| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 8.695652173913043, |
| "global_step": 1000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.04, |
| "learning_rate": 5.000000000000001e-07, |
| "loss": 0.7601, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.0000000000000002e-06, |
| "loss": 0.7662, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.5e-06, |
| "loss": 0.7385, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.9000000000000002e-06, |
| "loss": 0.707, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 2.4000000000000003e-06, |
| "loss": 0.6551, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 2.8000000000000003e-06, |
| "loss": 0.5934, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 3.3000000000000006e-06, |
| "loss": 0.5183, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 3.8000000000000005e-06, |
| "loss": 0.487, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.3e-06, |
| "loss": 0.4695, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.800000000000001e-06, |
| "loss": 0.4183, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 5.300000000000001e-06, |
| "loss": 0.4144, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 5.8e-06, |
| "loss": 0.3759, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 6.300000000000001e-06, |
| "loss": 0.3933, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.800000000000001e-06, |
| "loss": 0.406, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 7.3e-06, |
| "loss": 0.3771, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 7.800000000000002e-06, |
| "loss": 0.3719, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 8.3e-06, |
| "loss": 0.366, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 8.8e-06, |
| "loss": 0.3465, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 9.3e-06, |
| "loss": 0.3979, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 9.800000000000001e-06, |
| "loss": 0.3745, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 9.966666666666667e-06, |
| "loss": 0.388, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 9.911111111111113e-06, |
| "loss": 0.4033, |
| "step": 110 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 9.855555555555555e-06, |
| "loss": 0.3652, |
| "step": 115 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 9.800000000000001e-06, |
| "loss": 0.3252, |
| "step": 120 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 9.744444444444445e-06, |
| "loss": 0.357, |
| "step": 125 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 9.688888888888889e-06, |
| "loss": 0.3843, |
| "step": 130 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 9.633333333333335e-06, |
| "loss": 0.3602, |
| "step": 135 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 9.577777777777779e-06, |
| "loss": 0.2971, |
| "step": 140 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 9.522222222222223e-06, |
| "loss": 0.3397, |
| "step": 145 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 9.466666666666667e-06, |
| "loss": 0.3062, |
| "step": 150 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 9.411111111111113e-06, |
| "loss": 0.327, |
| "step": 155 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 9.355555555555557e-06, |
| "loss": 0.3006, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 9.3e-06, |
| "loss": 0.304, |
| "step": 165 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 9.244444444444445e-06, |
| "loss": 0.2943, |
| "step": 170 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 9.188888888888889e-06, |
| "loss": 0.252, |
| "step": 175 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 9.133333333333335e-06, |
| "loss": 0.3139, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 9.077777777777779e-06, |
| "loss": 0.2665, |
| "step": 185 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 9.022222222222223e-06, |
| "loss": 0.302, |
| "step": 190 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 8.966666666666667e-06, |
| "loss": 0.2248, |
| "step": 195 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 8.91111111111111e-06, |
| "loss": 0.2442, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 8.855555555555556e-06, |
| "loss": 0.2233, |
| "step": 205 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 8.8e-06, |
| "loss": 0.1981, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 8.744444444444446e-06, |
| "loss": 0.2188, |
| "step": 215 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 8.68888888888889e-06, |
| "loss": 0.2156, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 8.633333333333334e-06, |
| "loss": 0.1932, |
| "step": 225 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 8.577777777777778e-06, |
| "loss": 0.2024, |
| "step": 230 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 8.522222222222222e-06, |
| "loss": 0.2008, |
| "step": 235 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 8.466666666666668e-06, |
| "loss": 0.2096, |
| "step": 240 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 8.411111111111112e-06, |
| "loss": 0.1882, |
| "step": 245 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 8.355555555555556e-06, |
| "loss": 0.2387, |
| "step": 250 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 8.3e-06, |
| "loss": 0.1566, |
| "step": 255 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 8.244444444444444e-06, |
| "loss": 0.2002, |
| "step": 260 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 8.18888888888889e-06, |
| "loss": 0.1623, |
| "step": 265 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 8.133333333333334e-06, |
| "loss": 0.2073, |
| "step": 270 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 8.077777777777778e-06, |
| "loss": 0.2038, |
| "step": 275 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 8.022222222222222e-06, |
| "loss": 0.1519, |
| "step": 280 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 7.966666666666668e-06, |
| "loss": 0.1626, |
| "step": 285 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 7.911111111111112e-06, |
| "loss": 0.1737, |
| "step": 290 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.855555555555556e-06, |
| "loss": 0.1992, |
| "step": 295 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 7.800000000000002e-06, |
| "loss": 0.1539, |
| "step": 300 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 7.744444444444446e-06, |
| "loss": 0.1764, |
| "step": 305 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 7.68888888888889e-06, |
| "loss": 0.1601, |
| "step": 310 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 7.633333333333334e-06, |
| "loss": 0.1793, |
| "step": 315 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 7.5777777777777785e-06, |
| "loss": 0.1314, |
| "step": 320 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 7.5222222222222226e-06, |
| "loss": 0.1543, |
| "step": 325 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 7.4666666666666675e-06, |
| "loss": 0.1393, |
| "step": 330 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 7.411111111111112e-06, |
| "loss": 0.1847, |
| "step": 335 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 7.3555555555555555e-06, |
| "loss": 0.2043, |
| "step": 340 |
| }, |
| { |
| "epoch": 3.0, |
| "learning_rate": 7.3e-06, |
| "loss": 0.1308, |
| "step": 345 |
| }, |
| { |
| "epoch": 3.04, |
| "learning_rate": 7.244444444444445e-06, |
| "loss": 0.1352, |
| "step": 350 |
| }, |
| { |
| "epoch": 3.09, |
| "learning_rate": 7.188888888888889e-06, |
| "loss": 0.1378, |
| "step": 355 |
| }, |
| { |
| "epoch": 3.13, |
| "learning_rate": 7.133333333333334e-06, |
| "loss": 0.1581, |
| "step": 360 |
| }, |
| { |
| "epoch": 3.17, |
| "learning_rate": 7.077777777777778e-06, |
| "loss": 0.1399, |
| "step": 365 |
| }, |
| { |
| "epoch": 3.22, |
| "learning_rate": 7.022222222222222e-06, |
| "loss": 0.149, |
| "step": 370 |
| }, |
| { |
| "epoch": 3.26, |
| "learning_rate": 6.966666666666667e-06, |
| "loss": 0.0899, |
| "step": 375 |
| }, |
| { |
| "epoch": 3.3, |
| "learning_rate": 6.911111111111112e-06, |
| "loss": 0.1157, |
| "step": 380 |
| }, |
| { |
| "epoch": 3.35, |
| "learning_rate": 6.855555555555556e-06, |
| "loss": 0.137, |
| "step": 385 |
| }, |
| { |
| "epoch": 3.39, |
| "learning_rate": 6.800000000000001e-06, |
| "loss": 0.109, |
| "step": 390 |
| }, |
| { |
| "epoch": 3.43, |
| "learning_rate": 6.744444444444444e-06, |
| "loss": 0.0956, |
| "step": 395 |
| }, |
| { |
| "epoch": 3.48, |
| "learning_rate": 6.688888888888889e-06, |
| "loss": 0.1121, |
| "step": 400 |
| }, |
| { |
| "epoch": 3.52, |
| "learning_rate": 6.633333333333334e-06, |
| "loss": 0.125, |
| "step": 405 |
| }, |
| { |
| "epoch": 3.57, |
| "learning_rate": 6.577777777777779e-06, |
| "loss": 0.1222, |
| "step": 410 |
| }, |
| { |
| "epoch": 3.61, |
| "learning_rate": 6.522222222222223e-06, |
| "loss": 0.0806, |
| "step": 415 |
| }, |
| { |
| "epoch": 3.65, |
| "learning_rate": 6.466666666666667e-06, |
| "loss": 0.1144, |
| "step": 420 |
| }, |
| { |
| "epoch": 3.7, |
| "learning_rate": 6.411111111111111e-06, |
| "loss": 0.1139, |
| "step": 425 |
| }, |
| { |
| "epoch": 3.74, |
| "learning_rate": 6.355555555555556e-06, |
| "loss": 0.0836, |
| "step": 430 |
| }, |
| { |
| "epoch": 3.78, |
| "learning_rate": 6.300000000000001e-06, |
| "loss": 0.1699, |
| "step": 435 |
| }, |
| { |
| "epoch": 3.83, |
| "learning_rate": 6.2444444444444456e-06, |
| "loss": 0.1187, |
| "step": 440 |
| }, |
| { |
| "epoch": 3.87, |
| "learning_rate": 6.18888888888889e-06, |
| "loss": 0.0888, |
| "step": 445 |
| }, |
| { |
| "epoch": 3.91, |
| "learning_rate": 6.133333333333334e-06, |
| "loss": 0.1015, |
| "step": 450 |
| }, |
| { |
| "epoch": 3.96, |
| "learning_rate": 6.077777777777778e-06, |
| "loss": 0.0988, |
| "step": 455 |
| }, |
| { |
| "epoch": 4.0, |
| "learning_rate": 6.0222222222222225e-06, |
| "loss": 0.1092, |
| "step": 460 |
| }, |
| { |
| "epoch": 4.04, |
| "learning_rate": 5.966666666666667e-06, |
| "loss": 0.0985, |
| "step": 465 |
| }, |
| { |
| "epoch": 4.09, |
| "learning_rate": 5.911111111111112e-06, |
| "loss": 0.0934, |
| "step": 470 |
| }, |
| { |
| "epoch": 4.13, |
| "learning_rate": 5.855555555555556e-06, |
| "loss": 0.0739, |
| "step": 475 |
| }, |
| { |
| "epoch": 4.17, |
| "learning_rate": 5.8e-06, |
| "loss": 0.0905, |
| "step": 480 |
| }, |
| { |
| "epoch": 4.22, |
| "learning_rate": 5.744444444444444e-06, |
| "loss": 0.0695, |
| "step": 485 |
| }, |
| { |
| "epoch": 4.26, |
| "learning_rate": 5.688888888888889e-06, |
| "loss": 0.0875, |
| "step": 490 |
| }, |
| { |
| "epoch": 4.3, |
| "learning_rate": 5.633333333333334e-06, |
| "loss": 0.0896, |
| "step": 495 |
| }, |
| { |
| "epoch": 4.35, |
| "learning_rate": 5.577777777777778e-06, |
| "loss": 0.0751, |
| "step": 500 |
| }, |
| { |
| "epoch": 4.39, |
| "learning_rate": 5.522222222222222e-06, |
| "loss": 0.0923, |
| "step": 505 |
| }, |
| { |
| "epoch": 4.43, |
| "learning_rate": 5.466666666666667e-06, |
| "loss": 0.0841, |
| "step": 510 |
| }, |
| { |
| "epoch": 4.48, |
| "learning_rate": 5.411111111111111e-06, |
| "loss": 0.0978, |
| "step": 515 |
| }, |
| { |
| "epoch": 4.52, |
| "learning_rate": 5.355555555555556e-06, |
| "loss": 0.0831, |
| "step": 520 |
| }, |
| { |
| "epoch": 4.57, |
| "learning_rate": 5.300000000000001e-06, |
| "loss": 0.0817, |
| "step": 525 |
| }, |
| { |
| "epoch": 4.61, |
| "learning_rate": 5.244444444444445e-06, |
| "loss": 0.0612, |
| "step": 530 |
| }, |
| { |
| "epoch": 4.65, |
| "learning_rate": 5.188888888888889e-06, |
| "loss": 0.0601, |
| "step": 535 |
| }, |
| { |
| "epoch": 4.7, |
| "learning_rate": 5.133333333333334e-06, |
| "loss": 0.0664, |
| "step": 540 |
| }, |
| { |
| "epoch": 4.74, |
| "learning_rate": 5.077777777777778e-06, |
| "loss": 0.0893, |
| "step": 545 |
| }, |
| { |
| "epoch": 4.78, |
| "learning_rate": 5.022222222222223e-06, |
| "loss": 0.0725, |
| "step": 550 |
| }, |
| { |
| "epoch": 4.83, |
| "learning_rate": 4.966666666666667e-06, |
| "loss": 0.0761, |
| "step": 555 |
| }, |
| { |
| "epoch": 4.87, |
| "learning_rate": 4.911111111111112e-06, |
| "loss": 0.0509, |
| "step": 560 |
| }, |
| { |
| "epoch": 4.91, |
| "learning_rate": 4.855555555555556e-06, |
| "loss": 0.0841, |
| "step": 565 |
| }, |
| { |
| "epoch": 4.96, |
| "learning_rate": 4.800000000000001e-06, |
| "loss": 0.1143, |
| "step": 570 |
| }, |
| { |
| "epoch": 5.0, |
| "learning_rate": 4.744444444444445e-06, |
| "loss": 0.0867, |
| "step": 575 |
| }, |
| { |
| "epoch": 5.04, |
| "learning_rate": 4.6888888888888895e-06, |
| "loss": 0.0829, |
| "step": 580 |
| }, |
| { |
| "epoch": 5.09, |
| "learning_rate": 4.633333333333334e-06, |
| "loss": 0.0609, |
| "step": 585 |
| }, |
| { |
| "epoch": 5.13, |
| "learning_rate": 4.5777777777777785e-06, |
| "loss": 0.0551, |
| "step": 590 |
| }, |
| { |
| "epoch": 5.17, |
| "learning_rate": 4.5222222222222225e-06, |
| "loss": 0.0653, |
| "step": 595 |
| }, |
| { |
| "epoch": 5.22, |
| "learning_rate": 4.4666666666666665e-06, |
| "loss": 0.067, |
| "step": 600 |
| }, |
| { |
| "epoch": 5.26, |
| "learning_rate": 4.411111111111111e-06, |
| "loss": 0.0561, |
| "step": 605 |
| }, |
| { |
| "epoch": 5.3, |
| "learning_rate": 4.3555555555555555e-06, |
| "loss": 0.0465, |
| "step": 610 |
| }, |
| { |
| "epoch": 5.35, |
| "learning_rate": 4.3e-06, |
| "loss": 0.0484, |
| "step": 615 |
| }, |
| { |
| "epoch": 5.39, |
| "learning_rate": 4.244444444444445e-06, |
| "loss": 0.0667, |
| "step": 620 |
| }, |
| { |
| "epoch": 5.43, |
| "learning_rate": 4.188888888888889e-06, |
| "loss": 0.0692, |
| "step": 625 |
| }, |
| { |
| "epoch": 5.48, |
| "learning_rate": 4.133333333333333e-06, |
| "loss": 0.0602, |
| "step": 630 |
| }, |
| { |
| "epoch": 5.52, |
| "learning_rate": 4.077777777777778e-06, |
| "loss": 0.0665, |
| "step": 635 |
| }, |
| { |
| "epoch": 5.57, |
| "learning_rate": 4.022222222222222e-06, |
| "loss": 0.0974, |
| "step": 640 |
| }, |
| { |
| "epoch": 5.61, |
| "learning_rate": 3.966666666666667e-06, |
| "loss": 0.0573, |
| "step": 645 |
| }, |
| { |
| "epoch": 5.65, |
| "learning_rate": 3.911111111111112e-06, |
| "loss": 0.064, |
| "step": 650 |
| }, |
| { |
| "epoch": 5.7, |
| "learning_rate": 3.855555555555556e-06, |
| "loss": 0.0556, |
| "step": 655 |
| }, |
| { |
| "epoch": 5.74, |
| "learning_rate": 3.8000000000000005e-06, |
| "loss": 0.0401, |
| "step": 660 |
| }, |
| { |
| "epoch": 5.78, |
| "learning_rate": 3.744444444444445e-06, |
| "loss": 0.0626, |
| "step": 665 |
| }, |
| { |
| "epoch": 5.83, |
| "learning_rate": 3.688888888888889e-06, |
| "loss": 0.0483, |
| "step": 670 |
| }, |
| { |
| "epoch": 5.87, |
| "learning_rate": 3.633333333333334e-06, |
| "loss": 0.0847, |
| "step": 675 |
| }, |
| { |
| "epoch": 5.91, |
| "learning_rate": 3.577777777777778e-06, |
| "loss": 0.0492, |
| "step": 680 |
| }, |
| { |
| "epoch": 5.96, |
| "learning_rate": 3.5222222222222223e-06, |
| "loss": 0.0515, |
| "step": 685 |
| }, |
| { |
| "epoch": 6.0, |
| "learning_rate": 3.4666666666666672e-06, |
| "loss": 0.0372, |
| "step": 690 |
| }, |
| { |
| "epoch": 6.04, |
| "learning_rate": 3.4111111111111113e-06, |
| "loss": 0.0418, |
| "step": 695 |
| }, |
| { |
| "epoch": 6.09, |
| "learning_rate": 3.3555555555555557e-06, |
| "loss": 0.0368, |
| "step": 700 |
| }, |
| { |
| "epoch": 6.13, |
| "learning_rate": 3.3000000000000006e-06, |
| "loss": 0.0476, |
| "step": 705 |
| }, |
| { |
| "epoch": 6.17, |
| "learning_rate": 3.2444444444444446e-06, |
| "loss": 0.044, |
| "step": 710 |
| }, |
| { |
| "epoch": 6.22, |
| "learning_rate": 3.188888888888889e-06, |
| "loss": 0.0591, |
| "step": 715 |
| }, |
| { |
| "epoch": 6.26, |
| "learning_rate": 3.133333333333334e-06, |
| "loss": 0.0493, |
| "step": 720 |
| }, |
| { |
| "epoch": 6.3, |
| "learning_rate": 3.077777777777778e-06, |
| "loss": 0.0306, |
| "step": 725 |
| }, |
| { |
| "epoch": 6.35, |
| "learning_rate": 3.0222222222222225e-06, |
| "loss": 0.0481, |
| "step": 730 |
| }, |
| { |
| "epoch": 6.39, |
| "learning_rate": 2.9666666666666673e-06, |
| "loss": 0.0469, |
| "step": 735 |
| }, |
| { |
| "epoch": 6.43, |
| "learning_rate": 2.9111111111111114e-06, |
| "loss": 0.0412, |
| "step": 740 |
| }, |
| { |
| "epoch": 6.48, |
| "learning_rate": 2.855555555555556e-06, |
| "loss": 0.0325, |
| "step": 745 |
| }, |
| { |
| "epoch": 6.52, |
| "learning_rate": 2.8000000000000003e-06, |
| "loss": 0.0563, |
| "step": 750 |
| }, |
| { |
| "epoch": 6.57, |
| "learning_rate": 2.7444444444444448e-06, |
| "loss": 0.0402, |
| "step": 755 |
| }, |
| { |
| "epoch": 6.61, |
| "learning_rate": 2.6888888888888892e-06, |
| "loss": 0.0781, |
| "step": 760 |
| }, |
| { |
| "epoch": 6.65, |
| "learning_rate": 2.6333333333333332e-06, |
| "loss": 0.0459, |
| "step": 765 |
| }, |
| { |
| "epoch": 6.7, |
| "learning_rate": 2.577777777777778e-06, |
| "loss": 0.0339, |
| "step": 770 |
| }, |
| { |
| "epoch": 6.74, |
| "learning_rate": 2.5222222222222226e-06, |
| "loss": 0.0388, |
| "step": 775 |
| }, |
| { |
| "epoch": 6.78, |
| "learning_rate": 2.466666666666667e-06, |
| "loss": 0.0359, |
| "step": 780 |
| }, |
| { |
| "epoch": 6.83, |
| "learning_rate": 2.411111111111111e-06, |
| "loss": 0.088, |
| "step": 785 |
| }, |
| { |
| "epoch": 6.87, |
| "learning_rate": 2.3555555555555555e-06, |
| "loss": 0.0376, |
| "step": 790 |
| }, |
| { |
| "epoch": 6.91, |
| "learning_rate": 2.3000000000000004e-06, |
| "loss": 0.0361, |
| "step": 795 |
| }, |
| { |
| "epoch": 6.96, |
| "learning_rate": 2.2444444444444445e-06, |
| "loss": 0.0489, |
| "step": 800 |
| }, |
| { |
| "epoch": 7.0, |
| "learning_rate": 2.188888888888889e-06, |
| "loss": 0.0337, |
| "step": 805 |
| }, |
| { |
| "epoch": 7.04, |
| "learning_rate": 2.133333333333334e-06, |
| "loss": 0.0488, |
| "step": 810 |
| }, |
| { |
| "epoch": 7.09, |
| "learning_rate": 2.077777777777778e-06, |
| "loss": 0.0367, |
| "step": 815 |
| }, |
| { |
| "epoch": 7.13, |
| "learning_rate": 2.0222222222222223e-06, |
| "loss": 0.0514, |
| "step": 820 |
| }, |
| { |
| "epoch": 7.17, |
| "learning_rate": 1.9666666666666668e-06, |
| "loss": 0.0274, |
| "step": 825 |
| }, |
| { |
| "epoch": 7.22, |
| "learning_rate": 1.9111111111111112e-06, |
| "loss": 0.0356, |
| "step": 830 |
| }, |
| { |
| "epoch": 7.26, |
| "learning_rate": 1.8555555555555557e-06, |
| "loss": 0.03, |
| "step": 835 |
| }, |
| { |
| "epoch": 7.3, |
| "learning_rate": 1.8000000000000001e-06, |
| "loss": 0.0339, |
| "step": 840 |
| }, |
| { |
| "epoch": 7.35, |
| "learning_rate": 1.7444444444444448e-06, |
| "loss": 0.0269, |
| "step": 845 |
| }, |
| { |
| "epoch": 7.39, |
| "learning_rate": 1.688888888888889e-06, |
| "loss": 0.0379, |
| "step": 850 |
| }, |
| { |
| "epoch": 7.43, |
| "learning_rate": 1.6333333333333335e-06, |
| "loss": 0.0584, |
| "step": 855 |
| }, |
| { |
| "epoch": 7.48, |
| "learning_rate": 1.5777777777777778e-06, |
| "loss": 0.0282, |
| "step": 860 |
| }, |
| { |
| "epoch": 7.52, |
| "learning_rate": 1.5222222222222224e-06, |
| "loss": 0.0304, |
| "step": 865 |
| }, |
| { |
| "epoch": 7.57, |
| "learning_rate": 1.4666666666666669e-06, |
| "loss": 0.0289, |
| "step": 870 |
| }, |
| { |
| "epoch": 7.61, |
| "learning_rate": 1.4111111111111111e-06, |
| "loss": 0.0393, |
| "step": 875 |
| }, |
| { |
| "epoch": 7.65, |
| "learning_rate": 1.3555555555555558e-06, |
| "loss": 0.0318, |
| "step": 880 |
| }, |
| { |
| "epoch": 7.7, |
| "learning_rate": 1.3e-06, |
| "loss": 0.0704, |
| "step": 885 |
| }, |
| { |
| "epoch": 7.74, |
| "learning_rate": 1.2444444444444445e-06, |
| "loss": 0.0261, |
| "step": 890 |
| }, |
| { |
| "epoch": 7.78, |
| "learning_rate": 1.188888888888889e-06, |
| "loss": 0.0389, |
| "step": 895 |
| }, |
| { |
| "epoch": 7.83, |
| "learning_rate": 1.1333333333333334e-06, |
| "loss": 0.031, |
| "step": 900 |
| }, |
| { |
| "epoch": 7.87, |
| "learning_rate": 1.0777777777777779e-06, |
| "loss": 0.0261, |
| "step": 905 |
| }, |
| { |
| "epoch": 7.91, |
| "learning_rate": 1.0222222222222223e-06, |
| "loss": 0.0349, |
| "step": 910 |
| }, |
| { |
| "epoch": 7.96, |
| "learning_rate": 9.666666666666668e-07, |
| "loss": 0.0326, |
| "step": 915 |
| }, |
| { |
| "epoch": 8.0, |
| "learning_rate": 9.111111111111113e-07, |
| "loss": 0.0742, |
| "step": 920 |
| }, |
| { |
| "epoch": 8.04, |
| "learning_rate": 8.555555555555556e-07, |
| "loss": 0.0343, |
| "step": 925 |
| }, |
| { |
| "epoch": 8.09, |
| "learning_rate": 8.000000000000001e-07, |
| "loss": 0.0348, |
| "step": 930 |
| }, |
| { |
| "epoch": 8.13, |
| "learning_rate": 7.444444444444444e-07, |
| "loss": 0.0293, |
| "step": 935 |
| }, |
| { |
| "epoch": 8.17, |
| "learning_rate": 6.88888888888889e-07, |
| "loss": 0.0357, |
| "step": 940 |
| }, |
| { |
| "epoch": 8.22, |
| "learning_rate": 6.333333333333334e-07, |
| "loss": 0.024, |
| "step": 945 |
| }, |
| { |
| "epoch": 8.26, |
| "learning_rate": 5.777777777777778e-07, |
| "loss": 0.0336, |
| "step": 950 |
| }, |
| { |
| "epoch": 8.3, |
| "learning_rate": 5.222222222222223e-07, |
| "loss": 0.0202, |
| "step": 955 |
| }, |
| { |
| "epoch": 8.35, |
| "learning_rate": 4.666666666666667e-07, |
| "loss": 0.0379, |
| "step": 960 |
| }, |
| { |
| "epoch": 8.39, |
| "learning_rate": 4.111111111111112e-07, |
| "loss": 0.0388, |
| "step": 965 |
| }, |
| { |
| "epoch": 8.43, |
| "learning_rate": 3.555555555555556e-07, |
| "loss": 0.0287, |
| "step": 970 |
| }, |
| { |
| "epoch": 8.48, |
| "learning_rate": 3.0000000000000004e-07, |
| "loss": 0.021, |
| "step": 975 |
| }, |
| { |
| "epoch": 8.52, |
| "learning_rate": 2.444444444444445e-07, |
| "loss": 0.0342, |
| "step": 980 |
| }, |
| { |
| "epoch": 8.57, |
| "learning_rate": 1.888888888888889e-07, |
| "loss": 0.0329, |
| "step": 985 |
| }, |
| { |
| "epoch": 8.61, |
| "learning_rate": 1.3333333333333336e-07, |
| "loss": 0.0517, |
| "step": 990 |
| }, |
| { |
| "epoch": 8.65, |
| "learning_rate": 7.777777777777778e-08, |
| "loss": 0.024, |
| "step": 995 |
| }, |
| { |
| "epoch": 8.7, |
| "learning_rate": 2.2222222222222224e-08, |
| "loss": 0.0259, |
| "step": 1000 |
| }, |
| { |
| "epoch": 8.7, |
| "step": 1000, |
| "total_flos": 1850846955153408.0, |
| "train_loss": 0.15014069636166097, |
| "train_runtime": 142.3624, |
| "train_samples_per_second": 56.195, |
| "train_steps_per_second": 7.024 |
| } |
| ], |
| "max_steps": 1000, |
| "num_train_epochs": 9, |
| "total_flos": 1850846955153408.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
