| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 735, | |
| "global_step": 5874, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02502553626149132, | |
| "grad_norm": 65.55949401855469, | |
| "learning_rate": 4.834865509022812e-07, | |
| "loss": 16.851, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.05005107252298264, | |
| "grad_norm": 23.207971572875977, | |
| "learning_rate": 9.805924412665985e-07, | |
| "loss": 11.2787, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.07507660878447395, | |
| "grad_norm": 176.1532440185547, | |
| "learning_rate": 1.481103166496425e-06, | |
| "loss": 8.9166, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.10010214504596528, | |
| "grad_norm": 22.1564998626709, | |
| "learning_rate": 1.981613891726251e-06, | |
| "loss": 7.9463, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "grad_norm": 20.11876106262207, | |
| "learning_rate": 2.4821246169560777e-06, | |
| "loss": 7.2108, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_nli-pairs_loss": 6.905651569366455, | |
| "eval_nli-pairs_runtime": 4.0844, | |
| "eval_nli-pairs_samples_per_second": 36.725, | |
| "eval_nli-pairs_steps_per_second": 1.224, | |
| "eval_sts-test_pearson_cosine": 0.3740256550072784, | |
| "eval_sts-test_pearson_dot": 0.13384893803205677, | |
| "eval_sts-test_pearson_euclidean": 0.3912387619869807, | |
| "eval_sts-test_pearson_manhattan": 0.4202605137823524, | |
| "eval_sts-test_pearson_max": 0.4202605137823524, | |
| "eval_sts-test_spearman_cosine": 0.37210107338950205, | |
| "eval_sts-test_spearman_dot": 0.12092409843417483, | |
| "eval_sts-test_spearman_euclidean": 0.39172287978780546, | |
| "eval_sts-test_spearman_manhattan": 0.4169664738563951, | |
| "eval_sts-test_spearman_max": 0.4169664738563951, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_vitaminc-pairs_loss": 5.720878601074219, | |
| "eval_vitaminc-pairs_runtime": 2.1703, | |
| "eval_vitaminc-pairs_samples_per_second": 69.115, | |
| "eval_vitaminc-pairs_steps_per_second": 2.304, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_qnli-contrastive_loss": 8.1649751663208, | |
| "eval_qnli-contrastive_runtime": 0.4937, | |
| "eval_qnli-contrastive_samples_per_second": 303.841, | |
| "eval_qnli-contrastive_steps_per_second": 10.128, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_scitail-pairs-qa_loss": 3.7859296798706055, | |
| "eval_scitail-pairs-qa_runtime": 1.1509, | |
| "eval_scitail-pairs-qa_samples_per_second": 130.329, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.344, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_scitail-pairs-pos_loss": 3.9919917583465576, | |
| "eval_scitail-pairs-pos_runtime": 2.1442, | |
| "eval_scitail-pairs-pos_samples_per_second": 69.956, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.332, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_xsum-pairs_loss": 4.600368976593018, | |
| "eval_xsum-pairs_runtime": 2.26, | |
| "eval_xsum-pairs_samples_per_second": 66.371, | |
| "eval_xsum-pairs_steps_per_second": 2.212, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_compression-pairs_loss": 3.3037569522857666, | |
| "eval_compression-pairs_runtime": 0.449, | |
| "eval_compression-pairs_samples_per_second": 334.078, | |
| "eval_compression-pairs_steps_per_second": 11.136, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_sciq_pairs_loss": 10.214456558227539, | |
| "eval_sciq_pairs_runtime": 7.1179, | |
| "eval_sciq_pairs_samples_per_second": 21.074, | |
| "eval_sciq_pairs_steps_per_second": 0.702, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_qasc_pairs_loss": 10.58031940460205, | |
| "eval_qasc_pairs_runtime": 2.0175, | |
| "eval_qasc_pairs_samples_per_second": 74.348, | |
| "eval_qasc_pairs_steps_per_second": 2.478, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_openbookqa_pairs_loss": 7.862658977508545, | |
| "eval_openbookqa_pairs_runtime": 0.8571, | |
| "eval_openbookqa_pairs_samples_per_second": 120.168, | |
| "eval_openbookqa_pairs_steps_per_second": 4.667, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_msmarco_pairs_loss": 8.754273414611816, | |
| "eval_msmarco_pairs_runtime": 2.7533, | |
| "eval_msmarco_pairs_samples_per_second": 54.481, | |
| "eval_msmarco_pairs_steps_per_second": 1.816, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_nq_pairs_loss": 8.415486335754395, | |
| "eval_nq_pairs_runtime": 5.0894, | |
| "eval_nq_pairs_samples_per_second": 29.473, | |
| "eval_nq_pairs_steps_per_second": 0.982, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_trivia_pairs_loss": 9.051105499267578, | |
| "eval_trivia_pairs_runtime": 9.5498, | |
| "eval_trivia_pairs_samples_per_second": 15.707, | |
| "eval_trivia_pairs_steps_per_second": 0.524, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_quora_pairs_loss": 4.5232110023498535, | |
| "eval_quora_pairs_runtime": 1.1469, | |
| "eval_quora_pairs_samples_per_second": 130.785, | |
| "eval_quora_pairs_steps_per_second": 4.36, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.12512768130745658, | |
| "eval_gooaq_pairs_loss": 7.579105854034424, | |
| "eval_gooaq_pairs_runtime": 2.0491, | |
| "eval_gooaq_pairs_samples_per_second": 73.203, | |
| "eval_gooaq_pairs_steps_per_second": 2.44, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.1501532175689479, | |
| "grad_norm": 31.7736759185791, | |
| "learning_rate": 2.982635342185904e-06, | |
| "loss": 6.7709, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.1751787538304392, | |
| "grad_norm": 31.57339096069336, | |
| "learning_rate": 3.4831460674157306e-06, | |
| "loss": 6.1746, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.20020429009193055, | |
| "grad_norm": 25.392702102661133, | |
| "learning_rate": 3.9836567926455565e-06, | |
| "loss": 5.7706, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.22522982635342187, | |
| "grad_norm": 32.390472412109375, | |
| "learning_rate": 4.484167517875383e-06, | |
| "loss": 5.7283, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "grad_norm": 18.85039520263672, | |
| "learning_rate": 4.98467824310521e-06, | |
| "loss": 5.1856, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_nli-pairs_loss": 4.352054119110107, | |
| "eval_nli-pairs_runtime": 4.1476, | |
| "eval_nli-pairs_samples_per_second": 36.165, | |
| "eval_nli-pairs_steps_per_second": 1.206, | |
| "eval_sts-test_pearson_cosine": 0.6694155778571752, | |
| "eval_sts-test_pearson_dot": 0.5201102118957572, | |
| "eval_sts-test_pearson_euclidean": 0.6613028243200022, | |
| "eval_sts-test_pearson_manhattan": 0.6670710500315469, | |
| "eval_sts-test_pearson_max": 0.6694155778571752, | |
| "eval_sts-test_spearman_cosine": 0.6367853204388882, | |
| "eval_sts-test_spearman_dot": 0.4940207180607985, | |
| "eval_sts-test_spearman_euclidean": 0.6391132775161348, | |
| "eval_sts-test_spearman_manhattan": 0.6446159957787251, | |
| "eval_sts-test_spearman_max": 0.6446159957787251, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_vitaminc-pairs_loss": 3.4987735748291016, | |
| "eval_vitaminc-pairs_runtime": 2.1678, | |
| "eval_vitaminc-pairs_samples_per_second": 69.194, | |
| "eval_vitaminc-pairs_steps_per_second": 2.306, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_qnli-contrastive_loss": 12.915559768676758, | |
| "eval_qnli-contrastive_runtime": 0.4918, | |
| "eval_qnli-contrastive_samples_per_second": 304.99, | |
| "eval_qnli-contrastive_steps_per_second": 10.166, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_scitail-pairs-qa_loss": 1.3250077962875366, | |
| "eval_scitail-pairs-qa_runtime": 1.154, | |
| "eval_scitail-pairs-qa_samples_per_second": 129.984, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.333, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_scitail-pairs-pos_loss": 2.457335948944092, | |
| "eval_scitail-pairs-pos_runtime": 2.1475, | |
| "eval_scitail-pairs-pos_samples_per_second": 69.85, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.328, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_xsum-pairs_loss": 3.071201801300049, | |
| "eval_xsum-pairs_runtime": 2.2634, | |
| "eval_xsum-pairs_samples_per_second": 66.271, | |
| "eval_xsum-pairs_steps_per_second": 2.209, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_compression-pairs_loss": 2.0629916191101074, | |
| "eval_compression-pairs_runtime": 0.4529, | |
| "eval_compression-pairs_samples_per_second": 331.23, | |
| "eval_compression-pairs_steps_per_second": 11.041, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_sciq_pairs_loss": 9.06814193725586, | |
| "eval_sciq_pairs_runtime": 7.1445, | |
| "eval_sciq_pairs_samples_per_second": 20.995, | |
| "eval_sciq_pairs_steps_per_second": 0.7, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_qasc_pairs_loss": 9.245658874511719, | |
| "eval_qasc_pairs_runtime": 2.0471, | |
| "eval_qasc_pairs_samples_per_second": 73.274, | |
| "eval_qasc_pairs_steps_per_second": 2.442, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_openbookqa_pairs_loss": 5.652446746826172, | |
| "eval_openbookqa_pairs_runtime": 0.8946, | |
| "eval_openbookqa_pairs_samples_per_second": 115.14, | |
| "eval_openbookqa_pairs_steps_per_second": 4.471, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_msmarco_pairs_loss": 4.844855785369873, | |
| "eval_msmarco_pairs_runtime": 2.7887, | |
| "eval_msmarco_pairs_samples_per_second": 53.788, | |
| "eval_msmarco_pairs_steps_per_second": 1.793, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_nq_pairs_loss": 5.023958206176758, | |
| "eval_nq_pairs_runtime": 5.0823, | |
| "eval_nq_pairs_samples_per_second": 29.514, | |
| "eval_nq_pairs_steps_per_second": 0.984, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_trivia_pairs_loss": 5.2907304763793945, | |
| "eval_trivia_pairs_runtime": 9.6673, | |
| "eval_trivia_pairs_samples_per_second": 15.516, | |
| "eval_trivia_pairs_steps_per_second": 0.517, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_quora_pairs_loss": 1.5572240352630615, | |
| "eval_quora_pairs_runtime": 1.1979, | |
| "eval_quora_pairs_samples_per_second": 125.218, | |
| "eval_quora_pairs_steps_per_second": 4.174, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.25025536261491316, | |
| "eval_gooaq_pairs_loss": 3.970768928527832, | |
| "eval_gooaq_pairs_runtime": 2.117, | |
| "eval_gooaq_pairs_samples_per_second": 70.855, | |
| "eval_gooaq_pairs_steps_per_second": 2.362, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.2752808988764045, | |
| "grad_norm": 40.67585754394531, | |
| "learning_rate": 5.4851889683350365e-06, | |
| "loss": 4.185, | |
| "step": 1617 | |
| }, | |
| { | |
| "epoch": 0.3003064351378958, | |
| "grad_norm": 45.92570495605469, | |
| "learning_rate": 5.985699693564862e-06, | |
| "loss": 4.6367, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 0.32533197139938713, | |
| "grad_norm": 13.566838264465332, | |
| "learning_rate": 6.486210418794688e-06, | |
| "loss": 4.3615, | |
| "step": 1911 | |
| }, | |
| { | |
| "epoch": 0.3503575076608784, | |
| "grad_norm": 9.495999336242676, | |
| "learning_rate": 6.986721144024515e-06, | |
| "loss": 4.1791, | |
| "step": 2058 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "grad_norm": 32.735416412353516, | |
| "learning_rate": 7.487231869254341e-06, | |
| "loss": 4.1051, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_nli-pairs_loss": 3.2717113494873047, | |
| "eval_nli-pairs_runtime": 4.0124, | |
| "eval_nli-pairs_samples_per_second": 37.384, | |
| "eval_nli-pairs_steps_per_second": 1.246, | |
| "eval_sts-test_pearson_cosine": 0.6958570089637609, | |
| "eval_sts-test_pearson_dot": 0.5824298957890577, | |
| "eval_sts-test_pearson_euclidean": 0.6893962819387462, | |
| "eval_sts-test_pearson_manhattan": 0.6993681181979946, | |
| "eval_sts-test_pearson_max": 0.6993681181979946, | |
| "eval_sts-test_spearman_cosine": 0.6652712160836801, | |
| "eval_sts-test_spearman_dot": 0.5536505624407877, | |
| "eval_sts-test_spearman_euclidean": 0.6659844314307678, | |
| "eval_sts-test_spearman_manhattan": 0.675740852112121, | |
| "eval_sts-test_spearman_max": 0.675740852112121, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_vitaminc-pairs_loss": 2.7197911739349365, | |
| "eval_vitaminc-pairs_runtime": 2.1625, | |
| "eval_vitaminc-pairs_samples_per_second": 69.365, | |
| "eval_vitaminc-pairs_steps_per_second": 2.312, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_qnli-contrastive_loss": 9.638714790344238, | |
| "eval_qnli-contrastive_runtime": 0.4877, | |
| "eval_qnli-contrastive_samples_per_second": 307.567, | |
| "eval_qnli-contrastive_steps_per_second": 10.252, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_scitail-pairs-qa_loss": 0.8106752634048462, | |
| "eval_scitail-pairs-qa_runtime": 1.1588, | |
| "eval_scitail-pairs-qa_samples_per_second": 129.449, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.315, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_scitail-pairs-pos_loss": 1.8894625902175903, | |
| "eval_scitail-pairs-pos_runtime": 2.1181, | |
| "eval_scitail-pairs-pos_samples_per_second": 70.817, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.361, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_xsum-pairs_loss": 2.262718439102173, | |
| "eval_xsum-pairs_runtime": 2.2585, | |
| "eval_xsum-pairs_samples_per_second": 66.416, | |
| "eval_xsum-pairs_steps_per_second": 2.214, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_compression-pairs_loss": 1.4910633563995361, | |
| "eval_compression-pairs_runtime": 0.4462, | |
| "eval_compression-pairs_samples_per_second": 336.204, | |
| "eval_compression-pairs_steps_per_second": 11.207, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_sciq_pairs_loss": 8.59740161895752, | |
| "eval_sciq_pairs_runtime": 7.1845, | |
| "eval_sciq_pairs_samples_per_second": 20.878, | |
| "eval_sciq_pairs_steps_per_second": 0.696, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_qasc_pairs_loss": 8.103879928588867, | |
| "eval_qasc_pairs_runtime": 2.0762, | |
| "eval_qasc_pairs_samples_per_second": 72.246, | |
| "eval_qasc_pairs_steps_per_second": 2.408, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_openbookqa_pairs_loss": 5.090969562530518, | |
| "eval_openbookqa_pairs_runtime": 0.89, | |
| "eval_openbookqa_pairs_samples_per_second": 115.726, | |
| "eval_openbookqa_pairs_steps_per_second": 4.494, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_msmarco_pairs_loss": 3.9566943645477295, | |
| "eval_msmarco_pairs_runtime": 2.8183, | |
| "eval_msmarco_pairs_samples_per_second": 53.223, | |
| "eval_msmarco_pairs_steps_per_second": 1.774, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_nq_pairs_loss": 4.009054183959961, | |
| "eval_nq_pairs_runtime": 5.0219, | |
| "eval_nq_pairs_samples_per_second": 29.869, | |
| "eval_nq_pairs_steps_per_second": 0.996, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_trivia_pairs_loss": 4.286431312561035, | |
| "eval_trivia_pairs_runtime": 9.4975, | |
| "eval_trivia_pairs_samples_per_second": 15.794, | |
| "eval_trivia_pairs_steps_per_second": 0.526, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_quora_pairs_loss": 1.123273491859436, | |
| "eval_quora_pairs_runtime": 1.1487, | |
| "eval_quora_pairs_samples_per_second": 130.586, | |
| "eval_quora_pairs_steps_per_second": 4.353, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.37538304392236976, | |
| "eval_gooaq_pairs_loss": 3.222414255142212, | |
| "eval_gooaq_pairs_runtime": 2.0173, | |
| "eval_gooaq_pairs_samples_per_second": 74.357, | |
| "eval_gooaq_pairs_steps_per_second": 2.479, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.4004085801838611, | |
| "grad_norm": 218.56105041503906, | |
| "learning_rate": 7.987742594484168e-06, | |
| "loss": 3.7674, | |
| "step": 2352 | |
| }, | |
| { | |
| "epoch": 0.4254341164453524, | |
| "grad_norm": 27.877609252929688, | |
| "learning_rate": 8.488253319713993e-06, | |
| "loss": 3.8729, | |
| "step": 2499 | |
| }, | |
| { | |
| "epoch": 0.45045965270684374, | |
| "grad_norm": 33.50013732910156, | |
| "learning_rate": 8.988764044943822e-06, | |
| "loss": 3.4527, | |
| "step": 2646 | |
| }, | |
| { | |
| "epoch": 0.475485188968335, | |
| "grad_norm": 14.015911102294922, | |
| "learning_rate": 9.489274770173647e-06, | |
| "loss": 3.3545, | |
| "step": 2793 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "grad_norm": 33.59694290161133, | |
| "learning_rate": 9.989785495403473e-06, | |
| "loss": 3.3247, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_nli-pairs_loss": 2.7121565341949463, | |
| "eval_nli-pairs_runtime": 4.1564, | |
| "eval_nli-pairs_samples_per_second": 36.089, | |
| "eval_nli-pairs_steps_per_second": 1.203, | |
| "eval_sts-test_pearson_cosine": 0.716623047702725, | |
| "eval_sts-test_pearson_dot": 0.6128451070598809, | |
| "eval_sts-test_pearson_euclidean": 0.7138791236031807, | |
| "eval_sts-test_pearson_manhattan": 0.7213151818687454, | |
| "eval_sts-test_pearson_max": 0.7213151818687454, | |
| "eval_sts-test_spearman_cosine": 0.6919792400941177, | |
| "eval_sts-test_spearman_dot": 0.5867158357121192, | |
| "eval_sts-test_spearman_euclidean": 0.6925037259567834, | |
| "eval_sts-test_spearman_manhattan": 0.7008895667910079, | |
| "eval_sts-test_spearman_max": 0.7008895667910079, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_vitaminc-pairs_loss": 2.225992441177368, | |
| "eval_vitaminc-pairs_runtime": 2.253, | |
| "eval_vitaminc-pairs_samples_per_second": 66.577, | |
| "eval_vitaminc-pairs_steps_per_second": 2.219, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_qnli-contrastive_loss": 4.92629861831665, | |
| "eval_qnli-contrastive_runtime": 0.5005, | |
| "eval_qnli-contrastive_samples_per_second": 299.691, | |
| "eval_qnli-contrastive_steps_per_second": 9.99, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_scitail-pairs-qa_loss": 0.5898066163063049, | |
| "eval_scitail-pairs-qa_runtime": 1.2227, | |
| "eval_scitail-pairs-qa_samples_per_second": 122.682, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.089, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_scitail-pairs-pos_loss": 1.4237287044525146, | |
| "eval_scitail-pairs-pos_runtime": 2.4409, | |
| "eval_scitail-pairs-pos_samples_per_second": 61.452, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.048, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_xsum-pairs_loss": 1.8388895988464355, | |
| "eval_xsum-pairs_runtime": 2.2831, | |
| "eval_xsum-pairs_samples_per_second": 65.7, | |
| "eval_xsum-pairs_steps_per_second": 2.19, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_compression-pairs_loss": 1.1590967178344727, | |
| "eval_compression-pairs_runtime": 0.5152, | |
| "eval_compression-pairs_samples_per_second": 291.165, | |
| "eval_compression-pairs_steps_per_second": 9.706, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_sciq_pairs_loss": 8.282496452331543, | |
| "eval_sciq_pairs_runtime": 7.2871, | |
| "eval_sciq_pairs_samples_per_second": 20.584, | |
| "eval_sciq_pairs_steps_per_second": 0.686, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_qasc_pairs_loss": 7.817965507507324, | |
| "eval_qasc_pairs_runtime": 2.0211, | |
| "eval_qasc_pairs_samples_per_second": 74.218, | |
| "eval_qasc_pairs_steps_per_second": 2.474, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_openbookqa_pairs_loss": 4.619383811950684, | |
| "eval_openbookqa_pairs_runtime": 0.8531, | |
| "eval_openbookqa_pairs_samples_per_second": 120.731, | |
| "eval_openbookqa_pairs_steps_per_second": 4.689, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_msmarco_pairs_loss": 3.478559970855713, | |
| "eval_msmarco_pairs_runtime": 2.7512, | |
| "eval_msmarco_pairs_samples_per_second": 54.522, | |
| "eval_msmarco_pairs_steps_per_second": 1.817, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_nq_pairs_loss": 3.3449866771698, | |
| "eval_nq_pairs_runtime": 5.0591, | |
| "eval_nq_pairs_samples_per_second": 29.649, | |
| "eval_nq_pairs_steps_per_second": 0.988, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_trivia_pairs_loss": 3.524484872817993, | |
| "eval_trivia_pairs_runtime": 9.662, | |
| "eval_trivia_pairs_samples_per_second": 15.525, | |
| "eval_trivia_pairs_steps_per_second": 0.517, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_quora_pairs_loss": 0.9095575213432312, | |
| "eval_quora_pairs_runtime": 1.2482, | |
| "eval_quora_pairs_samples_per_second": 120.175, | |
| "eval_quora_pairs_steps_per_second": 4.006, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5005107252298263, | |
| "eval_gooaq_pairs_loss": 2.6586034297943115, | |
| "eval_gooaq_pairs_runtime": 2.1091, | |
| "eval_gooaq_pairs_samples_per_second": 71.12, | |
| "eval_gooaq_pairs_steps_per_second": 2.371, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.5255362614913177, | |
| "grad_norm": 35.33409118652344, | |
| "learning_rate": 1.04902962206333e-05, | |
| "loss": 3.116, | |
| "step": 3087 | |
| }, | |
| { | |
| "epoch": 0.550561797752809, | |
| "grad_norm": 22.29003143310547, | |
| "learning_rate": 1.0990806945863125e-05, | |
| "loss": 3.2418, | |
| "step": 3234 | |
| }, | |
| { | |
| "epoch": 0.5755873340143003, | |
| "grad_norm": 31.277965545654297, | |
| "learning_rate": 1.1491317671092953e-05, | |
| "loss": 3.0757, | |
| "step": 3381 | |
| }, | |
| { | |
| "epoch": 0.6006128702757916, | |
| "grad_norm": 24.612506866455078, | |
| "learning_rate": 1.1991828396322778e-05, | |
| "loss": 2.8524, | |
| "step": 3528 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "grad_norm": 25.11741065979004, | |
| "learning_rate": 1.2492339121552605e-05, | |
| "loss": 2.6875, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_nli-pairs_loss": 2.479051113128662, | |
| "eval_nli-pairs_runtime": 3.9943, | |
| "eval_nli-pairs_samples_per_second": 37.553, | |
| "eval_nli-pairs_steps_per_second": 1.252, | |
| "eval_sts-test_pearson_cosine": 0.7278742453545186, | |
| "eval_sts-test_pearson_dot": 0.6217650825208566, | |
| "eval_sts-test_pearson_euclidean": 0.7243228472931561, | |
| "eval_sts-test_pearson_manhattan": 0.7333297580184588, | |
| "eval_sts-test_pearson_max": 0.7333297580184588, | |
| "eval_sts-test_spearman_cosine": 0.7013110457844404, | |
| "eval_sts-test_spearman_dot": 0.5970993074902947, | |
| "eval_sts-test_spearman_euclidean": 0.701564129266252, | |
| "eval_sts-test_spearman_manhattan": 0.7116482009924582, | |
| "eval_sts-test_spearman_max": 0.7116482009924582, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_vitaminc-pairs_loss": 1.974273681640625, | |
| "eval_vitaminc-pairs_runtime": 2.1754, | |
| "eval_vitaminc-pairs_samples_per_second": 68.953, | |
| "eval_vitaminc-pairs_steps_per_second": 2.298, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_qnli-contrastive_loss": 1.7706010341644287, | |
| "eval_qnli-contrastive_runtime": 0.4866, | |
| "eval_qnli-contrastive_samples_per_second": 308.244, | |
| "eval_qnli-contrastive_steps_per_second": 10.275, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_scitail-pairs-qa_loss": 0.4400452673435211, | |
| "eval_scitail-pairs-qa_runtime": 1.1519, | |
| "eval_scitail-pairs-qa_samples_per_second": 130.222, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.341, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_scitail-pairs-pos_loss": 1.1909903287887573, | |
| "eval_scitail-pairs-pos_runtime": 2.1319, | |
| "eval_scitail-pairs-pos_samples_per_second": 70.36, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.345, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_xsum-pairs_loss": 1.4811985492706299, | |
| "eval_xsum-pairs_runtime": 2.254, | |
| "eval_xsum-pairs_samples_per_second": 66.548, | |
| "eval_xsum-pairs_steps_per_second": 2.218, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_compression-pairs_loss": 0.8453781008720398, | |
| "eval_compression-pairs_runtime": 0.4401, | |
| "eval_compression-pairs_samples_per_second": 340.826, | |
| "eval_compression-pairs_steps_per_second": 11.361, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_sciq_pairs_loss": 8.014656066894531, | |
| "eval_sciq_pairs_runtime": 7.0707, | |
| "eval_sciq_pairs_samples_per_second": 21.214, | |
| "eval_sciq_pairs_steps_per_second": 0.707, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_qasc_pairs_loss": 6.9316277503967285, | |
| "eval_qasc_pairs_runtime": 2.0338, | |
| "eval_qasc_pairs_samples_per_second": 73.752, | |
| "eval_qasc_pairs_steps_per_second": 2.458, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_openbookqa_pairs_loss": 4.21690034866333, | |
| "eval_openbookqa_pairs_runtime": 0.918, | |
| "eval_openbookqa_pairs_samples_per_second": 112.202, | |
| "eval_openbookqa_pairs_steps_per_second": 4.357, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_msmarco_pairs_loss": 3.0209598541259766, | |
| "eval_msmarco_pairs_runtime": 2.7749, | |
| "eval_msmarco_pairs_samples_per_second": 54.056, | |
| "eval_msmarco_pairs_steps_per_second": 1.802, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_nq_pairs_loss": 2.956088066101074, | |
| "eval_nq_pairs_runtime": 5.0024, | |
| "eval_nq_pairs_samples_per_second": 29.986, | |
| "eval_nq_pairs_steps_per_second": 1.0, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_trivia_pairs_loss": 3.17364501953125, | |
| "eval_trivia_pairs_runtime": 9.4856, | |
| "eval_trivia_pairs_samples_per_second": 15.813, | |
| "eval_trivia_pairs_steps_per_second": 0.527, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_quora_pairs_loss": 0.763593852519989, | |
| "eval_quora_pairs_runtime": 1.1441, | |
| "eval_quora_pairs_samples_per_second": 131.104, | |
| "eval_quora_pairs_steps_per_second": 4.37, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.625638406537283, | |
| "eval_gooaq_pairs_loss": 2.3524909019470215, | |
| "eval_gooaq_pairs_runtime": 2.0161, | |
| "eval_gooaq_pairs_samples_per_second": 74.4, | |
| "eval_gooaq_pairs_steps_per_second": 2.48, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.6506639427987743, | |
| "grad_norm": 31.163997650146484, | |
| "learning_rate": 1.2992849846782432e-05, | |
| "loss": 2.7808, | |
| "step": 3822 | |
| }, | |
| { | |
| "epoch": 0.6756894790602656, | |
| "grad_norm": 14.883658409118652, | |
| "learning_rate": 1.3493360572012258e-05, | |
| "loss": 2.5687, | |
| "step": 3969 | |
| }, | |
| { | |
| "epoch": 0.7007150153217568, | |
| "grad_norm": 5.874042987823486, | |
| "learning_rate": 1.3993871297242083e-05, | |
| "loss": 2.3034, | |
| "step": 4116 | |
| }, | |
| { | |
| "epoch": 0.7257405515832482, | |
| "grad_norm": 31.464054107666016, | |
| "learning_rate": 1.4494382022471912e-05, | |
| "loss": 2.4412, | |
| "step": 4263 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "grad_norm": 16.43915367126465, | |
| "learning_rate": 1.4994892747701737e-05, | |
| "loss": 2.3293, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_nli-pairs_loss": 2.3226094245910645, | |
| "eval_nli-pairs_runtime": 4.113, | |
| "eval_nli-pairs_samples_per_second": 36.47, | |
| "eval_nli-pairs_steps_per_second": 1.216, | |
| "eval_sts-test_pearson_cosine": 0.7356971966139032, | |
| "eval_sts-test_pearson_dot": 0.6150809513049869, | |
| "eval_sts-test_pearson_euclidean": 0.7330733579988641, | |
| "eval_sts-test_pearson_manhattan": 0.7423412248131348, | |
| "eval_sts-test_pearson_max": 0.7423412248131348, | |
| "eval_sts-test_spearman_cosine": 0.7121899723082045, | |
| "eval_sts-test_spearman_dot": 0.5926505936679538, | |
| "eval_sts-test_spearman_euclidean": 0.7130179905407037, | |
| "eval_sts-test_spearman_manhattan": 0.7227257562995023, | |
| "eval_sts-test_spearman_max": 0.7227257562995023, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_vitaminc-pairs_loss": 1.7956713438034058, | |
| "eval_vitaminc-pairs_runtime": 2.174, | |
| "eval_vitaminc-pairs_samples_per_second": 68.996, | |
| "eval_vitaminc-pairs_steps_per_second": 2.3, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_qnli-contrastive_loss": 1.0078614950180054, | |
| "eval_qnli-contrastive_runtime": 0.4874, | |
| "eval_qnli-contrastive_samples_per_second": 307.763, | |
| "eval_qnli-contrastive_steps_per_second": 10.259, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_scitail-pairs-qa_loss": 0.36971578001976013, | |
| "eval_scitail-pairs-qa_runtime": 1.164, | |
| "eval_scitail-pairs-qa_samples_per_second": 128.863, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.295, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_scitail-pairs-pos_loss": 1.0497769117355347, | |
| "eval_scitail-pairs-pos_runtime": 2.1205, | |
| "eval_scitail-pairs-pos_samples_per_second": 70.74, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.358, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_xsum-pairs_loss": 1.1691261529922485, | |
| "eval_xsum-pairs_runtime": 2.259, | |
| "eval_xsum-pairs_samples_per_second": 66.401, | |
| "eval_xsum-pairs_steps_per_second": 2.213, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_compression-pairs_loss": 0.5027483105659485, | |
| "eval_compression-pairs_runtime": 0.4403, | |
| "eval_compression-pairs_samples_per_second": 340.682, | |
| "eval_compression-pairs_steps_per_second": 11.356, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_sciq_pairs_loss": 7.823739528656006, | |
| "eval_sciq_pairs_runtime": 7.0738, | |
| "eval_sciq_pairs_samples_per_second": 21.205, | |
| "eval_sciq_pairs_steps_per_second": 0.707, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_qasc_pairs_loss": 6.404655933380127, | |
| "eval_qasc_pairs_runtime": 2.0346, | |
| "eval_qasc_pairs_samples_per_second": 73.723, | |
| "eval_qasc_pairs_steps_per_second": 2.457, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_openbookqa_pairs_loss": 3.857389211654663, | |
| "eval_openbookqa_pairs_runtime": 0.8544, | |
| "eval_openbookqa_pairs_samples_per_second": 120.547, | |
| "eval_openbookqa_pairs_steps_per_second": 4.681, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_msmarco_pairs_loss": 2.7028510570526123, | |
| "eval_msmarco_pairs_runtime": 2.7448, | |
| "eval_msmarco_pairs_samples_per_second": 54.649, | |
| "eval_msmarco_pairs_steps_per_second": 1.822, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_nq_pairs_loss": 2.679351329803467, | |
| "eval_nq_pairs_runtime": 5.067, | |
| "eval_nq_pairs_samples_per_second": 29.603, | |
| "eval_nq_pairs_steps_per_second": 0.987, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_trivia_pairs_loss": 2.8798065185546875, | |
| "eval_trivia_pairs_runtime": 9.5449, | |
| "eval_trivia_pairs_samples_per_second": 15.715, | |
| "eval_trivia_pairs_steps_per_second": 0.524, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_quora_pairs_loss": 0.6825175285339355, | |
| "eval_quora_pairs_runtime": 1.1431, | |
| "eval_quora_pairs_samples_per_second": 131.221, | |
| "eval_quora_pairs_steps_per_second": 4.374, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7507660878447395, | |
| "eval_gooaq_pairs_loss": 2.0472166538238525, | |
| "eval_gooaq_pairs_runtime": 2.0218, | |
| "eval_gooaq_pairs_samples_per_second": 74.191, | |
| "eval_gooaq_pairs_steps_per_second": 2.473, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.7757916241062308, | |
| "grad_norm": 4.2425055503845215, | |
| "learning_rate": 1.5495403472931565e-05, | |
| "loss": 2.3651, | |
| "step": 4557 | |
| }, | |
| { | |
| "epoch": 0.8008171603677222, | |
| "grad_norm": 22.42776107788086, | |
| "learning_rate": 1.5995914198161388e-05, | |
| "loss": 2.6296, | |
| "step": 4704 | |
| }, | |
| { | |
| "epoch": 0.8258426966292135, | |
| "grad_norm": 21.169517517089844, | |
| "learning_rate": 1.6496424923391215e-05, | |
| "loss": 2.2108, | |
| "step": 4851 | |
| }, | |
| { | |
| "epoch": 0.8508682328907048, | |
| "grad_norm": 23.326181411743164, | |
| "learning_rate": 1.699693564862104e-05, | |
| "loss": 2.1852, | |
| "step": 4998 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "grad_norm": 24.574176788330078, | |
| "learning_rate": 1.7497446373850868e-05, | |
| "loss": 2.2944, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_nli-pairs_loss": 2.0634915828704834, | |
| "eval_nli-pairs_runtime": 4.0019, | |
| "eval_nli-pairs_samples_per_second": 37.482, | |
| "eval_nli-pairs_steps_per_second": 1.249, | |
| "eval_sts-test_pearson_cosine": 0.7466390532977636, | |
| "eval_sts-test_pearson_dot": 0.612259458274589, | |
| "eval_sts-test_pearson_euclidean": 0.7432536346376271, | |
| "eval_sts-test_pearson_manhattan": 0.7500490179501229, | |
| "eval_sts-test_pearson_max": 0.7500490179501229, | |
| "eval_sts-test_spearman_cosine": 0.728273260456201, | |
| "eval_sts-test_spearman_dot": 0.5960115087190596, | |
| "eval_sts-test_spearman_euclidean": 0.7272394395622148, | |
| "eval_sts-test_spearman_manhattan": 0.7334149564445704, | |
| "eval_sts-test_spearman_max": 0.7334149564445704, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_vitaminc-pairs_loss": 1.638654112815857, | |
| "eval_vitaminc-pairs_runtime": 2.1637, | |
| "eval_vitaminc-pairs_samples_per_second": 69.327, | |
| "eval_vitaminc-pairs_steps_per_second": 2.311, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_qnli-contrastive_loss": 0.9639705419540405, | |
| "eval_qnli-contrastive_runtime": 0.4889, | |
| "eval_qnli-contrastive_samples_per_second": 306.825, | |
| "eval_qnli-contrastive_steps_per_second": 10.228, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_scitail-pairs-qa_loss": 0.31595128774642944, | |
| "eval_scitail-pairs-qa_runtime": 1.1467, | |
| "eval_scitail-pairs-qa_samples_per_second": 130.806, | |
| "eval_scitail-pairs-qa_steps_per_second": 4.36, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_scitail-pairs-pos_loss": 0.9187478423118591, | |
| "eval_scitail-pairs-pos_runtime": 2.1273, | |
| "eval_scitail-pairs-pos_samples_per_second": 70.512, | |
| "eval_scitail-pairs-pos_steps_per_second": 2.35, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_xsum-pairs_loss": 1.060194492340088, | |
| "eval_xsum-pairs_runtime": 2.2836, | |
| "eval_xsum-pairs_samples_per_second": 65.686, | |
| "eval_xsum-pairs_steps_per_second": 2.19, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_compression-pairs_loss": 0.41078585386276245, | |
| "eval_compression-pairs_runtime": 0.4434, | |
| "eval_compression-pairs_samples_per_second": 338.276, | |
| "eval_compression-pairs_steps_per_second": 11.276, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_sciq_pairs_loss": 7.577760696411133, | |
| "eval_sciq_pairs_runtime": 7.1025, | |
| "eval_sciq_pairs_samples_per_second": 21.119, | |
| "eval_sciq_pairs_steps_per_second": 0.704, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_qasc_pairs_loss": 6.353766918182373, | |
| "eval_qasc_pairs_runtime": 2.0113, | |
| "eval_qasc_pairs_samples_per_second": 74.58, | |
| "eval_qasc_pairs_steps_per_second": 2.486, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_openbookqa_pairs_loss": 3.7140932083129883, | |
| "eval_openbookqa_pairs_runtime": 0.8529, | |
| "eval_openbookqa_pairs_samples_per_second": 120.762, | |
| "eval_openbookqa_pairs_steps_per_second": 4.69, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_msmarco_pairs_loss": 2.3862576484680176, | |
| "eval_msmarco_pairs_runtime": 2.8953, | |
| "eval_msmarco_pairs_samples_per_second": 51.808, | |
| "eval_msmarco_pairs_steps_per_second": 1.727, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_nq_pairs_loss": 2.3543190956115723, | |
| "eval_nq_pairs_runtime": 5.0048, | |
| "eval_nq_pairs_samples_per_second": 29.971, | |
| "eval_nq_pairs_steps_per_second": 0.999, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_trivia_pairs_loss": 2.494807481765747, | |
| "eval_trivia_pairs_runtime": 9.5513, | |
| "eval_trivia_pairs_samples_per_second": 15.705, | |
| "eval_trivia_pairs_steps_per_second": 0.523, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_quora_pairs_loss": 0.6137441992759705, | |
| "eval_quora_pairs_runtime": 1.1541, | |
| "eval_quora_pairs_samples_per_second": 129.967, | |
| "eval_quora_pairs_steps_per_second": 4.332, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.8758937691521961, | |
| "eval_gooaq_pairs_loss": 1.8279658555984497, | |
| "eval_gooaq_pairs_runtime": 2.0951, | |
| "eval_gooaq_pairs_samples_per_second": 71.595, | |
| "eval_gooaq_pairs_steps_per_second": 2.387, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 0.9009193054136875, | |
| "grad_norm": 10.590804100036621, | |
| "learning_rate": 1.7997957099080695e-05, | |
| "loss": 2.2133, | |
| "step": 5292 | |
| }, | |
| { | |
| "epoch": 0.9259448416751788, | |
| "grad_norm": 18.527711868286133, | |
| "learning_rate": 1.849846782431052e-05, | |
| "loss": 2.2255, | |
| "step": 5439 | |
| }, | |
| { | |
| "epoch": 0.95097037793667, | |
| "grad_norm": 2.617710828781128, | |
| "learning_rate": 1.8995573714674838e-05, | |
| "loss": 2.3502, | |
| "step": 5586 | |
| }, | |
| { | |
| "epoch": 0.9759959141981613, | |
| "grad_norm": 19.551551818847656, | |
| "learning_rate": 1.9496084439904668e-05, | |
| "loss": 1.8964, | |
| "step": 5733 | |
| } | |
| ], | |
| "logging_steps": 147, | |
| "max_steps": 29370, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 2937, | |
| "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": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |