| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.12909680646775, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 3.0243, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.915, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.9116, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8826, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8696, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8775, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.87, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8411, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8488, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8009, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8595, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8559, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8263, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8642, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8383, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8285, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8655, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8234, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8558, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8231, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8069, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8212, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8382, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8344, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8534, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8426, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7892, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7954, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8582, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8365, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7891, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8129, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8047, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8331, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.851, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.829, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8266, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8251, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8473, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8351, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8375, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8011, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8072, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8397, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7985, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8076, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7993, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8001, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8046, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8017, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8004, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8409, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8321, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.818, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8402, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8167, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7762, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7975, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8285, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8029, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8491, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7931, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8352, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7869, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7956, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7798, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8087, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8393, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7998, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.835, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8198, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8249, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7963, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7589, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8159, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7879, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7873, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8207, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8378, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8275, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8256, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8334, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.806, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7845, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.795, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7606, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7933, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7927, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.79, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8054, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8285, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8354, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8047, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7817, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7981, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7823, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8003, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8004, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7759, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7948, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8154, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8316, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7714, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7947, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7863, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.806, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7746, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8075, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7656, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8109, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7928, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.822, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.861, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7977, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7805, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7758, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8053, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7491, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7826, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8382, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8119, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7602, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8249, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7729, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7897, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7743, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7666, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8326, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8229, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8069, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7894, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.787, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.785, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8032, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.761, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8074, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7925, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8167, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7805, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7819, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8112, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8142, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7858, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7584, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7949, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8499, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8228, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7731, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8033, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8054, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7958, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7941, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8277, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8018, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.819, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7754, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8026, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8033, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7695, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8164, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7647, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7884, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7753, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8015, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8319, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7853, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7749, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8053, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7519, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7748, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7965, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7741, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8151, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7982, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8229, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7984, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7661, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8516, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8034, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7805, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8021, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.783, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7751, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7926, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7954, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7673, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7672, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7545, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7972, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8428, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7968, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8249, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7491, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7843, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8082, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7685, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7934, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.796, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8012, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7579, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7828, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7891, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7988, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7918, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8269, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7916, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.785, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7495, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7931, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7967, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7271, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7938, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8055, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7623, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8052, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8144, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7964, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7778, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7869, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8252, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.783, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7544, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8041, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.772, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7963, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8056, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8026, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7501, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7515, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7931, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8126, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8232, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7556, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7862, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7879, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.806, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.758, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7859, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7884, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7934, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8048, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7897, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7762, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7546, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7745, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7908, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8209, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7881, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7929, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7735, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7656, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8099, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7826, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7748, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7853, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8411, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.747, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8118, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7836, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7898, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7546, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7698, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.6932, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7996, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7868, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7585, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7719, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7952, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7817, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7545, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7823, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.766, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7731, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7992, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8014, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7459, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7708, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.803, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7514, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7704, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7466, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7674, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7642, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8095, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7579, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8004, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7808, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.765, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7911, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8098, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7703, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8085, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7803, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7789, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7952, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7917, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7838, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.811, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7422, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.785, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8173, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7445, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7672, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8135, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7966, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8011, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7647, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.816, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7714, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7675, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7416, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7642, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7898, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7972, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7577, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8205, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7782, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8022, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.793, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7961, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7955, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.769, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7434, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7987, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7568, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7449, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7982, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7978, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8214, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.773, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8078, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7599, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7697, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7738, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7751, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7712, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7691, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7893, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8128, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.779, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8076, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7681, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7177, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8094, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7567, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8128, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.828, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8377, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8001, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.79, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7908, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7747, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7864, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8027, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7764, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7274, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7617, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8357, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7528, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7862, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7632, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.777, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7444, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8136, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7451, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7554, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7968, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7819, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7801, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7972, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7974, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7924, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8114, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7964, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7775, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7402, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7515, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7856, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7761, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8069, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7905, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7477, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7984, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7684, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7882, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7912, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7461, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.747, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7608, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7469, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7701, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7944, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7547, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7536, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7743, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8194, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7548, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7768, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.832, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7862, | |
| "step": 4000 | |
| } | |
| ], | |
| "max_steps": 30984, | |
| "num_train_epochs": 1, | |
| "total_flos": 2.7083329329168384e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |