| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.844791538556143, | |
| "global_step": 1500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8e-08, | |
| "loss": 1.1212, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.8e-07, | |
| "loss": 1.1134, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.1174, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 6.800000000000001e-07, | |
| "loss": 1.1076, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 8.8e-07, | |
| "loss": 1.1042, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.08e-06, | |
| "loss": 1.1135, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.28e-06, | |
| "loss": 1.1096, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.48e-06, | |
| "loss": 1.1024, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.6800000000000002e-06, | |
| "loss": 1.094, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.8600000000000002e-06, | |
| "loss": 1.1022, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.06e-06, | |
| "loss": 1.0981, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.2600000000000004e-06, | |
| "loss": 1.0915, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.46e-06, | |
| "loss": 1.0812, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.6600000000000004e-06, | |
| "loss": 1.0832, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.86e-06, | |
| "loss": 1.0731, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 1.0723, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.2600000000000006e-06, | |
| "loss": 1.0505, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.46e-06, | |
| "loss": 1.0352, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.66e-06, | |
| "loss": 0.9904, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.86e-06, | |
| "loss": 0.9888, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.9637, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.26e-06, | |
| "loss": 0.934, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.9193, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.66e-06, | |
| "loss": 0.877, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.86e-06, | |
| "loss": 0.8693, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.862, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 5.2600000000000005e-06, | |
| "loss": 0.7961, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.82, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 5.66e-06, | |
| "loss": 0.7964, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.86e-06, | |
| "loss": 0.7721, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.7876, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.26e-06, | |
| "loss": 0.7579, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.7519, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.741, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 0.7496, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.7587, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.260000000000001e-06, | |
| "loss": 0.7373, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.695, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.717, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 0.6464, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.7189, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.26e-06, | |
| "loss": 0.7103, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.7119, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.6992, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.860000000000002e-06, | |
| "loss": 0.7076, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.6979, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.260000000000001e-06, | |
| "loss": 0.7038, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.6698, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.6826, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.86e-06, | |
| "loss": 0.7071, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "eval_oasst_export_w_label_accuracy": 0.75, | |
| "eval_oasst_export_w_label_kendalltau": 0.4063736263736262, | |
| "eval_oasst_export_w_label_loss": 0.67431640625, | |
| "eval_oasst_export_w_label_neg_score": 0.7998046875, | |
| "eval_oasst_export_w_label_pos_score": 1.5751953125, | |
| "eval_oasst_export_w_label_runtime": 150.7481, | |
| "eval_oasst_export_w_label_samples_per_second": 12.073, | |
| "eval_oasst_export_w_label_score_diff": 0.775390625, | |
| "eval_oasst_export_w_label_steps_per_second": 6.037, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.98452810727179e-06, | |
| "loss": 0.7206, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.932955131511089e-06, | |
| "loss": 0.688, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.881382155750388e-06, | |
| "loss": 0.6715, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.829809179989687e-06, | |
| "loss": 0.6948, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.778236204228985e-06, | |
| "loss": 0.6869, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.726663228468284e-06, | |
| "loss": 0.6541, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.675090252707581e-06, | |
| "loss": 0.7238, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.62351727694688e-06, | |
| "loss": 0.6599, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.571944301186179e-06, | |
| "loss": 0.6549, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.520371325425478e-06, | |
| "loss": 0.6855, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.468798349664777e-06, | |
| "loss": 0.6486, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.417225373904075e-06, | |
| "loss": 0.64, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.365652398143374e-06, | |
| "loss": 0.7051, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.314079422382673e-06, | |
| "loss": 0.6982, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.26250644662197e-06, | |
| "loss": 0.6508, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.210933470861269e-06, | |
| "loss": 0.661, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.159360495100568e-06, | |
| "loss": 0.6907, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.107787519339866e-06, | |
| "loss": 0.6872, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.056214543579165e-06, | |
| "loss": 0.6979, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.004641567818464e-06, | |
| "loss": 0.6683, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.953068592057763e-06, | |
| "loss": 0.6973, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.901495616297062e-06, | |
| "loss": 0.6384, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.849922640536359e-06, | |
| "loss": 0.6686, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.798349664775658e-06, | |
| "loss": 0.6709, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.746776689014956e-06, | |
| "loss": 0.6693, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.695203713254255e-06, | |
| "loss": 0.6405, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.643630737493554e-06, | |
| "loss": 0.6937, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.592057761732853e-06, | |
| "loss": 0.6322, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.540484785972152e-06, | |
| "loss": 0.6862, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.48891181021145e-06, | |
| "loss": 0.6507, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.437338834450749e-06, | |
| "loss": 0.6452, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.385765858690046e-06, | |
| "loss": 0.6387, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.334192882929347e-06, | |
| "loss": 0.6422, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.282619907168644e-06, | |
| "loss": 0.6492, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.231046931407943e-06, | |
| "loss": 0.6533, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.179473955647241e-06, | |
| "loss": 0.6747, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.12790097988654e-06, | |
| "loss": 0.67, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.076328004125839e-06, | |
| "loss": 0.6337, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.024755028365138e-06, | |
| "loss": 0.6612, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.973182052604435e-06, | |
| "loss": 0.6354, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.921609076843735e-06, | |
| "loss": 0.6236, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.870036101083033e-06, | |
| "loss": 0.6394, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.818463125322331e-06, | |
| "loss": 0.631, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.76689014956163e-06, | |
| "loss": 0.6378, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.715317173800929e-06, | |
| "loss": 0.6328, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.663744198040228e-06, | |
| "loss": 0.6036, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.6121712222795265e-06, | |
| "loss": 0.6193, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.5605982465188245e-06, | |
| "loss": 0.6667, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.509025270758123e-06, | |
| "loss": 0.6434, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.457452294997421e-06, | |
| "loss": 0.5941, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "eval_oasst_export_w_label_accuracy": 0.7576923076923077, | |
| "eval_oasst_export_w_label_kendalltau": 0.3999267399267405, | |
| "eval_oasst_export_w_label_loss": 0.6787109375, | |
| "eval_oasst_export_w_label_neg_score": -1.3857421875, | |
| "eval_oasst_export_w_label_pos_score": -0.15380859375, | |
| "eval_oasst_export_w_label_runtime": 150.4368, | |
| "eval_oasst_export_w_label_samples_per_second": 12.098, | |
| "eval_oasst_export_w_label_score_diff": 1.232421875, | |
| "eval_oasst_export_w_label_steps_per_second": 6.049, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.405879319236721e-06, | |
| "loss": 0.6572, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.354306343476019e-06, | |
| "loss": 0.6466, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.302733367715318e-06, | |
| "loss": 0.6361, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.251160391954616e-06, | |
| "loss": 0.6522, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.199587416193915e-06, | |
| "loss": 0.6014, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.148014440433214e-06, | |
| "loss": 0.593, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.096441464672512e-06, | |
| "loss": 0.643, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 7.044868488911812e-06, | |
| "loss": 0.6474, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.9932955131511096e-06, | |
| "loss": 0.6338, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.941722537390408e-06, | |
| "loss": 0.6268, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.890149561629706e-06, | |
| "loss": 0.6358, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.838576585869006e-06, | |
| "loss": 0.6247, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.787003610108304e-06, | |
| "loss": 0.5791, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.735430634347603e-06, | |
| "loss": 0.6074, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.683857658586901e-06, | |
| "loss": 0.61, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 6.6322846828262e-06, | |
| "loss": 0.6167, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 6.580711707065498e-06, | |
| "loss": 0.6437, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 6.529138731304797e-06, | |
| "loss": 0.6182, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.477565755544095e-06, | |
| "loss": 0.6184, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.425992779783395e-06, | |
| "loss": 0.6191, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.374419804022693e-06, | |
| "loss": 0.6136, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.322846828261991e-06, | |
| "loss": 0.6355, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.271273852501289e-06, | |
| "loss": 0.5557, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.219700876740589e-06, | |
| "loss": 0.6393, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.168127900979887e-06, | |
| "loss": 0.5874, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.116554925219186e-06, | |
| "loss": 0.6041, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.064981949458484e-06, | |
| "loss": 0.6061, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.013408973697783e-06, | |
| "loss": 0.5858, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 5.961835997937081e-06, | |
| "loss": 0.6646, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 5.91026302217638e-06, | |
| "loss": 0.6314, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.858690046415678e-06, | |
| "loss": 0.6332, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 5.807117070654978e-06, | |
| "loss": 0.613, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 5.755544094894276e-06, | |
| "loss": 0.6441, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 5.7039711191335744e-06, | |
| "loss": 0.6369, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 5.652398143372873e-06, | |
| "loss": 0.6157, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 5.600825167612172e-06, | |
| "loss": 0.658, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 5.549252191851471e-06, | |
| "loss": 0.621, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.497679216090769e-06, | |
| "loss": 0.6134, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.446106240330068e-06, | |
| "loss": 0.6128, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.394533264569366e-06, | |
| "loss": 0.6307, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 5.342960288808665e-06, | |
| "loss": 0.6131, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.291387313047963e-06, | |
| "loss": 0.5757, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.239814337287262e-06, | |
| "loss": 0.7221, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 5.188241361526561e-06, | |
| "loss": 0.6262, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.1366683857658595e-06, | |
| "loss": 0.6628, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.0850954100051575e-06, | |
| "loss": 0.6105, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 5.033522434244456e-06, | |
| "loss": 0.6422, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.981949458483755e-06, | |
| "loss": 0.6443, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.930376482723053e-06, | |
| "loss": 0.6067, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.878803506962353e-06, | |
| "loss": 0.6264, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "eval_oasst_export_w_label_accuracy": 0.7708791208791209, | |
| "eval_oasst_export_w_label_kendalltau": 0.42186813186813227, | |
| "eval_oasst_export_w_label_loss": 0.63671875, | |
| "eval_oasst_export_w_label_neg_score": -0.201904296875, | |
| "eval_oasst_export_w_label_pos_score": 1.01171875, | |
| "eval_oasst_export_w_label_runtime": 150.6985, | |
| "eval_oasst_export_w_label_samples_per_second": 12.077, | |
| "eval_oasst_export_w_label_score_diff": 1.2138671875, | |
| "eval_oasst_export_w_label_steps_per_second": 6.039, | |
| "step": 1500 | |
| } | |
| ], | |
| "max_steps": 2439, | |
| "num_train_epochs": 3, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |