + "eval_sts-test_spearman_max": 0.7990335036358699, + "eval_vitaminc-pairs_loss": 4.480936050415039, + "eval_vitaminc-pairs_runtime": 5.4647, + "eval_vitaminc-pairs_samples_per_second": 23.423, + "eval_vitaminc-pairs_steps_per_second": 0.366, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_negation-triplets_loss": 2.1231603622436523, + "eval_negation-triplets_runtime": 1.101, + "eval_negation-triplets_samples_per_second": 116.261, + "eval_negation-triplets_steps_per_second": 1.817, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_scitail-pairs-pos_loss": 0.18876375257968903, + "eval_scitail-pairs-pos_runtime": 1.2357, + "eval_scitail-pairs-pos_samples_per_second": 103.582, + "eval_scitail-pairs-pos_steps_per_second": 1.618, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_scitail-pairs-qa_loss": 0.08243328332901001, + "eval_scitail-pairs-qa_runtime": 0.9098, + "eval_scitail-pairs-qa_samples_per_second": 140.687, + "eval_scitail-pairs-qa_steps_per_second": 2.198, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_xsum-pairs_loss": 0.6536691188812256, + "eval_xsum-pairs_runtime": 4.3213, + "eval_xsum-pairs_samples_per_second": 29.62, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_sciq_pairs_loss": 0.13692253828048706, + "eval_sciq_pairs_runtime": 5.5159, + "eval_sciq_pairs_samples_per_second": 23.206, + "eval_sciq_pairs_steps_per_second": 0.363, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_qasc_pairs_loss": 1.1581523418426514, + "eval_qasc_pairs_runtime": 0.9753, + "eval_qasc_pairs_samples_per_second": 131.245, + "eval_qasc_pairs_steps_per_second": 2.051, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_openbookqa_pairs_loss": 2.305072069168091, + "eval_openbookqa_pairs_runtime": 0.8571, + "eval_openbookqa_pairs_samples_per_second": 149.34, + "eval_openbookqa_pairs_steps_per_second": 2.333, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_msmarco_pairs_loss": 1.591927170753479, + "eval_msmarco_pairs_runtime": 1.9618, + "eval_msmarco_pairs_samples_per_second": 65.247, + "eval_msmarco_pairs_steps_per_second": 1.019, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_nq_pairs_loss": 2.2807977199554443, + "eval_nq_pairs_runtime": 4.4086, + "eval_nq_pairs_samples_per_second": 29.034, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_trivia_pairs_loss": 1.8372831344604492, + "eval_trivia_pairs_runtime": 4.5198, + "eval_trivia_pairs_samples_per_second": 28.32, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_gooaq_pairs_loss": 1.4098910093307495, + "eval_gooaq_pairs_runtime": 1.2673, + "eval_gooaq_pairs_samples_per_second": 101.002, + "eval_gooaq_pairs_steps_per_second": 1.578, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_paws-pos_loss": 0.05875324085354805, + "eval_paws-pos_runtime": 1.0192, + "eval_paws-pos_samples_per_second": 125.591, + "eval_paws-pos_steps_per_second": 1.962, + "step": 420 + }, + { + "epoch": 0.32583397982932505, + "eval_global_dataset_loss": 1.0107834339141846, + "eval_global_dataset_runtime": 21.1404, + "eval_global_dataset_samples_per_second": 19.678, + "eval_global_dataset_steps_per_second": 0.331, + "step": 420 + }, + { + "epoch": 0.32660977501939487, + "grad_norm": 14.164385795593262, + "learning_rate": 1.719153936545241e-05, + "loss": 1.1833, + "step": 421 + }, + { + "epoch": 0.3273855702094647, + "grad_norm": 13.15245246887207, + "learning_rate": 1.7232667450058754e-05, + "loss": 0.9057, + "step": 422 + }, + { + "epoch": 0.3281613653995345, + "grad_norm": 10.071792602539062, + "learning_rate": 1.7273795534665096e-05, + "loss": 0.6268, + "step": 423 + }, + { + "epoch": 0.32893716058960437, + "grad_norm": 8.69880485534668, + "learning_rate": 1.7314923619271445e-05, + "loss": 0.4343, + "step": 424 + }, + { + "epoch": 0.3297129557796742, + "grad_norm": 21.93539047241211, + "learning_rate": 1.735605170387779e-05, + "loss": 1.9356, + "step": 425 + }, + { + "epoch": 0.330488750969744, + "grad_norm": 12.004720687866211, + "learning_rate": 1.7397179788484135e-05, + "loss": 0.7565, + "step": 426 + }, + { + "epoch": 0.3312645461598138, + "grad_norm": 19.00567054748535, + "learning_rate": 1.743830787309048e-05, + "loss": 2.3958, + "step": 427 + }, + { + "epoch": 0.33204034134988364, + "grad_norm": 21.73807716369629, + "learning_rate": 1.7479435957696826e-05, + "loss": 2.2136, + "step": 428 + }, + { + "epoch": 0.33281613653995346, + "grad_norm": 14.654407501220703, + "learning_rate": 1.752056404230317e-05, + "loss": 0.9757, + "step": 429 + }, + { + "epoch": 0.3335919317300233, + "grad_norm": 17.83519744873047, + "learning_rate": 1.7561692126909516e-05, + "loss": 2.1771, + "step": 430 + }, + { + "epoch": 0.3343677269200931, + "grad_norm": 14.094572067260742, + "learning_rate": 1.7602820211515865e-05, + "loss": 1.4981, + "step": 431 + }, + { + "epoch": 0.3351435221101629, + "grad_norm": 9.891066551208496, + "learning_rate": 1.7643948296122207e-05, + "loss": 0.6229, + "step": 432 + }, + { + "epoch": 0.3359193173002327, + "grad_norm": 20.44182777404785, + "learning_rate": 1.7685076380728552e-05, + "loss": 3.4441, + "step": 433 + }, + { + "epoch": 0.33669511249030254, + "grad_norm": 16.11002540588379, + "learning_rate": 1.77262044653349e-05, + "loss": 1.9118, + "step": 434 + }, + { + "epoch": 0.33747090768037236, + "grad_norm": 15.73004150390625, + "learning_rate": 1.7767332549941243e-05, + "loss": 1.7004, + "step": 435 + }, + { + "epoch": 0.33824670287044223, + "grad_norm": 10.259026527404785, + "learning_rate": 1.780846063454759e-05, + "loss": 0.7453, + "step": 436 + }, + { + "epoch": 0.33902249806051205, + "grad_norm": 17.650493621826172, + "learning_rate": 1.7849588719153933e-05, + "loss": 2.1001, + "step": 437 + }, + { + "epoch": 0.33979829325058186, + "grad_norm": 13.035713195800781, + "learning_rate": 1.789071680376028e-05, + "loss": 1.4916, + "step": 438 + }, + { + "epoch": 0.3405740884406517, + "grad_norm": 8.519974708557129, + "learning_rate": 1.7931844888366627e-05, + "loss": 0.4525, + "step": 439 + }, + { + "epoch": 0.3413498836307215, + "grad_norm": 14.857168197631836, + "learning_rate": 1.797297297297297e-05, + "loss": 1.4354, + "step": 440 + }, + { + "epoch": 0.3421256788207913, + "grad_norm": 8.651593208312988, + "learning_rate": 1.8014101057579318e-05, + "loss": 0.6315, + "step": 441 + }, + { + "epoch": 0.34290147401086113, + "grad_norm": 11.607765197753906, + "learning_rate": 1.8055229142185663e-05, + "loss": 0.8246, + "step": 442 + }, + { + "epoch": 0.34367726920093095, + "grad_norm": 14.755167007446289, + "learning_rate": 1.8096357226792005e-05, + "loss": 2.0078, + "step": 443 + }, + { + "epoch": 0.34445306439100076, + "grad_norm": 6.7243804931640625, + "learning_rate": 1.8137485311398354e-05, + "loss": 0.2752, + "step": 444 + }, + { + "epoch": 0.3452288595810706, + "grad_norm": 19.55060577392578, + "learning_rate": 1.81786133960047e-05, + "loss": 1.8133, + "step": 445 + }, + { + "epoch": 0.3460046547711404, + "grad_norm": 22.428977966308594, + "learning_rate": 1.8219741480611044e-05, + "loss": 2.1664, + "step": 446 + }, + { + "epoch": 0.3467804499612102, + "grad_norm": 15.759836196899414, + "learning_rate": 1.826086956521739e-05, + "loss": 1.0758, + "step": 447 + }, + { + "epoch": 0.3475562451512801, + "grad_norm": 20.637527465820312, + "learning_rate": 1.830199764982374e-05, + "loss": 2.0718, + "step": 448 + }, + { + "epoch": 0.3483320403413499, + "grad_norm": 19.373313903808594, + "learning_rate": 1.834312573443008e-05, + "loss": 1.5408, + "step": 449 + }, + { + "epoch": 0.3491078355314197, + "grad_norm": 2.981391668319702, + "learning_rate": 1.8384253819036426e-05, + "loss": 0.1437, + "step": 450 + }, + { + "epoch": 0.34988363072148954, + "grad_norm": 11.054551124572754, + "learning_rate": 1.842538190364277e-05, + "loss": 0.9747, + "step": 451 + }, + { + "epoch": 0.35065942591155935, + "grad_norm": 14.374537467956543, + "learning_rate": 1.8466509988249116e-05, + "loss": 0.9244, + "step": 452 + }, + { + "epoch": 0.35143522110162917, + "grad_norm": 14.106080055236816, + "learning_rate": 1.8507638072855465e-05, + "loss": 1.0036, + "step": 453 + }, + { + "epoch": 0.352211016291699, + "grad_norm": 18.196334838867188, + "learning_rate": 1.8548766157461807e-05, + "loss": 1.9532, + "step": 454 + }, + { + "epoch": 0.3529868114817688, + "grad_norm": 15.430368423461914, + "learning_rate": 1.8589894242068152e-05, + "loss": 1.425, + "step": 455 + }, + { + "epoch": 0.3537626066718386, + "grad_norm": 12.661757469177246, + "learning_rate": 1.86310223266745e-05, + "loss": 0.7721, + "step": 456 + }, + { + "epoch": 0.35453840186190844, + "grad_norm": 12.894657135009766, + "learning_rate": 1.8672150411280843e-05, + "loss": 0.8837, + "step": 457 + }, + { + "epoch": 0.35531419705197825, + "grad_norm": 14.356178283691406, + "learning_rate": 1.871327849588719e-05, + "loss": 0.8375, + "step": 458 + }, + { + "epoch": 0.3560899922420481, + "grad_norm": 14.401089668273926, + "learning_rate": 1.8754406580493537e-05, + "loss": 1.5555, + "step": 459 + }, + { + "epoch": 0.35686578743211794, + "grad_norm": 13.208641052246094, + "learning_rate": 1.879553466509988e-05, + "loss": 1.9662, + "step": 460 + }, + { + "epoch": 0.35764158262218776, + "grad_norm": 18.4484806060791, + "learning_rate": 1.8836662749706227e-05, + "loss": 3.0869, + "step": 461 + }, + { + "epoch": 0.3584173778122576, + "grad_norm": 17.628223419189453, + "learning_rate": 1.887779083431257e-05, + "loss": 1.8195, + "step": 462 + }, + { + "epoch": 0.3591931730023274, + "grad_norm": 10.189529418945312, + "learning_rate": 1.8918918918918918e-05, + "loss": 0.7404, + "step": 463 + }, + { + "epoch": 0.3599689681923972, + "grad_norm": 11.85727596282959, + "learning_rate": 1.8960047003525263e-05, + "loss": 1.0893, + "step": 464 + }, + { + "epoch": 0.360744763382467, + "grad_norm": 9.262800216674805, + "learning_rate": 1.9001175088131608e-05, + "loss": 0.707, + "step": 465 + }, + { + "epoch": 0.36152055857253684, + "grad_norm": 9.710227966308594, + "learning_rate": 1.9042303172737954e-05, + "loss": 0.6543, + "step": 466 + }, + { + "epoch": 0.36229635376260666, + "grad_norm": 10.734469413757324, + "learning_rate": 1.90834312573443e-05, + "loss": 0.804, + "step": 467 + }, + { + "epoch": 0.3630721489526765, + "grad_norm": 10.810818672180176, + "learning_rate": 1.9124559341950644e-05, + "loss": 0.8668, + "step": 468 + }, + { + "epoch": 0.3638479441427463, + "grad_norm": 18.30976104736328, + "learning_rate": 1.916568742655699e-05, + "loss": 1.9092, + "step": 469 + }, + { + "epoch": 0.3646237393328161, + "grad_norm": 12.41912841796875, + "learning_rate": 1.9206815511163338e-05, + "loss": 0.8892, + "step": 470 + }, + { + "epoch": 0.365399534522886, + "grad_norm": 13.264505386352539, + "learning_rate": 1.924794359576968e-05, + "loss": 1.1507, + "step": 471 + }, + { + "epoch": 0.3661753297129558, + "grad_norm": 10.896422386169434, + "learning_rate": 1.9289071680376025e-05, + "loss": 0.7981, + "step": 472 + }, + { + "epoch": 0.3669511249030256, + "grad_norm": 11.71090030670166, + "learning_rate": 1.9330199764982374e-05, + "loss": 0.8476, + "step": 473 + }, + { + "epoch": 0.36772692009309543, + "grad_norm": 11.638192176818848, + "learning_rate": 1.9371327849588716e-05, + "loss": 0.8267, + "step": 474 + }, + { + "epoch": 0.36850271528316525, + "grad_norm": 9.334179878234863, + "learning_rate": 1.9412455934195065e-05, + "loss": 0.5709, + "step": 475 + }, + { + "epoch": 0.36927851047323507, + "grad_norm": 1.667304515838623, + "learning_rate": 1.9453584018801406e-05, + "loss": 0.0949, + "step": 476 + }, + { + "epoch": 0.3700543056633049, + "grad_norm": 7.3601250648498535, + "learning_rate": 1.9494712103407755e-05, + "loss": 0.3, + "step": 477 + }, + { + "epoch": 0.3708301008533747, + "grad_norm": 17.925691604614258, + "learning_rate": 1.95358401880141e-05, + "loss": 1.6049, + "step": 478 + }, + { + "epoch": 0.3716058960434445, + "grad_norm": 11.621016502380371, + "learning_rate": 1.9576968272620442e-05, + "loss": 0.5664, + "step": 479 + }, + { + "epoch": 0.37238169123351433, + "grad_norm": 18.168704986572266, + "learning_rate": 1.961809635722679e-05, + "loss": 1.7042, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_Qnli-dev_cosine_accuracy": 0.666015625, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.846991777420044, + "eval_Qnli-dev_cosine_ap": 0.6953468164893489, + "eval_Qnli-dev_cosine_f1": 0.6762360446570973, + "eval_Qnli-dev_cosine_f1_threshold": 0.7319585680961609, + "eval_Qnli-dev_cosine_precision": 0.5421994884910486, + "eval_Qnli-dev_cosine_recall": 0.8983050847457628, + "eval_Qnli-dev_dot_accuracy": 0.623046875, + "eval_Qnli-dev_dot_accuracy_threshold": 412.3604736328125, + "eval_Qnli-dev_dot_ap": 0.6263988112924701, + "eval_Qnli-dev_dot_f1": 0.6513157894736842, + "eval_Qnli-dev_dot_f1_threshold": 378.31402587890625, + "eval_Qnli-dev_dot_precision": 0.532258064516129, + "eval_Qnli-dev_dot_recall": 0.8389830508474576, + "eval_Qnli-dev_euclidean_accuracy": 0.673828125, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.859345436096191, + "eval_Qnli-dev_euclidean_ap": 0.690286173236297, + "eval_Qnli-dev_euclidean_f1": 0.6709677419354838, + "eval_Qnli-dev_euclidean_f1_threshold": 16.59159278869629, + "eval_Qnli-dev_euclidean_precision": 0.5416666666666666, + "eval_Qnli-dev_euclidean_recall": 0.8813559322033898, + "eval_Qnli-dev_manhattan_accuracy": 0.6640625, + "eval_Qnli-dev_manhattan_accuracy_threshold": 225.02207946777344, + "eval_Qnli-dev_manhattan_ap": 0.6878513586791358, + "eval_Qnli-dev_manhattan_f1": 0.672811059907834, + "eval_Qnli-dev_manhattan_f1_threshold": 314.00433349609375, + "eval_Qnli-dev_manhattan_precision": 0.5277108433734939, + "eval_Qnli-dev_manhattan_recall": 0.9279661016949152, + "eval_Qnli-dev_max_accuracy": 0.673828125, + "eval_Qnli-dev_max_accuracy_threshold": 412.3604736328125, + "eval_Qnli-dev_max_ap": 0.6953468164893489, + "eval_Qnli-dev_max_f1": 0.6762360446570973, + "eval_Qnli-dev_max_f1_threshold": 378.31402587890625, + "eval_Qnli-dev_max_precision": 0.5421994884910486, + "eval_Qnli-dev_max_recall": 0.9279661016949152, + "eval_allNLI-dev_cosine_accuracy": 0.7109375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8910847902297974, + "eval_allNLI-dev_cosine_ap": 0.5707600055515883, + "eval_allNLI-dev_cosine_f1": 0.5931372549019609, + "eval_allNLI-dev_cosine_f1_threshold": 0.8503602743148804, + "eval_allNLI-dev_cosine_precision": 0.5148936170212766, + "eval_allNLI-dev_cosine_recall": 0.6994219653179191, + "eval_allNLI-dev_dot_accuracy": 0.673828125, + "eval_allNLI-dev_dot_accuracy_threshold": 505.83184814453125, + "eval_allNLI-dev_dot_ap": 0.46124922397408263, + "eval_allNLI-dev_dot_f1": 0.535645472061657, + "eval_allNLI-dev_dot_f1_threshold": 381.40283203125, + "eval_allNLI-dev_dot_precision": 0.40173410404624277, + "eval_allNLI-dev_dot_recall": 0.8034682080924855, + "eval_allNLI-dev_euclidean_accuracy": 0.7109375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.794914245605469, + "eval_allNLI-dev_euclidean_ap": 0.5708890610588562, + "eval_allNLI-dev_euclidean_f1": 0.5940594059405941, + "eval_allNLI-dev_euclidean_f1_threshold": 12.236215591430664, + "eval_allNLI-dev_euclidean_precision": 0.5194805194805194, + "eval_allNLI-dev_euclidean_recall": 0.6936416184971098, + "eval_allNLI-dev_manhattan_accuracy": 0.705078125, + "eval_allNLI-dev_manhattan_accuracy_threshold": 182.04345703125, + "eval_allNLI-dev_manhattan_ap": 0.5681140609576328, + "eval_allNLI-dev_manhattan_f1": 0.5981735159817351, + "eval_allNLI-dev_manhattan_f1_threshold": 231.1646728515625, + "eval_allNLI-dev_manhattan_precision": 0.49433962264150944, + "eval_allNLI-dev_manhattan_recall": 0.7572254335260116, + "eval_allNLI-dev_max_accuracy": 0.7109375, + "eval_allNLI-dev_max_accuracy_threshold": 505.83184814453125, + "eval_allNLI-dev_max_ap": 0.5708890610588562, + "eval_allNLI-dev_max_f1": 0.5981735159817351, + "eval_allNLI-dev_max_f1_threshold": 381.40283203125, + "eval_allNLI-dev_max_precision": 0.5194805194805194, + "eval_allNLI-dev_max_recall": 0.8034682080924855, + "eval_sequential_score": 0.6953468164893489, + "eval_sts-test_pearson_cosine": 0.74918429599219, + "eval_sts-test_pearson_dot": 0.6827959529792219, + "eval_sts-test_pearson_euclidean": 0.7729030843418438, + "eval_sts-test_pearson_manhattan": 0.7725078896091997, + "eval_sts-test_pearson_max": 0.7729030843418438, + "eval_sts-test_spearman_cosine": 0.7867855309056578, + "eval_sts-test_spearman_dot": 0.6693821181882085, + "eval_sts-test_spearman_euclidean": 0.7744769649675218, + "eval_sts-test_spearman_manhattan": 0.7742014762023963, + "eval_sts-test_spearman_max": 0.7867855309056578, + "eval_vitaminc-pairs_loss": 3.861807346343994, + "eval_vitaminc-pairs_runtime": 5.4626, + "eval_vitaminc-pairs_samples_per_second": 23.432, + "eval_vitaminc-pairs_steps_per_second": 0.366, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_negation-triplets_loss": 2.006932020187378, + "eval_negation-triplets_runtime": 1.105, + "eval_negation-triplets_samples_per_second": 115.836, + "eval_negation-triplets_steps_per_second": 1.81, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_scitail-pairs-pos_loss": 0.1834138035774231, + "eval_scitail-pairs-pos_runtime": 1.2306, + "eval_scitail-pairs-pos_samples_per_second": 104.015, + "eval_scitail-pairs-pos_steps_per_second": 1.625, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_scitail-pairs-qa_loss": 0.044225938618183136, + "eval_scitail-pairs-qa_runtime": 0.9088, + "eval_scitail-pairs-qa_samples_per_second": 140.849, + "eval_scitail-pairs-qa_steps_per_second": 2.201, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_xsum-pairs_loss": 0.5437339544296265, + "eval_xsum-pairs_runtime": 4.3111, + "eval_xsum-pairs_samples_per_second": 29.691, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_sciq_pairs_loss": 0.14562037587165833, + "eval_sciq_pairs_runtime": 5.5577, + "eval_sciq_pairs_samples_per_second": 23.031, + "eval_sciq_pairs_steps_per_second": 0.36, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_qasc_pairs_loss": 0.8399978876113892, + "eval_qasc_pairs_runtime": 1.0153, + "eval_qasc_pairs_samples_per_second": 126.07, + "eval_qasc_pairs_steps_per_second": 1.97, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_openbookqa_pairs_loss": 2.159867763519287, + "eval_openbookqa_pairs_runtime": 0.8651, + "eval_openbookqa_pairs_samples_per_second": 147.961, + "eval_openbookqa_pairs_steps_per_second": 2.312, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_msmarco_pairs_loss": 1.5386474132537842, + "eval_msmarco_pairs_runtime": 1.9546, + "eval_msmarco_pairs_samples_per_second": 65.488, + "eval_msmarco_pairs_steps_per_second": 1.023, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_nq_pairs_loss": 1.7472549676895142, + "eval_nq_pairs_runtime": 4.4021, + "eval_nq_pairs_samples_per_second": 29.077, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_trivia_pairs_loss": 1.2336030006408691, + "eval_trivia_pairs_runtime": 4.525, + "eval_trivia_pairs_samples_per_second": 28.287, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_gooaq_pairs_loss": 1.1999468803405762, + "eval_gooaq_pairs_runtime": 1.2616, + "eval_gooaq_pairs_samples_per_second": 101.458, + "eval_gooaq_pairs_steps_per_second": 1.585, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_paws-pos_loss": 0.058183394372463226, + "eval_paws-pos_runtime": 1.0254, + "eval_paws-pos_samples_per_second": 124.829, + "eval_paws-pos_steps_per_second": 1.95, + "step": 480 + }, + { + "epoch": 0.37238169123351433, + "eval_global_dataset_loss": 0.8533011078834534, + "eval_global_dataset_runtime": 21.5193, + "eval_global_dataset_samples_per_second": 19.332, + "eval_global_dataset_steps_per_second": 0.325, + "step": 480 + }, + { + "epoch": 0.37315748642358415, + "grad_norm": 14.54818344116211, + "learning_rate": 1.9659224441833136e-05, + "loss": 0.9542, + "step": 481 + }, + { + "epoch": 0.373933281613654, + "grad_norm": 12.85341739654541, + "learning_rate": 1.970035252643948e-05, + "loss": 1.8103, + "step": 482 + }, + { + "epoch": 0.37470907680372384, + "grad_norm": 13.623619079589844, + "learning_rate": 1.9741480611045827e-05, + "loss": 0.7318, + "step": 483 + }, + { + "epoch": 0.37548487199379366, + "grad_norm": 18.0240535736084, + "learning_rate": 1.9782608695652172e-05, + "loss": 1.8031, + "step": 484 + }, + { + "epoch": 0.3762606671838635, + "grad_norm": 10.82210922241211, + "learning_rate": 1.9823736780258517e-05, + "loss": 0.8064, + "step": 485 + }, + { + "epoch": 0.3770364623739333, + "grad_norm": 11.726639747619629, + "learning_rate": 1.9864864864864863e-05, + "loss": 0.9347, + "step": 486 + }, + { + "epoch": 0.3778122575640031, + "grad_norm": 14.470053672790527, + "learning_rate": 1.990599294947121e-05, + "loss": 1.6457, + "step": 487 + }, + { + "epoch": 0.3785880527540729, + "grad_norm": 11.309890747070312, + "learning_rate": 1.9947121034077553e-05, + "loss": 0.6571, + "step": 488 + }, + { + "epoch": 0.37936384794414274, + "grad_norm": 10.566116333007812, + "learning_rate": 1.99882491186839e-05, + "loss": 0.6678, + "step": 489 + }, + { + "epoch": 0.38013964313421256, + "grad_norm": 15.370992660522461, + "learning_rate": 2.0029377203290244e-05, + "loss": 1.2484, + "step": 490 + }, + { + "epoch": 0.3809154383242824, + "grad_norm": 5.891637325286865, + "learning_rate": 2.007050528789659e-05, + "loss": 0.2608, + "step": 491 + }, + { + "epoch": 0.3816912335143522, + "grad_norm": 16.717676162719727, + "learning_rate": 2.0111633372502938e-05, + "loss": 1.5938, + "step": 492 + }, + { + "epoch": 0.382467028704422, + "grad_norm": 20.03619384765625, + "learning_rate": 2.015276145710928e-05, + "loss": 1.9322, + "step": 493 + }, + { + "epoch": 0.3832428238944919, + "grad_norm": 18.93113899230957, + "learning_rate": 2.019388954171563e-05, + "loss": 1.966, + "step": 494 + }, + { + "epoch": 0.3840186190845617, + "grad_norm": 18.57356834411621, + "learning_rate": 2.0235017626321974e-05, + "loss": 1.9759, + "step": 495 + }, + { + "epoch": 0.3847944142746315, + "grad_norm": 14.238194465637207, + "learning_rate": 2.0276145710928316e-05, + "loss": 1.7347, + "step": 496 + }, + { + "epoch": 0.38557020946470133, + "grad_norm": 14.526615142822266, + "learning_rate": 2.0317273795534664e-05, + "loss": 1.4667, + "step": 497 + }, + { + "epoch": 0.38634600465477115, + "grad_norm": 12.894027709960938, + "learning_rate": 2.035840188014101e-05, + "loss": 0.9997, + "step": 498 + }, + { + "epoch": 0.38712179984484096, + "grad_norm": 11.474167823791504, + "learning_rate": 2.0399529964747355e-05, + "loss": 0.9709, + "step": 499 + }, + { + "epoch": 0.3878975950349108, + "grad_norm": 12.648843765258789, + "learning_rate": 2.04406580493537e-05, + "loss": 1.2046, + "step": 500 + }, + { + "epoch": 0.3886733902249806, + "grad_norm": 11.636364936828613, + "learning_rate": 2.0481786133960042e-05, + "loss": 1.4984, + "step": 501 + }, + { + "epoch": 0.3894491854150504, + "grad_norm": 15.61589527130127, + "learning_rate": 2.052291421856639e-05, + "loss": 1.81, + "step": 502 + }, + { + "epoch": 0.39022498060512023, + "grad_norm": 14.492650032043457, + "learning_rate": 2.0564042303172736e-05, + "loss": 1.4561, + "step": 503 + }, + { + "epoch": 0.39100077579519005, + "grad_norm": 27.832653045654297, + "learning_rate": 2.060517038777908e-05, + "loss": 0.6121, + "step": 504 + }, + { + "epoch": 0.39177657098525986, + "grad_norm": 12.697063446044922, + "learning_rate": 2.0646298472385427e-05, + "loss": 1.4503, + "step": 505 + }, + { + "epoch": 0.39255236617532974, + "grad_norm": 10.05654525756836, + "learning_rate": 2.0687426556991772e-05, + "loss": 0.7027, + "step": 506 + }, + { + "epoch": 0.39332816136539955, + "grad_norm": 10.353205680847168, + "learning_rate": 2.0728554641598117e-05, + "loss": 0.9195, + "step": 507 + }, + { + "epoch": 0.39410395655546937, + "grad_norm": 8.179510116577148, + "learning_rate": 2.0769682726204463e-05, + "loss": 0.3975, + "step": 508 + }, + { + "epoch": 0.3948797517455392, + "grad_norm": 9.739526748657227, + "learning_rate": 2.081081081081081e-05, + "loss": 0.6786, + "step": 509 + }, + { + "epoch": 0.395655546935609, + "grad_norm": 18.27353858947754, + "learning_rate": 2.0851938895417153e-05, + "loss": 1.8421, + "step": 510 + }, + { + "epoch": 0.3964313421256788, + "grad_norm": 9.571355819702148, + "learning_rate": 2.0893066980023502e-05, + "loss": 0.4891, + "step": 511 + }, + { + "epoch": 0.39720713731574864, + "grad_norm": 10.137935638427734, + "learning_rate": 2.0934195064629847e-05, + "loss": 0.5814, + "step": 512 + }, + { + "epoch": 0.39798293250581845, + "grad_norm": 13.001891136169434, + "learning_rate": 2.097532314923619e-05, + "loss": 0.9891, + "step": 513 + }, + { + "epoch": 0.39875872769588827, + "grad_norm": 15.258377075195312, + "learning_rate": 2.1016451233842538e-05, + "loss": 1.4632, + "step": 514 + }, + { + "epoch": 0.3995345228859581, + "grad_norm": 16.525514602661133, + "learning_rate": 2.105757931844888e-05, + "loss": 1.6116, + "step": 515 + }, + { + "epoch": 0.4003103180760279, + "grad_norm": 15.327943801879883, + "learning_rate": 2.1098707403055228e-05, + "loss": 1.1468, + "step": 516 + }, + { + "epoch": 0.4010861132660978, + "grad_norm": 13.974370002746582, + "learning_rate": 2.1139835487661573e-05, + "loss": 1.7156, + "step": 517 + }, + { + "epoch": 0.4018619084561676, + "grad_norm": 11.866250991821289, + "learning_rate": 2.1180963572267915e-05, + "loss": 1.397, + "step": 518 + }, + { + "epoch": 0.4026377036462374, + "grad_norm": 10.632682800292969, + "learning_rate": 2.1222091656874264e-05, + "loss": 0.7668, + "step": 519 + }, + { + "epoch": 0.4034134988363072, + "grad_norm": 13.295013427734375, + "learning_rate": 2.126321974148061e-05, + "loss": 0.977, + "step": 520 + }, + { + "epoch": 0.40418929402637704, + "grad_norm": 16.0167236328125, + "learning_rate": 2.1304347826086955e-05, + "loss": 1.5923, + "step": 521 + }, + { + "epoch": 0.40496508921644686, + "grad_norm": 21.75667953491211, + "learning_rate": 2.13454759106933e-05, + "loss": 3.4076, + "step": 522 + }, + { + "epoch": 0.4057408844065167, + "grad_norm": 8.793801307678223, + "learning_rate": 2.1386603995299645e-05, + "loss": 1.0647, + "step": 523 + }, + { + "epoch": 0.4065166795965865, + "grad_norm": 13.385932922363281, + "learning_rate": 2.142773207990599e-05, + "loss": 1.1351, + "step": 524 + }, + { + "epoch": 0.4072924747866563, + "grad_norm": 10.233895301818848, + "learning_rate": 2.1468860164512336e-05, + "loss": 0.9047, + "step": 525 + }, + { + "epoch": 0.40806826997672613, + "grad_norm": 8.322166442871094, + "learning_rate": 2.1509988249118684e-05, + "loss": 0.5271, + "step": 526 + }, + { + "epoch": 0.40884406516679594, + "grad_norm": 15.468615531921387, + "learning_rate": 2.1551116333725026e-05, + "loss": 1.9047, + "step": 527 + }, + { + "epoch": 0.40961986035686576, + "grad_norm": 8.270638465881348, + "learning_rate": 2.1592244418331375e-05, + "loss": 0.5474, + "step": 528 + }, + { + "epoch": 0.41039565554693563, + "grad_norm": 12.269023895263672, + "learning_rate": 2.1633372502937717e-05, + "loss": 1.0393, + "step": 529 + }, + { + "epoch": 0.41117145073700545, + "grad_norm": 9.236624717712402, + "learning_rate": 2.1674500587544062e-05, + "loss": 0.5875, + "step": 530 + }, + { + "epoch": 0.41194724592707527, + "grad_norm": 8.601607322692871, + "learning_rate": 2.171562867215041e-05, + "loss": 0.5365, + "step": 531 + }, + { + "epoch": 0.4127230411171451, + "grad_norm": 11.163597106933594, + "learning_rate": 2.1756756756756753e-05, + "loss": 1.3738, + "step": 532 + }, + { + "epoch": 0.4134988363072149, + "grad_norm": 14.983156204223633, + "learning_rate": 2.17978848413631e-05, + "loss": 1.7889, + "step": 533 + }, + { + "epoch": 0.4142746314972847, + "grad_norm": 6.830989360809326, + "learning_rate": 2.1839012925969447e-05, + "loss": 0.2338, + "step": 534 + }, + { + "epoch": 0.41505042668735453, + "grad_norm": 14.335187911987305, + "learning_rate": 2.188014101057579e-05, + "loss": 1.3742, + "step": 535 + }, + { + "epoch": 0.41582622187742435, + "grad_norm": 14.836065292358398, + "learning_rate": 2.1921269095182137e-05, + "loss": 1.7282, + "step": 536 + }, + { + "epoch": 0.41660201706749417, + "grad_norm": 27.33770751953125, + "learning_rate": 2.1962397179788483e-05, + "loss": 3.8249, + "step": 537 + }, + { + "epoch": 0.417377812257564, + "grad_norm": 11.044896125793457, + "learning_rate": 2.2003525264394828e-05, + "loss": 0.626, + "step": 538 + }, + { + "epoch": 0.4181536074476338, + "grad_norm": 20.383256912231445, + "learning_rate": 2.2044653349001173e-05, + "loss": 2.1066, + "step": 539 + }, + { + "epoch": 0.4189294026377037, + "grad_norm": 12.21434497833252, + "learning_rate": 2.2085781433607515e-05, + "loss": 0.7842, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_Qnli-dev_cosine_accuracy": 0.66796875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8641210794448853, + "eval_Qnli-dev_cosine_ap": 0.6946738251666076, + "eval_Qnli-dev_cosine_f1": 0.6687797147385103, + "eval_Qnli-dev_cosine_f1_threshold": 0.7404845952987671, + "eval_Qnli-dev_cosine_precision": 0.5341772151898734, + "eval_Qnli-dev_cosine_recall": 0.8940677966101694, + "eval_Qnli-dev_dot_accuracy": 0.630859375, + "eval_Qnli-dev_dot_accuracy_threshold": 438.849853515625, + "eval_Qnli-dev_dot_ap": 0.6369433716268977, + "eval_Qnli-dev_dot_f1": 0.660436137071651, + "eval_Qnli-dev_dot_f1_threshold": 361.13909912109375, + "eval_Qnli-dev_dot_precision": 0.5221674876847291, + "eval_Qnli-dev_dot_recall": 0.8983050847457628, + "eval_Qnli-dev_euclidean_accuracy": 0.66015625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.017486572265625, + "eval_Qnli-dev_euclidean_ap": 0.687509829527032, + "eval_Qnli-dev_euclidean_f1": 0.6689895470383275, + "eval_Qnli-dev_euclidean_f1_threshold": 14.992185592651367, + "eval_Qnli-dev_euclidean_precision": 0.5680473372781065, + "eval_Qnli-dev_euclidean_recall": 0.8135593220338984, + "eval_Qnli-dev_manhattan_accuracy": 0.6640625, + "eval_Qnli-dev_manhattan_accuracy_threshold": 239.63194274902344, + "eval_Qnli-dev_manhattan_ap": 0.6895202032022496, + "eval_Qnli-dev_manhattan_f1": 0.6687797147385103, + "eval_Qnli-dev_manhattan_f1_threshold": 292.85833740234375, + "eval_Qnli-dev_manhattan_precision": 0.5341772151898734, + "eval_Qnli-dev_manhattan_recall": 0.8940677966101694, + "eval_Qnli-dev_max_accuracy": 0.66796875, + "eval_Qnli-dev_max_accuracy_threshold": 438.849853515625, + "eval_Qnli-dev_max_ap": 0.6946738251666076, + "eval_Qnli-dev_max_f1": 0.6689895470383275, + "eval_Qnli-dev_max_f1_threshold": 361.13909912109375, + "eval_Qnli-dev_max_precision": 0.5680473372781065, + "eval_Qnli-dev_max_recall": 0.8983050847457628, + "eval_allNLI-dev_cosine_accuracy": 0.708984375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8857862949371338, + "eval_allNLI-dev_cosine_ap": 0.5764885273482621, + "eval_allNLI-dev_cosine_f1": 0.6118721461187215, + "eval_allNLI-dev_cosine_f1_threshold": 0.826934814453125, + "eval_allNLI-dev_cosine_precision": 0.5056603773584906, + "eval_allNLI-dev_cosine_recall": 0.7745664739884393, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 423.199951171875, + "eval_allNLI-dev_dot_ap": 0.47407820835978787, + "eval_allNLI-dev_dot_f1": 0.5382830626450116, + "eval_allNLI-dev_dot_f1_threshold": 391.4878845214844, + "eval_allNLI-dev_dot_precision": 0.4496124031007752, + "eval_allNLI-dev_dot_recall": 0.6705202312138728, + "eval_allNLI-dev_euclidean_accuracy": 0.7109375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.394439697265625, + "eval_allNLI-dev_euclidean_ap": 0.5776242498946831, + "eval_allNLI-dev_euclidean_f1": 0.6112359550561798, + "eval_allNLI-dev_euclidean_f1_threshold": 12.974320411682129, + "eval_allNLI-dev_euclidean_precision": 0.5, + "eval_allNLI-dev_euclidean_recall": 0.7861271676300579, + "eval_allNLI-dev_manhattan_accuracy": 0.71484375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 186.4627685546875, + "eval_allNLI-dev_manhattan_ap": 0.5744859968569902, + "eval_allNLI-dev_manhattan_f1": 0.6153846153846154, + "eval_allNLI-dev_manhattan_f1_threshold": 228.9261016845703, + "eval_allNLI-dev_manhattan_precision": 0.515625, + "eval_allNLI-dev_manhattan_recall": 0.7630057803468208, + "eval_allNLI-dev_max_accuracy": 0.71484375, + "eval_allNLI-dev_max_accuracy_threshold": 423.199951171875, + "eval_allNLI-dev_max_ap": 0.5776242498946831, + "eval_allNLI-dev_max_f1": 0.6153846153846154, + "eval_allNLI-dev_max_f1_threshold": 391.4878845214844, + "eval_allNLI-dev_max_precision": 0.515625, + "eval_allNLI-dev_max_recall": 0.7861271676300579, + "eval_sequential_score": 0.6946738251666076, + "eval_sts-test_pearson_cosine": 0.7900763694225302, + "eval_sts-test_pearson_dot": 0.7154128969065976, + "eval_sts-test_pearson_euclidean": 0.8202843825511696, + "eval_sts-test_pearson_manhattan": 0.8207854898748433, + "eval_sts-test_pearson_max": 0.8207854898748433, + "eval_sts-test_spearman_cosine": 0.8264226181452482, + "eval_sts-test_spearman_dot": 0.7049591472380641, + "eval_sts-test_spearman_euclidean": 0.8181576419326547, + "eval_sts-test_spearman_manhattan": 0.8190534279319264, + "eval_sts-test_spearman_max": 0.8264226181452482, + "eval_vitaminc-pairs_loss": 3.8354361057281494, + "eval_vitaminc-pairs_runtime": 5.4878, + "eval_vitaminc-pairs_samples_per_second": 23.325, + "eval_vitaminc-pairs_steps_per_second": 0.364, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_negation-triplets_loss": 1.7570998668670654, + "eval_negation-triplets_runtime": 1.1247, + "eval_negation-triplets_samples_per_second": 113.81, + "eval_negation-triplets_steps_per_second": 1.778, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_scitail-pairs-pos_loss": 0.22965900599956512, + "eval_scitail-pairs-pos_runtime": 1.2444, + "eval_scitail-pairs-pos_samples_per_second": 102.863, + "eval_scitail-pairs-pos_steps_per_second": 1.607, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_scitail-pairs-qa_loss": 0.038437146693468094, + "eval_scitail-pairs-qa_runtime": 0.9089, + "eval_scitail-pairs-qa_samples_per_second": 140.826, + "eval_scitail-pairs-qa_steps_per_second": 2.2, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_xsum-pairs_loss": 0.5190384387969971, + "eval_xsum-pairs_runtime": 4.3124, + "eval_xsum-pairs_samples_per_second": 29.682, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_sciq_pairs_loss": 0.10552895069122314, + "eval_sciq_pairs_runtime": 5.5283, + "eval_sciq_pairs_samples_per_second": 23.154, + "eval_sciq_pairs_steps_per_second": 0.362, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_qasc_pairs_loss": 0.78385329246521, + "eval_qasc_pairs_runtime": 0.9741, + "eval_qasc_pairs_samples_per_second": 131.41, + "eval_qasc_pairs_steps_per_second": 2.053, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_openbookqa_pairs_loss": 1.882456660270691, + "eval_openbookqa_pairs_runtime": 0.864, + "eval_openbookqa_pairs_samples_per_second": 148.143, + "eval_openbookqa_pairs_steps_per_second": 2.315, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_msmarco_pairs_loss": 1.2333143949508667, + "eval_msmarco_pairs_runtime": 1.9596, + "eval_msmarco_pairs_samples_per_second": 65.32, + "eval_msmarco_pairs_steps_per_second": 1.021, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_nq_pairs_loss": 1.5782138109207153, + "eval_nq_pairs_runtime": 4.4034, + "eval_nq_pairs_samples_per_second": 29.068, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_trivia_pairs_loss": 1.1567370891571045, + "eval_trivia_pairs_runtime": 4.5196, + "eval_trivia_pairs_samples_per_second": 28.321, + "eval_trivia_pairs_steps_per_second": 0.443, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_gooaq_pairs_loss": 1.1066967248916626, + "eval_gooaq_pairs_runtime": 1.268, + "eval_gooaq_pairs_samples_per_second": 100.949, + "eval_gooaq_pairs_steps_per_second": 1.577, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_paws-pos_loss": 0.05742817744612694, + "eval_paws-pos_runtime": 1.025, + "eval_paws-pos_samples_per_second": 124.882, + "eval_paws-pos_steps_per_second": 1.951, + "step": 540 + }, + { + "epoch": 0.4189294026377037, + "eval_global_dataset_loss": 0.8099284768104553, + "eval_global_dataset_runtime": 21.1729, + "eval_global_dataset_samples_per_second": 19.648, + "eval_global_dataset_steps_per_second": 0.331, + "step": 540 + }, + { + "epoch": 0.4197051978277735, + "grad_norm": 11.069849014282227, + "learning_rate": 2.2126909518213864e-05, + "loss": 0.5988, + "step": 541 + }, + { + "epoch": 0.4204809930178433, + "grad_norm": 1.716650128364563, + "learning_rate": 2.216803760282021e-05, + "loss": 0.091, + "step": 542 + }, + { + "epoch": 0.4212567882079131, + "grad_norm": 6.249168872833252, + "learning_rate": 2.2209165687426558e-05, + "loss": 0.2432, + "step": 543 + }, + { + "epoch": 0.42203258339798294, + "grad_norm": 11.938614845275879, + "learning_rate": 2.22502937720329e-05, + "loss": 0.7155, + "step": 544 + }, + { + "epoch": 0.42280837858805276, + "grad_norm": 17.08501625061035, + "learning_rate": 2.229142185663925e-05, + "loss": 1.8321, + "step": 545 + }, + { + "epoch": 0.4235841737781226, + "grad_norm": 8.503474235534668, + "learning_rate": 2.233254994124559e-05, + "loss": 0.5349, + "step": 546 + }, + { + "epoch": 0.4243599689681924, + "grad_norm": 11.074429512023926, + "learning_rate": 2.2373678025851936e-05, + "loss": 0.5962, + "step": 547 + }, + { + "epoch": 0.4251357641582622, + "grad_norm": 12.107995986938477, + "learning_rate": 2.2414806110458284e-05, + "loss": 0.9328, + "step": 548 + }, + { + "epoch": 0.425911559348332, + "grad_norm": 14.911764144897461, + "learning_rate": 2.2455934195064626e-05, + "loss": 1.491, + "step": 549 + }, + { + "epoch": 0.42668735453840184, + "grad_norm": 12.876277923583984, + "learning_rate": 2.2497062279670975e-05, + "loss": 1.4732, + "step": 550 + }, + { + "epoch": 0.42746314972847166, + "grad_norm": 12.240724563598633, + "learning_rate": 2.253819036427732e-05, + "loss": 1.3947, + "step": 551 + }, + { + "epoch": 0.42823894491854153, + "grad_norm": 10.111100196838379, + "learning_rate": 2.2579318448883662e-05, + "loss": 0.8615, + "step": 552 + }, + { + "epoch": 0.42901474010861135, + "grad_norm": 15.11451530456543, + "learning_rate": 2.262044653349001e-05, + "loss": 1.4688, + "step": 553 + }, + { + "epoch": 0.42979053529868116, + "grad_norm": 15.309952735900879, + "learning_rate": 2.2661574618096353e-05, + "loss": 1.794, + "step": 554 + }, + { + "epoch": 0.430566330488751, + "grad_norm": 11.084304809570312, + "learning_rate": 2.27027027027027e-05, + "loss": 0.7028, + "step": 555 + }, + { + "epoch": 0.4313421256788208, + "grad_norm": 10.565820693969727, + "learning_rate": 2.2743830787309047e-05, + "loss": 0.7697, + "step": 556 + }, + { + "epoch": 0.4321179208688906, + "grad_norm": 10.25606918334961, + "learning_rate": 2.2784958871915395e-05, + "loss": 1.019, + "step": 557 + }, + { + "epoch": 0.43289371605896043, + "grad_norm": 9.988451957702637, + "learning_rate": 2.2826086956521737e-05, + "loss": 0.7246, + "step": 558 + }, + { + "epoch": 0.43366951124903025, + "grad_norm": 7.850244522094727, + "learning_rate": 2.2867215041128082e-05, + "loss": 0.5204, + "step": 559 + }, + { + "epoch": 0.43444530643910007, + "grad_norm": 12.435367584228516, + "learning_rate": 2.2908343125734428e-05, + "loss": 1.4801, + "step": 560 + }, + { + "epoch": 0.4352211016291699, + "grad_norm": 9.760563850402832, + "learning_rate": 2.2949471210340773e-05, + "loss": 0.7505, + "step": 561 + }, + { + "epoch": 0.4359968968192397, + "grad_norm": 9.784625053405762, + "learning_rate": 2.299059929494712e-05, + "loss": 0.657, + "step": 562 + }, + { + "epoch": 0.4367726920093095, + "grad_norm": 1.8895553350448608, + "learning_rate": 2.3031727379553464e-05, + "loss": 0.0683, + "step": 563 + }, + { + "epoch": 0.4375484871993794, + "grad_norm": 8.71015453338623, + "learning_rate": 2.307285546415981e-05, + "loss": 0.4504, + "step": 564 + }, + { + "epoch": 0.4383242823894492, + "grad_norm": 2.3740286827087402, + "learning_rate": 2.3113983548766158e-05, + "loss": 0.1295, + "step": 565 + }, + { + "epoch": 0.439100077579519, + "grad_norm": 10.439929962158203, + "learning_rate": 2.31551116333725e-05, + "loss": 0.6033, + "step": 566 + }, + { + "epoch": 0.43987587276958884, + "grad_norm": 13.831979751586914, + "learning_rate": 2.3196239717978848e-05, + "loss": 0.9079, + "step": 567 + }, + { + "epoch": 0.44065166795965865, + "grad_norm": 10.219809532165527, + "learning_rate": 2.323736780258519e-05, + "loss": 0.5326, + "step": 568 + }, + { + "epoch": 0.44142746314972847, + "grad_norm": 15.038511276245117, + "learning_rate": 2.3278495887191535e-05, + "loss": 1.2312, + "step": 569 + }, + { + "epoch": 0.4422032583397983, + "grad_norm": 24.031938552856445, + "learning_rate": 2.3319623971797884e-05, + "loss": 2.5394, + "step": 570 + }, + { + "epoch": 0.4429790535298681, + "grad_norm": 4.930804252624512, + "learning_rate": 2.3360752056404226e-05, + "loss": 0.1594, + "step": 571 + }, + { + "epoch": 0.4437548487199379, + "grad_norm": 4.852351665496826, + "learning_rate": 2.3401880141010575e-05, + "loss": 0.1987, + "step": 572 + }, + { + "epoch": 0.44453064391000774, + "grad_norm": 14.101249694824219, + "learning_rate": 2.344300822561692e-05, + "loss": 1.8328, + "step": 573 + }, + { + "epoch": 0.44530643910007756, + "grad_norm": 24.093111038208008, + "learning_rate": 2.3484136310223265e-05, + "loss": 2.5802, + "step": 574 + }, + { + "epoch": 0.44608223429014743, + "grad_norm": 10.027242660522461, + "learning_rate": 2.352526439482961e-05, + "loss": 0.9177, + "step": 575 + }, + { + "epoch": 0.44685802948021724, + "grad_norm": 11.803500175476074, + "learning_rate": 2.3566392479435956e-05, + "loss": 0.6669, + "step": 576 + }, + { + "epoch": 0.44763382467028706, + "grad_norm": 18.34312629699707, + "learning_rate": 2.36075205640423e-05, + "loss": 1.4284, + "step": 577 + }, + { + "epoch": 0.4484096198603569, + "grad_norm": 13.121169090270996, + "learning_rate": 2.3648648648648646e-05, + "loss": 0.8634, + "step": 578 + }, + { + "epoch": 0.4491854150504267, + "grad_norm": 19.3350772857666, + "learning_rate": 2.3689776733254995e-05, + "loss": 1.9154, + "step": 579 + }, + { + "epoch": 0.4499612102404965, + "grad_norm": 13.133557319641113, + "learning_rate": 2.3730904817861337e-05, + "loss": 0.803, + "step": 580 + }, + { + "epoch": 0.45073700543056633, + "grad_norm": 12.209722518920898, + "learning_rate": 2.3772032902467682e-05, + "loss": 0.881, + "step": 581 + }, + { + "epoch": 0.45151280062063615, + "grad_norm": 14.424088478088379, + "learning_rate": 2.381316098707403e-05, + "loss": 1.2738, + "step": 582 + }, + { + "epoch": 0.45228859581070596, + "grad_norm": 5.458702087402344, + "learning_rate": 2.3854289071680373e-05, + "loss": 0.2015, + "step": 583 + }, + { + "epoch": 0.4530643910007758, + "grad_norm": 10.050310134887695, + "learning_rate": 2.389541715628672e-05, + "loss": 0.7598, + "step": 584 + }, + { + "epoch": 0.4538401861908456, + "grad_norm": 9.868513107299805, + "learning_rate": 2.3936545240893063e-05, + "loss": 0.712, + "step": 585 + }, + { + "epoch": 0.4546159813809154, + "grad_norm": 13.284599304199219, + "learning_rate": 2.397767332549941e-05, + "loss": 1.4037, + "step": 586 + }, + { + "epoch": 0.4553917765709853, + "grad_norm": 9.94361400604248, + "learning_rate": 2.4018801410105757e-05, + "loss": 0.6935, + "step": 587 + }, + { + "epoch": 0.4561675717610551, + "grad_norm": 9.887158393859863, + "learning_rate": 2.40599294947121e-05, + "loss": 0.5348, + "step": 588 + }, + { + "epoch": 0.4569433669511249, + "grad_norm": 14.570903778076172, + "learning_rate": 2.4101057579318448e-05, + "loss": 1.6486, + "step": 589 + }, + { + "epoch": 0.45771916214119474, + "grad_norm": 19.21129608154297, + "learning_rate": 2.4142185663924793e-05, + "loss": 3.1141, + "step": 590 + }, + { + "epoch": 0.45849495733126455, + "grad_norm": 13.141075134277344, + "learning_rate": 2.418331374853114e-05, + "loss": 1.1971, + "step": 591 + }, + { + "epoch": 0.45927075252133437, + "grad_norm": 4.923962116241455, + "learning_rate": 2.4224441833137484e-05, + "loss": 0.1634, + "step": 592 + }, + { + "epoch": 0.4600465477114042, + "grad_norm": 18.966079711914062, + "learning_rate": 2.4265569917743826e-05, + "loss": 1.6485, + "step": 593 + }, + { + "epoch": 0.460822342901474, + "grad_norm": 11.013054847717285, + "learning_rate": 2.4306698002350174e-05, + "loss": 0.7393, + "step": 594 + }, + { + "epoch": 0.4615981380915438, + "grad_norm": 9.847901344299316, + "learning_rate": 2.434782608695652e-05, + "loss": 0.6824, + "step": 595 + }, + { + "epoch": 0.46237393328161364, + "grad_norm": 13.051827430725098, + "learning_rate": 2.4388954171562868e-05, + "loss": 1.4778, + "step": 596 + }, + { + "epoch": 0.46314972847168345, + "grad_norm": 11.05080509185791, + "learning_rate": 2.443008225616921e-05, + "loss": 0.7145, + "step": 597 + }, + { + "epoch": 0.46392552366175327, + "grad_norm": 12.709424018859863, + "learning_rate": 2.4471210340775555e-05, + "loss": 1.2698, + "step": 598 + }, + { + "epoch": 0.46470131885182314, + "grad_norm": 20.696352005004883, + "learning_rate": 2.45123384253819e-05, + "loss": 3.0779, + "step": 599 + }, + { + "epoch": 0.46547711404189296, + "grad_norm": 8.470314025878906, + "learning_rate": 2.4553466509988246e-05, + "loss": 0.3882, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_Qnli-dev_cosine_accuracy": 0.685546875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8817452192306519, + "eval_Qnli-dev_cosine_ap": 0.7173076918297527, + "eval_Qnli-dev_cosine_f1": 0.682142857142857, + "eval_Qnli-dev_cosine_f1_threshold": 0.8134192228317261, + "eval_Qnli-dev_cosine_precision": 0.5895061728395061, + "eval_Qnli-dev_cosine_recall": 0.809322033898305, + "eval_Qnli-dev_dot_accuracy": 0.6640625, + "eval_Qnli-dev_dot_accuracy_threshold": 456.8737487792969, + "eval_Qnli-dev_dot_ap": 0.620264943820585, + "eval_Qnli-dev_dot_f1": 0.6573643410852713, + "eval_Qnli-dev_dot_f1_threshold": 368.4569091796875, + "eval_Qnli-dev_dot_precision": 0.5183374083129584, + "eval_Qnli-dev_dot_recall": 0.8983050847457628, + "eval_Qnli-dev_euclidean_accuracy": 0.681640625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.752117156982422, + "eval_Qnli-dev_euclidean_ap": 0.7122592322530366, + "eval_Qnli-dev_euclidean_f1": 0.6925675675675677, + "eval_Qnli-dev_euclidean_f1_threshold": 14.497557640075684, + "eval_Qnli-dev_euclidean_precision": 0.5758426966292135, + "eval_Qnli-dev_euclidean_recall": 0.8686440677966102, + "eval_Qnli-dev_manhattan_accuracy": 0.677734375, + "eval_Qnli-dev_manhattan_accuracy_threshold": 211.67117309570312, + "eval_Qnli-dev_manhattan_ap": 0.7070467284308819, + "eval_Qnli-dev_manhattan_f1": 0.6795774647887325, + "eval_Qnli-dev_manhattan_f1_threshold": 258.075927734375, + "eval_Qnli-dev_manhattan_precision": 0.5813253012048193, + "eval_Qnli-dev_manhattan_recall": 0.8177966101694916, + "eval_Qnli-dev_max_accuracy": 0.685546875, + "eval_Qnli-dev_max_accuracy_threshold": 456.8737487792969, + "eval_Qnli-dev_max_ap": 0.7173076918297527, + "eval_Qnli-dev_max_f1": 0.6925675675675677, + "eval_Qnli-dev_max_f1_threshold": 368.4569091796875, + "eval_Qnli-dev_max_precision": 0.5895061728395061, + "eval_Qnli-dev_max_recall": 0.8983050847457628, + "eval_allNLI-dev_cosine_accuracy": 0.71484375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8874993324279785, + "eval_allNLI-dev_cosine_ap": 0.5763030708533948, + "eval_allNLI-dev_cosine_f1": 0.5936739659367396, + "eval_allNLI-dev_cosine_f1_threshold": 0.8460882902145386, + "eval_allNLI-dev_cosine_precision": 0.5126050420168067, + "eval_allNLI-dev_cosine_recall": 0.7052023121387283, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 481.656005859375, + "eval_allNLI-dev_dot_ap": 0.47418213859749314, + "eval_allNLI-dev_dot_f1": 0.5446428571428572, + "eval_allNLI-dev_dot_f1_threshold": 400.9698486328125, + "eval_allNLI-dev_dot_precision": 0.44363636363636366, + "eval_allNLI-dev_dot_recall": 0.7052023121387283, + "eval_allNLI-dev_euclidean_accuracy": 0.71484375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 11.067342758178711, + "eval_allNLI-dev_euclidean_ap": 0.575001939972708, + "eval_allNLI-dev_euclidean_f1": 0.5948275862068965, + "eval_allNLI-dev_euclidean_f1_threshold": 13.239274978637695, + "eval_allNLI-dev_euclidean_precision": 0.4742268041237113, + "eval_allNLI-dev_euclidean_recall": 0.7976878612716763, + "eval_allNLI-dev_manhattan_accuracy": 0.708984375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 188.52987670898438, + "eval_allNLI-dev_manhattan_ap": 0.5705463278324929, + "eval_allNLI-dev_manhattan_f1": 0.6053811659192826, + "eval_allNLI-dev_manhattan_f1_threshold": 231.40406799316406, + "eval_allNLI-dev_manhattan_precision": 0.4945054945054945, + "eval_allNLI-dev_manhattan_recall": 0.7803468208092486, + "eval_allNLI-dev_max_accuracy": 0.71484375, + "eval_allNLI-dev_max_accuracy_threshold": 481.656005859375, + "eval_allNLI-dev_max_ap": 0.5763030708533948, + "eval_allNLI-dev_max_f1": 0.6053811659192826, + "eval_allNLI-dev_max_f1_threshold": 400.9698486328125, + "eval_allNLI-dev_max_precision": 0.5126050420168067, + "eval_allNLI-dev_max_recall": 0.7976878612716763, + "eval_sequential_score": 0.7173076918297527, + "eval_sts-test_pearson_cosine": 0.788791334754521, + "eval_sts-test_pearson_dot": 0.7181977195216988, + "eval_sts-test_pearson_euclidean": 0.806798892548376, + "eval_sts-test_pearson_manhattan": 0.8023132648366168, + "eval_sts-test_pearson_max": 0.806798892548376, + "eval_sts-test_spearman_cosine": 0.8036047554844786, + "eval_sts-test_spearman_dot": 0.7062252051740883, + "eval_sts-test_spearman_euclidean": 0.7952549756315485, + "eval_sts-test_spearman_manhattan": 0.7927283194133901, + "eval_sts-test_spearman_max": 0.8036047554844786, + "eval_vitaminc-pairs_loss": 4.26703405380249, + "eval_vitaminc-pairs_runtime": 5.49, + "eval_vitaminc-pairs_samples_per_second": 23.315, + "eval_vitaminc-pairs_steps_per_second": 0.364, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_negation-triplets_loss": 1.652132272720337, + "eval_negation-triplets_runtime": 1.1099, + "eval_negation-triplets_samples_per_second": 115.324, + "eval_negation-triplets_steps_per_second": 1.802, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_scitail-pairs-pos_loss": 0.1807655543088913, + "eval_scitail-pairs-pos_runtime": 1.2321, + "eval_scitail-pairs-pos_samples_per_second": 103.885, + "eval_scitail-pairs-pos_steps_per_second": 1.623, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_scitail-pairs-qa_loss": 0.03043425641953945, + "eval_scitail-pairs-qa_runtime": 0.9169, + "eval_scitail-pairs-qa_samples_per_second": 139.594, + "eval_scitail-pairs-qa_steps_per_second": 2.181, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_xsum-pairs_loss": 0.4298955500125885, + "eval_xsum-pairs_runtime": 4.3135, + "eval_xsum-pairs_samples_per_second": 29.674, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_sciq_pairs_loss": 0.09356379508972168, + "eval_sciq_pairs_runtime": 5.5403, + "eval_sciq_pairs_samples_per_second": 23.103, + "eval_sciq_pairs_steps_per_second": 0.361, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_qasc_pairs_loss": 0.671562135219574, + "eval_qasc_pairs_runtime": 0.9763, + "eval_qasc_pairs_samples_per_second": 131.11, + "eval_qasc_pairs_steps_per_second": 2.049, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_openbookqa_pairs_loss": 1.7955065965652466, + "eval_openbookqa_pairs_runtime": 0.8595, + "eval_openbookqa_pairs_samples_per_second": 148.928, + "eval_openbookqa_pairs_steps_per_second": 2.327, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_msmarco_pairs_loss": 1.1523165702819824, + "eval_msmarco_pairs_runtime": 1.9624, + "eval_msmarco_pairs_samples_per_second": 65.227, + "eval_msmarco_pairs_steps_per_second": 1.019, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_nq_pairs_loss": 1.4328314065933228, + "eval_nq_pairs_runtime": 4.4012, + "eval_nq_pairs_samples_per_second": 29.083, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_trivia_pairs_loss": 1.181535243988037, + "eval_trivia_pairs_runtime": 4.517, + "eval_trivia_pairs_samples_per_second": 28.337, + "eval_trivia_pairs_steps_per_second": 0.443, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_gooaq_pairs_loss": 0.974396288394928, + "eval_gooaq_pairs_runtime": 1.3004, + "eval_gooaq_pairs_samples_per_second": 98.435, + "eval_gooaq_pairs_steps_per_second": 1.538, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_paws-pos_loss": 0.04967149719595909, + "eval_paws-pos_runtime": 1.0256, + "eval_paws-pos_samples_per_second": 124.809, + "eval_paws-pos_steps_per_second": 1.95, + "step": 600 + }, + { + "epoch": 0.46547711404189296, + "eval_global_dataset_loss": 0.7299151420593262, + "eval_global_dataset_runtime": 21.1401, + "eval_global_dataset_samples_per_second": 19.678, + "eval_global_dataset_steps_per_second": 0.331, + "step": 600 + }, + { + "epoch": 0.4662529092319628, + "grad_norm": 15.569989204406738, + "learning_rate": 2.4594594594594595e-05, + "loss": 1.3776, + "step": 601 + }, + { + "epoch": 0.4670287044220326, + "grad_norm": 8.396140098571777, + "learning_rate": 2.4635722679200937e-05, + "loss": 0.4241, + "step": 602 + }, + { + "epoch": 0.4678044996121024, + "grad_norm": 13.026421546936035, + "learning_rate": 2.4676850763807282e-05, + "loss": 1.2092, + "step": 603 + }, + { + "epoch": 0.4685802948021722, + "grad_norm": 7.66565465927124, + "learning_rate": 2.471797884841363e-05, + "loss": 0.3678, + "step": 604 + }, + { + "epoch": 0.46935608999224204, + "grad_norm": 8.589688301086426, + "learning_rate": 2.4759106933019972e-05, + "loss": 0.5244, + "step": 605 + }, + { + "epoch": 0.47013188518231186, + "grad_norm": 9.045538902282715, + "learning_rate": 2.480023501762632e-05, + "loss": 0.7051, + "step": 606 + }, + { + "epoch": 0.4709076803723817, + "grad_norm": 14.687758445739746, + "learning_rate": 2.4841363102232666e-05, + "loss": 1.6367, + "step": 607 + }, + { + "epoch": 0.4716834755624515, + "grad_norm": 1.2971081733703613, + "learning_rate": 2.4882491186839012e-05, + "loss": 0.0546, + "step": 608 + }, + { + "epoch": 0.4724592707525213, + "grad_norm": 10.166410446166992, + "learning_rate": 2.4923619271445357e-05, + "loss": 0.566, + "step": 609 + }, + { + "epoch": 0.4732350659425912, + "grad_norm": 3.489497661590576, + "learning_rate": 2.49647473560517e-05, + "loss": 0.1035, + "step": 610 + }, + { + "epoch": 0.474010861132661, + "grad_norm": 13.205260276794434, + "learning_rate": 2.5005875440658048e-05, + "loss": 1.5533, + "step": 611 + }, + { + "epoch": 0.4747866563227308, + "grad_norm": 12.985085487365723, + "learning_rate": 2.5047003525264393e-05, + "loss": 0.6382, + "step": 612 + }, + { + "epoch": 0.47556245151280063, + "grad_norm": 9.48460865020752, + "learning_rate": 2.5088131609870738e-05, + "loss": 0.6183, + "step": 613 + }, + { + "epoch": 0.47633824670287045, + "grad_norm": 11.595215797424316, + "learning_rate": 2.5129259694477083e-05, + "loss": 1.2768, + "step": 614 + }, + { + "epoch": 0.47711404189294027, + "grad_norm": 10.382658958435059, + "learning_rate": 2.517038777908343e-05, + "loss": 0.4508, + "step": 615 + }, + { + "epoch": 0.4778898370830101, + "grad_norm": 14.347338676452637, + "learning_rate": 2.5211515863689774e-05, + "loss": 0.7905, + "step": 616 + }, + { + "epoch": 0.4786656322730799, + "grad_norm": 16.12579345703125, + "learning_rate": 2.525264394829612e-05, + "loss": 1.0066, + "step": 617 + }, + { + "epoch": 0.4794414274631497, + "grad_norm": 10.241353988647461, + "learning_rate": 2.5293772032902468e-05, + "loss": 0.4964, + "step": 618 + }, + { + "epoch": 0.48021722265321953, + "grad_norm": 13.637825965881348, + "learning_rate": 2.533490011750881e-05, + "loss": 1.7319, + "step": 619 + }, + { + "epoch": 0.48099301784328935, + "grad_norm": 15.750479698181152, + "learning_rate": 2.5376028202115155e-05, + "loss": 1.6527, + "step": 620 + }, + { + "epoch": 0.48176881303335917, + "grad_norm": 1.8672064542770386, + "learning_rate": 2.5417156286721504e-05, + "loss": 0.019, + "step": 621 + }, + { + "epoch": 0.48254460822342904, + "grad_norm": 11.430908203125, + "learning_rate": 2.5458284371327846e-05, + "loss": 0.5658, + "step": 622 + }, + { + "epoch": 0.48332040341349886, + "grad_norm": 14.80313491821289, + "learning_rate": 2.5499412455934194e-05, + "loss": 0.9891, + "step": 623 + }, + { + "epoch": 0.4840961986035687, + "grad_norm": 10.941021919250488, + "learning_rate": 2.5540540540540536e-05, + "loss": 0.6243, + "step": 624 + }, + { + "epoch": 0.4848719937936385, + "grad_norm": 1.0070419311523438, + "learning_rate": 2.5581668625146885e-05, + "loss": 0.0219, + "step": 625 + }, + { + "epoch": 0.4856477889837083, + "grad_norm": 9.442733764648438, + "learning_rate": 2.562279670975323e-05, + "loss": 0.3771, + "step": 626 + }, + { + "epoch": 0.4864235841737781, + "grad_norm": 9.765592575073242, + "learning_rate": 2.5663924794359572e-05, + "loss": 0.5282, + "step": 627 + }, + { + "epoch": 0.48719937936384794, + "grad_norm": 14.095426559448242, + "learning_rate": 2.570505287896592e-05, + "loss": 1.5618, + "step": 628 + }, + { + "epoch": 0.48797517455391776, + "grad_norm": 11.686604499816895, + "learning_rate": 2.5746180963572266e-05, + "loss": 0.7537, + "step": 629 + }, + { + "epoch": 0.4887509697439876, + "grad_norm": 10.68002986907959, + "learning_rate": 2.578730904817861e-05, + "loss": 0.5743, + "step": 630 + }, + { + "epoch": 0.4895267649340574, + "grad_norm": 17.109416961669922, + "learning_rate": 2.5828437132784957e-05, + "loss": 1.431, + "step": 631 + }, + { + "epoch": 0.4903025601241272, + "grad_norm": 8.746842384338379, + "learning_rate": 2.5869565217391302e-05, + "loss": 0.4691, + "step": 632 + }, + { + "epoch": 0.4910783553141971, + "grad_norm": 8.443817138671875, + "learning_rate": 2.5910693301997647e-05, + "loss": 0.4524, + "step": 633 + }, + { + "epoch": 0.4918541505042669, + "grad_norm": 8.999419212341309, + "learning_rate": 2.5951821386603993e-05, + "loss": 0.35, + "step": 634 + }, + { + "epoch": 0.4926299456943367, + "grad_norm": 14.673932075500488, + "learning_rate": 2.599294947121034e-05, + "loss": 1.4527, + "step": 635 + }, + { + "epoch": 0.49340574088440653, + "grad_norm": 3.8801417350769043, + "learning_rate": 2.6034077555816683e-05, + "loss": 0.098, + "step": 636 + }, + { + "epoch": 0.49418153607447635, + "grad_norm": 4.879674434661865, + "learning_rate": 2.607520564042303e-05, + "loss": 0.1671, + "step": 637 + }, + { + "epoch": 0.49495733126454616, + "grad_norm": 1.2140374183654785, + "learning_rate": 2.6116333725029374e-05, + "loss": 0.0231, + "step": 638 + }, + { + "epoch": 0.495733126454616, + "grad_norm": 11.467575073242188, + "learning_rate": 2.615746180963572e-05, + "loss": 0.6049, + "step": 639 + }, + { + "epoch": 0.4965089216446858, + "grad_norm": 20.648147583007812, + "learning_rate": 2.6198589894242068e-05, + "loss": 1.5134, + "step": 640 + }, + { + "epoch": 0.4972847168347556, + "grad_norm": 18.041160583496094, + "learning_rate": 2.623971797884841e-05, + "loss": 1.3231, + "step": 641 + }, + { + "epoch": 0.49806051202482543, + "grad_norm": 10.526224136352539, + "learning_rate": 2.628084606345476e-05, + "loss": 0.6381, + "step": 642 + }, + { + "epoch": 0.49883630721489525, + "grad_norm": 10.155817985534668, + "learning_rate": 2.6321974148061104e-05, + "loss": 0.614, + "step": 643 + }, + { + "epoch": 0.49961210240496506, + "grad_norm": 15.734105110168457, + "learning_rate": 2.6363102232667446e-05, + "loss": 1.2059, + "step": 644 + }, + { + "epoch": 0.5003878975950349, + "grad_norm": 14.400376319885254, + "learning_rate": 2.6404230317273794e-05, + "loss": 1.1081, + "step": 645 + }, + { + "epoch": 0.5011636927851048, + "grad_norm": 10.921950340270996, + "learning_rate": 2.644535840188014e-05, + "loss": 0.5197, + "step": 646 + }, + { + "epoch": 0.5019394879751745, + "grad_norm": 8.452954292297363, + "learning_rate": 2.6486486486486485e-05, + "loss": 0.5318, + "step": 647 + }, + { + "epoch": 0.5027152831652444, + "grad_norm": 14.305099487304688, + "learning_rate": 2.652761457109283e-05, + "loss": 1.0322, + "step": 648 + }, + { + "epoch": 0.5034910783553141, + "grad_norm": 11.340603828430176, + "learning_rate": 2.6568742655699172e-05, + "loss": 1.1135, + "step": 649 + }, + { + "epoch": 0.504266873545384, + "grad_norm": 10.05394458770752, + "learning_rate": 2.660987074030552e-05, + "loss": 0.611, + "step": 650 + }, + { + "epoch": 0.5050426687354539, + "grad_norm": 9.249288558959961, + "learning_rate": 2.6650998824911866e-05, + "loss": 0.4931, + "step": 651 + }, + { + "epoch": 0.5058184639255237, + "grad_norm": 8.271453857421875, + "learning_rate": 2.669212690951821e-05, + "loss": 0.4262, + "step": 652 + }, + { + "epoch": 0.5065942591155935, + "grad_norm": 11.393213272094727, + "learning_rate": 2.6733254994124557e-05, + "loss": 0.8081, + "step": 653 + }, + { + "epoch": 0.5073700543056633, + "grad_norm": 13.045758247375488, + "learning_rate": 2.6774383078730905e-05, + "loss": 1.2332, + "step": 654 + }, + { + "epoch": 0.5081458494957332, + "grad_norm": 15.20844554901123, + "learning_rate": 2.6815511163337247e-05, + "loss": 1.3815, + "step": 655 + }, + { + "epoch": 0.5089216446858029, + "grad_norm": 8.675436973571777, + "learning_rate": 2.6856639247943592e-05, + "loss": 0.4555, + "step": 656 + }, + { + "epoch": 0.5096974398758728, + "grad_norm": 14.078372955322266, + "learning_rate": 2.689776733254994e-05, + "loss": 1.4717, + "step": 657 + }, + { + "epoch": 0.5104732350659426, + "grad_norm": 9.217131614685059, + "learning_rate": 2.6938895417156283e-05, + "loss": 0.6629, + "step": 658 + }, + { + "epoch": 0.5112490302560124, + "grad_norm": 7.337960720062256, + "learning_rate": 2.698002350176263e-05, + "loss": 0.2861, + "step": 659 + }, + { + "epoch": 0.5120248254460822, + "grad_norm": 8.492805480957031, + "learning_rate": 2.7021151586368977e-05, + "loss": 0.4551, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_Qnli-dev_cosine_accuracy": 0.671875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8589485883712769, + "eval_Qnli-dev_cosine_ap": 0.7007912514008645, + "eval_Qnli-dev_cosine_f1": 0.6754098360655738, + "eval_Qnli-dev_cosine_f1_threshold": 0.7456552386283875, + "eval_Qnli-dev_cosine_precision": 0.5508021390374331, + "eval_Qnli-dev_cosine_recall": 0.8728813559322034, + "eval_Qnli-dev_dot_accuracy": 0.625, + "eval_Qnli-dev_dot_accuracy_threshold": 420.0386962890625, + "eval_Qnli-dev_dot_ap": 0.6214107825707493, + "eval_Qnli-dev_dot_f1": 0.6604068857589983, + "eval_Qnli-dev_dot_f1_threshold": 358.24237060546875, + "eval_Qnli-dev_dot_precision": 0.5235732009925558, + "eval_Qnli-dev_dot_recall": 0.8940677966101694, + "eval_Qnli-dev_euclidean_accuracy": 0.669921875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.915521621704102, + "eval_Qnli-dev_euclidean_ap": 0.6960964915756133, + "eval_Qnli-dev_euclidean_f1": 0.6767676767676768, + "eval_Qnli-dev_euclidean_f1_threshold": 15.687938690185547, + "eval_Qnli-dev_euclidean_precision": 0.5614525139664804, + "eval_Qnli-dev_euclidean_recall": 0.8516949152542372, + "eval_Qnli-dev_manhattan_accuracy": 0.669921875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 238.73776245117188, + "eval_Qnli-dev_manhattan_ap": 0.6924771790396205, + "eval_Qnli-dev_manhattan_f1": 0.6758832565284177, + "eval_Qnli-dev_manhattan_f1_threshold": 310.7177734375, + "eval_Qnli-dev_manhattan_precision": 0.5301204819277109, + "eval_Qnli-dev_manhattan_recall": 0.9322033898305084, + "eval_Qnli-dev_max_accuracy": 0.671875, + "eval_Qnli-dev_max_accuracy_threshold": 420.0386962890625, + "eval_Qnli-dev_max_ap": 0.7007912514008645, + "eval_Qnli-dev_max_f1": 0.6767676767676768, + "eval_Qnli-dev_max_f1_threshold": 358.24237060546875, + "eval_Qnli-dev_max_precision": 0.5614525139664804, + "eval_Qnli-dev_max_recall": 0.9322033898305084, + "eval_allNLI-dev_cosine_accuracy": 0.705078125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.894379734992981, + "eval_allNLI-dev_cosine_ap": 0.58077636115338, + "eval_allNLI-dev_cosine_f1": 0.6057692307692306, + "eval_allNLI-dev_cosine_f1_threshold": 0.8448770046234131, + "eval_allNLI-dev_cosine_precision": 0.5185185185185185, + "eval_allNLI-dev_cosine_recall": 0.7283236994219653, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 479.90423583984375, + "eval_allNLI-dev_dot_ap": 0.47205439333282484, + "eval_allNLI-dev_dot_f1": 0.5433526011560694, + "eval_allNLI-dev_dot_f1_threshold": 374.1744384765625, + "eval_allNLI-dev_dot_precision": 0.407514450867052, + "eval_allNLI-dev_dot_recall": 0.815028901734104, + "eval_allNLI-dev_euclidean_accuracy": 0.7109375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.812898635864258, + "eval_allNLI-dev_euclidean_ap": 0.5811974122714074, + "eval_allNLI-dev_euclidean_f1": 0.6047619047619047, + "eval_allNLI-dev_euclidean_f1_threshold": 12.436492919921875, + "eval_allNLI-dev_euclidean_precision": 0.5141700404858299, + "eval_allNLI-dev_euclidean_recall": 0.7341040462427746, + "eval_allNLI-dev_manhattan_accuracy": 0.720703125, + "eval_allNLI-dev_manhattan_accuracy_threshold": 187.96119689941406, + "eval_allNLI-dev_manhattan_ap": 0.583329477014542, + "eval_allNLI-dev_manhattan_f1": 0.611353711790393, + "eval_allNLI-dev_manhattan_f1_threshold": 236.30606079101562, + "eval_allNLI-dev_manhattan_precision": 0.49122807017543857, + "eval_allNLI-dev_manhattan_recall": 0.8092485549132948, + "eval_allNLI-dev_max_accuracy": 0.720703125, + "eval_allNLI-dev_max_accuracy_threshold": 479.90423583984375, + "eval_allNLI-dev_max_ap": 0.583329477014542, + "eval_allNLI-dev_max_f1": 0.611353711790393, + "eval_allNLI-dev_max_f1_threshold": 374.1744384765625, + "eval_allNLI-dev_max_precision": 0.5185185185185185, + "eval_allNLI-dev_max_recall": 0.815028901734104, + "eval_sequential_score": 0.7007912514008645, + "eval_sts-test_pearson_cosine": 0.7918059509515682, + "eval_sts-test_pearson_dot": 0.7022968481900775, + "eval_sts-test_pearson_euclidean": 0.8153000179071552, + "eval_sts-test_pearson_manhattan": 0.8145147312460633, + "eval_sts-test_pearson_max": 0.8153000179071552, + "eval_sts-test_spearman_cosine": 0.8223630074681274, + "eval_sts-test_spearman_dot": 0.6867019303769535, + "eval_sts-test_spearman_euclidean": 0.8110723537216, + "eval_sts-test_spearman_manhattan": 0.8119889490574186, + "eval_sts-test_spearman_max": 0.8223630074681274, + "eval_vitaminc-pairs_loss": 3.940282106399536, + "eval_vitaminc-pairs_runtime": 5.4668, + "eval_vitaminc-pairs_samples_per_second": 23.414, + "eval_vitaminc-pairs_steps_per_second": 0.366, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_negation-triplets_loss": 1.8001011610031128, + "eval_negation-triplets_runtime": 1.1042, + "eval_negation-triplets_samples_per_second": 115.922, + "eval_negation-triplets_steps_per_second": 1.811, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_scitail-pairs-pos_loss": 0.18584367632865906, + "eval_scitail-pairs-pos_runtime": 1.2198, + "eval_scitail-pairs-pos_samples_per_second": 104.935, + "eval_scitail-pairs-pos_steps_per_second": 1.64, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_scitail-pairs-qa_loss": 0.019824935123324394, + "eval_scitail-pairs-qa_runtime": 0.9041, + "eval_scitail-pairs-qa_samples_per_second": 141.577, + "eval_scitail-pairs-qa_steps_per_second": 2.212, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_xsum-pairs_loss": 0.41752511262893677, + "eval_xsum-pairs_runtime": 4.3131, + "eval_xsum-pairs_samples_per_second": 29.677, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_sciq_pairs_loss": 0.09200021624565125, + "eval_sciq_pairs_runtime": 5.525, + "eval_sciq_pairs_samples_per_second": 23.168, + "eval_sciq_pairs_steps_per_second": 0.362, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_qasc_pairs_loss": 0.5905864834785461, + "eval_qasc_pairs_runtime": 0.9736, + "eval_qasc_pairs_samples_per_second": 131.469, + "eval_qasc_pairs_steps_per_second": 2.054, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_openbookqa_pairs_loss": 1.7676172256469727, + "eval_openbookqa_pairs_runtime": 0.8589, + "eval_openbookqa_pairs_samples_per_second": 149.024, + "eval_openbookqa_pairs_steps_per_second": 2.329, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_msmarco_pairs_loss": 1.1054226160049438, + "eval_msmarco_pairs_runtime": 1.9583, + "eval_msmarco_pairs_samples_per_second": 65.364, + "eval_msmarco_pairs_steps_per_second": 1.021, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_nq_pairs_loss": 1.3684680461883545, + "eval_nq_pairs_runtime": 4.4083, + "eval_nq_pairs_samples_per_second": 29.036, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_trivia_pairs_loss": 0.9785894751548767, + "eval_trivia_pairs_runtime": 4.5222, + "eval_trivia_pairs_samples_per_second": 28.305, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_gooaq_pairs_loss": 1.0878305435180664, + "eval_gooaq_pairs_runtime": 1.2657, + "eval_gooaq_pairs_samples_per_second": 101.127, + "eval_gooaq_pairs_steps_per_second": 1.58, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_paws-pos_loss": 0.05262936279177666, + "eval_paws-pos_runtime": 1.023, + "eval_paws-pos_samples_per_second": 125.119, + "eval_paws-pos_steps_per_second": 1.955, + "step": 660 + }, + { + "epoch": 0.5120248254460822, + "eval_global_dataset_loss": 0.7199053168296814, + "eval_global_dataset_runtime": 21.1232, + "eval_global_dataset_samples_per_second": 19.694, + "eval_global_dataset_steps_per_second": 0.331, + "step": 660 + }, + { + "epoch": 0.5128006206361521, + "grad_norm": 3.0742268562316895, + "learning_rate": 2.706227967097532e-05, + "loss": 0.0764, + "step": 661 + }, + { + "epoch": 0.5135764158262219, + "grad_norm": 10.229401588439941, + "learning_rate": 2.7103407755581668e-05, + "loss": 0.7882, + "step": 662 + }, + { + "epoch": 0.5143522110162917, + "grad_norm": 9.731830596923828, + "learning_rate": 2.714453584018801e-05, + "loss": 0.4966, + "step": 663 + }, + { + "epoch": 0.5151280062063616, + "grad_norm": 12.357343673706055, + "learning_rate": 2.7185663924794358e-05, + "loss": 1.3228, + "step": 664 + }, + { + "epoch": 0.5159038013964313, + "grad_norm": 13.702108383178711, + "learning_rate": 2.7226792009400703e-05, + "loss": 0.8, + "step": 665 + }, + { + "epoch": 0.5166795965865012, + "grad_norm": 12.134087562561035, + "learning_rate": 2.7267920094007045e-05, + "loss": 0.7538, + "step": 666 + }, + { + "epoch": 0.517455391776571, + "grad_norm": 15.904162406921387, + "learning_rate": 2.7309048178613394e-05, + "loss": 1.3167, + "step": 667 + }, + { + "epoch": 0.5182311869666408, + "grad_norm": 13.749979019165039, + "learning_rate": 2.735017626321974e-05, + "loss": 1.0212, + "step": 668 + }, + { + "epoch": 0.5190069821567106, + "grad_norm": 16.290987014770508, + "learning_rate": 2.7391304347826085e-05, + "loss": 1.2789, + "step": 669 + }, + { + "epoch": 0.5197827773467805, + "grad_norm": 11.391355514526367, + "learning_rate": 2.743243243243243e-05, + "loss": 1.013, + "step": 670 + }, + { + "epoch": 0.5205585725368502, + "grad_norm": 12.826869010925293, + "learning_rate": 2.747356051703878e-05, + "loss": 1.2767, + "step": 671 + }, + { + "epoch": 0.5213343677269201, + "grad_norm": 11.383997917175293, + "learning_rate": 2.751468860164512e-05, + "loss": 0.945, + "step": 672 + }, + { + "epoch": 0.5221101629169899, + "grad_norm": 13.27218246459961, + "learning_rate": 2.7555816686251466e-05, + "loss": 1.0292, + "step": 673 + }, + { + "epoch": 0.5228859581070597, + "grad_norm": 16.3438663482666, + "learning_rate": 2.7596944770857814e-05, + "loss": 1.6485, + "step": 674 + }, + { + "epoch": 0.5236617532971296, + "grad_norm": 12.95970344543457, + "learning_rate": 2.7638072855464156e-05, + "loss": 1.0887, + "step": 675 + }, + { + "epoch": 0.5244375484871994, + "grad_norm": 9.131745338439941, + "learning_rate": 2.7679200940070505e-05, + "loss": 0.5656, + "step": 676 + }, + { + "epoch": 0.5252133436772692, + "grad_norm": 9.181089401245117, + "learning_rate": 2.7720329024676847e-05, + "loss": 0.9879, + "step": 677 + }, + { + "epoch": 0.525989138867339, + "grad_norm": 13.082853317260742, + "learning_rate": 2.7761457109283192e-05, + "loss": 1.3261, + "step": 678 + }, + { + "epoch": 0.5267649340574089, + "grad_norm": 12.312566757202148, + "learning_rate": 2.780258519388954e-05, + "loss": 1.2894, + "step": 679 + }, + { + "epoch": 0.5275407292474786, + "grad_norm": 14.376317977905273, + "learning_rate": 2.7843713278495883e-05, + "loss": 1.715, + "step": 680 + }, + { + "epoch": 0.5283165244375485, + "grad_norm": 13.719832420349121, + "learning_rate": 2.788484136310223e-05, + "loss": 1.1868, + "step": 681 + }, + { + "epoch": 0.5290923196276183, + "grad_norm": 12.42406940460205, + "learning_rate": 2.7925969447708577e-05, + "loss": 0.9136, + "step": 682 + }, + { + "epoch": 0.5298681148176881, + "grad_norm": 3.575927257537842, + "learning_rate": 2.796709753231492e-05, + "loss": 0.1193, + "step": 683 + }, + { + "epoch": 0.5306439100077579, + "grad_norm": 3.044203996658325, + "learning_rate": 2.8008225616921267e-05, + "loss": 0.0927, + "step": 684 + }, + { + "epoch": 0.5314197051978278, + "grad_norm": 19.500102996826172, + "learning_rate": 2.8049353701527613e-05, + "loss": 1.9202, + "step": 685 + }, + { + "epoch": 0.5321955003878976, + "grad_norm": 14.444014549255371, + "learning_rate": 2.8090481786133958e-05, + "loss": 1.3207, + "step": 686 + }, + { + "epoch": 0.5329712955779674, + "grad_norm": 8.949565887451172, + "learning_rate": 2.8131609870740303e-05, + "loss": 0.585, + "step": 687 + }, + { + "epoch": 0.5337470907680373, + "grad_norm": 14.697336196899414, + "learning_rate": 2.8172737955346652e-05, + "loss": 1.5339, + "step": 688 + }, + { + "epoch": 0.534522885958107, + "grad_norm": 10.438421249389648, + "learning_rate": 2.8213866039952994e-05, + "loss": 0.457, + "step": 689 + }, + { + "epoch": 0.5352986811481769, + "grad_norm": 10.238954544067383, + "learning_rate": 2.825499412455934e-05, + "loss": 0.8198, + "step": 690 + }, + { + "epoch": 0.5360744763382467, + "grad_norm": 10.557405471801758, + "learning_rate": 2.8296122209165684e-05, + "loss": 0.5941, + "step": 691 + }, + { + "epoch": 0.5368502715283165, + "grad_norm": 21.442344665527344, + "learning_rate": 2.833725029377203e-05, + "loss": 2.7123, + "step": 692 + }, + { + "epoch": 0.5376260667183863, + "grad_norm": 15.35385513305664, + "learning_rate": 2.8378378378378378e-05, + "loss": 1.0445, + "step": 693 + }, + { + "epoch": 0.5384018619084562, + "grad_norm": 12.4437255859375, + "learning_rate": 2.841950646298472e-05, + "loss": 0.7682, + "step": 694 + }, + { + "epoch": 0.5391776570985259, + "grad_norm": 11.805971145629883, + "learning_rate": 2.8460634547591065e-05, + "loss": 0.8947, + "step": 695 + }, + { + "epoch": 0.5399534522885958, + "grad_norm": 14.908085823059082, + "learning_rate": 2.8501762632197414e-05, + "loss": 1.3467, + "step": 696 + }, + { + "epoch": 0.5407292474786657, + "grad_norm": 7.468909740447998, + "learning_rate": 2.8542890716803756e-05, + "loss": 0.5136, + "step": 697 + }, + { + "epoch": 0.5415050426687354, + "grad_norm": 10.362543106079102, + "learning_rate": 2.8584018801410105e-05, + "loss": 0.7952, + "step": 698 + }, + { + "epoch": 0.5422808378588053, + "grad_norm": 11.624421119689941, + "learning_rate": 2.862514688601645e-05, + "loss": 1.0138, + "step": 699 + }, + { + "epoch": 0.5430566330488751, + "grad_norm": 6.611544132232666, + "learning_rate": 2.8666274970622792e-05, + "loss": 0.3821, + "step": 700 + }, + { + "epoch": 0.543832428238945, + "grad_norm": 19.51932716369629, + "learning_rate": 2.870740305522914e-05, + "loss": 1.2379, + "step": 701 + }, + { + "epoch": 0.5446082234290147, + "grad_norm": 16.7391357421875, + "learning_rate": 2.8748531139835482e-05, + "loss": 1.5138, + "step": 702 + }, + { + "epoch": 0.5453840186190846, + "grad_norm": 7.159653186798096, + "learning_rate": 2.878965922444183e-05, + "loss": 0.3947, + "step": 703 + }, + { + "epoch": 0.5461598138091543, + "grad_norm": 9.305401802062988, + "learning_rate": 2.8830787309048176e-05, + "loss": 0.8279, + "step": 704 + }, + { + "epoch": 0.5469356089992242, + "grad_norm": 9.933472633361816, + "learning_rate": 2.8871915393654522e-05, + "loss": 1.0396, + "step": 705 + }, + { + "epoch": 0.547711404189294, + "grad_norm": 9.01205825805664, + "learning_rate": 2.8913043478260867e-05, + "loss": 0.3773, + "step": 706 + }, + { + "epoch": 0.5484871993793639, + "grad_norm": 7.620067119598389, + "learning_rate": 2.8954171562867212e-05, + "loss": 0.3855, + "step": 707 + }, + { + "epoch": 0.5492629945694336, + "grad_norm": 11.46169662475586, + "learning_rate": 2.8995299647473558e-05, + "loss": 0.528, + "step": 708 + }, + { + "epoch": 0.5500387897595035, + "grad_norm": 10.616952896118164, + "learning_rate": 2.9036427732079903e-05, + "loss": 1.2608, + "step": 709 + }, + { + "epoch": 0.5508145849495734, + "grad_norm": 11.644168853759766, + "learning_rate": 2.907755581668625e-05, + "loss": 0.8316, + "step": 710 + }, + { + "epoch": 0.5515903801396431, + "grad_norm": 17.670093536376953, + "learning_rate": 2.9118683901292593e-05, + "loss": 1.5927, + "step": 711 + }, + { + "epoch": 0.552366175329713, + "grad_norm": 9.694064140319824, + "learning_rate": 2.915981198589894e-05, + "loss": 0.5383, + "step": 712 + }, + { + "epoch": 0.5531419705197828, + "grad_norm": 14.904657363891602, + "learning_rate": 2.9200940070505287e-05, + "loss": 1.2653, + "step": 713 + }, + { + "epoch": 0.5539177657098526, + "grad_norm": 9.419154167175293, + "learning_rate": 2.924206815511163e-05, + "loss": 0.6491, + "step": 714 + }, + { + "epoch": 0.5546935608999224, + "grad_norm": 20.50204086303711, + "learning_rate": 2.9283196239717978e-05, + "loss": 3.2608, + "step": 715 + }, + { + "epoch": 0.5554693560899923, + "grad_norm": 8.960325241088867, + "learning_rate": 2.932432432432432e-05, + "loss": 0.4876, + "step": 716 + }, + { + "epoch": 0.556245151280062, + "grad_norm": 8.47724437713623, + "learning_rate": 2.9365452408930665e-05, + "loss": 0.4794, + "step": 717 + }, + { + "epoch": 0.5570209464701319, + "grad_norm": 14.06983470916748, + "learning_rate": 2.9406580493537014e-05, + "loss": 1.6532, + "step": 718 + }, + { + "epoch": 0.5577967416602017, + "grad_norm": 7.726792335510254, + "learning_rate": 2.9447708578143356e-05, + "loss": 0.4491, + "step": 719 + }, + { + "epoch": 0.5585725368502715, + "grad_norm": 8.665955543518066, + "learning_rate": 2.9488836662749704e-05, + "loss": 0.6076, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_Qnli-dev_cosine_accuracy": 0.689453125, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8603726625442505, + "eval_Qnli-dev_cosine_ap": 0.7013985251734445, + "eval_Qnli-dev_cosine_f1": 0.6775244299674267, + "eval_Qnli-dev_cosine_f1_threshold": 0.7960958480834961, + "eval_Qnli-dev_cosine_precision": 0.5502645502645502, + "eval_Qnli-dev_cosine_recall": 0.8813559322033898, + "eval_Qnli-dev_dot_accuracy": 0.634765625, + "eval_Qnli-dev_dot_accuracy_threshold": 438.27618408203125, + "eval_Qnli-dev_dot_ap": 0.6215014963741472, + "eval_Qnli-dev_dot_f1": 0.6554878048780487, + "eval_Qnli-dev_dot_f1_threshold": 382.665771484375, + "eval_Qnli-dev_dot_precision": 0.5119047619047619, + "eval_Qnli-dev_dot_recall": 0.9110169491525424, + "eval_Qnli-dev_euclidean_accuracy": 0.6796875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.89668083190918, + "eval_Qnli-dev_euclidean_ap": 0.6949958891404164, + "eval_Qnli-dev_euclidean_f1": 0.6774774774774774, + "eval_Qnli-dev_euclidean_f1_threshold": 13.534368515014648, + "eval_Qnli-dev_euclidean_precision": 0.5893416927899686, + "eval_Qnli-dev_euclidean_recall": 0.7966101694915254, + "eval_Qnli-dev_manhattan_accuracy": 0.6875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 207.3291473388672, + "eval_Qnli-dev_manhattan_ap": 0.6924403284902163, + "eval_Qnli-dev_manhattan_f1": 0.6741935483870968, + "eval_Qnli-dev_manhattan_f1_threshold": 269.037841796875, + "eval_Qnli-dev_manhattan_precision": 0.5442708333333334, + "eval_Qnli-dev_manhattan_recall": 0.885593220338983, + "eval_Qnli-dev_max_accuracy": 0.689453125, + "eval_Qnli-dev_max_accuracy_threshold": 438.27618408203125, + "eval_Qnli-dev_max_ap": 0.7013985251734445, + "eval_Qnli-dev_max_f1": 0.6775244299674267, + "eval_Qnli-dev_max_f1_threshold": 382.665771484375, + "eval_Qnli-dev_max_precision": 0.5893416927899686, + "eval_Qnli-dev_max_recall": 0.9110169491525424, + "eval_allNLI-dev_cosine_accuracy": 0.705078125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9376393556594849, + "eval_allNLI-dev_cosine_ap": 0.5798052690345175, + "eval_allNLI-dev_cosine_f1": 0.6208530805687204, + "eval_allNLI-dev_cosine_f1_threshold": 0.863029956817627, + "eval_allNLI-dev_cosine_precision": 0.5261044176706827, + "eval_allNLI-dev_cosine_recall": 0.7572254335260116, + "eval_allNLI-dev_dot_accuracy": 0.673828125, + "eval_allNLI-dev_dot_accuracy_threshold": 498.4715881347656, + "eval_allNLI-dev_dot_ap": 0.4700129297819463, + "eval_allNLI-dev_dot_f1": 0.5468451242829828, + "eval_allNLI-dev_dot_f1_threshold": 384.7823486328125, + "eval_allNLI-dev_dot_precision": 0.4085714285714286, + "eval_allNLI-dev_dot_recall": 0.8265895953757225, + "eval_allNLI-dev_euclidean_accuracy": 0.708984375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 8.113941192626953, + "eval_allNLI-dev_euclidean_ap": 0.5780604025406824, + "eval_allNLI-dev_euclidean_f1": 0.6063569682151589, + "eval_allNLI-dev_euclidean_f1_threshold": 11.373184204101562, + "eval_allNLI-dev_euclidean_precision": 0.5254237288135594, + "eval_allNLI-dev_euclidean_recall": 0.7167630057803468, + "eval_allNLI-dev_manhattan_accuracy": 0.712890625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 184.79954528808594, + "eval_allNLI-dev_manhattan_ap": 0.5781894267413036, + "eval_allNLI-dev_manhattan_f1": 0.6063569682151589, + "eval_allNLI-dev_manhattan_f1_threshold": 204.30731201171875, + "eval_allNLI-dev_manhattan_precision": 0.5254237288135594, + "eval_allNLI-dev_manhattan_recall": 0.7167630057803468, + "eval_allNLI-dev_max_accuracy": 0.712890625, + "eval_allNLI-dev_max_accuracy_threshold": 498.4715881347656, + "eval_allNLI-dev_max_ap": 0.5798052690345175, + "eval_allNLI-dev_max_f1": 0.6208530805687204, + "eval_allNLI-dev_max_f1_threshold": 384.7823486328125, + "eval_allNLI-dev_max_precision": 0.5261044176706827, + "eval_allNLI-dev_max_recall": 0.8265895953757225, + "eval_sequential_score": 0.7013985251734445, + "eval_sts-test_pearson_cosine": 0.8033284585505502, + "eval_sts-test_pearson_dot": 0.650022718025308, + "eval_sts-test_pearson_euclidean": 0.8326210219298027, + "eval_sts-test_pearson_manhattan": 0.8333016063834682, + "eval_sts-test_pearson_max": 0.8333016063834682, + "eval_sts-test_spearman_cosine": 0.8372359549377748, + "eval_sts-test_spearman_dot": 0.6482093118961868, + "eval_sts-test_spearman_euclidean": 0.8309703308303649, + "eval_sts-test_spearman_manhattan": 0.8326837671265189, + "eval_sts-test_spearman_max": 0.8372359549377748, + "eval_vitaminc-pairs_loss": 3.8188633918762207, + "eval_vitaminc-pairs_runtime": 5.4758, + "eval_vitaminc-pairs_samples_per_second": 23.376, + "eval_vitaminc-pairs_steps_per_second": 0.365, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_negation-triplets_loss": 1.716660499572754, + "eval_negation-triplets_runtime": 1.1172, + "eval_negation-triplets_samples_per_second": 114.568, + "eval_negation-triplets_steps_per_second": 1.79, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_scitail-pairs-pos_loss": 0.21144355833530426, + "eval_scitail-pairs-pos_runtime": 1.2353, + "eval_scitail-pairs-pos_samples_per_second": 103.618, + "eval_scitail-pairs-pos_steps_per_second": 1.619, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_scitail-pairs-qa_loss": 0.02602345310151577, + "eval_scitail-pairs-qa_runtime": 0.9031, + "eval_scitail-pairs-qa_samples_per_second": 141.726, + "eval_scitail-pairs-qa_steps_per_second": 2.214, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_xsum-pairs_loss": 0.4568767547607422, + "eval_xsum-pairs_runtime": 4.3183, + "eval_xsum-pairs_samples_per_second": 29.641, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_sciq_pairs_loss": 0.11070608347654343, + "eval_sciq_pairs_runtime": 5.5189, + "eval_sciq_pairs_samples_per_second": 23.193, + "eval_sciq_pairs_steps_per_second": 0.362, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_qasc_pairs_loss": 0.4799177646636963, + "eval_qasc_pairs_runtime": 0.9735, + "eval_qasc_pairs_samples_per_second": 131.485, + "eval_qasc_pairs_steps_per_second": 2.054, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_openbookqa_pairs_loss": 1.5125000476837158, + "eval_openbookqa_pairs_runtime": 0.8562, + "eval_openbookqa_pairs_samples_per_second": 149.495, + "eval_openbookqa_pairs_steps_per_second": 2.336, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_msmarco_pairs_loss": 0.911155104637146, + "eval_msmarco_pairs_runtime": 1.9594, + "eval_msmarco_pairs_samples_per_second": 65.325, + "eval_msmarco_pairs_steps_per_second": 1.021, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_nq_pairs_loss": 1.1344616413116455, + "eval_nq_pairs_runtime": 4.4011, + "eval_nq_pairs_samples_per_second": 29.083, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_trivia_pairs_loss": 0.8704670667648315, + "eval_trivia_pairs_runtime": 4.5324, + "eval_trivia_pairs_samples_per_second": 28.241, + "eval_trivia_pairs_steps_per_second": 0.441, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_gooaq_pairs_loss": 0.8927607536315918, + "eval_gooaq_pairs_runtime": 1.3065, + "eval_gooaq_pairs_samples_per_second": 97.974, + "eval_gooaq_pairs_steps_per_second": 1.531, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_paws-pos_loss": 0.05082692205905914, + "eval_paws-pos_runtime": 1.0223, + "eval_paws-pos_samples_per_second": 125.205, + "eval_paws-pos_steps_per_second": 1.956, + "step": 720 + }, + { + "epoch": 0.5585725368502715, + "eval_global_dataset_loss": 0.697550356388092, + "eval_global_dataset_runtime": 21.1305, + "eval_global_dataset_samples_per_second": 19.687, + "eval_global_dataset_steps_per_second": 0.331, + "step": 720 + }, + { + "epoch": 0.5593483320403414, + "grad_norm": 4.372936725616455, + "learning_rate": 2.952996474735605e-05, + "loss": 0.1445, + "step": 721 + }, + { + "epoch": 0.5601241272304112, + "grad_norm": 7.920864105224609, + "learning_rate": 2.9571092831962395e-05, + "loss": 0.5675, + "step": 722 + }, + { + "epoch": 0.560899922420481, + "grad_norm": 7.876180648803711, + "learning_rate": 2.961222091656874e-05, + "loss": 0.4063, + "step": 723 + }, + { + "epoch": 0.5616757176105508, + "grad_norm": 14.68873405456543, + "learning_rate": 2.9653349001175086e-05, + "loss": 1.5336, + "step": 724 + }, + { + "epoch": 0.5624515128006207, + "grad_norm": 12.422935485839844, + "learning_rate": 2.969447708578143e-05, + "loss": 1.1264, + "step": 725 + }, + { + "epoch": 0.5632273079906904, + "grad_norm": 0.8129711151123047, + "learning_rate": 2.9735605170387776e-05, + "loss": 0.038, + "step": 726 + }, + { + "epoch": 0.5640031031807603, + "grad_norm": 10.077345848083496, + "learning_rate": 2.9776733254994125e-05, + "loss": 0.4946, + "step": 727 + }, + { + "epoch": 0.5647788983708301, + "grad_norm": 8.413456916809082, + "learning_rate": 2.9817861339600467e-05, + "loss": 0.4399, + "step": 728 + }, + { + "epoch": 0.5655546935608999, + "grad_norm": 7.0985798835754395, + "learning_rate": 2.9858989424206812e-05, + "loss": 0.2663, + "step": 729 + }, + { + "epoch": 0.5663304887509697, + "grad_norm": 11.00217342376709, + "learning_rate": 2.9900117508813157e-05, + "loss": 0.5476, + "step": 730 + }, + { + "epoch": 0.5671062839410396, + "grad_norm": 1.1290919780731201, + "learning_rate": 2.9941245593419503e-05, + "loss": 0.0714, + "step": 731 + }, + { + "epoch": 0.5678820791311094, + "grad_norm": 17.3148193359375, + "learning_rate": 2.998237367802585e-05, + "loss": 1.2477, + "step": 732 + }, + { + "epoch": 0.5686578743211792, + "grad_norm": 8.444149017333984, + "learning_rate": 3.0023501762632193e-05, + "loss": 0.4663, + "step": 733 + }, + { + "epoch": 0.5694336695112491, + "grad_norm": 10.664148330688477, + "learning_rate": 3.0064629847238542e-05, + "loss": 0.3875, + "step": 734 + }, + { + "epoch": 0.5702094647013188, + "grad_norm": 2.331057071685791, + "learning_rate": 3.0105757931844887e-05, + "loss": 0.0636, + "step": 735 + }, + { + "epoch": 0.5709852598913887, + "grad_norm": 11.15227222442627, + "learning_rate": 3.014688601645123e-05, + "loss": 0.7118, + "step": 736 + }, + { + "epoch": 0.5717610550814585, + "grad_norm": 10.80561637878418, + "learning_rate": 3.0188014101057578e-05, + "loss": 0.5781, + "step": 737 + }, + { + "epoch": 0.5725368502715283, + "grad_norm": 9.853097915649414, + "learning_rate": 3.0229142185663923e-05, + "loss": 0.509, + "step": 738 + }, + { + "epoch": 0.5733126454615981, + "grad_norm": 6.646691799163818, + "learning_rate": 3.027027027027027e-05, + "loss": 0.314, + "step": 739 + }, + { + "epoch": 0.574088440651668, + "grad_norm": 2.3896520137786865, + "learning_rate": 3.0311398354876614e-05, + "loss": 0.0644, + "step": 740 + }, + { + "epoch": 0.5748642358417377, + "grad_norm": 6.928627014160156, + "learning_rate": 3.0352526439482956e-05, + "loss": 0.4537, + "step": 741 + }, + { + "epoch": 0.5756400310318076, + "grad_norm": 3.4702837467193604, + "learning_rate": 3.0393654524089304e-05, + "loss": 0.067, + "step": 742 + }, + { + "epoch": 0.5764158262218774, + "grad_norm": 11.393362998962402, + "learning_rate": 3.043478260869565e-05, + "loss": 0.63, + "step": 743 + }, + { + "epoch": 0.5771916214119472, + "grad_norm": 9.76422119140625, + "learning_rate": 3.0475910693301995e-05, + "loss": 0.9531, + "step": 744 + }, + { + "epoch": 0.5779674166020171, + "grad_norm": 11.860621452331543, + "learning_rate": 3.051703877790834e-05, + "loss": 0.9424, + "step": 745 + }, + { + "epoch": 0.5787432117920869, + "grad_norm": 9.523978233337402, + "learning_rate": 3.0558166862514685e-05, + "loss": 0.5714, + "step": 746 + }, + { + "epoch": 0.5795190069821567, + "grad_norm": 15.020031929016113, + "learning_rate": 3.059929494712103e-05, + "loss": 0.8998, + "step": 747 + }, + { + "epoch": 0.5802948021722265, + "grad_norm": 20.261709213256836, + "learning_rate": 3.0640423031727376e-05, + "loss": 2.9047, + "step": 748 + }, + { + "epoch": 0.5810705973622964, + "grad_norm": 10.935188293457031, + "learning_rate": 3.068155111633372e-05, + "loss": 1.1596, + "step": 749 + }, + { + "epoch": 0.5818463925523661, + "grad_norm": 8.534791946411133, + "learning_rate": 3.0722679200940067e-05, + "loss": 0.4542, + "step": 750 + }, + { + "epoch": 0.582622187742436, + "grad_norm": 9.784218788146973, + "learning_rate": 3.076380728554641e-05, + "loss": 0.4563, + "step": 751 + }, + { + "epoch": 0.5833979829325058, + "grad_norm": 10.46196460723877, + "learning_rate": 3.080493537015276e-05, + "loss": 0.7169, + "step": 752 + }, + { + "epoch": 0.5841737781225756, + "grad_norm": 12.059857368469238, + "learning_rate": 3.08460634547591e-05, + "loss": 0.9529, + "step": 753 + }, + { + "epoch": 0.5849495733126454, + "grad_norm": 15.391682624816895, + "learning_rate": 3.088719153936545e-05, + "loss": 1.8114, + "step": 754 + }, + { + "epoch": 0.5857253685027153, + "grad_norm": 11.071196556091309, + "learning_rate": 3.092831962397179e-05, + "loss": 0.6818, + "step": 755 + }, + { + "epoch": 0.5865011636927852, + "grad_norm": 14.564682006835938, + "learning_rate": 3.0969447708578145e-05, + "loss": 1.7219, + "step": 756 + }, + { + "epoch": 0.5872769588828549, + "grad_norm": 8.644091606140137, + "learning_rate": 3.1010575793184484e-05, + "loss": 0.483, + "step": 757 + }, + { + "epoch": 0.5880527540729248, + "grad_norm": 7.61334228515625, + "learning_rate": 3.105170387779083e-05, + "loss": 0.3853, + "step": 758 + }, + { + "epoch": 0.5888285492629945, + "grad_norm": 14.001456260681152, + "learning_rate": 3.109283196239718e-05, + "loss": 1.4371, + "step": 759 + }, + { + "epoch": 0.5896043444530644, + "grad_norm": 9.1812105178833, + "learning_rate": 3.113396004700352e-05, + "loss": 0.6702, + "step": 760 + }, + { + "epoch": 0.5903801396431342, + "grad_norm": 13.61975383758545, + "learning_rate": 3.117508813160987e-05, + "loss": 1.2968, + "step": 761 + }, + { + "epoch": 0.591155934833204, + "grad_norm": 7.879928112030029, + "learning_rate": 3.121621621621621e-05, + "loss": 0.5906, + "step": 762 + }, + { + "epoch": 0.5919317300232738, + "grad_norm": 11.044487953186035, + "learning_rate": 3.1257344300822555e-05, + "loss": 1.044, + "step": 763 + }, + { + "epoch": 0.5927075252133437, + "grad_norm": 9.801410675048828, + "learning_rate": 3.129847238542891e-05, + "loss": 0.5166, + "step": 764 + }, + { + "epoch": 0.5934833204034135, + "grad_norm": 9.44311237335205, + "learning_rate": 3.1339600470035246e-05, + "loss": 0.5071, + "step": 765 + }, + { + "epoch": 0.5942591155934833, + "grad_norm": 11.208498001098633, + "learning_rate": 3.13807285546416e-05, + "loss": 0.6988, + "step": 766 + }, + { + "epoch": 0.5950349107835532, + "grad_norm": 9.171442985534668, + "learning_rate": 3.142185663924794e-05, + "loss": 0.9081, + "step": 767 + }, + { + "epoch": 0.595810705973623, + "grad_norm": 6.937619686126709, + "learning_rate": 3.146298472385429e-05, + "loss": 0.3738, + "step": 768 + }, + { + "epoch": 0.5965865011636928, + "grad_norm": 8.843375205993652, + "learning_rate": 3.1504112808460634e-05, + "loss": 0.5956, + "step": 769 + }, + { + "epoch": 0.5973622963537626, + "grad_norm": 14.882801055908203, + "learning_rate": 3.154524089306698e-05, + "loss": 1.1079, + "step": 770 + }, + { + "epoch": 0.5981380915438325, + "grad_norm": 8.172786712646484, + "learning_rate": 3.1586368977673324e-05, + "loss": 0.6008, + "step": 771 + }, + { + "epoch": 0.5989138867339022, + "grad_norm": 12.259247779846191, + "learning_rate": 3.162749706227967e-05, + "loss": 1.522, + "step": 772 + }, + { + "epoch": 0.5996896819239721, + "grad_norm": 14.156942367553711, + "learning_rate": 3.1668625146886015e-05, + "loss": 1.3301, + "step": 773 + }, + { + "epoch": 0.6004654771140419, + "grad_norm": 3.3066277503967285, + "learning_rate": 3.170975323149236e-05, + "loss": 0.0772, + "step": 774 + }, + { + "epoch": 0.6012412723041117, + "grad_norm": 1.4857008457183838, + "learning_rate": 3.1750881316098706e-05, + "loss": 0.0563, + "step": 775 + }, + { + "epoch": 0.6020170674941815, + "grad_norm": 8.465863227844238, + "learning_rate": 3.179200940070505e-05, + "loss": 0.6185, + "step": 776 + }, + { + "epoch": 0.6027928626842514, + "grad_norm": 8.269394874572754, + "learning_rate": 3.1833137485311396e-05, + "loss": 0.465, + "step": 777 + }, + { + "epoch": 0.6035686578743211, + "grad_norm": 13.826294898986816, + "learning_rate": 3.187426556991774e-05, + "loss": 1.1821, + "step": 778 + }, + { + "epoch": 0.604344453064391, + "grad_norm": 11.957036018371582, + "learning_rate": 3.191539365452409e-05, + "loss": 1.108, + "step": 779 + }, + { + "epoch": 0.6051202482544609, + "grad_norm": 9.368646621704102, + "learning_rate": 3.195652173913043e-05, + "loss": 0.6595, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_Qnli-dev_cosine_accuracy": 0.666015625, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8268861770629883, + "eval_Qnli-dev_cosine_ap": 0.6887524572464852, + "eval_Qnli-dev_cosine_f1": 0.6734006734006733, + "eval_Qnli-dev_cosine_f1_threshold": 0.7692747116088867, + "eval_Qnli-dev_cosine_precision": 0.5586592178770949, + "eval_Qnli-dev_cosine_recall": 0.847457627118644, + "eval_Qnli-dev_dot_accuracy": 0.638671875, + "eval_Qnli-dev_dot_accuracy_threshold": 404.38848876953125, + "eval_Qnli-dev_dot_ap": 0.6348319628277213, + "eval_Qnli-dev_dot_f1": 0.6607669616519173, + "eval_Qnli-dev_dot_f1_threshold": 340.088623046875, + "eval_Qnli-dev_dot_precision": 0.5067873303167421, + "eval_Qnli-dev_dot_recall": 0.9491525423728814, + "eval_Qnli-dev_euclidean_accuracy": 0.6640625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.941691398620605, + "eval_Qnli-dev_euclidean_ap": 0.6816941758591539, + "eval_Qnli-dev_euclidean_f1": 0.6777408637873753, + "eval_Qnli-dev_euclidean_f1_threshold": 15.37027359008789, + "eval_Qnli-dev_euclidean_precision": 0.5573770491803278, + "eval_Qnli-dev_euclidean_recall": 0.864406779661017, + "eval_Qnli-dev_manhattan_accuracy": 0.662109375, + "eval_Qnli-dev_manhattan_accuracy_threshold": 201.82369995117188, + "eval_Qnli-dev_manhattan_ap": 0.6806650793482305, + "eval_Qnli-dev_manhattan_f1": 0.6807131280388979, + "eval_Qnli-dev_manhattan_f1_threshold": 282.18292236328125, + "eval_Qnli-dev_manhattan_precision": 0.5511811023622047, + "eval_Qnli-dev_manhattan_recall": 0.8898305084745762, + "eval_Qnli-dev_max_accuracy": 0.666015625, + "eval_Qnli-dev_max_accuracy_threshold": 404.38848876953125, + "eval_Qnli-dev_max_ap": 0.6887524572464852, + "eval_Qnli-dev_max_f1": 0.6807131280388979, + "eval_Qnli-dev_max_f1_threshold": 340.088623046875, + "eval_Qnli-dev_max_precision": 0.5586592178770949, + "eval_Qnli-dev_max_recall": 0.9491525423728814, + "eval_allNLI-dev_cosine_accuracy": 0.708984375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9096022844314575, + "eval_allNLI-dev_cosine_ap": 0.5767123578469748, + "eval_allNLI-dev_cosine_f1": 0.5936739659367396, + "eval_allNLI-dev_cosine_f1_threshold": 0.8360572457313538, + "eval_allNLI-dev_cosine_precision": 0.5126050420168067, + "eval_allNLI-dev_cosine_recall": 0.7052023121387283, + "eval_allNLI-dev_dot_accuracy": 0.677734375, + "eval_allNLI-dev_dot_accuracy_threshold": 454.55975341796875, + "eval_allNLI-dev_dot_ap": 0.479007773450769, + "eval_allNLI-dev_dot_f1": 0.5577264653641208, + "eval_allNLI-dev_dot_f1_threshold": 340.08404541015625, + "eval_allNLI-dev_dot_precision": 0.4025641025641026, + "eval_allNLI-dev_dot_recall": 0.9075144508670521, + "eval_allNLI-dev_euclidean_accuracy": 0.705078125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 8.928918838500977, + "eval_allNLI-dev_euclidean_ap": 0.577906322958973, + "eval_allNLI-dev_euclidean_f1": 0.6012793176972281, + "eval_allNLI-dev_euclidean_f1_threshold": 13.597992897033691, + "eval_allNLI-dev_euclidean_precision": 0.47635135135135137, + "eval_allNLI-dev_euclidean_recall": 0.815028901734104, + "eval_allNLI-dev_manhattan_accuracy": 0.712890625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 191.7563018798828, + "eval_allNLI-dev_manhattan_ap": 0.5728819124013971, + "eval_allNLI-dev_manhattan_f1": 0.5948275862068965, + "eval_allNLI-dev_manhattan_f1_threshold": 243.3690643310547, + "eval_allNLI-dev_manhattan_precision": 0.4742268041237113, + "eval_allNLI-dev_manhattan_recall": 0.7976878612716763, + "eval_allNLI-dev_max_accuracy": 0.712890625, + "eval_allNLI-dev_max_accuracy_threshold": 454.55975341796875, + "eval_allNLI-dev_max_ap": 0.577906322958973, + "eval_allNLI-dev_max_f1": 0.6012793176972281, + "eval_allNLI-dev_max_f1_threshold": 340.08404541015625, + "eval_allNLI-dev_max_precision": 0.5126050420168067, + "eval_allNLI-dev_max_recall": 0.9075144508670521, + "eval_sequential_score": 0.6887524572464852, + "eval_sts-test_pearson_cosine": 0.8157104398830798, + "eval_sts-test_pearson_dot": 0.755957162056917, + "eval_sts-test_pearson_euclidean": 0.8476799511024182, + "eval_sts-test_pearson_manhattan": 0.8488228621790745, + "eval_sts-test_pearson_max": 0.8488228621790745, + "eval_sts-test_spearman_cosine": 0.850806819705978, + "eval_sts-test_spearman_dot": 0.74277696144831, + "eval_sts-test_spearman_euclidean": 0.8431612457023031, + "eval_sts-test_spearman_manhattan": 0.8449485215797528, + "eval_sts-test_spearman_max": 0.850806819705978, + "eval_vitaminc-pairs_loss": 3.630547285079956, + "eval_vitaminc-pairs_runtime": 5.5616, + "eval_vitaminc-pairs_samples_per_second": 23.015, + "eval_vitaminc-pairs_steps_per_second": 0.36, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_negation-triplets_loss": 1.613579273223877, + "eval_negation-triplets_runtime": 1.215, + "eval_negation-triplets_samples_per_second": 105.35, + "eval_negation-triplets_steps_per_second": 1.646, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_scitail-pairs-pos_loss": 0.23045063018798828, + "eval_scitail-pairs-pos_runtime": 1.2974, + "eval_scitail-pairs-pos_samples_per_second": 98.662, + "eval_scitail-pairs-pos_steps_per_second": 1.542, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_scitail-pairs-qa_loss": 0.015527600422501564, + "eval_scitail-pairs-qa_runtime": 0.9734, + "eval_scitail-pairs-qa_samples_per_second": 131.504, + "eval_scitail-pairs-qa_steps_per_second": 2.055, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_xsum-pairs_loss": 0.34815362095832825, + "eval_xsum-pairs_runtime": 4.3525, + "eval_xsum-pairs_samples_per_second": 29.408, + "eval_xsum-pairs_steps_per_second": 0.46, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_sciq_pairs_loss": 0.09282456338405609, + "eval_sciq_pairs_runtime": 5.6003, + "eval_sciq_pairs_samples_per_second": 22.856, + "eval_sciq_pairs_steps_per_second": 0.357, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_qasc_pairs_loss": 0.49120184779167175, + "eval_qasc_pairs_runtime": 1.0343, + "eval_qasc_pairs_samples_per_second": 123.757, + "eval_qasc_pairs_steps_per_second": 1.934, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_openbookqa_pairs_loss": 1.6851919889450073, + "eval_openbookqa_pairs_runtime": 0.9198, + "eval_openbookqa_pairs_samples_per_second": 139.164, + "eval_openbookqa_pairs_steps_per_second": 2.174, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_msmarco_pairs_loss": 0.9934359192848206, + "eval_msmarco_pairs_runtime": 1.9806, + "eval_msmarco_pairs_samples_per_second": 64.627, + "eval_msmarco_pairs_steps_per_second": 1.01, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_nq_pairs_loss": 1.2469141483306885, + "eval_nq_pairs_runtime": 4.4543, + "eval_nq_pairs_samples_per_second": 28.736, + "eval_nq_pairs_steps_per_second": 0.449, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_trivia_pairs_loss": 0.8310095071792603, + "eval_trivia_pairs_runtime": 4.5718, + "eval_trivia_pairs_samples_per_second": 27.998, + "eval_trivia_pairs_steps_per_second": 0.437, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_gooaq_pairs_loss": 0.7958693504333496, + "eval_gooaq_pairs_runtime": 1.4868, + "eval_gooaq_pairs_samples_per_second": 86.093, + "eval_gooaq_pairs_steps_per_second": 1.345, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_paws-pos_loss": 0.04702610895037651, + "eval_paws-pos_runtime": 1.0246, + "eval_paws-pos_samples_per_second": 124.93, + "eval_paws-pos_steps_per_second": 1.952, + "step": 780 + }, + { + "epoch": 0.6051202482544609, + "eval_global_dataset_loss": 0.6912338137626648, + "eval_global_dataset_runtime": 21.1637, + "eval_global_dataset_samples_per_second": 19.656, + "eval_global_dataset_steps_per_second": 0.331, + "step": 780 + }, + { + "epoch": 0.6058960434445306, + "grad_norm": 1.2432920932769775, + "learning_rate": 3.199764982373678e-05, + "loss": 0.051, + "step": 781 + }, + { + "epoch": 0.6066718386346005, + "grad_norm": 1.139133334159851, + "learning_rate": 3.203877790834312e-05, + "loss": 0.0368, + "step": 782 + }, + { + "epoch": 0.6074476338246703, + "grad_norm": 9.700934410095215, + "learning_rate": 3.207990599294947e-05, + "loss": 0.5663, + "step": 783 + }, + { + "epoch": 0.6082234290147401, + "grad_norm": 9.31375503540039, + "learning_rate": 3.212103407755581e-05, + "loss": 0.5206, + "step": 784 + }, + { + "epoch": 0.6089992242048099, + "grad_norm": 8.267459869384766, + "learning_rate": 3.216216216216216e-05, + "loss": 0.509, + "step": 785 + }, + { + "epoch": 0.6097750193948798, + "grad_norm": 9.485785484313965, + "learning_rate": 3.2203290246768504e-05, + "loss": 0.8124, + "step": 786 + }, + { + "epoch": 0.6105508145849495, + "grad_norm": 8.342541694641113, + "learning_rate": 3.224441833137485e-05, + "loss": 0.6161, + "step": 787 + }, + { + "epoch": 0.6113266097750194, + "grad_norm": 15.334446907043457, + "learning_rate": 3.2285546415981194e-05, + "loss": 1.2831, + "step": 788 + }, + { + "epoch": 0.6121024049650892, + "grad_norm": 2.2616965770721436, + "learning_rate": 3.232667450058754e-05, + "loss": 0.059, + "step": 789 + }, + { + "epoch": 0.612878200155159, + "grad_norm": 7.60488224029541, + "learning_rate": 3.2367802585193885e-05, + "loss": 0.5068, + "step": 790 + }, + { + "epoch": 0.6136539953452289, + "grad_norm": 10.485137939453125, + "learning_rate": 3.240893066980023e-05, + "loss": 0.5579, + "step": 791 + }, + { + "epoch": 0.6144297905352987, + "grad_norm": 7.6277875900268555, + "learning_rate": 3.2450058754406575e-05, + "loss": 0.4626, + "step": 792 + }, + { + "epoch": 0.6152055857253685, + "grad_norm": 1.1966078281402588, + "learning_rate": 3.249118683901292e-05, + "loss": 0.0294, + "step": 793 + }, + { + "epoch": 0.6159813809154383, + "grad_norm": 11.492501258850098, + "learning_rate": 3.2532314923619266e-05, + "loss": 0.5729, + "step": 794 + }, + { + "epoch": 0.6167571761055082, + "grad_norm": 9.295225143432617, + "learning_rate": 3.257344300822562e-05, + "loss": 0.4606, + "step": 795 + }, + { + "epoch": 0.6175329712955779, + "grad_norm": 13.280166625976562, + "learning_rate": 3.2614571092831957e-05, + "loss": 0.7513, + "step": 796 + }, + { + "epoch": 0.6183087664856478, + "grad_norm": 16.385412216186523, + "learning_rate": 3.26556991774383e-05, + "loss": 1.2639, + "step": 797 + }, + { + "epoch": 0.6190845616757176, + "grad_norm": 21.18305015563965, + "learning_rate": 3.2696827262044654e-05, + "loss": 2.0057, + "step": 798 + }, + { + "epoch": 0.6198603568657874, + "grad_norm": 13.059396743774414, + "learning_rate": 3.273795534665099e-05, + "loss": 1.0553, + "step": 799 + }, + { + "epoch": 0.6206361520558572, + "grad_norm": 18.694307327270508, + "learning_rate": 3.2779083431257345e-05, + "loss": 1.3601, + "step": 800 + }, + { + "epoch": 0.6214119472459271, + "grad_norm": 10.15408706665039, + "learning_rate": 3.282021151586368e-05, + "loss": 0.498, + "step": 801 + }, + { + "epoch": 0.622187742435997, + "grad_norm": 1.4777809381484985, + "learning_rate": 3.2861339600470035e-05, + "loss": 0.0417, + "step": 802 + }, + { + "epoch": 0.6229635376260667, + "grad_norm": 7.932584285736084, + "learning_rate": 3.290246768507638e-05, + "loss": 0.5607, + "step": 803 + }, + { + "epoch": 0.6237393328161366, + "grad_norm": 7.78424072265625, + "learning_rate": 3.294359576968272e-05, + "loss": 0.4816, + "step": 804 + }, + { + "epoch": 0.6245151280062063, + "grad_norm": 9.77271842956543, + "learning_rate": 3.298472385428907e-05, + "loss": 0.5568, + "step": 805 + }, + { + "epoch": 0.6252909231962762, + "grad_norm": 3.6694772243499756, + "learning_rate": 3.3025851938895416e-05, + "loss": 0.1023, + "step": 806 + }, + { + "epoch": 0.626066718386346, + "grad_norm": 11.192282676696777, + "learning_rate": 3.306698002350176e-05, + "loss": 0.857, + "step": 807 + }, + { + "epoch": 0.6268425135764158, + "grad_norm": 6.97181510925293, + "learning_rate": 3.310810810810811e-05, + "loss": 0.3304, + "step": 808 + }, + { + "epoch": 0.6276183087664856, + "grad_norm": 8.356297492980957, + "learning_rate": 3.314923619271445e-05, + "loss": 0.38, + "step": 809 + }, + { + "epoch": 0.6283941039565555, + "grad_norm": 13.055412292480469, + "learning_rate": 3.31903642773208e-05, + "loss": 1.2422, + "step": 810 + }, + { + "epoch": 0.6291698991466252, + "grad_norm": 12.496575355529785, + "learning_rate": 3.323149236192714e-05, + "loss": 1.0591, + "step": 811 + }, + { + "epoch": 0.6299456943366951, + "grad_norm": 8.216036796569824, + "learning_rate": 3.327262044653349e-05, + "loss": 0.4432, + "step": 812 + }, + { + "epoch": 0.630721489526765, + "grad_norm": 9.185450553894043, + "learning_rate": 3.331374853113983e-05, + "loss": 0.6122, + "step": 813 + }, + { + "epoch": 0.6314972847168348, + "grad_norm": 0.9226237535476685, + "learning_rate": 3.335487661574618e-05, + "loss": 0.0425, + "step": 814 + }, + { + "epoch": 0.6322730799069046, + "grad_norm": 15.850872039794922, + "learning_rate": 3.3396004700352524e-05, + "loss": 1.5015, + "step": 815 + }, + { + "epoch": 0.6330488750969744, + "grad_norm": 7.984287261962891, + "learning_rate": 3.343713278495887e-05, + "loss": 0.4955, + "step": 816 + }, + { + "epoch": 0.6338246702870443, + "grad_norm": 16.100154876708984, + "learning_rate": 3.3478260869565214e-05, + "loss": 1.3967, + "step": 817 + }, + { + "epoch": 0.634600465477114, + "grad_norm": 6.398983001708984, + "learning_rate": 3.351938895417156e-05, + "loss": 0.2778, + "step": 818 + }, + { + "epoch": 0.6353762606671839, + "grad_norm": 8.271710395812988, + "learning_rate": 3.3560517038777905e-05, + "loss": 0.4304, + "step": 819 + }, + { + "epoch": 0.6361520558572537, + "grad_norm": 10.139490127563477, + "learning_rate": 3.360164512338425e-05, + "loss": 0.7362, + "step": 820 + }, + { + "epoch": 0.6369278510473235, + "grad_norm": 8.01966667175293, + "learning_rate": 3.3642773207990596e-05, + "loss": 0.5542, + "step": 821 + }, + { + "epoch": 0.6377036462373933, + "grad_norm": 2.7178361415863037, + "learning_rate": 3.368390129259694e-05, + "loss": 0.0701, + "step": 822 + }, + { + "epoch": 0.6384794414274632, + "grad_norm": 7.912689208984375, + "learning_rate": 3.3725029377203286e-05, + "loss": 0.5658, + "step": 823 + }, + { + "epoch": 0.6392552366175329, + "grad_norm": 6.892507553100586, + "learning_rate": 3.376615746180963e-05, + "loss": 0.3145, + "step": 824 + }, + { + "epoch": 0.6400310318076028, + "grad_norm": 6.382019519805908, + "learning_rate": 3.380728554641598e-05, + "loss": 0.2661, + "step": 825 + }, + { + "epoch": 0.6408068269976727, + "grad_norm": 11.254096984863281, + "learning_rate": 3.384841363102232e-05, + "loss": 1.0098, + "step": 826 + }, + { + "epoch": 0.6415826221877424, + "grad_norm": 2.080343246459961, + "learning_rate": 3.388954171562867e-05, + "loss": 0.1648, + "step": 827 + }, + { + "epoch": 0.6423584173778123, + "grad_norm": 6.863798141479492, + "learning_rate": 3.393066980023501e-05, + "loss": 0.3011, + "step": 828 + }, + { + "epoch": 0.6431342125678821, + "grad_norm": 6.703980922698975, + "learning_rate": 3.3971797884841365e-05, + "loss": 0.2553, + "step": 829 + }, + { + "epoch": 0.6439100077579519, + "grad_norm": 7.8994245529174805, + "learning_rate": 3.40129259694477e-05, + "loss": 0.4249, + "step": 830 + }, + { + "epoch": 0.6446858029480217, + "grad_norm": 9.544623374938965, + "learning_rate": 3.4054054054054055e-05, + "loss": 0.9364, + "step": 831 + }, + { + "epoch": 0.6454615981380916, + "grad_norm": 15.744454383850098, + "learning_rate": 3.4095182138660394e-05, + "loss": 1.5279, + "step": 832 + }, + { + "epoch": 0.6462373933281613, + "grad_norm": 11.321758270263672, + "learning_rate": 3.413631022326674e-05, + "loss": 0.8555, + "step": 833 + }, + { + "epoch": 0.6470131885182312, + "grad_norm": 9.652874946594238, + "learning_rate": 3.417743830787309e-05, + "loss": 0.4081, + "step": 834 + }, + { + "epoch": 0.647788983708301, + "grad_norm": 2.1973342895507812, + "learning_rate": 3.421856639247943e-05, + "loss": 0.0543, + "step": 835 + }, + { + "epoch": 0.6485647788983708, + "grad_norm": 8.358857154846191, + "learning_rate": 3.425969447708578e-05, + "loss": 0.3788, + "step": 836 + }, + { + "epoch": 0.6493405740884407, + "grad_norm": 10.387046813964844, + "learning_rate": 3.430082256169213e-05, + "loss": 0.7464, + "step": 837 + }, + { + "epoch": 0.6501163692785105, + "grad_norm": 10.43363094329834, + "learning_rate": 3.4341950646298466e-05, + "loss": 0.5038, + "step": 838 + }, + { + "epoch": 0.6508921644685803, + "grad_norm": 12.447196006774902, + "learning_rate": 3.438307873090482e-05, + "loss": 1.0619, + "step": 839 + }, + { + "epoch": 0.6516679596586501, + "grad_norm": 12.04200553894043, + "learning_rate": 3.4424206815511156e-05, + "loss": 1.1649, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_Qnli-dev_cosine_accuracy": 0.66015625, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8171392679214478, + "eval_Qnli-dev_cosine_ap": 0.7004489571844912, + "eval_Qnli-dev_cosine_f1": 0.6779661016949152, + "eval_Qnli-dev_cosine_f1_threshold": 0.7190055847167969, + "eval_Qnli-dev_cosine_precision": 0.5326876513317191, + "eval_Qnli-dev_cosine_recall": 0.9322033898305084, + "eval_Qnli-dev_dot_accuracy": 0.619140625, + "eval_Qnli-dev_dot_accuracy_threshold": 406.7884521484375, + "eval_Qnli-dev_dot_ap": 0.590894893134841, + "eval_Qnli-dev_dot_f1": 0.6617862371888725, + "eval_Qnli-dev_dot_f1_threshold": 330.5760803222656, + "eval_Qnli-dev_dot_precision": 0.5055928411633109, + "eval_Qnli-dev_dot_recall": 0.9576271186440678, + "eval_Qnli-dev_euclidean_accuracy": 0.666015625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.058063507080078, + "eval_Qnli-dev_euclidean_ap": 0.7011381154770583, + "eval_Qnli-dev_euclidean_f1": 0.674884437596302, + "eval_Qnli-dev_euclidean_f1_threshold": 17.170625686645508, + "eval_Qnli-dev_euclidean_precision": 0.5302663438256658, + "eval_Qnli-dev_euclidean_recall": 0.9279661016949152, + "eval_Qnli-dev_manhattan_accuracy": 0.66796875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 245.08718872070312, + "eval_Qnli-dev_manhattan_ap": 0.7015853838413395, + "eval_Qnli-dev_manhattan_f1": 0.68362480127186, + "eval_Qnli-dev_manhattan_f1_threshold": 298.2034912109375, + "eval_Qnli-dev_manhattan_precision": 0.5470737913486005, + "eval_Qnli-dev_manhattan_recall": 0.9110169491525424, + "eval_Qnli-dev_max_accuracy": 0.66796875, + "eval_Qnli-dev_max_accuracy_threshold": 406.7884521484375, + "eval_Qnli-dev_max_ap": 0.7015853838413395, + "eval_Qnli-dev_max_f1": 0.68362480127186, + "eval_Qnli-dev_max_f1_threshold": 330.5760803222656, + "eval_Qnli-dev_max_precision": 0.5470737913486005, + "eval_Qnli-dev_max_recall": 0.9576271186440678, + "eval_allNLI-dev_cosine_accuracy": 0.720703125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9094138741493225, + "eval_allNLI-dev_cosine_ap": 0.5949242895917363, + "eval_allNLI-dev_cosine_f1": 0.6045340050377833, + "eval_allNLI-dev_cosine_f1_threshold": 0.848529577255249, + "eval_allNLI-dev_cosine_precision": 0.5357142857142857, + "eval_allNLI-dev_cosine_recall": 0.6936416184971098, + "eval_allNLI-dev_dot_accuracy": 0.6875, + "eval_allNLI-dev_dot_accuracy_threshold": 465.26519775390625, + "eval_allNLI-dev_dot_ap": 0.46427556240386536, + "eval_allNLI-dev_dot_f1": 0.552, + "eval_allNLI-dev_dot_f1_threshold": 357.8551940917969, + "eval_allNLI-dev_dot_precision": 0.42201834862385323, + "eval_allNLI-dev_dot_recall": 0.7976878612716763, + "eval_allNLI-dev_euclidean_accuracy": 0.71875, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.009450912475586, + "eval_allNLI-dev_euclidean_ap": 0.594722905619412, + "eval_allNLI-dev_euclidean_f1": 0.6150234741784038, + "eval_allNLI-dev_euclidean_f1_threshold": 12.564327239990234, + "eval_allNLI-dev_euclidean_precision": 0.5177865612648221, + "eval_allNLI-dev_euclidean_recall": 0.7572254335260116, + "eval_allNLI-dev_manhattan_accuracy": 0.716796875, + "eval_allNLI-dev_manhattan_accuracy_threshold": 169.49075317382812, + "eval_allNLI-dev_manhattan_ap": 0.5947162276273193, + "eval_allNLI-dev_manhattan_f1": 0.6195121951219511, + "eval_allNLI-dev_manhattan_f1_threshold": 220.03335571289062, + "eval_allNLI-dev_manhattan_precision": 0.5358649789029536, + "eval_allNLI-dev_manhattan_recall": 0.7341040462427746, + "eval_allNLI-dev_max_accuracy": 0.720703125, + "eval_allNLI-dev_max_accuracy_threshold": 465.26519775390625, + "eval_allNLI-dev_max_ap": 0.5949242895917363, + "eval_allNLI-dev_max_f1": 0.6195121951219511, + "eval_allNLI-dev_max_f1_threshold": 357.8551940917969, + "eval_allNLI-dev_max_precision": 0.5358649789029536, + "eval_allNLI-dev_max_recall": 0.7976878612716763, + "eval_sequential_score": 0.7015853838413395, + "eval_sts-test_pearson_cosine": 0.8029319103047167, + "eval_sts-test_pearson_dot": 0.6656092981751172, + "eval_sts-test_pearson_euclidean": 0.8376380605294567, + "eval_sts-test_pearson_manhattan": 0.8358672042293024, + "eval_sts-test_pearson_max": 0.8376380605294567, + "eval_sts-test_spearman_cosine": 0.8419916714858389, + "eval_sts-test_spearman_dot": 0.6479591075937474, + "eval_sts-test_spearman_euclidean": 0.8343366549660103, + "eval_sts-test_spearman_manhattan": 0.8360729591567352, + "eval_sts-test_spearman_max": 0.8419916714858389, + "eval_vitaminc-pairs_loss": 3.7601094245910645, + "eval_vitaminc-pairs_runtime": 5.4742, + "eval_vitaminc-pairs_samples_per_second": 23.382, + "eval_vitaminc-pairs_steps_per_second": 0.365, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_negation-triplets_loss": 1.5633906126022339, + "eval_negation-triplets_runtime": 1.1066, + "eval_negation-triplets_samples_per_second": 115.666, + "eval_negation-triplets_steps_per_second": 1.807, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_scitail-pairs-pos_loss": 0.16826407611370087, + "eval_scitail-pairs-pos_runtime": 1.2384, + "eval_scitail-pairs-pos_samples_per_second": 103.361, + "eval_scitail-pairs-pos_steps_per_second": 1.615, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_scitail-pairs-qa_loss": 0.004816232714802027, + "eval_scitail-pairs-qa_runtime": 0.903, + "eval_scitail-pairs-qa_samples_per_second": 141.756, + "eval_scitail-pairs-qa_steps_per_second": 2.215, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_xsum-pairs_loss": 0.3080148994922638, + "eval_xsum-pairs_runtime": 4.3137, + "eval_xsum-pairs_samples_per_second": 29.673, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_sciq_pairs_loss": 0.08924353122711182, + "eval_sciq_pairs_runtime": 5.5271, + "eval_sciq_pairs_samples_per_second": 23.159, + "eval_sciq_pairs_steps_per_second": 0.362, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_qasc_pairs_loss": 0.5605460405349731, + "eval_qasc_pairs_runtime": 0.9785, + "eval_qasc_pairs_samples_per_second": 130.815, + "eval_qasc_pairs_steps_per_second": 2.044, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_openbookqa_pairs_loss": 1.6154065132141113, + "eval_openbookqa_pairs_runtime": 0.8681, + "eval_openbookqa_pairs_samples_per_second": 147.444, + "eval_openbookqa_pairs_steps_per_second": 2.304, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_msmarco_pairs_loss": 0.929015040397644, + "eval_msmarco_pairs_runtime": 1.9584, + "eval_msmarco_pairs_samples_per_second": 65.359, + "eval_msmarco_pairs_steps_per_second": 1.021, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_nq_pairs_loss": 1.1485563516616821, + "eval_nq_pairs_runtime": 4.409, + "eval_nq_pairs_samples_per_second": 29.032, + "eval_nq_pairs_steps_per_second": 0.454, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_trivia_pairs_loss": 0.8550643920898438, + "eval_trivia_pairs_runtime": 4.5197, + "eval_trivia_pairs_samples_per_second": 28.32, + "eval_trivia_pairs_steps_per_second": 0.443, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_gooaq_pairs_loss": 0.811971127986908, + "eval_gooaq_pairs_runtime": 1.2841, + "eval_gooaq_pairs_samples_per_second": 99.683, + "eval_gooaq_pairs_steps_per_second": 1.558, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_paws-pos_loss": 0.043758612126111984, + "eval_paws-pos_runtime": 1.0254, + "eval_paws-pos_samples_per_second": 124.823, + "eval_paws-pos_steps_per_second": 1.95, + "step": 840 + }, + { + "epoch": 0.6516679596586501, + "eval_global_dataset_loss": 0.6404618620872498, + "eval_global_dataset_runtime": 21.1283, + "eval_global_dataset_samples_per_second": 19.689, + "eval_global_dataset_steps_per_second": 0.331, + "step": 840 + }, + { + "epoch": 0.65244375484872, + "grad_norm": 1.4081767797470093, + "learning_rate": 3.446533490011751e-05, + "loss": 0.0749, + "step": 841 + }, + { + "epoch": 0.6532195500387897, + "grad_norm": 14.430989265441895, + "learning_rate": 3.4506462984723853e-05, + "loss": 1.5064, + "step": 842 + }, + { + "epoch": 0.6539953452288596, + "grad_norm": 11.624059677124023, + "learning_rate": 3.454759106933019e-05, + "loss": 1.1311, + "step": 843 + }, + { + "epoch": 0.6547711404189294, + "grad_norm": 11.544733047485352, + "learning_rate": 3.4588719153936544e-05, + "loss": 0.8582, + "step": 844 + }, + { + "epoch": 0.6555469356089992, + "grad_norm": 20.000152587890625, + "learning_rate": 3.462984723854289e-05, + "loss": 2.7894, + "step": 845 + }, + { + "epoch": 0.656322730799069, + "grad_norm": 10.974227905273438, + "learning_rate": 3.4670975323149235e-05, + "loss": 1.0835, + "step": 846 + }, + { + "epoch": 0.6570985259891389, + "grad_norm": 7.944055557250977, + "learning_rate": 3.471210340775558e-05, + "loss": 0.4166, + "step": 847 + }, + { + "epoch": 0.6578743211792087, + "grad_norm": 7.957513332366943, + "learning_rate": 3.4753231492361925e-05, + "loss": 0.8957, + "step": 848 + }, + { + "epoch": 0.6586501163692785, + "grad_norm": 10.607748985290527, + "learning_rate": 3.479435957696827e-05, + "loss": 1.203, + "step": 849 + }, + { + "epoch": 0.6594259115593484, + "grad_norm": 11.543784141540527, + "learning_rate": 3.4835487661574616e-05, + "loss": 1.1378, + "step": 850 + }, + { + "epoch": 0.6602017067494181, + "grad_norm": 10.6654052734375, + "learning_rate": 3.487661574618096e-05, + "loss": 1.144, + "step": 851 + }, + { + "epoch": 0.660977501939488, + "grad_norm": 7.211609840393066, + "learning_rate": 3.4917743830787306e-05, + "loss": 0.5134, + "step": 852 + }, + { + "epoch": 0.6617532971295578, + "grad_norm": 6.215585708618164, + "learning_rate": 3.495887191539365e-05, + "loss": 0.4396, + "step": 853 + }, + { + "epoch": 0.6625290923196276, + "grad_norm": 9.448208808898926, + "learning_rate": 3.5e-05, + "loss": 0.5735, + "step": 854 + }, + { + "epoch": 0.6633048875096974, + "grad_norm": 8.826326370239258, + "learning_rate": 3.499918927178749e-05, + "loss": 0.5101, + "step": 855 + }, + { + "epoch": 0.6640806826997673, + "grad_norm": 8.291367530822754, + "learning_rate": 3.4996757218605996e-05, + "loss": 0.4425, + "step": 856 + }, + { + "epoch": 0.664856477889837, + "grad_norm": 10.373847961425781, + "learning_rate": 3.499270423480235e-05, + "loss": 1.0071, + "step": 857 + }, + { + "epoch": 0.6656322730799069, + "grad_norm": 3.7650482654571533, + "learning_rate": 3.498703097755021e-05, + "loss": 0.1042, + "step": 858 + }, + { + "epoch": 0.6664080682699767, + "grad_norm": 10.80237865447998, + "learning_rate": 3.497973836674353e-05, + "loss": 0.6389, + "step": 859 + }, + { + "epoch": 0.6671838634600465, + "grad_norm": 12.34592342376709, + "learning_rate": 3.4970827584847364e-05, + "loss": 1.3249, + "step": 860 + }, + { + "epoch": 0.6679596586501164, + "grad_norm": 9.470741271972656, + "learning_rate": 3.496030007670617e-05, + "loss": 0.4195, + "step": 861 + }, + { + "epoch": 0.6687354538401862, + "grad_norm": 13.572481155395508, + "learning_rate": 3.494815754930952e-05, + "loss": 0.9902, + "step": 862 + }, + { + "epoch": 0.669511249030256, + "grad_norm": 7.040611267089844, + "learning_rate": 3.493440197151532e-05, + "loss": 0.4288, + "step": 863 + }, + { + "epoch": 0.6702870442203258, + "grad_norm": 7.9452972412109375, + "learning_rate": 3.491903557373057e-05, + "loss": 0.8166, + "step": 864 + }, + { + "epoch": 0.6710628394103957, + "grad_norm": 14.33533000946045, + "learning_rate": 3.4902060847549715e-05, + "loss": 1.4688, + "step": 865 + }, + { + "epoch": 0.6718386346004654, + "grad_norm": 4.840489387512207, + "learning_rate": 3.488348054535063e-05, + "loss": 0.2178, + "step": 866 + }, + { + "epoch": 0.6726144297905353, + "grad_norm": 9.400243759155273, + "learning_rate": 3.4863297679848356e-05, + "loss": 0.9398, + "step": 867 + }, + { + "epoch": 0.6733902249806051, + "grad_norm": 8.981668472290039, + "learning_rate": 3.484151552360659e-05, + "loss": 0.5052, + "step": 868 + }, + { + "epoch": 0.674166020170675, + "grad_norm": 10.604998588562012, + "learning_rate": 3.481813760850706e-05, + "loss": 0.4738, + "step": 869 + }, + { + "epoch": 0.6749418153607447, + "grad_norm": 15.958767890930176, + "learning_rate": 3.47931677251768e-05, + "loss": 1.6114, + "step": 870 + }, + { + "epoch": 0.6757176105508146, + "grad_norm": 9.349385261535645, + "learning_rate": 3.476660992237362e-05, + "loss": 0.5192, + "step": 871 + }, + { + "epoch": 0.6764934057408845, + "grad_norm": 15.109448432922363, + "learning_rate": 3.47384685063295e-05, + "loss": 1.3796, + "step": 872 + }, + { + "epoch": 0.6772692009309542, + "grad_norm": 8.498479843139648, + "learning_rate": 3.4708748040052434e-05, + "loss": 0.4999, + "step": 873 + }, + { + "epoch": 0.6780449961210241, + "grad_norm": 9.053614616394043, + "learning_rate": 3.467745334258652e-05, + "loss": 0.607, + "step": 874 + }, + { + "epoch": 0.6788207913110939, + "grad_norm": 10.35421085357666, + "learning_rate": 3.464458948823058e-05, + "loss": 0.7941, + "step": 875 + }, + { + "epoch": 0.6795965865011637, + "grad_norm": 11.33446979522705, + "learning_rate": 3.4610161805715394e-05, + "loss": 1.2409, + "step": 876 + }, + { + "epoch": 0.6803723816912335, + "grad_norm": 10.35372543334961, + "learning_rate": 3.457417587733967e-05, + "loss": 1.1035, + "step": 877 + }, + { + "epoch": 0.6811481768813034, + "grad_norm": 8.909985542297363, + "learning_rate": 3.453663753806488e-05, + "loss": 0.557, + "step": 878 + }, + { + "epoch": 0.6819239720713731, + "grad_norm": 6.681217670440674, + "learning_rate": 3.449755287456917e-05, + "loss": 0.3722, + "step": 879 + }, + { + "epoch": 0.682699767261443, + "grad_norm": 8.428468704223633, + "learning_rate": 3.4456928224260404e-05, + "loss": 0.6515, + "step": 880 + }, + { + "epoch": 0.6834755624515128, + "grad_norm": 7.6881937980651855, + "learning_rate": 3.441477017424862e-05, + "loss": 0.4847, + "step": 881 + }, + { + "epoch": 0.6842513576415826, + "grad_norm": 7.775545597076416, + "learning_rate": 3.437108556027792e-05, + "loss": 0.4722, + "step": 882 + }, + { + "epoch": 0.6850271528316525, + "grad_norm": 6.918392658233643, + "learning_rate": 3.4325881465618095e-05, + "loss": 0.4156, + "step": 883 + }, + { + "epoch": 0.6858029480217223, + "grad_norm": 12.802634239196777, + "learning_rate": 3.427916521991613e-05, + "loss": 1.2278, + "step": 884 + }, + { + "epoch": 0.6865787432117921, + "grad_norm": 9.124152183532715, + "learning_rate": 3.423094439800769e-05, + "loss": 0.6314, + "step": 885 + }, + { + "epoch": 0.6873545384018619, + "grad_norm": 13.434520721435547, + "learning_rate": 3.4181226818688936e-05, + "loss": 1.5736, + "step": 886 + }, + { + "epoch": 0.6881303335919318, + "grad_norm": 13.541108131408691, + "learning_rate": 3.41300205434487e-05, + "loss": 1.3482, + "step": 887 + }, + { + "epoch": 0.6889061287820015, + "grad_norm": 10.143789291381836, + "learning_rate": 3.4077333875161383e-05, + "loss": 1.2191, + "step": 888 + }, + { + "epoch": 0.6896819239720714, + "grad_norm": 7.859284400939941, + "learning_rate": 3.4023175356740667e-05, + "loss": 0.3178, + "step": 889 + }, + { + "epoch": 0.6904577191621412, + "grad_norm": 8.417284965515137, + "learning_rate": 3.3967553769754325e-05, + "loss": 0.4992, + "step": 890 + }, + { + "epoch": 0.691233514352211, + "grad_norm": 19.698083877563477, + "learning_rate": 3.391047813300031e-05, + "loss": 3.0199, + "step": 891 + }, + { + "epoch": 0.6920093095422808, + "grad_norm": 15.355926513671875, + "learning_rate": 3.385195770104441e-05, + "loss": 1.4033, + "step": 892 + }, + { + "epoch": 0.6927851047323507, + "grad_norm": 2.9927265644073486, + "learning_rate": 3.379200196271967e-05, + "loss": 0.0986, + "step": 893 + }, + { + "epoch": 0.6935608999224204, + "grad_norm": 8.0789213180542, + "learning_rate": 3.3730620639587795e-05, + "loss": 0.6785, + "step": 894 + }, + { + "epoch": 0.6943366951124903, + "grad_norm": 10.094559669494629, + "learning_rate": 3.366782368436288e-05, + "loss": 1.145, + "step": 895 + }, + { + "epoch": 0.6951124903025602, + "grad_norm": 11.94321346282959, + "learning_rate": 3.3603621279297556e-05, + "loss": 1.0562, + "step": 896 + }, + { + "epoch": 0.6958882854926299, + "grad_norm": 8.577461242675781, + "learning_rate": 3.353802383453205e-05, + "loss": 0.6484, + "step": 897 + }, + { + "epoch": 0.6966640806826998, + "grad_norm": 6.783735752105713, + "learning_rate": 3.34710419864062e-05, + "loss": 0.3522, + "step": 898 + }, + { + "epoch": 0.6974398758727696, + "grad_norm": 11.289161682128906, + "learning_rate": 3.340268659573479e-05, + "loss": 0.9692, + "step": 899 + }, + { + "epoch": 0.6982156710628394, + "grad_norm": 13.973520278930664, + "learning_rate": 3.333296874604657e-05, + "loss": 1.4612, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_Qnli-dev_cosine_accuracy": 0.669921875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8599634170532227, + "eval_Qnli-dev_cosine_ap": 0.6970089692030526, + "eval_Qnli-dev_cosine_f1": 0.6777609682299547, + "eval_Qnli-dev_cosine_f1_threshold": 0.7281531095504761, + "eval_Qnli-dev_cosine_precision": 0.5270588235294118, + "eval_Qnli-dev_cosine_recall": 0.9491525423728814, + "eval_Qnli-dev_dot_accuracy": 0.61328125, + "eval_Qnli-dev_dot_accuracy_threshold": 492.053466796875, + "eval_Qnli-dev_dot_ap": 0.6053211345300911, + "eval_Qnli-dev_dot_f1": 0.6645367412140575, + "eval_Qnli-dev_dot_f1_threshold": 393.0693664550781, + "eval_Qnli-dev_dot_precision": 0.5333333333333333, + "eval_Qnli-dev_dot_recall": 0.8813559322033898, + "eval_Qnli-dev_euclidean_accuracy": 0.673828125, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.88613510131836, + "eval_Qnli-dev_euclidean_ap": 0.6954464383341553, + "eval_Qnli-dev_euclidean_f1": 0.6738794435857806, + "eval_Qnli-dev_euclidean_f1_threshold": 16.379722595214844, + "eval_Qnli-dev_euclidean_precision": 0.5304136253041363, + "eval_Qnli-dev_euclidean_recall": 0.923728813559322, + "eval_Qnli-dev_manhattan_accuracy": 0.677734375, + "eval_Qnli-dev_manhattan_accuracy_threshold": 205.11907958984375, + "eval_Qnli-dev_manhattan_ap": 0.6971539032468665, + "eval_Qnli-dev_manhattan_f1": 0.6686838124054463, + "eval_Qnli-dev_manhattan_f1_threshold": 304.216796875, + "eval_Qnli-dev_manhattan_precision": 0.52, + "eval_Qnli-dev_manhattan_recall": 0.9364406779661016, + "eval_Qnli-dev_max_accuracy": 0.677734375, + "eval_Qnli-dev_max_accuracy_threshold": 492.053466796875, + "eval_Qnli-dev_max_ap": 0.6971539032468665, + "eval_Qnli-dev_max_f1": 0.6777609682299547, + "eval_Qnli-dev_max_f1_threshold": 393.0693664550781, + "eval_Qnli-dev_max_precision": 0.5333333333333333, + "eval_Qnli-dev_max_recall": 0.9491525423728814, + "eval_allNLI-dev_cosine_accuracy": 0.708984375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8860212564468384, + "eval_allNLI-dev_cosine_ap": 0.5822880366182692, + "eval_allNLI-dev_cosine_f1": 0.5902439024390244, + "eval_allNLI-dev_cosine_f1_threshold": 0.8557306528091431, + "eval_allNLI-dev_cosine_precision": 0.510548523206751, + "eval_allNLI-dev_cosine_recall": 0.6994219653179191, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 482.4095458984375, + "eval_allNLI-dev_dot_ap": 0.46741601433748925, + "eval_allNLI-dev_dot_f1": 0.5562130177514792, + "eval_allNLI-dev_dot_f1_threshold": 382.43414306640625, + "eval_allNLI-dev_dot_precision": 0.4221556886227545, + "eval_allNLI-dev_dot_recall": 0.815028901734104, + "eval_allNLI-dev_euclidean_accuracy": 0.70703125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.381266593933105, + "eval_allNLI-dev_euclidean_ap": 0.5841659284694745, + "eval_allNLI-dev_euclidean_f1": 0.6012269938650306, + "eval_allNLI-dev_euclidean_f1_threshold": 13.686376571655273, + "eval_allNLI-dev_euclidean_precision": 0.4651898734177215, + "eval_allNLI-dev_euclidean_recall": 0.8497109826589595, + "eval_allNLI-dev_manhattan_accuracy": 0.70703125, + "eval_allNLI-dev_manhattan_accuracy_threshold": 182.92349243164062, + "eval_allNLI-dev_manhattan_ap": 0.5850095772432237, + "eval_allNLI-dev_manhattan_f1": 0.6120689655172413, + "eval_allNLI-dev_manhattan_f1_threshold": 235.74771118164062, + "eval_allNLI-dev_manhattan_precision": 0.4879725085910653, + "eval_allNLI-dev_manhattan_recall": 0.8208092485549133, + "eval_allNLI-dev_max_accuracy": 0.708984375, + "eval_allNLI-dev_max_accuracy_threshold": 482.4095458984375, + "eval_allNLI-dev_max_ap": 0.5850095772432237, + "eval_allNLI-dev_max_f1": 0.6120689655172413, + "eval_allNLI-dev_max_f1_threshold": 382.43414306640625, + "eval_allNLI-dev_max_precision": 0.510548523206751, + "eval_allNLI-dev_max_recall": 0.8497109826589595, + "eval_sequential_score": 0.6971539032468665, + "eval_sts-test_pearson_cosine": 0.8130720582126042, + "eval_sts-test_pearson_dot": 0.7203056283513845, + "eval_sts-test_pearson_euclidean": 0.8460767778486171, + "eval_sts-test_pearson_manhattan": 0.8441398690416491, + "eval_sts-test_pearson_max": 0.8460767778486171, + "eval_sts-test_spearman_cosine": 0.8515779734422747, + "eval_sts-test_spearman_dot": 0.7067312971891635, + "eval_sts-test_spearman_euclidean": 0.8440778410381218, + "eval_sts-test_spearman_manhattan": 0.8434438499179897, + "eval_sts-test_spearman_max": 0.8515779734422747, + "eval_vitaminc-pairs_loss": 4.034060478210449, + "eval_vitaminc-pairs_runtime": 5.4578, + "eval_vitaminc-pairs_samples_per_second": 23.453, + "eval_vitaminc-pairs_steps_per_second": 0.366, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_negation-triplets_loss": 1.5019736289978027, + "eval_negation-triplets_runtime": 1.1091, + "eval_negation-triplets_samples_per_second": 115.411, + "eval_negation-triplets_steps_per_second": 1.803, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_scitail-pairs-pos_loss": 0.22966919839382172, + "eval_scitail-pairs-pos_runtime": 1.3428, + "eval_scitail-pairs-pos_samples_per_second": 95.323, + "eval_scitail-pairs-pos_steps_per_second": 1.489, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_scitail-pairs-qa_loss": 0.01810869202017784, + "eval_scitail-pairs-qa_runtime": 0.92, + "eval_scitail-pairs-qa_samples_per_second": 139.137, + "eval_scitail-pairs-qa_steps_per_second": 2.174, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_xsum-pairs_loss": 0.40340137481689453, + "eval_xsum-pairs_runtime": 4.3116, + "eval_xsum-pairs_samples_per_second": 29.687, + "eval_xsum-pairs_steps_per_second": 0.464, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_sciq_pairs_loss": 0.07373710721731186, + "eval_sciq_pairs_runtime": 5.5129, + "eval_sciq_pairs_samples_per_second": 23.218, + "eval_sciq_pairs_steps_per_second": 0.363, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_qasc_pairs_loss": 0.5565303564071655, + "eval_qasc_pairs_runtime": 0.9752, + "eval_qasc_pairs_samples_per_second": 131.262, + "eval_qasc_pairs_steps_per_second": 2.051, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_openbookqa_pairs_loss": 1.4768919944763184, + "eval_openbookqa_pairs_runtime": 0.8576, + "eval_openbookqa_pairs_samples_per_second": 149.25, + "eval_openbookqa_pairs_steps_per_second": 2.332, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_msmarco_pairs_loss": 0.998272716999054, + "eval_msmarco_pairs_runtime": 1.9564, + "eval_msmarco_pairs_samples_per_second": 65.426, + "eval_msmarco_pairs_steps_per_second": 1.022, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_nq_pairs_loss": 1.0959526300430298, + "eval_nq_pairs_runtime": 4.4205, + "eval_nq_pairs_samples_per_second": 28.956, + "eval_nq_pairs_steps_per_second": 0.452, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_trivia_pairs_loss": 0.7414935231208801, + "eval_trivia_pairs_runtime": 4.5257, + "eval_trivia_pairs_samples_per_second": 28.283, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_gooaq_pairs_loss": 0.8144569993019104, + "eval_gooaq_pairs_runtime": 1.266, + "eval_gooaq_pairs_samples_per_second": 101.104, + "eval_gooaq_pairs_steps_per_second": 1.58, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_paws-pos_loss": 0.04649306833744049, + "eval_paws-pos_runtime": 1.0246, + "eval_paws-pos_samples_per_second": 124.93, + "eval_paws-pos_steps_per_second": 1.952, + "step": 900 + }, + { + "epoch": 0.6982156710628394, + "eval_global_dataset_loss": 0.6019662618637085, + "eval_global_dataset_runtime": 21.1621, + "eval_global_dataset_samples_per_second": 19.658, + "eval_global_dataset_steps_per_second": 0.331, + "step": 900 + }, + { + "epoch": 0.6989914662529092, + "grad_norm": 6.40172004699707, + "learning_rate": 3.326189974178707e-05, + "loss": 0.2964, + "step": 901 + }, + { + "epoch": 0.6997672614429791, + "grad_norm": 7.125335693359375, + "learning_rate": 3.318949110648564e-05, + "loss": 0.3666, + "step": 902 + }, + { + "epoch": 0.7005430566330488, + "grad_norm": 6.8976850509643555, + "learning_rate": 3.311575458088698e-05, + "loss": 0.3556, + "step": 903 + }, + { + "epoch": 0.7013188518231187, + "grad_norm": 7.64393424987793, + "learning_rate": 3.3040702121047414e-05, + "loss": 0.4661, + "step": 904 + }, + { + "epoch": 0.7020946470131885, + "grad_norm": 8.273258209228516, + "learning_rate": 3.296434589639626e-05, + "loss": 0.3754, + "step": 905 + }, + { + "epoch": 0.7028704422032583, + "grad_norm": 7.339361667633057, + "learning_rate": 3.2886698287762624e-05, + "loss": 0.3114, + "step": 906 + }, + { + "epoch": 0.7036462373933282, + "grad_norm": 10.890863418579102, + "learning_rate": 3.2807771885367885e-05, + "loss": 1.0232, + "step": 907 + }, + { + "epoch": 0.704422032583398, + "grad_norm": 11.602267265319824, + "learning_rate": 3.272757948678429e-05, + "loss": 1.1533, + "step": 908 + }, + { + "epoch": 0.7051978277734678, + "grad_norm": 7.152297019958496, + "learning_rate": 3.264613409485981e-05, + "loss": 0.3414, + "step": 909 + }, + { + "epoch": 0.7059736229635376, + "grad_norm": 8.687931060791016, + "learning_rate": 3.2563448915609865e-05, + "loss": 0.5628, + "step": 910 + }, + { + "epoch": 0.7067494181536075, + "grad_norm": 12.789813041687012, + "learning_rate": 3.247953735607596e-05, + "loss": 1.3947, + "step": 911 + }, + { + "epoch": 0.7075252133436772, + "grad_norm": 8.150513648986816, + "learning_rate": 3.2394413022151834e-05, + "loss": 0.4052, + "step": 912 + }, + { + "epoch": 0.7083010085337471, + "grad_norm": 11.772473335266113, + "learning_rate": 3.23080897163773e-05, + "loss": 0.9239, + "step": 913 + }, + { + "epoch": 0.7090768037238169, + "grad_norm": 9.580940246582031, + "learning_rate": 3.222058143570024e-05, + "loss": 0.5245, + "step": 914 + }, + { + "epoch": 0.7098525989138867, + "grad_norm": 9.422074317932129, + "learning_rate": 3.213190236920705e-05, + "loss": 0.8049, + "step": 915 + }, + { + "epoch": 0.7106283941039565, + "grad_norm": 11.65447998046875, + "learning_rate": 3.204206689582193e-05, + "loss": 1.1185, + "step": 916 + }, + { + "epoch": 0.7114041892940264, + "grad_norm": 11.373902320861816, + "learning_rate": 3.195108958197544e-05, + "loss": 0.9616, + "step": 917 + }, + { + "epoch": 0.7121799844840963, + "grad_norm": 6.700024604797363, + "learning_rate": 3.18589851792426e-05, + "loss": 0.3811, + "step": 918 + }, + { + "epoch": 0.712955779674166, + "grad_norm": 6.899729251861572, + "learning_rate": 3.176576862195095e-05, + "loss": 0.3467, + "step": 919 + }, + { + "epoch": 0.7137315748642359, + "grad_norm": 11.678479194641113, + "learning_rate": 3.1671455024759065e-05, + "loss": 1.2782, + "step": 920 + }, + { + "epoch": 0.7145073700543056, + "grad_norm": 8.699654579162598, + "learning_rate": 3.157605968020577e-05, + "loss": 0.5922, + "step": 921 + }, + { + "epoch": 0.7152831652443755, + "grad_norm": 11.387981414794922, + "learning_rate": 3.147959805623049e-05, + "loss": 0.9406, + "step": 922 + }, + { + "epoch": 0.7160589604344453, + "grad_norm": 18.934412002563477, + "learning_rate": 3.138208579366523e-05, + "loss": 0.9116, + "step": 923 + }, + { + "epoch": 0.7168347556245152, + "grad_norm": 6.137956142425537, + "learning_rate": 3.1283538703698435e-05, + "loss": 0.3033, + "step": 924 + }, + { + "epoch": 0.7176105508145849, + "grad_norm": 8.020730972290039, + "learning_rate": 3.118397276531135e-05, + "loss": 0.5023, + "step": 925 + }, + { + "epoch": 0.7183863460046548, + "grad_norm": 2.2567999362945557, + "learning_rate": 3.1083404122687016e-05, + "loss": 0.0792, + "step": 926 + }, + { + "epoch": 0.7191621411947245, + "grad_norm": 8.43709659576416, + "learning_rate": 3.09818490825926e-05, + "loss": 0.4333, + "step": 927 + }, + { + "epoch": 0.7199379363847944, + "grad_norm": 12.493663787841797, + "learning_rate": 3.0879324111735335e-05, + "loss": 1.0547, + "step": 928 + }, + { + "epoch": 0.7207137315748642, + "grad_norm": 6.025222301483154, + "learning_rate": 3.0775845834092494e-05, + "loss": 0.232, + "step": 929 + }, + { + "epoch": 0.721489526764934, + "grad_norm": 12.016132354736328, + "learning_rate": 3.067143102821589e-05, + "loss": 0.9777, + "step": 930 + }, + { + "epoch": 0.7222653219550039, + "grad_norm": 0.7246423363685608, + "learning_rate": 3.05660966245113e-05, + "loss": 0.0284, + "step": 931 + }, + { + "epoch": 0.7230411171450737, + "grad_norm": 9.537558555603027, + "learning_rate": 3.04598597024933e-05, + "loss": 0.7501, + "step": 932 + }, + { + "epoch": 0.7238169123351436, + "grad_norm": 11.041244506835938, + "learning_rate": 3.0352737488015874e-05, + "loss": 0.7604, + "step": 933 + }, + { + "epoch": 0.7245927075252133, + "grad_norm": 10.2113037109375, + "learning_rate": 3.0244747350479293e-05, + "loss": 0.5379, + "step": 934 + }, + { + "epoch": 0.7253685027152832, + "grad_norm": 10.146904945373535, + "learning_rate": 3.013590680001381e-05, + "loss": 0.4961, + "step": 935 + }, + { + "epoch": 0.726144297905353, + "grad_norm": 13.39676284790039, + "learning_rate": 3.002623348464041e-05, + "loss": 1.1144, + "step": 936 + }, + { + "epoch": 0.7269200930954228, + "grad_norm": 8.934435844421387, + "learning_rate": 2.991574518740927e-05, + "loss": 0.2896, + "step": 937 + }, + { + "epoch": 0.7276958882854926, + "grad_norm": 7.973369598388672, + "learning_rate": 2.9804459823516345e-05, + "loss": 0.6437, + "step": 938 + }, + { + "epoch": 0.7284716834755625, + "grad_norm": 2.058208703994751, + "learning_rate": 2.969239543739845e-05, + "loss": 0.0315, + "step": 939 + }, + { + "epoch": 0.7292474786656322, + "grad_norm": 9.742630004882812, + "learning_rate": 2.9579570199807485e-05, + "loss": 0.4572, + "step": 940 + }, + { + "epoch": 0.7300232738557021, + "grad_norm": 15.449213981628418, + "learning_rate": 2.9466002404864087e-05, + "loss": 1.3591, + "step": 941 + }, + { + "epoch": 0.730799069045772, + "grad_norm": 12.771681785583496, + "learning_rate": 2.9351710467091334e-05, + "loss": 1.3463, + "step": 942 + }, + { + "epoch": 0.7315748642358417, + "grad_norm": 13.672371864318848, + "learning_rate": 2.9236712918428907e-05, + "loss": 1.0494, + "step": 943 + }, + { + "epoch": 0.7323506594259116, + "grad_norm": 8.407612800598145, + "learning_rate": 2.9121028405228227e-05, + "loss": 0.4497, + "step": 944 + }, + { + "epoch": 0.7331264546159814, + "grad_norm": 7.379111289978027, + "learning_rate": 2.9004675685229002e-05, + "loss": 0.2641, + "step": 945 + }, + { + "epoch": 0.7339022498060512, + "grad_norm": 13.915177345275879, + "learning_rate": 2.888767362451779e-05, + "loss": 0.9394, + "step": 946 + }, + { + "epoch": 0.734678044996121, + "grad_norm": 8.263601303100586, + "learning_rate": 2.8770041194468888e-05, + "loss": 0.5147, + "step": 947 + }, + { + "epoch": 0.7354538401861909, + "grad_norm": 7.944752216339111, + "learning_rate": 2.8651797468668243e-05, + "loss": 0.3103, + "step": 948 + }, + { + "epoch": 0.7362296353762606, + "grad_norm": 9.120464324951172, + "learning_rate": 2.853296161982076e-05, + "loss": 0.4554, + "step": 949 + }, + { + "epoch": 0.7370054305663305, + "grad_norm": 1.248524785041809, + "learning_rate": 2.8413552916641485e-05, + "loss": 0.0854, + "step": 950 + }, + { + "epoch": 0.7377812257564003, + "grad_norm": 17.106124877929688, + "learning_rate": 2.8293590720731328e-05, + "loss": 2.4657, + "step": 951 + }, + { + "epoch": 0.7385570209464701, + "grad_norm": 6.867243766784668, + "learning_rate": 2.817309448343762e-05, + "loss": 0.3617, + "step": 952 + }, + { + "epoch": 0.73933281613654, + "grad_norm": 8.823907852172852, + "learning_rate": 2.805208374270017e-05, + "loss": 0.3915, + "step": 953 + }, + { + "epoch": 0.7401086113266098, + "grad_norm": 7.573917388916016, + "learning_rate": 2.793057811988329e-05, + "loss": 0.33, + "step": 954 + }, + { + "epoch": 0.7408844065166796, + "grad_norm": 10.192573547363281, + "learning_rate": 2.7808597316594256e-05, + "loss": 0.6429, + "step": 955 + }, + { + "epoch": 0.7416602017067494, + "grad_norm": 8.470539093017578, + "learning_rate": 2.7686161111488793e-05, + "loss": 0.5283, + "step": 956 + }, + { + "epoch": 0.7424359968968193, + "grad_norm": 7.523329734802246, + "learning_rate": 2.7563289357064046e-05, + "loss": 0.3144, + "step": 957 + }, + { + "epoch": 0.743211792086889, + "grad_norm": 6.199915885925293, + "learning_rate": 2.744000197643958e-05, + "loss": 0.267, + "step": 958 + }, + { + "epoch": 0.7439875872769589, + "grad_norm": 1.560442328453064, + "learning_rate": 2.731631896012692e-05, + "loss": 0.0828, + "step": 959 + }, + { + "epoch": 0.7447633824670287, + "grad_norm": 5.971311092376709, + "learning_rate": 2.719226036278824e-05, + "loss": 0.2501, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_Qnli-dev_cosine_accuracy": 0.666015625, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8474994897842407, + "eval_Qnli-dev_cosine_ap": 0.694785363238134, + "eval_Qnli-dev_cosine_f1": 0.6824324324324325, + "eval_Qnli-dev_cosine_f1_threshold": 0.776332676410675, + "eval_Qnli-dev_cosine_precision": 0.5674157303370787, + "eval_Qnli-dev_cosine_recall": 0.8559322033898306, + "eval_Qnli-dev_dot_accuracy": 0.623046875, + "eval_Qnli-dev_dot_accuracy_threshold": 405.31781005859375, + "eval_Qnli-dev_dot_ap": 0.6139728282348256, + "eval_Qnli-dev_dot_f1": 0.6644628099173553, + "eval_Qnli-dev_dot_f1_threshold": 384.90911865234375, + "eval_Qnli-dev_dot_precision": 0.5447154471544715, + "eval_Qnli-dev_dot_recall": 0.8516949152542372, + "eval_Qnli-dev_euclidean_accuracy": 0.669921875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.822232246398926, + "eval_Qnli-dev_euclidean_ap": 0.6966373708949628, + "eval_Qnli-dev_euclidean_f1": 0.6881720430107527, + "eval_Qnli-dev_euclidean_f1_threshold": 14.471603393554688, + "eval_Qnli-dev_euclidean_precision": 0.5962732919254659, + "eval_Qnli-dev_euclidean_recall": 0.8135593220338984, + "eval_Qnli-dev_manhattan_accuracy": 0.67578125, + "eval_Qnli-dev_manhattan_accuracy_threshold": 232.7375030517578, + "eval_Qnli-dev_manhattan_ap": 0.6986863970447608, + "eval_Qnli-dev_manhattan_f1": 0.6803418803418803, + "eval_Qnli-dev_manhattan_f1_threshold": 269.2599182128906, + "eval_Qnli-dev_manhattan_precision": 0.5702005730659025, + "eval_Qnli-dev_manhattan_recall": 0.8432203389830508, + "eval_Qnli-dev_max_accuracy": 0.67578125, + "eval_Qnli-dev_max_accuracy_threshold": 405.31781005859375, + "eval_Qnli-dev_max_ap": 0.6986863970447608, + "eval_Qnli-dev_max_f1": 0.6881720430107527, + "eval_Qnli-dev_max_f1_threshold": 384.90911865234375, + "eval_Qnli-dev_max_precision": 0.5962732919254659, + "eval_Qnli-dev_max_recall": 0.8559322033898306, + "eval_allNLI-dev_cosine_accuracy": 0.716796875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.865006685256958, + "eval_allNLI-dev_cosine_ap": 0.5951891924090309, + "eval_allNLI-dev_cosine_f1": 0.6153846153846153, + "eval_allNLI-dev_cosine_f1_threshold": 0.8600099086761475, + "eval_allNLI-dev_cosine_precision": 0.5686274509803921, + "eval_allNLI-dev_cosine_recall": 0.6705202312138728, + "eval_allNLI-dev_dot_accuracy": 0.677734375, + "eval_allNLI-dev_dot_accuracy_threshold": 449.0525207519531, + "eval_allNLI-dev_dot_ap": 0.46854332466428344, + "eval_allNLI-dev_dot_f1": 0.5486381322957199, + "eval_allNLI-dev_dot_f1_threshold": 377.36712646484375, + "eval_allNLI-dev_dot_precision": 0.41348973607038125, + "eval_allNLI-dev_dot_recall": 0.815028901734104, + "eval_allNLI-dev_euclidean_accuracy": 0.708984375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 7.7555832862854, + "eval_allNLI-dev_euclidean_ap": 0.5970028100643333, + "eval_allNLI-dev_euclidean_f1": 0.6019417475728155, + "eval_allNLI-dev_euclidean_f1_threshold": 12.28809928894043, + "eval_allNLI-dev_euclidean_precision": 0.5188284518828452, + "eval_allNLI-dev_euclidean_recall": 0.7167630057803468, + "eval_allNLI-dev_manhattan_accuracy": 0.712890625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 152.32928466796875, + "eval_allNLI-dev_manhattan_ap": 0.5989976183013819, + "eval_allNLI-dev_manhattan_f1": 0.6045340050377833, + "eval_allNLI-dev_manhattan_f1_threshold": 216.45840454101562, + "eval_allNLI-dev_manhattan_precision": 0.5357142857142857, + "eval_allNLI-dev_manhattan_recall": 0.6936416184971098, + "eval_allNLI-dev_max_accuracy": 0.716796875, + "eval_allNLI-dev_max_accuracy_threshold": 449.0525207519531, + "eval_allNLI-dev_max_ap": 0.5989976183013819, + "eval_allNLI-dev_max_f1": 0.6153846153846153, + "eval_allNLI-dev_max_f1_threshold": 377.36712646484375, + "eval_allNLI-dev_max_precision": 0.5686274509803921, + "eval_allNLI-dev_max_recall": 0.815028901734104, + "eval_sequential_score": 0.6986863970447608, + "eval_sts-test_pearson_cosine": 0.8264792903026948, + "eval_sts-test_pearson_dot": 0.7246549410146222, + "eval_sts-test_pearson_euclidean": 0.8600755649878052, + "eval_sts-test_pearson_manhattan": 0.8594835704083789, + "eval_sts-test_pearson_max": 0.8600755649878052, + "eval_sts-test_spearman_cosine": 0.862024260388869, + "eval_sts-test_spearman_dot": 0.7217391249606447, + "eval_sts-test_spearman_euclidean": 0.8554596083528418, + "eval_sts-test_spearman_manhattan": 0.8565551639829647, + "eval_sts-test_spearman_max": 0.862024260388869, + "eval_vitaminc-pairs_loss": 3.7843265533447266, + "eval_vitaminc-pairs_runtime": 5.5207, + "eval_vitaminc-pairs_samples_per_second": 23.185, + "eval_vitaminc-pairs_steps_per_second": 0.362, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_negation-triplets_loss": 1.5218513011932373, + "eval_negation-triplets_runtime": 1.127, + "eval_negation-triplets_samples_per_second": 113.578, + "eval_negation-triplets_steps_per_second": 1.775, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_scitail-pairs-pos_loss": 0.19146156311035156, + "eval_scitail-pairs-pos_runtime": 1.2634, + "eval_scitail-pairs-pos_samples_per_second": 101.312, + "eval_scitail-pairs-pos_steps_per_second": 1.583, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_scitail-pairs-qa_loss": 0.015504817478358746, + "eval_scitail-pairs-qa_runtime": 0.9236, + "eval_scitail-pairs-qa_samples_per_second": 138.588, + "eval_scitail-pairs-qa_steps_per_second": 2.165, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_xsum-pairs_loss": 0.2775380611419678, + "eval_xsum-pairs_runtime": 4.3203, + "eval_xsum-pairs_samples_per_second": 29.627, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_sciq_pairs_loss": 0.08323033899068832, + "eval_sciq_pairs_runtime": 5.6073, + "eval_sciq_pairs_samples_per_second": 22.827, + "eval_sciq_pairs_steps_per_second": 0.357, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_qasc_pairs_loss": 0.4570181369781494, + "eval_qasc_pairs_runtime": 0.9876, + "eval_qasc_pairs_samples_per_second": 129.612, + "eval_qasc_pairs_steps_per_second": 2.025, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_openbookqa_pairs_loss": 1.298726201057434, + "eval_openbookqa_pairs_runtime": 0.8694, + "eval_openbookqa_pairs_samples_per_second": 147.23, + "eval_openbookqa_pairs_steps_per_second": 2.3, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_msmarco_pairs_loss": 0.885758638381958, + "eval_msmarco_pairs_runtime": 1.9716, + "eval_msmarco_pairs_samples_per_second": 64.921, + "eval_msmarco_pairs_steps_per_second": 1.014, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_nq_pairs_loss": 1.164872407913208, + "eval_nq_pairs_runtime": 4.4597, + "eval_nq_pairs_samples_per_second": 28.701, + "eval_nq_pairs_steps_per_second": 0.448, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_trivia_pairs_loss": 0.6729903221130371, + "eval_trivia_pairs_runtime": 4.5309, + "eval_trivia_pairs_samples_per_second": 28.251, + "eval_trivia_pairs_steps_per_second": 0.441, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_gooaq_pairs_loss": 0.7339937090873718, + "eval_gooaq_pairs_runtime": 1.2879, + "eval_gooaq_pairs_samples_per_second": 99.389, + "eval_gooaq_pairs_steps_per_second": 1.553, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_paws-pos_loss": 0.04508695751428604, + "eval_paws-pos_runtime": 1.037, + "eval_paws-pos_samples_per_second": 123.433, + "eval_paws-pos_steps_per_second": 1.929, + "step": 960 + }, + { + "epoch": 0.7447633824670287, + "eval_global_dataset_loss": 0.574341893196106, + "eval_global_dataset_runtime": 21.1497, + "eval_global_dataset_samples_per_second": 19.669, + "eval_global_dataset_steps_per_second": 0.331, + "step": 960 + }, + { + "epoch": 0.7455391776570985, + "grad_norm": 12.264106750488281, + "learning_rate": 2.706784629998451e-05, + "loss": 1.2049, + "step": 961 + }, + { + "epoch": 0.7463149728471683, + "grad_norm": 13.646756172180176, + "learning_rate": 2.6943096944913864e-05, + "loss": 1.4711, + "step": 962 + }, + { + "epoch": 0.7470907680372382, + "grad_norm": 14.10519790649414, + "learning_rate": 2.681803252514064e-05, + "loss": 1.3998, + "step": 963 + }, + { + "epoch": 0.747866563227308, + "grad_norm": 9.913697242736816, + "learning_rate": 2.6692673319315536e-05, + "loss": 0.8524, + "step": 964 + }, + { + "epoch": 0.7486423584173778, + "grad_norm": 11.893787384033203, + "learning_rate": 2.6567039653887533e-05, + "loss": 1.0452, + "step": 965 + }, + { + "epoch": 0.7494181536074477, + "grad_norm": 8.343534469604492, + "learning_rate": 2.6441151899808012e-05, + "loss": 0.4596, + "step": 966 + }, + { + "epoch": 0.7501939487975174, + "grad_norm": 13.511011123657227, + "learning_rate": 2.6315030469227762e-05, + "loss": 2.3406, + "step": 967 + }, + { + "epoch": 0.7509697439875873, + "grad_norm": 6.6399712562561035, + "learning_rate": 2.6188695812187162e-05, + "loss": 0.3107, + "step": 968 + }, + { + "epoch": 0.7517455391776571, + "grad_norm": 8.817875862121582, + "learning_rate": 2.606216841330036e-05, + "loss": 0.7968, + "step": 969 + }, + { + "epoch": 0.752521334367727, + "grad_norm": 11.40561580657959, + "learning_rate": 2.593546878843373e-05, + "loss": 1.0273, + "step": 970 + }, + { + "epoch": 0.7532971295577967, + "grad_norm": 7.930853366851807, + "learning_rate": 2.5808617481379366e-05, + "loss": 0.8244, + "step": 971 + }, + { + "epoch": 0.7540729247478666, + "grad_norm": 8.278972625732422, + "learning_rate": 2.568163506052394e-05, + "loss": 0.386, + "step": 972 + }, + { + "epoch": 0.7548487199379363, + "grad_norm": 9.533364295959473, + "learning_rate": 2.555454211551368e-05, + "loss": 0.5178, + "step": 973 + }, + { + "epoch": 0.7556245151280062, + "grad_norm": 14.524147033691406, + "learning_rate": 2.5427359253915796e-05, + "loss": 1.3655, + "step": 974 + }, + { + "epoch": 0.756400310318076, + "grad_norm": 16.180606842041016, + "learning_rate": 2.530010709787711e-05, + "loss": 1.6236, + "step": 975 + }, + { + "epoch": 0.7571761055081458, + "grad_norm": 15.53566837310791, + "learning_rate": 2.517280628078022e-05, + "loss": 2.4522, + "step": 976 + }, + { + "epoch": 0.7579519006982157, + "grad_norm": 12.565162658691406, + "learning_rate": 2.5045477443897906e-05, + "loss": 1.0867, + "step": 977 + }, + { + "epoch": 0.7587276958882855, + "grad_norm": 7.926860809326172, + "learning_rate": 2.4918141233046228e-05, + "loss": 0.484, + "step": 978 + }, + { + "epoch": 0.7595034910783554, + "grad_norm": 11.153532028198242, + "learning_rate": 2.4790818295236914e-05, + "loss": 0.8955, + "step": 979 + }, + { + "epoch": 0.7602792862684251, + "grad_norm": 12.23313045501709, + "learning_rate": 2.4663529275329524e-05, + "loss": 1.0314, + "step": 980 + }, + { + "epoch": 0.761055081458495, + "grad_norm": 9.183679580688477, + "learning_rate": 2.453629481268396e-05, + "loss": 0.7964, + "step": 981 + }, + { + "epoch": 0.7618308766485647, + "grad_norm": 6.668287754058838, + "learning_rate": 2.440913553781393e-05, + "loss": 0.4153, + "step": 982 + }, + { + "epoch": 0.7626066718386346, + "grad_norm": 8.554515838623047, + "learning_rate": 2.4282072069041753e-05, + "loss": 0.9775, + "step": 983 + }, + { + "epoch": 0.7633824670287044, + "grad_norm": 5.999274730682373, + "learning_rate": 2.4155125009155213e-05, + "loss": 0.339, + "step": 984 + }, + { + "epoch": 0.7641582622187743, + "grad_norm": 13.781418800354004, + "learning_rate": 2.402831494206689e-05, + "loss": 1.5816, + "step": 985 + }, + { + "epoch": 0.764934057408844, + "grad_norm": 8.647990226745605, + "learning_rate": 2.3901662429476605e-05, + "loss": 0.897, + "step": 986 + }, + { + "epoch": 0.7657098525989139, + "grad_norm": 12.957200050354004, + "learning_rate": 2.3775188007537373e-05, + "loss": 1.2145, + "step": 987 + }, + { + "epoch": 0.7664856477889838, + "grad_norm": 8.009690284729004, + "learning_rate": 2.3648912183525608e-05, + "loss": 0.3752, + "step": 988 + }, + { + "epoch": 0.7672614429790535, + "grad_norm": 6.878604412078857, + "learning_rate": 2.3522855432515916e-05, + "loss": 0.4954, + "step": 989 + }, + { + "epoch": 0.7680372381691234, + "grad_norm": 4.699245452880859, + "learning_rate": 2.339703819406119e-05, + "loss": 0.2914, + "step": 990 + }, + { + "epoch": 0.7688130333591932, + "grad_norm": 8.991491317749023, + "learning_rate": 2.3271480868878396e-05, + "loss": 0.8471, + "step": 991 + }, + { + "epoch": 0.769588828549263, + "grad_norm": 5.581956386566162, + "learning_rate": 2.31462038155407e-05, + "loss": 0.374, + "step": 992 + }, + { + "epoch": 0.7703646237393328, + "grad_norm": 5.642667293548584, + "learning_rate": 2.3021227347176396e-05, + "loss": 0.277, + "step": 993 + }, + { + "epoch": 0.7711404189294027, + "grad_norm": 11.739602088928223, + "learning_rate": 2.2896571728175252e-05, + "loss": 0.9676, + "step": 994 + }, + { + "epoch": 0.7719162141194724, + "grad_norm": 15.364945411682129, + "learning_rate": 2.27722571709027e-05, + "loss": 1.4121, + "step": 995 + }, + { + "epoch": 0.7726920093095423, + "grad_norm": 7.156294345855713, + "learning_rate": 2.2648303832422503e-05, + "loss": 0.2757, + "step": 996 + }, + { + "epoch": 0.773467804499612, + "grad_norm": 7.8592634201049805, + "learning_rate": 2.252473181122837e-05, + "loss": 0.5515, + "step": 997 + }, + { + "epoch": 0.7742435996896819, + "grad_norm": 10.447184562683105, + "learning_rate": 2.240156114398507e-05, + "loss": 1.0891, + "step": 998 + }, + { + "epoch": 0.7750193948797518, + "grad_norm": 11.452353477478027, + "learning_rate": 2.2278811802279586e-05, + "loss": 0.9895, + "step": 999 + }, + { + "epoch": 0.7757951900698216, + "grad_norm": 6.1940436363220215, + "learning_rate": 2.2156503689382787e-05, + "loss": 0.2428, + "step": 1000 + }, + { + "epoch": 0.7765709852598914, + "grad_norm": 8.439943313598633, + "learning_rate": 2.203465663702223e-05, + "loss": 0.3215, + "step": 1001 + }, + { + "epoch": 0.7773467804499612, + "grad_norm": 10.71116828918457, + "learning_rate": 2.1913290402166505e-05, + "loss": 0.9325, + "step": 1002 + }, + { + "epoch": 0.7781225756400311, + "grad_norm": 6.461543560028076, + "learning_rate": 2.1792424663821737e-05, + "loss": 0.3231, + "step": 1003 + }, + { + "epoch": 0.7788983708301008, + "grad_norm": 9.305453300476074, + "learning_rate": 2.1672079019840728e-05, + "loss": 0.2963, + "step": 1004 + }, + { + "epoch": 0.7796741660201707, + "grad_norm": 7.960659503936768, + "learning_rate": 2.1552272983745248e-05, + "loss": 0.5616, + "step": 1005 + }, + { + "epoch": 0.7804499612102405, + "grad_norm": 6.437761306762695, + "learning_rate": 2.143302598156199e-05, + "loss": 0.3755, + "step": 1006 + }, + { + "epoch": 0.7812257564003103, + "grad_norm": 10.811202049255371, + "learning_rate": 2.131435734867274e-05, + "loss": 0.8611, + "step": 1007 + }, + { + "epoch": 0.7820015515903801, + "grad_norm": 0.9098401665687561, + "learning_rate": 2.119628632667923e-05, + "loss": 0.0237, + "step": 1008 + }, + { + "epoch": 0.78277734678045, + "grad_norm": 12.607281684875488, + "learning_rate": 2.107883206028317e-05, + "loss": 1.222, + "step": 1009 + }, + { + "epoch": 0.7835531419705197, + "grad_norm": 13.561890602111816, + "learning_rate": 2.096201359418206e-05, + "loss": 1.1614, + "step": 1010 + }, + { + "epoch": 0.7843289371605896, + "grad_norm": 14.828733444213867, + "learning_rate": 2.0845849869981137e-05, + "loss": 1.4014, + "step": 1011 + }, + { + "epoch": 0.7851047323506595, + "grad_norm": 12.574007987976074, + "learning_rate": 2.0730359723122095e-05, + "loss": 1.0261, + "step": 1012 + }, + { + "epoch": 0.7858805275407292, + "grad_norm": 7.420691013336182, + "learning_rate": 2.0615561879829005e-05, + "loss": 0.3906, + "step": 1013 + }, + { + "epoch": 0.7866563227307991, + "grad_norm": 5.57167911529541, + "learning_rate": 2.0501474954071924e-05, + "loss": 0.1874, + "step": 1014 + }, + { + "epoch": 0.7874321179208689, + "grad_norm": 3.0548763275146484, + "learning_rate": 2.0388117444548727e-05, + "loss": 0.1013, + "step": 1015 + }, + { + "epoch": 0.7882079131109387, + "grad_norm": 6.536466598510742, + "learning_rate": 2.0275507731685632e-05, + "loss": 0.2955, + "step": 1016 + }, + { + "epoch": 0.7889837083010085, + "grad_norm": 11.016971588134766, + "learning_rate": 2.0163664074656877e-05, + "loss": 0.8856, + "step": 1017 + }, + { + "epoch": 0.7897595034910784, + "grad_norm": 13.149106979370117, + "learning_rate": 2.0052604608424097e-05, + "loss": 1.2317, + "step": 1018 + }, + { + "epoch": 0.7905352986811481, + "grad_norm": 7.8416748046875, + "learning_rate": 1.9942347340795804e-05, + "loss": 0.6771, + "step": 1019 + }, + { + "epoch": 0.791311093871218, + "grad_norm": 5.4342780113220215, + "learning_rate": 1.9832910149507487e-05, + "loss": 0.26, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_Qnli-dev_cosine_accuracy": 0.6796875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8606457710266113, + "eval_Qnli-dev_cosine_ap": 0.7134567054454279, + "eval_Qnli-dev_cosine_f1": 0.6862068965517241, + "eval_Qnli-dev_cosine_f1_threshold": 0.7837170362472534, + "eval_Qnli-dev_cosine_precision": 0.5784883720930233, + "eval_Qnli-dev_cosine_recall": 0.8432203389830508, + "eval_Qnli-dev_dot_accuracy": 0.625, + "eval_Qnli-dev_dot_accuracy_threshold": 448.73583984375, + "eval_Qnli-dev_dot_ap": 0.6131219760953251, + "eval_Qnli-dev_dot_f1": 0.6677018633540374, + "eval_Qnli-dev_dot_f1_threshold": 358.3121643066406, + "eval_Qnli-dev_dot_precision": 0.5269607843137255, + "eval_Qnli-dev_dot_recall": 0.9110169491525424, + "eval_Qnli-dev_euclidean_accuracy": 0.671875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.926061630249023, + "eval_Qnli-dev_euclidean_ap": 0.7107089574932685, + "eval_Qnli-dev_euclidean_f1": 0.6944444444444444, + "eval_Qnli-dev_euclidean_f1_threshold": 14.793469429016113, + "eval_Qnli-dev_euclidean_precision": 0.5882352941176471, + "eval_Qnli-dev_euclidean_recall": 0.847457627118644, + "eval_Qnli-dev_manhattan_accuracy": 0.671875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 216.6309356689453, + "eval_Qnli-dev_manhattan_ap": 0.7131558473266311, + "eval_Qnli-dev_manhattan_f1": 0.6949429037520392, + "eval_Qnli-dev_manhattan_f1_threshold": 280.2909240722656, + "eval_Qnli-dev_manhattan_precision": 0.5649867374005305, + "eval_Qnli-dev_manhattan_recall": 0.902542372881356, + "eval_Qnli-dev_max_accuracy": 0.6796875, + "eval_Qnli-dev_max_accuracy_threshold": 448.73583984375, + "eval_Qnli-dev_max_ap": 0.7134567054454279, + "eval_Qnli-dev_max_f1": 0.6949429037520392, + "eval_Qnli-dev_max_f1_threshold": 358.3121643066406, + "eval_Qnli-dev_max_precision": 0.5882352941176471, + "eval_Qnli-dev_max_recall": 0.9110169491525424, + "eval_allNLI-dev_cosine_accuracy": 0.71875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9162718057632446, + "eval_allNLI-dev_cosine_ap": 0.6098745566780684, + "eval_allNLI-dev_cosine_f1": 0.6258823529411764, + "eval_allNLI-dev_cosine_f1_threshold": 0.8387055397033691, + "eval_allNLI-dev_cosine_precision": 0.5277777777777778, + "eval_allNLI-dev_cosine_recall": 0.7687861271676301, + "eval_allNLI-dev_dot_accuracy": 0.67578125, + "eval_allNLI-dev_dot_accuracy_threshold": 477.22607421875, + "eval_allNLI-dev_dot_ap": 0.4797178064051032, + "eval_allNLI-dev_dot_f1": 0.5714285714285714, + "eval_allNLI-dev_dot_f1_threshold": 376.41058349609375, + "eval_allNLI-dev_dot_precision": 0.44516129032258067, + "eval_allNLI-dev_dot_recall": 0.7976878612716763, + "eval_allNLI-dev_euclidean_accuracy": 0.716796875, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.360190391540527, + "eval_allNLI-dev_euclidean_ap": 0.6097795347884272, + "eval_allNLI-dev_euclidean_f1": 0.6272727272727273, + "eval_allNLI-dev_euclidean_f1_threshold": 12.861324310302734, + "eval_allNLI-dev_euclidean_precision": 0.5168539325842697, + "eval_allNLI-dev_euclidean_recall": 0.7976878612716763, + "eval_allNLI-dev_manhattan_accuracy": 0.724609375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 202.03057861328125, + "eval_allNLI-dev_manhattan_ap": 0.6113200080785296, + "eval_allNLI-dev_manhattan_f1": 0.6229508196721311, + "eval_allNLI-dev_manhattan_f1_threshold": 225.83822631835938, + "eval_allNLI-dev_manhattan_precision": 0.5236220472440944, + "eval_allNLI-dev_manhattan_recall": 0.7687861271676301, + "eval_allNLI-dev_max_accuracy": 0.724609375, + "eval_allNLI-dev_max_accuracy_threshold": 477.22607421875, + "eval_allNLI-dev_max_ap": 0.6113200080785296, + "eval_allNLI-dev_max_f1": 0.6272727272727273, + "eval_allNLI-dev_max_f1_threshold": 376.41058349609375, + "eval_allNLI-dev_max_precision": 0.5277777777777778, + "eval_allNLI-dev_max_recall": 0.7976878612716763, + "eval_sequential_score": 0.7134567054454279, + "eval_sts-test_pearson_cosine": 0.8168834278479731, + "eval_sts-test_pearson_dot": 0.7256103117983841, + "eval_sts-test_pearson_euclidean": 0.8473072246533302, + "eval_sts-test_pearson_manhattan": 0.8440149277565967, + "eval_sts-test_pearson_max": 0.8473072246533302, + "eval_sts-test_spearman_cosine": 0.8531133892206136, + "eval_sts-test_spearman_dot": 0.7242559359012971, + "eval_sts-test_spearman_euclidean": 0.8447167995482682, + "eval_sts-test_spearman_manhattan": 0.8431346634530366, + "eval_sts-test_spearman_max": 0.8531133892206136, + "eval_vitaminc-pairs_loss": 3.8304741382598877, + "eval_vitaminc-pairs_runtime": 5.5124, + "eval_vitaminc-pairs_samples_per_second": 23.22, + "eval_vitaminc-pairs_steps_per_second": 0.363, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_negation-triplets_loss": 1.4637340307235718, + "eval_negation-triplets_runtime": 1.1176, + "eval_negation-triplets_samples_per_second": 114.536, + "eval_negation-triplets_steps_per_second": 1.79, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_scitail-pairs-pos_loss": 0.1727929562330246, + "eval_scitail-pairs-pos_runtime": 1.2624, + "eval_scitail-pairs-pos_samples_per_second": 101.393, + "eval_scitail-pairs-pos_steps_per_second": 1.584, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_scitail-pairs-qa_loss": 0.0069268993102014065, + "eval_scitail-pairs-qa_runtime": 0.9236, + "eval_scitail-pairs-qa_samples_per_second": 138.583, + "eval_scitail-pairs-qa_steps_per_second": 2.165, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_xsum-pairs_loss": 0.27876338362693787, + "eval_xsum-pairs_runtime": 4.3183, + "eval_xsum-pairs_samples_per_second": 29.641, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_sciq_pairs_loss": 0.07622424513101578, + "eval_sciq_pairs_runtime": 5.5909, + "eval_sciq_pairs_samples_per_second": 22.894, + "eval_sciq_pairs_steps_per_second": 0.358, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_qasc_pairs_loss": 0.36567261815071106, + "eval_qasc_pairs_runtime": 0.9909, + "eval_qasc_pairs_samples_per_second": 129.175, + "eval_qasc_pairs_steps_per_second": 2.018, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_openbookqa_pairs_loss": 1.257145881652832, + "eval_openbookqa_pairs_runtime": 0.8897, + "eval_openbookqa_pairs_samples_per_second": 143.875, + "eval_openbookqa_pairs_steps_per_second": 2.248, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_msmarco_pairs_loss": 0.7520588636398315, + "eval_msmarco_pairs_runtime": 1.9682, + "eval_msmarco_pairs_samples_per_second": 65.033, + "eval_msmarco_pairs_steps_per_second": 1.016, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_nq_pairs_loss": 0.9746608734130859, + "eval_nq_pairs_runtime": 4.4352, + "eval_nq_pairs_samples_per_second": 28.86, + "eval_nq_pairs_steps_per_second": 0.451, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_trivia_pairs_loss": 0.7279982566833496, + "eval_trivia_pairs_runtime": 4.5635, + "eval_trivia_pairs_samples_per_second": 28.048, + "eval_trivia_pairs_steps_per_second": 0.438, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_gooaq_pairs_loss": 0.5491998791694641, + "eval_gooaq_pairs_runtime": 1.2795, + "eval_gooaq_pairs_samples_per_second": 100.038, + "eval_gooaq_pairs_steps_per_second": 1.563, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_paws-pos_loss": 0.03706526383757591, + "eval_paws-pos_runtime": 1.0378, + "eval_paws-pos_samples_per_second": 123.332, + "eval_paws-pos_steps_per_second": 1.927, + "step": 1020 + }, + { + "epoch": 0.791311093871218, + "eval_global_dataset_loss": 0.524812638759613, + "eval_global_dataset_runtime": 21.1576, + "eval_global_dataset_samples_per_second": 19.662, + "eval_global_dataset_steps_per_second": 0.331, + "step": 1020 + }, + { + "epoch": 0.7920868890612878, + "grad_norm": 7.193615913391113, + "learning_rate": 1.972431077932285e-05, + "loss": 0.4263, + "step": 1021 + }, + { + "epoch": 0.7928626842513576, + "grad_norm": 10.385963439941406, + "learning_rate": 1.9616566839156542e-05, + "loss": 0.8427, + "step": 1022 + }, + { + "epoch": 0.7936384794414275, + "grad_norm": 5.188340187072754, + "learning_rate": 1.9509695799218964e-05, + "loss": 0.2991, + "step": 1023 + }, + { + "epoch": 0.7944142746314973, + "grad_norm": 0.7457648515701294, + "learning_rate": 1.9403714988183558e-05, + "loss": 0.0394, + "step": 1024 + }, + { + "epoch": 0.7951900698215671, + "grad_norm": 8.064065933227539, + "learning_rate": 1.929864159037703e-05, + "loss": 0.9928, + "step": 1025 + }, + { + "epoch": 0.7959658650116369, + "grad_norm": 7.173913478851318, + "learning_rate": 1.919449264299295e-05, + "loss": 0.3891, + "step": 1026 + }, + { + "epoch": 0.7967416602017068, + "grad_norm": 7.13004732131958, + "learning_rate": 1.9091285033329344e-05, + "loss": 0.2513, + "step": 1027 + }, + { + "epoch": 0.7975174553917765, + "grad_norm": 12.565049171447754, + "learning_rate": 1.898903549605036e-05, + "loss": 1.1539, + "step": 1028 + }, + { + "epoch": 0.7982932505818464, + "grad_norm": 12.020158767700195, + "learning_rate": 1.8887760610472935e-05, + "loss": 1.0283, + "step": 1029 + }, + { + "epoch": 0.7990690457719162, + "grad_norm": 10.133840560913086, + "learning_rate": 1.8787476797878453e-05, + "loss": 0.7875, + "step": 1030 + }, + { + "epoch": 0.799844840961986, + "grad_norm": 5.798900604248047, + "learning_rate": 1.8688200318850146e-05, + "loss": 0.2013, + "step": 1031 + }, + { + "epoch": 0.8006206361520558, + "grad_norm": 9.04205322265625, + "learning_rate": 1.858994727063648e-05, + "loss": 1.0791, + "step": 1032 + }, + { + "epoch": 0.8013964313421257, + "grad_norm": 9.078784942626953, + "learning_rate": 1.8492733584541093e-05, + "loss": 0.7575, + "step": 1033 + }, + { + "epoch": 0.8021722265321956, + "grad_norm": 8.002516746520996, + "learning_rate": 1.839657502333958e-05, + "loss": 0.3728, + "step": 1034 + }, + { + "epoch": 0.8029480217222653, + "grad_norm": 6.926529884338379, + "learning_rate": 1.8301487178723625e-05, + "loss": 0.6275, + "step": 1035 + }, + { + "epoch": 0.8037238169123352, + "grad_norm": 0.9333884119987488, + "learning_rate": 1.8207485468772882e-05, + "loss": 0.0223, + "step": 1036 + }, + { + "epoch": 0.804499612102405, + "grad_norm": 10.30510425567627, + "learning_rate": 1.8114585135455015e-05, + "loss": 0.7849, + "step": 1037 + }, + { + "epoch": 0.8052754072924748, + "grad_norm": 10.618290901184082, + "learning_rate": 1.802280124215426e-05, + "loss": 0.9409, + "step": 1038 + }, + { + "epoch": 0.8060512024825446, + "grad_norm": 8.816503524780273, + "learning_rate": 1.793214867122896e-05, + "loss": 0.4584, + "step": 1039 + }, + { + "epoch": 0.8068269976726145, + "grad_norm": 0.7741997838020325, + "learning_rate": 1.784264212159848e-05, + "loss": 0.0287, + "step": 1040 + }, + { + "epoch": 0.8076027928626842, + "grad_norm": 0.6247514486312866, + "learning_rate": 1.7754296106359812e-05, + "loss": 0.008, + "step": 1041 + }, + { + "epoch": 0.8083785880527541, + "grad_norm": 8.194477081298828, + "learning_rate": 1.766712495043436e-05, + "loss": 0.3427, + "step": 1042 + }, + { + "epoch": 0.8091543832428238, + "grad_norm": 10.494790077209473, + "learning_rate": 1.758114278824521e-05, + "loss": 0.6172, + "step": 1043 + }, + { + "epoch": 0.8099301784328937, + "grad_norm": 18.723392486572266, + "learning_rate": 1.7496363561425302e-05, + "loss": 2.497, + "step": 1044 + }, + { + "epoch": 0.8107059736229635, + "grad_norm": 12.240035057067871, + "learning_rate": 1.7412801016556827e-05, + "loss": 1.0139, + "step": 1045 + }, + { + "epoch": 0.8114817688130334, + "grad_norm": 1.670955777168274, + "learning_rate": 1.733046870294232e-05, + "loss": 0.0494, + "step": 1046 + }, + { + "epoch": 0.8122575640031032, + "grad_norm": 1.7689838409423828, + "learning_rate": 1.7249379970407667e-05, + "loss": 0.0508, + "step": 1047 + }, + { + "epoch": 0.813033359193173, + "grad_norm": 10.22224235534668, + "learning_rate": 1.7169547967137504e-05, + "loss": 0.8282, + "step": 1048 + }, + { + "epoch": 0.8138091543832429, + "grad_norm": 7.301149368286133, + "learning_rate": 1.7090985637543297e-05, + "loss": 0.3678, + "step": 1049 + }, + { + "epoch": 0.8145849495733126, + "grad_norm": 10.875919342041016, + "learning_rate": 1.7013705720164454e-05, + "loss": 1.0414, + "step": 1050 + }, + { + "epoch": 0.8153607447633825, + "grad_norm": 6.961636543273926, + "learning_rate": 1.6937720745602826e-05, + "loss": 0.2956, + "step": 1051 + }, + { + "epoch": 0.8161365399534523, + "grad_norm": 10.172094345092773, + "learning_rate": 1.6863043034490936e-05, + "loss": 0.6022, + "step": 1052 + }, + { + "epoch": 0.8169123351435221, + "grad_norm": 10.533774375915527, + "learning_rate": 1.678968469549424e-05, + "loss": 0.6047, + "step": 1053 + }, + { + "epoch": 0.8176881303335919, + "grad_norm": 12.16602611541748, + "learning_rate": 1.6717657623347738e-05, + "loss": 0.9537, + "step": 1054 + }, + { + "epoch": 0.8184639255236618, + "grad_norm": 7.5252251625061035, + "learning_rate": 1.6646973496927326e-05, + "loss": 0.4818, + "step": 1055 + }, + { + "epoch": 0.8192397207137315, + "grad_norm": 12.277027130126953, + "learning_rate": 1.6577643777356095e-05, + "loss": 0.9961, + "step": 1056 + }, + { + "epoch": 0.8200155159038014, + "grad_norm": 7.361176013946533, + "learning_rate": 1.650967970614597e-05, + "loss": 0.3835, + "step": 1057 + }, + { + "epoch": 0.8207913110938713, + "grad_norm": 10.20497989654541, + "learning_rate": 1.6443092303374943e-05, + "loss": 0.7192, + "step": 1058 + }, + { + "epoch": 0.821567106283941, + "grad_norm": 5.959074974060059, + "learning_rate": 1.6377892365900225e-05, + "loss": 0.3131, + "step": 1059 + }, + { + "epoch": 0.8223429014740109, + "grad_norm": 1.6121126413345337, + "learning_rate": 1.6314090465607563e-05, + "loss": 0.0402, + "step": 1060 + }, + { + "epoch": 0.8231186966640807, + "grad_norm": 6.287003993988037, + "learning_rate": 1.625169694769707e-05, + "loss": 0.247, + "step": 1061 + }, + { + "epoch": 0.8238944918541505, + "grad_norm": 9.430892944335938, + "learning_rate": 1.6190721929005806e-05, + "loss": 0.7557, + "step": 1062 + }, + { + "epoch": 0.8246702870442203, + "grad_norm": 1.6690380573272705, + "learning_rate": 1.6131175296367342e-05, + "loss": 0.0468, + "step": 1063 + }, + { + "epoch": 0.8254460822342902, + "grad_norm": 1.8117036819458008, + "learning_rate": 1.6073066705008697e-05, + "loss": 0.0421, + "step": 1064 + }, + { + "epoch": 0.8262218774243599, + "grad_norm": 2.686722993850708, + "learning_rate": 1.601640557698474e-05, + "loss": 0.0711, + "step": 1065 + }, + { + "epoch": 0.8269976726144298, + "grad_norm": 5.40755558013916, + "learning_rate": 1.596120109965049e-05, + "loss": 0.3368, + "step": 1066 + }, + { + "epoch": 0.8277734678044996, + "grad_norm": 7.796652793884277, + "learning_rate": 1.590746222417139e-05, + "loss": 0.5852, + "step": 1067 + }, + { + "epoch": 0.8285492629945694, + "grad_norm": 10.976995468139648, + "learning_rate": 1.5855197664071925e-05, + "loss": 0.999, + "step": 1068 + }, + { + "epoch": 0.8293250581846393, + "grad_norm": 8.075950622558594, + "learning_rate": 1.580441589382278e-05, + "loss": 0.4071, + "step": 1069 + }, + { + "epoch": 0.8301008533747091, + "grad_norm": 9.25891399383545, + "learning_rate": 1.5755125147466718e-05, + "loss": 0.4372, + "step": 1070 + }, + { + "epoch": 0.8308766485647789, + "grad_norm": 8.5850191116333, + "learning_rate": 1.5707333417283478e-05, + "loss": 0.6649, + "step": 1071 + }, + { + "epoch": 0.8316524437548487, + "grad_norm": 8.302315711975098, + "learning_rate": 1.566104845249385e-05, + "loss": 0.4461, + "step": 1072 + }, + { + "epoch": 0.8324282389449186, + "grad_norm": 10.136163711547852, + "learning_rate": 1.5616277758003197e-05, + "loss": 0.7703, + "step": 1073 + }, + { + "epoch": 0.8332040341349883, + "grad_norm": 7.116093635559082, + "learning_rate": 1.5573028593184537e-05, + "loss": 0.3798, + "step": 1074 + }, + { + "epoch": 0.8339798293250582, + "grad_norm": 8.765496253967285, + "learning_rate": 1.553130797070149e-05, + "loss": 0.4016, + "step": 1075 + }, + { + "epoch": 0.834755624515128, + "grad_norm": 8.114645004272461, + "learning_rate": 1.5491122655371192e-05, + "loss": 0.3554, + "step": 1076 + }, + { + "epoch": 0.8355314197051978, + "grad_norm": 12.428133010864258, + "learning_rate": 1.5452479163067416e-05, + "loss": 1.0995, + "step": 1077 + }, + { + "epoch": 0.8363072148952676, + "grad_norm": 12.10230541229248, + "learning_rate": 1.541538375966405e-05, + "loss": 1.0319, + "step": 1078 + }, + { + "epoch": 0.8370830100853375, + "grad_norm": 6.978891372680664, + "learning_rate": 1.5379842460019118e-05, + "loss": 0.3093, + "step": 1079 + }, + { + "epoch": 0.8378588052754073, + "grad_norm": 6.694214344024658, + "learning_rate": 1.534586102699948e-05, + "loss": 0.3652, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_Qnli-dev_cosine_accuracy": 0.677734375, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8653599619865417, + "eval_Qnli-dev_cosine_ap": 0.7179170174814624, + "eval_Qnli-dev_cosine_f1": 0.6870748299319728, + "eval_Qnli-dev_cosine_f1_threshold": 0.7731696367263794, + "eval_Qnli-dev_cosine_precision": 0.5738636363636364, + "eval_Qnli-dev_cosine_recall": 0.8559322033898306, + "eval_Qnli-dev_dot_accuracy": 0.6328125, + "eval_Qnli-dev_dot_accuracy_threshold": 413.5496520996094, + "eval_Qnli-dev_dot_ap": 0.6136794365202283, + "eval_Qnli-dev_dot_f1": 0.6655574043261231, + "eval_Qnli-dev_dot_f1_threshold": 362.39324951171875, + "eval_Qnli-dev_dot_precision": 0.547945205479452, + "eval_Qnli-dev_dot_recall": 0.847457627118644, + "eval_Qnli-dev_euclidean_accuracy": 0.68359375, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.387691497802734, + "eval_Qnli-dev_euclidean_ap": 0.7204084797433787, + "eval_Qnli-dev_euclidean_f1": 0.6920415224913494, + "eval_Qnli-dev_euclidean_f1_threshold": 14.871377944946289, + "eval_Qnli-dev_euclidean_precision": 0.5847953216374269, + "eval_Qnli-dev_euclidean_recall": 0.847457627118644, + "eval_Qnli-dev_manhattan_accuracy": 0.6796875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 237.66259765625, + "eval_Qnli-dev_manhattan_ap": 0.7189008928051237, + "eval_Qnli-dev_manhattan_f1": 0.68824531516184, + "eval_Qnli-dev_manhattan_f1_threshold": 268.883544921875, + "eval_Qnli-dev_manhattan_precision": 0.5754985754985755, + "eval_Qnli-dev_manhattan_recall": 0.8559322033898306, + "eval_Qnli-dev_max_accuracy": 0.68359375, + "eval_Qnli-dev_max_accuracy_threshold": 413.5496520996094, + "eval_Qnli-dev_max_ap": 0.7204084797433787, + "eval_Qnli-dev_max_f1": 0.6920415224913494, + "eval_Qnli-dev_max_f1_threshold": 362.39324951171875, + "eval_Qnli-dev_max_precision": 0.5847953216374269, + "eval_Qnli-dev_max_recall": 0.8559322033898306, + "eval_allNLI-dev_cosine_accuracy": 0.720703125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8944388031959534, + "eval_allNLI-dev_cosine_ap": 0.6097658196724249, + "eval_allNLI-dev_cosine_f1": 0.6146993318485523, + "eval_allNLI-dev_cosine_f1_threshold": 0.8173136711120605, + "eval_allNLI-dev_cosine_precision": 0.5, + "eval_allNLI-dev_cosine_recall": 0.7976878612716763, + "eval_allNLI-dev_dot_accuracy": 0.693359375, + "eval_allNLI-dev_dot_accuracy_threshold": 442.28173828125, + "eval_allNLI-dev_dot_ap": 0.5100599176517999, + "eval_allNLI-dev_dot_f1": 0.5625, + "eval_allNLI-dev_dot_f1_threshold": 350.533935546875, + "eval_allNLI-dev_dot_precision": 0.4247787610619469, + "eval_allNLI-dev_dot_recall": 0.8323699421965318, + "eval_allNLI-dev_euclidean_accuracy": 0.720703125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.057489395141602, + "eval_allNLI-dev_euclidean_ap": 0.6072363425301475, + "eval_allNLI-dev_euclidean_f1": 0.62004662004662, + "eval_allNLI-dev_euclidean_f1_threshold": 12.713096618652344, + "eval_allNLI-dev_euclidean_precision": 0.51953125, + "eval_allNLI-dev_euclidean_recall": 0.7687861271676301, + "eval_allNLI-dev_manhattan_accuracy": 0.71484375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 166.11390686035156, + "eval_allNLI-dev_manhattan_ap": 0.6062552226059159, + "eval_allNLI-dev_manhattan_f1": 0.6278026905829597, + "eval_allNLI-dev_manhattan_f1_threshold": 234.42974853515625, + "eval_allNLI-dev_manhattan_precision": 0.5128205128205128, + "eval_allNLI-dev_manhattan_recall": 0.8092485549132948, + "eval_allNLI-dev_max_accuracy": 0.720703125, + "eval_allNLI-dev_max_accuracy_threshold": 442.28173828125, + "eval_allNLI-dev_max_ap": 0.6097658196724249, + "eval_allNLI-dev_max_f1": 0.6278026905829597, + "eval_allNLI-dev_max_f1_threshold": 350.533935546875, + "eval_allNLI-dev_max_precision": 0.51953125, + "eval_allNLI-dev_max_recall": 0.8323699421965318, + "eval_sequential_score": 0.7204084797433787, + "eval_sts-test_pearson_cosine": 0.8254171973399721, + "eval_sts-test_pearson_dot": 0.749637475493375, + "eval_sts-test_pearson_euclidean": 0.8573189699008219, + "eval_sts-test_pearson_manhattan": 0.8548002587995798, + "eval_sts-test_pearson_max": 0.8573189699008219, + "eval_sts-test_spearman_cosine": 0.8614434785185315, + "eval_sts-test_spearman_dot": 0.7489230996878047, + "eval_sts-test_spearman_euclidean": 0.8550367536839545, + "eval_sts-test_spearman_manhattan": 0.8526494081425074, + "eval_sts-test_spearman_max": 0.8614434785185315, + "eval_vitaminc-pairs_loss": 3.844217300415039, + "eval_vitaminc-pairs_runtime": 5.5386, + "eval_vitaminc-pairs_samples_per_second": 23.111, + "eval_vitaminc-pairs_steps_per_second": 0.361, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_negation-triplets_loss": 1.432388186454773, + "eval_negation-triplets_runtime": 1.1286, + "eval_negation-triplets_samples_per_second": 113.418, + "eval_negation-triplets_steps_per_second": 1.772, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_scitail-pairs-pos_loss": 0.2345297634601593, + "eval_scitail-pairs-pos_runtime": 1.2964, + "eval_scitail-pairs-pos_samples_per_second": 98.737, + "eval_scitail-pairs-pos_steps_per_second": 1.543, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_scitail-pairs-qa_loss": 0.005646564066410065, + "eval_scitail-pairs-qa_runtime": 0.9286, + "eval_scitail-pairs-qa_samples_per_second": 137.837, + "eval_scitail-pairs-qa_steps_per_second": 2.154, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_xsum-pairs_loss": 0.267998605966568, + "eval_xsum-pairs_runtime": 4.3393, + "eval_xsum-pairs_samples_per_second": 29.498, + "eval_xsum-pairs_steps_per_second": 0.461, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_sciq_pairs_loss": 0.08142460882663727, + "eval_sciq_pairs_runtime": 5.6276, + "eval_sciq_pairs_samples_per_second": 22.745, + "eval_sciq_pairs_steps_per_second": 0.355, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_qasc_pairs_loss": 0.3024355471134186, + "eval_qasc_pairs_runtime": 0.9882, + "eval_qasc_pairs_samples_per_second": 129.524, + "eval_qasc_pairs_steps_per_second": 2.024, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_openbookqa_pairs_loss": 1.2320504188537598, + "eval_openbookqa_pairs_runtime": 0.8716, + "eval_openbookqa_pairs_samples_per_second": 146.857, + "eval_openbookqa_pairs_steps_per_second": 2.295, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_msmarco_pairs_loss": 0.6611219644546509, + "eval_msmarco_pairs_runtime": 1.9682, + "eval_msmarco_pairs_samples_per_second": 65.034, + "eval_msmarco_pairs_steps_per_second": 1.016, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_nq_pairs_loss": 0.9349308013916016, + "eval_nq_pairs_runtime": 4.419, + "eval_nq_pairs_samples_per_second": 28.966, + "eval_nq_pairs_steps_per_second": 0.453, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_trivia_pairs_loss": 0.6609914302825928, + "eval_trivia_pairs_runtime": 4.5704, + "eval_trivia_pairs_samples_per_second": 28.006, + "eval_trivia_pairs_steps_per_second": 0.438, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_gooaq_pairs_loss": 0.6042873859405518, + "eval_gooaq_pairs_runtime": 1.2862, + "eval_gooaq_pairs_samples_per_second": 99.522, + "eval_gooaq_pairs_steps_per_second": 1.555, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_paws-pos_loss": 0.036212559789419174, + "eval_paws-pos_runtime": 1.0416, + "eval_paws-pos_samples_per_second": 122.885, + "eval_paws-pos_steps_per_second": 1.92, + "step": 1080 + }, + { + "epoch": 0.8378588052754073, + "eval_global_dataset_loss": 0.5297088027000427, + "eval_global_dataset_runtime": 21.1789, + "eval_global_dataset_samples_per_second": 19.642, + "eval_global_dataset_steps_per_second": 0.331, + "step": 1080 + }, + { + "epoch": 0.8386346004654771, + "grad_norm": 4.742051124572754, + "learning_rate": 1.5313444970546426e-05, + "loss": 0.2184, + "step": 1081 + }, + { + "epoch": 0.839410395655547, + "grad_norm": 4.320516586303711, + "learning_rate": 1.5282599546782257e-05, + "loss": 0.18, + "step": 1082 + }, + { + "epoch": 0.8401861908456167, + "grad_norm": 6.53492546081543, + "learning_rate": 1.5253329757158036e-05, + "loss": 0.4299, + "step": 1083 + }, + { + "epoch": 0.8409619860356866, + "grad_norm": 6.300450801849365, + "learning_rate": 1.52256403476426e-05, + "loss": 0.2402, + "step": 1084 + }, + { + "epoch": 0.8417377812257564, + "grad_norm": 11.248185157775879, + "learning_rate": 1.5199535807953046e-05, + "loss": 0.805, + "step": 1085 + }, + { + "epoch": 0.8425135764158262, + "grad_norm": 8.108342170715332, + "learning_rate": 1.5175020370826736e-05, + "loss": 0.459, + "step": 1086 + }, + { + "epoch": 0.843289371605896, + "grad_norm": 6.837913990020752, + "learning_rate": 1.5152098011334971e-05, + "loss": 0.2368, + "step": 1087 + }, + { + "epoch": 0.8440651667959659, + "grad_norm": 5.785921573638916, + "learning_rate": 1.513077244623846e-05, + "loss": 0.2282, + "step": 1088 + }, + { + "epoch": 0.8448409619860356, + "grad_norm": 6.231807231903076, + "learning_rate": 1.5111047133384662e-05, + "loss": 0.3175, + "step": 1089 + }, + { + "epoch": 0.8456167571761055, + "grad_norm": 13.163674354553223, + "learning_rate": 1.5092925271147096e-05, + "loss": 1.3646, + "step": 1090 + }, + { + "epoch": 0.8463925523661753, + "grad_norm": 8.002300262451172, + "learning_rate": 1.5076409797906762e-05, + "loss": 0.3908, + "step": 1091 + }, + { + "epoch": 0.8471683475562451, + "grad_norm": 13.659926414489746, + "learning_rate": 1.5061503391575676e-05, + "loss": 1.1518, + "step": 1092 + }, + { + "epoch": 0.847944142746315, + "grad_norm": 8.983166694641113, + "learning_rate": 1.5048208469162676e-05, + "loss": 0.532, + "step": 1093 + }, + { + "epoch": 0.8487199379363848, + "grad_norm": 6.241405010223389, + "learning_rate": 1.5036527186381494e-05, + "loss": 0.2099, + "step": 1094 + }, + { + "epoch": 0.8494957331264547, + "grad_norm": 12.054598808288574, + "learning_rate": 1.5026461437301233e-05, + "loss": 0.9802, + "step": 1095 + }, + { + "epoch": 0.8502715283165244, + "grad_norm": 8.754733085632324, + "learning_rate": 1.5018012854039243e-05, + "loss": 0.6938, + "step": 1096 + }, + { + "epoch": 0.8510473235065943, + "grad_norm": 11.566781997680664, + "learning_rate": 1.5011182806496492e-05, + "loss": 1.0521, + "step": 1097 + }, + { + "epoch": 0.851823118696664, + "grad_norm": 11.473546981811523, + "learning_rate": 1.5005972402135419e-05, + "loss": 0.9221, + "step": 1098 + }, + { + "epoch": 0.8525989138867339, + "grad_norm": 8.2587251663208, + "learning_rate": 1.5002382485800392e-05, + "loss": 0.3857, + "step": 1099 + }, + { + "epoch": 0.8533747090768037, + "grad_norm": 7.858863830566406, + "learning_rate": 1.5000413639580697e-05, + "loss": 0.2906, + "step": 1100 + }, + { + "epoch": 0.8541505042668736, + "grad_norm": 7.137026786804199, + "learning_rate": 1.5000066182716175e-05, + "loss": 0.3088, + "step": 1101 + }, + { + "epoch": 0.8549262994569433, + "grad_norm": 10.72146224975586, + "learning_rate": 1.5001340171545438e-05, + "loss": 0.9182, + "step": 1102 + }, + { + "epoch": 0.8557020946470132, + "grad_norm": 4.1590495109558105, + "learning_rate": 1.5004235399496752e-05, + "loss": 0.1258, + "step": 1103 + }, + { + "epoch": 0.8564778898370831, + "grad_norm": 5.97036600112915, + "learning_rate": 1.5008751397121518e-05, + "loss": 0.1814, + "step": 1104 + }, + { + "epoch": 0.8572536850271528, + "grad_norm": 7.218429088592529, + "learning_rate": 1.50148874321704e-05, + "loss": 0.3998, + "step": 1105 + }, + { + "epoch": 0.8580294802172227, + "grad_norm": 5.190537452697754, + "learning_rate": 1.5022642509712055e-05, + "loss": 0.4094, + "step": 1106 + }, + { + "epoch": 0.8588052754072925, + "grad_norm": 7.47273588180542, + "learning_rate": 1.5032015372294449e-05, + "loss": 0.5049, + "step": 1107 + }, + { + "epoch": 0.8595810705973623, + "grad_norm": 7.2986626625061035, + "learning_rate": 1.504300450014876e-05, + "loss": 0.4656, + "step": 1108 + }, + { + "epoch": 0.8603568657874321, + "grad_norm": 12.720837593078613, + "learning_rate": 1.5055608111435787e-05, + "loss": 1.186, + "step": 1109 + }, + { + "epoch": 0.861132660977502, + "grad_norm": 7.21596622467041, + "learning_rate": 1.5069824162534886e-05, + "loss": 0.383, + "step": 1110 + }, + { + "epoch": 0.8619084561675717, + "grad_norm": 12.118733406066895, + "learning_rate": 1.5085650348375312e-05, + "loss": 1.0752, + "step": 1111 + }, + { + "epoch": 0.8626842513576416, + "grad_norm": 7.885045528411865, + "learning_rate": 1.5103084102809997e-05, + "loss": 0.4525, + "step": 1112 + }, + { + "epoch": 0.8634600465477114, + "grad_norm": 4.672414302825928, + "learning_rate": 1.5122122599031625e-05, + "loss": 0.1916, + "step": 1113 + }, + { + "epoch": 0.8642358417377812, + "grad_norm": 0.5589203834533691, + "learning_rate": 1.5142762750030996e-05, + "loss": 0.0106, + "step": 1114 + }, + { + "epoch": 0.8650116369278511, + "grad_norm": 3.377701759338379, + "learning_rate": 1.5165001209097561e-05, + "loss": 0.2125, + "step": 1115 + }, + { + "epoch": 0.8657874321179209, + "grad_norm": 10.249991416931152, + "learning_rate": 1.5188834370362092e-05, + "loss": 0.8533, + "step": 1116 + }, + { + "epoch": 0.8665632273079907, + "grad_norm": 5.519756317138672, + "learning_rate": 1.5214258369381337e-05, + "loss": 0.2105, + "step": 1117 + }, + { + "epoch": 0.8673390224980605, + "grad_norm": 11.169464111328125, + "learning_rate": 1.5241269083764644e-05, + "loss": 0.9959, + "step": 1118 + }, + { + "epoch": 0.8681148176881304, + "grad_norm": 8.748883247375488, + "learning_rate": 1.5269862133842377e-05, + "loss": 0.4655, + "step": 1119 + }, + { + "epoch": 0.8688906128782001, + "grad_norm": 0.7034602165222168, + "learning_rate": 1.5300032883376055e-05, + "loss": 0.0331, + "step": 1120 + }, + { + "epoch": 0.86966640806827, + "grad_norm": 9.566228866577148, + "learning_rate": 1.533177644031011e-05, + "loss": 0.4181, + "step": 1121 + }, + { + "epoch": 0.8704422032583398, + "grad_norm": 8.45223331451416, + "learning_rate": 1.5365087657565108e-05, + "loss": 0.2587, + "step": 1122 + }, + { + "epoch": 0.8712179984484096, + "grad_norm": 6.50554084777832, + "learning_rate": 1.539996113387233e-05, + "loss": 0.2525, + "step": 1123 + }, + { + "epoch": 0.8719937936384794, + "grad_norm": 0.9123693108558655, + "learning_rate": 1.543639121464955e-05, + "loss": 0.0117, + "step": 1124 + }, + { + "epoch": 0.8727695888285493, + "grad_norm": 7.313129425048828, + "learning_rate": 1.547437199291791e-05, + "loss": 0.2255, + "step": 1125 + }, + { + "epoch": 0.873545384018619, + "grad_norm": 6.8266754150390625, + "learning_rate": 1.551389731025972e-05, + "loss": 0.2137, + "step": 1126 + }, + { + "epoch": 0.8743211792086889, + "grad_norm": 9.998234748840332, + "learning_rate": 1.5554960757817003e-05, + "loss": 0.6591, + "step": 1127 + }, + { + "epoch": 0.8750969743987588, + "grad_norm": 5.652191162109375, + "learning_rate": 1.5597555677330677e-05, + "loss": 0.1919, + "step": 1128 + }, + { + "epoch": 0.8758727695888285, + "grad_norm": 10.237956047058105, + "learning_rate": 1.5641675162220136e-05, + "loss": 0.496, + "step": 1129 + }, + { + "epoch": 0.8766485647788984, + "grad_norm": 6.954111576080322, + "learning_rate": 1.5687312058703172e-05, + "loss": 0.1756, + "step": 1130 + }, + { + "epoch": 0.8774243599689682, + "grad_norm": 9.590442657470703, + "learning_rate": 1.5734458966955874e-05, + "loss": 0.769, + "step": 1131 + }, + { + "epoch": 0.878200155159038, + "grad_norm": 9.373478889465332, + "learning_rate": 1.578310824231251e-05, + "loss": 0.8033, + "step": 1132 + }, + { + "epoch": 0.8789759503491078, + "grad_norm": 10.543661117553711, + "learning_rate": 1.5833251996505076e-05, + "loss": 0.7703, + "step": 1133 + }, + { + "epoch": 0.8797517455391777, + "grad_norm": 6.544073581695557, + "learning_rate": 1.5884882098942327e-05, + "loss": 0.1636, + "step": 1134 + }, + { + "epoch": 0.8805275407292474, + "grad_norm": 5.172294616699219, + "learning_rate": 1.5937990178028133e-05, + "loss": 0.1441, + "step": 1135 + }, + { + "epoch": 0.8813033359193173, + "grad_norm": 16.34269905090332, + "learning_rate": 1.5992567622518885e-05, + "loss": 1.2285, + "step": 1136 + }, + { + "epoch": 0.8820791311093871, + "grad_norm": 9.113872528076172, + "learning_rate": 1.6048605582919785e-05, + "loss": 0.4601, + "step": 1137 + }, + { + "epoch": 0.8828549262994569, + "grad_norm": 11.077788352966309, + "learning_rate": 1.6106094972919734e-05, + "loss": 0.74, + "step": 1138 + }, + { + "epoch": 0.8836307214895268, + "grad_norm": 5.6204986572265625, + "learning_rate": 1.616502647086466e-05, + "loss": 0.2197, + "step": 1139 + }, + { + "epoch": 0.8844065166795966, + "grad_norm": 7.835806846618652, + "learning_rate": 1.6225390521268953e-05, + "loss": 0.3052, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_Qnli-dev_cosine_accuracy": 0.68359375, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8345149159431458, + "eval_Qnli-dev_cosine_ap": 0.7152209113976348, + "eval_Qnli-dev_cosine_f1": 0.680067001675042, + "eval_Qnli-dev_cosine_f1_threshold": 0.7428933382034302, + "eval_Qnli-dev_cosine_precision": 0.5623268698060941, + "eval_Qnli-dev_cosine_recall": 0.8601694915254238, + "eval_Qnli-dev_dot_accuracy": 0.630859375, + "eval_Qnli-dev_dot_accuracy_threshold": 396.8427734375, + "eval_Qnli-dev_dot_ap": 0.6042577502907576, + "eval_Qnli-dev_dot_f1": 0.6656534954407294, + "eval_Qnli-dev_dot_f1_threshold": 320.732177734375, + "eval_Qnli-dev_dot_precision": 0.518957345971564, + "eval_Qnli-dev_dot_recall": 0.9279661016949152, + "eval_Qnli-dev_euclidean_accuracy": 0.685546875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 13.120853424072266, + "eval_Qnli-dev_euclidean_ap": 0.7189115262994822, + "eval_Qnli-dev_euclidean_f1": 0.6818181818181818, + "eval_Qnli-dev_euclidean_f1_threshold": 16.184667587280273, + "eval_Qnli-dev_euclidean_precision": 0.5526315789473685, + "eval_Qnli-dev_euclidean_recall": 0.8898305084745762, + "eval_Qnli-dev_manhattan_accuracy": 0.68359375, + "eval_Qnli-dev_manhattan_accuracy_threshold": 233.14083862304688, + "eval_Qnli-dev_manhattan_ap": 0.7203492460774932, + "eval_Qnli-dev_manhattan_f1": 0.678513731825525, + "eval_Qnli-dev_manhattan_f1_threshold": 293.5357666015625, + "eval_Qnli-dev_manhattan_precision": 0.5483028720626631, + "eval_Qnli-dev_manhattan_recall": 0.8898305084745762, + "eval_Qnli-dev_max_accuracy": 0.685546875, + "eval_Qnli-dev_max_accuracy_threshold": 396.8427734375, + "eval_Qnli-dev_max_ap": 0.7203492460774932, + "eval_Qnli-dev_max_f1": 0.6818181818181818, + "eval_Qnli-dev_max_f1_threshold": 320.732177734375, + "eval_Qnli-dev_max_precision": 0.5623268698060941, + "eval_Qnli-dev_max_recall": 0.9279661016949152, + "eval_allNLI-dev_cosine_accuracy": 0.71875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8607510328292847, + "eval_allNLI-dev_cosine_ap": 0.5979509144611539, + "eval_allNLI-dev_cosine_f1": 0.6076923076923078, + "eval_allNLI-dev_cosine_f1_threshold": 0.7347486019134521, + "eval_allNLI-dev_cosine_precision": 0.4553314121037464, + "eval_allNLI-dev_cosine_recall": 0.9132947976878613, + "eval_allNLI-dev_dot_accuracy": 0.689453125, + "eval_allNLI-dev_dot_accuracy_threshold": 420.3511657714844, + "eval_allNLI-dev_dot_ap": 0.5034927351618514, + "eval_allNLI-dev_dot_f1": 0.5812619502868069, + "eval_allNLI-dev_dot_f1_threshold": 315.4132080078125, + "eval_allNLI-dev_dot_precision": 0.4342857142857143, + "eval_allNLI-dev_dot_recall": 0.8786127167630058, + "eval_allNLI-dev_euclidean_accuracy": 0.724609375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.658323287963867, + "eval_allNLI-dev_euclidean_ap": 0.5989846741191746, + "eval_allNLI-dev_euclidean_f1": 0.6125, + "eval_allNLI-dev_euclidean_f1_threshold": 14.394068717956543, + "eval_allNLI-dev_euclidean_precision": 0.4788273615635179, + "eval_allNLI-dev_euclidean_recall": 0.8497109826589595, + "eval_allNLI-dev_manhattan_accuracy": 0.7109375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 173.71578979492188, + "eval_allNLI-dev_manhattan_ap": 0.5979693176622969, + "eval_allNLI-dev_manhattan_f1": 0.6262135922330098, + "eval_allNLI-dev_manhattan_f1_threshold": 230.6854248046875, + "eval_allNLI-dev_manhattan_precision": 0.5397489539748954, + "eval_allNLI-dev_manhattan_recall": 0.7456647398843931, + "eval_allNLI-dev_max_accuracy": 0.724609375, + "eval_allNLI-dev_max_accuracy_threshold": 420.3511657714844, + "eval_allNLI-dev_max_ap": 0.5989846741191746, + "eval_allNLI-dev_max_f1": 0.6262135922330098, + "eval_allNLI-dev_max_f1_threshold": 315.4132080078125, + "eval_allNLI-dev_max_precision": 0.5397489539748954, + "eval_allNLI-dev_max_recall": 0.9132947976878613, + "eval_sequential_score": 0.7203492460774932, + "eval_sts-test_pearson_cosine": 0.8244380790655982, + "eval_sts-test_pearson_dot": 0.7800625356443638, + "eval_sts-test_pearson_euclidean": 0.8535878830570102, + "eval_sts-test_pearson_manhattan": 0.8516121660071754, + "eval_sts-test_pearson_max": 0.8535878830570102, + "eval_sts-test_spearman_cosine": 0.8607125114149644, + "eval_sts-test_spearman_dot": 0.7723289044207345, + "eval_sts-test_spearman_euclidean": 0.8532235668261593, + "eval_sts-test_spearman_manhattan": 0.8517668237653463, + "eval_sts-test_spearman_max": 0.8607125114149644, + "eval_vitaminc-pairs_loss": 3.8809142112731934, + "eval_vitaminc-pairs_runtime": 5.9076, + "eval_vitaminc-pairs_samples_per_second": 21.667, + "eval_vitaminc-pairs_steps_per_second": 0.339, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_negation-triplets_loss": 1.4353677034378052, + "eval_negation-triplets_runtime": 1.124, + "eval_negation-triplets_samples_per_second": 113.883, + "eval_negation-triplets_steps_per_second": 1.779, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_scitail-pairs-pos_loss": 0.24866406619548798, + "eval_scitail-pairs-pos_runtime": 1.2724, + "eval_scitail-pairs-pos_samples_per_second": 100.599, + "eval_scitail-pairs-pos_steps_per_second": 1.572, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_scitail-pairs-qa_loss": 0.003725280985236168, + "eval_scitail-pairs-qa_runtime": 0.9241, + "eval_scitail-pairs-qa_samples_per_second": 138.518, + "eval_scitail-pairs-qa_steps_per_second": 2.164, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_xsum-pairs_loss": 0.2623598575592041, + "eval_xsum-pairs_runtime": 4.32, + "eval_xsum-pairs_samples_per_second": 29.63, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_sciq_pairs_loss": 0.07306865602731705, + "eval_sciq_pairs_runtime": 5.6152, + "eval_sciq_pairs_samples_per_second": 22.795, + "eval_sciq_pairs_steps_per_second": 0.356, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_qasc_pairs_loss": 0.27778807282447815, + "eval_qasc_pairs_runtime": 0.9869, + "eval_qasc_pairs_samples_per_second": 129.705, + "eval_qasc_pairs_steps_per_second": 2.027, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_openbookqa_pairs_loss": 1.1855545043945312, + "eval_openbookqa_pairs_runtime": 0.8692, + "eval_openbookqa_pairs_samples_per_second": 147.267, + "eval_openbookqa_pairs_steps_per_second": 2.301, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_msmarco_pairs_loss": 0.654990553855896, + "eval_msmarco_pairs_runtime": 1.9711, + "eval_msmarco_pairs_samples_per_second": 64.938, + "eval_msmarco_pairs_steps_per_second": 1.015, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_nq_pairs_loss": 0.9545596837997437, + "eval_nq_pairs_runtime": 4.4163, + "eval_nq_pairs_samples_per_second": 28.984, + "eval_nq_pairs_steps_per_second": 0.453, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_trivia_pairs_loss": 0.7380074858665466, + "eval_trivia_pairs_runtime": 4.5277, + "eval_trivia_pairs_samples_per_second": 28.271, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_gooaq_pairs_loss": 0.5968726873397827, + "eval_gooaq_pairs_runtime": 1.2833, + "eval_gooaq_pairs_samples_per_second": 99.74, + "eval_gooaq_pairs_steps_per_second": 1.558, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_paws-pos_loss": 0.034593094140291214, + "eval_paws-pos_runtime": 1.0416, + "eval_paws-pos_samples_per_second": 122.882, + "eval_paws-pos_steps_per_second": 1.92, + "step": 1140 + }, + { + "epoch": 0.8844065166795966, + "eval_global_dataset_loss": 0.5444615483283997, + "eval_global_dataset_runtime": 21.1844, + "eval_global_dataset_samples_per_second": 19.637, + "eval_global_dataset_steps_per_second": 0.33, + "step": 1140 + }, + { + "epoch": 0.8851823118696664, + "grad_norm": 11.804282188415527, + "learning_rate": 1.6287177336364887e-05, + "loss": 0.6587, + "step": 1141 + }, + { + "epoch": 0.8859581070597362, + "grad_norm": 8.900123596191406, + "learning_rate": 1.635037689768963e-05, + "loss": 0.3953, + "step": 1142 + }, + { + "epoch": 0.8867339022498061, + "grad_norm": 8.825578689575195, + "learning_rate": 1.6414978957709702e-05, + "loss": 0.6124, + "step": 1143 + }, + { + "epoch": 0.8875096974398758, + "grad_norm": 12.496349334716797, + "learning_rate": 1.648097304148257e-05, + "loss": 0.952, + "step": 1144 + }, + { + "epoch": 0.8882854926299457, + "grad_norm": 9.799676895141602, + "learning_rate": 1.654834844835513e-05, + "loss": 0.9514, + "step": 1145 + }, + { + "epoch": 0.8890612878200155, + "grad_norm": 4.692087650299072, + "learning_rate": 1.6617094253698735e-05, + "loss": 0.2201, + "step": 1146 + }, + { + "epoch": 0.8898370830100854, + "grad_norm": 7.2312798500061035, + "learning_rate": 1.668719931068061e-05, + "loss": 0.3446, + "step": 1147 + }, + { + "epoch": 0.8906128782001551, + "grad_norm": 9.718889236450195, + "learning_rate": 1.6758652252071257e-05, + "loss": 0.7141, + "step": 1148 + }, + { + "epoch": 0.891388673390225, + "grad_norm": 9.90417766571045, + "learning_rate": 1.6831441492087577e-05, + "loss": 0.7227, + "step": 1149 + }, + { + "epoch": 0.8921644685802949, + "grad_norm": 8.608745574951172, + "learning_rate": 1.6905555228271483e-05, + "loss": 0.7811, + "step": 1150 + }, + { + "epoch": 0.8929402637703646, + "grad_norm": 5.626070976257324, + "learning_rate": 1.6980981443403598e-05, + "loss": 0.3033, + "step": 1151 + }, + { + "epoch": 0.8937160589604345, + "grad_norm": 10.469879150390625, + "learning_rate": 1.7057707907451818e-05, + "loss": 0.7532, + "step": 1152 + }, + { + "epoch": 0.8944918541505043, + "grad_norm": 0.7686850428581238, + "learning_rate": 1.7135722179554325e-05, + "loss": 0.0198, + "step": 1153 + }, + { + "epoch": 0.8952676493405741, + "grad_norm": 10.882933616638184, + "learning_rate": 1.721501161003685e-05, + "loss": 1.0878, + "step": 1154 + }, + { + "epoch": 0.8960434445306439, + "grad_norm": 11.984310150146484, + "learning_rate": 1.729556334246373e-05, + "loss": 0.9868, + "step": 1155 + }, + { + "epoch": 0.8968192397207138, + "grad_norm": 4.480854034423828, + "learning_rate": 1.737736431572258e-05, + "loss": 0.1472, + "step": 1156 + }, + { + "epoch": 0.8975950349107835, + "grad_norm": 7.384640693664551, + "learning_rate": 1.7460401266142013e-05, + "loss": 0.8511, + "step": 1157 + }, + { + "epoch": 0.8983708301008534, + "grad_norm": 7.989016532897949, + "learning_rate": 1.754466072964236e-05, + "loss": 0.3186, + "step": 1158 + }, + { + "epoch": 0.8991466252909232, + "grad_norm": 11.410018920898438, + "learning_rate": 1.7630129043918775e-05, + "loss": 0.7747, + "step": 1159 + }, + { + "epoch": 0.899922420480993, + "grad_norm": 1.7008391618728638, + "learning_rate": 1.7716792350656523e-05, + "loss": 0.0443, + "step": 1160 + }, + { + "epoch": 0.9006982156710628, + "grad_norm": 6.994379043579102, + "learning_rate": 1.7804636597778053e-05, + "loss": 0.2762, + "step": 1161 + }, + { + "epoch": 0.9014740108611327, + "grad_norm": 0.0, + "learning_rate": 1.7893647541721476e-05, + "loss": 0.0, + "step": 1162 + }, + { + "epoch": 0.9022498060512025, + "grad_norm": 6.318467617034912, + "learning_rate": 1.79838107497501e-05, + "loss": 0.2446, + "step": 1163 + }, + { + "epoch": 0.9030256012412723, + "grad_norm": 5.270249843597412, + "learning_rate": 1.8075111602292622e-05, + "loss": 0.1827, + "step": 1164 + }, + { + "epoch": 0.9038013964313422, + "grad_norm": 7.074584007263184, + "learning_rate": 1.8167535295313654e-05, + "loss": 0.3043, + "step": 1165 + }, + { + "epoch": 0.9045771916214119, + "grad_norm": 14.505364418029785, + "learning_rate": 1.8261066842714093e-05, + "loss": 1.1194, + "step": 1166 + }, + { + "epoch": 0.9053529868114818, + "grad_norm": 8.937790870666504, + "learning_rate": 1.835569107876111e-05, + "loss": 0.344, + "step": 1167 + }, + { + "epoch": 0.9061287820015516, + "grad_norm": 14.606131553649902, + "learning_rate": 1.8451392660547147e-05, + "loss": 1.1798, + "step": 1168 + }, + { + "epoch": 0.9069045771916214, + "grad_norm": 11.8740873336792, + "learning_rate": 1.8548156070477738e-05, + "loss": 0.9406, + "step": 1169 + }, + { + "epoch": 0.9076803723816912, + "grad_norm": 1.1290702819824219, + "learning_rate": 1.86459656187876e-05, + "loss": 0.0222, + "step": 1170 + }, + { + "epoch": 0.9084561675717611, + "grad_norm": 5.582184791564941, + "learning_rate": 1.8744805446084695e-05, + "loss": 0.1722, + "step": 1171 + }, + { + "epoch": 0.9092319627618308, + "grad_norm": 5.3683881759643555, + "learning_rate": 1.8844659525921707e-05, + "loss": 0.2667, + "step": 1172 + }, + { + "epoch": 0.9100077579519007, + "grad_norm": 7.846807956695557, + "learning_rate": 1.8945511667394715e-05, + "loss": 0.2907, + "step": 1173 + }, + { + "epoch": 0.9107835531419706, + "grad_norm": 10.131175994873047, + "learning_rate": 1.9047345517768438e-05, + "loss": 0.6799, + "step": 1174 + }, + { + "epoch": 0.9115593483320403, + "grad_norm": 16.07407569885254, + "learning_rate": 1.9150144565127777e-05, + "loss": 2.0621, + "step": 1175 + }, + { + "epoch": 0.9123351435221102, + "grad_norm": 7.773518085479736, + "learning_rate": 1.925389214105515e-05, + "loss": 0.396, + "step": 1176 + }, + { + "epoch": 0.91311093871218, + "grad_norm": 5.964335918426514, + "learning_rate": 1.9358571423333195e-05, + "loss": 0.3186, + "step": 1177 + }, + { + "epoch": 0.9138867339022498, + "grad_norm": 7.1874613761901855, + "learning_rate": 1.9464165438672444e-05, + "loss": 0.2532, + "step": 1178 + }, + { + "epoch": 0.9146625290923196, + "grad_norm": 1.6531507968902588, + "learning_rate": 1.9570657065463427e-05, + "loss": 0.0561, + "step": 1179 + }, + { + "epoch": 0.9154383242823895, + "grad_norm": 6.925435543060303, + "learning_rate": 1.9678029036552903e-05, + "loss": 0.24, + "step": 1180 + }, + { + "epoch": 0.9162141194724592, + "grad_norm": 0.3911866843700409, + "learning_rate": 1.9786263942043613e-05, + "loss": 0.0055, + "step": 1181 + }, + { + "epoch": 0.9169899146625291, + "grad_norm": 13.614795684814453, + "learning_rate": 1.989534423211729e-05, + "loss": 1.0443, + "step": 1182 + }, + { + "epoch": 0.9177657098525989, + "grad_norm": 9.041915893554688, + "learning_rate": 2.0005252219880197e-05, + "loss": 0.5319, + "step": 1183 + }, + { + "epoch": 0.9185415050426687, + "grad_norm": 32.4137077331543, + "learning_rate": 2.0115970084231057e-05, + "loss": 0.1738, + "step": 1184 + }, + { + "epoch": 0.9193173002327386, + "grad_norm": 19.339521408081055, + "learning_rate": 2.0227479872750607e-05, + "loss": 1.0717, + "step": 1185 + }, + { + "epoch": 0.9200930954228084, + "grad_norm": 0.8835486769676208, + "learning_rate": 2.033976350461255e-05, + "loss": 0.0151, + "step": 1186 + }, + { + "epoch": 0.9208688906128782, + "grad_norm": 8.5880765914917, + "learning_rate": 2.0452802773515245e-05, + "loss": 0.2959, + "step": 1187 + }, + { + "epoch": 0.921644685802948, + "grad_norm": 14.47278118133545, + "learning_rate": 2.0566579350633815e-05, + "loss": 1.1981, + "step": 1188 + }, + { + "epoch": 0.9224204809930179, + "grad_norm": 5.098402500152588, + "learning_rate": 2.068107478759206e-05, + "loss": 0.1575, + "step": 1189 + }, + { + "epoch": 0.9231962761830876, + "grad_norm": 8.61888313293457, + "learning_rate": 2.0796270519453798e-05, + "loss": 0.5027, + "step": 1190 + }, + { + "epoch": 0.9239720713731575, + "grad_norm": 13.766504287719727, + "learning_rate": 2.0912147867733068e-05, + "loss": 0.9443, + "step": 1191 + }, + { + "epoch": 0.9247478665632273, + "grad_norm": 9.993270874023438, + "learning_rate": 2.1028688043422773e-05, + "loss": 0.7371, + "step": 1192 + }, + { + "epoch": 0.9255236617532971, + "grad_norm": 10.331473350524902, + "learning_rate": 2.1145872150041262e-05, + "loss": 0.6714, + "step": 1193 + }, + { + "epoch": 0.9262994569433669, + "grad_norm": 0.0, + "learning_rate": 2.1263681186696266e-05, + "loss": 0.0, + "step": 1194 + }, + { + "epoch": 0.9270752521334368, + "grad_norm": 5.50454044342041, + "learning_rate": 2.1382096051165844e-05, + "loss": 0.2574, + "step": 1195 + }, + { + "epoch": 0.9278510473235065, + "grad_norm": 0.799031674861908, + "learning_rate": 2.1501097542995717e-05, + "loss": 0.011, + "step": 1196 + }, + { + "epoch": 0.9286268425135764, + "grad_norm": 7.612067222595215, + "learning_rate": 2.1620666366612538e-05, + "loss": 0.2799, + "step": 1197 + }, + { + "epoch": 0.9294026377036463, + "grad_norm": 12.70689868927002, + "learning_rate": 2.1740783134452586e-05, + "loss": 1.117, + "step": 1198 + }, + { + "epoch": 0.930178432893716, + "grad_norm": 7.956214427947998, + "learning_rate": 2.1861428370105344e-05, + "loss": 0.5258, + "step": 1199 + }, + { + "epoch": 0.9309542280837859, + "grad_norm": 0.0, + "learning_rate": 2.198258251147156e-05, + "loss": 0.0, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_Qnli-dev_cosine_accuracy": 0.673828125, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.819699764251709, + "eval_Qnli-dev_cosine_ap": 0.7140421117308015, + "eval_Qnli-dev_cosine_f1": 0.6794258373205743, + "eval_Qnli-dev_cosine_f1_threshold": 0.7247879505157471, + "eval_Qnli-dev_cosine_precision": 0.5447570332480819, + "eval_Qnli-dev_cosine_recall": 0.902542372881356, + "eval_Qnli-dev_dot_accuracy": 0.626953125, + "eval_Qnli-dev_dot_accuracy_threshold": 433.1668701171875, + "eval_Qnli-dev_dot_ap": 0.6284678333777363, + "eval_Qnli-dev_dot_f1": 0.6606875934230194, + "eval_Qnli-dev_dot_f1_threshold": 316.43701171875, + "eval_Qnli-dev_dot_precision": 0.5103926096997691, + "eval_Qnli-dev_dot_recall": 0.9364406779661016, + "eval_Qnli-dev_euclidean_accuracy": 0.6796875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 13.113260269165039, + "eval_Qnli-dev_euclidean_ap": 0.7157487199915942, + "eval_Qnli-dev_euclidean_f1": 0.679304897314376, + "eval_Qnli-dev_euclidean_f1_threshold": 16.397951126098633, + "eval_Qnli-dev_euclidean_precision": 0.5415617128463476, + "eval_Qnli-dev_euclidean_recall": 0.9110169491525424, + "eval_Qnli-dev_manhattan_accuracy": 0.6875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 235.4864959716797, + "eval_Qnli-dev_manhattan_ap": 0.7159983583480086, + "eval_Qnli-dev_manhattan_f1": 0.6831530139103554, + "eval_Qnli-dev_manhattan_f1_threshold": 305.32904052734375, + "eval_Qnli-dev_manhattan_precision": 0.537712895377129, + "eval_Qnli-dev_manhattan_recall": 0.9364406779661016, + "eval_Qnli-dev_max_accuracy": 0.6875, + "eval_Qnli-dev_max_accuracy_threshold": 433.1668701171875, + "eval_Qnli-dev_max_ap": 0.7159983583480086, + "eval_Qnli-dev_max_f1": 0.6831530139103554, + "eval_Qnli-dev_max_f1_threshold": 316.43701171875, + "eval_Qnli-dev_max_precision": 0.5447570332480819, + "eval_Qnli-dev_max_recall": 0.9364406779661016, + "eval_allNLI-dev_cosine_accuracy": 0.70703125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8791033029556274, + "eval_allNLI-dev_cosine_ap": 0.5812724479020475, + "eval_allNLI-dev_cosine_f1": 0.5897435897435896, + "eval_allNLI-dev_cosine_f1_threshold": 0.8391410112380981, + "eval_allNLI-dev_cosine_precision": 0.5299539170506913, + "eval_allNLI-dev_cosine_recall": 0.6647398843930635, + "eval_allNLI-dev_dot_accuracy": 0.681640625, + "eval_allNLI-dev_dot_accuracy_threshold": 421.3893737792969, + "eval_allNLI-dev_dot_ap": 0.4929905194023162, + "eval_allNLI-dev_dot_f1": 0.5665399239543726, + "eval_allNLI-dev_dot_f1_threshold": 312.20892333984375, + "eval_allNLI-dev_dot_precision": 0.42209631728045327, + "eval_allNLI-dev_dot_recall": 0.861271676300578, + "eval_allNLI-dev_euclidean_accuracy": 0.70703125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.258050918579102, + "eval_allNLI-dev_euclidean_ap": 0.5816093466349955, + "eval_allNLI-dev_euclidean_f1": 0.602020202020202, + "eval_allNLI-dev_euclidean_f1_threshold": 14.197221755981445, + "eval_allNLI-dev_euclidean_precision": 0.46273291925465837, + "eval_allNLI-dev_euclidean_recall": 0.861271676300578, + "eval_allNLI-dev_manhattan_accuracy": 0.708984375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 164.1345672607422, + "eval_allNLI-dev_manhattan_ap": 0.5823775541416165, + "eval_allNLI-dev_manhattan_f1": 0.6047619047619047, + "eval_allNLI-dev_manhattan_f1_threshold": 230.2345428466797, + "eval_allNLI-dev_manhattan_precision": 0.5141700404858299, + "eval_allNLI-dev_manhattan_recall": 0.7341040462427746, + "eval_allNLI-dev_max_accuracy": 0.708984375, + "eval_allNLI-dev_max_accuracy_threshold": 421.3893737792969, + "eval_allNLI-dev_max_ap": 0.5823775541416165, + "eval_allNLI-dev_max_f1": 0.6047619047619047, + "eval_allNLI-dev_max_f1_threshold": 312.20892333984375, + "eval_allNLI-dev_max_precision": 0.5299539170506913, + "eval_allNLI-dev_max_recall": 0.861271676300578, + "eval_sequential_score": 0.7159983583480086, + "eval_sts-test_pearson_cosine": 0.828512762156324, + "eval_sts-test_pearson_dot": 0.79157044100581, + "eval_sts-test_pearson_euclidean": 0.861384500873538, + "eval_sts-test_pearson_manhattan": 0.8607981389721269, + "eval_sts-test_pearson_max": 0.861384500873538, + "eval_sts-test_spearman_cosine": 0.8658471816444908, + "eval_sts-test_spearman_dot": 0.7803564304403852, + "eval_sts-test_spearman_euclidean": 0.8599981917806349, + "eval_sts-test_spearman_manhattan": 0.8604580109914315, + "eval_sts-test_spearman_max": 0.8658471816444908, + "eval_vitaminc-pairs_loss": 3.8138790130615234, + "eval_vitaminc-pairs_runtime": 5.4948, + "eval_vitaminc-pairs_samples_per_second": 23.295, + "eval_vitaminc-pairs_steps_per_second": 0.364, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_negation-triplets_loss": 1.4483975172042847, + "eval_negation-triplets_runtime": 1.1184, + "eval_negation-triplets_samples_per_second": 114.454, + "eval_negation-triplets_steps_per_second": 1.788, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_scitail-pairs-pos_loss": 0.1757652461528778, + "eval_scitail-pairs-pos_runtime": 1.2664, + "eval_scitail-pairs-pos_samples_per_second": 101.071, + "eval_scitail-pairs-pos_steps_per_second": 1.579, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_scitail-pairs-qa_loss": 0.0022748063784092665, + "eval_scitail-pairs-qa_runtime": 0.9284, + "eval_scitail-pairs-qa_samples_per_second": 137.87, + "eval_scitail-pairs-qa_steps_per_second": 2.154, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_xsum-pairs_loss": 0.26025938987731934, + "eval_xsum-pairs_runtime": 4.3247, + "eval_xsum-pairs_samples_per_second": 29.598, + "eval_xsum-pairs_steps_per_second": 0.462, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_sciq_pairs_loss": 0.07115399837493896, + "eval_sciq_pairs_runtime": 5.5994, + "eval_sciq_pairs_samples_per_second": 22.86, + "eval_sciq_pairs_steps_per_second": 0.357, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_qasc_pairs_loss": 0.2785109579563141, + "eval_qasc_pairs_runtime": 0.9873, + "eval_qasc_pairs_samples_per_second": 129.641, + "eval_qasc_pairs_steps_per_second": 2.026, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_openbookqa_pairs_loss": 1.2239012718200684, + "eval_openbookqa_pairs_runtime": 0.8696, + "eval_openbookqa_pairs_samples_per_second": 147.194, + "eval_openbookqa_pairs_steps_per_second": 2.3, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_msmarco_pairs_loss": 0.6812164783477783, + "eval_msmarco_pairs_runtime": 1.966, + "eval_msmarco_pairs_samples_per_second": 65.108, + "eval_msmarco_pairs_steps_per_second": 1.017, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_nq_pairs_loss": 0.9431746006011963, + "eval_nq_pairs_runtime": 4.4165, + "eval_nq_pairs_samples_per_second": 28.982, + "eval_nq_pairs_steps_per_second": 0.453, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_trivia_pairs_loss": 0.6358234882354736, + "eval_trivia_pairs_runtime": 4.5305, + "eval_trivia_pairs_samples_per_second": 28.253, + "eval_trivia_pairs_steps_per_second": 0.441, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_gooaq_pairs_loss": 0.5713344216346741, + "eval_gooaq_pairs_runtime": 1.299, + "eval_gooaq_pairs_samples_per_second": 98.537, + "eval_gooaq_pairs_steps_per_second": 1.54, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_paws-pos_loss": 0.036365460604429245, + "eval_paws-pos_runtime": 1.0357, + "eval_paws-pos_samples_per_second": 123.585, + "eval_paws-pos_steps_per_second": 1.931, + "step": 1200 + }, + { + "epoch": 0.9309542280837859, + "eval_global_dataset_loss": 0.5287986397743225, + "eval_global_dataset_runtime": 21.1696, + "eval_global_dataset_samples_per_second": 19.651, + "eval_global_dataset_steps_per_second": 0.331, + "step": 1200 + }, + { + "epoch": 0.9317300232738557, + "grad_norm": 6.235774993896484, + "learning_rate": 2.210422591393517e-05, + "loss": 0.3638, + "step": 1201 + }, + { + "epoch": 0.9325058184639256, + "grad_norm": 2.04115629196167, + "learning_rate": 2.22263388535485e-05, + "loss": 0.0419, + "step": 1202 + }, + { + "epoch": 0.9332816136539953, + "grad_norm": 12.374478340148926, + "learning_rate": 2.2348901530230507e-05, + "loss": 0.9931, + "step": 1203 + }, + { + "epoch": 0.9340574088440652, + "grad_norm": 7.619632720947266, + "learning_rate": 2.2471894070977232e-05, + "loss": 0.271, + "step": 1204 + }, + { + "epoch": 0.934833204034135, + "grad_norm": 10.664977073669434, + "learning_rate": 2.2595296533084132e-05, + "loss": 0.6739, + "step": 1205 + }, + { + "epoch": 0.9356089992242048, + "grad_norm": 12.465529441833496, + "learning_rate": 2.2719088907379706e-05, + "loss": 1.0256, + "step": 1206 + }, + { + "epoch": 0.9363847944142746, + "grad_norm": 11.22776985168457, + "learning_rate": 2.2843251121469876e-05, + "loss": 0.9012, + "step": 1207 + }, + { + "epoch": 0.9371605896043445, + "grad_norm": 1.4332178831100464, + "learning_rate": 2.2967763042992678e-05, + "loss": 0.0282, + "step": 1208 + }, + { + "epoch": 0.9379363847944143, + "grad_norm": 6.596785068511963, + "learning_rate": 2.309260448288259e-05, + "loss": 0.2159, + "step": 1209 + }, + { + "epoch": 0.9387121799844841, + "grad_norm": 5.84821891784668, + "learning_rate": 2.3217755198644134e-05, + "loss": 0.2428, + "step": 1210 + }, + { + "epoch": 0.939487975174554, + "grad_norm": 10.343107223510742, + "learning_rate": 2.3343194897634093e-05, + "loss": 0.7847, + "step": 1211 + }, + { + "epoch": 0.9402637703646237, + "grad_norm": 13.150612831115723, + "learning_rate": 2.3468903240351873e-05, + "loss": 1.1893, + "step": 1212 + }, + { + "epoch": 0.9410395655546936, + "grad_norm": 6.210269451141357, + "learning_rate": 2.3594859843737482e-05, + "loss": 0.2918, + "step": 1213 + }, + { + "epoch": 0.9418153607447634, + "grad_norm": 8.447084426879883, + "learning_rate": 2.3721044284476523e-05, + "loss": 0.697, + "step": 1214 + }, + { + "epoch": 0.9425911559348332, + "grad_norm": 8.582613945007324, + "learning_rate": 2.38474361023118e-05, + "loss": 0.7615, + "step": 1215 + }, + { + "epoch": 0.943366951124903, + "grad_norm": 0.0, + "learning_rate": 2.3974014803360795e-05, + "loss": 0.0, + "step": 1216 + }, + { + "epoch": 0.9441427463149729, + "grad_norm": 4.840447425842285, + "learning_rate": 2.4100759863438704e-05, + "loss": 0.1928, + "step": 1217 + }, + { + "epoch": 0.9449185415050426, + "grad_norm": 12.083598136901855, + "learning_rate": 2.4227650731386304e-05, + "loss": 0.8963, + "step": 1218 + }, + { + "epoch": 0.9456943366951125, + "grad_norm": 7.982798099517822, + "learning_rate": 2.4354666832402306e-05, + "loss": 0.3505, + "step": 1219 + }, + { + "epoch": 0.9464701318851824, + "grad_norm": 11.88045883178711, + "learning_rate": 2.44817875713794e-05, + "loss": 0.9786, + "step": 1220 + }, + { + "epoch": 0.9472459270752521, + "grad_norm": 7.532604217529297, + "learning_rate": 2.460899233624368e-05, + "loss": 0.6061, + "step": 1221 + }, + { + "epoch": 0.948021722265322, + "grad_norm": 6.655893325805664, + "learning_rate": 2.473626050129683e-05, + "loss": 0.2607, + "step": 1222 + }, + { + "epoch": 0.9487975174553918, + "grad_norm": 8.368523597717285, + "learning_rate": 2.4863571430560438e-05, + "loss": 0.5026, + "step": 1223 + }, + { + "epoch": 0.9495733126454616, + "grad_norm": 7.4424028396606445, + "learning_rate": 2.4990904481122106e-05, + "loss": 0.3365, + "step": 1224 + }, + { + "epoch": 0.9503491078355314, + "grad_norm": 8.568815231323242, + "learning_rate": 2.5118239006482526e-05, + "loss": 0.4172, + "step": 1225 + }, + { + "epoch": 0.9511249030256013, + "grad_norm": 7.632470607757568, + "learning_rate": 2.5245554359903254e-05, + "loss": 0.6302, + "step": 1226 + }, + { + "epoch": 0.951900698215671, + "grad_norm": 10.907419204711914, + "learning_rate": 2.5372829897754545e-05, + "loss": 0.8007, + "step": 1227 + }, + { + "epoch": 0.9526764934057409, + "grad_norm": 6.9367828369140625, + "learning_rate": 2.550004498286251e-05, + "loss": 0.358, + "step": 1228 + }, + { + "epoch": 0.9534522885958107, + "grad_norm": 13.178356170654297, + "learning_rate": 2.5627178987855455e-05, + "loss": 1.1527, + "step": 1229 + }, + { + "epoch": 0.9542280837858805, + "grad_norm": 5.810834884643555, + "learning_rate": 2.5754211298508453e-05, + "loss": 0.2319, + "step": 1230 + }, + { + "epoch": 0.9550038789759504, + "grad_norm": 8.390653610229492, + "learning_rate": 2.588112131708586e-05, + "loss": 0.4708, + "step": 1231 + }, + { + "epoch": 0.9557796741660202, + "grad_norm": 10.020398139953613, + "learning_rate": 2.6007888465681197e-05, + "loss": 0.6828, + "step": 1232 + }, + { + "epoch": 0.95655546935609, + "grad_norm": 7.844973564147949, + "learning_rate": 2.613449218955368e-05, + "loss": 0.3424, + "step": 1233 + }, + { + "epoch": 0.9573312645461598, + "grad_norm": 11.724699974060059, + "learning_rate": 2.6260911960461184e-05, + "loss": 0.933, + "step": 1234 + }, + { + "epoch": 0.9581070597362297, + "grad_norm": 7.519054412841797, + "learning_rate": 2.6387127279988734e-05, + "loss": 0.3672, + "step": 1235 + }, + { + "epoch": 0.9588828549262994, + "grad_norm": 8.2330904006958, + "learning_rate": 2.651311768287222e-05, + "loss": 0.4457, + "step": 1236 + }, + { + "epoch": 0.9596586501163693, + "grad_norm": 7.875716209411621, + "learning_rate": 2.663886274031683e-05, + "loss": 0.7641, + "step": 1237 + }, + { + "epoch": 0.9604344453064391, + "grad_norm": 8.273968696594238, + "learning_rate": 2.6764342063309447e-05, + "loss": 0.8338, + "step": 1238 + }, + { + "epoch": 0.9612102404965089, + "grad_norm": 10.740669250488281, + "learning_rate": 2.6889535305924614e-05, + "loss": 0.8975, + "step": 1239 + }, + { + "epoch": 0.9619860356865787, + "grad_norm": 7.118108749389648, + "learning_rate": 2.701442216862357e-05, + "loss": 0.4062, + "step": 1240 + }, + { + "epoch": 0.9627618308766486, + "grad_norm": 7.399252414703369, + "learning_rate": 2.7138982401545726e-05, + "loss": 0.3386, + "step": 1241 + }, + { + "epoch": 0.9635376260667183, + "grad_norm": 0.0, + "learning_rate": 2.7263195807792076e-05, + "loss": 0.0, + "step": 1242 + }, + { + "epoch": 0.9643134212567882, + "grad_norm": 6.8241376876831055, + "learning_rate": 2.7387042246700062e-05, + "loss": 0.2891, + "step": 1243 + }, + { + "epoch": 0.9650892164468581, + "grad_norm": 0.0, + "learning_rate": 2.7510501637109275e-05, + "loss": 0.0, + "step": 1244 + }, + { + "epoch": 0.9658650116369278, + "grad_norm": 5.6250224113464355, + "learning_rate": 2.763355396061754e-05, + "loss": 0.2005, + "step": 1245 + }, + { + "epoch": 0.9666408068269977, + "grad_norm": 11.53515911102295, + "learning_rate": 2.7756179264826792e-05, + "loss": 0.7817, + "step": 1246 + }, + { + "epoch": 0.9674166020170675, + "grad_norm": 2.6886539459228516, + "learning_rate": 2.78783576665783e-05, + "loss": 0.106, + "step": 1247 + }, + { + "epoch": 0.9681923972071373, + "grad_norm": 1.7198070287704468, + "learning_rate": 2.800006935517661e-05, + "loss": 0.0451, + "step": 1248 + }, + { + "epoch": 0.9689681923972071, + "grad_norm": 6.268177509307861, + "learning_rate": 2.812129459560177e-05, + "loss": 0.25, + "step": 1249 + }, + { + "epoch": 0.969743987587277, + "grad_norm": 16.950916290283203, + "learning_rate": 2.8242013731709283e-05, + "loss": 1.5482, + "step": 1250 + }, + { + "epoch": 0.9705197827773467, + "grad_norm": 4.664473533630371, + "learning_rate": 2.8362207189417272e-05, + "loss": 0.1751, + "step": 1251 + }, + { + "epoch": 0.9712955779674166, + "grad_norm": 6.2507524490356445, + "learning_rate": 2.8481855479880295e-05, + "loss": 0.4074, + "step": 1252 + }, + { + "epoch": 0.9720713731574864, + "grad_norm": 7.526296138763428, + "learning_rate": 2.860093920264944e-05, + "loss": 0.307, + "step": 1253 + }, + { + "epoch": 0.9728471683475562, + "grad_norm": 15.216961860656738, + "learning_rate": 2.8719439048817966e-05, + "loss": 0.9694, + "step": 1254 + }, + { + "epoch": 0.9736229635376261, + "grad_norm": 13.205751419067383, + "learning_rate": 2.8837335804152163e-05, + "loss": 1.0855, + "step": 1255 + }, + { + "epoch": 0.9743987587276959, + "grad_norm": 8.018424987792969, + "learning_rate": 2.8954610352206897e-05, + "loss": 0.36, + "step": 1256 + }, + { + "epoch": 0.9751745539177658, + "grad_norm": 9.845428466796875, + "learning_rate": 2.907124367742523e-05, + "loss": 0.4696, + "step": 1257 + }, + { + "epoch": 0.9759503491078355, + "grad_norm": 9.058723449707031, + "learning_rate": 2.9187216868221698e-05, + "loss": 0.2981, + "step": 1258 + }, + { + "epoch": 0.9767261442979054, + "grad_norm": 0.6596803665161133, + "learning_rate": 2.9302511120048773e-05, + "loss": 0.0108, + "step": 1259 + }, + { + "epoch": 0.9775019394879751, + "grad_norm": 14.962912559509277, + "learning_rate": 2.941710773844591e-05, + "loss": 1.1752, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_Qnli-dev_cosine_accuracy": 0.67578125, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8456684350967407, + "eval_Qnli-dev_cosine_ap": 0.7168685131801189, + "eval_Qnli-dev_cosine_f1": 0.678688524590164, + "eval_Qnli-dev_cosine_f1_threshold": 0.7042285203933716, + "eval_Qnli-dev_cosine_precision": 0.553475935828877, + "eval_Qnli-dev_cosine_recall": 0.8771186440677966, + "eval_Qnli-dev_dot_accuracy": 0.63671875, + "eval_Qnli-dev_dot_accuracy_threshold": 401.1481628417969, + "eval_Qnli-dev_dot_ap": 0.6371081438919197, + "eval_Qnli-dev_dot_f1": 0.684297520661157, + "eval_Qnli-dev_dot_f1_threshold": 335.7393798828125, + "eval_Qnli-dev_dot_precision": 0.5609756097560976, + "eval_Qnli-dev_dot_recall": 0.8771186440677966, + "eval_Qnli-dev_euclidean_accuracy": 0.67578125, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.61432933807373, + "eval_Qnli-dev_euclidean_ap": 0.7139502747319124, + "eval_Qnli-dev_euclidean_f1": 0.6787479406919276, + "eval_Qnli-dev_euclidean_f1_threshold": 16.70758819580078, + "eval_Qnli-dev_euclidean_precision": 0.555256064690027, + "eval_Qnli-dev_euclidean_recall": 0.8728813559322034, + "eval_Qnli-dev_manhattan_accuracy": 0.669921875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 216.93173217773438, + "eval_Qnli-dev_manhattan_ap": 0.7144934262831777, + "eval_Qnli-dev_manhattan_f1": 0.6819672131147542, + "eval_Qnli-dev_manhattan_f1_threshold": 302.77984619140625, + "eval_Qnli-dev_manhattan_precision": 0.5561497326203209, + "eval_Qnli-dev_manhattan_recall": 0.8813559322033898, + "eval_Qnli-dev_max_accuracy": 0.67578125, + "eval_Qnli-dev_max_accuracy_threshold": 401.1481628417969, + "eval_Qnli-dev_max_ap": 0.7168685131801189, + "eval_Qnli-dev_max_f1": 0.684297520661157, + "eval_Qnli-dev_max_f1_threshold": 335.7393798828125, + "eval_Qnli-dev_max_precision": 0.5609756097560976, + "eval_Qnli-dev_max_recall": 0.8813559322033898, + "eval_allNLI-dev_cosine_accuracy": 0.71875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8783560991287231, + "eval_allNLI-dev_cosine_ap": 0.6019568147315763, + "eval_allNLI-dev_cosine_f1": 0.6325167037861915, + "eval_allNLI-dev_cosine_f1_threshold": 0.8004631996154785, + "eval_allNLI-dev_cosine_precision": 0.5144927536231884, + "eval_allNLI-dev_cosine_recall": 0.8208092485549133, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 452.99359130859375, + "eval_allNLI-dev_dot_ap": 0.48922651308546194, + "eval_allNLI-dev_dot_f1": 0.5772357723577236, + "eval_allNLI-dev_dot_f1_threshold": 331.7955017089844, + "eval_allNLI-dev_dot_precision": 0.445141065830721, + "eval_allNLI-dev_dot_recall": 0.8208092485549133, + "eval_allNLI-dev_euclidean_accuracy": 0.7265625, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.467215538024902, + "eval_allNLI-dev_euclidean_ap": 0.6040519098735336, + "eval_allNLI-dev_euclidean_f1": 0.6327944572748269, + "eval_allNLI-dev_euclidean_f1_threshold": 13.209227561950684, + "eval_allNLI-dev_euclidean_precision": 0.5269230769230769, + "eval_allNLI-dev_euclidean_recall": 0.791907514450867, + "eval_allNLI-dev_manhattan_accuracy": 0.720703125, + "eval_allNLI-dev_manhattan_accuracy_threshold": 178.9223175048828, + "eval_allNLI-dev_manhattan_ap": 0.6028436956130316, + "eval_allNLI-dev_manhattan_f1": 0.6370023419203747, + "eval_allNLI-dev_manhattan_f1_threshold": 235.0154571533203, + "eval_allNLI-dev_manhattan_precision": 0.5354330708661418, + "eval_allNLI-dev_manhattan_recall": 0.7861271676300579, + "eval_allNLI-dev_max_accuracy": 0.7265625, + "eval_allNLI-dev_max_accuracy_threshold": 452.99359130859375, + "eval_allNLI-dev_max_ap": 0.6040519098735336, + "eval_allNLI-dev_max_f1": 0.6370023419203747, + "eval_allNLI-dev_max_f1_threshold": 331.7955017089844, + "eval_allNLI-dev_max_precision": 0.5354330708661418, + "eval_allNLI-dev_max_recall": 0.8208092485549133, + "eval_sequential_score": 0.7168685131801189, + "eval_sts-test_pearson_cosine": 0.815847362588151, + "eval_sts-test_pearson_dot": 0.7359966626313633, + "eval_sts-test_pearson_euclidean": 0.8535561946511334, + "eval_sts-test_pearson_manhattan": 0.8516017944909535, + "eval_sts-test_pearson_max": 0.8535561946511334, + "eval_sts-test_spearman_cosine": 0.8601639057015673, + "eval_sts-test_spearman_dot": 0.7255299595618487, + "eval_sts-test_spearman_euclidean": 0.8552219791548535, + "eval_sts-test_spearman_manhattan": 0.8544723060240249, + "eval_sts-test_spearman_max": 0.8601639057015673, + "eval_vitaminc-pairs_loss": 3.6225147247314453, + "eval_vitaminc-pairs_runtime": 5.5033, + "eval_vitaminc-pairs_samples_per_second": 23.259, + "eval_vitaminc-pairs_steps_per_second": 0.363, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_negation-triplets_loss": 1.5155601501464844, + "eval_negation-triplets_runtime": 1.1367, + "eval_negation-triplets_samples_per_second": 112.605, + "eval_negation-triplets_steps_per_second": 1.759, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_scitail-pairs-pos_loss": 0.143746480345726, + "eval_scitail-pairs-pos_runtime": 1.2843, + "eval_scitail-pairs-pos_samples_per_second": 99.663, + "eval_scitail-pairs-pos_steps_per_second": 1.557, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_scitail-pairs-qa_loss": 0.0021120132878422737, + "eval_scitail-pairs-qa_runtime": 0.926, + "eval_scitail-pairs-qa_samples_per_second": 138.234, + "eval_scitail-pairs-qa_steps_per_second": 2.16, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_xsum-pairs_loss": 0.20756767690181732, + "eval_xsum-pairs_runtime": 4.3251, + "eval_xsum-pairs_samples_per_second": 29.595, + "eval_xsum-pairs_steps_per_second": 0.462, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_sciq_pairs_loss": 0.07629822194576263, + "eval_sciq_pairs_runtime": 5.5649, + "eval_sciq_pairs_samples_per_second": 23.001, + "eval_sciq_pairs_steps_per_second": 0.359, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_qasc_pairs_loss": 0.244592547416687, + "eval_qasc_pairs_runtime": 0.9879, + "eval_qasc_pairs_samples_per_second": 129.565, + "eval_qasc_pairs_steps_per_second": 2.024, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_openbookqa_pairs_loss": 1.213144302368164, + "eval_openbookqa_pairs_runtime": 0.8738, + "eval_openbookqa_pairs_samples_per_second": 146.49, + "eval_openbookqa_pairs_steps_per_second": 2.289, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_msmarco_pairs_loss": 0.8372074961662292, + "eval_msmarco_pairs_runtime": 1.9706, + "eval_msmarco_pairs_samples_per_second": 64.954, + "eval_msmarco_pairs_steps_per_second": 1.015, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_nq_pairs_loss": 0.8869773745536804, + "eval_nq_pairs_runtime": 4.4247, + "eval_nq_pairs_samples_per_second": 28.929, + "eval_nq_pairs_steps_per_second": 0.452, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_trivia_pairs_loss": 0.676612377166748, + "eval_trivia_pairs_runtime": 4.5442, + "eval_trivia_pairs_samples_per_second": 28.168, + "eval_trivia_pairs_steps_per_second": 0.44, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_gooaq_pairs_loss": 0.5493606328964233, + "eval_gooaq_pairs_runtime": 1.2823, + "eval_gooaq_pairs_samples_per_second": 99.823, + "eval_gooaq_pairs_steps_per_second": 1.56, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_paws-pos_loss": 0.03376349061727524, + "eval_paws-pos_runtime": 1.0406, + "eval_paws-pos_samples_per_second": 123.004, + "eval_paws-pos_steps_per_second": 1.922, + "step": 1260 + }, + { + "epoch": 0.9775019394879751, + "eval_global_dataset_loss": 0.505013644695282, + "eval_global_dataset_runtime": 21.1732, + "eval_global_dataset_samples_per_second": 19.647, + "eval_global_dataset_steps_per_second": 0.331, + "step": 1260 + }, + { + "epoch": 0.978277734678045, + "grad_norm": 11.987425804138184, + "learning_rate": 2.9530988142070803e-05, + "loss": 0.8216, + "step": 1261 + }, + { + "epoch": 0.9790535298681148, + "grad_norm": 7.394922256469727, + "learning_rate": 2.964413386571222e-05, + "loss": 0.1737, + "step": 1262 + }, + { + "epoch": 0.9798293250581847, + "grad_norm": 8.055792808532715, + "learning_rate": 2.9756526563284106e-05, + "loss": 0.471, + "step": 1263 + }, + { + "epoch": 0.9806051202482544, + "grad_norm": 7.21699333190918, + "learning_rate": 2.9868148010800318e-05, + "loss": 0.4433, + "step": 1264 + }, + { + "epoch": 0.9813809154383243, + "grad_norm": 10.542901039123535, + "learning_rate": 2.9978980109329518e-05, + "loss": 0.7567, + "step": 1265 + }, + { + "epoch": 0.9821567106283942, + "grad_norm": 15.630145072937012, + "learning_rate": 3.0089004887929887e-05, + "loss": 0.9696, + "step": 1266 + }, + { + "epoch": 0.9829325058184639, + "grad_norm": 0.756580650806427, + "learning_rate": 3.0198204506563e-05, + "loss": 0.0401, + "step": 1267 + }, + { + "epoch": 0.9837083010085338, + "grad_norm": 8.775850296020508, + "learning_rate": 3.030656125898654e-05, + "loss": 0.5234, + "step": 1268 + }, + { + "epoch": 0.9844840961986036, + "grad_norm": 10.321736335754395, + "learning_rate": 3.041405757562526e-05, + "loss": 0.6081, + "step": 1269 + }, + { + "epoch": 0.9852598913886734, + "grad_norm": 6.888731479644775, + "learning_rate": 3.052067602641983e-05, + "loss": 0.2459, + "step": 1270 + }, + { + "epoch": 0.9860356865787432, + "grad_norm": 11.0391206741333, + "learning_rate": 3.062639932365305e-05, + "loss": 0.6489, + "step": 1271 + }, + { + "epoch": 0.9868114817688131, + "grad_norm": 0.0, + "learning_rate": 3.073121032475297e-05, + "loss": 0.0, + "step": 1272 + }, + { + "epoch": 0.9875872769588828, + "grad_norm": 19.85426139831543, + "learning_rate": 3.083509203507246e-05, + "loss": 1.4879, + "step": 1273 + }, + { + "epoch": 0.9883630721489527, + "grad_norm": 9.233138084411621, + "learning_rate": 3.093802761064485e-05, + "loss": 0.456, + "step": 1274 + }, + { + "epoch": 0.9891388673390225, + "grad_norm": 7.093190670013428, + "learning_rate": 3.104000036091513e-05, + "loss": 0.3709, + "step": 1275 + }, + { + "epoch": 0.9899146625290923, + "grad_norm": 0.4139319658279419, + "learning_rate": 3.114099375144616e-05, + "loss": 0.0063, + "step": 1276 + }, + { + "epoch": 0.9906904577191621, + "grad_norm": 8.267213821411133, + "learning_rate": 3.124099140659977e-05, + "loss": 0.42, + "step": 1277 + }, + { + "epoch": 0.991466252909232, + "grad_norm": 8.917819023132324, + "learning_rate": 3.13399771121919e-05, + "loss": 0.3612, + "step": 1278 + }, + { + "epoch": 0.9922420480993018, + "grad_norm": 0.03818356245756149, + "learning_rate": 3.143793481812172e-05, + "loss": 0.0003, + "step": 1279 + }, + { + "epoch": 0.9930178432893716, + "grad_norm": 5.075801849365234, + "learning_rate": 3.153484864097406e-05, + "loss": 0.1506, + "step": 1280 + }, + { + "epoch": 0.9937936384794415, + "grad_norm": 12.204947471618652, + "learning_rate": 3.163070286659486e-05, + "loss": 0.9189, + "step": 1281 + }, + { + "epoch": 0.9945694336695112, + "grad_norm": 9.532196044921875, + "learning_rate": 3.172548195263911e-05, + "loss": 0.4371, + "step": 1282 + }, + { + "epoch": 0.9953452288595811, + "grad_norm": 0.4736337661743164, + "learning_rate": 3.181917053109101e-05, + "loss": 0.0081, + "step": 1283 + }, + { + "epoch": 0.9961210240496509, + "grad_norm": 13.530466079711914, + "learning_rate": 3.1911753410755814e-05, + "loss": 1.0255, + "step": 1284 + }, + { + "epoch": 0.9968968192397207, + "grad_norm": 18.26179313659668, + "learning_rate": 3.200321557972301e-05, + "loss": 0.5756, + "step": 1285 + }, + { + "epoch": 0.9976726144297905, + "grad_norm": 9.082464218139648, + "learning_rate": 3.209354220780045e-05, + "loss": 0.3936, + "step": 1286 + }, + { + "epoch": 0.9984484096198604, + "grad_norm": 10.377568244934082, + "learning_rate": 3.218271864891898e-05, + "loss": 0.627, + "step": 1287 + }, + { + "epoch": 0.9992242048099301, + "grad_norm": 8.78341293334961, + "learning_rate": 3.2270730443507276e-05, + "loss": 0.047, + "step": 1288 + }, + { + "epoch": 1.0, + "grad_norm": 14.83316421508789, + "learning_rate": 3.2357563320836337e-05, + "loss": 0.8975, + "step": 1289 + }, + { + "epoch": 1.0007757951900698, + "grad_norm": 10.205459594726562, + "learning_rate": 3.2443203201333494e-05, + "loss": 0.8482, + "step": 1290 + }, + { + "epoch": 1.0015515903801397, + "grad_norm": 6.732105255126953, + "learning_rate": 3.252763619886529e-05, + "loss": 0.2354, + "step": 1291 + }, + { + "epoch": 1.0023273855702095, + "grad_norm": 11.613744735717773, + "learning_rate": 3.2610848622989095e-05, + "loss": 0.8971, + "step": 1292 + }, + { + "epoch": 1.0031031807602793, + "grad_norm": 10.947786331176758, + "learning_rate": 3.2692826981172945e-05, + "loss": 0.9631, + "step": 1293 + }, + { + "epoch": 1.003878975950349, + "grad_norm": 8.22778606414795, + "learning_rate": 3.2773557980983256e-05, + "loss": 0.7931, + "step": 1294 + }, + { + "epoch": 1.004654771140419, + "grad_norm": 6.610589981079102, + "learning_rate": 3.285302853224021e-05, + "loss": 0.3004, + "step": 1295 + }, + { + "epoch": 1.0054305663304888, + "grad_norm": 10.495040893554688, + "learning_rate": 3.293122574914023e-05, + "loss": 0.7473, + "step": 1296 + }, + { + "epoch": 1.0062063615205585, + "grad_norm": 11.122993469238281, + "learning_rate": 3.3008136952345304e-05, + "loss": 1.0953, + "step": 1297 + }, + { + "epoch": 1.0069821567106283, + "grad_norm": 9.837762832641602, + "learning_rate": 3.3083749671039e-05, + "loss": 1.0734, + "step": 1298 + }, + { + "epoch": 1.0077579519006983, + "grad_norm": 11.024539947509766, + "learning_rate": 3.315805164494846e-05, + "loss": 0.8006, + "step": 1299 + }, + { + "epoch": 1.008533747090768, + "grad_norm": 7.763734340667725, + "learning_rate": 3.323103082633237e-05, + "loss": 0.7081, + "step": 1300 + }, + { + "epoch": 1.0093095422808378, + "grad_norm": 1.9376087188720703, + "learning_rate": 3.33026753819345e-05, + "loss": 0.0588, + "step": 1301 + }, + { + "epoch": 1.0100853374709078, + "grad_norm": 1.388636589050293, + "learning_rate": 3.337297369490232e-05, + "loss": 0.0303, + "step": 1302 + }, + { + "epoch": 1.0108611326609775, + "grad_norm": 7.98875617980957, + "learning_rate": 3.344191436667073e-05, + "loss": 0.3468, + "step": 1303 + }, + { + "epoch": 1.0116369278510473, + "grad_norm": 11.81571102142334, + "learning_rate": 3.350948621881021e-05, + "loss": 0.6554, + "step": 1304 + }, + { + "epoch": 1.012412723041117, + "grad_norm": 7.929706573486328, + "learning_rate": 3.3575678294839366e-05, + "loss": 0.2907, + "step": 1305 + }, + { + "epoch": 1.013188518231187, + "grad_norm": 17.511756896972656, + "learning_rate": 3.364047986200151e-05, + "loss": 1.5075, + "step": 1306 + }, + { + "epoch": 1.0139643134212568, + "grad_norm": 15.651792526245117, + "learning_rate": 3.370388041300491e-05, + "loss": 1.059, + "step": 1307 + }, + { + "epoch": 1.0147401086113266, + "grad_norm": 3.5415265560150146, + "learning_rate": 3.3765869667726464e-05, + "loss": 0.0842, + "step": 1308 + }, + { + "epoch": 1.0155159038013963, + "grad_norm": 13.54472541809082, + "learning_rate": 3.3826437574878666e-05, + "loss": 1.0658, + "step": 1309 + }, + { + "epoch": 1.0162916989914663, + "grad_norm": 7.7867631912231445, + "learning_rate": 3.388557431363927e-05, + "loss": 0.2771, + "step": 1310 + }, + { + "epoch": 1.017067494181536, + "grad_norm": 6.247259616851807, + "learning_rate": 3.394327029524378e-05, + "loss": 0.2913, + "step": 1311 + }, + { + "epoch": 1.0178432893716058, + "grad_norm": 11.047322273254395, + "learning_rate": 3.3999516164540205e-05, + "loss": 0.6732, + "step": 1312 + }, + { + "epoch": 1.0186190845616758, + "grad_norm": 17.008304595947266, + "learning_rate": 3.4054302801505905e-05, + "loss": 2.5954, + "step": 1313 + }, + { + "epoch": 1.0193948797517456, + "grad_norm": 1.3579039573669434, + "learning_rate": 3.4107621322726454e-05, + "loss": 0.0352, + "step": 1314 + }, + { + "epoch": 1.0201706749418153, + "grad_norm": 14.844690322875977, + "learning_rate": 3.415946308283595e-05, + "loss": 1.2972, + "step": 1315 + }, + { + "epoch": 1.020946470131885, + "grad_norm": 7.01500129699707, + "learning_rate": 3.4209819675918906e-05, + "loss": 0.3018, + "step": 1316 + }, + { + "epoch": 1.021722265321955, + "grad_norm": 15.82321548461914, + "learning_rate": 3.4258682936873173e-05, + "loss": 2.6832, + "step": 1317 + }, + { + "epoch": 1.0224980605120249, + "grad_norm": 6.357627868652344, + "learning_rate": 3.430604494273392e-05, + "loss": 0.4663, + "step": 1318 + }, + { + "epoch": 1.0232738557020946, + "grad_norm": 7.564973831176758, + "learning_rate": 3.435189801395826e-05, + "loss": 0.402, + "step": 1319 + }, + { + "epoch": 1.0240496508921644, + "grad_norm": 9.514803886413574, + "learning_rate": 3.4396234715670525e-05, + "loss": 0.7501, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_Qnli-dev_cosine_accuracy": 0.693359375, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8683369159698486, + "eval_Qnli-dev_cosine_ap": 0.7435867988980178, + "eval_Qnli-dev_cosine_f1": 0.6852173913043478, + "eval_Qnli-dev_cosine_f1_threshold": 0.7871211171150208, + "eval_Qnli-dev_cosine_precision": 0.5811209439528023, + "eval_Qnli-dev_cosine_recall": 0.8347457627118644, + "eval_Qnli-dev_dot_accuracy": 0.638671875, + "eval_Qnli-dev_dot_accuracy_threshold": 408.35894775390625, + "eval_Qnli-dev_dot_ap": 0.6327868770210536, + "eval_Qnli-dev_dot_f1": 0.6622296173044925, + "eval_Qnli-dev_dot_f1_threshold": 369.78985595703125, + "eval_Qnli-dev_dot_precision": 0.5452054794520548, + "eval_Qnli-dev_dot_recall": 0.8432203389830508, + "eval_Qnli-dev_euclidean_accuracy": 0.68359375, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.23868179321289, + "eval_Qnli-dev_euclidean_ap": 0.7426810586444348, + "eval_Qnli-dev_euclidean_f1": 0.6897810218978102, + "eval_Qnli-dev_euclidean_f1_threshold": 13.970111846923828, + "eval_Qnli-dev_euclidean_precision": 0.6057692307692307, + "eval_Qnli-dev_euclidean_recall": 0.8008474576271186, + "eval_Qnli-dev_manhattan_accuracy": 0.685546875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 215.9767608642578, + "eval_Qnli-dev_manhattan_ap": 0.7382603173541741, + "eval_Qnli-dev_manhattan_f1": 0.6888111888111887, + "eval_Qnli-dev_manhattan_f1_threshold": 261.489501953125, + "eval_Qnli-dev_manhattan_precision": 0.5863095238095238, + "eval_Qnli-dev_manhattan_recall": 0.8347457627118644, + "eval_Qnli-dev_max_accuracy": 0.693359375, + "eval_Qnli-dev_max_accuracy_threshold": 408.35894775390625, + "eval_Qnli-dev_max_ap": 0.7435867988980178, + "eval_Qnli-dev_max_f1": 0.6897810218978102, + "eval_Qnli-dev_max_f1_threshold": 369.78985595703125, + "eval_Qnli-dev_max_precision": 0.6057692307692307, + "eval_Qnli-dev_max_recall": 0.8432203389830508, + "eval_allNLI-dev_cosine_accuracy": 0.71875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9041674137115479, + "eval_allNLI-dev_cosine_ap": 0.5986900766071175, + "eval_allNLI-dev_cosine_f1": 0.6069868995633189, + "eval_allNLI-dev_cosine_f1_threshold": 0.8182723522186279, + "eval_allNLI-dev_cosine_precision": 0.48771929824561405, + "eval_allNLI-dev_cosine_recall": 0.8034682080924855, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 450.2699890136719, + "eval_allNLI-dev_dot_ap": 0.49865615773331173, + "eval_allNLI-dev_dot_f1": 0.5749486652977412, + "eval_allNLI-dev_dot_f1_threshold": 364.66876220703125, + "eval_allNLI-dev_dot_precision": 0.445859872611465, + "eval_allNLI-dev_dot_recall": 0.8092485549132948, + "eval_allNLI-dev_euclidean_accuracy": 0.720703125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.100296020507812, + "eval_allNLI-dev_euclidean_ap": 0.5979238701971141, + "eval_allNLI-dev_euclidean_f1": 0.6092436974789915, + "eval_allNLI-dev_euclidean_f1_threshold": 13.368457794189453, + "eval_allNLI-dev_euclidean_precision": 0.47854785478547857, + "eval_allNLI-dev_euclidean_recall": 0.838150289017341, + "eval_allNLI-dev_manhattan_accuracy": 0.7265625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 182.22265625, + "eval_allNLI-dev_manhattan_ap": 0.5990786558063288, + "eval_allNLI-dev_manhattan_f1": 0.6229508196721311, + "eval_allNLI-dev_manhattan_f1_threshold": 223.14529418945312, + "eval_allNLI-dev_manhattan_precision": 0.5236220472440944, + "eval_allNLI-dev_manhattan_recall": 0.7687861271676301, + "eval_allNLI-dev_max_accuracy": 0.7265625, + "eval_allNLI-dev_max_accuracy_threshold": 450.2699890136719, + "eval_allNLI-dev_max_ap": 0.5990786558063288, + "eval_allNLI-dev_max_f1": 0.6229508196721311, + "eval_allNLI-dev_max_f1_threshold": 364.66876220703125, + "eval_allNLI-dev_max_precision": 0.5236220472440944, + "eval_allNLI-dev_max_recall": 0.838150289017341, + "eval_sequential_score": 0.7435867988980178, + "eval_sts-test_pearson_cosine": 0.8352857863696783, + "eval_sts-test_pearson_dot": 0.7708644479725002, + "eval_sts-test_pearson_euclidean": 0.8741684830174303, + "eval_sts-test_pearson_manhattan": 0.8702535948291277, + "eval_sts-test_pearson_max": 0.8741684830174303, + "eval_sts-test_spearman_cosine": 0.8800061674898998, + "eval_sts-test_spearman_dot": 0.76218365920952, + "eval_sts-test_spearman_euclidean": 0.875437197955622, + "eval_sts-test_spearman_manhattan": 0.8724729191523172, + "eval_sts-test_spearman_max": 0.8800061674898998, + "eval_vitaminc-pairs_loss": 3.6631999015808105, + "eval_vitaminc-pairs_runtime": 5.4926, + "eval_vitaminc-pairs_samples_per_second": 23.304, + "eval_vitaminc-pairs_steps_per_second": 0.364, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_negation-triplets_loss": 1.4271920919418335, + "eval_negation-triplets_runtime": 1.1185, + "eval_negation-triplets_samples_per_second": 114.435, + "eval_negation-triplets_steps_per_second": 1.788, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_scitail-pairs-pos_loss": 0.19838891923427582, + "eval_scitail-pairs-pos_runtime": 1.2663, + "eval_scitail-pairs-pos_samples_per_second": 101.085, + "eval_scitail-pairs-pos_steps_per_second": 1.579, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_scitail-pairs-qa_loss": 0.008117816410958767, + "eval_scitail-pairs-qa_runtime": 0.9161, + "eval_scitail-pairs-qa_samples_per_second": 139.717, + "eval_scitail-pairs-qa_steps_per_second": 2.183, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_xsum-pairs_loss": 0.25715702772140503, + "eval_xsum-pairs_runtime": 4.322, + "eval_xsum-pairs_samples_per_second": 29.616, + "eval_xsum-pairs_steps_per_second": 0.463, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_sciq_pairs_loss": 0.08018835633993149, + "eval_sciq_pairs_runtime": 5.5936, + "eval_sciq_pairs_samples_per_second": 22.883, + "eval_sciq_pairs_steps_per_second": 0.358, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_qasc_pairs_loss": 0.32260918617248535, + "eval_qasc_pairs_runtime": 0.9919, + "eval_qasc_pairs_samples_per_second": 129.04, + "eval_qasc_pairs_steps_per_second": 2.016, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_openbookqa_pairs_loss": 1.1771844625473022, + "eval_openbookqa_pairs_runtime": 0.8708, + "eval_openbookqa_pairs_samples_per_second": 146.989, + "eval_openbookqa_pairs_steps_per_second": 2.297, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_msmarco_pairs_loss": 0.7681916952133179, + "eval_msmarco_pairs_runtime": 1.9738, + "eval_msmarco_pairs_samples_per_second": 64.851, + "eval_msmarco_pairs_steps_per_second": 1.013, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_nq_pairs_loss": 0.8776646256446838, + "eval_nq_pairs_runtime": 4.4135, + "eval_nq_pairs_samples_per_second": 29.002, + "eval_nq_pairs_steps_per_second": 0.453, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_trivia_pairs_loss": 0.6701335310935974, + "eval_trivia_pairs_runtime": 4.5376, + "eval_trivia_pairs_samples_per_second": 28.209, + "eval_trivia_pairs_steps_per_second": 0.441, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_gooaq_pairs_loss": 0.61519455909729, + "eval_gooaq_pairs_runtime": 1.2839, + "eval_gooaq_pairs_samples_per_second": 99.693, + "eval_gooaq_pairs_steps_per_second": 1.558, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_paws-pos_loss": 0.03438161313533783, + "eval_paws-pos_runtime": 1.0421, + "eval_paws-pos_samples_per_second": 122.834, + "eval_paws-pos_steps_per_second": 1.919, + "step": 1320 + }, + { + "epoch": 1.0240496508921644, + "eval_global_dataset_loss": 0.5264055132865906, + "eval_global_dataset_runtime": 21.1711, + "eval_global_dataset_samples_per_second": 19.649, + "eval_global_dataset_steps_per_second": 0.331, + "step": 1320 + }, + { + "epoch": 1.0248254460822344, + "grad_norm": 1.4392180442810059, + "learning_rate": 3.4439047858867694e-05, + "loss": 0.048, + "step": 1321 + }, + { + "epoch": 1.0256012412723041, + "grad_norm": 6.716087818145752, + "learning_rate": 3.448033050158519e-05, + "loss": 0.3494, + "step": 1322 + }, + { + "epoch": 1.0263770364623739, + "grad_norm": 5.739917755126953, + "learning_rate": 3.452007595002236e-05, + "loss": 0.2134, + "step": 1323 + }, + { + "epoch": 1.0271528316524439, + "grad_norm": 7.634201526641846, + "learning_rate": 3.4558277759627944e-05, + "loss": 0.3184, + "step": 1324 + }, + { + "epoch": 1.0279286268425136, + "grad_norm": 7.894606113433838, + "learning_rate": 3.4594929736144966e-05, + "loss": 0.5951, + "step": 1325 + }, + { + "epoch": 1.0287044220325834, + "grad_norm": 15.105119705200195, + "learning_rate": 3.4630025936615166e-05, + "loss": 2.4519, + "step": 1326 + }, + { + "epoch": 1.0294802172226531, + "grad_norm": 7.895918369293213, + "learning_rate": 3.466356067034255e-05, + "loss": 0.5311, + "step": 1327 + }, + { + "epoch": 1.0302560124127231, + "grad_norm": 6.175323963165283, + "learning_rate": 3.469552849981618e-05, + "loss": 0.2615, + "step": 1328 + }, + { + "epoch": 1.031031807602793, + "grad_norm": 8.19454574584961, + "learning_rate": 3.4725924241591806e-05, + "loss": 0.6763, + "step": 1329 + }, + { + "epoch": 1.0318076027928627, + "grad_norm": 6.807870388031006, + "learning_rate": 3.475474296713235e-05, + "loss": 0.3215, + "step": 1330 + }, + { + "epoch": 1.0325833979829324, + "grad_norm": 6.442826747894287, + "learning_rate": 3.478198000360704e-05, + "loss": 0.2737, + "step": 1331 + }, + { + "epoch": 1.0333591931730024, + "grad_norm": 14.82455825805664, + "learning_rate": 3.48076309346491e-05, + "loss": 2.0641, + "step": 1332 + }, + { + "epoch": 1.0341349883630722, + "grad_norm": 7.081644535064697, + "learning_rate": 3.483169160107183e-05, + "loss": 0.2694, + "step": 1333 + }, + { + "epoch": 1.034910783553142, + "grad_norm": 6.481761932373047, + "learning_rate": 3.4854158101543015e-05, + "loss": 0.277, + "step": 1334 + }, + { + "epoch": 1.0356865787432117, + "grad_norm": 5.890316963195801, + "learning_rate": 3.48750267932175e-05, + "loss": 0.2947, + "step": 1335 + }, + { + "epoch": 1.0364623739332817, + "grad_norm": 4.919997692108154, + "learning_rate": 3.4894294292327874e-05, + "loss": 0.1652, + "step": 1336 + }, + { + "epoch": 1.0372381691233514, + "grad_norm": 7.105747699737549, + "learning_rate": 3.491195747473311e-05, + "loss": 0.4205, + "step": 1337 + }, + { + "epoch": 1.0380139643134212, + "grad_norm": 1.7777588367462158, + "learning_rate": 3.4928013476425144e-05, + "loss": 0.0427, + "step": 1338 + }, + { + "epoch": 1.0387897595034912, + "grad_norm": 7.158056259155273, + "learning_rate": 3.494245969399327e-05, + "loss": 0.3293, + "step": 1339 + }, + { + "epoch": 1.039565554693561, + "grad_norm": 7.921936511993408, + "learning_rate": 3.495529378504626e-05, + "loss": 0.3833, + "step": 1340 + }, + { + "epoch": 1.0403413498836307, + "grad_norm": 12.757912635803223, + "learning_rate": 3.496651366859217e-05, + "loss": 1.1121, + "step": 1341 + }, + { + "epoch": 1.0411171450737005, + "grad_norm": 9.888944625854492, + "learning_rate": 3.4976117525375776e-05, + "loss": 0.8354, + "step": 1342 + }, + { + "epoch": 1.0418929402637704, + "grad_norm": 13.719603538513184, + "learning_rate": 3.498410379817355e-05, + "loss": 2.2634, + "step": 1343 + }, + { + "epoch": 1.0426687354538402, + "grad_norm": 14.076007843017578, + "learning_rate": 3.499047119204615e-05, + "loss": 2.114, + "step": 1344 + }, + { + "epoch": 1.04344453064391, + "grad_norm": 11.876978874206543, + "learning_rate": 3.499521867454842e-05, + "loss": 1.175, + "step": 1345 + }, + { + "epoch": 1.0442203258339797, + "grad_norm": 4.729123592376709, + "learning_rate": 3.499834547589675e-05, + "loss": 0.1456, + "step": 1346 + }, + { + "epoch": 1.0449961210240497, + "grad_norm": 9.542031288146973, + "learning_rate": 3.499985108909392e-05, + "loss": 0.9623, + "step": 1347 + }, + { + "epoch": 1.0457719162141195, + "grad_norm": 9.251585960388184, + "learning_rate": 3.499973527001133e-05, + "loss": 0.5537, + "step": 1348 + }, + { + "epoch": 1.0465477114041892, + "grad_norm": 5.426143169403076, + "learning_rate": 3.499799803742852e-05, + "loss": 0.3204, + "step": 1349 + }, + { + "epoch": 1.0473235065942592, + "grad_norm": 0.9218572378158569, + "learning_rate": 3.49946396730302e-05, + "loss": 0.0416, + "step": 1350 + }, + { + "epoch": 1.048099301784329, + "grad_norm": 9.420156478881836, + "learning_rate": 3.498966072136051e-05, + "loss": 0.8396, + "step": 1351 + }, + { + "epoch": 1.0488750969743987, + "grad_norm": 11.664908409118652, + "learning_rate": 3.498306198973477e-05, + "loss": 1.2291, + "step": 1352 + }, + { + "epoch": 1.0496508921644685, + "grad_norm": 7.216692924499512, + "learning_rate": 3.4974844548108565e-05, + "loss": 0.4293, + "step": 1353 + }, + { + "epoch": 1.0504266873545385, + "grad_norm": 1.0699224472045898, + "learning_rate": 3.496500972890424e-05, + "loss": 0.033, + "step": 1354 + }, + { + "epoch": 1.0512024825446082, + "grad_norm": 7.099836826324463, + "learning_rate": 3.495355912679488e-05, + "loss": 0.3225, + "step": 1355 + }, + { + "epoch": 1.051978277734678, + "grad_norm": 9.31737232208252, + "learning_rate": 3.494049459844572e-05, + "loss": 0.6574, + "step": 1356 + }, + { + "epoch": 1.0527540729247478, + "grad_norm": 10.510056495666504, + "learning_rate": 3.49258182622131e-05, + "loss": 0.8099, + "step": 1357 + }, + { + "epoch": 1.0535298681148177, + "grad_norm": 1.0994374752044678, + "learning_rate": 3.490953249780099e-05, + "loss": 0.0201, + "step": 1358 + }, + { + "epoch": 1.0543056633048875, + "grad_norm": 13.770675659179688, + "learning_rate": 3.4891639945875116e-05, + "loss": 0.9445, + "step": 1359 + }, + { + "epoch": 1.0550814584949573, + "grad_norm": 7.7662248611450195, + "learning_rate": 3.4872143507634826e-05, + "loss": 0.3277, + "step": 1360 + }, + { + "epoch": 1.0558572536850273, + "grad_norm": 7.331538200378418, + "learning_rate": 3.485104634434261e-05, + "loss": 0.417, + "step": 1361 + }, + { + "epoch": 1.056633048875097, + "grad_norm": 7.422630310058594, + "learning_rate": 3.482835187681157e-05, + "loss": 0.3143, + "step": 1362 + }, + { + "epoch": 1.0574088440651668, + "grad_norm": 14.174910545349121, + "learning_rate": 3.4804063784850713e-05, + "loss": 1.0894, + "step": 1363 + }, + { + "epoch": 1.0581846392552365, + "grad_norm": 11.317131042480469, + "learning_rate": 3.477818600666833e-05, + "loss": 0.5934, + "step": 1364 + }, + { + "epoch": 1.0589604344453065, + "grad_norm": 9.18185806274414, + "learning_rate": 3.475072273823339e-05, + "loss": 0.5992, + "step": 1365 + }, + { + "epoch": 1.0597362296353763, + "grad_norm": 11.465063095092773, + "learning_rate": 3.472167843259519e-05, + "loss": 0.6675, + "step": 1366 + }, + { + "epoch": 1.060512024825446, + "grad_norm": 12.469922065734863, + "learning_rate": 3.4691057799161335e-05, + "loss": 1.0068, + "step": 1367 + }, + { + "epoch": 1.0612878200155158, + "grad_norm": 12.495719909667969, + "learning_rate": 3.465886580293411e-05, + "loss": 0.9513, + "step": 1368 + }, + { + "epoch": 1.0620636152055858, + "grad_norm": 7.234547138214111, + "learning_rate": 3.462510766370542e-05, + "loss": 0.2678, + "step": 1369 + }, + { + "epoch": 1.0628394103956555, + "grad_norm": 6.584646224975586, + "learning_rate": 3.458978885521043e-05, + "loss": 0.3386, + "step": 1370 + }, + { + "epoch": 1.0636152055857253, + "grad_norm": 11.376327514648438, + "learning_rate": 3.455291510424006e-05, + "loss": 0.9078, + "step": 1371 + }, + { + "epoch": 1.0643910007757953, + "grad_norm": 6.8868818283081055, + "learning_rate": 3.451449238971234e-05, + "loss": 0.3088, + "step": 1372 + }, + { + "epoch": 1.065166795965865, + "grad_norm": 15.791929244995117, + "learning_rate": 3.4474526941703e-05, + "loss": 2.1878, + "step": 1373 + }, + { + "epoch": 1.0659425911559348, + "grad_norm": 0.942842960357666, + "learning_rate": 3.44330252404353e-05, + "loss": 0.0486, + "step": 1374 + }, + { + "epoch": 1.0667183863460046, + "grad_norm": 7.561138153076172, + "learning_rate": 3.438999401522924e-05, + "loss": 0.3632, + "step": 1375 + }, + { + "epoch": 1.0674941815360746, + "grad_norm": 16.42315101623535, + "learning_rate": 3.434544024341048e-05, + "loss": 2.4621, + "step": 1376 + }, + { + "epoch": 1.0682699767261443, + "grad_norm": 0.918360710144043, + "learning_rate": 3.429937114917899e-05, + "loss": 0.0173, + "step": 1377 + }, + { + "epoch": 1.069045771916214, + "grad_norm": 11.115267753601074, + "learning_rate": 3.4251794202437635e-05, + "loss": 0.879, + "step": 1378 + }, + { + "epoch": 1.0698215671062838, + "grad_norm": 12.189352035522461, + "learning_rate": 3.420271711758103e-05, + "loss": 1.0652, + "step": 1379 + }, + { + "epoch": 1.0705973622963538, + "grad_norm": 8.32873249053955, + "learning_rate": 3.4152147852244636e-05, + "loss": 0.5605, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_Qnli-dev_cosine_accuracy": 0.67578125, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.838126003742218, + "eval_Qnli-dev_cosine_ap": 0.7349896668445257, + "eval_Qnli-dev_cosine_f1": 0.6823161189358372, + "eval_Qnli-dev_cosine_f1_threshold": 0.72300785779953, + "eval_Qnli-dev_cosine_precision": 0.5409429280397022, + "eval_Qnli-dev_cosine_recall": 0.923728813559322, + "eval_Qnli-dev_dot_accuracy": 0.626953125, + "eval_Qnli-dev_dot_accuracy_threshold": 400.089599609375, + "eval_Qnli-dev_dot_ap": 0.6266396315691469, + "eval_Qnli-dev_dot_f1": 0.6778625954198474, + "eval_Qnli-dev_dot_f1_threshold": 336.829833984375, + "eval_Qnli-dev_dot_precision": 0.5298329355608592, + "eval_Qnli-dev_dot_recall": 0.940677966101695, + "eval_Qnli-dev_euclidean_accuracy": 0.68359375, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.213363647460938, + "eval_Qnli-dev_euclidean_ap": 0.7364944213500426, + "eval_Qnli-dev_euclidean_f1": 0.6825396825396826, + "eval_Qnli-dev_euclidean_f1_threshold": 16.218490600585938, + "eval_Qnli-dev_euclidean_precision": 0.5456852791878173, + "eval_Qnli-dev_euclidean_recall": 0.9110169491525424, + "eval_Qnli-dev_manhattan_accuracy": 0.677734375, + "eval_Qnli-dev_manhattan_accuracy_threshold": 238.61953735351562, + "eval_Qnli-dev_manhattan_ap": 0.7351345160672922, + "eval_Qnli-dev_manhattan_f1": 0.68630849220104, + "eval_Qnli-dev_manhattan_f1_threshold": 274.7607421875, + "eval_Qnli-dev_manhattan_precision": 0.5806451612903226, + "eval_Qnli-dev_manhattan_recall": 0.8389830508474576, + "eval_Qnli-dev_max_accuracy": 0.68359375, + "eval_Qnli-dev_max_accuracy_threshold": 400.089599609375, + "eval_Qnli-dev_max_ap": 0.7364944213500426, + "eval_Qnli-dev_max_f1": 0.68630849220104, + "eval_Qnli-dev_max_f1_threshold": 336.829833984375, + "eval_Qnli-dev_max_precision": 0.5806451612903226, + "eval_Qnli-dev_max_recall": 0.940677966101695, + "eval_allNLI-dev_cosine_accuracy": 0.712890625, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8793672919273376, + "eval_allNLI-dev_cosine_ap": 0.594403484741277, + "eval_allNLI-dev_cosine_f1": 0.6244131455399061, + "eval_allNLI-dev_cosine_f1_threshold": 0.8185057640075684, + "eval_allNLI-dev_cosine_precision": 0.525691699604743, + "eval_allNLI-dev_cosine_recall": 0.7687861271676301, + "eval_allNLI-dev_dot_accuracy": 0.689453125, + "eval_allNLI-dev_dot_accuracy_threshold": 407.29168701171875, + "eval_allNLI-dev_dot_ap": 0.5168133509769519, + "eval_allNLI-dev_dot_f1": 0.5785123966942148, + "eval_allNLI-dev_dot_f1_threshold": 356.5469665527344, + "eval_allNLI-dev_dot_precision": 0.45016077170418006, + "eval_allNLI-dev_dot_recall": 0.8092485549132948, + "eval_allNLI-dev_euclidean_accuracy": 0.712890625, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.407035827636719, + "eval_allNLI-dev_euclidean_ap": 0.596297765521977, + "eval_allNLI-dev_euclidean_f1": 0.6238532110091742, + "eval_allNLI-dev_euclidean_f1_threshold": 13.30148983001709, + "eval_allNLI-dev_euclidean_precision": 0.5171102661596958, + "eval_allNLI-dev_euclidean_recall": 0.7861271676300579, + "eval_allNLI-dev_manhattan_accuracy": 0.720703125, + "eval_allNLI-dev_manhattan_accuracy_threshold": 189.25787353515625, + "eval_allNLI-dev_manhattan_ap": 0.5958479326536642, + "eval_allNLI-dev_manhattan_f1": 0.631578947368421, + "eval_allNLI-dev_manhattan_f1_threshold": 233.5772705078125, + "eval_allNLI-dev_manhattan_precision": 0.5387755102040817, + "eval_allNLI-dev_manhattan_recall": 0.7630057803468208, + "eval_allNLI-dev_max_accuracy": 0.720703125, + "eval_allNLI-dev_max_accuracy_threshold": 407.29168701171875, + "eval_allNLI-dev_max_ap": 0.596297765521977, + "eval_allNLI-dev_max_f1": 0.631578947368421, + "eval_allNLI-dev_max_f1_threshold": 356.5469665527344, + "eval_allNLI-dev_max_precision": 0.5387755102040817, + "eval_allNLI-dev_max_recall": 0.8092485549132948, + "eval_sequential_score": 0.7364944213500426, + "eval_sts-test_pearson_cosine": 0.8265920941552377, + "eval_sts-test_pearson_dot": 0.7535334340278752, + "eval_sts-test_pearson_euclidean": 0.8604460047066497, + "eval_sts-test_pearson_manhattan": 0.8547994587579331, + "eval_sts-test_pearson_max": 0.8604460047066497, + "eval_sts-test_spearman_cosine": 0.8623824047337846, + "eval_sts-test_spearman_dot": 0.7437214368199755, + "eval_sts-test_spearman_euclidean": 0.8595611957131479, + "eval_sts-test_spearman_manhattan": 0.8551257639428621, + "eval_sts-test_spearman_max": 0.8623824047337846, + "eval_vitaminc-pairs_loss": 3.3086421489715576, + "eval_vitaminc-pairs_runtime": 5.4853, + "eval_vitaminc-pairs_samples_per_second": 23.335, + "eval_vitaminc-pairs_steps_per_second": 0.365, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_negation-triplets_loss": 1.4226248264312744, + "eval_negation-triplets_runtime": 1.1117, + "eval_negation-triplets_samples_per_second": 115.138, + "eval_negation-triplets_steps_per_second": 1.799, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_scitail-pairs-pos_loss": 0.18248817324638367, + "eval_scitail-pairs-pos_runtime": 1.2825, + "eval_scitail-pairs-pos_samples_per_second": 99.803, + "eval_scitail-pairs-pos_steps_per_second": 1.559, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_scitail-pairs-qa_loss": 0.003936237655580044, + "eval_scitail-pairs-qa_runtime": 0.9135, + "eval_scitail-pairs-qa_samples_per_second": 140.121, + "eval_scitail-pairs-qa_steps_per_second": 2.189, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_xsum-pairs_loss": 0.1763661801815033, + "eval_xsum-pairs_runtime": 4.325, + "eval_xsum-pairs_samples_per_second": 29.595, + "eval_xsum-pairs_steps_per_second": 0.462, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_sciq_pairs_loss": 0.07912196964025497, + "eval_sciq_pairs_runtime": 5.5746, + "eval_sciq_pairs_samples_per_second": 22.961, + "eval_sciq_pairs_steps_per_second": 0.359, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_qasc_pairs_loss": 0.2345551699399948, + "eval_qasc_pairs_runtime": 0.9795, + "eval_qasc_pairs_samples_per_second": 130.682, + "eval_qasc_pairs_steps_per_second": 2.042, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_openbookqa_pairs_loss": 1.0948086977005005, + "eval_openbookqa_pairs_runtime": 0.8606, + "eval_openbookqa_pairs_samples_per_second": 148.726, + "eval_openbookqa_pairs_steps_per_second": 2.324, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_msmarco_pairs_loss": 0.6889941692352295, + "eval_msmarco_pairs_runtime": 1.9633, + "eval_msmarco_pairs_samples_per_second": 65.197, + "eval_msmarco_pairs_steps_per_second": 1.019, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_nq_pairs_loss": 0.7974749803543091, + "eval_nq_pairs_runtime": 4.4215, + "eval_nq_pairs_samples_per_second": 28.949, + "eval_nq_pairs_steps_per_second": 0.452, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_trivia_pairs_loss": 0.739738941192627, + "eval_trivia_pairs_runtime": 4.527, + "eval_trivia_pairs_samples_per_second": 28.275, + "eval_trivia_pairs_steps_per_second": 0.442, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_gooaq_pairs_loss": 0.5885088443756104, + "eval_gooaq_pairs_runtime": 1.2701, + "eval_gooaq_pairs_samples_per_second": 100.782, + "eval_gooaq_pairs_steps_per_second": 1.575, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_paws-pos_loss": 0.03457222506403923, + "eval_paws-pos_runtime": 1.0319, + "eval_paws-pos_samples_per_second": 124.043, + "eval_paws-pos_steps_per_second": 1.938, + "step": 1380 + }, + { + "epoch": 1.0705973622963538, + "eval_global_dataset_loss": 0.45676878094673157, + "eval_global_dataset_runtime": 21.1079, + "eval_global_dataset_samples_per_second": 19.708, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1380 + }, + { + "epoch": 1.0713731574864236, + "grad_norm": 7.572328090667725, + "learning_rate": 3.410009460601445e-05, + "loss": 0.6798, + "step": 1381 + }, + { + "epoch": 1.0721489526764933, + "grad_norm": 9.507984161376953, + "learning_rate": 3.404656581909754e-05, + "loss": 0.587, + "step": 1382 + }, + { + "epoch": 1.0729247478665633, + "grad_norm": 10.871009826660156, + "learning_rate": 3.3991570170953454e-05, + "loss": 0.728, + "step": 1383 + }, + { + "epoch": 1.073700543056633, + "grad_norm": 10.32335090637207, + "learning_rate": 3.3935116578886884e-05, + "loss": 0.9275, + "step": 1384 + }, + { + "epoch": 1.0744763382467029, + "grad_norm": 7.675612926483154, + "learning_rate": 3.38772141966018e-05, + "loss": 0.41, + "step": 1385 + }, + { + "epoch": 1.0752521334367726, + "grad_norm": 6.694522857666016, + "learning_rate": 3.381787241271717e-05, + "loss": 0.4197, + "step": 1386 + }, + { + "epoch": 1.0760279286268426, + "grad_norm": 15.044102668762207, + "learning_rate": 3.375710084924469e-05, + "loss": 1.1541, + "step": 1387 + }, + { + "epoch": 1.0768037238169124, + "grad_norm": 10.22745418548584, + "learning_rate": 3.3694909360028535e-05, + "loss": 0.7378, + "step": 1388 + }, + { + "epoch": 1.0775795190069821, + "grad_norm": 12.018893241882324, + "learning_rate": 3.36313080291477e-05, + "loss": 0.7358, + "step": 1389 + }, + { + "epoch": 1.0783553141970519, + "grad_norm": 7.781107425689697, + "learning_rate": 3.356630716928084e-05, + "loss": 0.3611, + "step": 1390 + }, + { + "epoch": 1.0791311093871219, + "grad_norm": 6.749479293823242, + "learning_rate": 3.349991732003414e-05, + "loss": 0.1857, + "step": 1391 + }, + { + "epoch": 1.0799069045771916, + "grad_norm": 6.715334415435791, + "learning_rate": 3.343214924623237e-05, + "loss": 0.2911, + "step": 1392 + }, + { + "epoch": 1.0806826997672614, + "grad_norm": 7.795923709869385, + "learning_rate": 3.336301393617339e-05, + "loss": 0.3216, + "step": 1393 + }, + { + "epoch": 1.0814584949573312, + "grad_norm": 8.67205810546875, + "learning_rate": 3.329252259984647e-05, + "loss": 0.6399, + "step": 1394 + }, + { + "epoch": 1.0822342901474011, + "grad_norm": 6.380029678344727, + "learning_rate": 3.322068666711463e-05, + "loss": 0.2129, + "step": 1395 + }, + { + "epoch": 1.083010085337471, + "grad_norm": 1.8791251182556152, + "learning_rate": 3.3147517785861346e-05, + "loss": 0.069, + "step": 1396 + }, + { + "epoch": 1.0837858805275407, + "grad_norm": 10.183939933776855, + "learning_rate": 3.307302782010186e-05, + "loss": 0.7572, + "step": 1397 + }, + { + "epoch": 1.0845616757176106, + "grad_norm": 9.175521850585938, + "learning_rate": 3.299722884805955e-05, + "loss": 0.7628, + "step": 1398 + }, + { + "epoch": 1.0853374709076804, + "grad_norm": 4.6771674156188965, + "learning_rate": 3.292013316020744e-05, + "loss": 0.1648, + "step": 1399 + }, + { + "epoch": 1.0861132660977502, + "grad_norm": 9.588823318481445, + "learning_rate": 3.284175325727535e-05, + "loss": 0.7332, + "step": 1400 + }, + { + "epoch": 1.08688906128782, + "grad_norm": 4.86440896987915, + "learning_rate": 3.2762101848223014e-05, + "loss": 0.2175, + "step": 1401 + }, + { + "epoch": 1.08766485647789, + "grad_norm": 8.658729553222656, + "learning_rate": 3.268119184817932e-05, + "loss": 0.7561, + "step": 1402 + }, + { + "epoch": 1.0884406516679597, + "grad_norm": 4.489306449890137, + "learning_rate": 3.259903637634822e-05, + "loss": 0.1919, + "step": 1403 + }, + { + "epoch": 1.0892164468580294, + "grad_norm": 12.452356338500977, + "learning_rate": 3.251564875388149e-05, + "loss": 1.1661, + "step": 1404 + }, + { + "epoch": 1.0899922420480994, + "grad_norm": 5.468027591705322, + "learning_rate": 3.2431042501718726e-05, + "loss": 0.2429, + "step": 1405 + }, + { + "epoch": 1.0907680372381692, + "grad_norm": 7.947955131530762, + "learning_rate": 3.2345231338395065e-05, + "loss": 0.427, + "step": 1406 + }, + { + "epoch": 1.091543832428239, + "grad_norm": 1.777831792831421, + "learning_rate": 3.225822917781671e-05, + "loss": 0.0602, + "step": 1407 + }, + { + "epoch": 1.0923196276183087, + "grad_norm": 9.821995735168457, + "learning_rate": 3.217005012700488e-05, + "loss": 0.8458, + "step": 1408 + }, + { + "epoch": 1.0930954228083787, + "grad_norm": 7.485621929168701, + "learning_rate": 3.208070848380843e-05, + "loss": 0.3433, + "step": 1409 + }, + { + "epoch": 1.0938712179984484, + "grad_norm": 6.117361068725586, + "learning_rate": 3.199021873458551e-05, + "loss": 0.2088, + "step": 1410 + }, + { + "epoch": 1.0946470131885182, + "grad_norm": 1.5002250671386719, + "learning_rate": 3.189859555185463e-05, + "loss": 0.0273, + "step": 1411 + }, + { + "epoch": 1.095422808378588, + "grad_norm": 6.309535026550293, + "learning_rate": 3.180585379191563e-05, + "loss": 0.3219, + "step": 1412 + }, + { + "epoch": 1.096198603568658, + "grad_norm": 8.092740058898926, + "learning_rate": 3.171200849244076e-05, + "loss": 0.3063, + "step": 1413 + }, + { + "epoch": 1.0969743987587277, + "grad_norm": 12.487458229064941, + "learning_rate": 3.1617074870036405e-05, + "loss": 0.9594, + "step": 1414 + }, + { + "epoch": 1.0977501939487975, + "grad_norm": 9.055601119995117, + "learning_rate": 3.152106831777575e-05, + "loss": 0.5197, + "step": 1415 + }, + { + "epoch": 1.0985259891388672, + "grad_norm": 0.17894873023033142, + "learning_rate": 3.142400440270291e-05, + "loss": 0.0035, + "step": 1416 + }, + { + "epoch": 1.0993017843289372, + "grad_norm": 14.980459213256836, + "learning_rate": 3.132589886330876e-05, + "loss": 1.2378, + "step": 1417 + }, + { + "epoch": 1.100077579519007, + "grad_norm": 14.346709251403809, + "learning_rate": 3.122676760697902e-05, + "loss": 1.2984, + "step": 1418 + }, + { + "epoch": 1.1008533747090767, + "grad_norm": 11.853586196899414, + "learning_rate": 3.112662670741491e-05, + "loss": 0.7661, + "step": 1419 + }, + { + "epoch": 1.1016291698991467, + "grad_norm": 8.279623031616211, + "learning_rate": 3.1025492402026964e-05, + "loss": 0.7097, + "step": 1420 + }, + { + "epoch": 1.1024049650892165, + "grad_norm": 5.802407741546631, + "learning_rate": 3.0923381089302105e-05, + "loss": 0.3042, + "step": 1421 + }, + { + "epoch": 1.1031807602792862, + "grad_norm": 2.0792477130889893, + "learning_rate": 3.0820309326144715e-05, + "loss": 0.0307, + "step": 1422 + }, + { + "epoch": 1.103956555469356, + "grad_norm": 0.25669577717781067, + "learning_rate": 3.071629382519209e-05, + "loss": 0.0041, + "step": 1423 + }, + { + "epoch": 1.104732350659426, + "grad_norm": 6.067919731140137, + "learning_rate": 3.0611351452104434e-05, + "loss": 0.1738, + "step": 1424 + }, + { + "epoch": 1.1055081458494957, + "grad_norm": 7.253147602081299, + "learning_rate": 3.0505499222830284e-05, + "loss": 0.2647, + "step": 1425 + }, + { + "epoch": 1.1062839410395655, + "grad_norm": 8.239336013793945, + "learning_rate": 3.039875430084734e-05, + "loss": 0.4175, + "step": 1426 + }, + { + "epoch": 1.1070597362296355, + "grad_norm": 7.247814655303955, + "learning_rate": 3.0291133994379576e-05, + "loss": 0.2879, + "step": 1427 + }, + { + "epoch": 1.1078355314197053, + "grad_norm": 7.090550422668457, + "learning_rate": 3.0182655753590724e-05, + "loss": 0.3564, + "step": 1428 + }, + { + "epoch": 1.108611326609775, + "grad_norm": 11.922552108764648, + "learning_rate": 3.007333716775484e-05, + "loss": 0.9345, + "step": 1429 + }, + { + "epoch": 1.1093871217998448, + "grad_norm": 11.5636625289917, + "learning_rate": 2.9963195962404247e-05, + "loss": 0.99, + "step": 1430 + }, + { + "epoch": 1.1101629169899148, + "grad_norm": 5.662031650543213, + "learning_rate": 2.9852249996455474e-05, + "loss": 0.2504, + "step": 1431 + }, + { + "epoch": 1.1109387121799845, + "grad_norm": 10.431690216064453, + "learning_rate": 2.9740517259313432e-05, + "loss": 0.9295, + "step": 1432 + }, + { + "epoch": 1.1117145073700543, + "grad_norm": 1.4973978996276855, + "learning_rate": 2.96280158679546e-05, + "loss": 0.0455, + "step": 1433 + }, + { + "epoch": 1.112490302560124, + "grad_norm": 10.173379898071289, + "learning_rate": 2.9514764063989333e-05, + "loss": 0.7356, + "step": 1434 + }, + { + "epoch": 1.113266097750194, + "grad_norm": 9.429365158081055, + "learning_rate": 2.9400780210704162e-05, + "loss": 0.8026, + "step": 1435 + }, + { + "epoch": 1.1140418929402638, + "grad_norm": 6.720676898956299, + "learning_rate": 2.9286082790084227e-05, + "loss": 0.3018, + "step": 1436 + }, + { + "epoch": 1.1148176881303336, + "grad_norm": 8.167119979858398, + "learning_rate": 2.9170690399816473e-05, + "loss": 0.4215, + "step": 1437 + }, + { + "epoch": 1.1155934833204033, + "grad_norm": 8.14616584777832, + "learning_rate": 2.9054621750274146e-05, + "loss": 0.4307, + "step": 1438 + }, + { + "epoch": 1.1163692785104733, + "grad_norm": 10.72919750213623, + "learning_rate": 2.8937895661483024e-05, + "loss": 0.9148, + "step": 1439 + }, + { + "epoch": 1.117145073700543, + "grad_norm": 6.909780502319336, + "learning_rate": 2.8820531060069775e-05, + "loss": 0.333, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_Qnli-dev_cosine_accuracy": 0.689453125, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8563566207885742, + "eval_Qnli-dev_cosine_ap": 0.737284844999962, + "eval_Qnli-dev_cosine_f1": 0.6916524701873935, + "eval_Qnli-dev_cosine_f1_threshold": 0.7498555183410645, + "eval_Qnli-dev_cosine_precision": 0.5783475783475783, + "eval_Qnli-dev_cosine_recall": 0.8601694915254238, + "eval_Qnli-dev_dot_accuracy": 0.654296875, + "eval_Qnli-dev_dot_accuracy_threshold": 365.72198486328125, + "eval_Qnli-dev_dot_ap": 0.6190767924770886, + "eval_Qnli-dev_dot_f1": 0.6688963210702342, + "eval_Qnli-dev_dot_f1_threshold": 333.119384765625, + "eval_Qnli-dev_dot_precision": 0.5524861878453039, + "eval_Qnli-dev_dot_recall": 0.847457627118644, + "eval_Qnli-dev_euclidean_accuracy": 0.69921875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.812474250793457, + "eval_Qnli-dev_euclidean_ap": 0.7432228058552675, + "eval_Qnli-dev_euclidean_f1": 0.6938775510204083, + "eval_Qnli-dev_euclidean_f1_threshold": 15.139213562011719, + "eval_Qnli-dev_euclidean_precision": 0.5795454545454546, + "eval_Qnli-dev_euclidean_recall": 0.864406779661017, + "eval_Qnli-dev_manhattan_accuracy": 0.693359375, + "eval_Qnli-dev_manhattan_accuracy_threshold": 233.50564575195312, + "eval_Qnli-dev_manhattan_ap": 0.7374577953888101, + "eval_Qnli-dev_manhattan_f1": 0.6908396946564884, + "eval_Qnli-dev_manhattan_f1_threshold": 257.90789794921875, + "eval_Qnli-dev_manhattan_precision": 0.6284722222222222, + "eval_Qnli-dev_manhattan_recall": 0.7669491525423728, + "eval_Qnli-dev_max_accuracy": 0.69921875, + "eval_Qnli-dev_max_accuracy_threshold": 365.72198486328125, + "eval_Qnli-dev_max_ap": 0.7432228058552675, + "eval_Qnli-dev_max_f1": 0.6938775510204083, + "eval_Qnli-dev_max_f1_threshold": 333.119384765625, + "eval_Qnli-dev_max_precision": 0.6284722222222222, + "eval_Qnli-dev_max_recall": 0.864406779661017, + "eval_allNLI-dev_cosine_accuracy": 0.71484375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8758759498596191, + "eval_allNLI-dev_cosine_ap": 0.5881676861465818, + "eval_allNLI-dev_cosine_f1": 0.5991189427312775, + "eval_allNLI-dev_cosine_f1_threshold": 0.7914148569107056, + "eval_allNLI-dev_cosine_precision": 0.48398576512455516, + "eval_allNLI-dev_cosine_recall": 0.7861271676300579, + "eval_allNLI-dev_dot_accuracy": 0.69140625, + "eval_allNLI-dev_dot_accuracy_threshold": 443.98907470703125, + "eval_allNLI-dev_dot_ap": 0.508547113644118, + "eval_allNLI-dev_dot_f1": 0.5805168986083499, + "eval_allNLI-dev_dot_f1_threshold": 329.53704833984375, + "eval_allNLI-dev_dot_precision": 0.44242424242424244, + "eval_allNLI-dev_dot_recall": 0.8439306358381503, + "eval_allNLI-dev_euclidean_accuracy": 0.716796875, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.91428279876709, + "eval_allNLI-dev_euclidean_ap": 0.5881883682252246, + "eval_allNLI-dev_euclidean_f1": 0.6036217303822938, + "eval_allNLI-dev_euclidean_f1_threshold": 14.558799743652344, + "eval_allNLI-dev_euclidean_precision": 0.46296296296296297, + "eval_allNLI-dev_euclidean_recall": 0.8670520231213873, + "eval_allNLI-dev_manhattan_accuracy": 0.72265625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 186.28140258789062, + "eval_allNLI-dev_manhattan_ap": 0.5912347318593615, + "eval_allNLI-dev_manhattan_f1": 0.6117647058823529, + "eval_allNLI-dev_manhattan_f1_threshold": 235.27369689941406, + "eval_allNLI-dev_manhattan_precision": 0.5158730158730159, + "eval_allNLI-dev_manhattan_recall": 0.7514450867052023, + "eval_allNLI-dev_max_accuracy": 0.72265625, + "eval_allNLI-dev_max_accuracy_threshold": 443.98907470703125, + "eval_allNLI-dev_max_ap": 0.5912347318593615, + "eval_allNLI-dev_max_f1": 0.6117647058823529, + "eval_allNLI-dev_max_f1_threshold": 329.53704833984375, + "eval_allNLI-dev_max_precision": 0.5158730158730159, + "eval_allNLI-dev_max_recall": 0.8670520231213873, + "eval_sequential_score": 0.7432228058552675, + "eval_sts-test_pearson_cosine": 0.8168412896244883, + "eval_sts-test_pearson_dot": 0.7602808644078655, + "eval_sts-test_pearson_euclidean": 0.8585768315446606, + "eval_sts-test_pearson_manhattan": 0.8544890138310812, + "eval_sts-test_pearson_max": 0.8585768315446606, + "eval_sts-test_spearman_cosine": 0.8625050231899969, + "eval_sts-test_spearman_dot": 0.7452011820291434, + "eval_sts-test_spearman_euclidean": 0.8612069233508166, + "eval_sts-test_spearman_manhattan": 0.8575647971912161, + "eval_sts-test_spearman_max": 0.8625050231899969, + "eval_vitaminc-pairs_loss": 3.4520416259765625, + "eval_vitaminc-pairs_runtime": 5.4306, + "eval_vitaminc-pairs_samples_per_second": 23.57, + "eval_vitaminc-pairs_steps_per_second": 0.368, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_negation-triplets_loss": 1.4830158948898315, + "eval_negation-triplets_runtime": 1.0967, + "eval_negation-triplets_samples_per_second": 116.719, + "eval_negation-triplets_steps_per_second": 1.824, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_scitail-pairs-pos_loss": 0.17665188014507294, + "eval_scitail-pairs-pos_runtime": 1.2137, + "eval_scitail-pairs-pos_samples_per_second": 105.462, + "eval_scitail-pairs-pos_steps_per_second": 1.648, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_scitail-pairs-qa_loss": 0.0034409333020448685, + "eval_scitail-pairs-qa_runtime": 0.9196, + "eval_scitail-pairs-qa_samples_per_second": 139.198, + "eval_scitail-pairs-qa_steps_per_second": 2.175, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_xsum-pairs_loss": 0.2398369163274765, + "eval_xsum-pairs_runtime": 4.2996, + "eval_xsum-pairs_samples_per_second": 29.77, + "eval_xsum-pairs_steps_per_second": 0.465, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_sciq_pairs_loss": 0.07447811216115952, + "eval_sciq_pairs_runtime": 5.5347, + "eval_sciq_pairs_samples_per_second": 23.127, + "eval_sciq_pairs_steps_per_second": 0.361, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_qasc_pairs_loss": 0.24804122745990753, + "eval_qasc_pairs_runtime": 0.9662, + "eval_qasc_pairs_samples_per_second": 132.472, + "eval_qasc_pairs_steps_per_second": 2.07, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_openbookqa_pairs_loss": 1.158809781074524, + "eval_openbookqa_pairs_runtime": 0.8488, + "eval_openbookqa_pairs_samples_per_second": 150.793, + "eval_openbookqa_pairs_steps_per_second": 2.356, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_msmarco_pairs_loss": 0.6112531423568726, + "eval_msmarco_pairs_runtime": 1.9454, + "eval_msmarco_pairs_samples_per_second": 65.796, + "eval_msmarco_pairs_steps_per_second": 1.028, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_nq_pairs_loss": 0.7372586131095886, + "eval_nq_pairs_runtime": 4.3896, + "eval_nq_pairs_samples_per_second": 29.16, + "eval_nq_pairs_steps_per_second": 0.456, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_trivia_pairs_loss": 0.6014779806137085, + "eval_trivia_pairs_runtime": 4.5108, + "eval_trivia_pairs_samples_per_second": 28.376, + "eval_trivia_pairs_steps_per_second": 0.443, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_gooaq_pairs_loss": 0.5425880551338196, + "eval_gooaq_pairs_runtime": 1.256, + "eval_gooaq_pairs_samples_per_second": 101.913, + "eval_gooaq_pairs_steps_per_second": 1.592, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_paws-pos_loss": 0.03510904312133789, + "eval_paws-pos_runtime": 1.0152, + "eval_paws-pos_samples_per_second": 126.089, + "eval_paws-pos_steps_per_second": 1.97, + "step": 1440 + }, + { + "epoch": 1.117145073700543, + "eval_global_dataset_loss": 0.44889751076698303, + "eval_global_dataset_runtime": 21.0775, + "eval_global_dataset_samples_per_second": 19.737, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1440 + }, + { + "epoch": 1.1179208688906128, + "grad_norm": 16.12545394897461, + "learning_rate": 2.870254697619307e-05, + "loss": 2.6087, + "step": 1441 + }, + { + "epoch": 1.1186966640806828, + "grad_norm": 7.544276237487793, + "learning_rate": 2.8583962540458027e-05, + "loss": 0.2893, + "step": 1442 + }, + { + "epoch": 1.1194724592707526, + "grad_norm": 0.6535027027130127, + "learning_rate": 2.8464796980814145e-05, + "loss": 0.015, + "step": 1443 + }, + { + "epoch": 1.1202482544608223, + "grad_norm": NaN, + "learning_rate": 2.8464796980814145e-05, + "loss": 0.766, + "step": 1444 + }, + { + "epoch": 1.121024049650892, + "grad_norm": 11.83984088897705, + "learning_rate": 2.8345069619437684e-05, + "loss": 1.0202, + "step": 1445 + }, + { + "epoch": 1.121799844840962, + "grad_norm": 6.6594953536987305, + "learning_rate": 2.8224799869598554e-05, + "loss": 0.2525, + "step": 1446 + }, + { + "epoch": 1.1225756400310318, + "grad_norm": 8.0680513381958, + "learning_rate": 2.8104007232512627e-05, + "loss": 0.3843, + "step": 1447 + }, + { + "epoch": 1.1233514352211016, + "grad_norm": 12.678421020507812, + "learning_rate": 2.7982711294179648e-05, + "loss": 1.3048, + "step": 1448 + }, + { + "epoch": 1.1241272304111714, + "grad_norm": 5.908491134643555, + "learning_rate": 2.786093172220749e-05, + "loss": 0.1779, + "step": 1449 + }, + { + "epoch": 1.1249030256012413, + "grad_norm": 11.057334899902344, + "learning_rate": 2.7738688262623053e-05, + "loss": 0.8845, + "step": 1450 + }, + { + "epoch": 1.125678820791311, + "grad_norm": 6.310258865356445, + "learning_rate": 2.7616000736670678e-05, + "loss": 0.3029, + "step": 1451 + }, + { + "epoch": 1.1264546159813809, + "grad_norm": 5.720364093780518, + "learning_rate": 2.7492889037598063e-05, + "loss": 0.2766, + "step": 1452 + }, + { + "epoch": 1.1272304111714506, + "grad_norm": 7.003293037414551, + "learning_rate": 2.736937312743075e-05, + "loss": 0.3545, + "step": 1453 + }, + { + "epoch": 1.1280062063615206, + "grad_norm": 10.090779304504395, + "learning_rate": 2.7245473033735376e-05, + "loss": 0.7774, + "step": 1454 + }, + { + "epoch": 1.1287820015515904, + "grad_norm": 6.065265655517578, + "learning_rate": 2.7121208846372206e-05, + "loss": 0.2583, + "step": 1455 + }, + { + "epoch": 1.1295577967416601, + "grad_norm": 4.22489595413208, + "learning_rate": 2.699660071423778e-05, + "loss": 0.1711, + "step": 1456 + }, + { + "epoch": 1.13033359193173, + "grad_norm": 9.47485637664795, + "learning_rate": 2.6871668841997724e-05, + "loss": 0.8988, + "step": 1457 + }, + { + "epoch": 1.1311093871217999, + "grad_norm": 6.85338830947876, + "learning_rate": 2.6746433486810708e-05, + "loss": 0.2521, + "step": 1458 + }, + { + "epoch": 1.1318851823118696, + "grad_norm": 10.234007835388184, + "learning_rate": 2.6620914955043898e-05, + "loss": 0.7194, + "step": 1459 + }, + { + "epoch": 1.1326609775019394, + "grad_norm": 6.6938252449035645, + "learning_rate": 2.6495133598980267e-05, + "loss": 0.283, + "step": 1460 + }, + { + "epoch": 1.1334367726920094, + "grad_norm": 5.914616584777832, + "learning_rate": 2.6369109813518587e-05, + "loss": 0.3763, + "step": 1461 + }, + { + "epoch": 1.1342125678820791, + "grad_norm": 0.40827468037605286, + "learning_rate": 2.6242864032866553e-05, + "loss": 0.0063, + "step": 1462 + }, + { + "epoch": 1.134988363072149, + "grad_norm": 10.07016658782959, + "learning_rate": 2.6116416727227357e-05, + "loss": 0.8174, + "step": 1463 + }, + { + "epoch": 1.1357641582622189, + "grad_norm": 4.915323257446289, + "learning_rate": 2.598978839948063e-05, + "loss": 0.2182, + "step": 1464 + }, + { + "epoch": 1.1365399534522886, + "grad_norm": 5.823632717132568, + "learning_rate": 2.5862999581857916e-05, + "loss": 0.2794, + "step": 1465 + }, + { + "epoch": 1.1373157486423584, + "grad_norm": 10.654269218444824, + "learning_rate": 2.573607083261353e-05, + "loss": 0.7149, + "step": 1466 + }, + { + "epoch": 1.1380915438324282, + "grad_norm": 6.282398700714111, + "learning_rate": 2.560902273269106e-05, + "loss": 0.1744, + "step": 1467 + }, + { + "epoch": 1.1388673390224981, + "grad_norm": 12.48471736907959, + "learning_rate": 2.5481875882386316e-05, + "loss": 0.9468, + "step": 1468 + }, + { + "epoch": 1.139643134212568, + "grad_norm": 7.10225248336792, + "learning_rate": 2.535465089800699e-05, + "loss": 0.4006, + "step": 1469 + }, + { + "epoch": 1.1404189294026377, + "grad_norm": 6.806367874145508, + "learning_rate": 2.522736840852996e-05, + "loss": 0.3854, + "step": 1470 + }, + { + "epoch": 1.1411947245927074, + "grad_norm": 8.481449127197266, + "learning_rate": 2.5100049052256235e-05, + "loss": 0.5539, + "step": 1471 + }, + { + "epoch": 1.1419705197827774, + "grad_norm": 6.161257266998291, + "learning_rate": 2.497271347346465e-05, + "loss": 0.2349, + "step": 1472 + }, + { + "epoch": 1.1427463149728472, + "grad_norm": 5.660153865814209, + "learning_rate": 2.4845382319064445e-05, + "loss": 0.1486, + "step": 1473 + }, + { + "epoch": 1.143522110162917, + "grad_norm": 0.8233408331871033, + "learning_rate": 2.4718076235247454e-05, + "loss": 0.0163, + "step": 1474 + }, + { + "epoch": 1.1442979053529867, + "grad_norm": 8.787489891052246, + "learning_rate": 2.4590815864140444e-05, + "loss": 0.3334, + "step": 1475 + }, + { + "epoch": 1.1450737005430567, + "grad_norm": 5.205694198608398, + "learning_rate": 2.4463621840458052e-05, + "loss": 0.3183, + "step": 1476 + }, + { + "epoch": 1.1458494957331264, + "grad_norm": 9.264506340026855, + "learning_rate": 2.433651478815694e-05, + "loss": 0.4481, + "step": 1477 + }, + { + "epoch": 1.1466252909231962, + "grad_norm": 13.73600959777832, + "learning_rate": 2.4209515317091802e-05, + "loss": 1.0267, + "step": 1478 + }, + { + "epoch": 1.1474010861132662, + "grad_norm": 8.001157760620117, + "learning_rate": 2.408264401967347e-05, + "loss": 0.3423, + "step": 1479 + }, + { + "epoch": 1.148176881303336, + "grad_norm": 5.337899208068848, + "learning_rate": 2.3955921467529955e-05, + "loss": 0.2553, + "step": 1480 + }, + { + "epoch": 1.1489526764934057, + "grad_norm": 10.858280181884766, + "learning_rate": 2.3829368208170935e-05, + "loss": 0.7728, + "step": 1481 + }, + { + "epoch": 1.1497284716834755, + "grad_norm": 11.384843826293945, + "learning_rate": 2.370300476165592e-05, + "loss": 0.8459, + "step": 1482 + }, + { + "epoch": 1.1505042668735455, + "grad_norm": 10.877307891845703, + "learning_rate": 2.357685161726715e-05, + "loss": 0.6853, + "step": 1483 + }, + { + "epoch": 1.1512800620636152, + "grad_norm": 9.117754936218262, + "learning_rate": 2.3450929230187284e-05, + "loss": 0.6812, + "step": 1484 + }, + { + "epoch": 1.152055857253685, + "grad_norm": 7.563530921936035, + "learning_rate": 2.3325258018182647e-05, + "loss": 0.2318, + "step": 1485 + }, + { + "epoch": 1.152831652443755, + "grad_norm": 1.230953574180603, + "learning_rate": 2.319985835829271e-05, + "loss": 0.023, + "step": 1486 + }, + { + "epoch": 1.1536074476338247, + "grad_norm": 10.394408226013184, + "learning_rate": 2.3074750583525844e-05, + "loss": 0.6199, + "step": 1487 + }, + { + "epoch": 1.1543832428238945, + "grad_norm": 8.30750560760498, + "learning_rate": 2.2949954979562575e-05, + "loss": 0.3305, + "step": 1488 + }, + { + "epoch": 1.1551590380139642, + "grad_norm": 6.444952964782715, + "learning_rate": 2.282549178146634e-05, + "loss": 0.2415, + "step": 1489 + }, + { + "epoch": 1.1559348332040342, + "grad_norm": 9.098458290100098, + "learning_rate": 2.2701381170402308e-05, + "loss": 0.5643, + "step": 1490 + }, + { + "epoch": 1.156710628394104, + "grad_norm": 6.67218017578125, + "learning_rate": 2.2577643270365266e-05, + "loss": 0.304, + "step": 1491 + }, + { + "epoch": 1.1574864235841738, + "grad_norm": 11.883628845214844, + "learning_rate": 2.245429814491653e-05, + "loss": 0.7578, + "step": 1492 + }, + { + "epoch": 1.1582622187742435, + "grad_norm": 4.988454818725586, + "learning_rate": 2.2331365793930695e-05, + "loss": 0.2544, + "step": 1493 + }, + { + "epoch": 1.1590380139643135, + "grad_norm": 0.8163090348243713, + "learning_rate": 2.2208866150352814e-05, + "loss": 0.0158, + "step": 1494 + }, + { + "epoch": 1.1598138091543833, + "grad_norm": 10.838471412658691, + "learning_rate": 2.2086819076966298e-05, + "loss": 0.6797, + "step": 1495 + }, + { + "epoch": 1.160589604344453, + "grad_norm": 8.346417427062988, + "learning_rate": 2.1965244363172266e-05, + "loss": 0.6421, + "step": 1496 + }, + { + "epoch": 1.1613653995345228, + "grad_norm": 8.896296501159668, + "learning_rate": 2.184416172178081e-05, + "loss": 0.5688, + "step": 1497 + }, + { + "epoch": 1.1621411947245928, + "grad_norm": 0.6204801201820374, + "learning_rate": 2.1723590785814613e-05, + "loss": 0.0122, + "step": 1498 + }, + { + "epoch": 1.1629169899146625, + "grad_norm": 13.445223808288574, + "learning_rate": 2.1603551105325545e-05, + "loss": 0.8166, + "step": 1499 + }, + { + "epoch": 1.1636927851047323, + "grad_norm": 0.03991204500198364, + "learning_rate": 2.1484062144224736e-05, + "loss": 0.0005, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_Qnli-dev_cosine_accuracy": 0.68359375, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8286857604980469, + "eval_Qnli-dev_cosine_ap": 0.736880493167421, + "eval_Qnli-dev_cosine_f1": 0.6936790923824959, + "eval_Qnli-dev_cosine_f1_threshold": 0.7110843658447266, + "eval_Qnli-dev_cosine_precision": 0.5616797900262467, + "eval_Qnli-dev_cosine_recall": 0.9067796610169492, + "eval_Qnli-dev_dot_accuracy": 0.658203125, + "eval_Qnli-dev_dot_accuracy_threshold": 374.8796081542969, + "eval_Qnli-dev_dot_ap": 0.6422818458297463, + "eval_Qnli-dev_dot_f1": 0.6703296703296704, + "eval_Qnli-dev_dot_f1_threshold": 353.6007080078125, + "eval_Qnli-dev_dot_precision": 0.5903225806451613, + "eval_Qnli-dev_dot_recall": 0.7754237288135594, + "eval_Qnli-dev_euclidean_accuracy": 0.6953125, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.278226852416992, + "eval_Qnli-dev_euclidean_ap": 0.7396819126166789, + "eval_Qnli-dev_euclidean_f1": 0.697754749568221, + "eval_Qnli-dev_euclidean_f1_threshold": 15.528464317321777, + "eval_Qnli-dev_euclidean_precision": 0.5889212827988338, + "eval_Qnli-dev_euclidean_recall": 0.8559322033898306, + "eval_Qnli-dev_manhattan_accuracy": 0.69140625, + "eval_Qnli-dev_manhattan_accuracy_threshold": 222.5366973876953, + "eval_Qnli-dev_manhattan_ap": 0.7328256012899059, + "eval_Qnli-dev_manhattan_f1": 0.6946983546617915, + "eval_Qnli-dev_manhattan_f1_threshold": 274.32061767578125, + "eval_Qnli-dev_manhattan_precision": 0.6109324758842444, + "eval_Qnli-dev_manhattan_recall": 0.8050847457627118, + "eval_Qnli-dev_max_accuracy": 0.6953125, + "eval_Qnli-dev_max_accuracy_threshold": 374.8796081542969, + "eval_Qnli-dev_max_ap": 0.7396819126166789, + "eval_Qnli-dev_max_f1": 0.697754749568221, + "eval_Qnli-dev_max_f1_threshold": 353.6007080078125, + "eval_Qnli-dev_max_precision": 0.6109324758842444, + "eval_Qnli-dev_max_recall": 0.9067796610169492, + "eval_allNLI-dev_cosine_accuracy": 0.71484375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8948071002960205, + "eval_allNLI-dev_cosine_ap": 0.593971811663098, + "eval_allNLI-dev_cosine_f1": 0.6214953271028038, + "eval_allNLI-dev_cosine_f1_threshold": 0.795933723449707, + "eval_allNLI-dev_cosine_precision": 0.5215686274509804, + "eval_allNLI-dev_cosine_recall": 0.7687861271676301, + "eval_allNLI-dev_dot_accuracy": 0.681640625, + "eval_allNLI-dev_dot_accuracy_threshold": 423.38787841796875, + "eval_allNLI-dev_dot_ap": 0.5114670298881441, + "eval_allNLI-dev_dot_f1": 0.5902912621359223, + "eval_allNLI-dev_dot_f1_threshold": 319.1331787109375, + "eval_allNLI-dev_dot_precision": 0.4444444444444444, + "eval_allNLI-dev_dot_recall": 0.8786127167630058, + "eval_allNLI-dev_euclidean_accuracy": 0.7265625, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.870847702026367, + "eval_allNLI-dev_euclidean_ap": 0.5929407548941804, + "eval_allNLI-dev_euclidean_f1": 0.6069868995633189, + "eval_allNLI-dev_euclidean_f1_threshold": 14.404670715332031, + "eval_allNLI-dev_euclidean_precision": 0.48771929824561405, + "eval_allNLI-dev_euclidean_recall": 0.8034682080924855, + "eval_allNLI-dev_manhattan_accuracy": 0.73046875, + "eval_allNLI-dev_manhattan_accuracy_threshold": 192.7371368408203, + "eval_allNLI-dev_manhattan_ap": 0.5944703592763119, + "eval_allNLI-dev_manhattan_f1": 0.6206896551724138, + "eval_allNLI-dev_manhattan_f1_threshold": 238.46380615234375, + "eval_allNLI-dev_manhattan_precision": 0.5407725321888412, + "eval_allNLI-dev_manhattan_recall": 0.7283236994219653, + "eval_allNLI-dev_max_accuracy": 0.73046875, + "eval_allNLI-dev_max_accuracy_threshold": 423.38787841796875, + "eval_allNLI-dev_max_ap": 0.5944703592763119, + "eval_allNLI-dev_max_f1": 0.6214953271028038, + "eval_allNLI-dev_max_f1_threshold": 319.1331787109375, + "eval_allNLI-dev_max_precision": 0.5407725321888412, + "eval_allNLI-dev_max_recall": 0.8786127167630058, + "eval_sequential_score": 0.7396819126166789, + "eval_sts-test_pearson_cosine": 0.8239486466163888, + "eval_sts-test_pearson_dot": 0.7681210471710991, + "eval_sts-test_pearson_euclidean": 0.8621935745766299, + "eval_sts-test_pearson_manhattan": 0.8565723751985608, + "eval_sts-test_pearson_max": 0.8621935745766299, + "eval_sts-test_spearman_cosine": 0.8663949370839216, + "eval_sts-test_spearman_dot": 0.7645901582908426, + "eval_sts-test_spearman_euclidean": 0.863693571941291, + "eval_sts-test_spearman_manhattan": 0.8587123537870758, + "eval_sts-test_spearman_max": 0.8663949370839216, + "eval_vitaminc-pairs_loss": 3.6729514598846436, + "eval_vitaminc-pairs_runtime": 5.4376, + "eval_vitaminc-pairs_samples_per_second": 23.54, + "eval_vitaminc-pairs_steps_per_second": 0.368, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_negation-triplets_loss": 1.4283326864242554, + "eval_negation-triplets_runtime": 1.1177, + "eval_negation-triplets_samples_per_second": 114.525, + "eval_negation-triplets_steps_per_second": 1.789, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_scitail-pairs-pos_loss": 0.13201840221881866, + "eval_scitail-pairs-pos_runtime": 1.2104, + "eval_scitail-pairs-pos_samples_per_second": 105.748, + "eval_scitail-pairs-pos_steps_per_second": 1.652, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_scitail-pairs-qa_loss": 0.0013827907387167215, + "eval_scitail-pairs-qa_runtime": 0.8964, + "eval_scitail-pairs-qa_samples_per_second": 142.788, + "eval_scitail-pairs-qa_steps_per_second": 2.231, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_xsum-pairs_loss": 0.14958305656909943, + "eval_xsum-pairs_runtime": 4.2923, + "eval_xsum-pairs_samples_per_second": 29.821, + "eval_xsum-pairs_steps_per_second": 0.466, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_sciq_pairs_loss": 0.07079197466373444, + "eval_sciq_pairs_runtime": 5.4853, + "eval_sciq_pairs_samples_per_second": 23.335, + "eval_sciq_pairs_steps_per_second": 0.365, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_qasc_pairs_loss": 0.19136451184749603, + "eval_qasc_pairs_runtime": 0.9681, + "eval_qasc_pairs_samples_per_second": 132.221, + "eval_qasc_pairs_steps_per_second": 2.066, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_openbookqa_pairs_loss": 1.1326899528503418, + "eval_openbookqa_pairs_runtime": 0.8686, + "eval_openbookqa_pairs_samples_per_second": 147.369, + "eval_openbookqa_pairs_steps_per_second": 2.303, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_msmarco_pairs_loss": 0.7426297068595886, + "eval_msmarco_pairs_runtime": 1.9461, + "eval_msmarco_pairs_samples_per_second": 65.774, + "eval_msmarco_pairs_steps_per_second": 1.028, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_nq_pairs_loss": 0.744480311870575, + "eval_nq_pairs_runtime": 4.3923, + "eval_nq_pairs_samples_per_second": 29.142, + "eval_nq_pairs_steps_per_second": 0.455, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_trivia_pairs_loss": 0.632261335849762, + "eval_trivia_pairs_runtime": 4.5087, + "eval_trivia_pairs_samples_per_second": 28.39, + "eval_trivia_pairs_steps_per_second": 0.444, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_gooaq_pairs_loss": 0.5575349926948547, + "eval_gooaq_pairs_runtime": 1.2542, + "eval_gooaq_pairs_samples_per_second": 102.057, + "eval_gooaq_pairs_steps_per_second": 1.595, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_paws-pos_loss": 0.034535180777311325, + "eval_paws-pos_runtime": 1.0175, + "eval_paws-pos_samples_per_second": 125.803, + "eval_paws-pos_steps_per_second": 1.966, + "step": 1500 + }, + { + "epoch": 1.1636927851047323, + "eval_global_dataset_loss": 0.4625393748283386, + "eval_global_dataset_runtime": 21.0957, + "eval_global_dataset_samples_per_second": 19.72, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1500 + }, + { + "epoch": 1.1644685802948023, + "grad_norm": 1.2380955219268799, + "learning_rate": 2.1365143277126562e-05, + "loss": 0.0475, + "step": 1501 + }, + { + "epoch": 1.165244375484872, + "grad_norm": 7.785430908203125, + "learning_rate": 2.1246813786207095e-05, + "loss": 0.4198, + "step": 1502 + }, + { + "epoch": 1.1660201706749418, + "grad_norm": 1.1769071817398071, + "learning_rate": 2.1129092858077727e-05, + "loss": 0.0112, + "step": 1503 + }, + { + "epoch": 1.1667959658650116, + "grad_norm": 7.791942596435547, + "learning_rate": 2.1011999580673922e-05, + "loss": 0.6125, + "step": 1504 + }, + { + "epoch": 1.1675717610550815, + "grad_norm": 7.968329429626465, + "learning_rate": 2.089555294016044e-05, + "loss": 0.2906, + "step": 1505 + }, + { + "epoch": 1.1683475562451513, + "grad_norm": 13.652813911437988, + "learning_rate": 2.0779771817852575e-05, + "loss": 1.1571, + "step": 1506 + }, + { + "epoch": 1.169123351435221, + "grad_norm": 8.930511474609375, + "learning_rate": 2.06646749871548e-05, + "loss": 0.5272, + "step": 1507 + }, + { + "epoch": 1.169899146625291, + "grad_norm": 5.6013922691345215, + "learning_rate": 2.0550281110516722e-05, + "loss": 0.2146, + "step": 1508 + }, + { + "epoch": 1.1706749418153608, + "grad_norm": 0.5505045652389526, + "learning_rate": 2.043660873640692e-05, + "loss": 0.0101, + "step": 1509 + }, + { + "epoch": 1.1714507370054306, + "grad_norm": 7.619174480438232, + "learning_rate": 2.032367629630554e-05, + "loss": 0.2705, + "step": 1510 + }, + { + "epoch": 1.1722265321955003, + "grad_norm": 9.072793960571289, + "learning_rate": 2.0211502101715647e-05, + "loss": 0.6662, + "step": 1511 + }, + { + "epoch": 1.17300232738557, + "grad_norm": 6.0196709632873535, + "learning_rate": 2.0100104341194096e-05, + "loss": 0.2002, + "step": 1512 + }, + { + "epoch": 1.17377812257564, + "grad_norm": 7.038978576660156, + "learning_rate": 1.9989501077402333e-05, + "loss": 0.3238, + "step": 1513 + }, + { + "epoch": 1.1745539177657098, + "grad_norm": 10.138888359069824, + "learning_rate": 1.987971024417764e-05, + "loss": 0.5931, + "step": 1514 + }, + { + "epoch": 1.1753297129557796, + "grad_norm": 12.482297897338867, + "learning_rate": 1.9770749643625196e-05, + "loss": 0.8702, + "step": 1515 + }, + { + "epoch": 1.1761055081458496, + "grad_norm": 11.27192211151123, + "learning_rate": 1.96626369432316e-05, + "loss": 0.7421, + "step": 1516 + }, + { + "epoch": 1.1768813033359193, + "grad_norm": 16.40995216369629, + "learning_rate": 1.9555389673000107e-05, + "loss": 2.2864, + "step": 1517 + }, + { + "epoch": 1.177657098525989, + "grad_norm": 8.062331199645996, + "learning_rate": 1.9449025222608266e-05, + "loss": 0.5332, + "step": 1518 + }, + { + "epoch": 1.1784328937160589, + "grad_norm": 0.15882490575313568, + "learning_rate": 1.934356083858822e-05, + "loss": 0.0027, + "step": 1519 + }, + { + "epoch": 1.1792086889061288, + "grad_norm": 7.452869892120361, + "learning_rate": 1.9239013621530292e-05, + "loss": 0.311, + "step": 1520 + }, + { + "epoch": 1.1799844840961986, + "grad_norm": 10.761478424072266, + "learning_rate": 1.9135400523310124e-05, + "loss": 0.6303, + "step": 1521 + }, + { + "epoch": 1.1807602792862684, + "grad_norm": 5.3508501052856445, + "learning_rate": 1.9032738344340146e-05, + "loss": 0.1651, + "step": 1522 + }, + { + "epoch": 1.1815360744763383, + "grad_norm": 5.727583408355713, + "learning_rate": 1.8931043730845295e-05, + "loss": 0.174, + "step": 1523 + }, + { + "epoch": 1.182311869666408, + "grad_norm": 7.603945255279541, + "learning_rate": 1.8830333172164026e-05, + "loss": 0.3305, + "step": 1524 + }, + { + "epoch": 1.1830876648564779, + "grad_norm": 7.93798303604126, + "learning_rate": 1.8730622998074585e-05, + "loss": 0.5773, + "step": 1525 + }, + { + "epoch": 1.1838634600465476, + "grad_norm": 3.5564613342285156, + "learning_rate": 1.86319293761472e-05, + "loss": 0.1239, + "step": 1526 + }, + { + "epoch": 1.1846392552366176, + "grad_norm": 0.5727388858795166, + "learning_rate": 1.8534268309122643e-05, + "loss": 0.0113, + "step": 1527 + }, + { + "epoch": 1.1854150504266874, + "grad_norm": 8.769505500793457, + "learning_rate": 1.843765563231734e-05, + "loss": 0.5165, + "step": 1528 + }, + { + "epoch": 1.1861908456167571, + "grad_norm": 0.21828579902648926, + "learning_rate": 1.8342107011055865e-05, + "loss": 0.0032, + "step": 1529 + }, + { + "epoch": 1.1869666408068271, + "grad_norm": 11.185166358947754, + "learning_rate": 1.8247637938130788e-05, + "loss": 0.7473, + "step": 1530 + }, + { + "epoch": 1.1877424359968969, + "grad_norm": 9.555970191955566, + "learning_rate": 1.8154263731290644e-05, + "loss": 0.6532, + "step": 1531 + }, + { + "epoch": 1.1885182311869666, + "grad_norm": 6.298295021057129, + "learning_rate": 1.8061999530756193e-05, + "loss": 0.1886, + "step": 1532 + }, + { + "epoch": 1.1892940263770364, + "grad_norm": 7.181407451629639, + "learning_rate": 1.7970860296765513e-05, + "loss": 0.3465, + "step": 1533 + }, + { + "epoch": 1.1900698215671062, + "grad_norm": 0.16474811732769012, + "learning_rate": 1.7880860807148267e-05, + "loss": 0.0026, + "step": 1534 + }, + { + "epoch": 1.1908456167571762, + "grad_norm": 7.771115779876709, + "learning_rate": 1.779201565492949e-05, + "loss": 0.3026, + "step": 1535 + }, + { + "epoch": 1.191621411947246, + "grad_norm": 9.012238502502441, + "learning_rate": 1.7704339245963497e-05, + "loss": 0.4752, + "step": 1536 + }, + { + "epoch": 1.1923972071373157, + "grad_norm": 9.43275260925293, + "learning_rate": 1.7617845796597952e-05, + "loss": 0.5882, + "step": 1537 + }, + { + "epoch": 1.1931730023273857, + "grad_norm": 8.851835250854492, + "learning_rate": 1.753254933136877e-05, + "loss": 0.666, + "step": 1538 + }, + { + "epoch": 1.1939487975174554, + "grad_norm": 10.922514915466309, + "learning_rate": 1.7448463680726122e-05, + "loss": 0.7948, + "step": 1539 + }, + { + "epoch": 1.1947245927075252, + "grad_norm": 10.574271202087402, + "learning_rate": 1.7365602478791822e-05, + "loss": 0.6849, + "step": 1540 + }, + { + "epoch": 1.195500387897595, + "grad_norm": 7.074329853057861, + "learning_rate": 1.7283979161148738e-05, + "loss": 0.4581, + "step": 1541 + }, + { + "epoch": 1.196276183087665, + "grad_norm": 5.094451904296875, + "learning_rate": 1.7203606962662122e-05, + "loss": 0.2484, + "step": 1542 + }, + { + "epoch": 1.1970519782777347, + "grad_norm": 14.666666984558105, + "learning_rate": 1.712449891533374e-05, + "loss": 2.1819, + "step": 1543 + }, + { + "epoch": 1.1978277734678044, + "grad_norm": 8.778453826904297, + "learning_rate": 1.704666784618878e-05, + "loss": 0.4339, + "step": 1544 + }, + { + "epoch": 1.1986035686578744, + "grad_norm": 7.847414493560791, + "learning_rate": 1.697012637519592e-05, + "loss": 0.6207, + "step": 1545 + }, + { + "epoch": 1.1993793638479442, + "grad_norm": 7.098594665527344, + "learning_rate": 1.6894886913221184e-05, + "loss": 0.5111, + "step": 1546 + }, + { + "epoch": 1.200155159038014, + "grad_norm": 1.4133108854293823, + "learning_rate": 1.682096166001545e-05, + "loss": 0.0276, + "step": 1547 + }, + { + "epoch": 1.2009309542280837, + "grad_norm": 5.427593231201172, + "learning_rate": 1.6748362602236404e-05, + "loss": 0.1887, + "step": 1548 + }, + { + "epoch": 1.2017067494181537, + "grad_norm": 8.035236358642578, + "learning_rate": 1.667710151150492e-05, + "loss": 0.4806, + "step": 1549 + }, + { + "epoch": 1.2024825446082235, + "grad_norm": 9.935614585876465, + "learning_rate": 1.6607189942496338e-05, + "loss": 0.4082, + "step": 1550 + }, + { + "epoch": 1.2032583397982932, + "grad_norm": 2.79838490486145, + "learning_rate": 1.653863923106693e-05, + "loss": 0.0667, + "step": 1551 + }, + { + "epoch": 1.204034134988363, + "grad_norm": 12.595191955566406, + "learning_rate": 1.6471460492415845e-05, + "loss": 1.0073, + "step": 1552 + }, + { + "epoch": 1.204809930178433, + "grad_norm": 7.502379894256592, + "learning_rate": 1.640566461928283e-05, + "loss": 0.3208, + "step": 1553 + }, + { + "epoch": 1.2055857253685027, + "grad_norm": 11.361260414123535, + "learning_rate": 1.6341262280182e-05, + "loss": 0.7815, + "step": 1554 + }, + { + "epoch": 1.2063615205585725, + "grad_norm": 8.955950736999512, + "learning_rate": 1.6278263917672014e-05, + "loss": 0.6104, + "step": 1555 + }, + { + "epoch": 1.2071373157486422, + "grad_norm": 8.359417915344238, + "learning_rate": 1.6216679746662832e-05, + "loss": 0.3631, + "step": 1556 + }, + { + "epoch": 1.2079131109387122, + "grad_norm": 9.19699764251709, + "learning_rate": 1.6156519752759436e-05, + "loss": 0.5627, + "step": 1557 + }, + { + "epoch": 1.208688906128782, + "grad_norm": 1.0740132331848145, + "learning_rate": 1.6097793690642694e-05, + "loss": 0.0198, + "step": 1558 + }, + { + "epoch": 1.2094647013188518, + "grad_norm": 8.647027969360352, + "learning_rate": 1.604051108248766e-05, + "loss": 0.4917, + "step": 1559 + }, + { + "epoch": 1.2102404965089217, + "grad_norm": 5.0128960609436035, + "learning_rate": 1.5984681216419663e-05, + "loss": 0.2046, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_Qnli-dev_cosine_accuracy": 0.6796875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8328875303268433, + "eval_Qnli-dev_cosine_ap": 0.7309377392302046, + "eval_Qnli-dev_cosine_f1": 0.6944908180300502, + "eval_Qnli-dev_cosine_f1_threshold": 0.7466595768928528, + "eval_Qnli-dev_cosine_precision": 0.5730027548209367, + "eval_Qnli-dev_cosine_recall": 0.8813559322033898, + "eval_Qnli-dev_dot_accuracy": 0.6484375, + "eval_Qnli-dev_dot_accuracy_threshold": 383.61895751953125, + "eval_Qnli-dev_dot_ap": 0.6360062257583865, + "eval_Qnli-dev_dot_f1": 0.67, + "eval_Qnli-dev_dot_f1_threshold": 351.2830810546875, + "eval_Qnli-dev_dot_precision": 0.5521978021978022, + "eval_Qnli-dev_dot_recall": 0.8516949152542372, + "eval_Qnli-dev_euclidean_accuracy": 0.685546875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.017932891845703, + "eval_Qnli-dev_euclidean_ap": 0.731910118681418, + "eval_Qnli-dev_euclidean_f1": 0.698581560283688, + "eval_Qnli-dev_euclidean_f1_threshold": 14.888692855834961, + "eval_Qnli-dev_euclidean_precision": 0.600609756097561, + "eval_Qnli-dev_euclidean_recall": 0.8347457627118644, + "eval_Qnli-dev_manhattan_accuracy": 0.6875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 243.98065185546875, + "eval_Qnli-dev_manhattan_ap": 0.7290599348441305, + "eval_Qnli-dev_manhattan_f1": 0.693950177935943, + "eval_Qnli-dev_manhattan_f1_threshold": 271.59228515625, + "eval_Qnli-dev_manhattan_precision": 0.598159509202454, + "eval_Qnli-dev_manhattan_recall": 0.826271186440678, + "eval_Qnli-dev_max_accuracy": 0.6875, + "eval_Qnli-dev_max_accuracy_threshold": 383.61895751953125, + "eval_Qnli-dev_max_ap": 0.731910118681418, + "eval_Qnli-dev_max_f1": 0.698581560283688, + "eval_Qnli-dev_max_f1_threshold": 351.2830810546875, + "eval_Qnli-dev_max_precision": 0.600609756097561, + "eval_Qnli-dev_max_recall": 0.8813559322033898, + "eval_allNLI-dev_cosine_accuracy": 0.71484375, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8919122815132141, + "eval_allNLI-dev_cosine_ap": 0.588113852043148, + "eval_allNLI-dev_cosine_f1": 0.6192468619246863, + "eval_allNLI-dev_cosine_f1_threshold": 0.7723808288574219, + "eval_allNLI-dev_cosine_precision": 0.4852459016393443, + "eval_allNLI-dev_cosine_recall": 0.8554913294797688, + "eval_allNLI-dev_dot_accuracy": 0.68359375, + "eval_allNLI-dev_dot_accuracy_threshold": 459.2498779296875, + "eval_allNLI-dev_dot_ap": 0.5044325365587718, + "eval_allNLI-dev_dot_f1": 0.5836909871244635, + "eval_allNLI-dev_dot_f1_threshold": 348.32275390625, + "eval_allNLI-dev_dot_precision": 0.4641638225255973, + "eval_allNLI-dev_dot_recall": 0.7861271676300579, + "eval_allNLI-dev_euclidean_accuracy": 0.716796875, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.666543006896973, + "eval_allNLI-dev_euclidean_ap": 0.5917175245530598, + "eval_allNLI-dev_euclidean_f1": 0.6160714285714286, + "eval_allNLI-dev_euclidean_f1_threshold": 13.741901397705078, + "eval_allNLI-dev_euclidean_precision": 0.5018181818181818, + "eval_allNLI-dev_euclidean_recall": 0.7976878612716763, + "eval_allNLI-dev_manhattan_accuracy": 0.73046875, + "eval_allNLI-dev_manhattan_accuracy_threshold": 181.99508666992188, + "eval_allNLI-dev_manhattan_ap": 0.5894122628144837, + "eval_allNLI-dev_manhattan_f1": 0.6144329896907217, + "eval_allNLI-dev_manhattan_f1_threshold": 265.2486877441406, + "eval_allNLI-dev_manhattan_precision": 0.4775641025641026, + "eval_allNLI-dev_manhattan_recall": 0.861271676300578, + "eval_allNLI-dev_max_accuracy": 0.73046875, + "eval_allNLI-dev_max_accuracy_threshold": 459.2498779296875, + "eval_allNLI-dev_max_ap": 0.5917175245530598, + "eval_allNLI-dev_max_f1": 0.6192468619246863, + "eval_allNLI-dev_max_f1_threshold": 348.32275390625, + "eval_allNLI-dev_max_precision": 0.5018181818181818, + "eval_allNLI-dev_max_recall": 0.861271676300578, + "eval_sequential_score": 0.731910118681418, + "eval_sts-test_pearson_cosine": 0.8319445503531696, + "eval_sts-test_pearson_dot": 0.7961187667090965, + "eval_sts-test_pearson_euclidean": 0.8723271159169943, + "eval_sts-test_pearson_manhattan": 0.8695138876381538, + "eval_sts-test_pearson_max": 0.8723271159169943, + "eval_sts-test_spearman_cosine": 0.8749276153421583, + "eval_sts-test_spearman_dot": 0.7859199520078254, + "eval_sts-test_spearman_euclidean": 0.8721875403584243, + "eval_sts-test_spearman_manhattan": 0.869902496200518, + "eval_sts-test_spearman_max": 0.8749276153421583, + "eval_vitaminc-pairs_loss": 3.7520534992218018, + "eval_vitaminc-pairs_runtime": 5.4241, + "eval_vitaminc-pairs_samples_per_second": 23.598, + "eval_vitaminc-pairs_steps_per_second": 0.369, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_negation-triplets_loss": 1.4263218641281128, + "eval_negation-triplets_runtime": 1.0958, + "eval_negation-triplets_samples_per_second": 116.811, + "eval_negation-triplets_steps_per_second": 1.825, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_scitail-pairs-pos_loss": 0.12758073210716248, + "eval_scitail-pairs-pos_runtime": 1.2148, + "eval_scitail-pairs-pos_samples_per_second": 105.37, + "eval_scitail-pairs-pos_steps_per_second": 1.646, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_scitail-pairs-qa_loss": 0.003996712155640125, + "eval_scitail-pairs-qa_runtime": 0.9035, + "eval_scitail-pairs-qa_samples_per_second": 141.67, + "eval_scitail-pairs-qa_steps_per_second": 2.214, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_xsum-pairs_loss": 0.15924642980098724, + "eval_xsum-pairs_runtime": 4.2961, + "eval_xsum-pairs_samples_per_second": 29.795, + "eval_xsum-pairs_steps_per_second": 0.466, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_sciq_pairs_loss": 0.052587881684303284, + "eval_sciq_pairs_runtime": 5.4798, + "eval_sciq_pairs_samples_per_second": 23.359, + "eval_sciq_pairs_steps_per_second": 0.365, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_qasc_pairs_loss": 0.23175887763500214, + "eval_qasc_pairs_runtime": 0.9659, + "eval_qasc_pairs_samples_per_second": 132.525, + "eval_qasc_pairs_steps_per_second": 2.071, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_openbookqa_pairs_loss": 1.0454384088516235, + "eval_openbookqa_pairs_runtime": 0.8546, + "eval_openbookqa_pairs_samples_per_second": 149.777, + "eval_openbookqa_pairs_steps_per_second": 2.34, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_msmarco_pairs_loss": 0.5985706448554993, + "eval_msmarco_pairs_runtime": 1.9601, + "eval_msmarco_pairs_samples_per_second": 65.303, + "eval_msmarco_pairs_steps_per_second": 1.02, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_nq_pairs_loss": 0.6394537687301636, + "eval_nq_pairs_runtime": 4.4213, + "eval_nq_pairs_samples_per_second": 28.951, + "eval_nq_pairs_steps_per_second": 0.452, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_trivia_pairs_loss": 0.4779373109340668, + "eval_trivia_pairs_runtime": 4.5027, + "eval_trivia_pairs_samples_per_second": 28.428, + "eval_trivia_pairs_steps_per_second": 0.444, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_gooaq_pairs_loss": 0.4913390278816223, + "eval_gooaq_pairs_runtime": 1.2621, + "eval_gooaq_pairs_samples_per_second": 101.422, + "eval_gooaq_pairs_steps_per_second": 1.585, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_paws-pos_loss": 0.03463022783398628, + "eval_paws-pos_runtime": 1.0144, + "eval_paws-pos_samples_per_second": 126.186, + "eval_paws-pos_steps_per_second": 1.972, + "step": 1560 + }, + { + "epoch": 1.2102404965089217, + "eval_global_dataset_loss": 0.4250268340110779, + "eval_global_dataset_runtime": 21.1059, + "eval_global_dataset_samples_per_second": 19.71, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1560 + }, + { + "epoch": 1.2110162916989915, + "grad_norm": 6.13113260269165, + "learning_rate": 1.593031314500819e-05, + "loss": 0.1724, + "step": 1561 + }, + { + "epoch": 1.2117920868890613, + "grad_norm": 7.434237957000732, + "learning_rate": 1.587741568379911e-05, + "loss": 0.4824, + "step": 1562 + }, + { + "epoch": 1.212567882079131, + "grad_norm": 11.15839672088623, + "learning_rate": 1.5825997409885276e-05, + "loss": 0.671, + "step": 1563 + }, + { + "epoch": 1.213343677269201, + "grad_norm": 10.371501922607422, + "learning_rate": 1.577606666051573e-05, + "loss": 1.0451, + "step": 1564 + }, + { + "epoch": 1.2141194724592708, + "grad_norm": 8.06485652923584, + "learning_rate": 1.5727631531743916e-05, + "loss": 0.3376, + "step": 1565 + }, + { + "epoch": 1.2148952676493405, + "grad_norm": 5.738529682159424, + "learning_rate": 1.568069987711491e-05, + "loss": 0.2706, + "step": 1566 + }, + { + "epoch": 1.2156710628394105, + "grad_norm": 14.194685935974121, + "learning_rate": 1.5635279306392e-05, + "loss": 2.1781, + "step": 1567 + }, + { + "epoch": 1.2164468580294803, + "grad_norm": 12.265534400939941, + "learning_rate": 1.5591377184322814e-05, + "loss": 0.9441, + "step": 1568 + }, + { + "epoch": 1.21722265321955, + "grad_norm": 9.075854301452637, + "learning_rate": 1.5549000629445138e-05, + "loss": 0.7491, + "step": 1569 + }, + { + "epoch": 1.2179984484096198, + "grad_norm": 10.4871826171875, + "learning_rate": 1.5508156512932695e-05, + "loss": 0.5274, + "step": 1570 + }, + { + "epoch": 1.2187742435996898, + "grad_norm": 1.3693101406097412, + "learning_rate": 1.5468851457480997e-05, + "loss": 0.0359, + "step": 1571 + }, + { + "epoch": 1.2195500387897595, + "grad_norm": 7.773709774017334, + "learning_rate": 1.5431091836233512e-05, + "loss": 0.6424, + "step": 1572 + }, + { + "epoch": 1.2203258339798293, + "grad_norm": 0.6953261494636536, + "learning_rate": 1.539488377174829e-05, + "loss": 0.0035, + "step": 1573 + }, + { + "epoch": 1.221101629169899, + "grad_norm": 9.300495147705078, + "learning_rate": 1.536023313500521e-05, + "loss": 0.3712, + "step": 1574 + }, + { + "epoch": 1.221877424359969, + "grad_norm": 6.792399883270264, + "learning_rate": 1.5327145544454033e-05, + "loss": 0.1988, + "step": 1575 + }, + { + "epoch": 1.2226532195500388, + "grad_norm": 13.107757568359375, + "learning_rate": 1.5295626365103375e-05, + "loss": 1.2197, + "step": 1576 + }, + { + "epoch": 1.2234290147401086, + "grad_norm": 8.831341743469238, + "learning_rate": 1.5265680707650844e-05, + "loss": 0.4283, + "step": 1577 + }, + { + "epoch": 1.2242048099301783, + "grad_norm": 9.626032829284668, + "learning_rate": 1.5237313427654283e-05, + "loss": 0.6134, + "step": 1578 + }, + { + "epoch": 1.2249806051202483, + "grad_norm": 7.858805179595947, + "learning_rate": 1.5210529124744557e-05, + "loss": 0.4049, + "step": 1579 + }, + { + "epoch": 1.225756400310318, + "grad_norm": 6.566247463226318, + "learning_rate": 1.5185332141879656e-05, + "loss": 0.3539, + "step": 1580 + }, + { + "epoch": 1.2265321955003878, + "grad_norm": 7.20612907409668, + "learning_rate": 1.5161726564640554e-05, + "loss": 0.3656, + "step": 1581 + }, + { + "epoch": 1.2273079906904578, + "grad_norm": 7.947671413421631, + "learning_rate": 1.5139716220568754e-05, + "loss": 0.2911, + "step": 1582 + }, + { + "epoch": 1.2280837858805276, + "grad_norm": 5.122169494628906, + "learning_rate": 1.5119304678545621e-05, + "loss": 0.1704, + "step": 1583 + }, + { + "epoch": 1.2288595810705973, + "grad_norm": 7.501595497131348, + "learning_rate": 1.5100495248213759e-05, + "loss": 0.4023, + "step": 1584 + }, + { + "epoch": 1.229635376260667, + "grad_norm": 6.04946231842041, + "learning_rate": 1.5083290979440336e-05, + "loss": 0.3373, + "step": 1585 + }, + { + "epoch": 1.230411171450737, + "grad_norm": 14.541313171386719, + "learning_rate": 1.5067694661822558e-05, + "loss": 2.163, + "step": 1586 + }, + { + "epoch": 1.2311869666408068, + "grad_norm": 5.8007049560546875, + "learning_rate": 1.5053708824235372e-05, + "loss": 0.2335, + "step": 1587 + }, + { + "epoch": 1.2319627618308766, + "grad_norm": 7.3891682624816895, + "learning_rate": 1.5041335734421398e-05, + "loss": 0.6351, + "step": 1588 + }, + { + "epoch": 1.2327385570209466, + "grad_norm": 8.230656623840332, + "learning_rate": 1.5030577398623234e-05, + "loss": 0.5066, + "step": 1589 + }, + { + "epoch": 1.2335143522110164, + "grad_norm": 11.284153938293457, + "learning_rate": 1.5021435561258152e-05, + "loss": 0.8229, + "step": 1590 + }, + { + "epoch": 1.2342901474010861, + "grad_norm": 7.5284423828125, + "learning_rate": 1.5013911704635239e-05, + "loss": 0.3195, + "step": 1591 + }, + { + "epoch": 1.2350659425911559, + "grad_norm": 0.373605877161026, + "learning_rate": 1.5008007048715069e-05, + "loss": 0.0061, + "step": 1592 + }, + { + "epoch": 1.2358417377812256, + "grad_norm": 11.610506057739258, + "learning_rate": 1.5003722550911864e-05, + "loss": 1.2257, + "step": 1593 + }, + { + "epoch": 1.2366175329712956, + "grad_norm": 6.2346930503845215, + "learning_rate": 1.5001058905938278e-05, + "loss": 0.293, + "step": 1594 + }, + { + "epoch": 1.2373933281613654, + "grad_norm": 0.7238097190856934, + "learning_rate": 1.5000016545692732e-05, + "loss": 0.0133, + "step": 1595 + }, + { + "epoch": 1.2381691233514351, + "grad_norm": 6.825682640075684, + "learning_rate": 1.50005956391894e-05, + "loss": 0.5551, + "step": 1596 + }, + { + "epoch": 1.2389449185415051, + "grad_norm": 1.8489340543746948, + "learning_rate": 1.5002796092530794e-05, + "loss": 0.0482, + "step": 1597 + }, + { + "epoch": 1.2397207137315749, + "grad_norm": 4.995535373687744, + "learning_rate": 1.5006617548922994e-05, + "loss": 0.1374, + "step": 1598 + }, + { + "epoch": 1.2404965089216446, + "grad_norm": 7.272418975830078, + "learning_rate": 1.5012059388733499e-05, + "loss": 0.2916, + "step": 1599 + }, + { + "epoch": 1.2412723041117144, + "grad_norm": 6.116049766540527, + "learning_rate": 1.5019120729591695e-05, + "loss": 0.2388, + "step": 1600 + }, + { + "epoch": 1.2420480993017844, + "grad_norm": 9.95783805847168, + "learning_rate": 1.5027800426531933e-05, + "loss": 0.6672, + "step": 1601 + }, + { + "epoch": 1.2428238944918542, + "grad_norm": 9.642383575439453, + "learning_rate": 1.5038097072179175e-05, + "loss": 0.4564, + "step": 1602 + }, + { + "epoch": 1.243599689681924, + "grad_norm": 8.286896705627441, + "learning_rate": 1.5050008996977195e-05, + "loss": 0.3358, + "step": 1603 + }, + { + "epoch": 1.244375484871994, + "grad_norm": 0.6107368469238281, + "learning_rate": 1.5063534269459296e-05, + "loss": 0.0297, + "step": 1604 + }, + { + "epoch": 1.2451512800620637, + "grad_norm": 6.165337562561035, + "learning_rate": 1.507867069656148e-05, + "loss": 0.1815, + "step": 1605 + }, + { + "epoch": 1.2459270752521334, + "grad_norm": 3.738879919052124, + "learning_rate": 1.509541582397805e-05, + "loss": 0.1539, + "step": 1606 + }, + { + "epoch": 1.2467028704422032, + "grad_norm": 5.302562236785889, + "learning_rate": 1.511376693655956e-05, + "loss": 0.1314, + "step": 1607 + }, + { + "epoch": 1.2474786656322732, + "grad_norm": 7.842299461364746, + "learning_rate": 1.5133721058753075e-05, + "loss": 0.2804, + "step": 1608 + }, + { + "epoch": 1.248254460822343, + "grad_norm": 7.157866954803467, + "learning_rate": 1.5155274955084625e-05, + "loss": 0.2451, + "step": 1609 + }, + { + "epoch": 1.2490302560124127, + "grad_norm": 12.686273574829102, + "learning_rate": 1.5178425130683847e-05, + "loss": 0.7975, + "step": 1610 + }, + { + "epoch": 1.2498060512024824, + "grad_norm": 11.311165809631348, + "learning_rate": 1.5203167831850639e-05, + "loss": 0.7371, + "step": 1611 + }, + { + "epoch": 1.2505818463925524, + "grad_norm": 9.781121253967285, + "learning_rate": 1.5229499046663827e-05, + "loss": 0.6379, + "step": 1612 + }, + { + "epoch": 1.2513576415826222, + "grad_norm": 8.047581672668457, + "learning_rate": 1.5257414505631661e-05, + "loss": 0.2899, + "step": 1613 + }, + { + "epoch": 1.252133436772692, + "grad_norm": 7.55552339553833, + "learning_rate": 1.5286909682384124e-05, + "loss": 0.4057, + "step": 1614 + }, + { + "epoch": 1.2529092319627617, + "grad_norm": 4.604458332061768, + "learning_rate": 1.5317979794406815e-05, + "loss": 0.1344, + "step": 1615 + }, + { + "epoch": 1.2536850271528317, + "grad_norm": 9.611407279968262, + "learning_rate": 1.535061980381647e-05, + "loss": 0.4448, + "step": 1616 + }, + { + "epoch": 1.2544608223429015, + "grad_norm": 11.091018676757812, + "learning_rate": 1.538482441817779e-05, + "loss": 0.7411, + "step": 1617 + }, + { + "epoch": 1.2552366175329712, + "grad_norm": 12.149458885192871, + "learning_rate": 1.5420588091361587e-05, + "loss": 0.7403, + "step": 1618 + }, + { + "epoch": 1.2560124127230412, + "grad_norm": 5.4338483810424805, + "learning_rate": 1.545790502444412e-05, + "loss": 0.1618, + "step": 1619 + }, + { + "epoch": 1.256788207913111, + "grad_norm": 1.1224780082702637, + "learning_rate": 1.5496769166647274e-05, + "loss": 0.0279, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_Qnli-dev_cosine_accuracy": 0.66796875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.7912746667861938, + "eval_Qnli-dev_cosine_ap": 0.7269333141698721, + "eval_Qnli-dev_cosine_f1": 0.6846543001686342, + "eval_Qnli-dev_cosine_f1_threshold": 0.7534282207489014, + "eval_Qnli-dev_cosine_precision": 0.5686274509803921, + "eval_Qnli-dev_cosine_recall": 0.8601694915254238, + "eval_Qnli-dev_dot_accuracy": 0.6328125, + "eval_Qnli-dev_dot_accuracy_threshold": 396.85113525390625, + "eval_Qnli-dev_dot_ap": 0.6127065597478176, + "eval_Qnli-dev_dot_f1": 0.6578171091445428, + "eval_Qnli-dev_dot_f1_threshold": 312.68878173828125, + "eval_Qnli-dev_dot_precision": 0.504524886877828, + "eval_Qnli-dev_dot_recall": 0.9449152542372882, + "eval_Qnli-dev_euclidean_accuracy": 0.689453125, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.999319076538086, + "eval_Qnli-dev_euclidean_ap": 0.7298614517636469, + "eval_Qnli-dev_euclidean_f1": 0.6860254083484574, + "eval_Qnli-dev_euclidean_f1_threshold": 14.55683708190918, + "eval_Qnli-dev_euclidean_precision": 0.6, + "eval_Qnli-dev_euclidean_recall": 0.8008474576271186, + "eval_Qnli-dev_manhattan_accuracy": 0.67578125, + "eval_Qnli-dev_manhattan_accuracy_threshold": 227.7866668701172, + "eval_Qnli-dev_manhattan_ap": 0.726705949328858, + "eval_Qnli-dev_manhattan_f1": 0.6830985915492958, + "eval_Qnli-dev_manhattan_f1_threshold": 273.657470703125, + "eval_Qnli-dev_manhattan_precision": 0.5843373493975904, + "eval_Qnli-dev_manhattan_recall": 0.8220338983050848, + "eval_Qnli-dev_max_accuracy": 0.689453125, + "eval_Qnli-dev_max_accuracy_threshold": 396.85113525390625, + "eval_Qnli-dev_max_ap": 0.7298614517636469, + "eval_Qnli-dev_max_f1": 0.6860254083484574, + "eval_Qnli-dev_max_f1_threshold": 312.68878173828125, + "eval_Qnli-dev_max_precision": 0.6, + "eval_Qnli-dev_max_recall": 0.9449152542372882, + "eval_allNLI-dev_cosine_accuracy": 0.720703125, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8801639080047607, + "eval_allNLI-dev_cosine_ap": 0.6013057434148124, + "eval_allNLI-dev_cosine_f1": 0.6150341685649203, + "eval_allNLI-dev_cosine_f1_threshold": 0.8005305528640747, + "eval_allNLI-dev_cosine_precision": 0.5075187969924813, + "eval_allNLI-dev_cosine_recall": 0.7803468208092486, + "eval_allNLI-dev_dot_accuracy": 0.6796875, + "eval_allNLI-dev_dot_accuracy_threshold": 406.8502197265625, + "eval_allNLI-dev_dot_ap": 0.5047656241371711, + "eval_allNLI-dev_dot_f1": 0.5961123110151189, + "eval_allNLI-dev_dot_f1_threshold": 335.1709899902344, + "eval_allNLI-dev_dot_precision": 0.47586206896551725, + "eval_allNLI-dev_dot_recall": 0.7976878612716763, + "eval_allNLI-dev_euclidean_accuracy": 0.724609375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.433222770690918, + "eval_allNLI-dev_euclidean_ap": 0.6030010010522653, + "eval_allNLI-dev_euclidean_f1": 0.6134969325153374, + "eval_allNLI-dev_euclidean_f1_threshold": 14.498795509338379, + "eval_allNLI-dev_euclidean_precision": 0.47468354430379744, + "eval_allNLI-dev_euclidean_recall": 0.8670520231213873, + "eval_allNLI-dev_manhattan_accuracy": 0.724609375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 200.98912048339844, + "eval_allNLI-dev_manhattan_ap": 0.6035152854209352, + "eval_allNLI-dev_manhattan_f1": 0.6247086247086246, + "eval_allNLI-dev_manhattan_f1_threshold": 237.9049072265625, + "eval_allNLI-dev_manhattan_precision": 0.5234375, + "eval_allNLI-dev_manhattan_recall": 0.7745664739884393, + "eval_allNLI-dev_max_accuracy": 0.724609375, + "eval_allNLI-dev_max_accuracy_threshold": 406.8502197265625, + "eval_allNLI-dev_max_ap": 0.6035152854209352, + "eval_allNLI-dev_max_f1": 0.6247086247086246, + "eval_allNLI-dev_max_f1_threshold": 335.1709899902344, + "eval_allNLI-dev_max_precision": 0.5234375, + "eval_allNLI-dev_max_recall": 0.8670520231213873, + "eval_sequential_score": 0.7298614517636469, + "eval_sts-test_pearson_cosine": 0.8262407698572868, + "eval_sts-test_pearson_dot": 0.7804964522349043, + "eval_sts-test_pearson_euclidean": 0.8647358778737688, + "eval_sts-test_pearson_manhattan": 0.8608486161981896, + "eval_sts-test_pearson_max": 0.8647358778737688, + "eval_sts-test_spearman_cosine": 0.8698426190127764, + "eval_sts-test_spearman_dot": 0.7725837628543586, + "eval_sts-test_spearman_euclidean": 0.8678749060493964, + "eval_sts-test_spearman_manhattan": 0.8645596878542602, + "eval_sts-test_spearman_max": 0.8698426190127764, + "eval_vitaminc-pairs_loss": 3.8826472759246826, + "eval_vitaminc-pairs_runtime": 5.4263, + "eval_vitaminc-pairs_samples_per_second": 23.589, + "eval_vitaminc-pairs_steps_per_second": 0.369, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_negation-triplets_loss": 1.425034523010254, + "eval_negation-triplets_runtime": 1.1135, + "eval_negation-triplets_samples_per_second": 114.949, + "eval_negation-triplets_steps_per_second": 1.796, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_scitail-pairs-pos_loss": 0.13596965372562408, + "eval_scitail-pairs-pos_runtime": 1.2164, + "eval_scitail-pairs-pos_samples_per_second": 105.227, + "eval_scitail-pairs-pos_steps_per_second": 1.644, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_scitail-pairs-qa_loss": 0.0026676501147449017, + "eval_scitail-pairs-qa_runtime": 0.9035, + "eval_scitail-pairs-qa_samples_per_second": 141.664, + "eval_scitail-pairs-qa_steps_per_second": 2.214, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_xsum-pairs_loss": 0.12921713292598724, + "eval_xsum-pairs_runtime": 4.2986, + "eval_xsum-pairs_samples_per_second": 29.777, + "eval_xsum-pairs_steps_per_second": 0.465, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_sciq_pairs_loss": 0.06194199621677399, + "eval_sciq_pairs_runtime": 5.5068, + "eval_sciq_pairs_samples_per_second": 23.244, + "eval_sciq_pairs_steps_per_second": 0.363, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_qasc_pairs_loss": 0.1831199675798416, + "eval_qasc_pairs_runtime": 0.9722, + "eval_qasc_pairs_samples_per_second": 131.655, + "eval_qasc_pairs_steps_per_second": 2.057, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_openbookqa_pairs_loss": 1.0066652297973633, + "eval_openbookqa_pairs_runtime": 0.8561, + "eval_openbookqa_pairs_samples_per_second": 149.516, + "eval_openbookqa_pairs_steps_per_second": 2.336, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_msmarco_pairs_loss": 0.44056645035743713, + "eval_msmarco_pairs_runtime": 1.9451, + "eval_msmarco_pairs_samples_per_second": 65.805, + "eval_msmarco_pairs_steps_per_second": 1.028, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_nq_pairs_loss": 0.6642189025878906, + "eval_nq_pairs_runtime": 4.3914, + "eval_nq_pairs_samples_per_second": 29.148, + "eval_nq_pairs_steps_per_second": 0.455, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_trivia_pairs_loss": 0.5387365818023682, + "eval_trivia_pairs_runtime": 4.4965, + "eval_trivia_pairs_samples_per_second": 28.467, + "eval_trivia_pairs_steps_per_second": 0.445, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_gooaq_pairs_loss": 0.5017899870872498, + "eval_gooaq_pairs_runtime": 1.2594, + "eval_gooaq_pairs_samples_per_second": 101.639, + "eval_gooaq_pairs_steps_per_second": 1.588, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_paws-pos_loss": 0.032512955367565155, + "eval_paws-pos_runtime": 1.021, + "eval_paws-pos_samples_per_second": 125.372, + "eval_paws-pos_steps_per_second": 1.959, + "step": 1620 + }, + { + "epoch": 1.256788207913111, + "eval_global_dataset_loss": 0.4466595947742462, + "eval_global_dataset_runtime": 21.0684, + "eval_global_dataset_samples_per_second": 19.745, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1620 + }, + { + "epoch": 1.2575640031031807, + "grad_norm": 0.9772026538848877, + "learning_rate": 1.553717421631976e-05, + "loss": 0.0186, + "step": 1621 + }, + { + "epoch": 1.2583397982932505, + "grad_norm": 5.760748863220215, + "learning_rate": 1.5579113621958828e-05, + "loss": 0.1856, + "step": 1622 + }, + { + "epoch": 1.2591155934833205, + "grad_norm": 8.94067668914795, + "learning_rate": 1.5622580583272607e-05, + "loss": 0.7787, + "step": 1623 + }, + { + "epoch": 1.2598913886733902, + "grad_norm": 5.811945915222168, + "learning_rate": 1.566756805228273e-05, + "loss": 0.2493, + "step": 1624 + }, + { + "epoch": 1.26066718386346, + "grad_norm": 2.895319938659668, + "learning_rate": 1.5714068734467125e-05, + "loss": 0.0744, + "step": 1625 + }, + { + "epoch": 1.26144297905353, + "grad_norm": 5.792057514190674, + "learning_rate": 1.5762075089942804e-05, + "loss": 0.2839, + "step": 1626 + }, + { + "epoch": 1.2622187742435997, + "grad_norm": 9.008331298828125, + "learning_rate": 1.581157933468842e-05, + "loss": 0.5277, + "step": 1627 + }, + { + "epoch": 1.2629945694336695, + "grad_norm": 0.6229547262191772, + "learning_rate": 1.586257344180639e-05, + "loss": 0.0128, + "step": 1628 + }, + { + "epoch": 1.2637703646237393, + "grad_norm": 9.951301574707031, + "learning_rate": 1.5915049142824463e-05, + "loss": 0.4677, + "step": 1629 + }, + { + "epoch": 1.264546159813809, + "grad_norm": 6.00922966003418, + "learning_rate": 1.5968997929036376e-05, + "loss": 0.1969, + "step": 1630 + }, + { + "epoch": 1.265321955003879, + "grad_norm": 1.3203481435775757, + "learning_rate": 1.6024411052881526e-05, + "loss": 0.0324, + "step": 1631 + }, + { + "epoch": 1.2660977501939488, + "grad_norm": 0.5914216041564941, + "learning_rate": 1.608127952936334e-05, + "loss": 0.0315, + "step": 1632 + }, + { + "epoch": 1.2668735453840188, + "grad_norm": 3.9921061992645264, + "learning_rate": 1.613959413750618e-05, + "loss": 0.0982, + "step": 1633 + }, + { + "epoch": 1.2676493405740885, + "grad_norm": 7.621471881866455, + "learning_rate": 1.6199345421850413e-05, + "loss": 0.4068, + "step": 1634 + }, + { + "epoch": 1.2684251357641583, + "grad_norm": 11.144283294677734, + "learning_rate": 1.6260523693985672e-05, + "loss": 0.8016, + "step": 1635 + }, + { + "epoch": 1.269200930954228, + "grad_norm": 8.5413818359375, + "learning_rate": 1.6323119034121714e-05, + "loss": 0.3118, + "step": 1636 + }, + { + "epoch": 1.2699767261442978, + "grad_norm": 2.7264978885650635, + "learning_rate": 1.638712129269687e-05, + "loss": 0.0459, + "step": 1637 + }, + { + "epoch": 1.2707525213343678, + "grad_norm": 0.4831717610359192, + "learning_rate": 1.645252009202383e-05, + "loss": 0.0275, + "step": 1638 + }, + { + "epoch": 1.2715283165244375, + "grad_norm": 6.617738246917725, + "learning_rate": 1.6519304827972236e-05, + "loss": 0.2523, + "step": 1639 + }, + { + "epoch": 1.2723041117145073, + "grad_norm": 5.507478713989258, + "learning_rate": 1.6587464671688184e-05, + "loss": 0.2622, + "step": 1640 + }, + { + "epoch": 1.2730799069045773, + "grad_norm": 6.155742168426514, + "learning_rate": 1.665698857135002e-05, + "loss": 0.2252, + "step": 1641 + }, + { + "epoch": 1.273855702094647, + "grad_norm": 0.3136428892612457, + "learning_rate": 1.6727865253960356e-05, + "loss": 0.0043, + "step": 1642 + }, + { + "epoch": 1.2746314972847168, + "grad_norm": 2.3330698013305664, + "learning_rate": 1.680008322717396e-05, + "loss": 0.0329, + "step": 1643 + }, + { + "epoch": 1.2754072924747866, + "grad_norm": 10.338220596313477, + "learning_rate": 1.6873630781161163e-05, + "loss": 0.7944, + "step": 1644 + }, + { + "epoch": 1.2761830876648566, + "grad_norm": 4.93920373916626, + "learning_rate": 1.6948495990506564e-05, + "loss": 0.1614, + "step": 1645 + }, + { + "epoch": 1.2769588828549263, + "grad_norm": 6.935882091522217, + "learning_rate": 1.7024666716142698e-05, + "loss": 0.4695, + "step": 1646 + }, + { + "epoch": 1.277734678044996, + "grad_norm": 4.372180461883545, + "learning_rate": 1.7102130607318316e-05, + "loss": 0.2149, + "step": 1647 + }, + { + "epoch": 1.278510473235066, + "grad_norm": 7.639461517333984, + "learning_rate": 1.718087510360101e-05, + "loss": 0.7172, + "step": 1648 + }, + { + "epoch": 1.2792862684251358, + "grad_norm": 3.653212308883667, + "learning_rate": 1.7260887436913842e-05, + "loss": 0.1386, + "step": 1649 + }, + { + "epoch": 1.2800620636152056, + "grad_norm": 9.81125545501709, + "learning_rate": 1.7342154633605614e-05, + "loss": 0.7124, + "step": 1650 + }, + { + "epoch": 1.2808378588052753, + "grad_norm": 0.37764161825180054, + "learning_rate": 1.7424663516554493e-05, + "loss": 0.0055, + "step": 1651 + }, + { + "epoch": 1.281613653995345, + "grad_norm": 0.5111414194107056, + "learning_rate": 1.7508400707304674e-05, + "loss": 0.0094, + "step": 1652 + }, + { + "epoch": 1.282389449185415, + "grad_norm": 6.858648300170898, + "learning_rate": 1.759335262823552e-05, + "loss": 0.2958, + "step": 1653 + }, + { + "epoch": 1.2831652443754848, + "grad_norm": 7.316114902496338, + "learning_rate": 1.7679505504763263e-05, + "loss": 0.3199, + "step": 1654 + }, + { + "epoch": 1.2839410395655548, + "grad_norm": 10.429588317871094, + "learning_rate": 1.7766845367574386e-05, + "loss": 0.5994, + "step": 1655 + }, + { + "epoch": 1.2847168347556246, + "grad_norm": 11.734217643737793, + "learning_rate": 1.7855358054890676e-05, + "loss": 1.0332, + "step": 1656 + }, + { + "epoch": 1.2854926299456944, + "grad_norm": 8.13221549987793, + "learning_rate": 1.794502921476562e-05, + "loss": 0.3709, + "step": 1657 + }, + { + "epoch": 1.2862684251357641, + "grad_norm": 9.168801307678223, + "learning_rate": 1.8035844307411375e-05, + "loss": 0.5758, + "step": 1658 + }, + { + "epoch": 1.2870442203258339, + "grad_norm": 9.61843490600586, + "learning_rate": 1.8127788607556394e-05, + "loss": 0.5492, + "step": 1659 + }, + { + "epoch": 1.2878200155159039, + "grad_norm": 9.393542289733887, + "learning_rate": 1.822084720683305e-05, + "loss": 0.5099, + "step": 1660 + }, + { + "epoch": 1.2885958107059736, + "grad_norm": 7.788226127624512, + "learning_rate": 1.831500501619498e-05, + "loss": 0.3742, + "step": 1661 + }, + { + "epoch": 1.2893716058960434, + "grad_norm": 0.09201373159885406, + "learning_rate": 1.8410246768363686e-05, + "loss": 0.0015, + "step": 1662 + }, + { + "epoch": 1.2901474010861134, + "grad_norm": 8.259096145629883, + "learning_rate": 1.8506557020304067e-05, + "loss": 0.4329, + "step": 1663 + }, + { + "epoch": 1.2909231962761831, + "grad_norm": 7.714040279388428, + "learning_rate": 1.8603920155728445e-05, + "loss": 0.4178, + "step": 1664 + }, + { + "epoch": 1.2916989914662529, + "grad_norm": 11.452858924865723, + "learning_rate": 1.8702320387628674e-05, + "loss": 1.1547, + "step": 1665 + }, + { + "epoch": 1.2924747866563226, + "grad_norm": 9.775408744812012, + "learning_rate": 1.880174176083592e-05, + "loss": 0.6708, + "step": 1666 + }, + { + "epoch": 1.2932505818463926, + "grad_norm": 9.986681938171387, + "learning_rate": 1.8902168154607758e-05, + "loss": 0.7185, + "step": 1667 + }, + { + "epoch": 1.2940263770364624, + "grad_norm": 4.5785017013549805, + "learning_rate": 1.9003583285242034e-05, + "loss": 0.1704, + "step": 1668 + }, + { + "epoch": 1.2948021722265322, + "grad_norm": 6.11545991897583, + "learning_rate": 1.9105970708717242e-05, + "loss": 0.4499, + "step": 1669 + }, + { + "epoch": 1.2955779674166021, + "grad_norm": 4.729278087615967, + "learning_rate": 1.9209313823358798e-05, + "loss": 0.1623, + "step": 1670 + }, + { + "epoch": 1.296353762606672, + "grad_norm": 5.819040775299072, + "learning_rate": 1.9313595872531025e-05, + "loss": 0.2804, + "step": 1671 + }, + { + "epoch": 1.2971295577967417, + "grad_norm": 3.81506609916687, + "learning_rate": 1.9418799947354003e-05, + "loss": 0.1968, + "step": 1672 + }, + { + "epoch": 1.2979053529868114, + "grad_norm": 4.604837894439697, + "learning_rate": 1.9524908989445466e-05, + "loss": 0.189, + "step": 1673 + }, + { + "epoch": 1.2986811481768812, + "grad_norm": 6.6130170822143555, + "learning_rate": 1.963190579368661e-05, + "loss": 0.2606, + "step": 1674 + }, + { + "epoch": 1.2994569433669512, + "grad_norm": 6.8984293937683105, + "learning_rate": 1.973977301101182e-05, + "loss": 0.5032, + "step": 1675 + }, + { + "epoch": 1.300232738557021, + "grad_norm": 10.33646297454834, + "learning_rate": 1.9848493151221885e-05, + "loss": 0.8025, + "step": 1676 + }, + { + "epoch": 1.3010085337470907, + "grad_norm": 4.21944522857666, + "learning_rate": 1.99580485858198e-05, + "loss": 0.1224, + "step": 1677 + }, + { + "epoch": 1.3017843289371607, + "grad_norm": 8.848977088928223, + "learning_rate": 2.0068421550869236e-05, + "loss": 0.5289, + "step": 1678 + }, + { + "epoch": 1.3025601241272304, + "grad_norm": 6.07085657119751, + "learning_rate": 2.0179594149874856e-05, + "loss": 0.2246, + "step": 1679 + }, + { + "epoch": 1.3033359193173002, + "grad_norm": 9.987034797668457, + "learning_rate": 2.029154835668417e-05, + "loss": 0.7786, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_Qnli-dev_cosine_accuracy": 0.6796875, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8678713440895081, + "eval_Qnli-dev_cosine_ap": 0.7335386252015701, + "eval_Qnli-dev_cosine_f1": 0.6856187290969901, + "eval_Qnli-dev_cosine_f1_threshold": 0.7640129327774048, + "eval_Qnli-dev_cosine_precision": 0.5662983425414365, + "eval_Qnli-dev_cosine_recall": 0.8686440677966102, + "eval_Qnli-dev_dot_accuracy": 0.638671875, + "eval_Qnli-dev_dot_accuracy_threshold": 436.5951843261719, + "eval_Qnli-dev_dot_ap": 0.6348585640812956, + "eval_Qnli-dev_dot_f1": 0.6623376623376623, + "eval_Qnli-dev_dot_f1_threshold": 362.1895446777344, + "eval_Qnli-dev_dot_precision": 0.5368421052631579, + "eval_Qnli-dev_dot_recall": 0.864406779661017, + "eval_Qnli-dev_euclidean_accuracy": 0.6875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.163410186767578, + "eval_Qnli-dev_euclidean_ap": 0.7353142597582815, + "eval_Qnli-dev_euclidean_f1": 0.6804123711340205, + "eval_Qnli-dev_euclidean_f1_threshold": 15.134939193725586, + "eval_Qnli-dev_euclidean_precision": 0.5722543352601156, + "eval_Qnli-dev_euclidean_recall": 0.8389830508474576, + "eval_Qnli-dev_manhattan_accuracy": 0.685546875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 207.81422424316406, + "eval_Qnli-dev_manhattan_ap": 0.732996804111243, + "eval_Qnli-dev_manhattan_f1": 0.6921739130434782, + "eval_Qnli-dev_manhattan_f1_threshold": 270.38861083984375, + "eval_Qnli-dev_manhattan_precision": 0.5870206489675516, + "eval_Qnli-dev_manhattan_recall": 0.8432203389830508, + "eval_Qnli-dev_max_accuracy": 0.6875, + "eval_Qnli-dev_max_accuracy_threshold": 436.5951843261719, + "eval_Qnli-dev_max_ap": 0.7353142597582815, + "eval_Qnli-dev_max_f1": 0.6921739130434782, + "eval_Qnli-dev_max_f1_threshold": 362.1895446777344, + "eval_Qnli-dev_max_precision": 0.5870206489675516, + "eval_Qnli-dev_max_recall": 0.8686440677966102, + "eval_allNLI-dev_cosine_accuracy": 0.712890625, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.9018930196762085, + "eval_allNLI-dev_cosine_ap": 0.6012500741760579, + "eval_allNLI-dev_cosine_f1": 0.6030368763557483, + "eval_allNLI-dev_cosine_f1_threshold": 0.789612889289856, + "eval_allNLI-dev_cosine_precision": 0.4826388888888889, + "eval_allNLI-dev_cosine_recall": 0.8034682080924855, + "eval_allNLI-dev_dot_accuracy": 0.6875, + "eval_allNLI-dev_dot_accuracy_threshold": 463.275146484375, + "eval_allNLI-dev_dot_ap": 0.5042959374161171, + "eval_allNLI-dev_dot_f1": 0.5862068965517241, + "eval_allNLI-dev_dot_f1_threshold": 354.99383544921875, + "eval_allNLI-dev_dot_precision": 0.46735395189003437, + "eval_allNLI-dev_dot_recall": 0.7861271676300579, + "eval_allNLI-dev_euclidean_accuracy": 0.71484375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 9.292919158935547, + "eval_allNLI-dev_euclidean_ap": 0.6011340408798549, + "eval_allNLI-dev_euclidean_f1": 0.6124197002141327, + "eval_allNLI-dev_euclidean_f1_threshold": 14.045066833496094, + "eval_allNLI-dev_euclidean_precision": 0.48639455782312924, + "eval_allNLI-dev_euclidean_recall": 0.8265895953757225, + "eval_allNLI-dev_manhattan_accuracy": 0.71484375, + "eval_allNLI-dev_manhattan_accuracy_threshold": 174.9363555908203, + "eval_allNLI-dev_manhattan_ap": 0.5998016759386099, + "eval_allNLI-dev_manhattan_f1": 0.6061946902654868, + "eval_allNLI-dev_manhattan_f1_threshold": 248.59786987304688, + "eval_allNLI-dev_manhattan_precision": 0.4910394265232975, + "eval_allNLI-dev_manhattan_recall": 0.791907514450867, + "eval_allNLI-dev_max_accuracy": 0.71484375, + "eval_allNLI-dev_max_accuracy_threshold": 463.275146484375, + "eval_allNLI-dev_max_ap": 0.6012500741760579, + "eval_allNLI-dev_max_f1": 0.6124197002141327, + "eval_allNLI-dev_max_f1_threshold": 354.99383544921875, + "eval_allNLI-dev_max_precision": 0.4910394265232975, + "eval_allNLI-dev_max_recall": 0.8265895953757225, + "eval_sequential_score": 0.7353142597582815, + "eval_sts-test_pearson_cosine": 0.8265328037382675, + "eval_sts-test_pearson_dot": 0.7723134672506842, + "eval_sts-test_pearson_euclidean": 0.8701273661823152, + "eval_sts-test_pearson_manhattan": 0.8676302226367159, + "eval_sts-test_pearson_max": 0.8701273661823152, + "eval_sts-test_spearman_cosine": 0.8743569920081566, + "eval_sts-test_spearman_dot": 0.7581232877609555, + "eval_sts-test_spearman_euclidean": 0.8735102533913699, + "eval_sts-test_spearman_manhattan": 0.8717104471607307, + "eval_sts-test_spearman_max": 0.8743569920081566, + "eval_vitaminc-pairs_loss": 3.5457091331481934, + "eval_vitaminc-pairs_runtime": 5.4144, + "eval_vitaminc-pairs_samples_per_second": 23.64, + "eval_vitaminc-pairs_steps_per_second": 0.369, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_negation-triplets_loss": 1.3875781297683716, + "eval_negation-triplets_runtime": 1.0947, + "eval_negation-triplets_samples_per_second": 116.925, + "eval_negation-triplets_steps_per_second": 1.827, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_scitail-pairs-pos_loss": 0.11780296266078949, + "eval_scitail-pairs-pos_runtime": 1.2372, + "eval_scitail-pairs-pos_samples_per_second": 103.458, + "eval_scitail-pairs-pos_steps_per_second": 1.617, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_scitail-pairs-qa_loss": 0.0030645374208688736, + "eval_scitail-pairs-qa_runtime": 0.8964, + "eval_scitail-pairs-qa_samples_per_second": 142.797, + "eval_scitail-pairs-qa_steps_per_second": 2.231, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_xsum-pairs_loss": 0.13688111305236816, + "eval_xsum-pairs_runtime": 4.3014, + "eval_xsum-pairs_samples_per_second": 29.757, + "eval_xsum-pairs_steps_per_second": 0.465, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_sciq_pairs_loss": 0.0627945214509964, + "eval_sciq_pairs_runtime": 5.5115, + "eval_sciq_pairs_samples_per_second": 23.224, + "eval_sciq_pairs_steps_per_second": 0.363, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_qasc_pairs_loss": 0.22918246686458588, + "eval_qasc_pairs_runtime": 0.9673, + "eval_qasc_pairs_samples_per_second": 132.321, + "eval_qasc_pairs_steps_per_second": 2.068, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_openbookqa_pairs_loss": 1.0260117053985596, + "eval_openbookqa_pairs_runtime": 0.8497, + "eval_openbookqa_pairs_samples_per_second": 150.643, + "eval_openbookqa_pairs_steps_per_second": 2.354, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_msmarco_pairs_loss": 0.4491913616657257, + "eval_msmarco_pairs_runtime": 1.9459, + "eval_msmarco_pairs_samples_per_second": 65.78, + "eval_msmarco_pairs_steps_per_second": 1.028, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_nq_pairs_loss": 0.6502649784088135, + "eval_nq_pairs_runtime": 4.3914, + "eval_nq_pairs_samples_per_second": 29.148, + "eval_nq_pairs_steps_per_second": 0.455, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_trivia_pairs_loss": 0.5434364080429077, + "eval_trivia_pairs_runtime": 4.5134, + "eval_trivia_pairs_samples_per_second": 28.36, + "eval_trivia_pairs_steps_per_second": 0.443, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_gooaq_pairs_loss": 0.49640563130378723, + "eval_gooaq_pairs_runtime": 1.2803, + "eval_gooaq_pairs_samples_per_second": 99.977, + "eval_gooaq_pairs_steps_per_second": 1.562, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_paws-pos_loss": 0.033405404537916183, + "eval_paws-pos_runtime": 1.0131, + "eval_paws-pos_samples_per_second": 126.346, + "eval_paws-pos_steps_per_second": 1.974, + "step": 1680 + }, + { + "epoch": 1.3033359193173002, + "eval_global_dataset_loss": 0.3779243528842926, + "eval_global_dataset_runtime": 21.0726, + "eval_global_dataset_samples_per_second": 19.741, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1680 + }, + { + "epoch": 1.30411171450737, + "grad_norm": 4.372567176818848, + "learning_rate": 2.0404266018410384e-05, + "loss": 0.1489, + "step": 1681 + }, + { + "epoch": 1.30488750969744, + "grad_norm": 7.9098734855651855, + "learning_rate": 2.0517728858375808e-05, + "loss": 0.4367, + "step": 1682 + }, + { + "epoch": 1.3056633048875097, + "grad_norm": 5.099017143249512, + "learning_rate": 2.0631918479075364e-05, + "loss": 0.2141, + "step": 1683 + }, + { + "epoch": 1.3064391000775795, + "grad_norm": 12.50365161895752, + "learning_rate": 2.0746816365159637e-05, + "loss": 1.8984, + "step": 1684 + }, + { + "epoch": 1.3072148952676494, + "grad_norm": 10.399861335754395, + "learning_rate": 2.086240388643706e-05, + "loss": 0.7923, + "step": 1685 + }, + { + "epoch": 1.3079906904577192, + "grad_norm": 0.4755062460899353, + "learning_rate": 2.097866230089473e-05, + "loss": 0.0053, + "step": 1686 + }, + { + "epoch": 1.308766485647789, + "grad_norm": 9.701809883117676, + "learning_rate": 2.109557275773733e-05, + "loss": 0.7178, + "step": 1687 + }, + { + "epoch": 1.3095422808378587, + "grad_norm": 0.819050133228302, + "learning_rate": 2.121311630044376e-05, + "loss": 0.0149, + "step": 1688 + }, + { + "epoch": 1.3103180760279285, + "grad_norm": 9.664173126220703, + "learning_rate": 2.133127386984071e-05, + "loss": 0.7261, + "step": 1689 + }, + { + "epoch": 1.3110938712179985, + "grad_norm": 10.702370643615723, + "learning_rate": 2.1450026307193226e-05, + "loss": 0.7316, + "step": 1690 + }, + { + "epoch": 1.3118696664080682, + "grad_norm": 3.725966215133667, + "learning_rate": 2.1569354357311018e-05, + "loss": 0.1864, + "step": 1691 + }, + { + "epoch": 1.3126454615981382, + "grad_norm": 0.31564074754714966, + "learning_rate": 2.168923867167072e-05, + "loss": 0.0042, + "step": 1692 + }, + { + "epoch": 1.313421256788208, + "grad_norm": 6.61835241317749, + "learning_rate": 2.180965981155322e-05, + "loss": 0.3548, + "step": 1693 + }, + { + "epoch": 1.3141970519782777, + "grad_norm": 4.3647356033325195, + "learning_rate": 2.1930598251195336e-05, + "loss": 0.131, + "step": 1694 + }, + { + "epoch": 1.3149728471683475, + "grad_norm": 4.319633483886719, + "learning_rate": 2.205203438095613e-05, + "loss": 0.1264, + "step": 1695 + }, + { + "epoch": 1.3157486423584173, + "grad_norm": 4.182954788208008, + "learning_rate": 2.2173948510496297e-05, + "loss": 0.1095, + "step": 1696 + }, + { + "epoch": 1.3165244375484872, + "grad_norm": 3.630993366241455, + "learning_rate": 2.2296320871970976e-05, + "loss": 0.1765, + "step": 1697 + }, + { + "epoch": 1.317300232738557, + "grad_norm": 10.283700942993164, + "learning_rate": 2.241913162323499e-05, + "loss": 0.5877, + "step": 1698 + }, + { + "epoch": 1.3180760279286268, + "grad_norm": 13.653228759765625, + "learning_rate": 2.254236085106017e-05, + "loss": 1.853, + "step": 1699 + }, + { + "epoch": 1.3188518231186968, + "grad_norm": 6.753418922424316, + "learning_rate": 2.26659885743642e-05, + "loss": 0.2694, + "step": 1700 + }, + { + "epoch": 1.3196276183087665, + "grad_norm": 4.196700572967529, + "learning_rate": 2.2789994747450442e-05, + "loss": 0.1205, + "step": 1701 + }, + { + "epoch": 1.3204034134988363, + "grad_norm": 6.408318519592285, + "learning_rate": 2.291435926325829e-05, + "loss": 0.2836, + "step": 1702 + }, + { + "epoch": 1.321179208688906, + "grad_norm": 10.550044059753418, + "learning_rate": 2.3039061956623433e-05, + "loss": 0.6221, + "step": 1703 + }, + { + "epoch": 1.321955003878976, + "grad_norm": 4.681580543518066, + "learning_rate": 2.316408260754752e-05, + "loss": 0.1223, + "step": 1704 + }, + { + "epoch": 1.3227307990690458, + "grad_norm": 1.2652360200881958, + "learning_rate": 2.3289400944476794e-05, + "loss": 0.0703, + "step": 1705 + }, + { + "epoch": 1.3235065942591155, + "grad_norm": 5.05535888671875, + "learning_rate": 2.3414996647588968e-05, + "loss": 0.1347, + "step": 1706 + }, + { + "epoch": 1.3242823894491855, + "grad_norm": 0.4351763427257538, + "learning_rate": 2.3540849352088127e-05, + "loss": 0.0044, + "step": 1707 + }, + { + "epoch": 1.3250581846392553, + "grad_norm": 6.179965019226074, + "learning_rate": 2.366693865150656e-05, + "loss": 0.2956, + "step": 1708 + }, + { + "epoch": 1.325833979829325, + "grad_norm": 8.875202178955078, + "learning_rate": 2.3793244101013878e-05, + "loss": 0.4716, + "step": 1709 + }, + { + "epoch": 1.3266097750193948, + "grad_norm": 7.6536970138549805, + "learning_rate": 2.3919745220731762e-05, + "loss": 0.4546, + "step": 1710 + }, + { + "epoch": 1.3273855702094646, + "grad_norm": 5.391899585723877, + "learning_rate": 2.404642149905488e-05, + "loss": 0.2143, + "step": 1711 + }, + { + "epoch": 1.3281613653995346, + "grad_norm": 3.491594076156616, + "learning_rate": 2.4173252395976746e-05, + "loss": 0.1181, + "step": 1712 + }, + { + "epoch": 1.3289371605896043, + "grad_norm": 1.023395299911499, + "learning_rate": 2.4300217346420017e-05, + "loss": 0.0194, + "step": 1713 + }, + { + "epoch": 1.3297129557796743, + "grad_norm": 11.769171714782715, + "learning_rate": 2.4427295763571286e-05, + "loss": 0.8112, + "step": 1714 + }, + { + "epoch": 1.330488750969744, + "grad_norm": 5.259401798248291, + "learning_rate": 2.4554467042218934e-05, + "loss": 0.1689, + "step": 1715 + }, + { + "epoch": 1.3312645461598138, + "grad_norm": 8.039584159851074, + "learning_rate": 2.4681710562094274e-05, + "loss": 0.5726, + "step": 1716 + }, + { + "epoch": 1.3320403413498836, + "grad_norm": 8.772049903869629, + "learning_rate": 2.4809005691215022e-05, + "loss": 0.5521, + "step": 1717 + }, + { + "epoch": 1.3328161365399533, + "grad_norm": 6.204926490783691, + "learning_rate": 2.4936331789230692e-05, + "loss": 0.2334, + "step": 1718 + }, + { + "epoch": 1.3335919317300233, + "grad_norm": 7.46255350112915, + "learning_rate": 2.5063668210769293e-05, + "loss": 0.4507, + "step": 1719 + }, + { + "epoch": 1.334367726920093, + "grad_norm": 8.235262870788574, + "learning_rate": 2.5190994308784967e-05, + "loss": 0.6141, + "step": 1720 + }, + { + "epoch": 1.3351435221101629, + "grad_norm": 3.6687986850738525, + "learning_rate": 2.5318289437905698e-05, + "loss": 0.1189, + "step": 1721 + }, + { + "epoch": 1.3359193173002328, + "grad_norm": 13.651668548583984, + "learning_rate": 2.5445532957781055e-05, + "loss": 1.846, + "step": 1722 + }, + { + "epoch": 1.3366951124903026, + "grad_norm": 12.404130935668945, + "learning_rate": 2.5572704236428703e-05, + "loss": 0.9293, + "step": 1723 + }, + { + "epoch": 1.3374709076803724, + "grad_norm": 8.230925559997559, + "learning_rate": 2.569978265357997e-05, + "loss": 0.4755, + "step": 1724 + }, + { + "epoch": 1.3382467028704421, + "grad_norm": 6.33299446105957, + "learning_rate": 2.5826747604023243e-05, + "loss": 0.2039, + "step": 1725 + }, + { + "epoch": 1.339022498060512, + "grad_norm": 9.209304809570312, + "learning_rate": 2.595357850094509e-05, + "loss": 0.6513, + "step": 1726 + }, + { + "epoch": 1.3397982932505819, + "grad_norm": 9.879910469055176, + "learning_rate": 2.6080254779268244e-05, + "loss": 0.7079, + "step": 1727 + }, + { + "epoch": 1.3405740884406516, + "grad_norm": 0.48255419731140137, + "learning_rate": 2.620675589898613e-05, + "loss": 0.0103, + "step": 1728 + }, + { + "epoch": 1.3413498836307216, + "grad_norm": 10.136865615844727, + "learning_rate": 2.6333061348493428e-05, + "loss": 0.5527, + "step": 1729 + }, + { + "epoch": 1.3421256788207914, + "grad_norm": 4.067371368408203, + "learning_rate": 2.645915064791186e-05, + "loss": 0.1503, + "step": 1730 + }, + { + "epoch": 1.3429014740108611, + "grad_norm": 4.70083475112915, + "learning_rate": 2.6585003352411004e-05, + "loss": 0.1049, + "step": 1731 + }, + { + "epoch": 1.343677269200931, + "grad_norm": 8.811023712158203, + "learning_rate": 2.6710599055523198e-05, + "loss": 0.7692, + "step": 1732 + }, + { + "epoch": 1.3444530643910007, + "grad_norm": 0.2850119471549988, + "learning_rate": 2.6835917392452484e-05, + "loss": 0.0056, + "step": 1733 + }, + { + "epoch": 1.3452288595810706, + "grad_norm": 11.363007545471191, + "learning_rate": 2.6960938043376556e-05, + "loss": 0.8317, + "step": 1734 + }, + { + "epoch": 1.3460046547711404, + "grad_norm": 11.469557762145996, + "learning_rate": 2.70856407367417e-05, + "loss": 0.6516, + "step": 1735 + }, + { + "epoch": 1.3467804499612102, + "grad_norm": 4.096419811248779, + "learning_rate": 2.7210005252549547e-05, + "loss": 0.1432, + "step": 1736 + }, + { + "epoch": 1.3475562451512801, + "grad_norm": 13.826745986938477, + "learning_rate": 2.733401142563577e-05, + "loss": 1.0962, + "step": 1737 + }, + { + "epoch": 1.34833204034135, + "grad_norm": 11.574246406555176, + "learning_rate": 2.7457639148939833e-05, + "loss": 0.5539, + "step": 1738 + }, + { + "epoch": 1.3491078355314197, + "grad_norm": 0.6562228202819824, + "learning_rate": 2.7580868376764996e-05, + "loss": 0.0322, + "step": 1739 + }, + { + "epoch": 1.3498836307214894, + "grad_norm": 6.574079513549805, + "learning_rate": 2.7703679128029013e-05, + "loss": 0.2232, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_Qnli-dev_cosine_accuracy": 0.681640625, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8642492294311523, + "eval_Qnli-dev_cosine_ap": 0.7223345765128643, + "eval_Qnli-dev_cosine_f1": 0.6855123674911661, + "eval_Qnli-dev_cosine_f1_threshold": 0.7717834115028381, + "eval_Qnli-dev_cosine_precision": 0.5878787878787879, + "eval_Qnli-dev_cosine_recall": 0.8220338983050848, + "eval_Qnli-dev_dot_accuracy": 0.63671875, + "eval_Qnli-dev_dot_accuracy_threshold": 382.30712890625, + "eval_Qnli-dev_dot_ap": 0.6191404470418108, + "eval_Qnli-dev_dot_f1": 0.675496688741722, + "eval_Qnli-dev_dot_f1_threshold": 346.96807861328125, + "eval_Qnli-dev_dot_precision": 0.5543478260869565, + "eval_Qnli-dev_dot_recall": 0.864406779661017, + "eval_Qnli-dev_euclidean_accuracy": 0.685546875, + "eval_Qnli-dev_euclidean_accuracy_threshold": 11.836234092712402, + "eval_Qnli-dev_euclidean_ap": 0.7248751235008488, + "eval_Qnli-dev_euclidean_f1": 0.6753246753246753, + "eval_Qnli-dev_euclidean_f1_threshold": 15.757074356079102, + "eval_Qnli-dev_euclidean_precision": 0.5473684210526316, + "eval_Qnli-dev_euclidean_recall": 0.8813559322033898, + "eval_Qnli-dev_manhattan_accuracy": 0.6796875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 206.25564575195312, + "eval_Qnli-dev_manhattan_ap": 0.7230975395076378, + "eval_Qnli-dev_manhattan_f1": 0.6779661016949152, + "eval_Qnli-dev_manhattan_f1_threshold": 279.01287841796875, + "eval_Qnli-dev_manhattan_precision": 0.5649717514124294, + "eval_Qnli-dev_manhattan_recall": 0.847457627118644, + "eval_Qnli-dev_max_accuracy": 0.685546875, + "eval_Qnli-dev_max_accuracy_threshold": 382.30712890625, + "eval_Qnli-dev_max_ap": 0.7248751235008488, + "eval_Qnli-dev_max_f1": 0.6855123674911661, + "eval_Qnli-dev_max_f1_threshold": 346.96807861328125, + "eval_Qnli-dev_max_precision": 0.5878787878787879, + "eval_Qnli-dev_max_recall": 0.8813559322033898, + "eval_allNLI-dev_cosine_accuracy": 0.7265625, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8744198679924011, + "eval_allNLI-dev_cosine_ap": 0.5998756685595877, + "eval_allNLI-dev_cosine_f1": 0.6106194690265486, + "eval_allNLI-dev_cosine_f1_threshold": 0.7878724336624146, + "eval_allNLI-dev_cosine_precision": 0.4946236559139785, + "eval_allNLI-dev_cosine_recall": 0.7976878612716763, + "eval_allNLI-dev_dot_accuracy": 0.68359375, + "eval_allNLI-dev_dot_accuracy_threshold": 440.6682434082031, + "eval_allNLI-dev_dot_ap": 0.5004578340237119, + "eval_allNLI-dev_dot_f1": 0.5892116182572614, + "eval_allNLI-dev_dot_f1_threshold": 332.84967041015625, + "eval_allNLI-dev_dot_precision": 0.459546925566343, + "eval_allNLI-dev_dot_recall": 0.8208092485549133, + "eval_allNLI-dev_euclidean_accuracy": 0.720703125, + "eval_allNLI-dev_euclidean_accuracy_threshold": 10.145851135253906, + "eval_allNLI-dev_euclidean_ap": 0.6020953375600313, + "eval_allNLI-dev_euclidean_f1": 0.6131687242798355, + "eval_allNLI-dev_euclidean_f1_threshold": 14.487386703491211, + "eval_allNLI-dev_euclidean_precision": 0.476038338658147, + "eval_allNLI-dev_euclidean_recall": 0.861271676300578, + "eval_allNLI-dev_manhattan_accuracy": 0.728515625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 189.596923828125, + "eval_allNLI-dev_manhattan_ap": 0.5994543245097959, + "eval_allNLI-dev_manhattan_f1": 0.6095717884130981, + "eval_allNLI-dev_manhattan_f1_threshold": 226.64190673828125, + "eval_allNLI-dev_manhattan_precision": 0.5401785714285714, + "eval_allNLI-dev_manhattan_recall": 0.6994219653179191, + "eval_allNLI-dev_max_accuracy": 0.728515625, + "eval_allNLI-dev_max_accuracy_threshold": 440.6682434082031, + "eval_allNLI-dev_max_ap": 0.6020953375600313, + "eval_allNLI-dev_max_f1": 0.6131687242798355, + "eval_allNLI-dev_max_f1_threshold": 332.84967041015625, + "eval_allNLI-dev_max_precision": 0.5401785714285714, + "eval_allNLI-dev_max_recall": 0.861271676300578, + "eval_sequential_score": 0.7248751235008488, + "eval_sts-test_pearson_cosine": 0.8192530568447666, + "eval_sts-test_pearson_dot": 0.7815830596765242, + "eval_sts-test_pearson_euclidean": 0.864450059424778, + "eval_sts-test_pearson_manhattan": 0.8627896607993385, + "eval_sts-test_pearson_max": 0.864450059424778, + "eval_sts-test_spearman_cosine": 0.8724459788929596, + "eval_sts-test_spearman_dot": 0.7754067619254509, + "eval_sts-test_spearman_euclidean": 0.8698076235263682, + "eval_sts-test_spearman_manhattan": 0.8673378909126997, + "eval_sts-test_spearman_max": 0.8724459788929596, + "eval_vitaminc-pairs_loss": 3.561216115951538, + "eval_vitaminc-pairs_runtime": 5.4277, + "eval_vitaminc-pairs_samples_per_second": 23.583, + "eval_vitaminc-pairs_steps_per_second": 0.368, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_negation-triplets_loss": 1.4608488082885742, + "eval_negation-triplets_runtime": 1.0951, + "eval_negation-triplets_samples_per_second": 116.889, + "eval_negation-triplets_steps_per_second": 1.826, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_scitail-pairs-pos_loss": 0.09265163540840149, + "eval_scitail-pairs-pos_runtime": 1.2069, + "eval_scitail-pairs-pos_samples_per_second": 106.059, + "eval_scitail-pairs-pos_steps_per_second": 1.657, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_scitail-pairs-qa_loss": 0.0019270963966846466, + "eval_scitail-pairs-qa_runtime": 0.9024, + "eval_scitail-pairs-qa_samples_per_second": 141.851, + "eval_scitail-pairs-qa_steps_per_second": 2.216, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_xsum-pairs_loss": 0.20208881795406342, + "eval_xsum-pairs_runtime": 4.2969, + "eval_xsum-pairs_samples_per_second": 29.789, + "eval_xsum-pairs_steps_per_second": 0.465, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_sciq_pairs_loss": 0.07007329910993576, + "eval_sciq_pairs_runtime": 5.5158, + "eval_sciq_pairs_samples_per_second": 23.206, + "eval_sciq_pairs_steps_per_second": 0.363, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_qasc_pairs_loss": 0.1795005351305008, + "eval_qasc_pairs_runtime": 0.9715, + "eval_qasc_pairs_samples_per_second": 131.761, + "eval_qasc_pairs_steps_per_second": 2.059, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_openbookqa_pairs_loss": 1.02101731300354, + "eval_openbookqa_pairs_runtime": 0.8497, + "eval_openbookqa_pairs_samples_per_second": 150.64, + "eval_openbookqa_pairs_steps_per_second": 2.354, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_msmarco_pairs_loss": 0.561587393283844, + "eval_msmarco_pairs_runtime": 1.9393, + "eval_msmarco_pairs_samples_per_second": 66.002, + "eval_msmarco_pairs_steps_per_second": 1.031, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_nq_pairs_loss": 0.6836109757423401, + "eval_nq_pairs_runtime": 4.3924, + "eval_nq_pairs_samples_per_second": 29.141, + "eval_nq_pairs_steps_per_second": 0.455, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_trivia_pairs_loss": 0.5307232737541199, + "eval_trivia_pairs_runtime": 4.503, + "eval_trivia_pairs_samples_per_second": 28.426, + "eval_trivia_pairs_steps_per_second": 0.444, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_gooaq_pairs_loss": 0.5363317728042603, + "eval_gooaq_pairs_runtime": 1.2537, + "eval_gooaq_pairs_samples_per_second": 102.099, + "eval_gooaq_pairs_steps_per_second": 1.595, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_paws-pos_loss": 0.03341594338417053, + "eval_paws-pos_runtime": 1.0222, + "eval_paws-pos_samples_per_second": 125.221, + "eval_paws-pos_steps_per_second": 1.957, + "step": 1740 + }, + { + "epoch": 1.3498836307214894, + "eval_global_dataset_loss": 0.4127369821071625, + "eval_global_dataset_runtime": 21.0747, + "eval_global_dataset_samples_per_second": 19.739, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1740 + }, + { + "epoch": 1.3506594259115594, + "grad_norm": 6.586140155792236, + "learning_rate": 2.7826051489503688e-05, + "loss": 0.3974, + "step": 1741 + }, + { + "epoch": 1.3514352211016292, + "grad_norm": 5.662478446960449, + "learning_rate": 2.7947965619043843e-05, + "loss": 0.2382, + "step": 1742 + }, + { + "epoch": 1.352211016291699, + "grad_norm": 10.956356048583984, + "learning_rate": 2.8069401748804636e-05, + "loss": 0.8594, + "step": 1743 + }, + { + "epoch": 1.352986811481769, + "grad_norm": 8.82938289642334, + "learning_rate": 2.8190340188446785e-05, + "loss": 0.521, + "step": 1744 + }, + { + "epoch": 1.3537626066718387, + "grad_norm": 5.948559761047363, + "learning_rate": 2.831076132832927e-05, + "loss": 0.1812, + "step": 1745 + }, + { + "epoch": 1.3545384018619084, + "grad_norm": 7.489203453063965, + "learning_rate": 2.843064564268897e-05, + "loss": 0.3721, + "step": 1746 + }, + { + "epoch": 1.3553141970519782, + "grad_norm": 6.20318603515625, + "learning_rate": 2.8549973692806763e-05, + "loss": 0.2256, + "step": 1747 + }, + { + "epoch": 1.3560899922420482, + "grad_norm": 8.90699577331543, + "learning_rate": 2.8668726130159264e-05, + "loss": 0.6265, + "step": 1748 + }, + { + "epoch": 1.356865787432118, + "grad_norm": 9.362919807434082, + "learning_rate": 2.8786883699556245e-05, + "loss": 0.6678, + "step": 1749 + }, + { + "epoch": 1.3576415826221877, + "grad_norm": 12.013951301574707, + "learning_rate": 2.8904427242262662e-05, + "loss": 1.6685, + "step": 1750 + }, + { + "epoch": 1.3584173778122577, + "grad_norm": 12.919083595275879, + "learning_rate": 2.9021337699105257e-05, + "loss": 0.9168, + "step": 1751 + }, + { + "epoch": 1.3591931730023274, + "grad_norm": 4.618101119995117, + "learning_rate": 2.913759611356293e-05, + "loss": 0.1295, + "step": 1752 + }, + { + "epoch": 1.3599689681923972, + "grad_norm": 6.827664852142334, + "learning_rate": 2.9253183634840352e-05, + "loss": 0.3444, + "step": 1753 + }, + { + "epoch": 1.360744763382467, + "grad_norm": 4.106960773468018, + "learning_rate": 2.9368081520924604e-05, + "loss": 0.1335, + "step": 1754 + }, + { + "epoch": 1.3615205585725367, + "grad_norm": 4.825080394744873, + "learning_rate": 2.9482271141624178e-05, + "loss": 0.1726, + "step": 1755 + }, + { + "epoch": 1.3622963537626067, + "grad_norm": 6.3333258628845215, + "learning_rate": 2.959573398158961e-05, + "loss": 0.2331, + "step": 1756 + }, + { + "epoch": 1.3630721489526765, + "grad_norm": 6.753283500671387, + "learning_rate": 2.970845164331582e-05, + "loss": 0.2975, + "step": 1757 + }, + { + "epoch": 1.3638479441427462, + "grad_norm": 7.906069278717041, + "learning_rate": 2.982040585012513e-05, + "loss": 0.6436, + "step": 1758 + }, + { + "epoch": 1.3646237393328162, + "grad_norm": 5.071579456329346, + "learning_rate": 2.9931578449130736e-05, + "loss": 0.2953, + "step": 1759 + }, + { + "epoch": 1.365399534522886, + "grad_norm": 4.928244113922119, + "learning_rate": 3.004195141418019e-05, + "loss": 0.1971, + "step": 1760 + }, + { + "epoch": 1.3661753297129557, + "grad_norm": 4.417056560516357, + "learning_rate": 3.0151506848778103e-05, + "loss": 0.1601, + "step": 1761 + }, + { + "epoch": 1.3669511249030255, + "grad_norm": 5.001020908355713, + "learning_rate": 3.0260226988988168e-05, + "loss": 0.1561, + "step": 1762 + }, + { + "epoch": 1.3677269200930955, + "grad_norm": 5.429628849029541, + "learning_rate": 3.036809420631338e-05, + "loss": 0.2363, + "step": 1763 + }, + { + "epoch": 1.3685027152831652, + "grad_norm": 5.347793102264404, + "learning_rate": 3.047509101055451e-05, + "loss": 0.1514, + "step": 1764 + }, + { + "epoch": 1.369278510473235, + "grad_norm": 0.8979998230934143, + "learning_rate": 3.058120005264599e-05, + "loss": 0.0549, + "step": 1765 + }, + { + "epoch": 1.370054305663305, + "grad_norm": 1.640284776687622, + "learning_rate": 3.068640412746898e-05, + "loss": 0.0335, + "step": 1766 + }, + { + "epoch": 1.3708301008533748, + "grad_norm": 9.631306648254395, + "learning_rate": 3.079068617664119e-05, + "loss": 0.6062, + "step": 1767 + }, + { + "epoch": 1.3716058960434445, + "grad_norm": 4.674038410186768, + "learning_rate": 3.0894029291282744e-05, + "loss": 0.1275, + "step": 1768 + }, + { + "epoch": 1.3723816912335143, + "grad_norm": 10.050506591796875, + "learning_rate": 3.0996416714757955e-05, + "loss": 0.5735, + "step": 1769 + }, + { + "epoch": 1.373157486423584, + "grad_norm": 8.444786071777344, + "learning_rate": 3.1097831845392235e-05, + "loss": 0.2689, + "step": 1770 + }, + { + "epoch": 1.373933281613654, + "grad_norm": 9.099848747253418, + "learning_rate": 3.1198258239164084e-05, + "loss": 0.6431, + "step": 1771 + }, + { + "epoch": 1.3747090768037238, + "grad_norm": 7.653295516967773, + "learning_rate": 3.129767961237132e-05, + "loss": 0.3176, + "step": 1772 + }, + { + "epoch": 1.3754848719937938, + "grad_norm": 8.819071769714355, + "learning_rate": 3.1396079844271544e-05, + "loss": 0.5344, + "step": 1773 + }, + { + "epoch": 1.3762606671838635, + "grad_norm": 4.571998119354248, + "learning_rate": 3.149344297969592e-05, + "loss": 0.1444, + "step": 1774 + }, + { + "epoch": 1.3770364623739333, + "grad_norm": 4.7068915367126465, + "learning_rate": 3.15897532316363e-05, + "loss": 0.2958, + "step": 1775 + }, + { + "epoch": 1.377812257564003, + "grad_norm": 8.494279861450195, + "learning_rate": 3.168499498380502e-05, + "loss": 0.5993, + "step": 1776 + }, + { + "epoch": 1.3785880527540728, + "grad_norm": 5.714775562286377, + "learning_rate": 3.1779152793166935e-05, + "loss": 0.1819, + "step": 1777 + }, + { + "epoch": 1.3793638479441428, + "grad_norm": 4.31868314743042, + "learning_rate": 3.187221139244359e-05, + "loss": 0.1459, + "step": 1778 + }, + { + "epoch": 1.3801396431342126, + "grad_norm": 8.578054428100586, + "learning_rate": 3.196415569258861e-05, + "loss": 0.5487, + "step": 1779 + }, + { + "epoch": 1.3809154383242823, + "grad_norm": 0.8803556561470032, + "learning_rate": 3.2054970785234354e-05, + "loss": 0.0287, + "step": 1780 + }, + { + "epoch": 1.3816912335143523, + "grad_norm": 12.123942375183105, + "learning_rate": 3.21446419451093e-05, + "loss": 1.0298, + "step": 1781 + }, + { + "epoch": 1.382467028704422, + "grad_norm": 12.323450088500977, + "learning_rate": 3.223315463242562e-05, + "loss": 0.9276, + "step": 1782 + }, + { + "epoch": 1.3832428238944918, + "grad_norm": 11.920193672180176, + "learning_rate": 3.2320494495236725e-05, + "loss": 0.6801, + "step": 1783 + }, + { + "epoch": 1.3840186190845616, + "grad_norm": 13.185144424438477, + "learning_rate": 3.2406647371764466e-05, + "loss": 0.7823, + "step": 1784 + }, + { + "epoch": 1.3847944142746316, + "grad_norm": 10.486449241638184, + "learning_rate": 3.2491599292695314e-05, + "loss": 0.6669, + "step": 1785 + }, + { + "epoch": 1.3855702094647013, + "grad_norm": 7.427524089813232, + "learning_rate": 3.257533648344549e-05, + "loss": 0.3824, + "step": 1786 + }, + { + "epoch": 1.386346004654771, + "grad_norm": 4.930051803588867, + "learning_rate": 3.265784536639439e-05, + "loss": 0.2028, + "step": 1787 + }, + { + "epoch": 1.387121799844841, + "grad_norm": 7.830265045166016, + "learning_rate": 3.2739112563086164e-05, + "loss": 0.3366, + "step": 1788 + }, + { + "epoch": 1.3878975950349108, + "grad_norm": 8.88143539428711, + "learning_rate": 3.2819124896398974e-05, + "loss": 0.6148, + "step": 1789 + }, + { + "epoch": 1.3886733902249806, + "grad_norm": 7.663028717041016, + "learning_rate": 3.289786939268167e-05, + "loss": 0.5542, + "step": 1790 + }, + { + "epoch": 1.3894491854150504, + "grad_norm": 9.246788024902344, + "learning_rate": 3.2975333283857295e-05, + "loss": 0.5155, + "step": 1791 + }, + { + "epoch": 1.3902249806051201, + "grad_norm": 9.66773509979248, + "learning_rate": 3.305150400949342e-05, + "loss": 0.4753, + "step": 1792 + }, + { + "epoch": 1.39100077579519, + "grad_norm": 6.699496746063232, + "learning_rate": 3.312636921883884e-05, + "loss": 0.2439, + "step": 1793 + }, + { + "epoch": 1.3917765709852599, + "grad_norm": 8.093302726745605, + "learning_rate": 3.319991677282603e-05, + "loss": 0.6011, + "step": 1794 + }, + { + "epoch": 1.3925523661753298, + "grad_norm": 6.0601911544799805, + "learning_rate": 3.327213474603963e-05, + "loss": 0.2492, + "step": 1795 + }, + { + "epoch": 1.3933281613653996, + "grad_norm": 4.722718715667725, + "learning_rate": 3.3343011428649974e-05, + "loss": 0.2637, + "step": 1796 + }, + { + "epoch": 1.3941039565554694, + "grad_norm": 1.3844106197357178, + "learning_rate": 3.34125353283118e-05, + "loss": 0.0424, + "step": 1797 + }, + { + "epoch": 1.3948797517455391, + "grad_norm": 5.794443130493164, + "learning_rate": 3.3480695172027756e-05, + "loss": 0.2346, + "step": 1798 + }, + { + "epoch": 1.395655546935609, + "grad_norm": 11.610705375671387, + "learning_rate": 3.3547479907976174e-05, + "loss": 0.5894, + "step": 1799 + }, + { + "epoch": 1.3964313421256789, + "grad_norm": 5.794182777404785, + "learning_rate": 3.3612878707303124e-05, + "loss": 0.1924, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_Qnli-dev_cosine_accuracy": 0.673828125, + "eval_Qnli-dev_cosine_accuracy_threshold": 0.8497576117515564, + "eval_Qnli-dev_cosine_ap": 0.7251059953375416, + "eval_Qnli-dev_cosine_f1": 0.6817391304347826, + "eval_Qnli-dev_cosine_f1_threshold": 0.7426311373710632, + "eval_Qnli-dev_cosine_precision": 0.5781710914454278, + "eval_Qnli-dev_cosine_recall": 0.8305084745762712, + "eval_Qnli-dev_dot_accuracy": 0.650390625, + "eval_Qnli-dev_dot_accuracy_threshold": 398.70751953125, + "eval_Qnli-dev_dot_ap": 0.658610065376225, + "eval_Qnli-dev_dot_f1": 0.6755407653910152, + "eval_Qnli-dev_dot_f1_threshold": 340.4079284667969, + "eval_Qnli-dev_dot_precision": 0.5561643835616439, + "eval_Qnli-dev_dot_recall": 0.8601694915254238, + "eval_Qnli-dev_euclidean_accuracy": 0.681640625, + "eval_Qnli-dev_euclidean_accuracy_threshold": 12.672463417053223, + "eval_Qnli-dev_euclidean_ap": 0.7200026523288421, + "eval_Qnli-dev_euclidean_f1": 0.6791808873720137, + "eval_Qnli-dev_euclidean_f1_threshold": 16.050626754760742, + "eval_Qnli-dev_euclidean_precision": 0.5685714285714286, + "eval_Qnli-dev_euclidean_recall": 0.8432203389830508, + "eval_Qnli-dev_manhattan_accuracy": 0.671875, + "eval_Qnli-dev_manhattan_accuracy_threshold": 251.18646240234375, + "eval_Qnli-dev_manhattan_ap": 0.7207212592599722, + "eval_Qnli-dev_manhattan_f1": 0.6755852842809364, + "eval_Qnli-dev_manhattan_f1_threshold": 296.843994140625, + "eval_Qnli-dev_manhattan_precision": 0.5580110497237569, + "eval_Qnli-dev_manhattan_recall": 0.8559322033898306, + "eval_Qnli-dev_max_accuracy": 0.681640625, + "eval_Qnli-dev_max_accuracy_threshold": 398.70751953125, + "eval_Qnli-dev_max_ap": 0.7251059953375416, + "eval_Qnli-dev_max_f1": 0.6817391304347826, + "eval_Qnli-dev_max_f1_threshold": 340.4079284667969, + "eval_Qnli-dev_max_precision": 0.5781710914454278, + "eval_Qnli-dev_max_recall": 0.8601694915254238, + "eval_allNLI-dev_cosine_accuracy": 0.73046875, + "eval_allNLI-dev_cosine_accuracy_threshold": 0.8663178086280823, + "eval_allNLI-dev_cosine_ap": 0.5986867023720726, + "eval_allNLI-dev_cosine_f1": 0.6085011185682326, + "eval_allNLI-dev_cosine_f1_threshold": 0.7677128314971924, + "eval_allNLI-dev_cosine_precision": 0.49635036496350365, + "eval_allNLI-dev_cosine_recall": 0.7861271676300579, + "eval_allNLI-dev_dot_accuracy": 0.677734375, + "eval_allNLI-dev_dot_accuracy_threshold": 460.61102294921875, + "eval_allNLI-dev_dot_ap": 0.49037841212851296, + "eval_allNLI-dev_dot_f1": 0.5566218809980806, + "eval_allNLI-dev_dot_f1_threshold": 301.651611328125, + "eval_allNLI-dev_dot_precision": 0.4166666666666667, + "eval_allNLI-dev_dot_recall": 0.838150289017341, + "eval_allNLI-dev_euclidean_accuracy": 0.724609375, + "eval_allNLI-dev_euclidean_accuracy_threshold": 11.271186828613281, + "eval_allNLI-dev_euclidean_ap": 0.6061194422116027, + "eval_allNLI-dev_euclidean_f1": 0.6197802197802197, + "eval_allNLI-dev_euclidean_f1_threshold": 14.510498046875, + "eval_allNLI-dev_euclidean_precision": 0.5, + "eval_allNLI-dev_euclidean_recall": 0.815028901734104, + "eval_allNLI-dev_manhattan_accuracy": 0.7265625, + "eval_allNLI-dev_manhattan_accuracy_threshold": 210.0970458984375, + "eval_allNLI-dev_manhattan_ap": 0.6054258645063132, + "eval_allNLI-dev_manhattan_f1": 0.6258823529411764, + "eval_allNLI-dev_manhattan_f1_threshold": 249.9814453125, + "eval_allNLI-dev_manhattan_precision": 0.5277777777777778, + "eval_allNLI-dev_manhattan_recall": 0.7687861271676301, + "eval_allNLI-dev_max_accuracy": 0.73046875, + "eval_allNLI-dev_max_accuracy_threshold": 460.61102294921875, + "eval_allNLI-dev_max_ap": 0.6061194422116027, + "eval_allNLI-dev_max_f1": 0.6258823529411764, + "eval_allNLI-dev_max_f1_threshold": 301.651611328125, + "eval_allNLI-dev_max_precision": 0.5277777777777778, + "eval_allNLI-dev_max_recall": 0.838150289017341, + "eval_sequential_score": 0.7251059953375416, + "eval_sts-test_pearson_cosine": 0.8242141695899022, + "eval_sts-test_pearson_dot": 0.7975513189284409, + "eval_sts-test_pearson_euclidean": 0.8621455942074856, + "eval_sts-test_pearson_manhattan": 0.8571515224396993, + "eval_sts-test_pearson_max": 0.8621455942074856, + "eval_sts-test_spearman_cosine": 0.8680163200353933, + "eval_sts-test_spearman_dot": 0.780866818576554, + "eval_sts-test_spearman_euclidean": 0.8633915009268991, + "eval_sts-test_spearman_manhattan": 0.8588204728345974, + "eval_sts-test_spearman_max": 0.8680163200353933, + "eval_vitaminc-pairs_loss": 3.566164970397949, + "eval_vitaminc-pairs_runtime": 5.4262, + "eval_vitaminc-pairs_samples_per_second": 23.589, + "eval_vitaminc-pairs_steps_per_second": 0.369, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_negation-triplets_loss": 1.3690544366836548, + "eval_negation-triplets_runtime": 1.099, + "eval_negation-triplets_samples_per_second": 116.466, + "eval_negation-triplets_steps_per_second": 1.82, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_scitail-pairs-pos_loss": 0.10022085160017014, + "eval_scitail-pairs-pos_runtime": 1.231, + "eval_scitail-pairs-pos_samples_per_second": 103.979, + "eval_scitail-pairs-pos_steps_per_second": 1.625, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_scitail-pairs-qa_loss": 0.004428846295922995, + "eval_scitail-pairs-qa_runtime": 0.8975, + "eval_scitail-pairs-qa_samples_per_second": 142.612, + "eval_scitail-pairs-qa_steps_per_second": 2.228, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_xsum-pairs_loss": 0.1744140237569809, + "eval_xsum-pairs_runtime": 4.3, + "eval_xsum-pairs_samples_per_second": 29.767, + "eval_xsum-pairs_steps_per_second": 0.465, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_sciq_pairs_loss": 0.08125954866409302, + "eval_sciq_pairs_runtime": 5.5173, + "eval_sciq_pairs_samples_per_second": 23.2, + "eval_sciq_pairs_steps_per_second": 0.362, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_qasc_pairs_loss": 0.1624910980463028, + "eval_qasc_pairs_runtime": 0.9714, + "eval_qasc_pairs_samples_per_second": 131.773, + "eval_qasc_pairs_steps_per_second": 2.059, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_openbookqa_pairs_loss": 1.105613350868225, + "eval_openbookqa_pairs_runtime": 0.8508, + "eval_openbookqa_pairs_samples_per_second": 150.44, + "eval_openbookqa_pairs_steps_per_second": 2.351, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_msmarco_pairs_loss": 0.6583799719810486, + "eval_msmarco_pairs_runtime": 1.9441, + "eval_msmarco_pairs_samples_per_second": 65.841, + "eval_msmarco_pairs_steps_per_second": 1.029, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_nq_pairs_loss": 0.852587103843689, + "eval_nq_pairs_runtime": 4.3875, + "eval_nq_pairs_samples_per_second": 29.174, + "eval_nq_pairs_steps_per_second": 0.456, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_trivia_pairs_loss": 0.5877262353897095, + "eval_trivia_pairs_runtime": 4.4983, + "eval_trivia_pairs_samples_per_second": 28.455, + "eval_trivia_pairs_steps_per_second": 0.445, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_gooaq_pairs_loss": 0.5990483164787292, + "eval_gooaq_pairs_runtime": 1.2784, + "eval_gooaq_pairs_samples_per_second": 100.126, + "eval_gooaq_pairs_steps_per_second": 1.564, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_paws-pos_loss": 0.032762546092271805, + "eval_paws-pos_runtime": 1.0193, + "eval_paws-pos_samples_per_second": 125.578, + "eval_paws-pos_steps_per_second": 1.962, + "step": 1800 + }, + { + "epoch": 1.3964313421256789, + "eval_global_dataset_loss": 0.4612788259983063, + "eval_global_dataset_runtime": 21.0841, + "eval_global_dataset_samples_per_second": 19.73, + "eval_global_dataset_steps_per_second": 0.332, + "step": 1800 + }, + { + "epoch": 1.3972071373157486, + "grad_norm": 5.059379577636719, + "learning_rate": 3.367688096587828e-05, + "loss": 0.1036, + "step": 1801 + }, + { + "epoch": 1.3979829325058184, + "grad_norm": 7.66911506652832, + "learning_rate": 3.3739476306014314e-05, + "loss": 0.3558, + "step": 1802 + }, + { + "epoch": 1.3987587276958884, + "grad_norm": 11.040764808654785, + "learning_rate": 3.380065457814957e-05, + "loss": 0.9031, + "step": 1803 + }, + { + "epoch": 1.3995345228859581, + "grad_norm": 10.618706703186035, + "learning_rate": 3.386040586249382e-05, + "loss": 0.8033, + "step": 1804 + }, + { + "epoch": 1.400310318076028, + "grad_norm": 10.616861343383789, + "learning_rate": 3.391872047063665e-05, + "loss": 0.5739, + "step": 1805 + }, + { + "epoch": 1.4010861132660977, + "grad_norm": 8.67395305633545, + "learning_rate": 3.397558894711846e-05, + "loss": 0.3957, + "step": 1806 + } + ], + "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 +}