| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.2, | |
| "eval_steps": 179, | |
| "global_step": 716, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005027932960893855, | |
| "grad_norm": 28.300092697143555, | |
| "learning_rate": 6.331471135940411e-07, | |
| "loss": 1.0198, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01005586592178771, | |
| "grad_norm": 1.059788465499878, | |
| "learning_rate": 1.303538175046555e-06, | |
| "loss": 0.8564, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.015083798882681564, | |
| "grad_norm": 9.368417739868164, | |
| "learning_rate": 1.9739292364990693e-06, | |
| "loss": 0.685, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.02011173184357542, | |
| "grad_norm": 3.4458816051483154, | |
| "learning_rate": 2.644320297951583e-06, | |
| "loss": 0.7667, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.025139664804469275, | |
| "grad_norm": 4.497154712677002, | |
| "learning_rate": 3.314711359404097e-06, | |
| "loss": 0.7907, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.030167597765363128, | |
| "grad_norm": 4.394736289978027, | |
| "learning_rate": 3.985102420856611e-06, | |
| "loss": 0.9719, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.03519553072625698, | |
| "grad_norm": 10.305765151977539, | |
| "learning_rate": 4.655493482309125e-06, | |
| "loss": 1.1097, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.04022346368715084, | |
| "grad_norm": 5.402688980102539, | |
| "learning_rate": 5.325884543761639e-06, | |
| "loss": 1.0463, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.045251396648044694, | |
| "grad_norm": 0.2811428904533386, | |
| "learning_rate": 5.996275605214154e-06, | |
| "loss": 0.8796, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_nli-pairs_loss": 0.9358150959014893, | |
| "eval_nli-pairs_runtime": 3.7163, | |
| "eval_nli-pairs_samples_per_second": 40.363, | |
| "eval_nli-pairs_steps_per_second": 0.538, | |
| "eval_sts-test_pearson_cosine": 0.7920907827756308, | |
| "eval_sts-test_pearson_dot": 0.553905804338793, | |
| "eval_sts-test_pearson_euclidean": 0.7477067417028286, | |
| "eval_sts-test_pearson_manhattan": 0.7454637009944656, | |
| "eval_sts-test_pearson_max": 0.7920907827756308, | |
| "eval_sts-test_spearman_cosine": 0.8050902340286232, | |
| "eval_sts-test_spearman_dot": 0.5325926730168974, | |
| "eval_sts-test_spearman_euclidean": 0.73705041377373, | |
| "eval_sts-test_spearman_manhattan": 0.7373939486807765, | |
| "eval_sts-test_spearman_max": 0.8050902340286232, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_vitaminc-pairs_loss": 4.57532262802124, | |
| "eval_vitaminc-pairs_runtime": 0.8386, | |
| "eval_vitaminc-pairs_samples_per_second": 149.066, | |
| "eval_vitaminc-pairs_steps_per_second": 2.385, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_qnli-contrastive_loss": 0.260235071182251, | |
| "eval_qnli-contrastive_runtime": 0.1227, | |
| "eval_qnli-contrastive_samples_per_second": 1222.824, | |
| "eval_qnli-contrastive_steps_per_second": 16.304, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_scitail-pairs-qa_loss": 0.06116430461406708, | |
| "eval_scitail-pairs-qa_runtime": 0.6972, | |
| "eval_scitail-pairs-qa_samples_per_second": 215.142, | |
| "eval_scitail-pairs-qa_steps_per_second": 2.869, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_scitail-pairs-pos_loss": 0.3001463711261749, | |
| "eval_scitail-pairs-pos_runtime": 1.8226, | |
| "eval_scitail-pairs-pos_samples_per_second": 82.299, | |
| "eval_scitail-pairs-pos_steps_per_second": 1.097, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_xsum-pairs_loss": 0.43996042013168335, | |
| "eval_xsum-pairs_runtime": 0.1548, | |
| "eval_xsum-pairs_samples_per_second": 529.555, | |
| "eval_xsum-pairs_steps_per_second": 6.458, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_compression-pairs_loss": 0.10088926553726196, | |
| "eval_compression-pairs_runtime": 0.1051, | |
| "eval_compression-pairs_samples_per_second": 1427.213, | |
| "eval_compression-pairs_steps_per_second": 19.03, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_sciq_pairs_loss": 0.15509434044361115, | |
| "eval_sciq_pairs_runtime": 4.9415, | |
| "eval_sciq_pairs_samples_per_second": 30.355, | |
| "eval_sciq_pairs_steps_per_second": 0.405, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_qasc_pairs_loss": 0.24072103202342987, | |
| "eval_qasc_pairs_runtime": 0.6453, | |
| "eval_qasc_pairs_samples_per_second": 232.455, | |
| "eval_qasc_pairs_steps_per_second": 3.099, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_openbookqa_pairs_loss": 1.5241280794143677, | |
| "eval_openbookqa_pairs_runtime": 0.7059, | |
| "eval_openbookqa_pairs_samples_per_second": 212.48, | |
| "eval_openbookqa_pairs_steps_per_second": 2.833, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_msmarco_pairs_loss": 0.6149919629096985, | |
| "eval_msmarco_pairs_runtime": 1.2013, | |
| "eval_msmarco_pairs_samples_per_second": 124.863, | |
| "eval_msmarco_pairs_steps_per_second": 1.665, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_nq_pairs_loss": 0.591522216796875, | |
| "eval_nq_pairs_runtime": 2.2917, | |
| "eval_nq_pairs_samples_per_second": 65.455, | |
| "eval_nq_pairs_steps_per_second": 0.873, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_trivia_pairs_loss": 1.0899044275283813, | |
| "eval_trivia_pairs_runtime": 3.1583, | |
| "eval_trivia_pairs_samples_per_second": 47.494, | |
| "eval_trivia_pairs_steps_per_second": 0.633, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_quora_pairs_loss": 0.21252885460853577, | |
| "eval_quora_pairs_runtime": 0.3861, | |
| "eval_quora_pairs_samples_per_second": 388.452, | |
| "eval_quora_pairs_steps_per_second": 5.179, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_gooaq_pairs_loss": 0.5593109130859375, | |
| "eval_gooaq_pairs_runtime": 0.7918, | |
| "eval_gooaq_pairs_samples_per_second": 189.449, | |
| "eval_gooaq_pairs_steps_per_second": 2.526, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_mrpc_pairs_loss": 0.055620357394218445, | |
| "eval_mrpc_pairs_runtime": 0.1048, | |
| "eval_mrpc_pairs_samples_per_second": 1431.291, | |
| "eval_mrpc_pairs_steps_per_second": 19.084, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.05027932960893855, | |
| "grad_norm": 9.769979476928711, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.7777, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0553072625698324, | |
| "grad_norm": 9.294490814208984, | |
| "learning_rate": 7.337057728119181e-06, | |
| "loss": 0.9075, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.060335195530726256, | |
| "grad_norm": 0.6179952621459961, | |
| "learning_rate": 8.007448789571696e-06, | |
| "loss": 0.8239, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.06536312849162011, | |
| "grad_norm": 6.369142532348633, | |
| "learning_rate": 8.677839851024209e-06, | |
| "loss": 0.949, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.07039106145251396, | |
| "grad_norm": 1.7066636085510254, | |
| "learning_rate": 9.348230912476724e-06, | |
| "loss": 1.2046, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.07541899441340782, | |
| "grad_norm": 0.33892622590065, | |
| "learning_rate": 1.0018621973929237e-05, | |
| "loss": 0.8526, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.08044692737430167, | |
| "grad_norm": 24.46626091003418, | |
| "learning_rate": 1.0689013035381753e-05, | |
| "loss": 0.8629, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.08547486033519552, | |
| "grad_norm": 9.228001594543457, | |
| "learning_rate": 1.1359404096834266e-05, | |
| "loss": 1.1474, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.09050279329608939, | |
| "grad_norm": 3.441566228866577, | |
| "learning_rate": 1.2029795158286779e-05, | |
| "loss": 0.9141, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.09553072625698324, | |
| "grad_norm": 0.3621586859226227, | |
| "learning_rate": 1.2700186219739294e-05, | |
| "loss": 0.9756, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_nli-pairs_loss": 0.9455356001853943, | |
| "eval_nli-pairs_runtime": 3.6115, | |
| "eval_nli-pairs_samples_per_second": 41.533, | |
| "eval_nli-pairs_steps_per_second": 0.554, | |
| "eval_sts-test_pearson_cosine": 0.7944563743777492, | |
| "eval_sts-test_pearson_dot": 0.554185693724103, | |
| "eval_sts-test_pearson_euclidean": 0.7483225064388481, | |
| "eval_sts-test_pearson_manhattan": 0.7454936340776467, | |
| "eval_sts-test_pearson_max": 0.7944563743777492, | |
| "eval_sts-test_spearman_cosine": 0.8064910811776218, | |
| "eval_sts-test_spearman_dot": 0.5333761592947532, | |
| "eval_sts-test_spearman_euclidean": 0.7382698586949932, | |
| "eval_sts-test_spearman_manhattan": 0.7378149008082869, | |
| "eval_sts-test_spearman_max": 0.8064910811776218, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_vitaminc-pairs_loss": 4.517832279205322, | |
| "eval_vitaminc-pairs_runtime": 0.8908, | |
| "eval_vitaminc-pairs_samples_per_second": 140.33, | |
| "eval_vitaminc-pairs_steps_per_second": 2.245, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_qnli-contrastive_loss": 0.2627922296524048, | |
| "eval_qnli-contrastive_runtime": 0.1229, | |
| "eval_qnli-contrastive_samples_per_second": 1220.886, | |
| "eval_qnli-contrastive_steps_per_second": 16.278, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_scitail-pairs-qa_loss": 0.06381849944591522, | |
| "eval_scitail-pairs-qa_runtime": 0.7125, | |
| "eval_scitail-pairs-qa_samples_per_second": 210.516, | |
| "eval_scitail-pairs-qa_steps_per_second": 2.807, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_scitail-pairs-pos_loss": 0.30350035429000854, | |
| "eval_scitail-pairs-pos_runtime": 1.877, | |
| "eval_scitail-pairs-pos_samples_per_second": 79.916, | |
| "eval_scitail-pairs-pos_steps_per_second": 1.066, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_xsum-pairs_loss": 0.44631534814834595, | |
| "eval_xsum-pairs_runtime": 0.1512, | |
| "eval_xsum-pairs_samples_per_second": 542.377, | |
| "eval_xsum-pairs_steps_per_second": 6.614, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_compression-pairs_loss": 0.10376789420843124, | |
| "eval_compression-pairs_runtime": 0.1034, | |
| "eval_compression-pairs_samples_per_second": 1450.105, | |
| "eval_compression-pairs_steps_per_second": 19.335, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_sciq_pairs_loss": 0.15615364909172058, | |
| "eval_sciq_pairs_runtime": 5.1236, | |
| "eval_sciq_pairs_samples_per_second": 29.276, | |
| "eval_sciq_pairs_steps_per_second": 0.39, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_qasc_pairs_loss": 0.24106402695178986, | |
| "eval_qasc_pairs_runtime": 0.6663, | |
| "eval_qasc_pairs_samples_per_second": 225.133, | |
| "eval_qasc_pairs_steps_per_second": 3.002, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_openbookqa_pairs_loss": 1.5405189990997314, | |
| "eval_openbookqa_pairs_runtime": 0.7365, | |
| "eval_openbookqa_pairs_samples_per_second": 203.658, | |
| "eval_openbookqa_pairs_steps_per_second": 2.715, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_msmarco_pairs_loss": 0.6067730188369751, | |
| "eval_msmarco_pairs_runtime": 1.2272, | |
| "eval_msmarco_pairs_samples_per_second": 122.233, | |
| "eval_msmarco_pairs_steps_per_second": 1.63, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_nq_pairs_loss": 0.5683658123016357, | |
| "eval_nq_pairs_runtime": 2.3358, | |
| "eval_nq_pairs_samples_per_second": 64.219, | |
| "eval_nq_pairs_steps_per_second": 0.856, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_trivia_pairs_loss": 1.1103042364120483, | |
| "eval_trivia_pairs_runtime": 3.2191, | |
| "eval_trivia_pairs_samples_per_second": 46.596, | |
| "eval_trivia_pairs_steps_per_second": 0.621, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_quora_pairs_loss": 0.2291153520345688, | |
| "eval_quora_pairs_runtime": 0.39, | |
| "eval_quora_pairs_samples_per_second": 384.605, | |
| "eval_quora_pairs_steps_per_second": 5.128, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_gooaq_pairs_loss": 0.5594914555549622, | |
| "eval_gooaq_pairs_runtime": 0.814, | |
| "eval_gooaq_pairs_samples_per_second": 184.276, | |
| "eval_gooaq_pairs_steps_per_second": 2.457, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_mrpc_pairs_loss": 0.056830935180187225, | |
| "eval_mrpc_pairs_runtime": 0.1099, | |
| "eval_mrpc_pairs_samples_per_second": 1364.673, | |
| "eval_mrpc_pairs_steps_per_second": 18.196, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.1005586592178771, | |
| "grad_norm": 0.40709275007247925, | |
| "learning_rate": 1.3370577281191808e-05, | |
| "loss": 0.7904, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.10558659217877095, | |
| "grad_norm": 8.056321144104004, | |
| "learning_rate": 1.4040968342644321e-05, | |
| "loss": 1.2543, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.1106145251396648, | |
| "grad_norm": 6.643378257751465, | |
| "learning_rate": 1.4711359404096834e-05, | |
| "loss": 0.854, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.11564245810055866, | |
| "grad_norm": 7.473087310791016, | |
| "learning_rate": 1.538175046554935e-05, | |
| "loss": 0.4719, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.12067039106145251, | |
| "grad_norm": 8.68940258026123, | |
| "learning_rate": 1.6052141527001864e-05, | |
| "loss": 1.4229, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.12569832402234637, | |
| "grad_norm": 8.564693450927734, | |
| "learning_rate": 1.6722532588454377e-05, | |
| "loss": 0.7646, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.13072625698324022, | |
| "grad_norm": 0.39765194058418274, | |
| "learning_rate": 1.7392923649906893e-05, | |
| "loss": 0.8512, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.13575418994413407, | |
| "grad_norm": 7.3941192626953125, | |
| "learning_rate": 1.8063314711359406e-05, | |
| "loss": 1.1333, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.14078212290502792, | |
| "grad_norm": 3.183729410171509, | |
| "learning_rate": 1.873370577281192e-05, | |
| "loss": 0.4451, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.1458100558659218, | |
| "grad_norm": 1.2496144771575928, | |
| "learning_rate": 1.9404096834264436e-05, | |
| "loss": 0.8859, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_nli-pairs_loss": 0.9459429979324341, | |
| "eval_nli-pairs_runtime": 3.418, | |
| "eval_nli-pairs_samples_per_second": 43.886, | |
| "eval_nli-pairs_steps_per_second": 0.585, | |
| "eval_sts-test_pearson_cosine": 0.7949855875446147, | |
| "eval_sts-test_pearson_dot": 0.5640384325880634, | |
| "eval_sts-test_pearson_euclidean": 0.751389277231376, | |
| "eval_sts-test_pearson_manhattan": 0.7487497192657817, | |
| "eval_sts-test_pearson_max": 0.7949855875446147, | |
| "eval_sts-test_spearman_cosine": 0.8064088099185094, | |
| "eval_sts-test_spearman_dot": 0.5419498621910633, | |
| "eval_sts-test_spearman_euclidean": 0.7411803691316964, | |
| "eval_sts-test_spearman_manhattan": 0.7408925147897409, | |
| "eval_sts-test_spearman_max": 0.8064088099185094, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_vitaminc-pairs_loss": 4.557371139526367, | |
| "eval_vitaminc-pairs_runtime": 0.8813, | |
| "eval_vitaminc-pairs_samples_per_second": 141.843, | |
| "eval_vitaminc-pairs_steps_per_second": 2.269, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_qnli-contrastive_loss": 0.2963089942932129, | |
| "eval_qnli-contrastive_runtime": 0.1173, | |
| "eval_qnli-contrastive_samples_per_second": 1279.24, | |
| "eval_qnli-contrastive_steps_per_second": 17.057, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_scitail-pairs-qa_loss": 0.06569412350654602, | |
| "eval_scitail-pairs-qa_runtime": 0.6754, | |
| "eval_scitail-pairs-qa_samples_per_second": 222.091, | |
| "eval_scitail-pairs-qa_steps_per_second": 2.961, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_scitail-pairs-pos_loss": 0.3104495108127594, | |
| "eval_scitail-pairs-pos_runtime": 1.7852, | |
| "eval_scitail-pairs-pos_samples_per_second": 84.022, | |
| "eval_scitail-pairs-pos_steps_per_second": 1.12, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_xsum-pairs_loss": 0.44305890798568726, | |
| "eval_xsum-pairs_runtime": 0.1668, | |
| "eval_xsum-pairs_samples_per_second": 491.658, | |
| "eval_xsum-pairs_steps_per_second": 5.996, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_compression-pairs_loss": 0.10611961036920547, | |
| "eval_compression-pairs_runtime": 0.1033, | |
| "eval_compression-pairs_samples_per_second": 1451.474, | |
| "eval_compression-pairs_steps_per_second": 19.353, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_sciq_pairs_loss": 0.15942400693893433, | |
| "eval_sciq_pairs_runtime": 4.8457, | |
| "eval_sciq_pairs_samples_per_second": 30.955, | |
| "eval_sciq_pairs_steps_per_second": 0.413, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_qasc_pairs_loss": 0.2380143702030182, | |
| "eval_qasc_pairs_runtime": 0.6528, | |
| "eval_qasc_pairs_samples_per_second": 229.78, | |
| "eval_qasc_pairs_steps_per_second": 3.064, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_openbookqa_pairs_loss": 1.556572437286377, | |
| "eval_openbookqa_pairs_runtime": 0.6905, | |
| "eval_openbookqa_pairs_samples_per_second": 217.246, | |
| "eval_openbookqa_pairs_steps_per_second": 2.897, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_msmarco_pairs_loss": 0.5950079560279846, | |
| "eval_msmarco_pairs_runtime": 1.1541, | |
| "eval_msmarco_pairs_samples_per_second": 129.968, | |
| "eval_msmarco_pairs_steps_per_second": 1.733, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_nq_pairs_loss": 0.5976797938346863, | |
| "eval_nq_pairs_runtime": 2.2683, | |
| "eval_nq_pairs_samples_per_second": 66.128, | |
| "eval_nq_pairs_steps_per_second": 0.882, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_trivia_pairs_loss": 1.1377041339874268, | |
| "eval_trivia_pairs_runtime": 3.1791, | |
| "eval_trivia_pairs_samples_per_second": 47.183, | |
| "eval_trivia_pairs_steps_per_second": 0.629, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_quora_pairs_loss": 0.20836791396141052, | |
| "eval_quora_pairs_runtime": 0.374, | |
| "eval_quora_pairs_samples_per_second": 401.115, | |
| "eval_quora_pairs_steps_per_second": 5.348, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_gooaq_pairs_loss": 0.5545207262039185, | |
| "eval_gooaq_pairs_runtime": 0.7667, | |
| "eval_gooaq_pairs_samples_per_second": 195.642, | |
| "eval_gooaq_pairs_steps_per_second": 2.609, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_mrpc_pairs_loss": 0.060290463268756866, | |
| "eval_mrpc_pairs_runtime": 0.1026, | |
| "eval_mrpc_pairs_samples_per_second": 1462.106, | |
| "eval_mrpc_pairs_steps_per_second": 19.495, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.15083798882681565, | |
| "grad_norm": 1.3315377235412598, | |
| "learning_rate": 1.9999898748390674e-05, | |
| "loss": 0.9245, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.1558659217877095, | |
| "grad_norm": 8.015869140625, | |
| "learning_rate": 1.9989876718445097e-05, | |
| "loss": 1.0384, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.16089385474860335, | |
| "grad_norm": 1.3641886711120605, | |
| "learning_rate": 1.9963472835306562e-05, | |
| "loss": 0.8826, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.1659217877094972, | |
| "grad_norm": 7.676197528839111, | |
| "learning_rate": 1.9920735213598995e-05, | |
| "loss": 0.6693, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.17094972067039105, | |
| "grad_norm": 8.144271850585938, | |
| "learning_rate": 1.9861741732192866e-05, | |
| "loss": 0.8437, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.17597765363128492, | |
| "grad_norm": 11.723029136657715, | |
| "learning_rate": 1.9786599892290035e-05, | |
| "loss": 1.2241, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.18100558659217877, | |
| "grad_norm": 12.107389450073242, | |
| "learning_rate": 1.9695446621529053e-05, | |
| "loss": 0.7169, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.18603351955307262, | |
| "grad_norm": 0.671999454498291, | |
| "learning_rate": 1.9588448024468015e-05, | |
| "loss": 0.7937, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.19106145251396647, | |
| "grad_norm": 2.9247829914093018, | |
| "learning_rate": 1.9465799079899608e-05, | |
| "loss": 0.8666, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.19608938547486032, | |
| "grad_norm": 9.634854316711426, | |
| "learning_rate": 1.932772328554989e-05, | |
| "loss": 0.8002, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_nli-pairs_loss": 0.9524073004722595, | |
| "eval_nli-pairs_runtime": 3.3652, | |
| "eval_nli-pairs_samples_per_second": 44.573, | |
| "eval_nli-pairs_steps_per_second": 0.594, | |
| "eval_sts-test_pearson_cosine": 0.7960974627777679, | |
| "eval_sts-test_pearson_dot": 0.5653927619143873, | |
| "eval_sts-test_pearson_euclidean": 0.7510324476182012, | |
| "eval_sts-test_pearson_manhattan": 0.7478581635465753, | |
| "eval_sts-test_pearson_max": 0.7960974627777679, | |
| "eval_sts-test_spearman_cosine": 0.8076217698074869, | |
| "eval_sts-test_spearman_dot": 0.5437275169754562, | |
| "eval_sts-test_spearman_euclidean": 0.7394612354069316, | |
| "eval_sts-test_spearman_manhattan": 0.7383580125266631, | |
| "eval_sts-test_spearman_max": 0.8076217698074869, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_vitaminc-pairs_loss": 4.544824600219727, | |
| "eval_vitaminc-pairs_runtime": 0.8209, | |
| "eval_vitaminc-pairs_samples_per_second": 152.27, | |
| "eval_vitaminc-pairs_steps_per_second": 2.436, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_qnli-contrastive_loss": 0.2854141891002655, | |
| "eval_qnli-contrastive_runtime": 0.116, | |
| "eval_qnli-contrastive_samples_per_second": 1293.335, | |
| "eval_qnli-contrastive_steps_per_second": 17.244, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_scitail-pairs-qa_loss": 0.06479831039905548, | |
| "eval_scitail-pairs-qa_runtime": 0.6668, | |
| "eval_scitail-pairs-qa_samples_per_second": 224.963, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.0, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_scitail-pairs-pos_loss": 0.30356064438819885, | |
| "eval_scitail-pairs-pos_runtime": 1.7895, | |
| "eval_scitail-pairs-pos_samples_per_second": 83.823, | |
| "eval_scitail-pairs-pos_steps_per_second": 1.118, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_xsum-pairs_loss": 0.4510384202003479, | |
| "eval_xsum-pairs_runtime": 0.1513, | |
| "eval_xsum-pairs_samples_per_second": 541.869, | |
| "eval_xsum-pairs_steps_per_second": 6.608, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_compression-pairs_loss": 0.10218393057584763, | |
| "eval_compression-pairs_runtime": 0.1063, | |
| "eval_compression-pairs_samples_per_second": 1410.495, | |
| "eval_compression-pairs_steps_per_second": 18.807, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_sciq_pairs_loss": 0.16117030382156372, | |
| "eval_sciq_pairs_runtime": 4.8842, | |
| "eval_sciq_pairs_samples_per_second": 30.711, | |
| "eval_sciq_pairs_steps_per_second": 0.409, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_qasc_pairs_loss": 0.25312450528144836, | |
| "eval_qasc_pairs_runtime": 0.626, | |
| "eval_qasc_pairs_samples_per_second": 239.619, | |
| "eval_qasc_pairs_steps_per_second": 3.195, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_openbookqa_pairs_loss": 1.560429573059082, | |
| "eval_openbookqa_pairs_runtime": 0.6856, | |
| "eval_openbookqa_pairs_samples_per_second": 218.787, | |
| "eval_openbookqa_pairs_steps_per_second": 2.917, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_msmarco_pairs_loss": 0.6264744997024536, | |
| "eval_msmarco_pairs_runtime": 1.1496, | |
| "eval_msmarco_pairs_samples_per_second": 130.479, | |
| "eval_msmarco_pairs_steps_per_second": 1.74, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_nq_pairs_loss": 0.602968692779541, | |
| "eval_nq_pairs_runtime": 2.2714, | |
| "eval_nq_pairs_samples_per_second": 66.038, | |
| "eval_nq_pairs_steps_per_second": 0.881, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_trivia_pairs_loss": 1.1763343811035156, | |
| "eval_trivia_pairs_runtime": 3.1779, | |
| "eval_trivia_pairs_samples_per_second": 47.201, | |
| "eval_trivia_pairs_steps_per_second": 0.629, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_quora_pairs_loss": 0.25559327006340027, | |
| "eval_quora_pairs_runtime": 0.3926, | |
| "eval_quora_pairs_samples_per_second": 382.056, | |
| "eval_quora_pairs_steps_per_second": 5.094, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_gooaq_pairs_loss": 0.5762031078338623, | |
| "eval_gooaq_pairs_runtime": 0.7962, | |
| "eval_gooaq_pairs_samples_per_second": 188.399, | |
| "eval_gooaq_pairs_steps_per_second": 2.512, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_mrpc_pairs_loss": 0.05901935696601868, | |
| "eval_mrpc_pairs_runtime": 0.1022, | |
| "eval_mrpc_pairs_samples_per_second": 1467.265, | |
| "eval_mrpc_pairs_steps_per_second": 19.564, | |
| "step": 716 | |
| } | |
| ], | |
| "logging_steps": 18, | |
| "max_steps": 7160, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 716, | |
| "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": 82, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |