| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 45, | |
| "global_step": 445, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0022471910112359553, | |
| "grad_norm": 3.9492883682250977, | |
| "learning_rate": 1.9662921348314604e-07, | |
| "loss": 0.8103, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0044943820224719105, | |
| "grad_norm": 4.117438793182373, | |
| "learning_rate": 3.932584269662921e-07, | |
| "loss": 0.8803, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006741573033707865, | |
| "grad_norm": 3.809002161026001, | |
| "learning_rate": 5.898876404494381e-07, | |
| "loss": 0.8219, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.008988764044943821, | |
| "grad_norm": 0.7417504787445068, | |
| "learning_rate": 7.865168539325842e-07, | |
| "loss": 0.0574, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.011235955056179775, | |
| "grad_norm": 2.707460403442383, | |
| "learning_rate": 9.831460674157302e-07, | |
| "loss": 0.3044, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01348314606741573, | |
| "grad_norm": 3.082705497741699, | |
| "learning_rate": 1.1797752808988763e-06, | |
| "loss": 0.3306, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.015730337078651686, | |
| "grad_norm": 3.102416753768921, | |
| "learning_rate": 1.3764044943820223e-06, | |
| "loss": 0.759, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.017977528089887642, | |
| "grad_norm": 0.6271047592163086, | |
| "learning_rate": 1.5730337078651683e-06, | |
| "loss": 0.0472, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.020224719101123594, | |
| "grad_norm": 3.1362593173980713, | |
| "learning_rate": 1.7696629213483144e-06, | |
| "loss": 0.7782, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02247191011235955, | |
| "grad_norm": 1.124997615814209, | |
| "learning_rate": 1.9662921348314604e-06, | |
| "loss": 0.0757, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.024719101123595506, | |
| "grad_norm": 3.194413185119629, | |
| "learning_rate": 2.1629213483146067e-06, | |
| "loss": 0.7778, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02696629213483146, | |
| "grad_norm": 3.966202974319458, | |
| "learning_rate": 2.3595505617977525e-06, | |
| "loss": 0.7111, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.029213483146067417, | |
| "grad_norm": 3.63393235206604, | |
| "learning_rate": 2.5561797752808988e-06, | |
| "loss": 0.6598, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03146067415730337, | |
| "grad_norm": 4.087065696716309, | |
| "learning_rate": 2.7528089887640446e-06, | |
| "loss": 0.8901, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.033707865168539325, | |
| "grad_norm": 2.769573211669922, | |
| "learning_rate": 2.949438202247191e-06, | |
| "loss": 0.3206, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.035955056179775284, | |
| "grad_norm": 2.630620002746582, | |
| "learning_rate": 3.1460674157303367e-06, | |
| "loss": 0.3408, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.038202247191011236, | |
| "grad_norm": 2.9570937156677246, | |
| "learning_rate": 3.342696629213483e-06, | |
| "loss": 0.5623, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.04044943820224719, | |
| "grad_norm": 1.0999970436096191, | |
| "learning_rate": 3.5393258426966288e-06, | |
| "loss": 0.0758, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04269662921348315, | |
| "grad_norm": 5.516472816467285, | |
| "learning_rate": 3.735955056179775e-06, | |
| "loss": 0.994, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0449438202247191, | |
| "grad_norm": 6.245299816131592, | |
| "learning_rate": 3.932584269662921e-06, | |
| "loss": 2.4196, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04719101123595506, | |
| "grad_norm": 0.546605110168457, | |
| "learning_rate": 4.129213483146067e-06, | |
| "loss": 0.0561, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04943820224719101, | |
| "grad_norm": 0.7049635648727417, | |
| "learning_rate": 4.325842696629213e-06, | |
| "loss": 0.0827, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.051685393258426963, | |
| "grad_norm": 3.1022439002990723, | |
| "learning_rate": 4.522471910112359e-06, | |
| "loss": 0.7405, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05393258426966292, | |
| "grad_norm": 4.534759044647217, | |
| "learning_rate": 4.719101123595505e-06, | |
| "loss": 0.9656, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.056179775280898875, | |
| "grad_norm": 3.0486032962799072, | |
| "learning_rate": 4.915730337078652e-06, | |
| "loss": 0.7855, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.058426966292134834, | |
| "grad_norm": 3.7457478046417236, | |
| "learning_rate": 5.1123595505617975e-06, | |
| "loss": 0.6349, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.060674157303370786, | |
| "grad_norm": 3.2051479816436768, | |
| "learning_rate": 5.308988764044943e-06, | |
| "loss": 0.8087, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06292134831460675, | |
| "grad_norm": 4.389094829559326, | |
| "learning_rate": 5.505617977528089e-06, | |
| "loss": 0.9282, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0651685393258427, | |
| "grad_norm": 2.920410394668579, | |
| "learning_rate": 5.702247191011236e-06, | |
| "loss": 0.3377, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06741573033707865, | |
| "grad_norm": 2.7193148136138916, | |
| "learning_rate": 5.898876404494382e-06, | |
| "loss": 0.3289, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0696629213483146, | |
| "grad_norm": 4.0008225440979, | |
| "learning_rate": 6.0955056179775275e-06, | |
| "loss": 0.6314, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07191011235955057, | |
| "grad_norm": 0.5842159390449524, | |
| "learning_rate": 6.292134831460673e-06, | |
| "loss": 0.0611, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07415730337078652, | |
| "grad_norm": 3.1256043910980225, | |
| "learning_rate": 6.48876404494382e-06, | |
| "loss": 0.8942, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.07640449438202247, | |
| "grad_norm": 0.9526051878929138, | |
| "learning_rate": 6.685393258426966e-06, | |
| "loss": 0.0701, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07865168539325842, | |
| "grad_norm": 4.061926364898682, | |
| "learning_rate": 6.882022471910112e-06, | |
| "loss": 0.8506, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08089887640449438, | |
| "grad_norm": 2.8898491859436035, | |
| "learning_rate": 7.0786516853932575e-06, | |
| "loss": 0.3386, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08314606741573034, | |
| "grad_norm": 0.9806709289550781, | |
| "learning_rate": 7.275280898876404e-06, | |
| "loss": 0.0701, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0853932584269663, | |
| "grad_norm": 3.8004391193389893, | |
| "learning_rate": 7.47191011235955e-06, | |
| "loss": 0.8042, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.08764044943820225, | |
| "grad_norm": 4.089083194732666, | |
| "learning_rate": 7.668539325842697e-06, | |
| "loss": 0.8744, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0898876404494382, | |
| "grad_norm": 3.419440984725952, | |
| "learning_rate": 7.865168539325842e-06, | |
| "loss": 0.8644, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09213483146067415, | |
| "grad_norm": 4.094921588897705, | |
| "learning_rate": 8.061797752808988e-06, | |
| "loss": 0.8647, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09438202247191012, | |
| "grad_norm": 3.9199764728546143, | |
| "learning_rate": 8.258426966292133e-06, | |
| "loss": 0.7916, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.09662921348314607, | |
| "grad_norm": 4.082360744476318, | |
| "learning_rate": 8.45505617977528e-06, | |
| "loss": 0.8599, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.09887640449438202, | |
| "grad_norm": 0.6443855166435242, | |
| "learning_rate": 8.651685393258427e-06, | |
| "loss": 0.0523, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "grad_norm": 4.051048278808594, | |
| "learning_rate": 8.848314606741572e-06, | |
| "loss": 0.6968, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_VitaminC_cosine_accuracy": 0.55859375, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8300318121910095, | |
| "eval_VitaminC_cosine_ap": 0.5514483751609435, | |
| "eval_VitaminC_cosine_f1": 0.6657718120805369, | |
| "eval_VitaminC_cosine_f1_threshold": 0.37456807494163513, | |
| "eval_VitaminC_cosine_precision": 0.5020242914979757, | |
| "eval_VitaminC_cosine_recall": 0.9880478087649402, | |
| "eval_VitaminC_dot_accuracy": 0.5546875, | |
| "eval_VitaminC_dot_accuracy_threshold": 311.380615234375, | |
| "eval_VitaminC_dot_ap": 0.5333497363350208, | |
| "eval_VitaminC_dot_f1": 0.6684709066305818, | |
| "eval_VitaminC_dot_f1_threshold": 144.8927001953125, | |
| "eval_VitaminC_dot_precision": 0.5061475409836066, | |
| "eval_VitaminC_dot_recall": 0.9840637450199203, | |
| "eval_VitaminC_euclidean_accuracy": 0.556640625, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 11.028482437133789, | |
| "eval_VitaminC_euclidean_ap": 0.5544340410314673, | |
| "eval_VitaminC_euclidean_f1": 0.6649006622516557, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.38451385498047, | |
| "eval_VitaminC_euclidean_precision": 0.498015873015873, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.556640625, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 232.38790893554688, | |
| "eval_VitaminC_manhattan_ap": 0.5515569514532939, | |
| "eval_VitaminC_manhattan_f1": 0.6649006622516557, | |
| "eval_VitaminC_manhattan_f1_threshold": 498.126220703125, | |
| "eval_VitaminC_manhattan_precision": 0.498015873015873, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.55859375, | |
| "eval_VitaminC_max_accuracy_threshold": 311.380615234375, | |
| "eval_VitaminC_max_ap": 0.5544340410314673, | |
| "eval_VitaminC_max_f1": 0.6684709066305818, | |
| "eval_VitaminC_max_f1_threshold": 498.126220703125, | |
| "eval_VitaminC_max_precision": 0.5061475409836066, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5544340410314673, | |
| "eval_sts-test_pearson_cosine": 0.8803067271464453, | |
| "eval_sts-test_pearson_dot": 0.8698285291814508, | |
| "eval_sts-test_pearson_euclidean": 0.9023937835918766, | |
| "eval_sts-test_pearson_manhattan": 0.9020751259156048, | |
| "eval_sts-test_pearson_max": 0.9023937835918766, | |
| "eval_sts-test_spearman_cosine": 0.9038005474254912, | |
| "eval_sts-test_spearman_dot": 0.8707897794601254, | |
| "eval_sts-test_spearman_euclidean": 0.8989733631129851, | |
| "eval_sts-test_spearman_manhattan": 0.8980189529612906, | |
| "eval_sts-test_spearman_max": 0.9038005474254912, | |
| "eval_vitaminc-pairs_loss": 1.7273772954940796, | |
| "eval_vitaminc-pairs_runtime": 1.8924, | |
| "eval_vitaminc-pairs_samples_per_second": 57.071, | |
| "eval_vitaminc-pairs_steps_per_second": 1.057, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_negation-triplets_loss": 0.9174526929855347, | |
| "eval_negation-triplets_runtime": 0.2972, | |
| "eval_negation-triplets_samples_per_second": 215.314, | |
| "eval_negation-triplets_steps_per_second": 3.364, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_scitail-pairs-pos_loss": 0.07368183881044388, | |
| "eval_scitail-pairs-pos_runtime": 0.379, | |
| "eval_scitail-pairs-pos_samples_per_second": 142.492, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.639, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_scitail-pairs-qa_loss": 0.001584450714290142, | |
| "eval_scitail-pairs-qa_runtime": 0.5178, | |
| "eval_scitail-pairs-qa_samples_per_second": 247.198, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.862, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_xsum-pairs_loss": 0.038235221058130264, | |
| "eval_xsum-pairs_runtime": 2.7268, | |
| "eval_xsum-pairs_samples_per_second": 46.941, | |
| "eval_xsum-pairs_steps_per_second": 0.733, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_sciq_pairs_loss": 0.01538097020238638, | |
| "eval_sciq_pairs_runtime": 2.7808, | |
| "eval_sciq_pairs_samples_per_second": 46.029, | |
| "eval_sciq_pairs_steps_per_second": 0.719, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_qasc_pairs_loss": 0.09078988432884216, | |
| "eval_qasc_pairs_runtime": 0.6473, | |
| "eval_qasc_pairs_samples_per_second": 197.758, | |
| "eval_qasc_pairs_steps_per_second": 3.09, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_openbookqa_pairs_loss": 0.6754768490791321, | |
| "eval_openbookqa_pairs_runtime": 0.573, | |
| "eval_openbookqa_pairs_samples_per_second": 223.397, | |
| "eval_openbookqa_pairs_steps_per_second": 3.491, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_msmarco_pairs_loss": 0.15991328656673431, | |
| "eval_msmarco_pairs_runtime": 1.487, | |
| "eval_msmarco_pairs_samples_per_second": 86.078, | |
| "eval_msmarco_pairs_steps_per_second": 1.345, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_nq_pairs_loss": 0.09591890126466751, | |
| "eval_nq_pairs_runtime": 2.3943, | |
| "eval_nq_pairs_samples_per_second": 53.459, | |
| "eval_nq_pairs_steps_per_second": 0.835, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_trivia_pairs_loss": 0.5305934548377991, | |
| "eval_trivia_pairs_runtime": 3.5752, | |
| "eval_trivia_pairs_samples_per_second": 35.802, | |
| "eval_trivia_pairs_steps_per_second": 0.559, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_gooaq_pairs_loss": 0.29681000113487244, | |
| "eval_gooaq_pairs_runtime": 0.9087, | |
| "eval_gooaq_pairs_samples_per_second": 140.861, | |
| "eval_gooaq_pairs_steps_per_second": 2.201, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10112359550561797, | |
| "eval_paws-pos_loss": 0.024501051753759384, | |
| "eval_paws-pos_runtime": 0.6773, | |
| "eval_paws-pos_samples_per_second": 188.996, | |
| "eval_paws-pos_steps_per_second": 2.953, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10337078651685393, | |
| "grad_norm": 2.9021923542022705, | |
| "learning_rate": 9.044943820224718e-06, | |
| "loss": 0.3376, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.10561797752808989, | |
| "grad_norm": 3.179288625717163, | |
| "learning_rate": 9.241573033707863e-06, | |
| "loss": 0.5174, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.10786516853932585, | |
| "grad_norm": 3.1919493675231934, | |
| "learning_rate": 9.43820224719101e-06, | |
| "loss": 0.8162, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1101123595505618, | |
| "grad_norm": 2.8602521419525146, | |
| "learning_rate": 9.634831460674157e-06, | |
| "loss": 0.3545, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.11235955056179775, | |
| "grad_norm": 2.7570478916168213, | |
| "learning_rate": 9.831460674157303e-06, | |
| "loss": 0.315, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1146067415730337, | |
| "grad_norm": 0.8641514778137207, | |
| "learning_rate": 1.0028089887640448e-05, | |
| "loss": 0.0627, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.11685393258426967, | |
| "grad_norm": 3.9437484741210938, | |
| "learning_rate": 1.0224719101123595e-05, | |
| "loss": 0.8851, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.11910112359550562, | |
| "grad_norm": 4.144773006439209, | |
| "learning_rate": 1.042134831460674e-05, | |
| "loss": 0.8382, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12134831460674157, | |
| "grad_norm": 4.277736186981201, | |
| "learning_rate": 1.0617977528089887e-05, | |
| "loss": 0.733, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12359550561797752, | |
| "grad_norm": 4.025904178619385, | |
| "learning_rate": 1.0814606741573032e-05, | |
| "loss": 0.7173, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1258426966292135, | |
| "grad_norm": 3.923046827316284, | |
| "learning_rate": 1.1011235955056178e-05, | |
| "loss": 0.7659, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.12808988764044943, | |
| "grad_norm": 3.2707138061523438, | |
| "learning_rate": 1.1207865168539325e-05, | |
| "loss": 0.793, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1303370786516854, | |
| "grad_norm": 3.1660959720611572, | |
| "learning_rate": 1.1404494382022472e-05, | |
| "loss": 0.5426, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13258426966292136, | |
| "grad_norm": 4.5236663818359375, | |
| "learning_rate": 1.1601123595505617e-05, | |
| "loss": 0.7641, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1348314606741573, | |
| "grad_norm": 0.5771021246910095, | |
| "learning_rate": 1.1797752808988763e-05, | |
| "loss": 0.0657, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13707865168539327, | |
| "grad_norm": 3.8541343212127686, | |
| "learning_rate": 1.1994382022471908e-05, | |
| "loss": 0.7836, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1393258426966292, | |
| "grad_norm": 4.284148693084717, | |
| "learning_rate": 1.2191011235955055e-05, | |
| "loss": 0.9306, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14157303370786517, | |
| "grad_norm": 4.175032615661621, | |
| "learning_rate": 1.23876404494382e-05, | |
| "loss": 0.8673, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14382022471910114, | |
| "grad_norm": 5.025452136993408, | |
| "learning_rate": 1.2584269662921347e-05, | |
| "loss": 0.9296, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.14606741573033707, | |
| "grad_norm": 3.970745086669922, | |
| "learning_rate": 1.2780898876404493e-05, | |
| "loss": 0.8211, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14831460674157304, | |
| "grad_norm": 3.150197744369507, | |
| "learning_rate": 1.297752808988764e-05, | |
| "loss": 0.7685, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.15056179775280898, | |
| "grad_norm": 4.280994415283203, | |
| "learning_rate": 1.3174157303370785e-05, | |
| "loss": 0.7139, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.15280898876404495, | |
| "grad_norm": 4.288730621337891, | |
| "learning_rate": 1.3370786516853932e-05, | |
| "loss": 0.8241, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1550561797752809, | |
| "grad_norm": 3.7402424812316895, | |
| "learning_rate": 1.3567415730337077e-05, | |
| "loss": 0.6256, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.15730337078651685, | |
| "grad_norm": 4.478890895843506, | |
| "learning_rate": 1.3764044943820223e-05, | |
| "loss": 0.8842, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15955056179775282, | |
| "grad_norm": 3.8147876262664795, | |
| "learning_rate": 1.3960674157303368e-05, | |
| "loss": 0.804, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16179775280898875, | |
| "grad_norm": 0.7314035296440125, | |
| "learning_rate": 1.4157303370786515e-05, | |
| "loss": 0.0989, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.16404494382022472, | |
| "grad_norm": 3.074303150177002, | |
| "learning_rate": 1.4353932584269662e-05, | |
| "loss": 0.332, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.1662921348314607, | |
| "grad_norm": 3.414987325668335, | |
| "learning_rate": 1.4550561797752808e-05, | |
| "loss": 0.5736, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.16853932584269662, | |
| "grad_norm": 3.7946674823760986, | |
| "learning_rate": 1.4747191011235953e-05, | |
| "loss": 0.8285, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1707865168539326, | |
| "grad_norm": 4.310474395751953, | |
| "learning_rate": 1.49438202247191e-05, | |
| "loss": 0.9561, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.17303370786516853, | |
| "grad_norm": 0.9791378974914551, | |
| "learning_rate": 1.5140449438202245e-05, | |
| "loss": 0.0633, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1752808988764045, | |
| "grad_norm": 0.6351795196533203, | |
| "learning_rate": 1.5337078651685393e-05, | |
| "loss": 0.0848, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.17752808988764046, | |
| "grad_norm": 3.4832303524017334, | |
| "learning_rate": 1.553370786516854e-05, | |
| "loss": 0.8325, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1797752808988764, | |
| "grad_norm": 5.115800380706787, | |
| "learning_rate": 1.5730337078651683e-05, | |
| "loss": 1.0011, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18202247191011237, | |
| "grad_norm": 3.552396297454834, | |
| "learning_rate": 1.592696629213483e-05, | |
| "loss": 0.8697, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.1842696629213483, | |
| "grad_norm": 4.491541862487793, | |
| "learning_rate": 1.6123595505617977e-05, | |
| "loss": 0.8344, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.18651685393258427, | |
| "grad_norm": 4.73278284072876, | |
| "learning_rate": 1.6320224719101122e-05, | |
| "loss": 0.9967, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.18876404494382024, | |
| "grad_norm": 2.994192123413086, | |
| "learning_rate": 1.6516853932584267e-05, | |
| "loss": 0.4638, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.19101123595505617, | |
| "grad_norm": 4.142394542694092, | |
| "learning_rate": 1.6713483146067415e-05, | |
| "loss": 0.8994, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.19325842696629214, | |
| "grad_norm": 4.149839401245117, | |
| "learning_rate": 1.691011235955056e-05, | |
| "loss": 0.7789, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.19550561797752808, | |
| "grad_norm": 0.45795938372612, | |
| "learning_rate": 1.7106741573033705e-05, | |
| "loss": 0.0555, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.19775280898876405, | |
| "grad_norm": 3.4293618202209473, | |
| "learning_rate": 1.7303370786516853e-05, | |
| "loss": 0.3778, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 4.041529655456543, | |
| "learning_rate": 1.75e-05, | |
| "loss": 0.708, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "grad_norm": 0.6160458922386169, | |
| "learning_rate": 1.7696629213483143e-05, | |
| "loss": 0.0689, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_VitaminC_cosine_accuracy": 0.556640625, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.74173903465271, | |
| "eval_VitaminC_cosine_ap": 0.5513770735348443, | |
| "eval_VitaminC_cosine_f1": 0.6675531914893617, | |
| "eval_VitaminC_cosine_f1_threshold": 0.32480987906455994, | |
| "eval_VitaminC_cosine_precision": 0.500998003992016, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.560546875, | |
| "eval_VitaminC_dot_accuracy_threshold": 297.664794921875, | |
| "eval_VitaminC_dot_ap": 0.5340088824099496, | |
| "eval_VitaminC_dot_f1": 0.6666666666666667, | |
| "eval_VitaminC_dot_f1_threshold": 126.67618560791016, | |
| "eval_VitaminC_dot_precision": 0.501002004008016, | |
| "eval_VitaminC_dot_recall": 0.9960159362549801, | |
| "eval_VitaminC_euclidean_accuracy": 0.55859375, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 14.345688819885254, | |
| "eval_VitaminC_euclidean_ap": 0.5542145004976253, | |
| "eval_VitaminC_euclidean_f1": 0.6675531914893617, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.381019592285156, | |
| "eval_VitaminC_euclidean_precision": 0.500998003992016, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.552734375, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 232.7296142578125, | |
| "eval_VitaminC_manhattan_ap": 0.5523953693907266, | |
| "eval_VitaminC_manhattan_f1": 0.6657824933687002, | |
| "eval_VitaminC_manhattan_f1_threshold": 496.4290466308594, | |
| "eval_VitaminC_manhattan_precision": 0.4990059642147117, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.560546875, | |
| "eval_VitaminC_max_accuracy_threshold": 297.664794921875, | |
| "eval_VitaminC_max_ap": 0.5542145004976253, | |
| "eval_VitaminC_max_f1": 0.6675531914893617, | |
| "eval_VitaminC_max_f1_threshold": 496.4290466308594, | |
| "eval_VitaminC_max_precision": 0.501002004008016, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5542145004976253, | |
| "eval_sts-test_pearson_cosine": 0.8800782580988616, | |
| "eval_sts-test_pearson_dot": 0.8687642290872662, | |
| "eval_sts-test_pearson_euclidean": 0.9034088230546415, | |
| "eval_sts-test_pearson_manhattan": 0.9030146212284895, | |
| "eval_sts-test_pearson_max": 0.9034088230546415, | |
| "eval_sts-test_spearman_cosine": 0.904560289590133, | |
| "eval_sts-test_spearman_dot": 0.8705944849554133, | |
| "eval_sts-test_spearman_euclidean": 0.8998959103665689, | |
| "eval_sts-test_spearman_manhattan": 0.8995891404697307, | |
| "eval_sts-test_spearman_max": 0.904560289590133, | |
| "eval_vitaminc-pairs_loss": 1.6141985654830933, | |
| "eval_vitaminc-pairs_runtime": 1.864, | |
| "eval_vitaminc-pairs_samples_per_second": 57.94, | |
| "eval_vitaminc-pairs_steps_per_second": 1.073, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_negation-triplets_loss": 0.9220322370529175, | |
| "eval_negation-triplets_runtime": 0.3199, | |
| "eval_negation-triplets_samples_per_second": 200.043, | |
| "eval_negation-triplets_steps_per_second": 3.126, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_scitail-pairs-pos_loss": 0.0654294565320015, | |
| "eval_scitail-pairs-pos_runtime": 0.4625, | |
| "eval_scitail-pairs-pos_samples_per_second": 116.76, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.162, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_scitail-pairs-qa_loss": 0.0015887805493548512, | |
| "eval_scitail-pairs-qa_runtime": 0.5768, | |
| "eval_scitail-pairs-qa_samples_per_second": 221.899, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.467, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_xsum-pairs_loss": 0.03991687670350075, | |
| "eval_xsum-pairs_runtime": 2.7403, | |
| "eval_xsum-pairs_samples_per_second": 46.71, | |
| "eval_xsum-pairs_steps_per_second": 0.73, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_sciq_pairs_loss": 0.01584962010383606, | |
| "eval_sciq_pairs_runtime": 2.8429, | |
| "eval_sciq_pairs_samples_per_second": 45.024, | |
| "eval_sciq_pairs_steps_per_second": 0.703, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_qasc_pairs_loss": 0.09112343192100525, | |
| "eval_qasc_pairs_runtime": 0.6492, | |
| "eval_qasc_pairs_samples_per_second": 197.154, | |
| "eval_qasc_pairs_steps_per_second": 3.081, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_openbookqa_pairs_loss": 0.7132729887962341, | |
| "eval_openbookqa_pairs_runtime": 0.5847, | |
| "eval_openbookqa_pairs_samples_per_second": 218.922, | |
| "eval_openbookqa_pairs_steps_per_second": 3.421, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_msmarco_pairs_loss": 0.15173853933811188, | |
| "eval_msmarco_pairs_runtime": 1.4966, | |
| "eval_msmarco_pairs_samples_per_second": 85.527, | |
| "eval_msmarco_pairs_steps_per_second": 1.336, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_nq_pairs_loss": 0.09653442353010178, | |
| "eval_nq_pairs_runtime": 2.3749, | |
| "eval_nq_pairs_samples_per_second": 53.897, | |
| "eval_nq_pairs_steps_per_second": 0.842, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_trivia_pairs_loss": 0.5191965699195862, | |
| "eval_trivia_pairs_runtime": 3.6006, | |
| "eval_trivia_pairs_samples_per_second": 35.55, | |
| "eval_trivia_pairs_steps_per_second": 0.555, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_gooaq_pairs_loss": 0.30713126063346863, | |
| "eval_gooaq_pairs_runtime": 0.9131, | |
| "eval_gooaq_pairs_samples_per_second": 140.178, | |
| "eval_gooaq_pairs_steps_per_second": 2.19, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20224719101123595, | |
| "eval_paws-pos_loss": 0.024471310898661613, | |
| "eval_paws-pos_runtime": 0.6872, | |
| "eval_paws-pos_samples_per_second": 186.254, | |
| "eval_paws-pos_steps_per_second": 2.91, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20449438202247192, | |
| "grad_norm": 6.209661483764648, | |
| "learning_rate": 1.7893258426966292e-05, | |
| "loss": 2.3489, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.20674157303370785, | |
| "grad_norm": 3.1821141242980957, | |
| "learning_rate": 1.8089887640449437e-05, | |
| "loss": 0.741, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.20898876404494382, | |
| "grad_norm": 3.871994972229004, | |
| "learning_rate": 1.8286516853932585e-05, | |
| "loss": 0.7729, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.21123595505617979, | |
| "grad_norm": 0.5280765891075134, | |
| "learning_rate": 1.8483146067415727e-05, | |
| "loss": 0.0631, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.21348314606741572, | |
| "grad_norm": 4.475915431976318, | |
| "learning_rate": 1.8679775280898875e-05, | |
| "loss": 0.9342, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2157303370786517, | |
| "grad_norm": 3.949381113052368, | |
| "learning_rate": 1.887640449438202e-05, | |
| "loss": 0.8581, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.21797752808988763, | |
| "grad_norm": 2.910426616668701, | |
| "learning_rate": 1.907303370786517e-05, | |
| "loss": 0.5198, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.2202247191011236, | |
| "grad_norm": 4.028941631317139, | |
| "learning_rate": 1.9269662921348313e-05, | |
| "loss": 0.846, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.22247191011235956, | |
| "grad_norm": 4.183433532714844, | |
| "learning_rate": 1.946629213483146e-05, | |
| "loss": 0.6581, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2247191011235955, | |
| "grad_norm": 3.348114252090454, | |
| "learning_rate": 1.9662921348314607e-05, | |
| "loss": 0.3579, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22696629213483147, | |
| "grad_norm": 4.055211544036865, | |
| "learning_rate": 1.9859550561797752e-05, | |
| "loss": 0.908, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.2292134831460674, | |
| "grad_norm": 1.0024710893630981, | |
| "learning_rate": 2.0056179775280897e-05, | |
| "loss": 0.0664, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.23146067415730337, | |
| "grad_norm": 3.582249641418457, | |
| "learning_rate": 2.0252808988764042e-05, | |
| "loss": 0.5411, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.23370786516853934, | |
| "grad_norm": 4.226349830627441, | |
| "learning_rate": 2.044943820224719e-05, | |
| "loss": 0.9163, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.23595505617977527, | |
| "grad_norm": 3.002727508544922, | |
| "learning_rate": 2.0646067415730335e-05, | |
| "loss": 0.7975, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.23820224719101124, | |
| "grad_norm": 3.5497515201568604, | |
| "learning_rate": 2.084269662921348e-05, | |
| "loss": 0.37, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.24044943820224718, | |
| "grad_norm": 4.381045341491699, | |
| "learning_rate": 2.103932584269663e-05, | |
| "loss": 0.8495, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.24269662921348314, | |
| "grad_norm": 3.926840305328369, | |
| "learning_rate": 2.1235955056179773e-05, | |
| "loss": 0.8073, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.2449438202247191, | |
| "grad_norm": 3.0835390090942383, | |
| "learning_rate": 2.1432584269662922e-05, | |
| "loss": 0.7563, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.24719101123595505, | |
| "grad_norm": 4.230669975280762, | |
| "learning_rate": 2.1629213483146063e-05, | |
| "loss": 0.6585, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24943820224719102, | |
| "grad_norm": 2.8849070072174072, | |
| "learning_rate": 2.1825842696629212e-05, | |
| "loss": 0.3246, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.251685393258427, | |
| "grad_norm": 4.796951770782471, | |
| "learning_rate": 2.2022471910112357e-05, | |
| "loss": 0.9718, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2539325842696629, | |
| "grad_norm": 4.60318660736084, | |
| "learning_rate": 2.2219101123595505e-05, | |
| "loss": 0.8584, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.25617977528089886, | |
| "grad_norm": 3.098703384399414, | |
| "learning_rate": 2.241573033707865e-05, | |
| "loss": 0.3385, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.25842696629213485, | |
| "grad_norm": 2.9519224166870117, | |
| "learning_rate": 2.2612359550561795e-05, | |
| "loss": 0.323, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2606741573033708, | |
| "grad_norm": 2.913742780685425, | |
| "learning_rate": 2.2808988764044944e-05, | |
| "loss": 0.3359, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.26292134831460673, | |
| "grad_norm": 4.148440837860107, | |
| "learning_rate": 2.300561797752809e-05, | |
| "loss": 0.6955, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2651685393258427, | |
| "grad_norm": 0.8463248610496521, | |
| "learning_rate": 2.3202247191011234e-05, | |
| "loss": 0.0539, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.26741573033707866, | |
| "grad_norm": 0.7284589409828186, | |
| "learning_rate": 2.339887640449438e-05, | |
| "loss": 0.0507, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2696629213483146, | |
| "grad_norm": 3.615086317062378, | |
| "learning_rate": 2.3595505617977527e-05, | |
| "loss": 0.314, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.27191011235955054, | |
| "grad_norm": 5.229820728302002, | |
| "learning_rate": 2.3792134831460672e-05, | |
| "loss": 1.0339, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.27415730337078653, | |
| "grad_norm": 3.6847782135009766, | |
| "learning_rate": 2.3988764044943817e-05, | |
| "loss": 0.3158, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.27640449438202247, | |
| "grad_norm": 4.280517578125, | |
| "learning_rate": 2.4185393258426965e-05, | |
| "loss": 0.7809, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.2786516853932584, | |
| "grad_norm": 4.476150035858154, | |
| "learning_rate": 2.438202247191011e-05, | |
| "loss": 0.9516, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2808988764044944, | |
| "grad_norm": 2.7380239963531494, | |
| "learning_rate": 2.457865168539326e-05, | |
| "loss": 0.3117, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28314606741573034, | |
| "grad_norm": 3.9667162895202637, | |
| "learning_rate": 2.47752808988764e-05, | |
| "loss": 0.8366, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2853932584269663, | |
| "grad_norm": 4.552999019622803, | |
| "learning_rate": 2.497191011235955e-05, | |
| "loss": 0.8033, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.2876404494382023, | |
| "grad_norm": 3.4238576889038086, | |
| "learning_rate": 2.5168539325842694e-05, | |
| "loss": 0.7253, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.2898876404494382, | |
| "grad_norm": 4.677807331085205, | |
| "learning_rate": 2.5365168539325842e-05, | |
| "loss": 0.8345, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.29213483146067415, | |
| "grad_norm": 4.282113075256348, | |
| "learning_rate": 2.5561797752808987e-05, | |
| "loss": 0.7532, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2943820224719101, | |
| "grad_norm": 4.375221252441406, | |
| "learning_rate": 2.5758426966292132e-05, | |
| "loss": 0.8247, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2966292134831461, | |
| "grad_norm": 3.2591633796691895, | |
| "learning_rate": 2.595505617977528e-05, | |
| "loss": 0.5175, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.298876404494382, | |
| "grad_norm": 4.146636962890625, | |
| "learning_rate": 2.6151685393258425e-05, | |
| "loss": 0.7813, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.30112359550561796, | |
| "grad_norm": 4.2413249015808105, | |
| "learning_rate": 2.634831460674157e-05, | |
| "loss": 0.6582, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "grad_norm": 4.541455268859863, | |
| "learning_rate": 2.6544943820224715e-05, | |
| "loss": 0.3484, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_VitaminC_cosine_accuracy": 0.560546875, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.7956135272979736, | |
| "eval_VitaminC_cosine_ap": 0.5505565383154402, | |
| "eval_VitaminC_cosine_f1": 0.6684709066305818, | |
| "eval_VitaminC_cosine_f1_threshold": 0.40466147661209106, | |
| "eval_VitaminC_cosine_precision": 0.5061475409836066, | |
| "eval_VitaminC_cosine_recall": 0.9840637450199203, | |
| "eval_VitaminC_dot_accuracy": 0.5546875, | |
| "eval_VitaminC_dot_accuracy_threshold": 312.2774658203125, | |
| "eval_VitaminC_dot_ap": 0.5365135091766033, | |
| "eval_VitaminC_dot_f1": 0.6684856753069577, | |
| "eval_VitaminC_dot_f1_threshold": 157.33203125, | |
| "eval_VitaminC_dot_precision": 0.508298755186722, | |
| "eval_VitaminC_dot_recall": 0.9760956175298805, | |
| "eval_VitaminC_euclidean_accuracy": 0.5546875, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 12.184114456176758, | |
| "eval_VitaminC_euclidean_ap": 0.5517706579195627, | |
| "eval_VitaminC_euclidean_f1": 0.6649006622516557, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.68879508972168, | |
| "eval_VitaminC_euclidean_precision": 0.498015873015873, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.5546875, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 241.22061157226562, | |
| "eval_VitaminC_manhattan_ap": 0.5494156168773414, | |
| "eval_VitaminC_manhattan_f1": 0.6649006622516557, | |
| "eval_VitaminC_manhattan_f1_threshold": 510.2530212402344, | |
| "eval_VitaminC_manhattan_precision": 0.498015873015873, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.560546875, | |
| "eval_VitaminC_max_accuracy_threshold": 312.2774658203125, | |
| "eval_VitaminC_max_ap": 0.5517706579195627, | |
| "eval_VitaminC_max_f1": 0.6684856753069577, | |
| "eval_VitaminC_max_f1_threshold": 510.2530212402344, | |
| "eval_VitaminC_max_precision": 0.508298755186722, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5517706579195627, | |
| "eval_sts-test_pearson_cosine": 0.8812438499723412, | |
| "eval_sts-test_pearson_dot": 0.8695651753004092, | |
| "eval_sts-test_pearson_euclidean": 0.9036940037118162, | |
| "eval_sts-test_pearson_manhattan": 0.9035516699922166, | |
| "eval_sts-test_pearson_max": 0.9036940037118162, | |
| "eval_sts-test_spearman_cosine": 0.9049742835092648, | |
| "eval_sts-test_spearman_dot": 0.8707925987895928, | |
| "eval_sts-test_spearman_euclidean": 0.9003956924537878, | |
| "eval_sts-test_spearman_manhattan": 0.9002747745455083, | |
| "eval_sts-test_spearman_max": 0.9049742835092648, | |
| "eval_vitaminc-pairs_loss": 1.5520410537719727, | |
| "eval_vitaminc-pairs_runtime": 1.8323, | |
| "eval_vitaminc-pairs_samples_per_second": 58.943, | |
| "eval_vitaminc-pairs_steps_per_second": 1.092, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_negation-triplets_loss": 0.9211694002151489, | |
| "eval_negation-triplets_runtime": 0.2923, | |
| "eval_negation-triplets_samples_per_second": 218.93, | |
| "eval_negation-triplets_steps_per_second": 3.421, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_scitail-pairs-pos_loss": 0.07377135753631592, | |
| "eval_scitail-pairs-pos_runtime": 0.3681, | |
| "eval_scitail-pairs-pos_samples_per_second": 146.691, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.716, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_scitail-pairs-qa_loss": 0.00150959100574255, | |
| "eval_scitail-pairs-qa_runtime": 0.5123, | |
| "eval_scitail-pairs-qa_samples_per_second": 249.842, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.904, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_xsum-pairs_loss": 0.036599572747945786, | |
| "eval_xsum-pairs_runtime": 2.7238, | |
| "eval_xsum-pairs_samples_per_second": 46.994, | |
| "eval_xsum-pairs_steps_per_second": 0.734, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_sciq_pairs_loss": 0.01615014858543873, | |
| "eval_sciq_pairs_runtime": 2.8064, | |
| "eval_sciq_pairs_samples_per_second": 45.61, | |
| "eval_sciq_pairs_steps_per_second": 0.713, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_qasc_pairs_loss": 0.09235507994890213, | |
| "eval_qasc_pairs_runtime": 0.6488, | |
| "eval_qasc_pairs_samples_per_second": 197.276, | |
| "eval_qasc_pairs_steps_per_second": 3.082, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_openbookqa_pairs_loss": 0.6891775727272034, | |
| "eval_openbookqa_pairs_runtime": 0.5698, | |
| "eval_openbookqa_pairs_samples_per_second": 224.641, | |
| "eval_openbookqa_pairs_steps_per_second": 3.51, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_msmarco_pairs_loss": 0.16766037046909332, | |
| "eval_msmarco_pairs_runtime": 1.4798, | |
| "eval_msmarco_pairs_samples_per_second": 86.499, | |
| "eval_msmarco_pairs_steps_per_second": 1.352, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_nq_pairs_loss": 0.09737721085548401, | |
| "eval_nq_pairs_runtime": 2.3409, | |
| "eval_nq_pairs_samples_per_second": 54.68, | |
| "eval_nq_pairs_steps_per_second": 0.854, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_trivia_pairs_loss": 0.5458433032035828, | |
| "eval_trivia_pairs_runtime": 3.5771, | |
| "eval_trivia_pairs_samples_per_second": 35.783, | |
| "eval_trivia_pairs_steps_per_second": 0.559, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_gooaq_pairs_loss": 0.3082329332828522, | |
| "eval_gooaq_pairs_runtime": 0.9181, | |
| "eval_gooaq_pairs_samples_per_second": 139.413, | |
| "eval_gooaq_pairs_steps_per_second": 2.178, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30337078651685395, | |
| "eval_paws-pos_loss": 0.02423396334052086, | |
| "eval_paws-pos_runtime": 0.6827, | |
| "eval_paws-pos_samples_per_second": 187.501, | |
| "eval_paws-pos_steps_per_second": 2.93, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3056179775280899, | |
| "grad_norm": 4.549901485443115, | |
| "learning_rate": 2.6741573033707864e-05, | |
| "loss": 0.7648, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.30786516853932583, | |
| "grad_norm": 3.225851535797119, | |
| "learning_rate": 2.693820224719101e-05, | |
| "loss": 0.7554, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3101123595505618, | |
| "grad_norm": 0.6228423118591309, | |
| "learning_rate": 2.7134831460674154e-05, | |
| "loss": 0.0753, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.31235955056179776, | |
| "grad_norm": 3.12802791595459, | |
| "learning_rate": 2.7331460674157302e-05, | |
| "loss": 0.4987, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3146067415730337, | |
| "grad_norm": 4.1997880935668945, | |
| "learning_rate": 2.7528089887640447e-05, | |
| "loss": 0.8543, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31685393258426964, | |
| "grad_norm": 4.3362860679626465, | |
| "learning_rate": 2.7724719101123595e-05, | |
| "loss": 0.9425, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.31910112359550563, | |
| "grad_norm": 0.5599316954612732, | |
| "learning_rate": 2.7921348314606737e-05, | |
| "loss": 0.0472, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.32134831460674157, | |
| "grad_norm": 3.503603458404541, | |
| "learning_rate": 2.8117977528089885e-05, | |
| "loss": 0.848, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.3235955056179775, | |
| "grad_norm": 4.712310314178467, | |
| "learning_rate": 2.831460674157303e-05, | |
| "loss": 0.8946, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3258426966292135, | |
| "grad_norm": 3.1823527812957764, | |
| "learning_rate": 2.851123595505618e-05, | |
| "loss": 0.7841, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.32808988764044944, | |
| "grad_norm": 4.423196315765381, | |
| "learning_rate": 2.8707865168539324e-05, | |
| "loss": 0.6653, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.3303370786516854, | |
| "grad_norm": 4.137822151184082, | |
| "learning_rate": 2.890449438202247e-05, | |
| "loss": 0.3522, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.3325842696629214, | |
| "grad_norm": 2.997777223587036, | |
| "learning_rate": 2.9101123595505617e-05, | |
| "loss": 0.4853, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.3348314606741573, | |
| "grad_norm": 2.89650559425354, | |
| "learning_rate": 2.9297752808988762e-05, | |
| "loss": 0.4726, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.33707865168539325, | |
| "grad_norm": 5.486624717712402, | |
| "learning_rate": 2.9494382022471907e-05, | |
| "loss": 0.8693, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3393258426966292, | |
| "grad_norm": 4.800889015197754, | |
| "learning_rate": 2.9691011235955052e-05, | |
| "loss": 0.8124, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.3415730337078652, | |
| "grad_norm": 4.188066005706787, | |
| "learning_rate": 2.98876404494382e-05, | |
| "loss": 0.8206, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.3438202247191011, | |
| "grad_norm": 4.340461254119873, | |
| "learning_rate": 3.0084269662921345e-05, | |
| "loss": 0.9406, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.34606741573033706, | |
| "grad_norm": 4.658304214477539, | |
| "learning_rate": 3.028089887640449e-05, | |
| "loss": 0.7944, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.34831460674157305, | |
| "grad_norm": 0.6266987919807434, | |
| "learning_rate": 3.047752808988764e-05, | |
| "loss": 0.0766, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.350561797752809, | |
| "grad_norm": 4.252346515655518, | |
| "learning_rate": 3.067415730337079e-05, | |
| "loss": 0.8609, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.35280898876404493, | |
| "grad_norm": 4.9649658203125, | |
| "learning_rate": 3.087078651685393e-05, | |
| "loss": 1.0533, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3550561797752809, | |
| "grad_norm": 4.485607624053955, | |
| "learning_rate": 3.106741573033708e-05, | |
| "loss": 0.8396, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.35730337078651686, | |
| "grad_norm": 3.241231918334961, | |
| "learning_rate": 3.126404494382022e-05, | |
| "loss": 0.7865, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3595505617977528, | |
| "grad_norm": 6.846582889556885, | |
| "learning_rate": 3.146067415730337e-05, | |
| "loss": 2.4616, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.36179775280898874, | |
| "grad_norm": 0.5514687895774841, | |
| "learning_rate": 3.165730337078651e-05, | |
| "loss": 0.0556, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.36404494382022473, | |
| "grad_norm": 3.7877562046051025, | |
| "learning_rate": 3.185393258426966e-05, | |
| "loss": 0.3758, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.36629213483146067, | |
| "grad_norm": 5.397939682006836, | |
| "learning_rate": 3.205056179775281e-05, | |
| "loss": 0.9312, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.3685393258426966, | |
| "grad_norm": 4.301459312438965, | |
| "learning_rate": 3.2247191011235954e-05, | |
| "loss": 0.7993, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.3707865168539326, | |
| "grad_norm": 4.49428129196167, | |
| "learning_rate": 3.24438202247191e-05, | |
| "loss": 0.8104, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.37303370786516854, | |
| "grad_norm": 3.2210912704467773, | |
| "learning_rate": 3.2640449438202244e-05, | |
| "loss": 0.8199, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.3752808988764045, | |
| "grad_norm": 5.359859466552734, | |
| "learning_rate": 3.283707865168539e-05, | |
| "loss": 1.0724, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.3775280898876405, | |
| "grad_norm": 4.00059700012207, | |
| "learning_rate": 3.3033707865168534e-05, | |
| "loss": 0.3521, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.3797752808988764, | |
| "grad_norm": 4.418768882751465, | |
| "learning_rate": 3.3230337078651685e-05, | |
| "loss": 0.8536, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.38202247191011235, | |
| "grad_norm": 4.15454626083374, | |
| "learning_rate": 3.342696629213483e-05, | |
| "loss": 0.872, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3842696629213483, | |
| "grad_norm": 3.8060054779052734, | |
| "learning_rate": 3.3623595505617975e-05, | |
| "loss": 0.8009, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3865168539325843, | |
| "grad_norm": 3.584745407104492, | |
| "learning_rate": 3.382022471910112e-05, | |
| "loss": 0.7798, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3887640449438202, | |
| "grad_norm": 4.861410140991211, | |
| "learning_rate": 3.4016853932584265e-05, | |
| "loss": 0.5953, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.39101123595505616, | |
| "grad_norm": 3.983793020248413, | |
| "learning_rate": 3.421348314606741e-05, | |
| "loss": 0.7562, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.39325842696629215, | |
| "grad_norm": 4.841738224029541, | |
| "learning_rate": 3.4410112359550555e-05, | |
| "loss": 0.7227, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3955056179775281, | |
| "grad_norm": 4.787370204925537, | |
| "learning_rate": 3.460674157303371e-05, | |
| "loss": 0.8953, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.39775280898876403, | |
| "grad_norm": 4.337812900543213, | |
| "learning_rate": 3.480337078651685e-05, | |
| "loss": 0.7102, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.9599294662475586, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.0667, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.40224719101123596, | |
| "grad_norm": 0.6864398717880249, | |
| "learning_rate": 3.4999863718440846e-05, | |
| "loss": 0.0528, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "grad_norm": 4.738316059112549, | |
| "learning_rate": 3.499945487641664e-05, | |
| "loss": 0.7312, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_VitaminC_cosine_accuracy": 0.556640625, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8256886005401611, | |
| "eval_VitaminC_cosine_ap": 0.5557251062538118, | |
| "eval_VitaminC_cosine_f1": 0.6666666666666667, | |
| "eval_VitaminC_cosine_f1_threshold": 0.4391498863697052, | |
| "eval_VitaminC_cosine_precision": 0.5051334702258727, | |
| "eval_VitaminC_cosine_recall": 0.9800796812749004, | |
| "eval_VitaminC_dot_accuracy": 0.556640625, | |
| "eval_VitaminC_dot_accuracy_threshold": 314.2790832519531, | |
| "eval_VitaminC_dot_ap": 0.5397120960874565, | |
| "eval_VitaminC_dot_f1": 0.6684636118598383, | |
| "eval_VitaminC_dot_f1_threshold": 144.02464294433594, | |
| "eval_VitaminC_dot_precision": 0.505091649694501, | |
| "eval_VitaminC_dot_recall": 0.9880478087649402, | |
| "eval_VitaminC_euclidean_accuracy": 0.560546875, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 13.859346389770508, | |
| "eval_VitaminC_euclidean_ap": 0.5582755831276058, | |
| "eval_VitaminC_euclidean_f1": 0.667605633802817, | |
| "eval_VitaminC_euclidean_f1_threshold": 18.874879837036133, | |
| "eval_VitaminC_euclidean_precision": 0.5163398692810458, | |
| "eval_VitaminC_euclidean_recall": 0.9442231075697212, | |
| "eval_VitaminC_manhattan_accuracy": 0.560546875, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 239.6153564453125, | |
| "eval_VitaminC_manhattan_ap": 0.5569115785564898, | |
| "eval_VitaminC_manhattan_f1": 0.6649006622516557, | |
| "eval_VitaminC_manhattan_f1_threshold": 501.158447265625, | |
| "eval_VitaminC_manhattan_precision": 0.498015873015873, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.560546875, | |
| "eval_VitaminC_max_accuracy_threshold": 314.2790832519531, | |
| "eval_VitaminC_max_ap": 0.5582755831276058, | |
| "eval_VitaminC_max_f1": 0.6684636118598383, | |
| "eval_VitaminC_max_f1_threshold": 501.158447265625, | |
| "eval_VitaminC_max_precision": 0.5163398692810458, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5582755831276058, | |
| "eval_sts-test_pearson_cosine": 0.8825432226222443, | |
| "eval_sts-test_pearson_dot": 0.8720125241659442, | |
| "eval_sts-test_pearson_euclidean": 0.9053801707227738, | |
| "eval_sts-test_pearson_manhattan": 0.9060044572091359, | |
| "eval_sts-test_pearson_max": 0.9060044572091359, | |
| "eval_sts-test_spearman_cosine": 0.9055030196626042, | |
| "eval_sts-test_spearman_dot": 0.8729395405548455, | |
| "eval_sts-test_spearman_euclidean": 0.9013990604854444, | |
| "eval_sts-test_spearman_manhattan": 0.9021052353902007, | |
| "eval_sts-test_spearman_max": 0.9055030196626042, | |
| "eval_vitaminc-pairs_loss": 1.5215541124343872, | |
| "eval_vitaminc-pairs_runtime": 1.8745, | |
| "eval_vitaminc-pairs_samples_per_second": 57.614, | |
| "eval_vitaminc-pairs_steps_per_second": 1.067, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_negation-triplets_loss": 0.9813100099563599, | |
| "eval_negation-triplets_runtime": 0.3009, | |
| "eval_negation-triplets_samples_per_second": 212.73, | |
| "eval_negation-triplets_steps_per_second": 3.324, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_scitail-pairs-pos_loss": 0.09161412715911865, | |
| "eval_scitail-pairs-pos_runtime": 0.3936, | |
| "eval_scitail-pairs-pos_samples_per_second": 137.188, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.541, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_scitail-pairs-qa_loss": 0.0013133077882230282, | |
| "eval_scitail-pairs-qa_runtime": 0.5286, | |
| "eval_scitail-pairs-qa_samples_per_second": 242.147, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.784, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_xsum-pairs_loss": 0.049595557153224945, | |
| "eval_xsum-pairs_runtime": 2.7447, | |
| "eval_xsum-pairs_samples_per_second": 46.636, | |
| "eval_xsum-pairs_steps_per_second": 0.729, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_sciq_pairs_loss": 0.017273178324103355, | |
| "eval_sciq_pairs_runtime": 2.8401, | |
| "eval_sciq_pairs_samples_per_second": 45.069, | |
| "eval_sciq_pairs_steps_per_second": 0.704, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_qasc_pairs_loss": 0.09485691040754318, | |
| "eval_qasc_pairs_runtime": 0.6594, | |
| "eval_qasc_pairs_samples_per_second": 194.113, | |
| "eval_qasc_pairs_steps_per_second": 3.033, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_openbookqa_pairs_loss": 0.7253161072731018, | |
| "eval_openbookqa_pairs_runtime": 0.5801, | |
| "eval_openbookqa_pairs_samples_per_second": 220.633, | |
| "eval_openbookqa_pairs_steps_per_second": 3.447, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_msmarco_pairs_loss": 0.17383378744125366, | |
| "eval_msmarco_pairs_runtime": 1.4824, | |
| "eval_msmarco_pairs_samples_per_second": 86.346, | |
| "eval_msmarco_pairs_steps_per_second": 1.349, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_nq_pairs_loss": 0.10324681550264359, | |
| "eval_nq_pairs_runtime": 2.3542, | |
| "eval_nq_pairs_samples_per_second": 54.372, | |
| "eval_nq_pairs_steps_per_second": 0.85, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_trivia_pairs_loss": 0.5358972549438477, | |
| "eval_trivia_pairs_runtime": 3.5881, | |
| "eval_trivia_pairs_samples_per_second": 35.673, | |
| "eval_trivia_pairs_steps_per_second": 0.557, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_gooaq_pairs_loss": 0.3070329427719116, | |
| "eval_gooaq_pairs_runtime": 0.9009, | |
| "eval_gooaq_pairs_samples_per_second": 142.079, | |
| "eval_gooaq_pairs_steps_per_second": 2.22, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4044943820224719, | |
| "eval_paws-pos_loss": 0.024055125191807747, | |
| "eval_paws-pos_runtime": 0.6792, | |
| "eval_paws-pos_samples_per_second": 188.469, | |
| "eval_paws-pos_steps_per_second": 2.945, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4067415730337079, | |
| "grad_norm": 5.063413143157959, | |
| "learning_rate": 3.4998773481887046e-05, | |
| "loss": 0.7809, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.40898876404494383, | |
| "grad_norm": 4.108719825744629, | |
| "learning_rate": 3.499781954811798e-05, | |
| "loss": 0.8333, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.41123595505617977, | |
| "grad_norm": 4.6362104415893555, | |
| "learning_rate": 3.499659309368139e-05, | |
| "loss": 0.9283, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.4134831460674157, | |
| "grad_norm": 4.432968616485596, | |
| "learning_rate": 3.499509414245486e-05, | |
| "loss": 0.7011, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.4157303370786517, | |
| "grad_norm": 4.040768623352051, | |
| "learning_rate": 3.4993322723621164e-05, | |
| "loss": 0.8413, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.41797752808988764, | |
| "grad_norm": 5.797406196594238, | |
| "learning_rate": 3.499127887166769e-05, | |
| "loss": 1.1679, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.4202247191011236, | |
| "grad_norm": 4.275143623352051, | |
| "learning_rate": 3.498896262638578e-05, | |
| "loss": 0.8701, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.42247191011235957, | |
| "grad_norm": 3.920672655105591, | |
| "learning_rate": 3.498637403286993e-05, | |
| "loss": 0.8139, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.4247191011235955, | |
| "grad_norm": 4.049210071563721, | |
| "learning_rate": 3.498351314151693e-05, | |
| "loss": 0.664, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.42696629213483145, | |
| "grad_norm": 4.007586479187012, | |
| "learning_rate": 3.498038000802489e-05, | |
| "loss": 0.3835, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.42921348314606744, | |
| "grad_norm": 3.7303507328033447, | |
| "learning_rate": 3.497697469339215e-05, | |
| "loss": 0.8516, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.4314606741573034, | |
| "grad_norm": 2.96820330619812, | |
| "learning_rate": 3.497329726391606e-05, | |
| "loss": 0.5479, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.4337078651685393, | |
| "grad_norm": 5.242271423339844, | |
| "learning_rate": 3.496934779119175e-05, | |
| "loss": 0.8642, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.43595505617977526, | |
| "grad_norm": 2.740006685256958, | |
| "learning_rate": 3.496512635211069e-05, | |
| "loss": 0.3121, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.43820224719101125, | |
| "grad_norm": 4.162242889404297, | |
| "learning_rate": 3.496063302885921e-05, | |
| "loss": 0.6932, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4404494382022472, | |
| "grad_norm": 0.632938027381897, | |
| "learning_rate": 3.495586790891689e-05, | |
| "loss": 0.0647, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.44269662921348313, | |
| "grad_norm": 4.595058917999268, | |
| "learning_rate": 3.495083108505487e-05, | |
| "loss": 0.8173, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.4449438202247191, | |
| "grad_norm": 3.102372646331787, | |
| "learning_rate": 3.494552265533404e-05, | |
| "loss": 0.3122, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.44719101123595506, | |
| "grad_norm": 4.9895830154418945, | |
| "learning_rate": 3.493994272310313e-05, | |
| "loss": 0.7852, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.449438202247191, | |
| "grad_norm": 4.032258987426758, | |
| "learning_rate": 3.493409139699669e-05, | |
| "loss": 0.811, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.451685393258427, | |
| "grad_norm": 4.17324161529541, | |
| "learning_rate": 3.4927968790932973e-05, | |
| "loss": 0.7564, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.45393258426966293, | |
| "grad_norm": 0.49707159399986267, | |
| "learning_rate": 3.492157502411174e-05, | |
| "loss": 0.0541, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.45617977528089887, | |
| "grad_norm": 3.847059965133667, | |
| "learning_rate": 3.491491022101194e-05, | |
| "loss": 0.9085, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.4584269662921348, | |
| "grad_norm": 4.565647602081299, | |
| "learning_rate": 3.4907974511389224e-05, | |
| "loss": 0.8416, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.4606741573033708, | |
| "grad_norm": 0.8872150778770447, | |
| "learning_rate": 3.4900768030273515e-05, | |
| "loss": 0.0569, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.46292134831460674, | |
| "grad_norm": 3.2797999382019043, | |
| "learning_rate": 3.4893290917966305e-05, | |
| "loss": 0.7998, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.4651685393258427, | |
| "grad_norm": 5.683195114135742, | |
| "learning_rate": 3.4885543320037956e-05, | |
| "loss": 0.7218, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.46741573033707867, | |
| "grad_norm": 5.348382949829102, | |
| "learning_rate": 3.4877525387324844e-05, | |
| "loss": 0.9292, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.4696629213483146, | |
| "grad_norm": 4.3047099113464355, | |
| "learning_rate": 3.486923727592647e-05, | |
| "loss": 0.8279, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.47191011235955055, | |
| "grad_norm": 4.425166130065918, | |
| "learning_rate": 3.486067914720236e-05, | |
| "loss": 0.8452, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47415730337078654, | |
| "grad_norm": 5.7947916984558105, | |
| "learning_rate": 3.485185116776896e-05, | |
| "loss": 1.1099, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.4764044943820225, | |
| "grad_norm": 4.257087230682373, | |
| "learning_rate": 3.4842753509496385e-05, | |
| "loss": 0.9436, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.4786516853932584, | |
| "grad_norm": 4.357375144958496, | |
| "learning_rate": 3.483338634950507e-05, | |
| "loss": 0.8389, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.48089887640449436, | |
| "grad_norm": 3.666268825531006, | |
| "learning_rate": 3.482374987016233e-05, | |
| "loss": 0.3297, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.48314606741573035, | |
| "grad_norm": 3.0593607425689697, | |
| "learning_rate": 3.481384425907879e-05, | |
| "loss": 0.8098, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4853932584269663, | |
| "grad_norm": 0.4539957344532013, | |
| "learning_rate": 3.480366970910476e-05, | |
| "loss": 0.0386, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.48764044943820223, | |
| "grad_norm": 3.3102784156799316, | |
| "learning_rate": 3.479322641832646e-05, | |
| "loss": 0.7752, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.4898876404494382, | |
| "grad_norm": 3.8798298835754395, | |
| "learning_rate": 3.4782514590062165e-05, | |
| "loss": 0.8071, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.49213483146067416, | |
| "grad_norm": 6.300197124481201, | |
| "learning_rate": 3.4771534432858255e-05, | |
| "loss": 2.571, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.4943820224719101, | |
| "grad_norm": 4.163381099700928, | |
| "learning_rate": 3.4760286160485145e-05, | |
| "loss": 0.5912, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4966292134831461, | |
| "grad_norm": 3.5834686756134033, | |
| "learning_rate": 3.474876999193314e-05, | |
| "loss": 0.3792, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.49887640449438203, | |
| "grad_norm": 4.494593143463135, | |
| "learning_rate": 3.473698615140816e-05, | |
| "loss": 0.7456, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.501123595505618, | |
| "grad_norm": 3.909142017364502, | |
| "learning_rate": 3.4724934868327366e-05, | |
| "loss": 0.7207, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.503370786516854, | |
| "grad_norm": 3.0387282371520996, | |
| "learning_rate": 3.47126163773147e-05, | |
| "loss": 0.3254, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "grad_norm": 0.6529088616371155, | |
| "learning_rate": 3.4700030918196344e-05, | |
| "loss": 0.0461, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_VitaminC_cosine_accuracy": 0.556640625, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8303268551826477, | |
| "eval_VitaminC_cosine_ap": 0.5509523400010791, | |
| "eval_VitaminC_cosine_f1": 0.6657824933687002, | |
| "eval_VitaminC_cosine_f1_threshold": 0.2634955048561096, | |
| "eval_VitaminC_cosine_precision": 0.4990059642147117, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.55078125, | |
| "eval_VitaminC_dot_accuracy_threshold": 281.011474609375, | |
| "eval_VitaminC_dot_ap": 0.5281394234221073, | |
| "eval_VitaminC_dot_f1": 0.6711772665764546, | |
| "eval_VitaminC_dot_f1_threshold": 141.11529541015625, | |
| "eval_VitaminC_dot_precision": 0.5081967213114754, | |
| "eval_VitaminC_dot_recall": 0.9880478087649402, | |
| "eval_VitaminC_euclidean_accuracy": 0.556640625, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 11.405111312866211, | |
| "eval_VitaminC_euclidean_ap": 0.5573376843815556, | |
| "eval_VitaminC_euclidean_f1": 0.6640211640211641, | |
| "eval_VitaminC_euclidean_f1_threshold": 24.63976287841797, | |
| "eval_VitaminC_euclidean_precision": 0.497029702970297, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.556640625, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 349.33441162109375, | |
| "eval_VitaminC_manhattan_ap": 0.5561637270496671, | |
| "eval_VitaminC_manhattan_f1": 0.6657824933687002, | |
| "eval_VitaminC_manhattan_f1_threshold": 505.0340270996094, | |
| "eval_VitaminC_manhattan_precision": 0.4990059642147117, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.556640625, | |
| "eval_VitaminC_max_accuracy_threshold": 349.33441162109375, | |
| "eval_VitaminC_max_ap": 0.5573376843815556, | |
| "eval_VitaminC_max_f1": 0.6711772665764546, | |
| "eval_VitaminC_max_f1_threshold": 505.0340270996094, | |
| "eval_VitaminC_max_precision": 0.5081967213114754, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5573376843815556, | |
| "eval_sts-test_pearson_cosine": 0.8848200869109313, | |
| "eval_sts-test_pearson_dot": 0.8723563516714744, | |
| "eval_sts-test_pearson_euclidean": 0.9070688973489409, | |
| "eval_sts-test_pearson_manhattan": 0.9073961699007848, | |
| "eval_sts-test_pearson_max": 0.9073961699007848, | |
| "eval_sts-test_spearman_cosine": 0.9050875937031079, | |
| "eval_sts-test_spearman_dot": 0.8699468894518183, | |
| "eval_sts-test_spearman_euclidean": 0.9020747597811932, | |
| "eval_sts-test_spearman_manhattan": 0.9019608230696907, | |
| "eval_sts-test_spearman_max": 0.9050875937031079, | |
| "eval_vitaminc-pairs_loss": 1.4897230863571167, | |
| "eval_vitaminc-pairs_runtime": 1.8927, | |
| "eval_vitaminc-pairs_samples_per_second": 57.062, | |
| "eval_vitaminc-pairs_steps_per_second": 1.057, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_negation-triplets_loss": 0.9457363486289978, | |
| "eval_negation-triplets_runtime": 0.3019, | |
| "eval_negation-triplets_samples_per_second": 212.002, | |
| "eval_negation-triplets_steps_per_second": 3.313, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_scitail-pairs-pos_loss": 0.07606112211942673, | |
| "eval_scitail-pairs-pos_runtime": 0.3972, | |
| "eval_scitail-pairs-pos_samples_per_second": 135.938, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.517, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_scitail-pairs-qa_loss": 0.001212431932799518, | |
| "eval_scitail-pairs-qa_runtime": 0.5348, | |
| "eval_scitail-pairs-qa_samples_per_second": 239.347, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.74, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_xsum-pairs_loss": 0.02758924476802349, | |
| "eval_xsum-pairs_runtime": 2.767, | |
| "eval_xsum-pairs_samples_per_second": 46.26, | |
| "eval_xsum-pairs_steps_per_second": 0.723, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_sciq_pairs_loss": 0.016450434923171997, | |
| "eval_sciq_pairs_runtime": 2.8812, | |
| "eval_sciq_pairs_samples_per_second": 44.426, | |
| "eval_sciq_pairs_steps_per_second": 0.694, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_qasc_pairs_loss": 0.09214109182357788, | |
| "eval_qasc_pairs_runtime": 0.6597, | |
| "eval_qasc_pairs_samples_per_second": 194.029, | |
| "eval_qasc_pairs_steps_per_second": 3.032, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_openbookqa_pairs_loss": 0.7429620623588562, | |
| "eval_openbookqa_pairs_runtime": 0.5947, | |
| "eval_openbookqa_pairs_samples_per_second": 215.22, | |
| "eval_openbookqa_pairs_steps_per_second": 3.363, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_msmarco_pairs_loss": 0.17871831357479095, | |
| "eval_msmarco_pairs_runtime": 1.5003, | |
| "eval_msmarco_pairs_samples_per_second": 85.314, | |
| "eval_msmarco_pairs_steps_per_second": 1.333, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_nq_pairs_loss": 0.09803248196840286, | |
| "eval_nq_pairs_runtime": 2.3587, | |
| "eval_nq_pairs_samples_per_second": 54.267, | |
| "eval_nq_pairs_steps_per_second": 0.848, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_trivia_pairs_loss": 0.5323590636253357, | |
| "eval_trivia_pairs_runtime": 3.6206, | |
| "eval_trivia_pairs_samples_per_second": 35.354, | |
| "eval_trivia_pairs_steps_per_second": 0.552, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_gooaq_pairs_loss": 0.2667708098888397, | |
| "eval_gooaq_pairs_runtime": 0.9171, | |
| "eval_gooaq_pairs_samples_per_second": 139.573, | |
| "eval_gooaq_pairs_steps_per_second": 2.181, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5056179775280899, | |
| "eval_paws-pos_loss": 0.0236118845641613, | |
| "eval_paws-pos_runtime": 0.6973, | |
| "eval_paws-pos_samples_per_second": 183.563, | |
| "eval_paws-pos_steps_per_second": 2.868, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5078651685393258, | |
| "grad_norm": 3.5867371559143066, | |
| "learning_rate": 3.4687178735995997e-05, | |
| "loss": 0.347, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.5101123595505618, | |
| "grad_norm": 0.37994861602783203, | |
| "learning_rate": 3.467406008093016e-05, | |
| "loss": 0.0417, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.5123595505617977, | |
| "grad_norm": 4.081336975097656, | |
| "learning_rate": 3.466067520840322e-05, | |
| "loss": 0.7783, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.5146067415730337, | |
| "grad_norm": 4.306976795196533, | |
| "learning_rate": 3.46470243790025e-05, | |
| "loss": 0.9027, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.5168539325842697, | |
| "grad_norm": 4.0280022621154785, | |
| "learning_rate": 3.4633107858493206e-05, | |
| "loss": 0.7166, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5191011235955056, | |
| "grad_norm": 3.4807679653167725, | |
| "learning_rate": 3.461892591781319e-05, | |
| "loss": 0.705, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.5213483146067416, | |
| "grad_norm": 4.166563510894775, | |
| "learning_rate": 3.4604478833067756e-05, | |
| "loss": 0.8425, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.5235955056179775, | |
| "grad_norm": 3.828537940979004, | |
| "learning_rate": 3.4589766885524204e-05, | |
| "loss": 0.5362, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.5258426966292135, | |
| "grad_norm": 4.316190242767334, | |
| "learning_rate": 3.4574790361606435e-05, | |
| "loss": 0.7869, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.5280898876404494, | |
| "grad_norm": 4.244805335998535, | |
| "learning_rate": 3.4559549552889285e-05, | |
| "loss": 0.88, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5303370786516854, | |
| "grad_norm": 4.208700656890869, | |
| "learning_rate": 3.454404475609294e-05, | |
| "loss": 0.8077, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5325842696629214, | |
| "grad_norm": 3.1473183631896973, | |
| "learning_rate": 3.4528276273077094e-05, | |
| "loss": 0.8145, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.5348314606741573, | |
| "grad_norm": 3.798297166824341, | |
| "learning_rate": 3.4512244410835094e-05, | |
| "loss": 0.78, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5370786516853933, | |
| "grad_norm": 0.535529375076294, | |
| "learning_rate": 3.449594948148796e-05, | |
| "loss": 0.0536, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.5393258426966292, | |
| "grad_norm": 3.2119970321655273, | |
| "learning_rate": 3.447939180227833e-05, | |
| "loss": 0.7975, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5415730337078651, | |
| "grad_norm": 4.725860118865967, | |
| "learning_rate": 3.446257169556425e-05, | |
| "loss": 0.8932, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.5438202247191011, | |
| "grad_norm": 3.867676258087158, | |
| "learning_rate": 3.4445489488812906e-05, | |
| "loss": 0.3386, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.5460674157303371, | |
| "grad_norm": 3.981114387512207, | |
| "learning_rate": 3.4428145514594274e-05, | |
| "loss": 0.7741, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.5483146067415731, | |
| "grad_norm": 4.034990310668945, | |
| "learning_rate": 3.4410540110574616e-05, | |
| "loss": 0.7439, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.550561797752809, | |
| "grad_norm": 4.209812641143799, | |
| "learning_rate": 3.4392673619509916e-05, | |
| "loss": 0.7999, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5528089887640449, | |
| "grad_norm": 3.942631244659424, | |
| "learning_rate": 3.437454638923921e-05, | |
| "loss": 0.8542, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.5550561797752809, | |
| "grad_norm": 4.087955951690674, | |
| "learning_rate": 3.435615877267783e-05, | |
| "loss": 0.6992, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.5573033707865168, | |
| "grad_norm": 3.885822057723999, | |
| "learning_rate": 3.4337511127810466e-05, | |
| "loss": 0.8579, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.5595505617977528, | |
| "grad_norm": 5.198770523071289, | |
| "learning_rate": 3.431860381768431e-05, | |
| "loss": 1.0221, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.5617977528089888, | |
| "grad_norm": 4.321418285369873, | |
| "learning_rate": 3.4299437210401866e-05, | |
| "loss": 0.699, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5640449438202247, | |
| "grad_norm": 3.1992154121398926, | |
| "learning_rate": 3.4280011679113884e-05, | |
| "loss": 0.8523, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.5662921348314607, | |
| "grad_norm": 4.94226598739624, | |
| "learning_rate": 3.4260327602012027e-05, | |
| "loss": 1.0307, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.5685393258426966, | |
| "grad_norm": 3.958935499191284, | |
| "learning_rate": 3.424038536232154e-05, | |
| "loss": 0.846, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.5707865168539326, | |
| "grad_norm": 4.023487091064453, | |
| "learning_rate": 3.4220185348293775e-05, | |
| "loss": 0.8361, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.5730337078651685, | |
| "grad_norm": 3.275102138519287, | |
| "learning_rate": 3.4199727953198665e-05, | |
| "loss": 0.8224, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5752808988764045, | |
| "grad_norm": 3.6130261421203613, | |
| "learning_rate": 3.417901357531701e-05, | |
| "loss": 0.5301, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.5775280898876405, | |
| "grad_norm": 4.571770668029785, | |
| "learning_rate": 3.415804261793277e-05, | |
| "loss": 0.3795, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.5797752808988764, | |
| "grad_norm": 3.1884663105010986, | |
| "learning_rate": 3.413681548932521e-05, | |
| "loss": 0.5434, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.5820224719101124, | |
| "grad_norm": 4.795211315155029, | |
| "learning_rate": 3.411533260276091e-05, | |
| "loss": 0.847, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.5842696629213483, | |
| "grad_norm": 4.761318206787109, | |
| "learning_rate": 3.409359437648579e-05, | |
| "loss": 0.7323, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5865168539325842, | |
| "grad_norm": 4.4683098793029785, | |
| "learning_rate": 3.407160123371687e-05, | |
| "loss": 0.6606, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.5887640449438202, | |
| "grad_norm": 0.7677178382873535, | |
| "learning_rate": 3.404935360263415e-05, | |
| "loss": 0.0543, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5910112359550562, | |
| "grad_norm": 4.110381126403809, | |
| "learning_rate": 3.4026851916372166e-05, | |
| "loss": 0.6709, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5932584269662922, | |
| "grad_norm": 4.766375541687012, | |
| "learning_rate": 3.400409661301162e-05, | |
| "loss": 0.809, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5955056179775281, | |
| "grad_norm": 5.389264106750488, | |
| "learning_rate": 3.398108813557082e-05, | |
| "loss": 1.0391, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.597752808988764, | |
| "grad_norm": 3.8780810832977295, | |
| "learning_rate": 3.3957826931997094e-05, | |
| "loss": 0.7396, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 4.399974822998047, | |
| "learning_rate": 3.393431345515801e-05, | |
| "loss": 0.7839, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.6022471910112359, | |
| "grad_norm": 3.2098612785339355, | |
| "learning_rate": 3.391054816283262e-05, | |
| "loss": 0.3054, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.604494382022472, | |
| "grad_norm": 3.606182098388672, | |
| "learning_rate": 3.3886531517702505e-05, | |
| "loss": 0.5258, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "grad_norm": 4.3564934730529785, | |
| "learning_rate": 3.3862263987342784e-05, | |
| "loss": 0.7367, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_VitaminC_cosine_accuracy": 0.552734375, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.814909815788269, | |
| "eval_VitaminC_cosine_ap": 0.5506214433093293, | |
| "eval_VitaminC_cosine_f1": 0.664886515353805, | |
| "eval_VitaminC_cosine_f1_threshold": 0.3506072461605072, | |
| "eval_VitaminC_cosine_precision": 0.5, | |
| "eval_VitaminC_cosine_recall": 0.9920318725099602, | |
| "eval_VitaminC_dot_accuracy": 0.55078125, | |
| "eval_VitaminC_dot_accuracy_threshold": 316.90899658203125, | |
| "eval_VitaminC_dot_ap": 0.5353657977329522, | |
| "eval_VitaminC_dot_f1": 0.6666666666666667, | |
| "eval_VitaminC_dot_f1_threshold": 155.67796325683594, | |
| "eval_VitaminC_dot_precision": 0.506198347107438, | |
| "eval_VitaminC_dot_recall": 0.9760956175298805, | |
| "eval_VitaminC_euclidean_accuracy": 0.55078125, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 10.77621841430664, | |
| "eval_VitaminC_euclidean_ap": 0.550546292530568, | |
| "eval_VitaminC_euclidean_f1": 0.6666666666666666, | |
| "eval_VitaminC_euclidean_f1_threshold": 24.22284698486328, | |
| "eval_VitaminC_euclidean_precision": 0.5, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.5546875, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 335.6986389160156, | |
| "eval_VitaminC_manhattan_ap": 0.5497325043939846, | |
| "eval_VitaminC_manhattan_f1": 0.6640211640211641, | |
| "eval_VitaminC_manhattan_f1_threshold": 513.494873046875, | |
| "eval_VitaminC_manhattan_precision": 0.497029702970297, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.5546875, | |
| "eval_VitaminC_max_accuracy_threshold": 335.6986389160156, | |
| "eval_VitaminC_max_ap": 0.5506214433093293, | |
| "eval_VitaminC_max_f1": 0.6666666666666667, | |
| "eval_VitaminC_max_f1_threshold": 513.494873046875, | |
| "eval_VitaminC_max_precision": 0.506198347107438, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5506214433093293, | |
| "eval_sts-test_pearson_cosine": 0.8848372816940555, | |
| "eval_sts-test_pearson_dot": 0.8774995772730847, | |
| "eval_sts-test_pearson_euclidean": 0.9058906663416005, | |
| "eval_sts-test_pearson_manhattan": 0.9066316554236529, | |
| "eval_sts-test_pearson_max": 0.9066316554236529, | |
| "eval_sts-test_spearman_cosine": 0.9085018016884417, | |
| "eval_sts-test_spearman_dot": 0.8776881864036095, | |
| "eval_sts-test_spearman_euclidean": 0.903223569412372, | |
| "eval_sts-test_spearman_manhattan": 0.9037578547221237, | |
| "eval_sts-test_spearman_max": 0.9085018016884417, | |
| "eval_vitaminc-pairs_loss": 1.4935871362686157, | |
| "eval_vitaminc-pairs_runtime": 1.8963, | |
| "eval_vitaminc-pairs_samples_per_second": 56.952, | |
| "eval_vitaminc-pairs_steps_per_second": 1.055, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_negation-triplets_loss": 0.9505463242530823, | |
| "eval_negation-triplets_runtime": 0.3041, | |
| "eval_negation-triplets_samples_per_second": 210.485, | |
| "eval_negation-triplets_steps_per_second": 3.289, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_scitail-pairs-pos_loss": 0.09635873883962631, | |
| "eval_scitail-pairs-pos_runtime": 0.4048, | |
| "eval_scitail-pairs-pos_samples_per_second": 133.396, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.47, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_scitail-pairs-qa_loss": 0.0009468490607105196, | |
| "eval_scitail-pairs-qa_runtime": 0.5341, | |
| "eval_scitail-pairs-qa_samples_per_second": 239.65, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.745, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_xsum-pairs_loss": 0.026903513818979263, | |
| "eval_xsum-pairs_runtime": 2.7518, | |
| "eval_xsum-pairs_samples_per_second": 46.514, | |
| "eval_xsum-pairs_steps_per_second": 0.727, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_sciq_pairs_loss": 0.01619444414973259, | |
| "eval_sciq_pairs_runtime": 2.8856, | |
| "eval_sciq_pairs_samples_per_second": 44.358, | |
| "eval_sciq_pairs_steps_per_second": 0.693, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_qasc_pairs_loss": 0.09130185097455978, | |
| "eval_qasc_pairs_runtime": 0.6645, | |
| "eval_qasc_pairs_samples_per_second": 192.631, | |
| "eval_qasc_pairs_steps_per_second": 3.01, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_openbookqa_pairs_loss": 0.7336423397064209, | |
| "eval_openbookqa_pairs_runtime": 0.5935, | |
| "eval_openbookqa_pairs_samples_per_second": 215.687, | |
| "eval_openbookqa_pairs_steps_per_second": 3.37, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_msmarco_pairs_loss": 0.15868164598941803, | |
| "eval_msmarco_pairs_runtime": 1.5086, | |
| "eval_msmarco_pairs_samples_per_second": 84.844, | |
| "eval_msmarco_pairs_steps_per_second": 1.326, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_nq_pairs_loss": 0.10780799388885498, | |
| "eval_nq_pairs_runtime": 2.3746, | |
| "eval_nq_pairs_samples_per_second": 53.905, | |
| "eval_nq_pairs_steps_per_second": 0.842, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_trivia_pairs_loss": 0.49691149592399597, | |
| "eval_trivia_pairs_runtime": 3.5992, | |
| "eval_trivia_pairs_samples_per_second": 35.563, | |
| "eval_trivia_pairs_steps_per_second": 0.556, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_gooaq_pairs_loss": 0.3025541603565216, | |
| "eval_gooaq_pairs_runtime": 0.9181, | |
| "eval_gooaq_pairs_samples_per_second": 139.423, | |
| "eval_gooaq_pairs_steps_per_second": 2.178, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6067415730337079, | |
| "eval_paws-pos_loss": 0.024440350010991096, | |
| "eval_paws-pos_runtime": 0.7046, | |
| "eval_paws-pos_samples_per_second": 181.67, | |
| "eval_paws-pos_steps_per_second": 2.839, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6089887640449438, | |
| "grad_norm": 3.21183705329895, | |
| "learning_rate": 3.383774604421301e-05, | |
| "loss": 0.747, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.6112359550561798, | |
| "grad_norm": 4.403411865234375, | |
| "learning_rate": 3.3812978165647975e-05, | |
| "loss": 0.7855, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.6134831460674157, | |
| "grad_norm": 0.46612274646759033, | |
| "learning_rate": 3.3787960833848405e-05, | |
| "loss": 0.0473, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.6157303370786517, | |
| "grad_norm": 3.30610990524292, | |
| "learning_rate": 3.3762694535871584e-05, | |
| "loss": 0.4378, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.6179775280898876, | |
| "grad_norm": 3.7408640384674072, | |
| "learning_rate": 3.373717976362187e-05, | |
| "loss": 0.8767, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6202247191011236, | |
| "grad_norm": 5.345012187957764, | |
| "learning_rate": 3.3711417013841105e-05, | |
| "loss": 1.0345, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.6224719101123596, | |
| "grad_norm": 3.518765449523926, | |
| "learning_rate": 3.368540678809897e-05, | |
| "loss": 0.5182, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.6247191011235955, | |
| "grad_norm": 6.666887283325195, | |
| "learning_rate": 3.3659149592783186e-05, | |
| "loss": 2.5949, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.6269662921348315, | |
| "grad_norm": 3.197411298751831, | |
| "learning_rate": 3.363264593908969e-05, | |
| "loss": 0.833, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.6292134831460674, | |
| "grad_norm": 0.6012090444564819, | |
| "learning_rate": 3.360589634301267e-05, | |
| "loss": 0.0778, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6314606741573033, | |
| "grad_norm": 4.5016188621521, | |
| "learning_rate": 3.357890132533449e-05, | |
| "loss": 0.8048, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.6337078651685393, | |
| "grad_norm": 3.865889072418213, | |
| "learning_rate": 3.35516614116156e-05, | |
| "loss": 0.7524, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.6359550561797753, | |
| "grad_norm": 3.2998361587524414, | |
| "learning_rate": 3.3524177132184266e-05, | |
| "loss": 0.3246, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.6382022471910113, | |
| "grad_norm": 0.6418587565422058, | |
| "learning_rate": 3.349644902212628e-05, | |
| "loss": 0.0728, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.6404494382022472, | |
| "grad_norm": 5.772351264953613, | |
| "learning_rate": 3.34684776212745e-05, | |
| "loss": 2.3619, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6426966292134831, | |
| "grad_norm": 3.769488573074341, | |
| "learning_rate": 3.3440263474198376e-05, | |
| "loss": 0.7464, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.6449438202247191, | |
| "grad_norm": 4.559601783752441, | |
| "learning_rate": 3.3411807130193325e-05, | |
| "loss": 0.6691, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.647191011235955, | |
| "grad_norm": 0.45337462425231934, | |
| "learning_rate": 3.338310914327005e-05, | |
| "loss": 0.059, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.6494382022471911, | |
| "grad_norm": 4.7184553146362305, | |
| "learning_rate": 3.3354170072143766e-05, | |
| "loss": 0.7841, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.651685393258427, | |
| "grad_norm": 3.886216640472412, | |
| "learning_rate": 3.332499048022328e-05, | |
| "loss": 0.647, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6539325842696629, | |
| "grad_norm": 4.497567176818848, | |
| "learning_rate": 3.329557093560006e-05, | |
| "loss": 0.8814, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.6561797752808989, | |
| "grad_norm": 3.995391368865967, | |
| "learning_rate": 3.326591201103716e-05, | |
| "loss": 0.7247, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.6584269662921348, | |
| "grad_norm": 0.4348815083503723, | |
| "learning_rate": 3.323601428395809e-05, | |
| "loss": 0.059, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.6606741573033708, | |
| "grad_norm": 3.6197896003723145, | |
| "learning_rate": 3.320587833643554e-05, | |
| "loss": 0.8317, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.6629213483146067, | |
| "grad_norm": 4.4088215827941895, | |
| "learning_rate": 3.317550475518006e-05, | |
| "loss": 0.8548, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6651685393258427, | |
| "grad_norm": 4.541014194488525, | |
| "learning_rate": 3.314489413152867e-05, | |
| "loss": 0.9213, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.6674157303370787, | |
| "grad_norm": 3.067857265472412, | |
| "learning_rate": 3.311404706143329e-05, | |
| "loss": 0.6923, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.6696629213483146, | |
| "grad_norm": 4.037753582000732, | |
| "learning_rate": 3.3082964145449174e-05, | |
| "loss": 0.7777, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.6719101123595506, | |
| "grad_norm": 4.280182838439941, | |
| "learning_rate": 3.305164598872322e-05, | |
| "loss": 0.7496, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.6741573033707865, | |
| "grad_norm": 4.357325077056885, | |
| "learning_rate": 3.302009320098218e-05, | |
| "loss": 0.7636, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6764044943820224, | |
| "grad_norm": 4.007940292358398, | |
| "learning_rate": 3.2988306396520775e-05, | |
| "loss": 0.6867, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.6786516853932584, | |
| "grad_norm": 0.8544747233390808, | |
| "learning_rate": 3.295628619418977e-05, | |
| "loss": 0.0506, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.6808988764044944, | |
| "grad_norm": 3.34498929977417, | |
| "learning_rate": 3.292403321738387e-05, | |
| "loss": 0.3346, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.6831460674157304, | |
| "grad_norm": 2.441420316696167, | |
| "learning_rate": 3.289154809402967e-05, | |
| "loss": 0.2485, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.6853932584269663, | |
| "grad_norm": 4.533839702606201, | |
| "learning_rate": 3.285883145657334e-05, | |
| "loss": 0.8508, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6876404494382022, | |
| "grad_norm": 3.2033944129943848, | |
| "learning_rate": 3.2825883941968346e-05, | |
| "loss": 0.8464, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.6898876404494382, | |
| "grad_norm": 3.6305220127105713, | |
| "learning_rate": 3.279270619166309e-05, | |
| "loss": 0.3385, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.6921348314606741, | |
| "grad_norm": 4.438405990600586, | |
| "learning_rate": 3.2759298851588336e-05, | |
| "loss": 0.8837, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.6943820224719102, | |
| "grad_norm": 4.252586841583252, | |
| "learning_rate": 3.272566257214474e-05, | |
| "loss": 0.9019, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.6966292134831461, | |
| "grad_norm": 4.231752872467041, | |
| "learning_rate": 3.2691798008190096e-05, | |
| "loss": 0.6922, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.698876404494382, | |
| "grad_norm": 3.862682342529297, | |
| "learning_rate": 3.265770581902662e-05, | |
| "loss": 0.6348, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.701123595505618, | |
| "grad_norm": 3.783026933670044, | |
| "learning_rate": 3.262338666838813e-05, | |
| "loss": 0.7522, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.7033707865168539, | |
| "grad_norm": 4.141933917999268, | |
| "learning_rate": 3.25888412244271e-05, | |
| "loss": 0.7843, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.7056179775280899, | |
| "grad_norm": 0.7638006210327148, | |
| "learning_rate": 3.2554070159701684e-05, | |
| "loss": 0.0493, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "grad_norm": 3.7285079956054688, | |
| "learning_rate": 3.2519074151162564e-05, | |
| "loss": 0.357, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_VitaminC_cosine_accuracy": 0.556640625, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.8318675756454468, | |
| "eval_VitaminC_cosine_ap": 0.553255462027648, | |
| "eval_VitaminC_cosine_f1": 0.6666666666666666, | |
| "eval_VitaminC_cosine_f1_threshold": 0.3080925941467285, | |
| "eval_VitaminC_cosine_precision": 0.5, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.5546875, | |
| "eval_VitaminC_dot_accuracy_threshold": 284.4936218261719, | |
| "eval_VitaminC_dot_ap": 0.5335304755231123, | |
| "eval_VitaminC_dot_f1": 0.6675531914893617, | |
| "eval_VitaminC_dot_f1_threshold": 117.11366271972656, | |
| "eval_VitaminC_dot_precision": 0.500998003992016, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.556640625, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 14.916669845581055, | |
| "eval_VitaminC_euclidean_ap": 0.5560392780320775, | |
| "eval_VitaminC_euclidean_f1": 0.6657824933687002, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.758323669433594, | |
| "eval_VitaminC_euclidean_precision": 0.4990059642147117, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.5546875, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 303.769775390625, | |
| "eval_VitaminC_manhattan_ap": 0.5575735035337728, | |
| "eval_VitaminC_manhattan_f1": 0.6666666666666666, | |
| "eval_VitaminC_manhattan_f1_threshold": 500.6726989746094, | |
| "eval_VitaminC_manhattan_precision": 0.5, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.556640625, | |
| "eval_VitaminC_max_accuracy_threshold": 303.769775390625, | |
| "eval_VitaminC_max_ap": 0.5575735035337728, | |
| "eval_VitaminC_max_f1": 0.6675531914893617, | |
| "eval_VitaminC_max_f1_threshold": 500.6726989746094, | |
| "eval_VitaminC_max_precision": 0.500998003992016, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5575735035337728, | |
| "eval_sts-test_pearson_cosine": 0.884017793393225, | |
| "eval_sts-test_pearson_dot": 0.8725802033594147, | |
| "eval_sts-test_pearson_euclidean": 0.9065592531799239, | |
| "eval_sts-test_pearson_manhattan": 0.9070236641674441, | |
| "eval_sts-test_pearson_max": 0.9070236641674441, | |
| "eval_sts-test_spearman_cosine": 0.9067846957888538, | |
| "eval_sts-test_spearman_dot": 0.8716365180769119, | |
| "eval_sts-test_spearman_euclidean": 0.9026938039800204, | |
| "eval_sts-test_spearman_manhattan": 0.903306941012344, | |
| "eval_sts-test_spearman_max": 0.9067846957888538, | |
| "eval_vitaminc-pairs_loss": 1.4885247945785522, | |
| "eval_vitaminc-pairs_runtime": 1.9137, | |
| "eval_vitaminc-pairs_samples_per_second": 56.436, | |
| "eval_vitaminc-pairs_steps_per_second": 1.045, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_negation-triplets_loss": 0.9597576856613159, | |
| "eval_negation-triplets_runtime": 0.3023, | |
| "eval_negation-triplets_samples_per_second": 211.742, | |
| "eval_negation-triplets_steps_per_second": 3.308, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_scitail-pairs-pos_loss": 0.09951130300760269, | |
| "eval_scitail-pairs-pos_runtime": 0.3896, | |
| "eval_scitail-pairs-pos_samples_per_second": 138.608, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.567, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_scitail-pairs-qa_loss": 0.0010157548822462559, | |
| "eval_scitail-pairs-qa_runtime": 0.5373, | |
| "eval_scitail-pairs-qa_samples_per_second": 238.245, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.723, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_xsum-pairs_loss": 0.027823584154248238, | |
| "eval_xsum-pairs_runtime": 2.7408, | |
| "eval_xsum-pairs_samples_per_second": 46.701, | |
| "eval_xsum-pairs_steps_per_second": 0.73, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_sciq_pairs_loss": 0.015241424553096294, | |
| "eval_sciq_pairs_runtime": 2.8458, | |
| "eval_sciq_pairs_samples_per_second": 44.978, | |
| "eval_sciq_pairs_steps_per_second": 0.703, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_qasc_pairs_loss": 0.09173130989074707, | |
| "eval_qasc_pairs_runtime": 0.6608, | |
| "eval_qasc_pairs_samples_per_second": 193.694, | |
| "eval_qasc_pairs_steps_per_second": 3.026, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_openbookqa_pairs_loss": 0.6921954154968262, | |
| "eval_openbookqa_pairs_runtime": 0.5893, | |
| "eval_openbookqa_pairs_samples_per_second": 217.196, | |
| "eval_openbookqa_pairs_steps_per_second": 3.394, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_msmarco_pairs_loss": 0.15177518129348755, | |
| "eval_msmarco_pairs_runtime": 1.494, | |
| "eval_msmarco_pairs_samples_per_second": 85.673, | |
| "eval_msmarco_pairs_steps_per_second": 1.339, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_nq_pairs_loss": 0.10136909037828445, | |
| "eval_nq_pairs_runtime": 2.3524, | |
| "eval_nq_pairs_samples_per_second": 54.413, | |
| "eval_nq_pairs_steps_per_second": 0.85, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_trivia_pairs_loss": 0.5301617980003357, | |
| "eval_trivia_pairs_runtime": 3.5809, | |
| "eval_trivia_pairs_samples_per_second": 35.745, | |
| "eval_trivia_pairs_steps_per_second": 0.559, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_gooaq_pairs_loss": 0.28424739837646484, | |
| "eval_gooaq_pairs_runtime": 0.9167, | |
| "eval_gooaq_pairs_samples_per_second": 139.635, | |
| "eval_gooaq_pairs_steps_per_second": 2.182, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7078651685393258, | |
| "eval_paws-pos_loss": 0.023981213569641113, | |
| "eval_paws-pos_runtime": 0.6966, | |
| "eval_paws-pos_samples_per_second": 183.744, | |
| "eval_paws-pos_steps_per_second": 2.871, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7101123595505618, | |
| "grad_norm": 3.6374969482421875, | |
| "learning_rate": 3.248385388013984e-05, | |
| "loss": 0.841, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.7123595505617978, | |
| "grad_norm": 4.251607418060303, | |
| "learning_rate": 3.2448410032329716e-05, | |
| "loss": 0.5849, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.7146067415730337, | |
| "grad_norm": 4.323038101196289, | |
| "learning_rate": 3.241274329778117e-05, | |
| "loss": 0.6818, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.7168539325842697, | |
| "grad_norm": 4.027289867401123, | |
| "learning_rate": 3.237685437088251e-05, | |
| "loss": 0.8269, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.7191011235955056, | |
| "grad_norm": 3.014479875564575, | |
| "learning_rate": 3.234074395034787e-05, | |
| "loss": 0.6979, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7213483146067415, | |
| "grad_norm": 3.5980277061462402, | |
| "learning_rate": 3.2304412739203595e-05, | |
| "loss": 0.3218, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.7235955056179775, | |
| "grad_norm": 3.2924134731292725, | |
| "learning_rate": 3.226786144477456e-05, | |
| "loss": 0.8206, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.7258426966292135, | |
| "grad_norm": 2.524231195449829, | |
| "learning_rate": 3.2231090778670385e-05, | |
| "loss": 0.2106, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.7280898876404495, | |
| "grad_norm": 5.464061260223389, | |
| "learning_rate": 3.2194101456771604e-05, | |
| "loss": 1.0524, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.7303370786516854, | |
| "grad_norm": 3.4692578315734863, | |
| "learning_rate": 3.215689419921572e-05, | |
| "loss": 0.3774, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7325842696629213, | |
| "grad_norm": 4.947183132171631, | |
| "learning_rate": 3.211946973038315e-05, | |
| "loss": 0.9098, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.7348314606741573, | |
| "grad_norm": 4.432866096496582, | |
| "learning_rate": 3.208182877888319e-05, | |
| "loss": 0.7988, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.7370786516853932, | |
| "grad_norm": 4.585951328277588, | |
| "learning_rate": 3.204397207753978e-05, | |
| "loss": 0.7916, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.7393258426966293, | |
| "grad_norm": 3.7288637161254883, | |
| "learning_rate": 3.200590036337724e-05, | |
| "loss": 0.6314, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.7415730337078652, | |
| "grad_norm": 3.840074300765991, | |
| "learning_rate": 3.196761437760593e-05, | |
| "loss": 0.8628, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7438202247191011, | |
| "grad_norm": 0.6423048377037048, | |
| "learning_rate": 3.192911486560784e-05, | |
| "loss": 0.0688, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.7460674157303371, | |
| "grad_norm": 4.148509502410889, | |
| "learning_rate": 3.1890402576922036e-05, | |
| "loss": 0.7386, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.748314606741573, | |
| "grad_norm": 4.7345147132873535, | |
| "learning_rate": 3.1851478265230103e-05, | |
| "loss": 0.8458, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.750561797752809, | |
| "grad_norm": 0.695708155632019, | |
| "learning_rate": 3.181234268834144e-05, | |
| "loss": 0.0442, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.7528089887640449, | |
| "grad_norm": 3.434741735458374, | |
| "learning_rate": 3.177299660817856e-05, | |
| "loss": 0.317, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.755056179775281, | |
| "grad_norm": 3.306964874267578, | |
| "learning_rate": 3.1733440790762176e-05, | |
| "loss": 0.8087, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.7573033707865169, | |
| "grad_norm": 3.010828733444214, | |
| "learning_rate": 3.169367600619637e-05, | |
| "loss": 0.3398, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.7595505617977528, | |
| "grad_norm": 4.152151584625244, | |
| "learning_rate": 3.1653703028653545e-05, | |
| "loss": 0.699, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.7617977528089888, | |
| "grad_norm": 4.073326110839844, | |
| "learning_rate": 3.161352263635937e-05, | |
| "loss": 0.7901, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.7640449438202247, | |
| "grad_norm": 4.365633487701416, | |
| "learning_rate": 3.157313561157764e-05, | |
| "loss": 0.8072, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7662921348314606, | |
| "grad_norm": 3.506556272506714, | |
| "learning_rate": 3.153254274059501e-05, | |
| "loss": 0.5939, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.7685393258426966, | |
| "grad_norm": 4.319092273712158, | |
| "learning_rate": 3.149174481370575e-05, | |
| "loss": 0.6933, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.7707865168539326, | |
| "grad_norm": 0.6184964179992676, | |
| "learning_rate": 3.145074262519629e-05, | |
| "loss": 0.0437, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.7730337078651686, | |
| "grad_norm": 4.866581916809082, | |
| "learning_rate": 3.140953697332979e-05, | |
| "loss": 0.9882, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.7752808988764045, | |
| "grad_norm": 3.9585559368133545, | |
| "learning_rate": 3.136812866033063e-05, | |
| "loss": 0.3707, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7775280898876404, | |
| "grad_norm": 4.253391265869141, | |
| "learning_rate": 3.132651849236871e-05, | |
| "loss": 0.7103, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.7797752808988764, | |
| "grad_norm": 0.5847011208534241, | |
| "learning_rate": 3.128470727954383e-05, | |
| "loss": 0.0372, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.7820224719101123, | |
| "grad_norm": 0.5127836465835571, | |
| "learning_rate": 3.124269583586989e-05, | |
| "loss": 0.028, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.7842696629213484, | |
| "grad_norm": 4.145182132720947, | |
| "learning_rate": 3.120048497925904e-05, | |
| "loss": 0.7676, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.7865168539325843, | |
| "grad_norm": 4.833105087280273, | |
| "learning_rate": 3.1158075531505755e-05, | |
| "loss": 0.6754, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7887640449438202, | |
| "grad_norm": 0.49345946311950684, | |
| "learning_rate": 3.1115468318270844e-05, | |
| "loss": 0.0439, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.7910112359550562, | |
| "grad_norm": 3.357720375061035, | |
| "learning_rate": 3.107266416906538e-05, | |
| "loss": 0.8039, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.7932584269662921, | |
| "grad_norm": 0.2371903359889984, | |
| "learning_rate": 3.1029663917234514e-05, | |
| "loss": 0.0104, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.7955056179775281, | |
| "grad_norm": 0.48881796002388, | |
| "learning_rate": 3.098646839994132e-05, | |
| "loss": 0.0555, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.797752808988764, | |
| "grad_norm": 3.3021090030670166, | |
| "learning_rate": 3.094307845815042e-05, | |
| "loss": 0.8646, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 3.0412533283233643, | |
| "learning_rate": 3.0899494936611663e-05, | |
| "loss": 0.7781, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.802247191011236, | |
| "grad_norm": 0.30917835235595703, | |
| "learning_rate": 3.085571868384366e-05, | |
| "loss": 0.011, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.8044943820224719, | |
| "grad_norm": 3.6957950592041016, | |
| "learning_rate": 3.081175055211726e-05, | |
| "loss": 0.3267, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.8067415730337079, | |
| "grad_norm": 7.202300071716309, | |
| "learning_rate": 3.0767591397438974e-05, | |
| "loss": 2.5281, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "grad_norm": 2.9833834171295166, | |
| "learning_rate": 3.072324207953429e-05, | |
| "loss": 0.301, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_VitaminC_cosine_accuracy": 0.55859375, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.6793336868286133, | |
| "eval_VitaminC_cosine_ap": 0.5555632752592039, | |
| "eval_VitaminC_cosine_f1": 0.6657824933687002, | |
| "eval_VitaminC_cosine_f1_threshold": 0.28029173612594604, | |
| "eval_VitaminC_cosine_precision": 0.4990059642147117, | |
| "eval_VitaminC_cosine_recall": 1.0, | |
| "eval_VitaminC_dot_accuracy": 0.55078125, | |
| "eval_VitaminC_dot_accuracy_threshold": 265.5102844238281, | |
| "eval_VitaminC_dot_ap": 0.5326105108889087, | |
| "eval_VitaminC_dot_f1": 0.6675531914893617, | |
| "eval_VitaminC_dot_f1_threshold": 106.37774658203125, | |
| "eval_VitaminC_dot_precision": 0.500998003992016, | |
| "eval_VitaminC_dot_recall": 1.0, | |
| "eval_VitaminC_euclidean_accuracy": 0.55859375, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 15.296594619750977, | |
| "eval_VitaminC_euclidean_ap": 0.5592294311948881, | |
| "eval_VitaminC_euclidean_f1": 0.6657824933687002, | |
| "eval_VitaminC_euclidean_f1_threshold": 23.58568572998047, | |
| "eval_VitaminC_euclidean_precision": 0.4990059642147117, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.556640625, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 306.79913330078125, | |
| "eval_VitaminC_manhattan_ap": 0.5598941655081213, | |
| "eval_VitaminC_manhattan_f1": 0.6649006622516557, | |
| "eval_VitaminC_manhattan_f1_threshold": 512.0101318359375, | |
| "eval_VitaminC_manhattan_precision": 0.498015873015873, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.55859375, | |
| "eval_VitaminC_max_accuracy_threshold": 306.79913330078125, | |
| "eval_VitaminC_max_ap": 0.5598941655081213, | |
| "eval_VitaminC_max_f1": 0.6675531914893617, | |
| "eval_VitaminC_max_f1_threshold": 512.0101318359375, | |
| "eval_VitaminC_max_precision": 0.500998003992016, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5598941655081213, | |
| "eval_sts-test_pearson_cosine": 0.8832151520369376, | |
| "eval_sts-test_pearson_dot": 0.8763916954110884, | |
| "eval_sts-test_pearson_euclidean": 0.9046869354209082, | |
| "eval_sts-test_pearson_manhattan": 0.9047119917370259, | |
| "eval_sts-test_pearson_max": 0.9047119917370259, | |
| "eval_sts-test_spearman_cosine": 0.9054341922225841, | |
| "eval_sts-test_spearman_dot": 0.8786041104705073, | |
| "eval_sts-test_spearman_euclidean": 0.9002407635868509, | |
| "eval_sts-test_spearman_manhattan": 0.9006719867416183, | |
| "eval_sts-test_spearman_max": 0.9054341922225841, | |
| "eval_vitaminc-pairs_loss": 1.4290639162063599, | |
| "eval_vitaminc-pairs_runtime": 1.8905, | |
| "eval_vitaminc-pairs_samples_per_second": 57.128, | |
| "eval_vitaminc-pairs_steps_per_second": 1.058, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_negation-triplets_loss": 0.9030703902244568, | |
| "eval_negation-triplets_runtime": 0.2986, | |
| "eval_negation-triplets_samples_per_second": 214.299, | |
| "eval_negation-triplets_steps_per_second": 3.348, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_scitail-pairs-pos_loss": 0.10728535801172256, | |
| "eval_scitail-pairs-pos_runtime": 0.3831, | |
| "eval_scitail-pairs-pos_samples_per_second": 140.965, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.61, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_scitail-pairs-qa_loss": 0.0005650219391100109, | |
| "eval_scitail-pairs-qa_runtime": 0.5259, | |
| "eval_scitail-pairs-qa_samples_per_second": 243.397, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.803, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_xsum-pairs_loss": 0.025990577414631844, | |
| "eval_xsum-pairs_runtime": 2.734, | |
| "eval_xsum-pairs_samples_per_second": 46.818, | |
| "eval_xsum-pairs_steps_per_second": 0.732, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_sciq_pairs_loss": 0.016017427667975426, | |
| "eval_sciq_pairs_runtime": 2.8252, | |
| "eval_sciq_pairs_samples_per_second": 45.307, | |
| "eval_sciq_pairs_steps_per_second": 0.708, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_qasc_pairs_loss": 0.10250324755907059, | |
| "eval_qasc_pairs_runtime": 0.6511, | |
| "eval_qasc_pairs_samples_per_second": 196.585, | |
| "eval_qasc_pairs_steps_per_second": 3.072, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_openbookqa_pairs_loss": 0.6710968613624573, | |
| "eval_openbookqa_pairs_runtime": 0.5776, | |
| "eval_openbookqa_pairs_samples_per_second": 221.625, | |
| "eval_openbookqa_pairs_steps_per_second": 3.463, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_msmarco_pairs_loss": 0.14522777497768402, | |
| "eval_msmarco_pairs_runtime": 1.4981, | |
| "eval_msmarco_pairs_samples_per_second": 85.441, | |
| "eval_msmarco_pairs_steps_per_second": 1.335, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_nq_pairs_loss": 0.10225611180067062, | |
| "eval_nq_pairs_runtime": 2.3595, | |
| "eval_nq_pairs_samples_per_second": 54.248, | |
| "eval_nq_pairs_steps_per_second": 0.848, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_trivia_pairs_loss": 0.5312957167625427, | |
| "eval_trivia_pairs_runtime": 3.5813, | |
| "eval_trivia_pairs_samples_per_second": 35.741, | |
| "eval_trivia_pairs_steps_per_second": 0.558, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_gooaq_pairs_loss": 0.27713337540626526, | |
| "eval_gooaq_pairs_runtime": 0.9166, | |
| "eval_gooaq_pairs_samples_per_second": 139.645, | |
| "eval_gooaq_pairs_steps_per_second": 2.182, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8089887640449438, | |
| "eval_paws-pos_loss": 0.024326296523213387, | |
| "eval_paws-pos_runtime": 0.6893, | |
| "eval_paws-pos_samples_per_second": 185.682, | |
| "eval_paws-pos_steps_per_second": 2.901, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8112359550561797, | |
| "grad_norm": 4.372533798217773, | |
| "learning_rate": 3.067870346183096e-05, | |
| "loss": 0.7533, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.8134831460674158, | |
| "grad_norm": 2.6585452556610107, | |
| "learning_rate": 3.063397641144216e-05, | |
| "loss": 0.2958, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.8157303370786517, | |
| "grad_norm": 4.378647327423096, | |
| "learning_rate": 3.058906179914962e-05, | |
| "loss": 0.8296, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.8179775280898877, | |
| "grad_norm": 3.1601309776306152, | |
| "learning_rate": 3.0543960499386694e-05, | |
| "loss": 0.3191, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.8202247191011236, | |
| "grad_norm": 3.446498394012451, | |
| "learning_rate": 3.049867339022129e-05, | |
| "loss": 0.7866, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.8224719101123595, | |
| "grad_norm": 3.0058486461639404, | |
| "learning_rate": 3.0453201353338826e-05, | |
| "loss": 0.3157, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.8247191011235955, | |
| "grad_norm": 4.380611419677734, | |
| "learning_rate": 3.040754527402502e-05, | |
| "loss": 0.7402, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.8269662921348314, | |
| "grad_norm": 3.8081209659576416, | |
| "learning_rate": 3.036170604114869e-05, | |
| "loss": 0.4957, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.8292134831460675, | |
| "grad_norm": 4.2056989669799805, | |
| "learning_rate": 3.031568454714442e-05, | |
| "loss": 0.8505, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.8314606741573034, | |
| "grad_norm": 3.101804733276367, | |
| "learning_rate": 3.0269481687995207e-05, | |
| "loss": 0.7702, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8337078651685393, | |
| "grad_norm": 4.0704345703125, | |
| "learning_rate": 3.0223098363215002e-05, | |
| "loss": 0.7591, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.8359550561797753, | |
| "grad_norm": 2.9631364345550537, | |
| "learning_rate": 3.0176535475831208e-05, | |
| "loss": 0.727, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.8382022471910112, | |
| "grad_norm": 3.3760929107666016, | |
| "learning_rate": 3.01297939323671e-05, | |
| "loss": 0.3233, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.8404494382022472, | |
| "grad_norm": 4.116260051727295, | |
| "learning_rate": 3.0082874642824164e-05, | |
| "loss": 0.8738, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.8426966292134831, | |
| "grad_norm": 0.40298929810523987, | |
| "learning_rate": 3.0035778520664388e-05, | |
| "loss": 0.0393, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8449438202247191, | |
| "grad_norm": 3.0647614002227783, | |
| "learning_rate": 2.9988506482792485e-05, | |
| "loss": 0.7454, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.8471910112359551, | |
| "grad_norm": 2.951953649520874, | |
| "learning_rate": 2.994105944953803e-05, | |
| "loss": 0.8297, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.849438202247191, | |
| "grad_norm": 4.049951553344727, | |
| "learning_rate": 2.9893438344637538e-05, | |
| "loss": 0.7802, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.851685393258427, | |
| "grad_norm": 3.7383949756622314, | |
| "learning_rate": 2.984564409521651e-05, | |
| "loss": 0.6229, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.8539325842696629, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.979767763177134e-05, | |
| "loss": 0.0, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8561797752808988, | |
| "grad_norm": 3.399641513824463, | |
| "learning_rate": 2.9749539888151244e-05, | |
| "loss": 0.3506, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.8584269662921349, | |
| "grad_norm": 0.48723292350769043, | |
| "learning_rate": 2.9701231801540032e-05, | |
| "loss": 0.041, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.8606741573033708, | |
| "grad_norm": 3.1171765327453613, | |
| "learning_rate": 2.9652754312437897e-05, | |
| "loss": 0.725, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.8629213483146068, | |
| "grad_norm": 2.6491808891296387, | |
| "learning_rate": 2.9604108364643112e-05, | |
| "loss": 0.257, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.8651685393258427, | |
| "grad_norm": 4.025605201721191, | |
| "learning_rate": 2.9555294905233606e-05, | |
| "loss": 0.7912, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.8674157303370786, | |
| "grad_norm": 4.142299652099609, | |
| "learning_rate": 2.9506314884548583e-05, | |
| "loss": 0.8915, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.8696629213483146, | |
| "grad_norm": 2.943582534790039, | |
| "learning_rate": 2.945716925616998e-05, | |
| "loss": 0.779, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.8719101123595505, | |
| "grad_norm": 4.478114604949951, | |
| "learning_rate": 2.9407858976903913e-05, | |
| "loss": 0.7828, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.8741573033707866, | |
| "grad_norm": 3.9878995418548584, | |
| "learning_rate": 2.935838500676207e-05, | |
| "loss": 0.7462, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.8764044943820225, | |
| "grad_norm": 3.7733311653137207, | |
| "learning_rate": 2.9308748308942983e-05, | |
| "loss": 0.7913, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8786516853932584, | |
| "grad_norm": 3.179732322692871, | |
| "learning_rate": 2.9258949849813315e-05, | |
| "loss": 0.3209, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.8808988764044944, | |
| "grad_norm": 3.6665351390838623, | |
| "learning_rate": 2.9208990598889008e-05, | |
| "loss": 0.5932, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.8831460674157303, | |
| "grad_norm": 0.545093834400177, | |
| "learning_rate": 2.9158871528816442e-05, | |
| "loss": 0.0613, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.8853932584269663, | |
| "grad_norm": 5.226474285125732, | |
| "learning_rate": 2.9108593615353467e-05, | |
| "loss": 0.8802, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.8876404494382022, | |
| "grad_norm": 3.691817283630371, | |
| "learning_rate": 2.9058157837350437e-05, | |
| "loss": 0.6116, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8898876404494382, | |
| "grad_norm": 0.4754512906074524, | |
| "learning_rate": 2.900756517673113e-05, | |
| "loss": 0.0537, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.8921348314606742, | |
| "grad_norm": 2.874117374420166, | |
| "learning_rate": 2.8956816618473647e-05, | |
| "loss": 0.3006, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.8943820224719101, | |
| "grad_norm": 3.8957912921905518, | |
| "learning_rate": 2.890591315059121e-05, | |
| "loss": 0.7636, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.8966292134831461, | |
| "grad_norm": 3.7385432720184326, | |
| "learning_rate": 2.8854855764112973e-05, | |
| "loss": 0.612, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.898876404494382, | |
| "grad_norm": 3.7403082847595215, | |
| "learning_rate": 2.880364545306468e-05, | |
| "loss": 0.54, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9011235955056179, | |
| "grad_norm": 2.7360849380493164, | |
| "learning_rate": 2.8752283214449328e-05, | |
| "loss": 0.2761, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.903370786516854, | |
| "grad_norm": 8.988025665283203, | |
| "learning_rate": 2.8700770048227775e-05, | |
| "loss": 1.2668, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.9056179775280899, | |
| "grad_norm": 3.411295175552368, | |
| "learning_rate": 2.864910695729925e-05, | |
| "loss": 0.8066, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.9078651685393259, | |
| "grad_norm": 0.3018481135368347, | |
| "learning_rate": 2.8597294947481834e-05, | |
| "loss": 0.0094, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "grad_norm": 4.116438388824463, | |
| "learning_rate": 2.8545335027492885e-05, | |
| "loss": 0.673, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_VitaminC_cosine_accuracy": 0.55859375, | |
| "eval_VitaminC_cosine_accuracy_threshold": 0.7188639044761658, | |
| "eval_VitaminC_cosine_ap": 0.5516905675485202, | |
| "eval_VitaminC_cosine_f1": 0.6675712347354138, | |
| "eval_VitaminC_cosine_f1_threshold": 0.42514583468437195, | |
| "eval_VitaminC_cosine_precision": 0.5061728395061729, | |
| "eval_VitaminC_cosine_recall": 0.9800796812749004, | |
| "eval_VitaminC_dot_accuracy": 0.548828125, | |
| "eval_VitaminC_dot_accuracy_threshold": 320.3775329589844, | |
| "eval_VitaminC_dot_ap": 0.5343066680873013, | |
| "eval_VitaminC_dot_f1": 0.6720867208672087, | |
| "eval_VitaminC_dot_f1_threshold": 152.709716796875, | |
| "eval_VitaminC_dot_precision": 0.5092402464065708, | |
| "eval_VitaminC_dot_recall": 0.9880478087649402, | |
| "eval_VitaminC_euclidean_accuracy": 0.556640625, | |
| "eval_VitaminC_euclidean_accuracy_threshold": 15.12228775024414, | |
| "eval_VitaminC_euclidean_ap": 0.5542894540784595, | |
| "eval_VitaminC_euclidean_f1": 0.6640211640211641, | |
| "eval_VitaminC_euclidean_f1_threshold": 24.3716983795166, | |
| "eval_VitaminC_euclidean_precision": 0.497029702970297, | |
| "eval_VitaminC_euclidean_recall": 1.0, | |
| "eval_VitaminC_manhattan_accuracy": 0.55859375, | |
| "eval_VitaminC_manhattan_accuracy_threshold": 305.93597412109375, | |
| "eval_VitaminC_manhattan_ap": 0.5533328154567183, | |
| "eval_VitaminC_manhattan_f1": 0.6649006622516557, | |
| "eval_VitaminC_manhattan_f1_threshold": 509.4247741699219, | |
| "eval_VitaminC_manhattan_precision": 0.498015873015873, | |
| "eval_VitaminC_manhattan_recall": 1.0, | |
| "eval_VitaminC_max_accuracy": 0.55859375, | |
| "eval_VitaminC_max_accuracy_threshold": 320.3775329589844, | |
| "eval_VitaminC_max_ap": 0.5542894540784595, | |
| "eval_VitaminC_max_f1": 0.6720867208672087, | |
| "eval_VitaminC_max_f1_threshold": 509.4247741699219, | |
| "eval_VitaminC_max_precision": 0.5092402464065708, | |
| "eval_VitaminC_max_recall": 1.0, | |
| "eval_sequential_score": 0.5542894540784595, | |
| "eval_sts-test_pearson_cosine": 0.8820726638294588, | |
| "eval_sts-test_pearson_dot": 0.8723940521896922, | |
| "eval_sts-test_pearson_euclidean": 0.9038814103150634, | |
| "eval_sts-test_pearson_manhattan": 0.904449390563823, | |
| "eval_sts-test_pearson_max": 0.904449390563823, | |
| "eval_sts-test_spearman_cosine": 0.9051641183600871, | |
| "eval_sts-test_spearman_dot": 0.8721959088443044, | |
| "eval_sts-test_spearman_euclidean": 0.8999642007914521, | |
| "eval_sts-test_spearman_manhattan": 0.9005904051921018, | |
| "eval_sts-test_spearman_max": 0.9051641183600871, | |
| "eval_vitaminc-pairs_loss": 1.48486328125, | |
| "eval_vitaminc-pairs_runtime": 1.8874, | |
| "eval_vitaminc-pairs_samples_per_second": 57.222, | |
| "eval_vitaminc-pairs_steps_per_second": 1.06, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_negation-triplets_loss": 0.9023827314376831, | |
| "eval_negation-triplets_runtime": 0.302, | |
| "eval_negation-triplets_samples_per_second": 211.927, | |
| "eval_negation-triplets_steps_per_second": 3.311, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_scitail-pairs-pos_loss": 0.10495099425315857, | |
| "eval_scitail-pairs-pos_runtime": 0.3856, | |
| "eval_scitail-pairs-pos_samples_per_second": 140.031, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.593, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_scitail-pairs-qa_loss": 0.0008332311408594251, | |
| "eval_scitail-pairs-qa_runtime": 0.5224, | |
| "eval_scitail-pairs-qa_samples_per_second": 245.005, | |
| "eval_scitail-pairs-qa_steps_per_second": 3.828, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_xsum-pairs_loss": 0.028531953692436218, | |
| "eval_xsum-pairs_runtime": 2.7425, | |
| "eval_xsum-pairs_samples_per_second": 46.672, | |
| "eval_xsum-pairs_steps_per_second": 0.729, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_sciq_pairs_loss": 0.015175853855907917, | |
| "eval_sciq_pairs_runtime": 2.8294, | |
| "eval_sciq_pairs_samples_per_second": 45.239, | |
| "eval_sciq_pairs_steps_per_second": 0.707, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_qasc_pairs_loss": 0.09416583180427551, | |
| "eval_qasc_pairs_runtime": 0.6538, | |
| "eval_qasc_pairs_samples_per_second": 195.781, | |
| "eval_qasc_pairs_steps_per_second": 3.059, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_openbookqa_pairs_loss": 0.715216875076294, | |
| "eval_openbookqa_pairs_runtime": 0.578, | |
| "eval_openbookqa_pairs_samples_per_second": 221.449, | |
| "eval_openbookqa_pairs_steps_per_second": 3.46, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_msmarco_pairs_loss": 0.1417744755744934, | |
| "eval_msmarco_pairs_runtime": 1.4882, | |
| "eval_msmarco_pairs_samples_per_second": 86.012, | |
| "eval_msmarco_pairs_steps_per_second": 1.344, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_nq_pairs_loss": 0.10870223492383957, | |
| "eval_nq_pairs_runtime": 2.3451, | |
| "eval_nq_pairs_samples_per_second": 54.583, | |
| "eval_nq_pairs_steps_per_second": 0.853, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_trivia_pairs_loss": 0.49194595217704773, | |
| "eval_trivia_pairs_runtime": 3.5796, | |
| "eval_trivia_pairs_samples_per_second": 35.759, | |
| "eval_trivia_pairs_steps_per_second": 0.559, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_gooaq_pairs_loss": 0.2616226375102997, | |
| "eval_gooaq_pairs_runtime": 0.9137, | |
| "eval_gooaq_pairs_samples_per_second": 140.093, | |
| "eval_gooaq_pairs_steps_per_second": 2.189, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9101123595505618, | |
| "eval_paws-pos_loss": 0.02422034554183483, | |
| "eval_paws-pos_runtime": 0.6895, | |
| "eval_paws-pos_samples_per_second": 185.641, | |
| "eval_paws-pos_steps_per_second": 2.901, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9123595505617977, | |
| "grad_norm": 3.427104949951172, | |
| "learning_rate": 2.8493228208929387e-05, | |
| "loss": 0.5189, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.9146067415730337, | |
| "grad_norm": 4.941195487976074, | |
| "learning_rate": 2.8440975506248268e-05, | |
| "loss": 0.649, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.9168539325842696, | |
| "grad_norm": 2.7992403507232666, | |
| "learning_rate": 2.8388577936746633e-05, | |
| "loss": 0.2982, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.9191011235955057, | |
| "grad_norm": 3.8877484798431396, | |
| "learning_rate": 2.833603652054199e-05, | |
| "loss": 0.7511, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.9213483146067416, | |
| "grad_norm": 3.2458090782165527, | |
| "learning_rate": 2.8283352280552348e-05, | |
| "loss": 0.5164, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.9235955056179775, | |
| "grad_norm": 3.7385945320129395, | |
| "learning_rate": 2.8230526242476332e-05, | |
| "loss": 0.5924, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.9258426966292135, | |
| "grad_norm": 4.369627952575684, | |
| "learning_rate": 2.8177559434773203e-05, | |
| "loss": 0.8191, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.9280898876404494, | |
| "grad_norm": 2.95206356048584, | |
| "learning_rate": 2.8124452888642838e-05, | |
| "loss": 0.2311, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.9303370786516854, | |
| "grad_norm": 3.984375238418579, | |
| "learning_rate": 2.8071207638005662e-05, | |
| "loss": 0.7421, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.9325842696629213, | |
| "grad_norm": 3.0188541412353516, | |
| "learning_rate": 2.801782471948248e-05, | |
| "loss": 0.2936, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.9348314606741573, | |
| "grad_norm": 4.104308605194092, | |
| "learning_rate": 2.7964305172374362e-05, | |
| "loss": 0.737, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.9370786516853933, | |
| "grad_norm": 3.686523675918579, | |
| "learning_rate": 2.791065003864235e-05, | |
| "loss": 0.6539, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.9393258426966292, | |
| "grad_norm": 3.839590311050415, | |
| "learning_rate": 2.785686036288719e-05, | |
| "loss": 0.6855, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.9415730337078652, | |
| "grad_norm": 4.174718856811523, | |
| "learning_rate": 2.780293719232902e-05, | |
| "loss": 0.8134, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.9438202247191011, | |
| "grad_norm": 4.046380043029785, | |
| "learning_rate": 2.7748881576786946e-05, | |
| "loss": 0.6885, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.946067415730337, | |
| "grad_norm": 3.4202940464019775, | |
| "learning_rate": 2.7694694568658613e-05, | |
| "loss": 0.5581, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.9483146067415731, | |
| "grad_norm": 3.787081718444824, | |
| "learning_rate": 2.764037722289973e-05, | |
| "loss": 0.8029, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.950561797752809, | |
| "grad_norm": 3.870718240737915, | |
| "learning_rate": 2.7585930597003524e-05, | |
| "loss": 0.8126, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.952808988764045, | |
| "grad_norm": 3.1959424018859863, | |
| "learning_rate": 2.753135575098015e-05, | |
| "loss": 0.8425, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.9550561797752809, | |
| "grad_norm": 0.4186573922634125, | |
| "learning_rate": 2.7476653747336047e-05, | |
| "loss": 0.049, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9573033707865168, | |
| "grad_norm": 4.299917697906494, | |
| "learning_rate": 2.7421825651053265e-05, | |
| "loss": 0.7849, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.9595505617977528, | |
| "grad_norm": 2.6435227394104004, | |
| "learning_rate": 2.736687252956873e-05, | |
| "loss": 0.068, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.9617977528089887, | |
| "grad_norm": 2.717653274536133, | |
| "learning_rate": 2.7311795452753443e-05, | |
| "loss": 0.2925, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.9640449438202248, | |
| "grad_norm": 3.6929807662963867, | |
| "learning_rate": 2.7256595492891683e-05, | |
| "loss": 0.777, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.9662921348314607, | |
| "grad_norm": 2.8760790824890137, | |
| "learning_rate": 2.720127372466011e-05, | |
| "loss": 0.7397, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9685393258426966, | |
| "grad_norm": 0.03685740381479263, | |
| "learning_rate": 2.714583122510683e-05, | |
| "loss": 0.0007, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.9707865168539326, | |
| "grad_norm": 4.058692455291748, | |
| "learning_rate": 2.709026907363047e-05, | |
| "loss": 0.8535, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.9730337078651685, | |
| "grad_norm": 4.2914276123046875, | |
| "learning_rate": 2.703458835195911e-05, | |
| "loss": 0.7026, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.9752808988764045, | |
| "grad_norm": 3.735518217086792, | |
| "learning_rate": 2.6978790144129262e-05, | |
| "loss": 0.7557, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.9775280898876404, | |
| "grad_norm": 4.058504104614258, | |
| "learning_rate": 2.6922875536464747e-05, | |
| "loss": 0.7225, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9797752808988764, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6866845617555555e-05, | |
| "loss": 0.0, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.9820224719101124, | |
| "grad_norm": 5.648872375488281, | |
| "learning_rate": 2.6810701478236642e-05, | |
| "loss": 0.4131, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.9842696629213483, | |
| "grad_norm": 2.7032744884490967, | |
| "learning_rate": 2.6754444211566702e-05, | |
| "loss": 0.2824, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.9865168539325843, | |
| "grad_norm": 3.150801420211792, | |
| "learning_rate": 2.6698074912806882e-05, | |
| "loss": 0.3144, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.9887640449438202, | |
| "grad_norm": 2.3572490215301514, | |
| "learning_rate": 2.6641594679399448e-05, | |
| "loss": 0.0509, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9910112359550561, | |
| "grad_norm": 3.2544448375701904, | |
| "learning_rate": 2.6585004610946452e-05, | |
| "loss": 0.7645, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.9932584269662922, | |
| "grad_norm": 4.310440540313721, | |
| "learning_rate": 2.6528305809188273e-05, | |
| "loss": 0.2787, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.9955056179775281, | |
| "grad_norm": 3.863487482070923, | |
| "learning_rate": 2.6471499377982225e-05, | |
| "loss": 0.64, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.9977528089887641, | |
| "grad_norm": 6.1020612716674805, | |
| "learning_rate": 2.6414586423281017e-05, | |
| "loss": 0.4045, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 3.1245224475860596, | |
| "learning_rate": 2.6357568053111255e-05, | |
| "loss": 0.7661, | |
| "step": 445 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 890, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 89, | |
| "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": 640, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |