diff --git "a/checkpoint-774/trainer_state.json" "b/checkpoint-774/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-774/trainer_state.json" @@ -0,0 +1,7767 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.6004654771140419, + "eval_steps": 60, + "global_step": 774, + "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 + } + ], + "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 +}