| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.7474886236799176, |
| "global_step": 2000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.01, |
| "learning_rate": 1.40353507565337e-05, |
| "loss": 0.7119, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.39707015130674e-05, |
| "loss": 0.7003, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.39060522696011e-05, |
| "loss": 0.7064, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.3841403026134801e-05, |
| "loss": 0.7039, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.3776753782668502e-05, |
| "loss": 0.7, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.3712104539202202e-05, |
| "loss": 0.6913, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.3647455295735902e-05, |
| "loss": 0.6942, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.3582806052269601e-05, |
| "loss": 0.6986, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.3518156808803301e-05, |
| "loss": 0.6973, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.3453507565337e-05, |
| "loss": 0.6986, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.3388858321870702e-05, |
| "loss": 0.694, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.3324209078404402e-05, |
| "loss": 0.7043, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.3259559834938103e-05, |
| "loss": 0.7008, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.3194910591471803e-05, |
| "loss": 0.7012, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.3130261348005503e-05, |
| "loss": 0.6962, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.3065612104539202e-05, |
| "loss": 0.7042, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.3000962861072902e-05, |
| "loss": 0.6915, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.2936313617606603e-05, |
| "loss": 0.7044, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.2871664374140303e-05, |
| "loss": 0.6995, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.2807015130674003e-05, |
| "loss": 0.6963, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.2742365887207704e-05, |
| "loss": 0.7051, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.2677716643741404e-05, |
| "loss": 0.7008, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.2613067400275103e-05, |
| "loss": 0.6967, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.2548418156808803e-05, |
| "loss": 0.6978, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.2483768913342505e-05, |
| "loss": 0.7031, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.2419119669876204e-05, |
| "loss": 0.7012, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.2354470426409904e-05, |
| "loss": 0.6968, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.2289821182943604e-05, |
| "loss": 0.7052, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.2225171939477305e-05, |
| "loss": 0.6951, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.2160522696011005e-05, |
| "loss": 0.7078, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2095873452544704e-05, |
| "loss": 0.6983, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2031224209078406e-05, |
| "loss": 0.6925, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.1966574965612105e-05, |
| "loss": 0.7019, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1901925722145805e-05, |
| "loss": 0.6932, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1837276478679505e-05, |
| "loss": 0.699, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.1772627235213205e-05, |
| "loss": 0.6918, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.1707977991746906e-05, |
| "loss": 0.6999, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 1.1643328748280606e-05, |
| "loss": 0.6911, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 1.1578679504814307e-05, |
| "loss": 0.6997, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 1.1514030261348007e-05, |
| "loss": 0.6928, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 1.1449381017881706e-05, |
| "loss": 0.6917, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 1.1384731774415406e-05, |
| "loss": 0.6941, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 1.1320082530949106e-05, |
| "loss": 0.6985, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 1.1255433287482805e-05, |
| "loss": 0.6969, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 1.1190784044016507e-05, |
| "loss": 0.6848, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 1.1126134800550206e-05, |
| "loss": 0.6953, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 1.1061485557083908e-05, |
| "loss": 0.6946, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 1.0996836313617608e-05, |
| "loss": 0.6964, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 1.0932187070151307e-05, |
| "loss": 0.6868, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 1.0867537826685007e-05, |
| "loss": 0.6945, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 1.0802888583218707e-05, |
| "loss": 0.692, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 1.0738239339752406e-05, |
| "loss": 0.6914, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 1.0673590096286108e-05, |
| "loss": 0.6932, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 1.0608940852819809e-05, |
| "loss": 0.6939, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 1.0544291609353509e-05, |
| "loss": 0.6959, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 1.0479642365887208e-05, |
| "loss": 0.693, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 1.0414993122420908e-05, |
| "loss": 0.6946, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 1.0350343878954608e-05, |
| "loss": 0.6902, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.0285694635488307e-05, |
| "loss": 0.6912, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.0221045392022009e-05, |
| "loss": 0.7023, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.0156396148555709e-05, |
| "loss": 0.6944, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.009174690508941e-05, |
| "loss": 0.7002, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 1.002709766162311e-05, |
| "loss": 0.6895, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 9.96244841815681e-06, |
| "loss": 0.6933, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 9.897799174690509e-06, |
| "loss": 0.6882, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 9.833149931224209e-06, |
| "loss": 0.6901, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 9.76850068775791e-06, |
| "loss": 0.6918, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 9.70385144429161e-06, |
| "loss": 0.6953, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 9.63920220082531e-06, |
| "loss": 0.6919, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 9.57455295735901e-06, |
| "loss": 0.701, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 9.50990371389271e-06, |
| "loss": 0.6953, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 9.44525447042641e-06, |
| "loss": 0.6874, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 9.38060522696011e-06, |
| "loss": 0.6977, |
| "step": 730 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 9.31595598349381e-06, |
| "loss": 0.6777, |
| "step": 740 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 9.251306740027511e-06, |
| "loss": 0.7029, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 9.18665749656121e-06, |
| "loss": 0.69, |
| "step": 760 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 9.12200825309491e-06, |
| "loss": 0.7011, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 9.057359009628612e-06, |
| "loss": 0.691, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 8.992709766162311e-06, |
| "loss": 0.6957, |
| "step": 790 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 8.928060522696011e-06, |
| "loss": 0.6954, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 8.86341127922971e-06, |
| "loss": 0.7002, |
| "step": 810 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 8.798762035763412e-06, |
| "loss": 0.6882, |
| "step": 820 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 8.734112792297112e-06, |
| "loss": 0.6939, |
| "step": 830 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 8.669463548830812e-06, |
| "loss": 0.6968, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 8.604814305364511e-06, |
| "loss": 0.6938, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 8.540165061898213e-06, |
| "loss": 0.6882, |
| "step": 860 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 8.475515818431912e-06, |
| "loss": 0.6958, |
| "step": 870 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 8.410866574965612e-06, |
| "loss": 0.696, |
| "step": 880 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 8.346217331499313e-06, |
| "loss": 0.6878, |
| "step": 890 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 8.281568088033013e-06, |
| "loss": 0.7001, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 8.216918844566713e-06, |
| "loss": 0.6959, |
| "step": 910 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 8.152269601100412e-06, |
| "loss": 0.6886, |
| "step": 920 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 8.087620357634112e-06, |
| "loss": 0.7038, |
| "step": 930 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 8.022971114167813e-06, |
| "loss": 0.6999, |
| "step": 940 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 7.958321870701513e-06, |
| "loss": 0.6996, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 7.893672627235215e-06, |
| "loss": 0.6958, |
| "step": 960 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 7.829023383768914e-06, |
| "loss": 0.6939, |
| "step": 970 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 7.764374140302614e-06, |
| "loss": 0.6922, |
| "step": 980 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 7.699724896836314e-06, |
| "loss": 0.6954, |
| "step": 990 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 7.635075653370013e-06, |
| "loss": 0.6976, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 7.570426409903714e-06, |
| "loss": 0.6981, |
| "step": 1010 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 7.505777166437415e-06, |
| "loss": 0.6983, |
| "step": 1020 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 7.441127922971115e-06, |
| "loss": 0.691, |
| "step": 1030 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 7.3764786795048146e-06, |
| "loss": 0.6851, |
| "step": 1040 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 7.311829436038515e-06, |
| "loss": 0.6823, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 7.247180192572215e-06, |
| "loss": 0.6921, |
| "step": 1060 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 7.1825309491059145e-06, |
| "loss": 0.6895, |
| "step": 1070 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 7.117881705639615e-06, |
| "loss": 0.6966, |
| "step": 1080 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 7.053232462173315e-06, |
| "loss": 0.6908, |
| "step": 1090 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 6.988583218707015e-06, |
| "loss": 0.6833, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 6.923933975240715e-06, |
| "loss": 0.6916, |
| "step": 1110 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 6.859284731774416e-06, |
| "loss": 0.6958, |
| "step": 1120 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 6.794635488308116e-06, |
| "loss": 0.6974, |
| "step": 1130 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 6.729986244841816e-06, |
| "loss": 0.6889, |
| "step": 1140 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 6.665337001375516e-06, |
| "loss": 0.6871, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 6.600687757909217e-06, |
| "loss": 0.6987, |
| "step": 1160 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 6.536038514442916e-06, |
| "loss": 0.6899, |
| "step": 1170 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 6.471389270976616e-06, |
| "loss": 0.6975, |
| "step": 1180 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 6.406740027510317e-06, |
| "loss": 0.6931, |
| "step": 1190 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 6.342090784044016e-06, |
| "loss": 0.7096, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 6.277441540577717e-06, |
| "loss": 0.6917, |
| "step": 1210 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 6.212792297111417e-06, |
| "loss": 0.6978, |
| "step": 1220 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 6.148143053645117e-06, |
| "loss": 0.6972, |
| "step": 1230 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 6.083493810178817e-06, |
| "loss": 0.6813, |
| "step": 1240 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 6.018844566712517e-06, |
| "loss": 0.6972, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 5.954195323246218e-06, |
| "loss": 0.6895, |
| "step": 1260 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 5.8895460797799175e-06, |
| "loss": 0.6912, |
| "step": 1270 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 5.824896836313617e-06, |
| "loss": 0.6952, |
| "step": 1280 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 5.760247592847319e-06, |
| "loss": 0.6963, |
| "step": 1290 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 5.695598349381018e-06, |
| "loss": 0.6969, |
| "step": 1300 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 5.630949105914718e-06, |
| "loss": 0.7003, |
| "step": 1310 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 5.566299862448418e-06, |
| "loss": 0.697, |
| "step": 1320 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 5.501650618982119e-06, |
| "loss": 0.6929, |
| "step": 1330 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 5.437001375515819e-06, |
| "loss": 0.6892, |
| "step": 1340 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 5.372352132049518e-06, |
| "loss": 0.6805, |
| "step": 1350 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 5.307702888583219e-06, |
| "loss": 0.6887, |
| "step": 1360 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 5.2430536451169195e-06, |
| "loss": 0.6905, |
| "step": 1370 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 5.178404401650619e-06, |
| "loss": 0.6874, |
| "step": 1380 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 5.113755158184319e-06, |
| "loss": 0.6949, |
| "step": 1390 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 5.049105914718019e-06, |
| "loss": 0.6851, |
| "step": 1400 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 4.98445667125172e-06, |
| "loss": 0.6964, |
| "step": 1410 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 4.91980742778542e-06, |
| "loss": 0.688, |
| "step": 1420 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 4.85515818431912e-06, |
| "loss": 0.7013, |
| "step": 1430 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 4.79050894085282e-06, |
| "loss": 0.6907, |
| "step": 1440 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 4.72585969738652e-06, |
| "loss": 0.6969, |
| "step": 1450 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 4.66121045392022e-06, |
| "loss": 0.6897, |
| "step": 1460 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 4.596561210453921e-06, |
| "loss": 0.6924, |
| "step": 1470 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 4.53191196698762e-06, |
| "loss": 0.6923, |
| "step": 1480 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 4.467262723521321e-06, |
| "loss": 0.6911, |
| "step": 1490 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 4.4026134800550205e-06, |
| "loss": 0.6942, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 4.337964236588721e-06, |
| "loss": 0.684, |
| "step": 1510 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 4.273314993122421e-06, |
| "loss": 0.687, |
| "step": 1520 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 4.208665749656121e-06, |
| "loss": 0.6998, |
| "step": 1530 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 4.144016506189822e-06, |
| "loss": 0.6833, |
| "step": 1540 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 4.0793672627235215e-06, |
| "loss": 0.6946, |
| "step": 1550 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 4.014718019257221e-06, |
| "loss": 0.6914, |
| "step": 1560 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 3.950068775790922e-06, |
| "loss": 0.6875, |
| "step": 1570 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 3.885419532324622e-06, |
| "loss": 0.6988, |
| "step": 1580 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 3.820770288858322e-06, |
| "loss": 0.683, |
| "step": 1590 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 3.756121045392022e-06, |
| "loss": 0.6894, |
| "step": 1600 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 3.6914718019257225e-06, |
| "loss": 0.6967, |
| "step": 1610 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 3.6268225584594227e-06, |
| "loss": 0.6912, |
| "step": 1620 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 3.5621733149931223e-06, |
| "loss": 0.6913, |
| "step": 1630 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 3.497524071526823e-06, |
| "loss": 0.6895, |
| "step": 1640 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 3.432874828060523e-06, |
| "loss": 0.6977, |
| "step": 1650 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 3.368225584594223e-06, |
| "loss": 0.6884, |
| "step": 1660 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 3.303576341127923e-06, |
| "loss": 0.6889, |
| "step": 1670 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 3.2389270976616233e-06, |
| "loss": 0.6887, |
| "step": 1680 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 3.1742778541953234e-06, |
| "loss": 0.6947, |
| "step": 1690 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 3.1096286107290235e-06, |
| "loss": 0.6933, |
| "step": 1700 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 3.0449793672627237e-06, |
| "loss": 0.6972, |
| "step": 1710 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 2.9803301237964238e-06, |
| "loss": 0.6986, |
| "step": 1720 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 2.915680880330124e-06, |
| "loss": 0.6965, |
| "step": 1730 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 2.8510316368638244e-06, |
| "loss": 0.6912, |
| "step": 1740 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 2.786382393397524e-06, |
| "loss": 0.6956, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 2.7217331499312246e-06, |
| "loss": 0.6889, |
| "step": 1760 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 2.6570839064649243e-06, |
| "loss": 0.6861, |
| "step": 1770 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 2.5924346629986244e-06, |
| "loss": 0.691, |
| "step": 1780 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 2.5277854195323245e-06, |
| "loss": 0.6862, |
| "step": 1790 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 2.4631361760660246e-06, |
| "loss": 0.6893, |
| "step": 1800 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 2.398486932599725e-06, |
| "loss": 0.6878, |
| "step": 1810 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 2.333837689133425e-06, |
| "loss": 0.6837, |
| "step": 1820 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 2.2691884456671254e-06, |
| "loss": 0.6873, |
| "step": 1830 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 2.204539202200825e-06, |
| "loss": 0.6986, |
| "step": 1840 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 2.1398899587345256e-06, |
| "loss": 0.7065, |
| "step": 1850 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 2.0752407152682253e-06, |
| "loss": 0.6907, |
| "step": 1860 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 2.010591471801926e-06, |
| "loss": 0.6934, |
| "step": 1870 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 1.945942228335626e-06, |
| "loss": 0.7035, |
| "step": 1880 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 1.881292984869326e-06, |
| "loss": 0.6919, |
| "step": 1890 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 1.816643741403026e-06, |
| "loss": 0.6858, |
| "step": 1900 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 1.7519944979367263e-06, |
| "loss": 0.6902, |
| "step": 1910 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 1.6873452544704266e-06, |
| "loss": 0.6996, |
| "step": 1920 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 1.6226960110041267e-06, |
| "loss": 0.6841, |
| "step": 1930 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 1.5580467675378268e-06, |
| "loss": 0.6934, |
| "step": 1940 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 1.493397524071527e-06, |
| "loss": 0.6828, |
| "step": 1950 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 1.428748280605227e-06, |
| "loss": 0.6993, |
| "step": 1960 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 1.3640990371389272e-06, |
| "loss": 0.695, |
| "step": 1970 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 1.2994497936726273e-06, |
| "loss": 0.7036, |
| "step": 1980 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 1.2348005502063276e-06, |
| "loss": 0.6844, |
| "step": 1990 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 1.1701513067400277e-06, |
| "loss": 0.6908, |
| "step": 2000 |
| } |
| ], |
| "max_steps": 2181, |
| "num_train_epochs": 3, |
| "total_flos": 0.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
