| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5019181585677749, | |
| "eval_steps": 157, | |
| "global_step": 785, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010230179028132993, | |
| "grad_norm": 127.07197570800781, | |
| "learning_rate": 2.129925452609159e-07, | |
| "loss": 7.0465, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.020460358056265986, | |
| "grad_norm": 193.34689331054688, | |
| "learning_rate": 5.537806176783813e-07, | |
| "loss": 9.0583, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.030690537084398978, | |
| "grad_norm": 183.065185546875, | |
| "learning_rate": 8.732694355697552e-07, | |
| "loss": 8.6271, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.04092071611253197, | |
| "grad_norm": 150.517333984375, | |
| "learning_rate": 1.2140575079872206e-06, | |
| "loss": 5.6026, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.05115089514066496, | |
| "grad_norm": 78.38551330566406, | |
| "learning_rate": 1.5335463258785943e-06, | |
| "loss": 4.6908, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.061381074168797956, | |
| "grad_norm": 106.0997314453125, | |
| "learning_rate": 1.8743343982960599e-06, | |
| "loss": 3.1954, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07161125319693094, | |
| "grad_norm": 47.50294876098633, | |
| "learning_rate": 2.215122470713525e-06, | |
| "loss": 2.1179, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08184143222506395, | |
| "grad_norm": 24.581972122192383, | |
| "learning_rate": 2.5559105431309904e-06, | |
| "loss": 1.2636, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.09207161125319693, | |
| "grad_norm": 219.01101684570312, | |
| "learning_rate": 2.8966986155484563e-06, | |
| "loss": 0.8325, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_Qnli-dev_cosine_accuracy": 0.65234375, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.8284608125686646, | |
| "eval_Qnli-dev_cosine_ap": 0.6663718886737695, | |
| "eval_Qnli-dev_cosine_f1": 0.6469673405909797, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.702437698841095, | |
| "eval_Qnli-dev_cosine_precision": 0.5110565110565111, | |
| "eval_Qnli-dev_cosine_recall": 0.8813559322033898, | |
| "eval_Qnli-dev_dot_accuracy": 0.654296875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 899.9195556640625, | |
| "eval_Qnli-dev_dot_ap": 0.6265727343636489, | |
| "eval_Qnli-dev_dot_f1": 0.6462480857580397, | |
| "eval_Qnli-dev_dot_f1_threshold": 708.03076171875, | |
| "eval_Qnli-dev_dot_precision": 0.5059952038369304, | |
| "eval_Qnli-dev_dot_recall": 0.8940677966101694, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.654296875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 17.465450286865234, | |
| "eval_Qnli-dev_euclidean_ap": 0.6678841105609593, | |
| "eval_Qnli-dev_euclidean_f1": 0.6485568760611206, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 22.50777244567871, | |
| "eval_Qnli-dev_euclidean_precision": 0.5410764872521246, | |
| "eval_Qnli-dev_euclidean_recall": 0.809322033898305, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.654296875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 541.6007080078125, | |
| "eval_Qnli-dev_manhattan_ap": 0.6671352135215549, | |
| "eval_Qnli-dev_manhattan_f1": 0.6472602739726028, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 693.766845703125, | |
| "eval_Qnli-dev_manhattan_precision": 0.5431034482758621, | |
| "eval_Qnli-dev_manhattan_recall": 0.8008474576271186, | |
| "eval_Qnli-dev_max_accuracy": 0.654296875, | |
| "eval_Qnli-dev_max_accuracy_threshold": 899.9195556640625, | |
| "eval_Qnli-dev_max_ap": 0.6678841105609593, | |
| "eval_Qnli-dev_max_f1": 0.6485568760611206, | |
| "eval_Qnli-dev_max_f1_threshold": 708.03076171875, | |
| "eval_Qnli-dev_max_precision": 0.5431034482758621, | |
| "eval_Qnli-dev_max_recall": 0.8940677966101694, | |
| "eval_allNLI-dev_cosine_accuracy": 0.69140625, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8886671662330627, | |
| "eval_allNLI-dev_cosine_ap": 0.5184375875165067, | |
| "eval_allNLI-dev_cosine_f1": 0.5584415584415584, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7475587725639343, | |
| "eval_allNLI-dev_cosine_precision": 0.4463667820069204, | |
| "eval_allNLI-dev_cosine_recall": 0.7456647398843931, | |
| "eval_allNLI-dev_dot_accuracy": 0.689453125, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 935.1204223632812, | |
| "eval_allNLI-dev_dot_ap": 0.4904119502439996, | |
| "eval_allNLI-dev_dot_f1": 0.5509259259259259, | |
| "eval_allNLI-dev_dot_f1_threshold": 803.4815673828125, | |
| "eval_allNLI-dev_dot_precision": 0.4594594594594595, | |
| "eval_allNLI-dev_dot_recall": 0.6878612716763006, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.6953125, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 15.452061653137207, | |
| "eval_allNLI-dev_euclidean_ap": 0.5162188247814465, | |
| "eval_allNLI-dev_euclidean_f1": 0.5557986870897156, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 22.695791244506836, | |
| "eval_allNLI-dev_euclidean_precision": 0.4471830985915493, | |
| "eval_allNLI-dev_euclidean_recall": 0.7341040462427746, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.693359375, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 484.3291015625, | |
| "eval_allNLI-dev_manhattan_ap": 0.5171097761848361, | |
| "eval_allNLI-dev_manhattan_f1": 0.5550660792951542, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 707.2572631835938, | |
| "eval_allNLI-dev_manhattan_precision": 0.4483985765124555, | |
| "eval_allNLI-dev_manhattan_recall": 0.7283236994219653, | |
| "eval_allNLI-dev_max_accuracy": 0.6953125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 935.1204223632812, | |
| "eval_allNLI-dev_max_ap": 0.5184375875165067, | |
| "eval_allNLI-dev_max_f1": 0.5584415584415584, | |
| "eval_allNLI-dev_max_f1_threshold": 803.4815673828125, | |
| "eval_allNLI-dev_max_precision": 0.4594594594594595, | |
| "eval_allNLI-dev_max_recall": 0.7456647398843931, | |
| "eval_sequential_score": 0.6678841105609593, | |
| "eval_sts-test_pearson_cosine": 0.8691399425075377, | |
| "eval_sts-test_pearson_dot": 0.84633945645717, | |
| "eval_sts-test_pearson_euclidean": 0.8863973274037125, | |
| "eval_sts-test_pearson_manhattan": 0.887205994467606, | |
| "eval_sts-test_pearson_max": 0.887205994467606, | |
| "eval_sts-test_spearman_cosine": 0.8753474716765325, | |
| "eval_sts-test_spearman_dot": 0.8345210301629428, | |
| "eval_sts-test_spearman_euclidean": 0.8755160049269324, | |
| "eval_sts-test_spearman_manhattan": 0.8763851191744148, | |
| "eval_sts-test_spearman_max": 0.8763851191744148, | |
| "eval_vitaminc-pairs_loss": 3.764756917953491, | |
| "eval_vitaminc-pairs_runtime": 4.7024, | |
| "eval_vitaminc-pairs_samples_per_second": 27.22, | |
| "eval_vitaminc-pairs_steps_per_second": 0.213, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_negation-triplets_loss": 1.4374932050704956, | |
| "eval_negation-triplets_runtime": 3.2218, | |
| "eval_negation-triplets_samples_per_second": 39.73, | |
| "eval_negation-triplets_steps_per_second": 0.31, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_scitail-pairs-pos_loss": 0.07158112525939941, | |
| "eval_scitail-pairs-pos_runtime": 2.7614, | |
| "eval_scitail-pairs-pos_samples_per_second": 46.353, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.362, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_scitail-pairs-qa_loss": 0.11121156811714172, | |
| "eval_scitail-pairs-qa_runtime": 2.2983, | |
| "eval_scitail-pairs-qa_samples_per_second": 55.694, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.435, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_xsum-pairs_loss": 0.28410428762435913, | |
| "eval_xsum-pairs_runtime": 3.1437, | |
| "eval_xsum-pairs_samples_per_second": 40.717, | |
| "eval_xsum-pairs_steps_per_second": 0.318, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_sciq_pairs_loss": 0.15557295083999634, | |
| "eval_sciq_pairs_runtime": 4.4086, | |
| "eval_sciq_pairs_samples_per_second": 29.034, | |
| "eval_sciq_pairs_steps_per_second": 0.227, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_qasc_pairs_loss": 0.9191333651542664, | |
| "eval_qasc_pairs_runtime": 2.1323, | |
| "eval_qasc_pairs_samples_per_second": 60.03, | |
| "eval_qasc_pairs_steps_per_second": 0.469, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_openbookqa_pairs_loss": 0.8187942504882812, | |
| "eval_openbookqa_pairs_runtime": 2.2771, | |
| "eval_openbookqa_pairs_samples_per_second": 56.213, | |
| "eval_openbookqa_pairs_steps_per_second": 0.439, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_msmarco_pairs_loss": 0.6015267372131348, | |
| "eval_msmarco_pairs_runtime": 2.2879, | |
| "eval_msmarco_pairs_samples_per_second": 55.946, | |
| "eval_msmarco_pairs_steps_per_second": 0.437, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_nq_pairs_loss": 0.9822638630867004, | |
| "eval_nq_pairs_runtime": 3.3732, | |
| "eval_nq_pairs_samples_per_second": 37.946, | |
| "eval_nq_pairs_steps_per_second": 0.296, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_trivia_pairs_loss": 0.3045535683631897, | |
| "eval_trivia_pairs_runtime": 3.3462, | |
| "eval_trivia_pairs_samples_per_second": 38.253, | |
| "eval_trivia_pairs_steps_per_second": 0.299, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_gooaq_pairs_loss": 0.4534245729446411, | |
| "eval_gooaq_pairs_runtime": 2.1775, | |
| "eval_gooaq_pairs_samples_per_second": 58.783, | |
| "eval_gooaq_pairs_steps_per_second": 0.459, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_paws-pos_loss": 0.03816518932580948, | |
| "eval_paws-pos_runtime": 2.3035, | |
| "eval_paws-pos_samples_per_second": 55.567, | |
| "eval_paws-pos_steps_per_second": 0.434, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.100383631713555, | |
| "eval_global_dataset_loss": 0.4413764476776123, | |
| "eval_global_dataset_runtime": 8.8968, | |
| "eval_global_dataset_samples_per_second": 36.53, | |
| "eval_global_dataset_steps_per_second": 0.337, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.10230179028132992, | |
| "grad_norm": 4.65382719039917, | |
| "learning_rate": 3.2374866879659217e-06, | |
| "loss": 0.5674, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11253196930946291, | |
| "grad_norm": 16.067514419555664, | |
| "learning_rate": 3.5782747603833866e-06, | |
| "loss": 0.6077, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.12276214833759591, | |
| "grad_norm": 13.119791984558105, | |
| "learning_rate": 3.919062832800852e-06, | |
| "loss": 0.4102, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1329923273657289, | |
| "grad_norm": 27.51828384399414, | |
| "learning_rate": 4.259850905218318e-06, | |
| "loss": 0.4442, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.1432225063938619, | |
| "grad_norm": 76.78152465820312, | |
| "learning_rate": 4.600638977635783e-06, | |
| "loss": 0.3306, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.1534526854219949, | |
| "grad_norm": 9.617141723632812, | |
| "learning_rate": 4.941427050053248e-06, | |
| "loss": 0.3002, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.1636828644501279, | |
| "grad_norm": 0.007532110437750816, | |
| "learning_rate": 5.2822151224707144e-06, | |
| "loss": 0.2485, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.17391304347826086, | |
| "grad_norm": 9.366889953613281, | |
| "learning_rate": 5.623003194888179e-06, | |
| "loss": 0.274, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.18414322250639387, | |
| "grad_norm": 2.958397626876831, | |
| "learning_rate": 5.963791267305645e-06, | |
| "loss": 0.2093, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.19437340153452684, | |
| "grad_norm": 0.2244662344455719, | |
| "learning_rate": 6.304579339723111e-06, | |
| "loss": 0.3521, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_Qnli-dev_cosine_accuracy": 0.6796875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.750099241733551, | |
| "eval_Qnli-dev_cosine_ap": 0.7156359777017246, | |
| "eval_Qnli-dev_cosine_f1": 0.6803418803418803, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6485489010810852, | |
| "eval_Qnli-dev_cosine_precision": 0.5702005730659025, | |
| "eval_Qnli-dev_cosine_recall": 0.8432203389830508, | |
| "eval_Qnli-dev_dot_accuracy": 0.681640625, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 724.4273681640625, | |
| "eval_Qnli-dev_dot_ap": 0.7008881822413993, | |
| "eval_Qnli-dev_dot_f1": 0.6768189509306259, | |
| "eval_Qnli-dev_dot_f1_threshold": 601.5821533203125, | |
| "eval_Qnli-dev_dot_precision": 0.5633802816901409, | |
| "eval_Qnli-dev_dot_recall": 0.847457627118644, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.6796875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 21.8489990234375, | |
| "eval_Qnli-dev_euclidean_ap": 0.7160572797142014, | |
| "eval_Qnli-dev_euclidean_f1": 0.6843033509700176, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 25.212352752685547, | |
| "eval_Qnli-dev_euclidean_precision": 0.5861027190332326, | |
| "eval_Qnli-dev_euclidean_recall": 0.8220338983050848, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.6796875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 667.1034545898438, | |
| "eval_Qnli-dev_manhattan_ap": 0.7161884702166013, | |
| "eval_Qnli-dev_manhattan_f1": 0.6877192982456141, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 791.084716796875, | |
| "eval_Qnli-dev_manhattan_precision": 0.5868263473053892, | |
| "eval_Qnli-dev_manhattan_recall": 0.8305084745762712, | |
| "eval_Qnli-dev_max_accuracy": 0.681640625, | |
| "eval_Qnli-dev_max_accuracy_threshold": 724.4273681640625, | |
| "eval_Qnli-dev_max_ap": 0.7161884702166013, | |
| "eval_Qnli-dev_max_f1": 0.6877192982456141, | |
| "eval_Qnli-dev_max_f1_threshold": 791.084716796875, | |
| "eval_Qnli-dev_max_precision": 0.5868263473053892, | |
| "eval_Qnli-dev_max_recall": 0.847457627118644, | |
| "eval_allNLI-dev_cosine_accuracy": 0.71875, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.834626317024231, | |
| "eval_allNLI-dev_cosine_ap": 0.5835584554560046, | |
| "eval_allNLI-dev_cosine_f1": 0.6094808126410836, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6867948174476624, | |
| "eval_allNLI-dev_cosine_precision": 0.5, | |
| "eval_allNLI-dev_cosine_recall": 0.7803468208092486, | |
| "eval_allNLI-dev_dot_accuracy": 0.716796875, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 792.005615234375, | |
| "eval_allNLI-dev_dot_ap": 0.5699574122566177, | |
| "eval_allNLI-dev_dot_f1": 0.6130434782608695, | |
| "eval_allNLI-dev_dot_f1_threshold": 633.9556274414062, | |
| "eval_allNLI-dev_dot_precision": 0.4912891986062718, | |
| "eval_allNLI-dev_dot_recall": 0.815028901734104, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.716796875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 17.933330535888672, | |
| "eval_allNLI-dev_euclidean_ap": 0.583705987979273, | |
| "eval_allNLI-dev_euclidean_f1": 0.602510460251046, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 25.284334182739258, | |
| "eval_allNLI-dev_euclidean_precision": 0.4721311475409836, | |
| "eval_allNLI-dev_euclidean_recall": 0.8323699421965318, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.720703125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 559.5651245117188, | |
| "eval_allNLI-dev_manhattan_ap": 0.5843640673658059, | |
| "eval_allNLI-dev_manhattan_f1": 0.6072186836518046, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 784.5065307617188, | |
| "eval_allNLI-dev_manhattan_precision": 0.4798657718120805, | |
| "eval_allNLI-dev_manhattan_recall": 0.8265895953757225, | |
| "eval_allNLI-dev_max_accuracy": 0.720703125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 792.005615234375, | |
| "eval_allNLI-dev_max_ap": 0.5843640673658059, | |
| "eval_allNLI-dev_max_f1": 0.6130434782608695, | |
| "eval_allNLI-dev_max_f1_threshold": 784.5065307617188, | |
| "eval_allNLI-dev_max_precision": 0.5, | |
| "eval_allNLI-dev_max_recall": 0.8323699421965318, | |
| "eval_sequential_score": 0.7161884702166013, | |
| "eval_sts-test_pearson_cosine": 0.9083914036604525, | |
| "eval_sts-test_pearson_dot": 0.8981723163151856, | |
| "eval_sts-test_pearson_euclidean": 0.9173432574357514, | |
| "eval_sts-test_pearson_manhattan": 0.9168751297887556, | |
| "eval_sts-test_pearson_max": 0.9173432574357514, | |
| "eval_sts-test_spearman_cosine": 0.9121196279131069, | |
| "eval_sts-test_spearman_dot": 0.894677286770672, | |
| "eval_sts-test_spearman_euclidean": 0.9124939274633844, | |
| "eval_sts-test_spearman_manhattan": 0.9125480317384066, | |
| "eval_sts-test_spearman_max": 0.9125480317384066, | |
| "eval_vitaminc-pairs_loss": 3.233806610107422, | |
| "eval_vitaminc-pairs_runtime": 4.507, | |
| "eval_vitaminc-pairs_samples_per_second": 28.401, | |
| "eval_vitaminc-pairs_steps_per_second": 0.222, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_negation-triplets_loss": 0.7851067781448364, | |
| "eval_negation-triplets_runtime": 3.158, | |
| "eval_negation-triplets_samples_per_second": 40.532, | |
| "eval_negation-triplets_steps_per_second": 0.317, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_scitail-pairs-pos_loss": 0.021762849763035774, | |
| "eval_scitail-pairs-pos_runtime": 2.6751, | |
| "eval_scitail-pairs-pos_samples_per_second": 47.849, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.374, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_scitail-pairs-qa_loss": 0.010233570821583271, | |
| "eval_scitail-pairs-qa_runtime": 2.2385, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.181, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.447, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_xsum-pairs_loss": 0.07829513400793076, | |
| "eval_xsum-pairs_runtime": 3.1055, | |
| "eval_xsum-pairs_samples_per_second": 41.217, | |
| "eval_xsum-pairs_steps_per_second": 0.322, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_sciq_pairs_loss": 0.06304117292165756, | |
| "eval_sciq_pairs_runtime": 4.3422, | |
| "eval_sciq_pairs_samples_per_second": 29.478, | |
| "eval_sciq_pairs_steps_per_second": 0.23, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_qasc_pairs_loss": 0.11734917759895325, | |
| "eval_qasc_pairs_runtime": 2.1055, | |
| "eval_qasc_pairs_samples_per_second": 60.793, | |
| "eval_qasc_pairs_steps_per_second": 0.475, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_openbookqa_pairs_loss": 0.45362916588783264, | |
| "eval_openbookqa_pairs_runtime": 2.2519, | |
| "eval_openbookqa_pairs_samples_per_second": 56.84, | |
| "eval_openbookqa_pairs_steps_per_second": 0.444, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_msmarco_pairs_loss": 0.1522417515516281, | |
| "eval_msmarco_pairs_runtime": 2.2506, | |
| "eval_msmarco_pairs_samples_per_second": 56.874, | |
| "eval_msmarco_pairs_steps_per_second": 0.444, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_nq_pairs_loss": 0.18318293988704681, | |
| "eval_nq_pairs_runtime": 3.3369, | |
| "eval_nq_pairs_samples_per_second": 38.359, | |
| "eval_nq_pairs_steps_per_second": 0.3, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_trivia_pairs_loss": 0.13312742114067078, | |
| "eval_trivia_pairs_runtime": 3.3622, | |
| "eval_trivia_pairs_samples_per_second": 38.07, | |
| "eval_trivia_pairs_steps_per_second": 0.297, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_gooaq_pairs_loss": 0.13270226120948792, | |
| "eval_gooaq_pairs_runtime": 2.2422, | |
| "eval_gooaq_pairs_samples_per_second": 57.086, | |
| "eval_gooaq_pairs_steps_per_second": 0.446, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_paws-pos_loss": 0.02471838891506195, | |
| "eval_paws-pos_runtime": 2.3061, | |
| "eval_paws-pos_samples_per_second": 55.506, | |
| "eval_paws-pos_steps_per_second": 0.434, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20076726342711, | |
| "eval_global_dataset_loss": 0.23317644000053406, | |
| "eval_global_dataset_runtime": 8.8835, | |
| "eval_global_dataset_samples_per_second": 36.585, | |
| "eval_global_dataset_steps_per_second": 0.338, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.20460358056265984, | |
| "grad_norm": 14.963007926940918, | |
| "learning_rate": 6.645367412140575e-06, | |
| "loss": 0.2628, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.21483375959079284, | |
| "grad_norm": 33.163265228271484, | |
| "learning_rate": 6.986155484558041e-06, | |
| "loss": 0.2167, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.22506393861892582, | |
| "grad_norm": 239.8547821044922, | |
| "learning_rate": 7.326943556975507e-06, | |
| "loss": 0.1816, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 6.623231410980225, | |
| "learning_rate": 7.667731629392972e-06, | |
| "loss": 0.1327, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.24552429667519182, | |
| "grad_norm": 9.30941390991211, | |
| "learning_rate": 8.008519701810438e-06, | |
| "loss": 0.2178, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.2557544757033248, | |
| "grad_norm": 4.750399112701416, | |
| "learning_rate": 8.32800851970181e-06, | |
| "loss": 0.0993, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2659846547314578, | |
| "grad_norm": 6.0740461349487305, | |
| "learning_rate": 8.668796592119276e-06, | |
| "loss": 0.2293, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.27621483375959077, | |
| "grad_norm": 39.3314323425293, | |
| "learning_rate": 9.009584664536743e-06, | |
| "loss": 0.4346, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.2864450127877238, | |
| "grad_norm": 1288.867431640625, | |
| "learning_rate": 9.329073482428116e-06, | |
| "loss": 1.0288, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.2966751918158568, | |
| "grad_norm": 9.365520477294922, | |
| "learning_rate": 9.64856230031949e-06, | |
| "loss": 0.4893, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_Qnli-dev_cosine_accuracy": 0.69921875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7410688400268555, | |
| "eval_Qnli-dev_cosine_ap": 0.7331344916543854, | |
| "eval_Qnli-dev_cosine_f1": 0.6888111888111887, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6412021517753601, | |
| "eval_Qnli-dev_cosine_precision": 0.5863095238095238, | |
| "eval_Qnli-dev_cosine_recall": 0.8347457627118644, | |
| "eval_Qnli-dev_dot_accuracy": 0.69140625, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 673.3492431640625, | |
| "eval_Qnli-dev_dot_ap": 0.702049578471496, | |
| "eval_Qnli-dev_dot_f1": 0.6897810218978102, | |
| "eval_Qnli-dev_dot_f1_threshold": 611.4298706054688, | |
| "eval_Qnli-dev_dot_precision": 0.6057692307692307, | |
| "eval_Qnli-dev_dot_recall": 0.8008474576271186, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.697265625, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 22.84914207458496, | |
| "eval_Qnli-dev_euclidean_ap": 0.7357836494194735, | |
| "eval_Qnli-dev_euclidean_f1": 0.6843177189409368, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 23.908557891845703, | |
| "eval_Qnli-dev_euclidean_precision": 0.6588235294117647, | |
| "eval_Qnli-dev_euclidean_recall": 0.711864406779661, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.69921875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 712.7897338867188, | |
| "eval_Qnli-dev_manhattan_ap": 0.736884863348749, | |
| "eval_Qnli-dev_manhattan_f1": 0.6865079365079366, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 753.1321411132812, | |
| "eval_Qnli-dev_manhattan_precision": 0.6455223880597015, | |
| "eval_Qnli-dev_manhattan_recall": 0.7330508474576272, | |
| "eval_Qnli-dev_max_accuracy": 0.69921875, | |
| "eval_Qnli-dev_max_accuracy_threshold": 712.7897338867188, | |
| "eval_Qnli-dev_max_ap": 0.736884863348749, | |
| "eval_Qnli-dev_max_f1": 0.6897810218978102, | |
| "eval_Qnli-dev_max_f1_threshold": 753.1321411132812, | |
| "eval_Qnli-dev_max_precision": 0.6588235294117647, | |
| "eval_Qnli-dev_max_recall": 0.8347457627118644, | |
| "eval_allNLI-dev_cosine_accuracy": 0.716796875, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.835934042930603, | |
| "eval_allNLI-dev_cosine_ap": 0.5816321594253542, | |
| "eval_allNLI-dev_cosine_f1": 0.6029106029106028, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6776033043861389, | |
| "eval_allNLI-dev_cosine_precision": 0.4707792207792208, | |
| "eval_allNLI-dev_cosine_recall": 0.838150289017341, | |
| "eval_allNLI-dev_dot_accuracy": 0.703125, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 810.56591796875, | |
| "eval_allNLI-dev_dot_ap": 0.5677886575854101, | |
| "eval_allNLI-dev_dot_f1": 0.6004228329809725, | |
| "eval_allNLI-dev_dot_f1_threshold": 647.8101806640625, | |
| "eval_allNLI-dev_dot_precision": 0.47333333333333333, | |
| "eval_allNLI-dev_dot_recall": 0.8208092485549133, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.71875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 17.530292510986328, | |
| "eval_allNLI-dev_euclidean_ap": 0.5790276724859877, | |
| "eval_allNLI-dev_euclidean_f1": 0.6070686070686071, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 24.67755126953125, | |
| "eval_allNLI-dev_euclidean_precision": 0.474025974025974, | |
| "eval_allNLI-dev_euclidean_recall": 0.8439306358381503, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.71875, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 547.3074951171875, | |
| "eval_allNLI-dev_manhattan_ap": 0.5803704124451106, | |
| "eval_allNLI-dev_manhattan_f1": 0.6008583690987125, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 760.1802368164062, | |
| "eval_allNLI-dev_manhattan_precision": 0.4778156996587031, | |
| "eval_allNLI-dev_manhattan_recall": 0.8092485549132948, | |
| "eval_allNLI-dev_max_accuracy": 0.71875, | |
| "eval_allNLI-dev_max_accuracy_threshold": 810.56591796875, | |
| "eval_allNLI-dev_max_ap": 0.5816321594253542, | |
| "eval_allNLI-dev_max_f1": 0.6070686070686071, | |
| "eval_allNLI-dev_max_f1_threshold": 760.1802368164062, | |
| "eval_allNLI-dev_max_precision": 0.4778156996587031, | |
| "eval_allNLI-dev_max_recall": 0.8439306358381503, | |
| "eval_sequential_score": 0.736884863348749, | |
| "eval_sts-test_pearson_cosine": 0.9094229035285761, | |
| "eval_sts-test_pearson_dot": 0.8982590057670237, | |
| "eval_sts-test_pearson_euclidean": 0.9199260651959468, | |
| "eval_sts-test_pearson_manhattan": 0.9194219727619668, | |
| "eval_sts-test_pearson_max": 0.9199260651959468, | |
| "eval_sts-test_spearman_cosine": 0.9158958736030971, | |
| "eval_sts-test_spearman_dot": 0.8963403778977608, | |
| "eval_sts-test_spearman_euclidean": 0.9158618178931783, | |
| "eval_sts-test_spearman_manhattan": 0.9156180577723795, | |
| "eval_sts-test_spearman_max": 0.9158958736030971, | |
| "eval_vitaminc-pairs_loss": 3.671621084213257, | |
| "eval_vitaminc-pairs_runtime": 4.6059, | |
| "eval_vitaminc-pairs_samples_per_second": 27.791, | |
| "eval_vitaminc-pairs_steps_per_second": 0.217, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_negation-triplets_loss": 0.6801220774650574, | |
| "eval_negation-triplets_runtime": 3.2119, | |
| "eval_negation-triplets_samples_per_second": 39.852, | |
| "eval_negation-triplets_steps_per_second": 0.311, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_scitail-pairs-pos_loss": 0.01225399598479271, | |
| "eval_scitail-pairs-pos_runtime": 2.7794, | |
| "eval_scitail-pairs-pos_samples_per_second": 46.052, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.36, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_scitail-pairs-qa_loss": 0.00501797953620553, | |
| "eval_scitail-pairs-qa_runtime": 2.3318, | |
| "eval_scitail-pairs-qa_samples_per_second": 54.893, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.429, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_xsum-pairs_loss": 0.05804339051246643, | |
| "eval_xsum-pairs_runtime": 3.1527, | |
| "eval_xsum-pairs_samples_per_second": 40.6, | |
| "eval_xsum-pairs_steps_per_second": 0.317, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_sciq_pairs_loss": 0.04724589362740517, | |
| "eval_sciq_pairs_runtime": 4.3481, | |
| "eval_sciq_pairs_samples_per_second": 29.438, | |
| "eval_sciq_pairs_steps_per_second": 0.23, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_qasc_pairs_loss": 0.04424997419118881, | |
| "eval_qasc_pairs_runtime": 2.1476, | |
| "eval_qasc_pairs_samples_per_second": 59.6, | |
| "eval_qasc_pairs_steps_per_second": 0.466, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_openbookqa_pairs_loss": 0.46001699566841125, | |
| "eval_openbookqa_pairs_runtime": 2.3534, | |
| "eval_openbookqa_pairs_samples_per_second": 54.389, | |
| "eval_openbookqa_pairs_steps_per_second": 0.425, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_msmarco_pairs_loss": 0.22556719183921814, | |
| "eval_msmarco_pairs_runtime": 2.2946, | |
| "eval_msmarco_pairs_samples_per_second": 55.782, | |
| "eval_msmarco_pairs_steps_per_second": 0.436, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_nq_pairs_loss": 0.13123252987861633, | |
| "eval_nq_pairs_runtime": 3.3642, | |
| "eval_nq_pairs_samples_per_second": 38.048, | |
| "eval_nq_pairs_steps_per_second": 0.297, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_trivia_pairs_loss": 0.11183707416057587, | |
| "eval_trivia_pairs_runtime": 3.3501, | |
| "eval_trivia_pairs_samples_per_second": 38.208, | |
| "eval_trivia_pairs_steps_per_second": 0.299, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_gooaq_pairs_loss": 0.17444811761379242, | |
| "eval_gooaq_pairs_runtime": 2.2434, | |
| "eval_gooaq_pairs_samples_per_second": 57.056, | |
| "eval_gooaq_pairs_steps_per_second": 0.446, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_paws-pos_loss": 0.02457336150109768, | |
| "eval_paws-pos_runtime": 2.3035, | |
| "eval_paws-pos_samples_per_second": 55.567, | |
| "eval_paws-pos_steps_per_second": 0.434, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.30115089514066495, | |
| "eval_global_dataset_loss": 0.259395956993103, | |
| "eval_global_dataset_runtime": 8.8545, | |
| "eval_global_dataset_samples_per_second": 36.704, | |
| "eval_global_dataset_steps_per_second": 0.339, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.3069053708439898, | |
| "grad_norm": 0.012338381260633469, | |
| "learning_rate": 9.989350372736955e-06, | |
| "loss": 0.1706, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3171355498721228, | |
| "grad_norm": 2.509495973587036, | |
| "learning_rate": 1.033013844515442e-05, | |
| "loss": 0.1083, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.3273657289002558, | |
| "grad_norm": 58.62932586669922, | |
| "learning_rate": 1.0670926517571887e-05, | |
| "loss": 0.1036, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.3375959079283887, | |
| "grad_norm": 57.70167922973633, | |
| "learning_rate": 1.101171458998935e-05, | |
| "loss": 0.4057, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.34782608695652173, | |
| "grad_norm": 5.249549388885498, | |
| "learning_rate": 1.1352502662406816e-05, | |
| "loss": 0.1629, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.35805626598465473, | |
| "grad_norm": 744.4637451171875, | |
| "learning_rate": 1.1693290734824283e-05, | |
| "loss": 0.6291, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.36828644501278773, | |
| "grad_norm": 3727.07373046875, | |
| "learning_rate": 1.2012779552715656e-05, | |
| "loss": 0.7985, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.37851662404092073, | |
| "grad_norm": 1.6333959102630615, | |
| "learning_rate": 1.233226837060703e-05, | |
| "loss": 0.3209, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.3887468030690537, | |
| "grad_norm": 71.8410873413086, | |
| "learning_rate": 1.2651757188498404e-05, | |
| "loss": 0.2933, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.3989769820971867, | |
| "grad_norm": 15.115602493286133, | |
| "learning_rate": 1.299254526091587e-05, | |
| "loss": 0.1887, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_Qnli-dev_cosine_accuracy": 0.6796875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7387017011642456, | |
| "eval_Qnli-dev_cosine_ap": 0.7232227658329593, | |
| "eval_Qnli-dev_cosine_f1": 0.6812816188870151, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.5850255489349365, | |
| "eval_Qnli-dev_cosine_precision": 0.5658263305322129, | |
| "eval_Qnli-dev_cosine_recall": 0.8559322033898306, | |
| "eval_Qnli-dev_dot_accuracy": 0.6796875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 610.0311279296875, | |
| "eval_Qnli-dev_dot_ap": 0.7080904391300955, | |
| "eval_Qnli-dev_dot_f1": 0.6843853820598007, | |
| "eval_Qnli-dev_dot_f1_threshold": 513.2543334960938, | |
| "eval_Qnli-dev_dot_precision": 0.5628415300546448, | |
| "eval_Qnli-dev_dot_recall": 0.8728813559322034, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.68359375, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 22.151939392089844, | |
| "eval_Qnli-dev_euclidean_ap": 0.7234064908940119, | |
| "eval_Qnli-dev_euclidean_f1": 0.6771653543307087, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 28.56058692932129, | |
| "eval_Qnli-dev_euclidean_precision": 0.5388471177944862, | |
| "eval_Qnli-dev_euclidean_recall": 0.9110169491525424, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.68359375, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 695.5964965820312, | |
| "eval_Qnli-dev_manhattan_ap": 0.7224600475409794, | |
| "eval_Qnli-dev_manhattan_f1": 0.679245283018868, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 890.5444946289062, | |
| "eval_Qnli-dev_manhattan_precision": 0.54, | |
| "eval_Qnli-dev_manhattan_recall": 0.9152542372881356, | |
| "eval_Qnli-dev_max_accuracy": 0.68359375, | |
| "eval_Qnli-dev_max_accuracy_threshold": 695.5964965820312, | |
| "eval_Qnli-dev_max_ap": 0.7234064908940119, | |
| "eval_Qnli-dev_max_f1": 0.6843853820598007, | |
| "eval_Qnli-dev_max_f1_threshold": 890.5444946289062, | |
| "eval_Qnli-dev_max_precision": 0.5658263305322129, | |
| "eval_Qnli-dev_max_recall": 0.9152542372881356, | |
| "eval_allNLI-dev_cosine_accuracy": 0.71484375, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8244803547859192, | |
| "eval_allNLI-dev_cosine_ap": 0.5598399809421326, | |
| "eval_allNLI-dev_cosine_f1": 0.6026785714285714, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6709309816360474, | |
| "eval_allNLI-dev_cosine_precision": 0.4909090909090909, | |
| "eval_allNLI-dev_cosine_recall": 0.7803468208092486, | |
| "eval_allNLI-dev_dot_accuracy": 0.7109375, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 769.110107421875, | |
| "eval_allNLI-dev_dot_ap": 0.5522830058603818, | |
| "eval_allNLI-dev_dot_f1": 0.5976470588235294, | |
| "eval_allNLI-dev_dot_f1_threshold": 639.0211181640625, | |
| "eval_allNLI-dev_dot_precision": 0.503968253968254, | |
| "eval_allNLI-dev_dot_recall": 0.7341040462427746, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.705078125, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 15.881658554077148, | |
| "eval_allNLI-dev_euclidean_ap": 0.5573704039417041, | |
| "eval_allNLI-dev_euclidean_f1": 0.589010989010989, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 24.65237808227539, | |
| "eval_allNLI-dev_euclidean_precision": 0.475177304964539, | |
| "eval_allNLI-dev_euclidean_recall": 0.7745664739884393, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.705078125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 494.8561706542969, | |
| "eval_allNLI-dev_manhattan_ap": 0.5596244537722644, | |
| "eval_allNLI-dev_manhattan_f1": 0.593886462882096, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 772.1622314453125, | |
| "eval_allNLI-dev_manhattan_precision": 0.47719298245614034, | |
| "eval_allNLI-dev_manhattan_recall": 0.7861271676300579, | |
| "eval_allNLI-dev_max_accuracy": 0.71484375, | |
| "eval_allNLI-dev_max_accuracy_threshold": 769.110107421875, | |
| "eval_allNLI-dev_max_ap": 0.5598399809421326, | |
| "eval_allNLI-dev_max_f1": 0.6026785714285714, | |
| "eval_allNLI-dev_max_f1_threshold": 772.1622314453125, | |
| "eval_allNLI-dev_max_precision": 0.503968253968254, | |
| "eval_allNLI-dev_max_recall": 0.7861271676300579, | |
| "eval_sequential_score": 0.7234064908940119, | |
| "eval_sts-test_pearson_cosine": 0.9059824822945504, | |
| "eval_sts-test_pearson_dot": 0.8934118861402478, | |
| "eval_sts-test_pearson_euclidean": 0.9143064951884331, | |
| "eval_sts-test_pearson_manhattan": 0.9138764030584118, | |
| "eval_sts-test_pearson_max": 0.9143064951884331, | |
| "eval_sts-test_spearman_cosine": 0.9099404652396526, | |
| "eval_sts-test_spearman_dot": 0.8907050753070487, | |
| "eval_sts-test_spearman_euclidean": 0.9102612870325675, | |
| "eval_sts-test_spearman_manhattan": 0.9097054966996683, | |
| "eval_sts-test_spearman_max": 0.9102612870325675, | |
| "eval_vitaminc-pairs_loss": 3.6695005893707275, | |
| "eval_vitaminc-pairs_runtime": 4.5674, | |
| "eval_vitaminc-pairs_samples_per_second": 28.025, | |
| "eval_vitaminc-pairs_steps_per_second": 0.219, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_negation-triplets_loss": 0.8490238189697266, | |
| "eval_negation-triplets_runtime": 3.2004, | |
| "eval_negation-triplets_samples_per_second": 39.995, | |
| "eval_negation-triplets_steps_per_second": 0.312, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_scitail-pairs-pos_loss": 0.017992086708545685, | |
| "eval_scitail-pairs-pos_runtime": 2.8, | |
| "eval_scitail-pairs-pos_samples_per_second": 45.714, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.357, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_scitail-pairs-qa_loss": 0.06842579692602158, | |
| "eval_scitail-pairs-qa_runtime": 2.2721, | |
| "eval_scitail-pairs-qa_samples_per_second": 56.337, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.44, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_xsum-pairs_loss": 1.8987103700637817, | |
| "eval_xsum-pairs_runtime": 3.1443, | |
| "eval_xsum-pairs_samples_per_second": 40.709, | |
| "eval_xsum-pairs_steps_per_second": 0.318, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_sciq_pairs_loss": 0.2226540595293045, | |
| "eval_sciq_pairs_runtime": 4.4099, | |
| "eval_sciq_pairs_samples_per_second": 29.025, | |
| "eval_sciq_pairs_steps_per_second": 0.227, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_qasc_pairs_loss": 0.0974484235048294, | |
| "eval_qasc_pairs_runtime": 2.1808, | |
| "eval_qasc_pairs_samples_per_second": 58.693, | |
| "eval_qasc_pairs_steps_per_second": 0.459, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_openbookqa_pairs_loss": 0.614979088306427, | |
| "eval_openbookqa_pairs_runtime": 2.2849, | |
| "eval_openbookqa_pairs_samples_per_second": 56.02, | |
| "eval_openbookqa_pairs_steps_per_second": 0.438, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_msmarco_pairs_loss": 2.7714712619781494, | |
| "eval_msmarco_pairs_runtime": 2.2847, | |
| "eval_msmarco_pairs_samples_per_second": 56.025, | |
| "eval_msmarco_pairs_steps_per_second": 0.438, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_nq_pairs_loss": 1.1084551811218262, | |
| "eval_nq_pairs_runtime": 3.3615, | |
| "eval_nq_pairs_samples_per_second": 38.079, | |
| "eval_nq_pairs_steps_per_second": 0.297, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_trivia_pairs_loss": 1.0560091733932495, | |
| "eval_trivia_pairs_runtime": 3.3645, | |
| "eval_trivia_pairs_samples_per_second": 38.044, | |
| "eval_trivia_pairs_steps_per_second": 0.297, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_gooaq_pairs_loss": 0.7939258813858032, | |
| "eval_gooaq_pairs_runtime": 2.197, | |
| "eval_gooaq_pairs_samples_per_second": 58.261, | |
| "eval_gooaq_pairs_steps_per_second": 0.455, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_paws-pos_loss": 0.17634068429470062, | |
| "eval_paws-pos_runtime": 2.2907, | |
| "eval_paws-pos_samples_per_second": 55.878, | |
| "eval_paws-pos_steps_per_second": 0.437, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.40153452685422, | |
| "eval_global_dataset_loss": 0.678693413734436, | |
| "eval_global_dataset_runtime": 8.8475, | |
| "eval_global_dataset_samples_per_second": 36.733, | |
| "eval_global_dataset_steps_per_second": 0.339, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.4092071611253197, | |
| "grad_norm": 440.4800720214844, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 1.0156, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.4194373401534527, | |
| "grad_norm": 398.4678955078125, | |
| "learning_rate": 1.36741214057508e-05, | |
| "loss": 1.3699, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.4296675191815857, | |
| "grad_norm": 4009.26220703125, | |
| "learning_rate": 1.3993610223642173e-05, | |
| "loss": 3.8892, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.4398976982097187, | |
| "grad_norm": 643.8749389648438, | |
| "learning_rate": 1.4334398296059639e-05, | |
| "loss": 2.5961, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.45012787723785164, | |
| "grad_norm": 2570.24560546875, | |
| "learning_rate": 1.4675186368477104e-05, | |
| "loss": 7.094, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.46035805626598464, | |
| "grad_norm": 5654.052734375, | |
| "learning_rate": 1.501597444089457e-05, | |
| "loss": 2.8695, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 710.1712646484375, | |
| "learning_rate": 1.5356762513312037e-05, | |
| "loss": 5.7516, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.48081841432225064, | |
| "grad_norm": 186.94851684570312, | |
| "learning_rate": 1.56975505857295e-05, | |
| "loss": 1.1622, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.49104859335038364, | |
| "grad_norm": 418.9023742675781, | |
| "learning_rate": 1.6038338658146964e-05, | |
| "loss": 1.2898, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.5012787723785166, | |
| "grad_norm": 59.84318542480469, | |
| "learning_rate": 1.637912673056443e-05, | |
| "loss": 0.5779, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_Qnli-dev_cosine_accuracy": 0.701171875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.6924287676811218, | |
| "eval_Qnli-dev_cosine_ap": 0.7385336798634711, | |
| "eval_Qnli-dev_cosine_f1": 0.6933867735470942, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6914424300193787, | |
| "eval_Qnli-dev_cosine_precision": 0.6577946768060836, | |
| "eval_Qnli-dev_cosine_recall": 0.7330508474576272, | |
| "eval_Qnli-dev_dot_accuracy": 0.671875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 727.0087890625, | |
| "eval_Qnli-dev_dot_ap": 0.6770156987156393, | |
| "eval_Qnli-dev_dot_f1": 0.6799999999999999, | |
| "eval_Qnli-dev_dot_f1_threshold": 679.4116821289062, | |
| "eval_Qnli-dev_dot_precision": 0.5955414012738853, | |
| "eval_Qnli-dev_dot_recall": 0.7923728813559322, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.701171875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 23.944599151611328, | |
| "eval_Qnli-dev_euclidean_ap": 0.748116028021377, | |
| "eval_Qnli-dev_euclidean_f1": 0.69215291750503, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 25.11563491821289, | |
| "eval_Qnli-dev_euclidean_precision": 0.6590038314176245, | |
| "eval_Qnli-dev_euclidean_recall": 0.7288135593220338, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.703125, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 755.9312744140625, | |
| "eval_Qnli-dev_manhattan_ap": 0.7493754470561521, | |
| "eval_Qnli-dev_manhattan_f1": 0.6960000000000001, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 786.2037353515625, | |
| "eval_Qnli-dev_manhattan_precision": 0.6590909090909091, | |
| "eval_Qnli-dev_manhattan_recall": 0.7372881355932204, | |
| "eval_Qnli-dev_max_accuracy": 0.703125, | |
| "eval_Qnli-dev_max_accuracy_threshold": 755.9312744140625, | |
| "eval_Qnli-dev_max_ap": 0.7493754470561521, | |
| "eval_Qnli-dev_max_f1": 0.6960000000000001, | |
| "eval_Qnli-dev_max_f1_threshold": 786.2037353515625, | |
| "eval_Qnli-dev_max_precision": 0.6590909090909091, | |
| "eval_Qnli-dev_max_recall": 0.7923728813559322, | |
| "eval_allNLI-dev_cosine_accuracy": 0.708984375, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8311391472816467, | |
| "eval_allNLI-dev_cosine_ap": 0.5827534981088864, | |
| "eval_allNLI-dev_cosine_f1": 0.5991379310344827, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6599106788635254, | |
| "eval_allNLI-dev_cosine_precision": 0.47766323024054985, | |
| "eval_allNLI-dev_cosine_recall": 0.8034682080924855, | |
| "eval_allNLI-dev_dot_accuracy": 0.708984375, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 879.6187744140625, | |
| "eval_allNLI-dev_dot_ap": 0.5812507953653799, | |
| "eval_allNLI-dev_dot_f1": 0.597590361445783, | |
| "eval_allNLI-dev_dot_f1_threshold": 726.0135498046875, | |
| "eval_allNLI-dev_dot_precision": 0.512396694214876, | |
| "eval_allNLI-dev_dot_recall": 0.7167630057803468, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.712890625, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 20.80200958251953, | |
| "eval_allNLI-dev_euclidean_ap": 0.5793171372613979, | |
| "eval_allNLI-dev_euclidean_f1": 0.5988023952095808, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 27.593257904052734, | |
| "eval_allNLI-dev_euclidean_precision": 0.4573170731707317, | |
| "eval_allNLI-dev_euclidean_recall": 0.8670520231213873, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.7109375, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 592.0584716796875, | |
| "eval_allNLI-dev_manhattan_ap": 0.5794717044021251, | |
| "eval_allNLI-dev_manhattan_f1": 0.5980198019801979, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 856.8595581054688, | |
| "eval_allNLI-dev_manhattan_precision": 0.45481927710843373, | |
| "eval_allNLI-dev_manhattan_recall": 0.8728323699421965, | |
| "eval_allNLI-dev_max_accuracy": 0.712890625, | |
| "eval_allNLI-dev_max_accuracy_threshold": 879.6187744140625, | |
| "eval_allNLI-dev_max_ap": 0.5827534981088864, | |
| "eval_allNLI-dev_max_f1": 0.5991379310344827, | |
| "eval_allNLI-dev_max_f1_threshold": 856.8595581054688, | |
| "eval_allNLI-dev_max_precision": 0.512396694214876, | |
| "eval_allNLI-dev_max_recall": 0.8728323699421965, | |
| "eval_sequential_score": 0.7493754470561521, | |
| "eval_sts-test_pearson_cosine": 0.9091021775992931, | |
| "eval_sts-test_pearson_dot": 0.8832539076390011, | |
| "eval_sts-test_pearson_euclidean": 0.9096800553065446, | |
| "eval_sts-test_pearson_manhattan": 0.9116788038260619, | |
| "eval_sts-test_pearson_max": 0.9116788038260619, | |
| "eval_sts-test_spearman_cosine": 0.9058912347278699, | |
| "eval_sts-test_spearman_dot": 0.8744089034713706, | |
| "eval_sts-test_spearman_euclidean": 0.9063140467728005, | |
| "eval_sts-test_spearman_manhattan": 0.9082527161675373, | |
| "eval_sts-test_spearman_max": 0.9082527161675373, | |
| "eval_vitaminc-pairs_loss": 3.191007614135742, | |
| "eval_vitaminc-pairs_runtime": 4.5691, | |
| "eval_vitaminc-pairs_samples_per_second": 28.014, | |
| "eval_vitaminc-pairs_steps_per_second": 0.219, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_negation-triplets_loss": 0.8027606010437012, | |
| "eval_negation-triplets_runtime": 3.1899, | |
| "eval_negation-triplets_samples_per_second": 40.127, | |
| "eval_negation-triplets_steps_per_second": 0.313, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_scitail-pairs-pos_loss": 0.011304985731840134, | |
| "eval_scitail-pairs-pos_runtime": 2.7258, | |
| "eval_scitail-pairs-pos_samples_per_second": 46.958, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.367, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_scitail-pairs-qa_loss": 0.009512349031865597, | |
| "eval_scitail-pairs-qa_runtime": 2.2719, | |
| "eval_scitail-pairs-qa_samples_per_second": 56.341, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.44, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_xsum-pairs_loss": 0.11288230121135712, | |
| "eval_xsum-pairs_runtime": 3.1424, | |
| "eval_xsum-pairs_samples_per_second": 40.734, | |
| "eval_xsum-pairs_steps_per_second": 0.318, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_sciq_pairs_loss": 0.062423791736364365, | |
| "eval_sciq_pairs_runtime": 4.3404, | |
| "eval_sciq_pairs_samples_per_second": 29.491, | |
| "eval_sciq_pairs_steps_per_second": 0.23, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_qasc_pairs_loss": 0.0833595022559166, | |
| "eval_qasc_pairs_runtime": 2.153, | |
| "eval_qasc_pairs_samples_per_second": 59.453, | |
| "eval_qasc_pairs_steps_per_second": 0.464, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_openbookqa_pairs_loss": 0.47392797470092773, | |
| "eval_openbookqa_pairs_runtime": 2.286, | |
| "eval_openbookqa_pairs_samples_per_second": 55.994, | |
| "eval_openbookqa_pairs_steps_per_second": 0.437, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_msmarco_pairs_loss": 0.1399916559457779, | |
| "eval_msmarco_pairs_runtime": 2.3254, | |
| "eval_msmarco_pairs_samples_per_second": 55.045, | |
| "eval_msmarco_pairs_steps_per_second": 0.43, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_nq_pairs_loss": 0.2359679490327835, | |
| "eval_nq_pairs_runtime": 3.3719, | |
| "eval_nq_pairs_samples_per_second": 37.961, | |
| "eval_nq_pairs_steps_per_second": 0.297, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_trivia_pairs_loss": 0.3237164318561554, | |
| "eval_trivia_pairs_runtime": 3.3441, | |
| "eval_trivia_pairs_samples_per_second": 38.276, | |
| "eval_trivia_pairs_steps_per_second": 0.299, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_gooaq_pairs_loss": 0.2629774510860443, | |
| "eval_gooaq_pairs_runtime": 2.1943, | |
| "eval_gooaq_pairs_samples_per_second": 58.333, | |
| "eval_gooaq_pairs_steps_per_second": 0.456, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_paws-pos_loss": 0.02829882688820362, | |
| "eval_paws-pos_runtime": 2.3487, | |
| "eval_paws-pos_samples_per_second": 54.497, | |
| "eval_paws-pos_steps_per_second": 0.426, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5019181585677749, | |
| "eval_global_dataset_loss": 0.24213480949401855, | |
| "eval_global_dataset_runtime": 8.8463, | |
| "eval_global_dataset_samples_per_second": 36.738, | |
| "eval_global_dataset_steps_per_second": 0.339, | |
| "step": 785 | |
| } | |
| ], | |
| "logging_steps": 16, | |
| "max_steps": 3128, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 157, | |
| "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": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |