| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.8615040953090096, | |
| "eval_steps": 500, | |
| "global_step": 2500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.98758997269794e-05, | |
| "loss": 1.8076, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.97517994539588e-05, | |
| "loss": 1.792, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.96276991809382e-05, | |
| "loss": 1.7935, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.95035989079176e-05, | |
| "loss": 1.7432, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9379498634897e-05, | |
| "loss": 1.7369, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.92553983618764e-05, | |
| "loss": 1.7506, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.91312980888558e-05, | |
| "loss": 1.6795, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.90071978158352e-05, | |
| "loss": 1.721, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.88830975428146e-05, | |
| "loss": 1.6954, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.875899726979399e-05, | |
| "loss": 1.6715, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.86348969967734e-05, | |
| "loss": 1.6799, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8510796723752796e-05, | |
| "loss": 1.7203, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8386696450732196e-05, | |
| "loss": 1.6643, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8262596177711595e-05, | |
| "loss": 1.6655, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8138495904690995e-05, | |
| "loss": 1.681, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8014395631670394e-05, | |
| "loss": 1.7232, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.789029535864979e-05, | |
| "loss": 1.6658, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7766195085629186e-05, | |
| "loss": 1.6965, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.764209481260859e-05, | |
| "loss": 1.6745, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.751799453958799e-05, | |
| "loss": 1.6881, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.739389426656739e-05, | |
| "loss": 1.6899, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.726979399354679e-05, | |
| "loss": 1.6825, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7145693720526184e-05, | |
| "loss": 1.6894, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.702159344750558e-05, | |
| "loss": 1.7016, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.689749317448498e-05, | |
| "loss": 1.6556, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.677339290146439e-05, | |
| "loss": 1.715, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.664929262844379e-05, | |
| "loss": 1.6415, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.652519235542319e-05, | |
| "loss": 1.6986, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.640109208240258e-05, | |
| "loss": 1.6766, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.627699180938198e-05, | |
| "loss": 1.7016, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.615289153636138e-05, | |
| "loss": 1.698, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.602879126334078e-05, | |
| "loss": 1.6692, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.590469099032018e-05, | |
| "loss": 1.6386, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5780590717299585e-05, | |
| "loss": 1.7369, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.5656490444278984e-05, | |
| "loss": 1.6997, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.553239017125838e-05, | |
| "loss": 1.6983, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5408289898237776e-05, | |
| "loss": 1.6702, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5284189625217176e-05, | |
| "loss": 1.6901, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5160089352196575e-05, | |
| "loss": 1.6916, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5035989079175975e-05, | |
| "loss": 1.682, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.491188880615538e-05, | |
| "loss": 1.6756, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4787788533134774e-05, | |
| "loss": 1.6632, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.466368826011417e-05, | |
| "loss": 1.67, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.453958798709357e-05, | |
| "loss": 1.6581, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.441548771407297e-05, | |
| "loss": 1.7106, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.429138744105237e-05, | |
| "loss": 1.6757, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.416728716803177e-05, | |
| "loss": 1.6315, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.404318689501117e-05, | |
| "loss": 1.66, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.391908662199057e-05, | |
| "loss": 1.6966, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.379498634896997e-05, | |
| "loss": 1.6642, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.367088607594937e-05, | |
| "loss": 1.6882, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.354678580292877e-05, | |
| "loss": 1.6741, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.342268552990817e-05, | |
| "loss": 1.6859, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.329858525688757e-05, | |
| "loss": 1.6925, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.317448498386697e-05, | |
| "loss": 1.6748, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.3050384710846366e-05, | |
| "loss": 1.674, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.2926284437825766e-05, | |
| "loss": 1.6723, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.2802184164805165e-05, | |
| "loss": 1.688, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.2678083891784565e-05, | |
| "loss": 1.6585, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2553983618763964e-05, | |
| "loss": 1.6902, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2429883345743364e-05, | |
| "loss": 1.6565, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2305783072722757e-05, | |
| "loss": 1.6897, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.218168279970216e-05, | |
| "loss": 1.6698, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.205758252668156e-05, | |
| "loss": 1.6712, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.193348225366096e-05, | |
| "loss": 1.6653, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.180938198064036e-05, | |
| "loss": 1.683, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.168528170761976e-05, | |
| "loss": 1.6498, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1561181434599153e-05, | |
| "loss": 1.6939, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.143708116157855e-05, | |
| "loss": 1.7123, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.131298088855796e-05, | |
| "loss": 1.669, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.118888061553736e-05, | |
| "loss": 1.7012, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.106478034251676e-05, | |
| "loss": 1.668, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.094068006949616e-05, | |
| "loss": 1.6845, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.081657979647556e-05, | |
| "loss": 1.6371, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.069247952345495e-05, | |
| "loss": 1.6871, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.056837925043435e-05, | |
| "loss": 1.6605, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.044427897741375e-05, | |
| "loss": 1.6838, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0320178704393155e-05, | |
| "loss": 1.6586, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.0196078431372555e-05, | |
| "loss": 1.7265, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0071978158351954e-05, | |
| "loss": 1.6616, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.994787788533135e-05, | |
| "loss": 1.6445, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9823777612310746e-05, | |
| "loss": 1.6916, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9699677339290146e-05, | |
| "loss": 1.6986, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9575577066269545e-05, | |
| "loss": 1.6776, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.945147679324895e-05, | |
| "loss": 1.6629, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.932737652022835e-05, | |
| "loss": 1.6789, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9203276247207744e-05, | |
| "loss": 1.6432, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.907917597418714e-05, | |
| "loss": 1.6934, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.895507570116654e-05, | |
| "loss": 1.7064, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.883097542814594e-05, | |
| "loss": 1.6626, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.870687515512534e-05, | |
| "loss": 1.6482, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.858277488210474e-05, | |
| "loss": 1.674, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.845867460908415e-05, | |
| "loss": 1.6474, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.833457433606354e-05, | |
| "loss": 1.6719, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.821047406304294e-05, | |
| "loss": 1.6694, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.808637379002234e-05, | |
| "loss": 1.6876, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.796227351700174e-05, | |
| "loss": 1.6902, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.783817324398114e-05, | |
| "loss": 1.67, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.771407297096054e-05, | |
| "loss": 1.6488, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.758997269793994e-05, | |
| "loss": 1.6153, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7465872424919336e-05, | |
| "loss": 1.6346, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.7341772151898736e-05, | |
| "loss": 1.7009, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7217671878878135e-05, | |
| "loss": 1.6873, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7093571605857535e-05, | |
| "loss": 1.6742, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.6969471332836934e-05, | |
| "loss": 1.645, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6845371059816334e-05, | |
| "loss": 1.7089, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.672127078679573e-05, | |
| "loss": 1.6473, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.659717051377513e-05, | |
| "loss": 1.6846, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.647307024075453e-05, | |
| "loss": 1.7159, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.634896996773393e-05, | |
| "loss": 1.6727, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.622486969471333e-05, | |
| "loss": 1.6722, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.610076942169273e-05, | |
| "loss": 1.6785, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.597666914867212e-05, | |
| "loss": 1.68, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.585256887565153e-05, | |
| "loss": 1.6698, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.572846860263093e-05, | |
| "loss": 1.7116, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.560436832961033e-05, | |
| "loss": 1.6635, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.548026805658973e-05, | |
| "loss": 1.6821, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.535616778356913e-05, | |
| "loss": 1.6669, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.523206751054853e-05, | |
| "loss": 1.6733, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.510796723752792e-05, | |
| "loss": 1.66, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.498386696450732e-05, | |
| "loss": 1.6447, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.4859766691486725e-05, | |
| "loss": 1.5991, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4735666418466125e-05, | |
| "loss": 1.6811, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4611566145445524e-05, | |
| "loss": 1.7129, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4487465872424924e-05, | |
| "loss": 1.6621, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.4363365599404317e-05, | |
| "loss": 1.7028, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4239265326383716e-05, | |
| "loss": 1.6428, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4115165053363115e-05, | |
| "loss": 1.6931, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.399106478034252e-05, | |
| "loss": 1.6831, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.386696450732192e-05, | |
| "loss": 1.6676, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.374286423430132e-05, | |
| "loss": 1.6861, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.3618763961280713e-05, | |
| "loss": 1.6731, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.349466368826011e-05, | |
| "loss": 1.6459, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.337056341523951e-05, | |
| "loss": 1.7026, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.324646314221891e-05, | |
| "loss": 1.6455, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.312236286919831e-05, | |
| "loss": 1.6502, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.299826259617772e-05, | |
| "loss": 1.6802, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.287416232315712e-05, | |
| "loss": 1.6756, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.275006205013651e-05, | |
| "loss": 1.615, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.262596177711591e-05, | |
| "loss": 1.6637, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.250186150409531e-05, | |
| "loss": 1.6727, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.237776123107471e-05, | |
| "loss": 1.6169, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.225366095805411e-05, | |
| "loss": 1.6792, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2129560685033514e-05, | |
| "loss": 1.6745, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.200546041201291e-05, | |
| "loss": 1.6932, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1881360138992306e-05, | |
| "loss": 1.6967, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1757259865971706e-05, | |
| "loss": 1.6416, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1633159592951105e-05, | |
| "loss": 1.6618, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.1509059319930505e-05, | |
| "loss": 1.6343, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1384959046909904e-05, | |
| "loss": 1.6633, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1260858773889304e-05, | |
| "loss": 1.7296, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.11367585008687e-05, | |
| "loss": 1.6811, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.10126582278481e-05, | |
| "loss": 1.6472, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.08885579548275e-05, | |
| "loss": 1.6914, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.07644576818069e-05, | |
| "loss": 1.6557, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.06403574087863e-05, | |
| "loss": 1.6505, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0516257135765704e-05, | |
| "loss": 1.6881, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0392156862745097e-05, | |
| "loss": 1.6816, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0268056589724496e-05, | |
| "loss": 1.6364, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.01439563167039e-05, | |
| "loss": 1.6724, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.00198560436833e-05, | |
| "loss": 1.6698, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9895755770662698e-05, | |
| "loss": 1.6864, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9771655497642097e-05, | |
| "loss": 1.6614, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9647555224621497e-05, | |
| "loss": 1.6181, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9523454951600893e-05, | |
| "loss": 1.697, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9399354678580292e-05, | |
| "loss": 1.6462, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9275254405559692e-05, | |
| "loss": 1.6583, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9151154132539095e-05, | |
| "loss": 1.6781, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.9027053859518494e-05, | |
| "loss": 1.6672, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8902953586497894e-05, | |
| "loss": 1.6406, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.877885331347729e-05, | |
| "loss": 1.7013, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.865475304045669e-05, | |
| "loss": 1.6636, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.853065276743609e-05, | |
| "loss": 1.6721, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8406552494415488e-05, | |
| "loss": 1.6605, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.828245222139489e-05, | |
| "loss": 1.6329, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.815835194837429e-05, | |
| "loss": 1.6444, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8034251675353683e-05, | |
| "loss": 1.6278, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7910151402333086e-05, | |
| "loss": 1.6724, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7786051129312486e-05, | |
| "loss": 1.7456, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7661950856291885e-05, | |
| "loss": 1.6874, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7537850583271285e-05, | |
| "loss": 1.6249, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7413750310250684e-05, | |
| "loss": 1.6814, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7289650037230087e-05, | |
| "loss": 1.6465, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.716554976420948e-05, | |
| "loss": 1.6471, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.704144949118888e-05, | |
| "loss": 1.6207, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6917349218168282e-05, | |
| "loss": 1.6443, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.679324894514768e-05, | |
| "loss": 1.6533, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.666914867212708e-05, | |
| "loss": 1.7088, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.654504839910648e-05, | |
| "loss": 1.6959, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6420948126085876e-05, | |
| "loss": 1.7019, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6296847853065276e-05, | |
| "loss": 1.6753, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6172747580044675e-05, | |
| "loss": 1.7044, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.6048647307024078e-05, | |
| "loss": 1.6777, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.5924547034003478e-05, | |
| "loss": 1.6419, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5800446760982877e-05, | |
| "loss": 1.6893, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5676346487962277e-05, | |
| "loss": 1.6555, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5552246214941673e-05, | |
| "loss": 1.6814, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5428145941921072e-05, | |
| "loss": 1.7269, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5304045668900472e-05, | |
| "loss": 1.6449, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.517994539587987e-05, | |
| "loss": 1.6843, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.5055845122859274e-05, | |
| "loss": 1.6782, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.493174484983867e-05, | |
| "loss": 1.7249, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.480764457681807e-05, | |
| "loss": 1.6899, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.468354430379747e-05, | |
| "loss": 1.6383, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.455944403077687e-05, | |
| "loss": 1.6221, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4435343757756268e-05, | |
| "loss": 1.6997, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4311243484735668e-05, | |
| "loss": 1.6394, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4187143211715067e-05, | |
| "loss": 1.6809, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4063042938694467e-05, | |
| "loss": 1.6683, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3938942665673866e-05, | |
| "loss": 1.6625, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3814842392653262e-05, | |
| "loss": 1.6692, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3690742119632665e-05, | |
| "loss": 1.6996, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3566641846612065e-05, | |
| "loss": 1.6653, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3442541573591464e-05, | |
| "loss": 1.6903, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3318441300570863e-05, | |
| "loss": 1.6639, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3194341027550263e-05, | |
| "loss": 1.6501, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.3070240754529662e-05, | |
| "loss": 1.6298, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.294614048150906e-05, | |
| "loss": 1.6438, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.282204020848846e-05, | |
| "loss": 1.6265, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.269793993546786e-05, | |
| "loss": 1.6856, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2573839662447257e-05, | |
| "loss": 1.6824, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2449739389426656e-05, | |
| "loss": 1.6899, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.232563911640606e-05, | |
| "loss": 1.6264, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2201538843385455e-05, | |
| "loss": 1.6794, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.2077438570364855e-05, | |
| "loss": 1.6568, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.1953338297344254e-05, | |
| "loss": 1.6893, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1829238024323654e-05, | |
| "loss": 1.696, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1705137751303053e-05, | |
| "loss": 1.6312, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1581037478282453e-05, | |
| "loss": 1.6189, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1456937205261852e-05, | |
| "loss": 1.6403, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1332836932241252e-05, | |
| "loss": 1.6639, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.120873665922065e-05, | |
| "loss": 1.7026, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.108463638620005e-05, | |
| "loss": 1.6919, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.096053611317945e-05, | |
| "loss": 1.6642, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.083643584015885e-05, | |
| "loss": 1.6531, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.071233556713825e-05, | |
| "loss": 1.6271, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.058823529411765e-05, | |
| "loss": 1.6252, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0464135021097048e-05, | |
| "loss": 1.6681, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0340034748076448e-05, | |
| "loss": 1.6417, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0215934475055844e-05, | |
| "loss": 1.6386, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0091834202035247e-05, | |
| "loss": 1.6616, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.9967733929014646e-05, | |
| "loss": 1.629, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9843633655994042e-05, | |
| "loss": 1.711, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.971953338297344e-05, | |
| "loss": 1.6446, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9595433109952844e-05, | |
| "loss": 1.6744, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9471332836932244e-05, | |
| "loss": 1.6503, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.934723256391164e-05, | |
| "loss": 1.6351, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.922313229089104e-05, | |
| "loss": 1.6266, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9099032017870442e-05, | |
| "loss": 1.6997, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.897493174484984e-05, | |
| "loss": 1.62, | |
| "step": 2500 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 4029, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 1.5800620548096e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |