| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9986516853932583, | |
| "global_step": 3336, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00014285714285714287, | |
| "loss": 2.1522, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1153, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1966, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0445, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.129, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1375, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.222, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1855, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0923, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1476, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1455, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0955, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1059, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1632, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1201, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0771, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0987, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1079, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0977, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1799, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1506, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1915, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1582, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.152, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1891, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.134, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0751, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1839, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0991, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1025, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0104, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0413, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2118, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0967, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1071, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0002, | |
| "loss": 2.133, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9729, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1274, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1024, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1811, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002, | |
| "loss": 2.185, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1036, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1459, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0876, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0875, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002, | |
| "loss": 2.055, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1818, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1068, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2196, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1927, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002, | |
| "loss": 2.036, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1142, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1268, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1791, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1283, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0002, | |
| "loss": 2.193, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0152, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0009, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1249, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1407, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1295, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002, | |
| "loss": 2.089, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0329, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0177, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0617, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1508, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0777, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1076, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0268, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0524, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2323, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0497, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1242, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0884, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1387, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0996, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0888, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0002, | |
| "loss": 2.075, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0399, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0794, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1025, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0167, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1276, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1315, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1187, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1076, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0695, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0748, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2029, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1045, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1248, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1507, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1449, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1644, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0002, | |
| "loss": 2.032, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0764, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1192, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0544, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1685, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0982, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1693, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1791, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0983, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1951, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1016, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0002, | |
| "loss": 2.036, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1191, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1521, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0011, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1262, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0556, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1164, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9819, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0956, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0392, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9776, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0342, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2182, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0083, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0577, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0761, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0859, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1006, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002, | |
| "loss": 1.995, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0788, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1673, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0002, | |
| "loss": 2.139, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1512, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1254, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0226, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0838, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0932, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1345, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0827, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1414, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0599, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1055, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0964, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0999, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0562, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1069, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0947, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1343, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0173, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1017, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9641, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1025, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0687, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002, | |
| "loss": 2.236, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0522, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0771, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0485, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8949, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9603, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002, | |
| "loss": 2.074, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0713, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0783, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9597, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0893, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0602, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0002, | |
| "loss": 2.162, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1747, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1334, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0435, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1426, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1423, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002, | |
| "loss": 2.2027, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9743, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0254, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0267, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1039, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0299, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1488, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0275, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1143, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1606, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0503, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002, | |
| "loss": 2.04, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0776, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1319, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0666, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0002, | |
| "loss": 2.053, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9928, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1031, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0299, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0994, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0448, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0595, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0684, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0864, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002, | |
| "loss": 2.124, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9532, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0002, | |
| "loss": 2.12, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0919, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1219, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0002, | |
| "loss": 2.064, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0373, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1743, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1573, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1227, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0709, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0339, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1041, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0002, | |
| "loss": 1.964, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0002, | |
| "loss": 2.08, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0518, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0307, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0245, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0705, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0527, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1128, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1393, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9537, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0637, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0353, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9832, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9752, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0359, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0974, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0753, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1015, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.066, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0608, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9895, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9744, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.054, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9988, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0002, | |
| "loss": 2.053, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0016, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0695, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0228, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0002, | |
| "loss": 2.057, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0661, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0485, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9971, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0407, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1142, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0038, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0038, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0773, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0318, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0458, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9554, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0387, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0064, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0005, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0559, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0194, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0007, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0846, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1337, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0797, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.0002, | |
| "loss": 1.987, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0475, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0013, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0002, | |
| "loss": 2.015, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0226, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9841, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0383, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0002, | |
| "loss": 2.124, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0644, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9905, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0737, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1003, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0002, | |
| "loss": 2.004, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9829, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0634, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0428, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9416, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1038, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9866, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0276, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0687, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9582, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9865, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0927, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1238, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8871, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0573, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0741, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0955, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0308, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9725, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0641, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1123, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0281, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0856, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0672, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0245, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1526, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9798, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0419, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0524, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0126, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0409, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0467, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1037, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1238, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0907, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0154, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9919, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0587, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0638, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0735, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9985, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0722, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0665, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0002, | |
| "loss": 2.042, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0074, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9094, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9634, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9839, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1577, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0551, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0602, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0428, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1107, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9711, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0454, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0748, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0038, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0406, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9877, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0979, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0185, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9865, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.0002, | |
| "loss": 2.005, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0063, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0485, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0386, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0974, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0669, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9982, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0033, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.958, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9564, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0332, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9938, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9808, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0822, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0447, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9504, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0112, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0605, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0779, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0668, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0865, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0448, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0022, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0191, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9709, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0996, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0665, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0026, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.99, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9911, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9545, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9896, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0052, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0018, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9814, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0657, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0972, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9325, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0156, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1249, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0976, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0244, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0838, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9799, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9803, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9679, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9555, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0569, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0188, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0643, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0002, | |
| "loss": 2.14, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9972, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0366, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0284, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0579, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0083, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0264, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9787, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0181, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0741, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0002, | |
| "loss": 2.109, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0251, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0554, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0581, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9969, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0673, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1152, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9933, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9623, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9943, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9933, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9461, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0254, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1032, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9983, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9862, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1021, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0499, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9565, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9829, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0002, | |
| "loss": 2.081, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0852, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0039, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0727, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0764, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0426, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1461, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0314, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0846, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1664, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0705, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0439, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0807, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0002, | |
| "loss": 1.997, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0189, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0002, | |
| "loss": 2.056, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0832, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0163, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0145, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0765, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0208, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0408, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9439, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.978, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1421, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0027, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0091, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9229, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9744, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0321, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0983, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1434, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0524, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1093, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9859, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9931, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0115, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9579, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.042, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0394, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9881, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9385, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0002, | |
| "loss": 1.984, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8972, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0127, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0322, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9344, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9156, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9947, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9557, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9373, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0865, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9885, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0106, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9263, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8822, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9703, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9723, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0002, | |
| "loss": 1.955, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9522, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0376, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8136, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8973, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8794, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0367, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9562, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0002, | |
| "loss": 1.99, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9503, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0289, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9522, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9947, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0269, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9271, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0773, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9706, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0688, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9785, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8934, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.0002, | |
| "loss": 2.059, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8678, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0496, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0301, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0092, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9622, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.0002, | |
| "loss": 2.033, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0002, | |
| "loss": 2.065, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0195, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9856, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9478, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9775, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0245, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9808, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9839, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0303, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0041, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9821, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0558, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9337, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9905, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.0002, | |
| "loss": 1.944, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9393, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9812, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9049, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9862, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0057, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9763, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9989, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9061, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9701, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9051, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9026, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9964, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9455, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9355, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9866, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.0002, | |
| "loss": 1.921, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0324, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9441, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9736, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8812, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0055, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9584, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9264, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9187, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9087, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.888, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9745, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9592, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9549, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9745, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9604, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8755, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0011, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9574, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0206, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9416, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9429, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9836, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0279, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0464, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9094, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9515, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9832, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9372, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1184, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9291, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 0.0002, | |
| "loss": 1.961, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9648, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9476, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9632, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9584, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9579, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9163, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9172, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9536, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9738, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.0002, | |
| "loss": 1.847, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0218, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0131, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0069, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0603, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0213, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.0002, | |
| "loss": 1.952, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9705, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.0002, | |
| "loss": 1.99, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9214, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.0002, | |
| "loss": 2.028, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9956, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0084, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9649, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9657, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9632, | |
| "step": 2895 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9227, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9293, | |
| "step": 2905 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0111, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9686, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9098, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9402, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0028, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9038, | |
| "step": 2935 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9543, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0392, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.0002, | |
| "loss": 1.905, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9441, | |
| "step": 2955 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9979, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0091, | |
| "step": 2965 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9602, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9426, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0169, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0368, | |
| "step": 2985 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8992, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9961, | |
| "step": 2995 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.0002, | |
| "loss": 2.1249, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9761, | |
| "step": 3005 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9954, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9454, | |
| "step": 3015 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0621, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8487, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0469, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9774, | |
| "step": 3035 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0672, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9935, | |
| "step": 3045 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9712, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9269, | |
| "step": 3055 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9989, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9844, | |
| "step": 3065 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0057, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8694, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0094, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9451, | |
| "step": 3085 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0165, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9604, | |
| "step": 3095 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9604, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8861, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9491, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9272, | |
| "step": 3115 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0349, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9834, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8975, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8226, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0628, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8673, | |
| "step": 3145 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8963, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9782, | |
| "step": 3155 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9048, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0262, | |
| "step": 3165 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0755, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0465, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 0.0002, | |
| "loss": 1.906, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9611, | |
| "step": 3185 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9635, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0385, | |
| "step": 3195 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0048, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9428, | |
| "step": 3205 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9685, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0215, | |
| "step": 3215 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0445, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9155, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9542, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0736, | |
| "step": 3235 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8556, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9885, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0219, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9969, | |
| "step": 3255 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9887, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9237, | |
| "step": 3265 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0694, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9643, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0543, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0497, | |
| "step": 3285 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9816, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0116, | |
| "step": 3295 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8615, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0354, | |
| "step": 3305 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0116, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9864, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0002, | |
| "loss": 1.8841, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0151, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.0002, | |
| "loss": 2.0346, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0002, | |
| "loss": 1.9278, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 3336, | |
| "total_flos": 9.82091491287552e+17, | |
| "train_loss": 1.5420540891033854, | |
| "train_runtime": 40504.6107, | |
| "train_samples_per_second": 0.33, | |
| "train_steps_per_second": 0.082 | |
| } | |
| ], | |
| "max_steps": 3336, | |
| "num_train_epochs": 3, | |
| "total_flos": 9.82091491287552e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |