| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.40102827763496146, | |
| "eval_steps": 26, | |
| "global_step": 208, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005784061696658098, | |
| "grad_norm": NaN, | |
| "learning_rate": 0.0, | |
| "loss": 7.8503, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.011568123393316195, | |
| "grad_norm": 127.71849060058594, | |
| "learning_rate": 5.791505791505792e-08, | |
| "loss": 8.4022, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.017352185089974295, | |
| "grad_norm": 244.984619140625, | |
| "learning_rate": 1.7374517374517373e-07, | |
| "loss": 11.1776, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02313624678663239, | |
| "grad_norm": 425.6427917480469, | |
| "learning_rate": 3.4749034749034746e-07, | |
| "loss": 9.7845, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02892030848329049, | |
| "grad_norm": 238.8634490966797, | |
| "learning_rate": 5.212355212355212e-07, | |
| "loss": 8.9224, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03470437017994859, | |
| "grad_norm": 173.4210968017578, | |
| "learning_rate": 6.949806949806949e-07, | |
| "loss": 11.1202, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04048843187660668, | |
| "grad_norm": 77.89308166503906, | |
| "learning_rate": 8.687258687258688e-07, | |
| "loss": 7.413, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04627249357326478, | |
| "grad_norm": Infinity, | |
| "learning_rate": 9.845559845559844e-07, | |
| "loss": 7.7803, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_Qnli-dev_cosine_accuracy": 0.609375, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.9602330923080444, | |
| "eval_Qnli-dev_cosine_ap": 0.587737015984062, | |
| "eval_Qnli-dev_cosine_f1": 0.6358695652173914, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.7741116285324097, | |
| "eval_Qnli-dev_cosine_precision": 0.468, | |
| "eval_Qnli-dev_cosine_recall": 0.9915254237288136, | |
| "eval_Qnli-dev_dot_accuracy": 0.56640625, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 750.123779296875, | |
| "eval_Qnli-dev_dot_ap": 0.5198069471819926, | |
| "eval_Qnli-dev_dot_f1": 0.6318607764390897, | |
| "eval_Qnli-dev_dot_f1_threshold": 387.37371826171875, | |
| "eval_Qnli-dev_dot_precision": 0.461839530332681, | |
| "eval_Qnli-dev_dot_recall": 1.0, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.6171875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 7.253280162811279, | |
| "eval_Qnli-dev_euclidean_ap": 0.5993562103857255, | |
| "eval_Qnli-dev_euclidean_f1": 0.6376021798365122, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 17.42431640625, | |
| "eval_Qnli-dev_euclidean_precision": 0.46987951807228917, | |
| "eval_Qnli-dev_euclidean_recall": 0.9915254237288136, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.625, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 212.0604705810547, | |
| "eval_Qnli-dev_manhattan_ap": 0.609409682371795, | |
| "eval_Qnli-dev_manhattan_f1": 0.6358695652173914, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 461.66192626953125, | |
| "eval_Qnli-dev_manhattan_precision": 0.468, | |
| "eval_Qnli-dev_manhattan_recall": 0.9915254237288136, | |
| "eval_Qnli-dev_max_accuracy": 0.625, | |
| "eval_Qnli-dev_max_accuracy_threshold": 750.123779296875, | |
| "eval_Qnli-dev_max_ap": 0.609409682371795, | |
| "eval_Qnli-dev_max_f1": 0.6376021798365122, | |
| "eval_Qnli-dev_max_f1_threshold": 461.66192626953125, | |
| "eval_Qnli-dev_max_precision": 0.46987951807228917, | |
| "eval_Qnli-dev_max_recall": 1.0, | |
| "eval_allNLI-dev_cosine_accuracy": 0.666015625, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.9931652545928955, | |
| "eval_allNLI-dev_cosine_ap": 0.3330277035108763, | |
| "eval_allNLI-dev_cosine_f1": 0.5088235294117648, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.40660202503204346, | |
| "eval_allNLI-dev_cosine_precision": 0.34122287968441817, | |
| "eval_allNLI-dev_cosine_recall": 1.0, | |
| "eval_allNLI-dev_dot_accuracy": 0.662109375, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 810.7718505859375, | |
| "eval_allNLI-dev_dot_ap": 0.2980492563498839, | |
| "eval_allNLI-dev_dot_f1": 0.5088235294117648, | |
| "eval_allNLI-dev_dot_f1_threshold": 294.67071533203125, | |
| "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": 3.3851852416992188, | |
| "eval_allNLI-dev_euclidean_ap": 0.33790858355605125, | |
| "eval_allNLI-dev_euclidean_f1": 0.5095729013254787, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 28.78042984008789, | |
| "eval_allNLI-dev_euclidean_precision": 0.34189723320158105, | |
| "eval_allNLI-dev_euclidean_recall": 1.0, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.666015625, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 98.2369613647461, | |
| "eval_allNLI-dev_manhattan_ap": 0.3452833632811151, | |
| "eval_allNLI-dev_manhattan_f1": 0.5091463414634146, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 626.5145874023438, | |
| "eval_allNLI-dev_manhattan_precision": 0.34575569358178054, | |
| "eval_allNLI-dev_manhattan_recall": 0.9653179190751445, | |
| "eval_allNLI-dev_max_accuracy": 0.666015625, | |
| "eval_allNLI-dev_max_accuracy_threshold": 810.7718505859375, | |
| "eval_allNLI-dev_max_ap": 0.3452833632811151, | |
| "eval_allNLI-dev_max_f1": 0.5095729013254787, | |
| "eval_allNLI-dev_max_f1_threshold": 626.5145874023438, | |
| "eval_allNLI-dev_max_precision": 0.34575569358178054, | |
| "eval_allNLI-dev_max_recall": 1.0, | |
| "eval_sequential_score": 0.609409682371795, | |
| "eval_sts-test_pearson_cosine": 0.24255956652335486, | |
| "eval_sts-test_pearson_dot": 0.18022025813169615, | |
| "eval_sts-test_pearson_euclidean": 0.28657386934462287, | |
| "eval_sts-test_pearson_manhattan": 0.32587411772454034, | |
| "eval_sts-test_pearson_max": 0.32587411772454034, | |
| "eval_sts-test_spearman_cosine": 0.32225123770630715, | |
| "eval_sts-test_spearman_dot": 0.17184726031108927, | |
| "eval_sts-test_spearman_euclidean": 0.3279675400801376, | |
| "eval_sts-test_spearman_manhattan": 0.3635547276663291, | |
| "eval_sts-test_spearman_max": 0.3635547276663291, | |
| "eval_vitaminc-pairs_loss": 4.122440338134766, | |
| "eval_vitaminc-pairs_runtime": 4.7337, | |
| "eval_vitaminc-pairs_samples_per_second": 27.04, | |
| "eval_vitaminc-pairs_steps_per_second": 0.211, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_negation-triplets_loss": 3.917590379714966, | |
| "eval_negation-triplets_runtime": 3.3679, | |
| "eval_negation-triplets_samples_per_second": 38.006, | |
| "eval_negation-triplets_steps_per_second": 0.297, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_scitail-pairs-pos_loss": 2.5062460899353027, | |
| "eval_scitail-pairs-pos_runtime": 2.7274, | |
| "eval_scitail-pairs-pos_samples_per_second": 46.931, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.367, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_scitail-pairs-qa_loss": 2.978612184524536, | |
| "eval_scitail-pairs-qa_runtime": 2.2452, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.011, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.445, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_xsum-pairs_loss": 5.394700050354004, | |
| "eval_xsum-pairs_runtime": 3.0886, | |
| "eval_xsum-pairs_samples_per_second": 41.442, | |
| "eval_xsum-pairs_steps_per_second": 0.324, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_sciq_pairs_loss": 0.8977569937705994, | |
| "eval_sciq_pairs_runtime": 4.7516, | |
| "eval_sciq_pairs_samples_per_second": 26.938, | |
| "eval_sciq_pairs_steps_per_second": 0.21, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_qasc_pairs_loss": 6.6808977127075195, | |
| "eval_qasc_pairs_runtime": 2.1087, | |
| "eval_qasc_pairs_samples_per_second": 60.701, | |
| "eval_qasc_pairs_steps_per_second": 0.474, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_openbookqa_pairs_loss": 6.402599334716797, | |
| "eval_openbookqa_pairs_runtime": 2.2222, | |
| "eval_openbookqa_pairs_samples_per_second": 57.601, | |
| "eval_openbookqa_pairs_steps_per_second": 0.45, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_msmarco_pairs_loss": 15.782674789428711, | |
| "eval_msmarco_pairs_runtime": 2.307, | |
| "eval_msmarco_pairs_samples_per_second": 55.483, | |
| "eval_msmarco_pairs_steps_per_second": 0.433, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_nq_pairs_loss": 10.447425842285156, | |
| "eval_nq_pairs_runtime": 3.6239, | |
| "eval_nq_pairs_samples_per_second": 35.321, | |
| "eval_nq_pairs_steps_per_second": 0.276, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_trivia_pairs_loss": 6.107333183288574, | |
| "eval_trivia_pairs_runtime": 3.1299, | |
| "eval_trivia_pairs_samples_per_second": 40.895, | |
| "eval_trivia_pairs_steps_per_second": 0.319, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_gooaq_pairs_loss": 8.095273971557617, | |
| "eval_gooaq_pairs_runtime": 2.1098, | |
| "eval_gooaq_pairs_samples_per_second": 60.669, | |
| "eval_gooaq_pairs_steps_per_second": 0.474, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_paws-pos_loss": 1.6077361106872559, | |
| "eval_paws-pos_runtime": 2.2412, | |
| "eval_paws-pos_samples_per_second": 57.112, | |
| "eval_paws-pos_steps_per_second": 0.446, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05012853470437018, | |
| "eval_global_dataset_loss": 6.348417282104492, | |
| "eval_global_dataset_runtime": 8.3683, | |
| "eval_global_dataset_samples_per_second": 38.837, | |
| "eval_global_dataset_steps_per_second": 0.358, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05205655526992288, | |
| "grad_norm": 172.1571807861328, | |
| "learning_rate": 1.1583011583011583e-06, | |
| "loss": 7.9729, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.05784061696658098, | |
| "grad_norm": 106.302001953125, | |
| "learning_rate": 1.3320463320463321e-06, | |
| "loss": 6.0587, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06362467866323908, | |
| "grad_norm": 72.31057739257812, | |
| "learning_rate": 1.5057915057915057e-06, | |
| "loss": 5.6742, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06940874035989718, | |
| "grad_norm": 258.2577209472656, | |
| "learning_rate": 1.6795366795366796e-06, | |
| "loss": 6.5406, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07519280205655526, | |
| "grad_norm": 156.0019073486328, | |
| "learning_rate": 1.7953667953667953e-06, | |
| "loss": 5.4429, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.08097686375321336, | |
| "grad_norm": 227.6062469482422, | |
| "learning_rate": 1.969111969111969e-06, | |
| "loss": 6.7855, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.08676092544987146, | |
| "grad_norm": 221.12347412109375, | |
| "learning_rate": 2.1428571428571427e-06, | |
| "loss": 5.3403, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.09254498714652956, | |
| "grad_norm": 363.6319274902344, | |
| "learning_rate": 2.3166023166023166e-06, | |
| "loss": 4.2282, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09832904884318766, | |
| "grad_norm": 119.22310638427734, | |
| "learning_rate": 2.4903474903474904e-06, | |
| "loss": 4.7411, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_Qnli-dev_cosine_accuracy": 0.58984375, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.9407284259796143, | |
| "eval_Qnli-dev_cosine_ap": 0.5957496162735323, | |
| "eval_Qnli-dev_cosine_f1": 0.6314363143631436, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.7475543022155762, | |
| "eval_Qnli-dev_cosine_precision": 0.4641434262948207, | |
| "eval_Qnli-dev_cosine_recall": 0.9872881355932204, | |
| "eval_Qnli-dev_dot_accuracy": 0.591796875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 787.171630859375, | |
| "eval_Qnli-dev_dot_ap": 0.556540262374452, | |
| "eval_Qnli-dev_dot_f1": 0.6328767123287672, | |
| "eval_Qnli-dev_dot_f1_threshold": 576.0565795898438, | |
| "eval_Qnli-dev_dot_precision": 0.4676113360323887, | |
| "eval_Qnli-dev_dot_recall": 0.9788135593220338, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.587890625, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 9.92269515991211, | |
| "eval_Qnli-dev_euclidean_ap": 0.5913529071853763, | |
| "eval_Qnli-dev_euclidean_f1": 0.6322930800542741, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 19.546676635742188, | |
| "eval_Qnli-dev_euclidean_precision": 0.46506986027944114, | |
| "eval_Qnli-dev_euclidean_recall": 0.9872881355932204, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.591796875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 314.9248352050781, | |
| "eval_Qnli-dev_manhattan_ap": 0.5944730102616272, | |
| "eval_Qnli-dev_manhattan_f1": 0.632952691680261, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 424.9676818847656, | |
| "eval_Qnli-dev_manhattan_precision": 0.5145888594164456, | |
| "eval_Qnli-dev_manhattan_recall": 0.8220338983050848, | |
| "eval_Qnli-dev_max_accuracy": 0.591796875, | |
| "eval_Qnli-dev_max_accuracy_threshold": 787.171630859375, | |
| "eval_Qnli-dev_max_ap": 0.5957496162735323, | |
| "eval_Qnli-dev_max_f1": 0.632952691680261, | |
| "eval_Qnli-dev_max_f1_threshold": 576.0565795898438, | |
| "eval_Qnli-dev_max_precision": 0.5145888594164456, | |
| "eval_Qnli-dev_max_recall": 0.9872881355932204, | |
| "eval_allNLI-dev_cosine_accuracy": 0.666015625, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.9878466129302979, | |
| "eval_allNLI-dev_cosine_ap": 0.3898337405414008, | |
| "eval_allNLI-dev_cosine_f1": 0.5065885797950219, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6534674167633057, | |
| "eval_allNLI-dev_cosine_precision": 0.3392156862745098, | |
| "eval_allNLI-dev_cosine_recall": 1.0, | |
| "eval_allNLI-dev_dot_accuracy": 0.666015625, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 925.5989990234375, | |
| "eval_allNLI-dev_dot_ap": 0.3391870470132937, | |
| "eval_allNLI-dev_dot_f1": 0.5046153846153847, | |
| "eval_allNLI-dev_dot_f1_threshold": 608.7012939453125, | |
| "eval_allNLI-dev_dot_precision": 0.3438155136268344, | |
| "eval_allNLI-dev_dot_recall": 0.9479768786127167, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.666015625, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 4.708977699279785, | |
| "eval_allNLI-dev_euclidean_ap": 0.39556533875401867, | |
| "eval_allNLI-dev_euclidean_f1": 0.5080763582966226, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 23.160057067871094, | |
| "eval_allNLI-dev_euclidean_precision": 0.3405511811023622, | |
| "eval_allNLI-dev_euclidean_recall": 1.0, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.666015625, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 142.7260284423828, | |
| "eval_allNLI-dev_manhattan_ap": 0.4050661514301679, | |
| "eval_allNLI-dev_manhattan_f1": 0.5118483412322274, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 504.7154846191406, | |
| "eval_allNLI-dev_manhattan_precision": 0.3521739130434783, | |
| "eval_allNLI-dev_manhattan_recall": 0.9364161849710982, | |
| "eval_allNLI-dev_max_accuracy": 0.666015625, | |
| "eval_allNLI-dev_max_accuracy_threshold": 925.5989990234375, | |
| "eval_allNLI-dev_max_ap": 0.4050661514301679, | |
| "eval_allNLI-dev_max_f1": 0.5118483412322274, | |
| "eval_allNLI-dev_max_f1_threshold": 608.7012939453125, | |
| "eval_allNLI-dev_max_precision": 0.3521739130434783, | |
| "eval_allNLI-dev_max_recall": 1.0, | |
| "eval_sequential_score": 0.5957496162735323, | |
| "eval_sts-test_pearson_cosine": 0.5067228548513025, | |
| "eval_sts-test_pearson_dot": 0.28867418819353924, | |
| "eval_sts-test_pearson_euclidean": 0.5315956249457565, | |
| "eval_sts-test_pearson_manhattan": 0.5620907203006935, | |
| "eval_sts-test_pearson_max": 0.5620907203006935, | |
| "eval_sts-test_spearman_cosine": 0.5257409961657202, | |
| "eval_sts-test_spearman_dot": 0.280153994622986, | |
| "eval_sts-test_spearman_euclidean": 0.5306475244646732, | |
| "eval_sts-test_spearman_manhattan": 0.5605804796765843, | |
| "eval_sts-test_spearman_max": 0.5605804796765843, | |
| "eval_vitaminc-pairs_loss": 3.929797887802124, | |
| "eval_vitaminc-pairs_runtime": 4.6018, | |
| "eval_vitaminc-pairs_samples_per_second": 27.815, | |
| "eval_vitaminc-pairs_steps_per_second": 0.217, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_negation-triplets_loss": 3.267047166824341, | |
| "eval_negation-triplets_runtime": 3.2813, | |
| "eval_negation-triplets_samples_per_second": 39.009, | |
| "eval_negation-triplets_steps_per_second": 0.305, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_scitail-pairs-pos_loss": 1.134178876876831, | |
| "eval_scitail-pairs-pos_runtime": 2.6261, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.742, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.381, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_scitail-pairs-qa_loss": 2.0085601806640625, | |
| "eval_scitail-pairs-qa_runtime": 2.2161, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.759, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.451, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_xsum-pairs_loss": 4.462350845336914, | |
| "eval_xsum-pairs_runtime": 3.1126, | |
| "eval_xsum-pairs_samples_per_second": 41.124, | |
| "eval_xsum-pairs_steps_per_second": 0.321, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_sciq_pairs_loss": 0.71834796667099, | |
| "eval_sciq_pairs_runtime": 4.7505, | |
| "eval_sciq_pairs_samples_per_second": 26.944, | |
| "eval_sciq_pairs_steps_per_second": 0.211, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_qasc_pairs_loss": 4.889453887939453, | |
| "eval_qasc_pairs_runtime": 2.1101, | |
| "eval_qasc_pairs_samples_per_second": 60.662, | |
| "eval_qasc_pairs_steps_per_second": 0.474, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_openbookqa_pairs_loss": 3.4273369312286377, | |
| "eval_openbookqa_pairs_runtime": 2.2311, | |
| "eval_openbookqa_pairs_samples_per_second": 57.371, | |
| "eval_openbookqa_pairs_steps_per_second": 0.448, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_msmarco_pairs_loss": 7.532540321350098, | |
| "eval_msmarco_pairs_runtime": 2.2959, | |
| "eval_msmarco_pairs_samples_per_second": 55.751, | |
| "eval_msmarco_pairs_steps_per_second": 0.436, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_nq_pairs_loss": 6.29541015625, | |
| "eval_nq_pairs_runtime": 3.661, | |
| "eval_nq_pairs_samples_per_second": 34.963, | |
| "eval_nq_pairs_steps_per_second": 0.273, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_trivia_pairs_loss": 5.789977073669434, | |
| "eval_trivia_pairs_runtime": 3.0826, | |
| "eval_trivia_pairs_samples_per_second": 41.523, | |
| "eval_trivia_pairs_steps_per_second": 0.324, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_gooaq_pairs_loss": 5.785163879394531, | |
| "eval_gooaq_pairs_runtime": 2.1382, | |
| "eval_gooaq_pairs_samples_per_second": 59.863, | |
| "eval_gooaq_pairs_steps_per_second": 0.468, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_paws-pos_loss": 0.47830095887184143, | |
| "eval_paws-pos_runtime": 2.3113, | |
| "eval_paws-pos_samples_per_second": 55.379, | |
| "eval_paws-pos_steps_per_second": 0.433, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10025706940874037, | |
| "eval_global_dataset_loss": 3.8913991451263428, | |
| "eval_global_dataset_runtime": 8.3623, | |
| "eval_global_dataset_samples_per_second": 38.865, | |
| "eval_global_dataset_steps_per_second": 0.359, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10411311053984576, | |
| "grad_norm": 55.83478546142578, | |
| "learning_rate": 2.6640926640926642e-06, | |
| "loss": 3.9082, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.10989717223650386, | |
| "grad_norm": 57.4188117980957, | |
| "learning_rate": 2.7799227799227797e-06, | |
| "loss": 4.3922, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11568123393316196, | |
| "grad_norm": 116.89701843261719, | |
| "learning_rate": 2.9536679536679535e-06, | |
| "loss": 3.2655, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12146529562982006, | |
| "grad_norm": 53.49580001831055, | |
| "learning_rate": 3.1274131274131274e-06, | |
| "loss": 3.1043, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12724935732647816, | |
| "grad_norm": 51.044410705566406, | |
| "learning_rate": 3.3011583011583012e-06, | |
| "loss": 2.2074, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.13303341902313626, | |
| "grad_norm": 139.9288787841797, | |
| "learning_rate": 3.474903474903475e-06, | |
| "loss": 1.4414, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13881748071979436, | |
| "grad_norm": 37.221168518066406, | |
| "learning_rate": 3.5907335907335905e-06, | |
| "loss": 1.5937, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.14460154241645246, | |
| "grad_norm": 25.855661392211914, | |
| "learning_rate": 3.7644787644787644e-06, | |
| "loss": 1.0306, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "grad_norm": 29.79730796813965, | |
| "learning_rate": 3.938223938223938e-06, | |
| "loss": 1.0784, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_Qnli-dev_cosine_accuracy": 0.638671875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.856202244758606, | |
| "eval_Qnli-dev_cosine_ap": 0.6292803059732273, | |
| "eval_Qnli-dev_cosine_f1": 0.6408839779005525, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6206527352333069, | |
| "eval_Qnli-dev_cosine_precision": 0.47540983606557374, | |
| "eval_Qnli-dev_cosine_recall": 0.9830508474576272, | |
| "eval_Qnli-dev_dot_accuracy": 0.626953125, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 887.830810546875, | |
| "eval_Qnli-dev_dot_ap": 0.592075047542057, | |
| "eval_Qnli-dev_dot_f1": 0.640990371389271, | |
| "eval_Qnli-dev_dot_f1_threshold": 601.355224609375, | |
| "eval_Qnli-dev_dot_precision": 0.4745417515274949, | |
| "eval_Qnli-dev_dot_recall": 0.9872881355932204, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.638671875, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 17.094371795654297, | |
| "eval_Qnli-dev_euclidean_ap": 0.6289192660680781, | |
| "eval_Qnli-dev_euclidean_f1": 0.6423357664233577, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 24.894683837890625, | |
| "eval_Qnli-dev_euclidean_precision": 0.48997772828507796, | |
| "eval_Qnli-dev_euclidean_recall": 0.9322033898305084, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.640625, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 526.0111083984375, | |
| "eval_Qnli-dev_manhattan_ap": 0.6256143574081243, | |
| "eval_Qnli-dev_manhattan_f1": 0.6417704011065007, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 853.242431640625, | |
| "eval_Qnli-dev_manhattan_precision": 0.47638603696098564, | |
| "eval_Qnli-dev_manhattan_recall": 0.9830508474576272, | |
| "eval_Qnli-dev_max_accuracy": 0.640625, | |
| "eval_Qnli-dev_max_accuracy_threshold": 887.830810546875, | |
| "eval_Qnli-dev_max_ap": 0.6292803059732273, | |
| "eval_Qnli-dev_max_f1": 0.6423357664233577, | |
| "eval_Qnli-dev_max_f1_threshold": 853.242431640625, | |
| "eval_Qnli-dev_max_precision": 0.48997772828507796, | |
| "eval_Qnli-dev_max_recall": 0.9872881355932204, | |
| "eval_allNLI-dev_cosine_accuracy": 0.6796875, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.9426460862159729, | |
| "eval_allNLI-dev_cosine_ap": 0.473508692504868, | |
| "eval_allNLI-dev_cosine_f1": 0.5328947368421053, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6884603500366211, | |
| "eval_allNLI-dev_cosine_precision": 0.3724137931034483, | |
| "eval_allNLI-dev_cosine_recall": 0.9364161849710982, | |
| "eval_allNLI-dev_dot_accuracy": 0.673828125, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 980.9593505859375, | |
| "eval_allNLI-dev_dot_ap": 0.44188020452374555, | |
| "eval_allNLI-dev_dot_f1": 0.53156146179402, | |
| "eval_allNLI-dev_dot_f1_threshold": 683.7774658203125, | |
| "eval_allNLI-dev_dot_precision": 0.372960372960373, | |
| "eval_allNLI-dev_dot_recall": 0.9248554913294798, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.6796875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 10.647405624389648, | |
| "eval_allNLI-dev_euclidean_ap": 0.47440611686548606, | |
| "eval_allNLI-dev_euclidean_f1": 0.5292207792207793, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 25.08370590209961, | |
| "eval_allNLI-dev_euclidean_precision": 0.36794582392776526, | |
| "eval_allNLI-dev_euclidean_recall": 0.9421965317919075, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.6796875, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 318.408203125, | |
| "eval_allNLI-dev_manhattan_ap": 0.4747137405292771, | |
| "eval_allNLI-dev_manhattan_f1": 0.5305343511450381, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 697.8466796875, | |
| "eval_allNLI-dev_manhattan_precision": 0.396011396011396, | |
| "eval_allNLI-dev_manhattan_recall": 0.8034682080924855, | |
| "eval_allNLI-dev_max_accuracy": 0.6796875, | |
| "eval_allNLI-dev_max_accuracy_threshold": 980.9593505859375, | |
| "eval_allNLI-dev_max_ap": 0.4747137405292771, | |
| "eval_allNLI-dev_max_f1": 0.5328947368421053, | |
| "eval_allNLI-dev_max_f1_threshold": 697.8466796875, | |
| "eval_allNLI-dev_max_precision": 0.396011396011396, | |
| "eval_allNLI-dev_max_recall": 0.9421965317919075, | |
| "eval_sequential_score": 0.6292803059732273, | |
| "eval_sts-test_pearson_cosine": 0.8216305666941104, | |
| "eval_sts-test_pearson_dot": 0.7893058557928705, | |
| "eval_sts-test_pearson_euclidean": 0.841701309677291, | |
| "eval_sts-test_pearson_manhattan": 0.8438559225680149, | |
| "eval_sts-test_pearson_max": 0.8438559225680149, | |
| "eval_sts-test_spearman_cosine": 0.8291168678373173, | |
| "eval_sts-test_spearman_dot": 0.7782024179758329, | |
| "eval_sts-test_spearman_euclidean": 0.829029110613729, | |
| "eval_sts-test_spearman_manhattan": 0.8308224279114665, | |
| "eval_sts-test_spearman_max": 0.8308224279114665, | |
| "eval_vitaminc-pairs_loss": 3.763850212097168, | |
| "eval_vitaminc-pairs_runtime": 4.6142, | |
| "eval_vitaminc-pairs_samples_per_second": 27.741, | |
| "eval_vitaminc-pairs_steps_per_second": 0.217, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_negation-triplets_loss": 1.4772971868515015, | |
| "eval_negation-triplets_runtime": 3.4145, | |
| "eval_negation-triplets_samples_per_second": 37.488, | |
| "eval_negation-triplets_steps_per_second": 0.293, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_scitail-pairs-pos_loss": 0.11086193472146988, | |
| "eval_scitail-pairs-pos_runtime": 2.8091, | |
| "eval_scitail-pairs-pos_samples_per_second": 45.565, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.356, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_scitail-pairs-qa_loss": 0.12822097539901733, | |
| "eval_scitail-pairs-qa_runtime": 2.312, | |
| "eval_scitail-pairs-qa_samples_per_second": 55.364, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.433, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_xsum-pairs_loss": 0.43598121404647827, | |
| "eval_xsum-pairs_runtime": 3.1471, | |
| "eval_xsum-pairs_samples_per_second": 40.673, | |
| "eval_xsum-pairs_steps_per_second": 0.318, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_sciq_pairs_loss": 0.20338551700115204, | |
| "eval_sciq_pairs_runtime": 4.8592, | |
| "eval_sciq_pairs_samples_per_second": 26.342, | |
| "eval_sciq_pairs_steps_per_second": 0.206, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_qasc_pairs_loss": 1.3059492111206055, | |
| "eval_qasc_pairs_runtime": 2.1191, | |
| "eval_qasc_pairs_samples_per_second": 60.403, | |
| "eval_qasc_pairs_steps_per_second": 0.472, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_openbookqa_pairs_loss": 1.1711227893829346, | |
| "eval_openbookqa_pairs_runtime": 2.2105, | |
| "eval_openbookqa_pairs_samples_per_second": 57.907, | |
| "eval_openbookqa_pairs_steps_per_second": 0.452, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_msmarco_pairs_loss": 1.3856205940246582, | |
| "eval_msmarco_pairs_runtime": 2.3065, | |
| "eval_msmarco_pairs_samples_per_second": 55.494, | |
| "eval_msmarco_pairs_steps_per_second": 0.434, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_nq_pairs_loss": 1.2713654041290283, | |
| "eval_nq_pairs_runtime": 3.6729, | |
| "eval_nq_pairs_samples_per_second": 34.85, | |
| "eval_nq_pairs_steps_per_second": 0.272, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_trivia_pairs_loss": 1.082784652709961, | |
| "eval_trivia_pairs_runtime": 3.145, | |
| "eval_trivia_pairs_samples_per_second": 40.7, | |
| "eval_trivia_pairs_steps_per_second": 0.318, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_gooaq_pairs_loss": 0.703183650970459, | |
| "eval_gooaq_pairs_runtime": 2.1975, | |
| "eval_gooaq_pairs_samples_per_second": 58.248, | |
| "eval_gooaq_pairs_steps_per_second": 0.455, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_paws-pos_loss": 0.04489962384104729, | |
| "eval_paws-pos_runtime": 2.3636, | |
| "eval_paws-pos_samples_per_second": 54.155, | |
| "eval_paws-pos_steps_per_second": 0.423, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15038560411311053, | |
| "eval_global_dataset_loss": 0.6713590621948242, | |
| "eval_global_dataset_runtime": 8.4648, | |
| "eval_global_dataset_samples_per_second": 38.394, | |
| "eval_global_dataset_steps_per_second": 0.354, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15616966580976863, | |
| "grad_norm": 22.163389205932617, | |
| "learning_rate": 4.1119691119691125e-06, | |
| "loss": 0.9674, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.16195372750642673, | |
| "grad_norm": 25.121131896972656, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 0.9335, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.16773778920308482, | |
| "grad_norm": 44.308528900146484, | |
| "learning_rate": 4.45945945945946e-06, | |
| "loss": 0.8806, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.17352185089974292, | |
| "grad_norm": 9.972362518310547, | |
| "learning_rate": 4.633204633204633e-06, | |
| "loss": 0.631, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17930591259640102, | |
| "grad_norm": 13.609597206115723, | |
| "learning_rate": 4.806949806949808e-06, | |
| "loss": 0.3384, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.18508997429305912, | |
| "grad_norm": 6.892751693725586, | |
| "learning_rate": 4.980694980694981e-06, | |
| "loss": 0.404, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.19087403598971722, | |
| "grad_norm": 7.414526462554932, | |
| "learning_rate": 5.154440154440155e-06, | |
| "loss": 0.6488, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.19665809768637532, | |
| "grad_norm": 23.75099754333496, | |
| "learning_rate": 5.3281853281853285e-06, | |
| "loss": 0.4728, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_Qnli-dev_cosine_accuracy": 0.693359375, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.8276629447937012, | |
| "eval_Qnli-dev_cosine_ap": 0.7127503402574797, | |
| "eval_Qnli-dev_cosine_f1": 0.6730038022813688, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.7761298418045044, | |
| "eval_Qnli-dev_cosine_precision": 0.6103448275862069, | |
| "eval_Qnli-dev_cosine_recall": 0.75, | |
| "eval_Qnli-dev_dot_accuracy": 0.681640625, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 857.5540771484375, | |
| "eval_Qnli-dev_dot_ap": 0.6617256136301952, | |
| "eval_Qnli-dev_dot_f1": 0.6678141135972461, | |
| "eval_Qnli-dev_dot_f1_threshold": 756.1309204101562, | |
| "eval_Qnli-dev_dot_precision": 0.5623188405797102, | |
| "eval_Qnli-dev_dot_recall": 0.8220338983050848, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.69140625, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 18.45541763305664, | |
| "eval_Qnli-dev_euclidean_ap": 0.7154480594709788, | |
| "eval_Qnli-dev_euclidean_f1": 0.6725043782837128, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 22.723068237304688, | |
| "eval_Qnli-dev_euclidean_precision": 0.573134328358209, | |
| "eval_Qnli-dev_euclidean_recall": 0.8135593220338984, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.69140625, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 585.3442993164062, | |
| "eval_Qnli-dev_manhattan_ap": 0.713289230749026, | |
| "eval_Qnli-dev_manhattan_f1": 0.6733333333333333, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 733.2077026367188, | |
| "eval_Qnli-dev_manhattan_precision": 0.554945054945055, | |
| "eval_Qnli-dev_manhattan_recall": 0.8559322033898306, | |
| "eval_Qnli-dev_max_accuracy": 0.693359375, | |
| "eval_Qnli-dev_max_accuracy_threshold": 857.5540771484375, | |
| "eval_Qnli-dev_max_ap": 0.7154480594709788, | |
| "eval_Qnli-dev_max_f1": 0.6733333333333333, | |
| "eval_Qnli-dev_max_f1_threshold": 756.1309204101562, | |
| "eval_Qnli-dev_max_precision": 0.6103448275862069, | |
| "eval_Qnli-dev_max_recall": 0.8559322033898306, | |
| "eval_allNLI-dev_cosine_accuracy": 0.693359375, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8820043802261353, | |
| "eval_allNLI-dev_cosine_ap": 0.5380653098565689, | |
| "eval_allNLI-dev_cosine_f1": 0.5689655172413792, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7308224439620972, | |
| "eval_allNLI-dev_cosine_precision": 0.4536082474226804, | |
| "eval_allNLI-dev_cosine_recall": 0.7630057803468208, | |
| "eval_allNLI-dev_dot_accuracy": 0.6875, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 904.0134887695312, | |
| "eval_allNLI-dev_dot_ap": 0.5175334662627952, | |
| "eval_allNLI-dev_dot_f1": 0.5675675675675675, | |
| "eval_allNLI-dev_dot_f1_threshold": 762.652587890625, | |
| "eval_allNLI-dev_dot_precision": 0.46494464944649444, | |
| "eval_allNLI-dev_dot_recall": 0.7283236994219653, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.6953125, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 15.50227165222168, | |
| "eval_allNLI-dev_euclidean_ap": 0.5410854453600531, | |
| "eval_allNLI-dev_euclidean_f1": 0.5645514223194749, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 23.451454162597656, | |
| "eval_allNLI-dev_euclidean_precision": 0.45422535211267606, | |
| "eval_allNLI-dev_euclidean_recall": 0.7456647398843931, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.6953125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 518.75146484375, | |
| "eval_allNLI-dev_manhattan_ap": 0.5430149701961973, | |
| "eval_allNLI-dev_manhattan_f1": 0.5639913232104121, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 733.8897705078125, | |
| "eval_allNLI-dev_manhattan_precision": 0.4513888888888889, | |
| "eval_allNLI-dev_manhattan_recall": 0.7514450867052023, | |
| "eval_allNLI-dev_max_accuracy": 0.6953125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 904.0134887695312, | |
| "eval_allNLI-dev_max_ap": 0.5430149701961973, | |
| "eval_allNLI-dev_max_f1": 0.5689655172413792, | |
| "eval_allNLI-dev_max_f1_threshold": 762.652587890625, | |
| "eval_allNLI-dev_max_precision": 0.46494464944649444, | |
| "eval_allNLI-dev_max_recall": 0.7630057803468208, | |
| "eval_sequential_score": 0.7154480594709788, | |
| "eval_sts-test_pearson_cosine": 0.8841456597076958, | |
| "eval_sts-test_pearson_dot": 0.8597263949983345, | |
| "eval_sts-test_pearson_euclidean": 0.8977156894521804, | |
| "eval_sts-test_pearson_manhattan": 0.8973099980705126, | |
| "eval_sts-test_pearson_max": 0.8977156894521804, | |
| "eval_sts-test_spearman_cosine": 0.8862653474156692, | |
| "eval_sts-test_spearman_dot": 0.8451249310521576, | |
| "eval_sts-test_spearman_euclidean": 0.8866065310325173, | |
| "eval_sts-test_spearman_manhattan": 0.8860841943095571, | |
| "eval_sts-test_spearman_max": 0.8866065310325173, | |
| "eval_vitaminc-pairs_loss": 3.314167022705078, | |
| "eval_vitaminc-pairs_runtime": 4.4886, | |
| "eval_vitaminc-pairs_samples_per_second": 28.517, | |
| "eval_vitaminc-pairs_steps_per_second": 0.223, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_negation-triplets_loss": 1.3400739431381226, | |
| "eval_negation-triplets_runtime": 3.3068, | |
| "eval_negation-triplets_samples_per_second": 38.708, | |
| "eval_negation-triplets_steps_per_second": 0.302, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_scitail-pairs-pos_loss": 0.07263079285621643, | |
| "eval_scitail-pairs-pos_runtime": 2.6213, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.83, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.381, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_scitail-pairs-qa_loss": 0.048478204756975174, | |
| "eval_scitail-pairs-qa_runtime": 2.202, | |
| "eval_scitail-pairs-qa_samples_per_second": 58.13, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.454, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_xsum-pairs_loss": 0.14091075956821442, | |
| "eval_xsum-pairs_runtime": 3.0891, | |
| "eval_xsum-pairs_samples_per_second": 41.436, | |
| "eval_xsum-pairs_steps_per_second": 0.324, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_sciq_pairs_loss": 0.1505264937877655, | |
| "eval_sciq_pairs_runtime": 4.762, | |
| "eval_sciq_pairs_samples_per_second": 26.879, | |
| "eval_sciq_pairs_steps_per_second": 0.21, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_qasc_pairs_loss": 0.4244489073753357, | |
| "eval_qasc_pairs_runtime": 2.0783, | |
| "eval_qasc_pairs_samples_per_second": 61.589, | |
| "eval_qasc_pairs_steps_per_second": 0.481, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_openbookqa_pairs_loss": 0.5923758745193481, | |
| "eval_openbookqa_pairs_runtime": 2.1938, | |
| "eval_openbookqa_pairs_samples_per_second": 58.345, | |
| "eval_openbookqa_pairs_steps_per_second": 0.456, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_msmarco_pairs_loss": 0.5882836580276489, | |
| "eval_msmarco_pairs_runtime": 2.2483, | |
| "eval_msmarco_pairs_samples_per_second": 56.933, | |
| "eval_msmarco_pairs_steps_per_second": 0.445, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_nq_pairs_loss": 0.5424538254737854, | |
| "eval_nq_pairs_runtime": 3.6424, | |
| "eval_nq_pairs_samples_per_second": 35.142, | |
| "eval_nq_pairs_steps_per_second": 0.275, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_trivia_pairs_loss": 0.6437886953353882, | |
| "eval_trivia_pairs_runtime": 3.0809, | |
| "eval_trivia_pairs_samples_per_second": 41.546, | |
| "eval_trivia_pairs_steps_per_second": 0.325, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_gooaq_pairs_loss": 0.22378715872764587, | |
| "eval_gooaq_pairs_runtime": 2.101, | |
| "eval_gooaq_pairs_samples_per_second": 60.924, | |
| "eval_gooaq_pairs_steps_per_second": 0.476, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_paws-pos_loss": 0.03581737354397774, | |
| "eval_paws-pos_runtime": 2.2447, | |
| "eval_paws-pos_samples_per_second": 57.024, | |
| "eval_paws-pos_steps_per_second": 0.445, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20051413881748073, | |
| "eval_global_dataset_loss": 0.3989756405353546, | |
| "eval_global_dataset_runtime": 8.402, | |
| "eval_global_dataset_samples_per_second": 38.681, | |
| "eval_global_dataset_steps_per_second": 0.357, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20244215938303342, | |
| "grad_norm": 96.23986053466797, | |
| "learning_rate": 5.5019305019305015e-06, | |
| "loss": 0.5094, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.20822622107969152, | |
| "grad_norm": 33.88426208496094, | |
| "learning_rate": 5.675675675675676e-06, | |
| "loss": 0.8002, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.21401028277634962, | |
| "grad_norm": 7.319549560546875, | |
| "learning_rate": 5.849420849420849e-06, | |
| "loss": 0.3886, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.21979434447300772, | |
| "grad_norm": 8.65904712677002, | |
| "learning_rate": 6.023166023166023e-06, | |
| "loss": 0.6937, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.22557840616966582, | |
| "grad_norm": 376.61663818359375, | |
| "learning_rate": 6.196911196911197e-06, | |
| "loss": 0.2909, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.23136246786632392, | |
| "grad_norm": 10.809043884277344, | |
| "learning_rate": 6.370656370656371e-06, | |
| "loss": 0.3885, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23714652956298202, | |
| "grad_norm": 8.564398765563965, | |
| "learning_rate": 6.5444015444015445e-06, | |
| "loss": 0.29, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.24293059125964012, | |
| "grad_norm": 1.5215222835540771, | |
| "learning_rate": 6.718146718146718e-06, | |
| "loss": 0.3485, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2487146529562982, | |
| "grad_norm": 181.20809936523438, | |
| "learning_rate": 6.891891891891892e-06, | |
| "loss": 0.3931, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_Qnli-dev_cosine_accuracy": 0.67578125, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7986456155776978, | |
| "eval_Qnli-dev_cosine_ap": 0.6985736393833883, | |
| "eval_Qnli-dev_cosine_f1": 0.6777777777777777, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.726032018661499, | |
| "eval_Qnli-dev_cosine_precision": 0.6019736842105263, | |
| "eval_Qnli-dev_cosine_recall": 0.7754237288135594, | |
| "eval_Qnli-dev_dot_accuracy": 0.66796875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 840.5162353515625, | |
| "eval_Qnli-dev_dot_ap": 0.6656432640480556, | |
| "eval_Qnli-dev_dot_f1": 0.662020905923345, | |
| "eval_Qnli-dev_dot_f1_threshold": 715.1839599609375, | |
| "eval_Qnli-dev_dot_precision": 0.5621301775147929, | |
| "eval_Qnli-dev_dot_recall": 0.8050847457627118, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.677734375, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 20.397323608398438, | |
| "eval_Qnli-dev_euclidean_ap": 0.6992667316315335, | |
| "eval_Qnli-dev_euclidean_f1": 0.67296786389414, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 23.44580841064453, | |
| "eval_Qnli-dev_euclidean_precision": 0.6075085324232082, | |
| "eval_Qnli-dev_euclidean_recall": 0.7542372881355932, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.671875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 628.4102172851562, | |
| "eval_Qnli-dev_manhattan_ap": 0.6982133105013253, | |
| "eval_Qnli-dev_manhattan_f1": 0.673992673992674, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 746.4862060546875, | |
| "eval_Qnli-dev_manhattan_precision": 0.5935483870967742, | |
| "eval_Qnli-dev_manhattan_recall": 0.7796610169491526, | |
| "eval_Qnli-dev_max_accuracy": 0.677734375, | |
| "eval_Qnli-dev_max_accuracy_threshold": 840.5162353515625, | |
| "eval_Qnli-dev_max_ap": 0.6992667316315335, | |
| "eval_Qnli-dev_max_f1": 0.6777777777777777, | |
| "eval_Qnli-dev_max_f1_threshold": 746.4862060546875, | |
| "eval_Qnli-dev_max_precision": 0.6075085324232082, | |
| "eval_Qnli-dev_max_recall": 0.8050847457627118, | |
| "eval_allNLI-dev_cosine_accuracy": 0.7109375, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8466876745223999, | |
| "eval_allNLI-dev_cosine_ap": 0.5614885136859779, | |
| "eval_allNLI-dev_cosine_f1": 0.5809128630705395, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.6695160269737244, | |
| "eval_allNLI-dev_cosine_precision": 0.45307443365695793, | |
| "eval_allNLI-dev_cosine_recall": 0.8092485549132948, | |
| "eval_allNLI-dev_dot_accuracy": 0.69921875, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 901.9229736328125, | |
| "eval_allNLI-dev_dot_ap": 0.5637771052921187, | |
| "eval_allNLI-dev_dot_f1": 0.5812807881773399, | |
| "eval_allNLI-dev_dot_f1_threshold": 752.6895141601562, | |
| "eval_allNLI-dev_dot_precision": 0.5064377682403434, | |
| "eval_allNLI-dev_dot_recall": 0.6820809248554913, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.708984375, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 17.844379425048828, | |
| "eval_allNLI-dev_euclidean_ap": 0.5616160818796474, | |
| "eval_allNLI-dev_euclidean_f1": 0.5838509316770187, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 26.068946838378906, | |
| "eval_allNLI-dev_euclidean_precision": 0.45483870967741935, | |
| "eval_allNLI-dev_euclidean_recall": 0.815028901734104, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.7109375, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 553.34228515625, | |
| "eval_allNLI-dev_manhattan_ap": 0.5624349287984178, | |
| "eval_allNLI-dev_manhattan_f1": 0.5778688524590163, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 816.2254028320312, | |
| "eval_allNLI-dev_manhattan_precision": 0.44761904761904764, | |
| "eval_allNLI-dev_manhattan_recall": 0.815028901734104, | |
| "eval_allNLI-dev_max_accuracy": 0.7109375, | |
| "eval_allNLI-dev_max_accuracy_threshold": 901.9229736328125, | |
| "eval_allNLI-dev_max_ap": 0.5637771052921187, | |
| "eval_allNLI-dev_max_f1": 0.5838509316770187, | |
| "eval_allNLI-dev_max_f1_threshold": 816.2254028320312, | |
| "eval_allNLI-dev_max_precision": 0.5064377682403434, | |
| "eval_allNLI-dev_max_recall": 0.815028901734104, | |
| "eval_sequential_score": 0.6992667316315335, | |
| "eval_sts-test_pearson_cosine": 0.9024404439516205, | |
| "eval_sts-test_pearson_dot": 0.8941554024395801, | |
| "eval_sts-test_pearson_euclidean": 0.9164387414940136, | |
| "eval_sts-test_pearson_manhattan": 0.9158406136217865, | |
| "eval_sts-test_pearson_max": 0.9164387414940136, | |
| "eval_sts-test_spearman_cosine": 0.9081672412582797, | |
| "eval_sts-test_spearman_dot": 0.8855668249766109, | |
| "eval_sts-test_spearman_euclidean": 0.9076990535616029, | |
| "eval_sts-test_spearman_manhattan": 0.907316833038059, | |
| "eval_sts-test_spearman_max": 0.9081672412582797, | |
| "eval_vitaminc-pairs_loss": 3.8795158863067627, | |
| "eval_vitaminc-pairs_runtime": 4.6143, | |
| "eval_vitaminc-pairs_samples_per_second": 27.74, | |
| "eval_vitaminc-pairs_steps_per_second": 0.217, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_negation-triplets_loss": 1.1134737730026245, | |
| "eval_negation-triplets_runtime": 3.4103, | |
| "eval_negation-triplets_samples_per_second": 37.533, | |
| "eval_negation-triplets_steps_per_second": 0.293, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_scitail-pairs-pos_loss": 0.045735035091638565, | |
| "eval_scitail-pairs-pos_runtime": 2.7664, | |
| "eval_scitail-pairs-pos_samples_per_second": 46.27, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.361, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_scitail-pairs-qa_loss": 0.036478620022535324, | |
| "eval_scitail-pairs-qa_runtime": 2.2905, | |
| "eval_scitail-pairs-qa_samples_per_second": 55.884, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.437, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_xsum-pairs_loss": 0.05133158713579178, | |
| "eval_xsum-pairs_runtime": 3.1323, | |
| "eval_xsum-pairs_samples_per_second": 40.865, | |
| "eval_xsum-pairs_steps_per_second": 0.319, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_sciq_pairs_loss": 0.12314368784427643, | |
| "eval_sciq_pairs_runtime": 4.8477, | |
| "eval_sciq_pairs_samples_per_second": 26.404, | |
| "eval_sciq_pairs_steps_per_second": 0.206, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_qasc_pairs_loss": 0.22055241465568542, | |
| "eval_qasc_pairs_runtime": 2.1243, | |
| "eval_qasc_pairs_samples_per_second": 60.255, | |
| "eval_qasc_pairs_steps_per_second": 0.471, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_openbookqa_pairs_loss": 0.4567987024784088, | |
| "eval_openbookqa_pairs_runtime": 2.3078, | |
| "eval_openbookqa_pairs_samples_per_second": 55.465, | |
| "eval_openbookqa_pairs_steps_per_second": 0.433, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_msmarco_pairs_loss": 0.21620944142341614, | |
| "eval_msmarco_pairs_runtime": 2.3399, | |
| "eval_msmarco_pairs_samples_per_second": 54.703, | |
| "eval_msmarco_pairs_steps_per_second": 0.427, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_nq_pairs_loss": 0.26776355504989624, | |
| "eval_nq_pairs_runtime": 3.7038, | |
| "eval_nq_pairs_samples_per_second": 34.559, | |
| "eval_nq_pairs_steps_per_second": 0.27, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_trivia_pairs_loss": 0.15397313237190247, | |
| "eval_trivia_pairs_runtime": 3.1256, | |
| "eval_trivia_pairs_samples_per_second": 40.952, | |
| "eval_trivia_pairs_steps_per_second": 0.32, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_gooaq_pairs_loss": 0.12057099491357803, | |
| "eval_gooaq_pairs_runtime": 2.1825, | |
| "eval_gooaq_pairs_samples_per_second": 58.648, | |
| "eval_gooaq_pairs_steps_per_second": 0.458, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_paws-pos_loss": 0.026105860248208046, | |
| "eval_paws-pos_runtime": 2.2995, | |
| "eval_paws-pos_samples_per_second": 55.665, | |
| "eval_paws-pos_steps_per_second": 0.435, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2506426735218509, | |
| "eval_global_dataset_loss": 0.3591004014015198, | |
| "eval_global_dataset_runtime": 8.4153, | |
| "eval_global_dataset_samples_per_second": 38.62, | |
| "eval_global_dataset_steps_per_second": 0.356, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2544987146529563, | |
| "grad_norm": 11.167648315429688, | |
| "learning_rate": 7.065637065637066e-06, | |
| "loss": 0.3394, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.2602827763496144, | |
| "grad_norm": 3.4576871395111084, | |
| "learning_rate": 7.239382239382239e-06, | |
| "loss": 0.1276, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2660668380462725, | |
| "grad_norm": 9.594191551208496, | |
| "learning_rate": 7.413127413127414e-06, | |
| "loss": 0.3569, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.2718508997429306, | |
| "grad_norm": 4.255222320556641, | |
| "learning_rate": 7.586872586872587e-06, | |
| "loss": 0.1231, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.2776349614395887, | |
| "grad_norm": 6.074405670166016, | |
| "learning_rate": 7.76061776061776e-06, | |
| "loss": 0.3086, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.2834190231362468, | |
| "grad_norm": 19.442323684692383, | |
| "learning_rate": 7.934362934362936e-06, | |
| "loss": 0.3541, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.2892030848329049, | |
| "grad_norm": 5.141862392425537, | |
| "learning_rate": 8.108108108108109e-06, | |
| "loss": 0.2597, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.294987146529563, | |
| "grad_norm": 5.942401885986328, | |
| "learning_rate": 8.281853281853282e-06, | |
| "loss": 0.1585, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "grad_norm": 97.85575103759766, | |
| "learning_rate": 8.455598455598455e-06, | |
| "loss": 0.1436, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_Qnli-dev_cosine_accuracy": 0.693359375, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7395063638687134, | |
| "eval_Qnli-dev_cosine_ap": 0.7187155891006661, | |
| "eval_Qnli-dev_cosine_f1": 0.6837606837606838, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.660994291305542, | |
| "eval_Qnli-dev_cosine_precision": 0.5730659025787965, | |
| "eval_Qnli-dev_cosine_recall": 0.847457627118644, | |
| "eval_Qnli-dev_dot_accuracy": 0.669921875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 701.7633056640625, | |
| "eval_Qnli-dev_dot_ap": 0.6888124099944863, | |
| "eval_Qnli-dev_dot_f1": 0.6812816188870151, | |
| "eval_Qnli-dev_dot_f1_threshold": 633.31884765625, | |
| "eval_Qnli-dev_dot_precision": 0.5658263305322129, | |
| "eval_Qnli-dev_dot_recall": 0.8559322033898306, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.697265625, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 22.954120635986328, | |
| "eval_Qnli-dev_euclidean_ap": 0.721593279891293, | |
| "eval_Qnli-dev_euclidean_f1": 0.6762075134168157, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 24.89653968811035, | |
| "eval_Qnli-dev_euclidean_precision": 0.5851393188854489, | |
| "eval_Qnli-dev_euclidean_recall": 0.8008474576271186, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.6953125, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 718.2297973632812, | |
| "eval_Qnli-dev_manhattan_ap": 0.7239405100010182, | |
| "eval_Qnli-dev_manhattan_f1": 0.6807017543859649, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 784.701171875, | |
| "eval_Qnli-dev_manhattan_precision": 0.5808383233532934, | |
| "eval_Qnli-dev_manhattan_recall": 0.8220338983050848, | |
| "eval_Qnli-dev_max_accuracy": 0.697265625, | |
| "eval_Qnli-dev_max_accuracy_threshold": 718.2297973632812, | |
| "eval_Qnli-dev_max_ap": 0.7239405100010182, | |
| "eval_Qnli-dev_max_f1": 0.6837606837606838, | |
| "eval_Qnli-dev_max_f1_threshold": 784.701171875, | |
| "eval_Qnli-dev_max_precision": 0.5851393188854489, | |
| "eval_Qnli-dev_max_recall": 0.8559322033898306, | |
| "eval_allNLI-dev_cosine_accuracy": 0.71875, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8460206985473633, | |
| "eval_allNLI-dev_cosine_ap": 0.5796621470925665, | |
| "eval_allNLI-dev_cosine_f1": 0.6139954853273137, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7097446322441101, | |
| "eval_allNLI-dev_cosine_precision": 0.5037037037037037, | |
| "eval_allNLI-dev_cosine_recall": 0.7861271676300579, | |
| "eval_allNLI-dev_dot_accuracy": 0.7109375, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 837.8741455078125, | |
| "eval_allNLI-dev_dot_ap": 0.5763352242724786, | |
| "eval_allNLI-dev_dot_f1": 0.6179245283018867, | |
| "eval_allNLI-dev_dot_f1_threshold": 712.9136352539062, | |
| "eval_allNLI-dev_dot_precision": 0.5219123505976095, | |
| "eval_allNLI-dev_dot_recall": 0.7572254335260116, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.724609375, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 18.82543182373047, | |
| "eval_allNLI-dev_euclidean_ap": 0.5803015452539461, | |
| "eval_allNLI-dev_euclidean_f1": 0.6085011185682326, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 24.0869197845459, | |
| "eval_allNLI-dev_euclidean_precision": 0.49635036496350365, | |
| "eval_allNLI-dev_euclidean_recall": 0.7861271676300579, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.720703125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 586.8082275390625, | |
| "eval_allNLI-dev_manhattan_ap": 0.5806697040885319, | |
| "eval_allNLI-dev_manhattan_f1": 0.6035242290748899, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 756.8704833984375, | |
| "eval_allNLI-dev_manhattan_precision": 0.4875444839857651, | |
| "eval_allNLI-dev_manhattan_recall": 0.791907514450867, | |
| "eval_allNLI-dev_max_accuracy": 0.724609375, | |
| "eval_allNLI-dev_max_accuracy_threshold": 837.8741455078125, | |
| "eval_allNLI-dev_max_ap": 0.5806697040885319, | |
| "eval_allNLI-dev_max_f1": 0.6179245283018867, | |
| "eval_allNLI-dev_max_f1_threshold": 756.8704833984375, | |
| "eval_allNLI-dev_max_precision": 0.5219123505976095, | |
| "eval_allNLI-dev_max_recall": 0.791907514450867, | |
| "eval_sequential_score": 0.7239405100010182, | |
| "eval_sts-test_pearson_cosine": 0.9086022204177822, | |
| "eval_sts-test_pearson_dot": 0.8963707820565703, | |
| "eval_sts-test_pearson_euclidean": 0.9226263845809013, | |
| "eval_sts-test_pearson_manhattan": 0.9216624653503607, | |
| "eval_sts-test_pearson_max": 0.9226263845809013, | |
| "eval_sts-test_spearman_cosine": 0.9176886091344368, | |
| "eval_sts-test_spearman_dot": 0.8941249219513181, | |
| "eval_sts-test_spearman_euclidean": 0.9176283739366038, | |
| "eval_sts-test_spearman_manhattan": 0.9167815458172944, | |
| "eval_sts-test_spearman_max": 0.9176886091344368, | |
| "eval_vitaminc-pairs_loss": 3.8346316814422607, | |
| "eval_vitaminc-pairs_runtime": 4.4216, | |
| "eval_vitaminc-pairs_samples_per_second": 28.949, | |
| "eval_vitaminc-pairs_steps_per_second": 0.226, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_negation-triplets_loss": 0.9441184997558594, | |
| "eval_negation-triplets_runtime": 3.3236, | |
| "eval_negation-triplets_samples_per_second": 38.512, | |
| "eval_negation-triplets_steps_per_second": 0.301, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_scitail-pairs-pos_loss": 0.03759249672293663, | |
| "eval_scitail-pairs-pos_runtime": 2.6436, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.418, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.378, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_scitail-pairs-qa_loss": 0.04104204475879669, | |
| "eval_scitail-pairs-qa_runtime": 2.205, | |
| "eval_scitail-pairs-qa_samples_per_second": 58.05, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.454, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_xsum-pairs_loss": 0.05032652989029884, | |
| "eval_xsum-pairs_runtime": 3.0619, | |
| "eval_xsum-pairs_samples_per_second": 41.805, | |
| "eval_xsum-pairs_steps_per_second": 0.327, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_sciq_pairs_loss": 0.08709195256233215, | |
| "eval_sciq_pairs_runtime": 4.8405, | |
| "eval_sciq_pairs_samples_per_second": 26.444, | |
| "eval_sciq_pairs_steps_per_second": 0.207, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_qasc_pairs_loss": 0.17168982326984406, | |
| "eval_qasc_pairs_runtime": 2.078, | |
| "eval_qasc_pairs_samples_per_second": 61.599, | |
| "eval_qasc_pairs_steps_per_second": 0.481, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_openbookqa_pairs_loss": 0.4304167926311493, | |
| "eval_openbookqa_pairs_runtime": 2.1991, | |
| "eval_openbookqa_pairs_samples_per_second": 58.205, | |
| "eval_openbookqa_pairs_steps_per_second": 0.455, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_msmarco_pairs_loss": 0.13501481711864471, | |
| "eval_msmarco_pairs_runtime": 2.2186, | |
| "eval_msmarco_pairs_samples_per_second": 57.694, | |
| "eval_msmarco_pairs_steps_per_second": 0.451, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_nq_pairs_loss": 0.20035718381404877, | |
| "eval_nq_pairs_runtime": 3.6367, | |
| "eval_nq_pairs_samples_per_second": 35.197, | |
| "eval_nq_pairs_steps_per_second": 0.275, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_trivia_pairs_loss": 0.08228898048400879, | |
| "eval_trivia_pairs_runtime": 3.086, | |
| "eval_trivia_pairs_samples_per_second": 41.477, | |
| "eval_trivia_pairs_steps_per_second": 0.324, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_gooaq_pairs_loss": 0.0767039954662323, | |
| "eval_gooaq_pairs_runtime": 2.1123, | |
| "eval_gooaq_pairs_samples_per_second": 60.599, | |
| "eval_gooaq_pairs_steps_per_second": 0.473, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_paws-pos_loss": 0.02134825475513935, | |
| "eval_paws-pos_runtime": 2.2234, | |
| "eval_paws-pos_samples_per_second": 57.571, | |
| "eval_paws-pos_steps_per_second": 0.45, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.30077120822622105, | |
| "eval_global_dataset_loss": 0.2958550453186035, | |
| "eval_global_dataset_runtime": 8.3542, | |
| "eval_global_dataset_samples_per_second": 38.903, | |
| "eval_global_dataset_steps_per_second": 0.359, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3065552699228792, | |
| "grad_norm": 47.52979278564453, | |
| "learning_rate": 8.62934362934363e-06, | |
| "loss": 0.1941, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.31233933161953725, | |
| "grad_norm": 1234.25927734375, | |
| "learning_rate": 8.803088803088804e-06, | |
| "loss": 0.3041, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3181233933161954, | |
| "grad_norm": 185.80238342285156, | |
| "learning_rate": 8.976833976833977e-06, | |
| "loss": 0.2358, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.32390745501285345, | |
| "grad_norm": 7.233102798461914, | |
| "learning_rate": 9.15057915057915e-06, | |
| "loss": 0.2148, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.3296915167095116, | |
| "grad_norm": 342.77685546875, | |
| "learning_rate": 9.324324324324325e-06, | |
| "loss": 0.8567, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.33547557840616965, | |
| "grad_norm": 181.07650756835938, | |
| "learning_rate": 9.498069498069498e-06, | |
| "loss": 0.3668, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3412596401028278, | |
| "grad_norm": 198.4124755859375, | |
| "learning_rate": 9.671814671814673e-06, | |
| "loss": 0.3278, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.34704370179948585, | |
| "grad_norm": 91.80472564697266, | |
| "learning_rate": 9.845559845559846e-06, | |
| "loss": 0.474, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_Qnli-dev_cosine_accuracy": 0.6953125, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.7320406436920166, | |
| "eval_Qnli-dev_cosine_ap": 0.7163901550352688, | |
| "eval_Qnli-dev_cosine_f1": 0.6857142857142857, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.6526604890823364, | |
| "eval_Qnli-dev_cosine_precision": 0.5682451253481894, | |
| "eval_Qnli-dev_cosine_recall": 0.864406779661017, | |
| "eval_Qnli-dev_dot_accuracy": 0.67578125, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 759.5008544921875, | |
| "eval_Qnli-dev_dot_ap": 0.6857309318375049, | |
| "eval_Qnli-dev_dot_f1": 0.6857142857142857, | |
| "eval_Qnli-dev_dot_f1_threshold": 653.0701904296875, | |
| "eval_Qnli-dev_dot_precision": 0.5682451253481894, | |
| "eval_Qnli-dev_dot_recall": 0.864406779661017, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.689453125, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 23.348251342773438, | |
| "eval_Qnli-dev_euclidean_ap": 0.7186338110228241, | |
| "eval_Qnli-dev_euclidean_f1": 0.6819672131147542, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 26.874767303466797, | |
| "eval_Qnli-dev_euclidean_precision": 0.5561497326203209, | |
| "eval_Qnli-dev_euclidean_recall": 0.8813559322033898, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.6875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 695.7578125, | |
| "eval_Qnli-dev_manhattan_ap": 0.7187237394606566, | |
| "eval_Qnli-dev_manhattan_f1": 0.6815286624203821, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 856.7666015625, | |
| "eval_Qnli-dev_manhattan_precision": 0.5459183673469388, | |
| "eval_Qnli-dev_manhattan_recall": 0.9067796610169492, | |
| "eval_Qnli-dev_max_accuracy": 0.6953125, | |
| "eval_Qnli-dev_max_accuracy_threshold": 759.5008544921875, | |
| "eval_Qnli-dev_max_ap": 0.7187237394606566, | |
| "eval_Qnli-dev_max_f1": 0.6857142857142857, | |
| "eval_Qnli-dev_max_f1_threshold": 856.7666015625, | |
| "eval_Qnli-dev_max_precision": 0.5682451253481894, | |
| "eval_Qnli-dev_max_recall": 0.9067796610169492, | |
| "eval_allNLI-dev_cosine_accuracy": 0.705078125, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8531240224838257, | |
| "eval_allNLI-dev_cosine_ap": 0.5555367525973895, | |
| "eval_allNLI-dev_cosine_f1": 0.5972222222222222, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7072098255157471, | |
| "eval_allNLI-dev_cosine_precision": 0.4980694980694981, | |
| "eval_allNLI-dev_cosine_recall": 0.7456647398843931, | |
| "eval_allNLI-dev_dot_accuracy": 0.705078125, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 858.6700439453125, | |
| "eval_allNLI-dev_dot_ap": 0.5501313597611072, | |
| "eval_allNLI-dev_dot_f1": 0.5992217898832685, | |
| "eval_allNLI-dev_dot_f1_threshold": 657.9625244140625, | |
| "eval_allNLI-dev_dot_precision": 0.45161290322580644, | |
| "eval_allNLI-dev_dot_recall": 0.8901734104046243, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.705078125, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 16.42839813232422, | |
| "eval_allNLI-dev_euclidean_ap": 0.5519081828104085, | |
| "eval_allNLI-dev_euclidean_f1": 0.5919282511210763, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 24.696884155273438, | |
| "eval_allNLI-dev_euclidean_precision": 0.4835164835164835, | |
| "eval_allNLI-dev_euclidean_recall": 0.7630057803468208, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.705078125, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 517.3240356445312, | |
| "eval_allNLI-dev_manhattan_ap": 0.5527648330520126, | |
| "eval_allNLI-dev_manhattan_f1": 0.5900900900900902, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 765.703857421875, | |
| "eval_allNLI-dev_manhattan_precision": 0.4833948339483395, | |
| "eval_allNLI-dev_manhattan_recall": 0.7572254335260116, | |
| "eval_allNLI-dev_max_accuracy": 0.705078125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 858.6700439453125, | |
| "eval_allNLI-dev_max_ap": 0.5555367525973895, | |
| "eval_allNLI-dev_max_f1": 0.5992217898832685, | |
| "eval_allNLI-dev_max_f1_threshold": 765.703857421875, | |
| "eval_allNLI-dev_max_precision": 0.4980694980694981, | |
| "eval_allNLI-dev_max_recall": 0.8901734104046243, | |
| "eval_sequential_score": 0.7187237394606566, | |
| "eval_sts-test_pearson_cosine": 0.8999110842263669, | |
| "eval_sts-test_pearson_dot": 0.8830335713257453, | |
| "eval_sts-test_pearson_euclidean": 0.9123289195839425, | |
| "eval_sts-test_pearson_manhattan": 0.913131760355957, | |
| "eval_sts-test_pearson_max": 0.913131760355957, | |
| "eval_sts-test_spearman_cosine": 0.9102556931248936, | |
| "eval_sts-test_spearman_dot": 0.8782251567890449, | |
| "eval_sts-test_spearman_euclidean": 0.9078544746924153, | |
| "eval_sts-test_spearman_manhattan": 0.9084604962741777, | |
| "eval_sts-test_spearman_max": 0.9102556931248936, | |
| "eval_vitaminc-pairs_loss": 3.8443844318389893, | |
| "eval_vitaminc-pairs_runtime": 4.62, | |
| "eval_vitaminc-pairs_samples_per_second": 27.706, | |
| "eval_vitaminc-pairs_steps_per_second": 0.216, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_negation-triplets_loss": 0.8706199526786804, | |
| "eval_negation-triplets_runtime": 3.4865, | |
| "eval_negation-triplets_samples_per_second": 36.714, | |
| "eval_negation-triplets_steps_per_second": 0.287, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_scitail-pairs-pos_loss": 0.05496061593294144, | |
| "eval_scitail-pairs-pos_runtime": 2.8238, | |
| "eval_scitail-pairs-pos_samples_per_second": 45.328, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.354, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_scitail-pairs-qa_loss": 0.12511999905109406, | |
| "eval_scitail-pairs-qa_runtime": 2.2972, | |
| "eval_scitail-pairs-qa_samples_per_second": 55.721, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.435, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_xsum-pairs_loss": 0.03821578621864319, | |
| "eval_xsum-pairs_runtime": 3.1015, | |
| "eval_xsum-pairs_samples_per_second": 41.27, | |
| "eval_xsum-pairs_steps_per_second": 0.322, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_sciq_pairs_loss": 0.08246304094791412, | |
| "eval_sciq_pairs_runtime": 4.8955, | |
| "eval_sciq_pairs_samples_per_second": 26.147, | |
| "eval_sciq_pairs_steps_per_second": 0.204, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_qasc_pairs_loss": 0.15105530619621277, | |
| "eval_qasc_pairs_runtime": 2.2079, | |
| "eval_qasc_pairs_samples_per_second": 57.972, | |
| "eval_qasc_pairs_steps_per_second": 0.453, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_openbookqa_pairs_loss": 0.7893790602684021, | |
| "eval_openbookqa_pairs_runtime": 2.2579, | |
| "eval_openbookqa_pairs_samples_per_second": 56.69, | |
| "eval_openbookqa_pairs_steps_per_second": 0.443, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_msmarco_pairs_loss": 0.41756388545036316, | |
| "eval_msmarco_pairs_runtime": 2.2908, | |
| "eval_msmarco_pairs_samples_per_second": 55.877, | |
| "eval_msmarco_pairs_steps_per_second": 0.437, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_nq_pairs_loss": 0.2524656355381012, | |
| "eval_nq_pairs_runtime": 3.6631, | |
| "eval_nq_pairs_samples_per_second": 34.943, | |
| "eval_nq_pairs_steps_per_second": 0.273, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_trivia_pairs_loss": 0.062444861978292465, | |
| "eval_trivia_pairs_runtime": 3.115, | |
| "eval_trivia_pairs_samples_per_second": 41.091, | |
| "eval_trivia_pairs_steps_per_second": 0.321, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_gooaq_pairs_loss": 0.12118146568536758, | |
| "eval_gooaq_pairs_runtime": 2.236, | |
| "eval_gooaq_pairs_samples_per_second": 57.245, | |
| "eval_gooaq_pairs_steps_per_second": 0.447, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_paws-pos_loss": 0.020900897681713104, | |
| "eval_paws-pos_runtime": 2.2981, | |
| "eval_paws-pos_samples_per_second": 55.698, | |
| "eval_paws-pos_steps_per_second": 0.435, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3508997429305913, | |
| "eval_global_dataset_loss": 0.38996464014053345, | |
| "eval_global_dataset_runtime": 8.423, | |
| "eval_global_dataset_samples_per_second": 38.585, | |
| "eval_global_dataset_steps_per_second": 0.356, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.352827763496144, | |
| "grad_norm": 412.62738037109375, | |
| "learning_rate": 1.0019305019305019e-05, | |
| "loss": 0.5365, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.35861182519280205, | |
| "grad_norm": 693.9551391601562, | |
| "learning_rate": 1.0193050193050193e-05, | |
| "loss": 0.6902, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.3643958868894602, | |
| "grad_norm": 739.5757446289062, | |
| "learning_rate": 1.0366795366795366e-05, | |
| "loss": 0.4105, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.37017994858611825, | |
| "grad_norm": 174.94882202148438, | |
| "learning_rate": 1.0540540540540541e-05, | |
| "loss": 0.2434, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.3759640102827764, | |
| "grad_norm": 5.913711071014404, | |
| "learning_rate": 1.0714285714285714e-05, | |
| "loss": 0.1521, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.38174807197943444, | |
| "grad_norm": 34.247581481933594, | |
| "learning_rate": 1.0888030888030889e-05, | |
| "loss": 0.1878, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.38753213367609257, | |
| "grad_norm": 50.26247787475586, | |
| "learning_rate": 1.1061776061776062e-05, | |
| "loss": 0.3544, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.39331619537275064, | |
| "grad_norm": 227.15286254882812, | |
| "learning_rate": 1.1235521235521235e-05, | |
| "loss": 0.1397, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.3991002570694087, | |
| "grad_norm": 95.1203384399414, | |
| "learning_rate": 1.140926640926641e-05, | |
| "loss": 0.2982, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_Qnli-dev_cosine_accuracy": 0.685546875, | |
| "eval_Qnli-dev_cosine_accuracy_threshold": 0.715149462223053, | |
| "eval_Qnli-dev_cosine_ap": 0.7217325275419479, | |
| "eval_Qnli-dev_cosine_f1": 0.6854130052724077, | |
| "eval_Qnli-dev_cosine_f1_threshold": 0.658507227897644, | |
| "eval_Qnli-dev_cosine_precision": 0.5855855855855856, | |
| "eval_Qnli-dev_cosine_recall": 0.826271186440678, | |
| "eval_Qnli-dev_dot_accuracy": 0.669921875, | |
| "eval_Qnli-dev_dot_accuracy_threshold": 717.3741455078125, | |
| "eval_Qnli-dev_dot_ap": 0.6857348319823914, | |
| "eval_Qnli-dev_dot_f1": 0.6757679180887372, | |
| "eval_Qnli-dev_dot_f1_threshold": 627.569580078125, | |
| "eval_Qnli-dev_dot_precision": 0.5657142857142857, | |
| "eval_Qnli-dev_dot_recall": 0.8389830508474576, | |
| "eval_Qnli-dev_euclidean_accuracy": 0.68359375, | |
| "eval_Qnli-dev_euclidean_accuracy_threshold": 23.550647735595703, | |
| "eval_Qnli-dev_euclidean_ap": 0.7266186529674616, | |
| "eval_Qnli-dev_euclidean_f1": 0.6959578207381371, | |
| "eval_Qnli-dev_euclidean_f1_threshold": 25.634620666503906, | |
| "eval_Qnli-dev_euclidean_precision": 0.5945945945945946, | |
| "eval_Qnli-dev_euclidean_recall": 0.8389830508474576, | |
| "eval_Qnli-dev_manhattan_accuracy": 0.6796875, | |
| "eval_Qnli-dev_manhattan_accuracy_threshold": 681.6762084960938, | |
| "eval_Qnli-dev_manhattan_ap": 0.7264245248641615, | |
| "eval_Qnli-dev_manhattan_f1": 0.6938053097345133, | |
| "eval_Qnli-dev_manhattan_f1_threshold": 793.83837890625, | |
| "eval_Qnli-dev_manhattan_precision": 0.5957446808510638, | |
| "eval_Qnli-dev_manhattan_recall": 0.8305084745762712, | |
| "eval_Qnli-dev_max_accuracy": 0.685546875, | |
| "eval_Qnli-dev_max_accuracy_threshold": 717.3741455078125, | |
| "eval_Qnli-dev_max_ap": 0.7266186529674616, | |
| "eval_Qnli-dev_max_f1": 0.6959578207381371, | |
| "eval_Qnli-dev_max_f1_threshold": 793.83837890625, | |
| "eval_Qnli-dev_max_precision": 0.5957446808510638, | |
| "eval_Qnli-dev_max_recall": 0.8389830508474576, | |
| "eval_allNLI-dev_cosine_accuracy": 0.720703125, | |
| "eval_allNLI-dev_cosine_accuracy_threshold": 0.8279609680175781, | |
| "eval_allNLI-dev_cosine_ap": 0.5853174176053627, | |
| "eval_allNLI-dev_cosine_f1": 0.6189376443418013, | |
| "eval_allNLI-dev_cosine_f1_threshold": 0.7112143039703369, | |
| "eval_allNLI-dev_cosine_precision": 0.5153846153846153, | |
| "eval_allNLI-dev_cosine_recall": 0.7745664739884393, | |
| "eval_allNLI-dev_dot_accuracy": 0.708984375, | |
| "eval_allNLI-dev_dot_accuracy_threshold": 844.9638061523438, | |
| "eval_allNLI-dev_dot_ap": 0.5767480495013521, | |
| "eval_allNLI-dev_dot_f1": 0.620525059665871, | |
| "eval_allNLI-dev_dot_f1_threshold": 698.743408203125, | |
| "eval_allNLI-dev_dot_precision": 0.5284552845528455, | |
| "eval_allNLI-dev_dot_recall": 0.7514450867052023, | |
| "eval_allNLI-dev_euclidean_accuracy": 0.71875, | |
| "eval_allNLI-dev_euclidean_accuracy_threshold": 17.55453109741211, | |
| "eval_allNLI-dev_euclidean_ap": 0.5835774601767848, | |
| "eval_allNLI-dev_euclidean_f1": 0.6063569682151589, | |
| "eval_allNLI-dev_euclidean_f1_threshold": 22.958927154541016, | |
| "eval_allNLI-dev_euclidean_precision": 0.5254237288135594, | |
| "eval_allNLI-dev_euclidean_recall": 0.7167630057803468, | |
| "eval_allNLI-dev_manhattan_accuracy": 0.71875, | |
| "eval_allNLI-dev_manhattan_accuracy_threshold": 546.237060546875, | |
| "eval_allNLI-dev_manhattan_ap": 0.5834282573106326, | |
| "eval_allNLI-dev_manhattan_f1": 0.6093023255813954, | |
| "eval_allNLI-dev_manhattan_f1_threshold": 738.0977172851562, | |
| "eval_allNLI-dev_manhattan_precision": 0.5097276264591439, | |
| "eval_allNLI-dev_manhattan_recall": 0.7572254335260116, | |
| "eval_allNLI-dev_max_accuracy": 0.720703125, | |
| "eval_allNLI-dev_max_accuracy_threshold": 844.9638061523438, | |
| "eval_allNLI-dev_max_ap": 0.5853174176053627, | |
| "eval_allNLI-dev_max_f1": 0.620525059665871, | |
| "eval_allNLI-dev_max_f1_threshold": 738.0977172851562, | |
| "eval_allNLI-dev_max_precision": 0.5284552845528455, | |
| "eval_allNLI-dev_max_recall": 0.7745664739884393, | |
| "eval_sequential_score": 0.7266186529674616, | |
| "eval_sts-test_pearson_cosine": 0.9110444991658042, | |
| "eval_sts-test_pearson_dot": 0.9027007693117498, | |
| "eval_sts-test_pearson_euclidean": 0.9255983263878029, | |
| "eval_sts-test_pearson_manhattan": 0.925028098980626, | |
| "eval_sts-test_pearson_max": 0.9255983263878029, | |
| "eval_sts-test_spearman_cosine": 0.9211899030744385, | |
| "eval_sts-test_spearman_dot": 0.9034949858127107, | |
| "eval_sts-test_spearman_euclidean": 0.9209135640353466, | |
| "eval_sts-test_spearman_manhattan": 0.9206623304538951, | |
| "eval_sts-test_spearman_max": 0.9211899030744385, | |
| "eval_vitaminc-pairs_loss": 3.210468292236328, | |
| "eval_vitaminc-pairs_runtime": 4.4941, | |
| "eval_vitaminc-pairs_samples_per_second": 28.482, | |
| "eval_vitaminc-pairs_steps_per_second": 0.223, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_negation-triplets_loss": 0.6793166995048523, | |
| "eval_negation-triplets_runtime": 3.2613, | |
| "eval_negation-triplets_samples_per_second": 39.249, | |
| "eval_negation-triplets_steps_per_second": 0.307, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_scitail-pairs-pos_loss": 0.020910607650876045, | |
| "eval_scitail-pairs-pos_runtime": 2.6259, | |
| "eval_scitail-pairs-pos_samples_per_second": 48.745, | |
| "eval_scitail-pairs-pos_steps_per_second": 0.381, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_scitail-pairs-qa_loss": 0.024287551641464233, | |
| "eval_scitail-pairs-qa_runtime": 2.2326, | |
| "eval_scitail-pairs-qa_samples_per_second": 57.331, | |
| "eval_scitail-pairs-qa_steps_per_second": 0.448, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_xsum-pairs_loss": 0.028975818306207657, | |
| "eval_xsum-pairs_runtime": 3.0795, | |
| "eval_xsum-pairs_samples_per_second": 41.565, | |
| "eval_xsum-pairs_steps_per_second": 0.325, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_sciq_pairs_loss": 0.08505217730998993, | |
| "eval_sciq_pairs_runtime": 4.7361, | |
| "eval_sciq_pairs_samples_per_second": 27.026, | |
| "eval_sciq_pairs_steps_per_second": 0.211, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_qasc_pairs_loss": 0.1334802210330963, | |
| "eval_qasc_pairs_runtime": 2.0711, | |
| "eval_qasc_pairs_samples_per_second": 61.804, | |
| "eval_qasc_pairs_steps_per_second": 0.483, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_openbookqa_pairs_loss": 0.4361065924167633, | |
| "eval_openbookqa_pairs_runtime": 2.2402, | |
| "eval_openbookqa_pairs_samples_per_second": 57.137, | |
| "eval_openbookqa_pairs_steps_per_second": 0.446, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_msmarco_pairs_loss": 0.09543590992689133, | |
| "eval_msmarco_pairs_runtime": 2.2223, | |
| "eval_msmarco_pairs_samples_per_second": 57.599, | |
| "eval_msmarco_pairs_steps_per_second": 0.45, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_nq_pairs_loss": 0.24286797642707825, | |
| "eval_nq_pairs_runtime": 3.617, | |
| "eval_nq_pairs_samples_per_second": 35.389, | |
| "eval_nq_pairs_steps_per_second": 0.276, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_trivia_pairs_loss": 0.053676724433898926, | |
| "eval_trivia_pairs_runtime": 3.0737, | |
| "eval_trivia_pairs_samples_per_second": 41.644, | |
| "eval_trivia_pairs_steps_per_second": 0.325, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_gooaq_pairs_loss": 0.07278460264205933, | |
| "eval_gooaq_pairs_runtime": 2.1378, | |
| "eval_gooaq_pairs_samples_per_second": 59.874, | |
| "eval_gooaq_pairs_steps_per_second": 0.468, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_paws-pos_loss": 0.022397885099053383, | |
| "eval_paws-pos_runtime": 2.253, | |
| "eval_paws-pos_samples_per_second": 56.812, | |
| "eval_paws-pos_steps_per_second": 0.444, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40102827763496146, | |
| "eval_global_dataset_loss": 0.250370055437088, | |
| "eval_global_dataset_runtime": 8.321, | |
| "eval_global_dataset_samples_per_second": 39.058, | |
| "eval_global_dataset_steps_per_second": 0.361, | |
| "step": 208 | |
| } | |
| ], | |
| "logging_steps": 3, | |
| "max_steps": 1036, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 52, | |
| "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 | |
| } | |