| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9801980198019802, | |
| "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": 1.0000000000000002e-06, | |
| "loss": 0.692, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.6851, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3e-06, | |
| "loss": 0.684, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.6851, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6784, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6e-06, | |
| "loss": 0.6873, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 0.6819, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.6911, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9e-06, | |
| "loss": 0.6832, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6941, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 0.6733, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.6941, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 0.6695, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.6911, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.5e-05, | |
| "loss": 0.6834, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.6549, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 0.6843, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.6793, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9e-05, | |
| "loss": 0.6626, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2e-05, | |
| "loss": 0.5984, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.1e-05, | |
| "loss": 0.7177, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.6791, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 0.6775, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.6688, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.6781, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.6575, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 0.7072, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.6703, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.9e-05, | |
| "loss": 0.6599, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3e-05, | |
| "loss": 0.6333, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.1e-05, | |
| "loss": 0.6763, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.6848, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.3e-05, | |
| "loss": 0.6731, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.6705, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.6575, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.6181, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.67, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.6074, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.6562, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4e-05, | |
| "loss": 0.7, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.1e-05, | |
| "loss": 0.6448, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.6828, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.6688, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.6626, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.6477, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.6886, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.6824, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.6429, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.7269, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5e-05, | |
| "loss": 0.6604, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.967105263157895e-05, | |
| "loss": 0.6495, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.9342105263157894e-05, | |
| "loss": 0.6372, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.901315789473684e-05, | |
| "loss": 0.6697, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.868421052631579e-05, | |
| "loss": 0.6671, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.8355263157894734e-05, | |
| "loss": 0.617, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.802631578947368e-05, | |
| "loss": 0.7253, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.769736842105263e-05, | |
| "loss": 0.6568, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.736842105263158e-05, | |
| "loss": 0.6646, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.703947368421053e-05, | |
| "loss": 0.7026, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.671052631578948e-05, | |
| "loss": 0.6863, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.638157894736843e-05, | |
| "loss": 0.6684, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.605263157894737e-05, | |
| "loss": 0.6482, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.572368421052632e-05, | |
| "loss": 0.6287, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.539473684210527e-05, | |
| "loss": 0.6767, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.506578947368421e-05, | |
| "loss": 0.6524, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.473684210526316e-05, | |
| "loss": 0.6543, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.440789473684211e-05, | |
| "loss": 0.6408, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.407894736842105e-05, | |
| "loss": 0.672, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.6361, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.342105263157895e-05, | |
| "loss": 0.6891, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.30921052631579e-05, | |
| "loss": 0.6911, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.2763157894736847e-05, | |
| "loss": 0.6503, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.2434210526315796e-05, | |
| "loss": 0.6552, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.210526315789474e-05, | |
| "loss": 0.6684, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.177631578947369e-05, | |
| "loss": 0.6592, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.1447368421052636e-05, | |
| "loss": 0.6478, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.111842105263158e-05, | |
| "loss": 0.6628, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.078947368421053e-05, | |
| "loss": 0.6485, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.0460526315789476e-05, | |
| "loss": 0.6567, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.0131578947368425e-05, | |
| "loss": 0.6694, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.980263157894737e-05, | |
| "loss": 0.6078, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.9473684210526316e-05, | |
| "loss": 0.6832, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.9144736842105265e-05, | |
| "loss": 0.6638, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.8815789473684214e-05, | |
| "loss": 0.7009, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.848684210526316e-05, | |
| "loss": 0.6482, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.815789473684211e-05, | |
| "loss": 0.6194, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.7828947368421054e-05, | |
| "loss": 0.6492, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.7175, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.717105263157895e-05, | |
| "loss": 0.6696, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.6842105263157895e-05, | |
| "loss": 0.6697, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.6513157894736844e-05, | |
| "loss": 0.6839, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.618421052631579e-05, | |
| "loss": 0.6369, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.5855263157894735e-05, | |
| "loss": 0.6564, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.5526315789473684e-05, | |
| "loss": 0.65, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.519736842105263e-05, | |
| "loss": 0.6178, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4868421052631575e-05, | |
| "loss": 0.6656, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.4539473684210524e-05, | |
| "loss": 0.6427, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.421052631578947e-05, | |
| "loss": 0.644, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.388157894736842e-05, | |
| "loss": 0.6889, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.355263157894737e-05, | |
| "loss": 0.6547, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.322368421052632e-05, | |
| "loss": 0.6458, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.289473684210527e-05, | |
| "loss": 0.6092, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.256578947368421e-05, | |
| "loss": 0.6023, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.223684210526316e-05, | |
| "loss": 0.6188, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.190789473684211e-05, | |
| "loss": 0.6435, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.157894736842105e-05, | |
| "loss": 0.611, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.125e-05, | |
| "loss": 0.645, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.092105263157895e-05, | |
| "loss": 0.641, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.059210526315789e-05, | |
| "loss": 0.5954, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.0263157894736844e-05, | |
| "loss": 0.6351, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 2.9934210526315793e-05, | |
| "loss": 0.6116, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 2.9605263157894735e-05, | |
| "loss": 0.6003, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 2.9276315789473684e-05, | |
| "loss": 0.6181, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.8947368421052634e-05, | |
| "loss": 0.6172, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 2.861842105263158e-05, | |
| "loss": 0.6334, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.8289473684210528e-05, | |
| "loss": 0.5976, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 2.7960526315789477e-05, | |
| "loss": 0.6194, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 2.7631578947368426e-05, | |
| "loss": 0.5842, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 2.730263157894737e-05, | |
| "loss": 0.6007, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 2.6973684210526317e-05, | |
| "loss": 0.6574, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.6644736842105266e-05, | |
| "loss": 0.5696, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.6315789473684212e-05, | |
| "loss": 0.5755, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.598684210526316e-05, | |
| "loss": 0.6179, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.565789473684211e-05, | |
| "loss": 0.6103, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.5328947368421052e-05, | |
| "loss": 0.5706, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.5457, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.4671052631578947e-05, | |
| "loss": 0.6563, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.4342105263157896e-05, | |
| "loss": 0.5547, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.401315789473684e-05, | |
| "loss": 0.5982, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.368421052631579e-05, | |
| "loss": 0.6063, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.335526315789474e-05, | |
| "loss": 0.6005, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.3026315789473685e-05, | |
| "loss": 0.5698, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.2697368421052634e-05, | |
| "loss": 0.6047, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.236842105263158e-05, | |
| "loss": 0.6283, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.2039473684210525e-05, | |
| "loss": 0.6027, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.1710526315789474e-05, | |
| "loss": 0.5837, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.1381578947368423e-05, | |
| "loss": 0.6238, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.105263157894737e-05, | |
| "loss": 0.6703, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.0723684210526318e-05, | |
| "loss": 0.6224, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.0394736842105264e-05, | |
| "loss": 0.5902, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.0065789473684213e-05, | |
| "loss": 0.587, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.9736842105263158e-05, | |
| "loss": 0.5555, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.9407894736842107e-05, | |
| "loss": 0.561, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.9078947368421056e-05, | |
| "loss": 0.5143, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.5531, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.8421052631578947e-05, | |
| "loss": 0.6924, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.8092105263157896e-05, | |
| "loss": 0.6052, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.7763157894736842e-05, | |
| "loss": 0.548, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.7434210526315788e-05, | |
| "loss": 0.6726, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.7105263157894737e-05, | |
| "loss": 0.5311, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.6776315789473686e-05, | |
| "loss": 0.6106, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.6447368421052635e-05, | |
| "loss": 0.5802, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.611842105263158e-05, | |
| "loss": 0.5713, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.5789473684210526e-05, | |
| "loss": 0.6444, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.5460526315789475e-05, | |
| "loss": 0.5585, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.5131578947368422e-05, | |
| "loss": 0.624, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4802631578947368e-05, | |
| "loss": 0.6592, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.4473684210526317e-05, | |
| "loss": 0.5805, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.4144736842105264e-05, | |
| "loss": 0.6499, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3815789473684213e-05, | |
| "loss": 0.6087, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.3486842105263159e-05, | |
| "loss": 0.5951, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3157894736842106e-05, | |
| "loss": 0.5929, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.2828947368421055e-05, | |
| "loss": 0.584, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.6071, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.2171052631578948e-05, | |
| "loss": 0.5658, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.1842105263157895e-05, | |
| "loss": 0.5683, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.1513157894736843e-05, | |
| "loss": 0.5722, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.118421052631579e-05, | |
| "loss": 0.5975, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.0855263157894737e-05, | |
| "loss": 0.6027, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.0526315789473684e-05, | |
| "loss": 0.6929, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.0197368421052632e-05, | |
| "loss": 0.5979, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 9.868421052631579e-06, | |
| "loss": 0.5386, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 9.539473684210528e-06, | |
| "loss": 0.5994, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.210526315789474e-06, | |
| "loss": 0.7043, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.881578947368421e-06, | |
| "loss": 0.5805, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.552631578947368e-06, | |
| "loss": 0.5956, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.223684210526317e-06, | |
| "loss": 0.5831, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.894736842105263e-06, | |
| "loss": 0.5695, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.565789473684211e-06, | |
| "loss": 0.6033, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 7.236842105263158e-06, | |
| "loss": 0.5645, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.9078947368421065e-06, | |
| "loss": 0.5921, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 6.578947368421053e-06, | |
| "loss": 0.617, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.6565, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 5.921052631578948e-06, | |
| "loss": 0.5636, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 5.592105263157895e-06, | |
| "loss": 0.6248, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.263157894736842e-06, | |
| "loss": 0.6867, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.9342105263157895e-06, | |
| "loss": 0.5723, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.605263157894737e-06, | |
| "loss": 0.5877, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.276315789473684e-06, | |
| "loss": 0.583, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.9473684210526315e-06, | |
| "loss": 0.6015, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 3.618421052631579e-06, | |
| "loss": 0.5597, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.2894736842105265e-06, | |
| "loss": 0.6101, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.960526315789474e-06, | |
| "loss": 0.5713, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.631578947368421e-06, | |
| "loss": 0.5499, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.3026315789473684e-06, | |
| "loss": 0.5892, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.9736842105263157e-06, | |
| "loss": 0.6348, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.6447368421052632e-06, | |
| "loss": 0.5842, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.3157894736842106e-06, | |
| "loss": 0.557, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.868421052631579e-07, | |
| "loss": 0.6027, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.578947368421053e-07, | |
| "loss": 0.6697, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 2020, | |
| "num_train_epochs": 2, | |
| "total_flos": 4238029485201408.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |