| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 22486, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.990957336416734e-07, | |
| "loss": 0.1541, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.981469951673633e-07, | |
| "loss": 0.0788, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.971982566930536e-07, | |
| "loss": 0.2255, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.962495182187434e-07, | |
| "loss": 0.1454, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.953007797444335e-07, | |
| "loss": 0.081, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.943520412701236e-07, | |
| "loss": 0.1782, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.934033027958137e-07, | |
| "loss": 0.1119, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.924545643215036e-07, | |
| "loss": 0.1648, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.915058258471937e-07, | |
| "loss": 0.1536, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.905570873728838e-07, | |
| "loss": 0.1527, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.896083488985739e-07, | |
| "loss": 0.1153, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.88659610424264e-07, | |
| "loss": 0.0826, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.87710871949954e-07, | |
| "loss": 0.135, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.867621334756441e-07, | |
| "loss": 0.1657, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.85813395001334e-07, | |
| "loss": 0.1777, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.848646565270243e-07, | |
| "loss": 0.0975, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.839159180527142e-07, | |
| "loss": 0.1336, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.829671795784043e-07, | |
| "loss": 0.0785, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.820184411040944e-07, | |
| "loss": 0.164, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.810697026297845e-07, | |
| "loss": 0.0864, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.801209641554746e-07, | |
| "loss": 0.128, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.791722256811644e-07, | |
| "loss": 0.0897, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.782234872068545e-07, | |
| "loss": 0.0989, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.772747487325446e-07, | |
| "loss": 0.1309, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.763408342968958e-07, | |
| "loss": 0.2144, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.75392095822586e-07, | |
| "loss": 0.0691, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.744433573482758e-07, | |
| "loss": 0.1476, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.73494618873966e-07, | |
| "loss": 0.1541, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.72545880399656e-07, | |
| "loss": 0.1108, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.71597141925346e-07, | |
| "loss": 0.112, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.706484034510362e-07, | |
| "loss": 0.1901, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.696996649767262e-07, | |
| "loss": 0.1207, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.687509265024163e-07, | |
| "loss": 0.128, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.678021880281062e-07, | |
| "loss": 0.2029, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.668534495537965e-07, | |
| "loss": 0.1629, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.659047110794864e-07, | |
| "loss": 0.1662, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.649559726051765e-07, | |
| "loss": 0.1392, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.640072341308666e-07, | |
| "loss": 0.081, | |
| "step": 2432 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.630584956565567e-07, | |
| "loss": 0.0894, | |
| "step": 2496 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.621097571822468e-07, | |
| "loss": 0.1017, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.611610187079368e-07, | |
| "loss": 0.1476, | |
| "step": 2624 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.602122802336267e-07, | |
| "loss": 0.1592, | |
| "step": 2688 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.592635417593168e-07, | |
| "loss": 0.146, | |
| "step": 2752 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.58314803285007e-07, | |
| "loss": 0.1552, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.57366064810697e-07, | |
| "loss": 0.0498, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.56417326336387e-07, | |
| "loss": 0.0999, | |
| "step": 2944 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.55468587862077e-07, | |
| "loss": 0.1196, | |
| "step": 3008 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.545198493877673e-07, | |
| "loss": 0.1172, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.535711109134573e-07, | |
| "loss": 0.1107, | |
| "step": 3136 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.526223724391472e-07, | |
| "loss": 0.182, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.516736339648373e-07, | |
| "loss": 0.0687, | |
| "step": 3264 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.507248954905273e-07, | |
| "loss": 0.1274, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.497761570162175e-07, | |
| "loss": 0.1374, | |
| "step": 3392 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.488274185419075e-07, | |
| "loss": 0.159, | |
| "step": 3456 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.478786800675975e-07, | |
| "loss": 0.0687, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.469299415932877e-07, | |
| "loss": 0.1446, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.459812031189777e-07, | |
| "loss": 0.1047, | |
| "step": 3648 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.450324646446678e-07, | |
| "loss": 0.1445, | |
| "step": 3712 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.440837261703578e-07, | |
| "loss": 0.1033, | |
| "step": 3776 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.431349876960478e-07, | |
| "loss": 0.1514, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.421862492217379e-07, | |
| "loss": 0.1245, | |
| "step": 3904 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.41237510747428e-07, | |
| "loss": 0.2032, | |
| "step": 3968 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.402887722731181e-07, | |
| "loss": 0.0935, | |
| "step": 4032 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.393400337988081e-07, | |
| "loss": 0.1492, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.383912953244981e-07, | |
| "loss": 0.1011, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.374425568501883e-07, | |
| "loss": 0.1279, | |
| "step": 4224 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.364938183758783e-07, | |
| "loss": 0.0842, | |
| "step": 4288 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.355599039402295e-07, | |
| "loss": 0.1482, | |
| "step": 4352 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.346111654659194e-07, | |
| "loss": 0.1444, | |
| "step": 4416 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.336624269916095e-07, | |
| "loss": 0.1124, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.327285125559607e-07, | |
| "loss": 0.1552, | |
| "step": 4544 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.317797740816507e-07, | |
| "loss": 0.113, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.308310356073408e-07, | |
| "loss": 0.1406, | |
| "step": 4672 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.298822971330309e-07, | |
| "loss": 0.1464, | |
| "step": 4736 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.289335586587209e-07, | |
| "loss": 0.0889, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.279848201844111e-07, | |
| "loss": 0.1028, | |
| "step": 4864 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.270360817101011e-07, | |
| "loss": 0.1816, | |
| "step": 4928 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.26087343235791e-07, | |
| "loss": 0.0867, | |
| "step": 4992 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.251386047614812e-07, | |
| "loss": 0.1426, | |
| "step": 5056 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.241898662871712e-07, | |
| "loss": 0.1404, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.232411278128613e-07, | |
| "loss": 0.0599, | |
| "step": 5184 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.222923893385513e-07, | |
| "loss": 0.0753, | |
| "step": 5248 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.213436508642414e-07, | |
| "loss": 0.1109, | |
| "step": 5312 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.203949123899315e-07, | |
| "loss": 0.1836, | |
| "step": 5376 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.194461739156215e-07, | |
| "loss": 0.0633, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.184974354413117e-07, | |
| "loss": 0.1392, | |
| "step": 5504 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.175486969670017e-07, | |
| "loss": 0.11, | |
| "step": 5568 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.165999584926916e-07, | |
| "loss": 0.1901, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.156512200183818e-07, | |
| "loss": 0.1399, | |
| "step": 5696 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.147024815440718e-07, | |
| "loss": 0.1312, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.137537430697619e-07, | |
| "loss": 0.0884, | |
| "step": 5824 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.12805004595452e-07, | |
| "loss": 0.1663, | |
| "step": 5888 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.11856266121142e-07, | |
| "loss": 0.0728, | |
| "step": 5952 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.109075276468321e-07, | |
| "loss": 0.1709, | |
| "step": 6016 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.099587891725221e-07, | |
| "loss": 0.1476, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.090100506982122e-07, | |
| "loss": 0.0778, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.080613122239022e-07, | |
| "loss": 0.1081, | |
| "step": 6208 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.071125737495922e-07, | |
| "loss": 0.1059, | |
| "step": 6272 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.061638352752824e-07, | |
| "loss": 0.1143, | |
| "step": 6336 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.052150968009724e-07, | |
| "loss": 0.1205, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.042663583266624e-07, | |
| "loss": 0.1479, | |
| "step": 6464 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.033176198523526e-07, | |
| "loss": 0.0817, | |
| "step": 6528 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.023688813780426e-07, | |
| "loss": 0.1398, | |
| "step": 6592 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.014201429037327e-07, | |
| "loss": 0.1608, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.004714044294228e-07, | |
| "loss": 0.1345, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.995226659551127e-07, | |
| "loss": 0.1773, | |
| "step": 6784 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.985739274808028e-07, | |
| "loss": 0.1769, | |
| "step": 6848 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.976251890064928e-07, | |
| "loss": 0.1345, | |
| "step": 6912 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.96676450532183e-07, | |
| "loss": 0.1293, | |
| "step": 6976 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.95727712057873e-07, | |
| "loss": 0.1388, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.94778973583563e-07, | |
| "loss": 0.1004, | |
| "step": 7104 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.938302351092532e-07, | |
| "loss": 0.1363, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.928814966349432e-07, | |
| "loss": 0.0933, | |
| "step": 7232 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.919327581606333e-07, | |
| "loss": 0.1617, | |
| "step": 7296 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.909840196863233e-07, | |
| "loss": 0.1085, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.900352812120133e-07, | |
| "loss": 0.1836, | |
| "step": 7424 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.890865427377034e-07, | |
| "loss": 0.1464, | |
| "step": 7488 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.881378042633935e-07, | |
| "loss": 0.1166, | |
| "step": 7552 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.871890657890836e-07, | |
| "loss": 0.1071, | |
| "step": 7616 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.862403273147736e-07, | |
| "loss": 0.1564, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.852915888404636e-07, | |
| "loss": 0.0937, | |
| "step": 7744 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.843428503661538e-07, | |
| "loss": 0.1095, | |
| "step": 7808 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.833941118918438e-07, | |
| "loss": 0.1313, | |
| "step": 7872 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.824453734175337e-07, | |
| "loss": 0.0819, | |
| "step": 7936 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.814966349432239e-07, | |
| "loss": 0.1273, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.805478964689139e-07, | |
| "loss": 0.1079, | |
| "step": 8064 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.79599157994604e-07, | |
| "loss": 0.0566, | |
| "step": 8128 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.786504195202941e-07, | |
| "loss": 0.0882, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.777016810459841e-07, | |
| "loss": 0.1801, | |
| "step": 8256 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.767529425716742e-07, | |
| "loss": 0.1006, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.758042040973643e-07, | |
| "loss": 0.1033, | |
| "step": 8384 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.748554656230544e-07, | |
| "loss": 0.0907, | |
| "step": 8448 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.739067271487443e-07, | |
| "loss": 0.0548, | |
| "step": 8512 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.729579886744343e-07, | |
| "loss": 0.1173, | |
| "step": 8576 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.720092502001245e-07, | |
| "loss": 0.1415, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.710605117258145e-07, | |
| "loss": 0.1035, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.701117732515046e-07, | |
| "loss": 0.1016, | |
| "step": 8768 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.691630347771947e-07, | |
| "loss": 0.111, | |
| "step": 8832 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.682291203415458e-07, | |
| "loss": 0.1427, | |
| "step": 8896 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.672803818672359e-07, | |
| "loss": 0.1181, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.66331643392926e-07, | |
| "loss": 0.097, | |
| "step": 9024 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.65382904918616e-07, | |
| "loss": 0.1367, | |
| "step": 9088 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.64434166444306e-07, | |
| "loss": 0.1212, | |
| "step": 9152 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.634854279699961e-07, | |
| "loss": 0.1393, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.625366894956861e-07, | |
| "loss": 0.1322, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.615879510213763e-07, | |
| "loss": 0.0786, | |
| "step": 9344 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.606392125470663e-07, | |
| "loss": 0.1263, | |
| "step": 9408 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.596904740727563e-07, | |
| "loss": 0.1537, | |
| "step": 9472 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.587417355984465e-07, | |
| "loss": 0.1267, | |
| "step": 9536 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.577929971241365e-07, | |
| "loss": 0.1087, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.568442586498266e-07, | |
| "loss": 0.1097, | |
| "step": 9664 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.558955201755165e-07, | |
| "loss": 0.1423, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.549467817012066e-07, | |
| "loss": 0.0911, | |
| "step": 9792 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.539980432268967e-07, | |
| "loss": 0.0748, | |
| "step": 9856 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.530493047525867e-07, | |
| "loss": 0.112, | |
| "step": 9920 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.521005662782769e-07, | |
| "loss": 0.1765, | |
| "step": 9984 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.511518278039669e-07, | |
| "loss": 0.1539, | |
| "step": 10048 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.502030893296569e-07, | |
| "loss": 0.1382, | |
| "step": 10112 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.492543508553471e-07, | |
| "loss": 0.1146, | |
| "step": 10176 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.48305612381037e-07, | |
| "loss": 0.1589, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.473568739067271e-07, | |
| "loss": 0.1184, | |
| "step": 10304 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.464081354324172e-07, | |
| "loss": 0.0892, | |
| "step": 10368 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.454593969581072e-07, | |
| "loss": 0.1883, | |
| "step": 10432 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.445106584837973e-07, | |
| "loss": 0.1441, | |
| "step": 10496 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.435619200094873e-07, | |
| "loss": 0.1233, | |
| "step": 10560 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.426131815351774e-07, | |
| "loss": 0.1376, | |
| "step": 10624 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.416644430608675e-07, | |
| "loss": 0.057, | |
| "step": 10688 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.407157045865575e-07, | |
| "loss": 0.0966, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.397669661122477e-07, | |
| "loss": 0.1026, | |
| "step": 10816 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.388182276379376e-07, | |
| "loss": 0.1161, | |
| "step": 10880 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.378694891636276e-07, | |
| "loss": 0.1889, | |
| "step": 10944 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.369207506893178e-07, | |
| "loss": 0.0808, | |
| "step": 11008 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.359868362536689e-07, | |
| "loss": 0.8512, | |
| "step": 11072 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.35038097779359e-07, | |
| "loss": 0.1572, | |
| "step": 11136 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.34089359305049e-07, | |
| "loss": 0.1883, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.331406208307391e-07, | |
| "loss": 0.0772, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.321918823564291e-07, | |
| "loss": 0.0924, | |
| "step": 11328 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.312431438821193e-07, | |
| "loss": 0.1308, | |
| "step": 11392 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.302944054078093e-07, | |
| "loss": 0.2317, | |
| "step": 11456 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.293456669334992e-07, | |
| "loss": 0.1581, | |
| "step": 11520 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.283969284591894e-07, | |
| "loss": 0.1068, | |
| "step": 11584 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.274481899848794e-07, | |
| "loss": 0.0793, | |
| "step": 11648 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.264994515105695e-07, | |
| "loss": 0.1407, | |
| "step": 11712 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.255507130362596e-07, | |
| "loss": 0.2219, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.246019745619496e-07, | |
| "loss": 0.1364, | |
| "step": 11840 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.236532360876397e-07, | |
| "loss": 0.0842, | |
| "step": 11904 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.227193216519909e-07, | |
| "loss": 0.0893, | |
| "step": 11968 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.217705831776809e-07, | |
| "loss": 0.1798, | |
| "step": 12032 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.20821844703371e-07, | |
| "loss": 0.1263, | |
| "step": 12096 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.19873106229061e-07, | |
| "loss": 0.1289, | |
| "step": 12160 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.18924367754751e-07, | |
| "loss": 0.1712, | |
| "step": 12224 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.179756292804412e-07, | |
| "loss": 0.2207, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.170268908061312e-07, | |
| "loss": 0.1219, | |
| "step": 12352 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.160781523318212e-07, | |
| "loss": 0.1446, | |
| "step": 12416 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.151294138575113e-07, | |
| "loss": 0.076, | |
| "step": 12480 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.141806753832014e-07, | |
| "loss": 0.1448, | |
| "step": 12544 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.132319369088915e-07, | |
| "loss": 0.1613, | |
| "step": 12608 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.122831984345814e-07, | |
| "loss": 0.1241, | |
| "step": 12672 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.113344599602715e-07, | |
| "loss": 0.1833, | |
| "step": 12736 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.103857214859616e-07, | |
| "loss": 0.2071, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.094369830116516e-07, | |
| "loss": 0.1467, | |
| "step": 12864 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.084882445373418e-07, | |
| "loss": 0.0889, | |
| "step": 12928 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.075395060630318e-07, | |
| "loss": 0.141, | |
| "step": 12992 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.065907675887218e-07, | |
| "loss": 0.106, | |
| "step": 13056 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.05642029114412e-07, | |
| "loss": 0.1359, | |
| "step": 13120 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.04693290640102e-07, | |
| "loss": 0.2273, | |
| "step": 13184 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.03744552165792e-07, | |
| "loss": 0.1071, | |
| "step": 13248 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.02795813691482e-07, | |
| "loss": 0.1106, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.018470752171721e-07, | |
| "loss": 0.1035, | |
| "step": 13376 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.008983367428622e-07, | |
| "loss": 0.1459, | |
| "step": 13440 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.999495982685522e-07, | |
| "loss": 0.1739, | |
| "step": 13504 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.990008597942423e-07, | |
| "loss": 0.116, | |
| "step": 13568 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.980521213199324e-07, | |
| "loss": 0.1017, | |
| "step": 13632 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.971033828456224e-07, | |
| "loss": 0.1414, | |
| "step": 13696 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.961546443713126e-07, | |
| "loss": 0.2619, | |
| "step": 13760 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.952059058970026e-07, | |
| "loss": 0.1166, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.942571674226925e-07, | |
| "loss": 0.142, | |
| "step": 13888 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.933084289483827e-07, | |
| "loss": 0.2068, | |
| "step": 13952 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.923596904740727e-07, | |
| "loss": 0.1119, | |
| "step": 14016 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.914109519997628e-07, | |
| "loss": 0.143, | |
| "step": 14080 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.904770375641139e-07, | |
| "loss": 0.1718, | |
| "step": 14144 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.89528299089804e-07, | |
| "loss": 0.0871, | |
| "step": 14208 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.88579560615494e-07, | |
| "loss": 0.1059, | |
| "step": 14272 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.876308221411842e-07, | |
| "loss": 0.1276, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.866820836668742e-07, | |
| "loss": 0.1503, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.857333451925641e-07, | |
| "loss": 0.2201, | |
| "step": 14464 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.847846067182543e-07, | |
| "loss": 0.1737, | |
| "step": 14528 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.838358682439443e-07, | |
| "loss": 0.1133, | |
| "step": 14592 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.828871297696344e-07, | |
| "loss": 0.1378, | |
| "step": 14656 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.819383912953245e-07, | |
| "loss": 0.166, | |
| "step": 14720 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.809896528210145e-07, | |
| "loss": 0.2001, | |
| "step": 14784 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.800409143467046e-07, | |
| "loss": 0.158, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.790921758723946e-07, | |
| "loss": 0.1009, | |
| "step": 14912 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.781434373980848e-07, | |
| "loss": 0.2009, | |
| "step": 14976 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.771946989237747e-07, | |
| "loss": 0.1437, | |
| "step": 15040 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.762459604494647e-07, | |
| "loss": 0.1653, | |
| "step": 15104 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.752972219751549e-07, | |
| "loss": 0.1736, | |
| "step": 15168 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.743484835008449e-07, | |
| "loss": 0.2137, | |
| "step": 15232 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.73399745026535e-07, | |
| "loss": 0.1282, | |
| "step": 15296 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.724510065522251e-07, | |
| "loss": 0.1766, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.715022680779151e-07, | |
| "loss": 0.0876, | |
| "step": 15424 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.705535296036052e-07, | |
| "loss": 0.1338, | |
| "step": 15488 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.696047911292953e-07, | |
| "loss": 0.2006, | |
| "step": 15552 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.686560526549854e-07, | |
| "loss": 0.1505, | |
| "step": 15616 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.677073141806753e-07, | |
| "loss": 0.1223, | |
| "step": 15680 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.667585757063653e-07, | |
| "loss": 0.1693, | |
| "step": 15744 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.658098372320555e-07, | |
| "loss": 0.1872, | |
| "step": 15808 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.648610987577455e-07, | |
| "loss": 0.1109, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.639123602834355e-07, | |
| "loss": 0.1351, | |
| "step": 15936 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.629636218091257e-07, | |
| "loss": 0.1336, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.620148833348157e-07, | |
| "loss": 0.2153, | |
| "step": 16064 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.610661448605058e-07, | |
| "loss": 0.2274, | |
| "step": 16128 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.601174063861959e-07, | |
| "loss": 0.105, | |
| "step": 16192 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.591686679118858e-07, | |
| "loss": 0.1016, | |
| "step": 16256 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.582199294375759e-07, | |
| "loss": 0.116, | |
| "step": 16320 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.57271190963266e-07, | |
| "loss": 0.1086, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.563224524889561e-07, | |
| "loss": 0.1153, | |
| "step": 16448 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.553737140146461e-07, | |
| "loss": 0.1044, | |
| "step": 16512 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.544249755403361e-07, | |
| "loss": 0.2228, | |
| "step": 16576 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.534762370660263e-07, | |
| "loss": 0.1002, | |
| "step": 16640 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.525274985917163e-07, | |
| "loss": 0.1919, | |
| "step": 16704 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.515787601174063e-07, | |
| "loss": 0.13, | |
| "step": 16768 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.506300216430964e-07, | |
| "loss": 0.1165, | |
| "step": 16832 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.496812831687864e-07, | |
| "loss": 0.1385, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.487325446944765e-07, | |
| "loss": 0.1836, | |
| "step": 16960 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.477838062201666e-07, | |
| "loss": 0.1852, | |
| "step": 17024 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.468498917845177e-07, | |
| "loss": 0.1721, | |
| "step": 17088 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.459011533102078e-07, | |
| "loss": 0.1164, | |
| "step": 17152 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.449524148358979e-07, | |
| "loss": 0.0989, | |
| "step": 17216 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.440036763615879e-07, | |
| "loss": 0.1524, | |
| "step": 17280 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.430549378872781e-07, | |
| "loss": 0.174, | |
| "step": 17344 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.42106199412968e-07, | |
| "loss": 0.1509, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.41157460938658e-07, | |
| "loss": 0.1264, | |
| "step": 17472 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.402087224643482e-07, | |
| "loss": 0.1148, | |
| "step": 17536 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.392599839900382e-07, | |
| "loss": 0.1227, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.383112455157283e-07, | |
| "loss": 0.1462, | |
| "step": 17664 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.373625070414183e-07, | |
| "loss": 0.1673, | |
| "step": 17728 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.364137685671084e-07, | |
| "loss": 0.1274, | |
| "step": 17792 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.354650300927985e-07, | |
| "loss": 0.162, | |
| "step": 17856 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.345162916184885e-07, | |
| "loss": 0.148, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.335675531441787e-07, | |
| "loss": 0.157, | |
| "step": 17984 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.326188146698686e-07, | |
| "loss": 0.1419, | |
| "step": 18048 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.316700761955586e-07, | |
| "loss": 0.1916, | |
| "step": 18112 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.307213377212488e-07, | |
| "loss": 0.1772, | |
| "step": 18176 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.297725992469388e-07, | |
| "loss": 0.1243, | |
| "step": 18240 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.288238607726288e-07, | |
| "loss": 0.1738, | |
| "step": 18304 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.27875122298319e-07, | |
| "loss": 0.165, | |
| "step": 18368 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.26926383824009e-07, | |
| "loss": 0.0869, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.259776453496991e-07, | |
| "loss": 0.1613, | |
| "step": 18496 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.25028906875389e-07, | |
| "loss": 0.1646, | |
| "step": 18560 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.240801684010791e-07, | |
| "loss": 0.1232, | |
| "step": 18624 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.231314299267692e-07, | |
| "loss": 0.1695, | |
| "step": 18688 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.221826914524592e-07, | |
| "loss": 0.1305, | |
| "step": 18752 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.212339529781494e-07, | |
| "loss": 0.2047, | |
| "step": 18816 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.202852145038394e-07, | |
| "loss": 0.1881, | |
| "step": 18880 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.193364760295294e-07, | |
| "loss": 0.1513, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.183877375552196e-07, | |
| "loss": 0.1355, | |
| "step": 19008 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.174389990809096e-07, | |
| "loss": 0.1733, | |
| "step": 19072 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.164902606065996e-07, | |
| "loss": 0.1647, | |
| "step": 19136 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.155415221322897e-07, | |
| "loss": 0.132, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.145927836579797e-07, | |
| "loss": 0.2464, | |
| "step": 19264 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.136440451836698e-07, | |
| "loss": 0.1954, | |
| "step": 19328 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.126953067093598e-07, | |
| "loss": 0.1751, | |
| "step": 19392 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.1174656823505e-07, | |
| "loss": 0.1182, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.1079782976074e-07, | |
| "loss": 0.0678, | |
| "step": 19520 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.0984909128643e-07, | |
| "loss": 0.1656, | |
| "step": 19584 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.089003528121202e-07, | |
| "loss": 0.219, | |
| "step": 19648 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.079516143378101e-07, | |
| "loss": 0.15, | |
| "step": 19712 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.070028758635002e-07, | |
| "loss": 0.1845, | |
| "step": 19776 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.060541373891903e-07, | |
| "loss": 0.1766, | |
| "step": 19840 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.051053989148803e-07, | |
| "loss": 0.1355, | |
| "step": 19904 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.041566604405704e-07, | |
| "loss": 0.1268, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.032079219662605e-07, | |
| "loss": 0.191, | |
| "step": 20032 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.022591834919506e-07, | |
| "loss": 0.2032, | |
| "step": 20096 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.013104450176406e-07, | |
| "loss": 0.2632, | |
| "step": 20160 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.003617065433306e-07, | |
| "loss": 0.2028, | |
| "step": 20224 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.994129680690208e-07, | |
| "loss": 0.135, | |
| "step": 20288 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.984642295947107e-07, | |
| "loss": 0.1635, | |
| "step": 20352 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.975154911204007e-07, | |
| "loss": 0.1986, | |
| "step": 20416 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.965667526460909e-07, | |
| "loss": 0.1786, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.956180141717809e-07, | |
| "loss": 0.116, | |
| "step": 20544 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.94669275697471e-07, | |
| "loss": 0.2206, | |
| "step": 20608 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.937205372231611e-07, | |
| "loss": 0.0995, | |
| "step": 20672 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.927717987488511e-07, | |
| "loss": 0.1634, | |
| "step": 20736 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.918230602745412e-07, | |
| "loss": 0.1429, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.908743218002312e-07, | |
| "loss": 0.2244, | |
| "step": 20864 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.899255833259213e-07, | |
| "loss": 0.1666, | |
| "step": 20928 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.889768448516113e-07, | |
| "loss": 0.1413, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.880281063773013e-07, | |
| "loss": 0.1824, | |
| "step": 21056 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.870941919416525e-07, | |
| "loss": 0.1737, | |
| "step": 21120 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.861454534673426e-07, | |
| "loss": 0.171, | |
| "step": 21184 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.851967149930327e-07, | |
| "loss": 0.1571, | |
| "step": 21248 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.842479765187227e-07, | |
| "loss": 0.115, | |
| "step": 21312 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.832992380444128e-07, | |
| "loss": 0.2192, | |
| "step": 21376 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.823504995701029e-07, | |
| "loss": 0.146, | |
| "step": 21440 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.81401761095793e-07, | |
| "loss": 0.1586, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.804530226214829e-07, | |
| "loss": 0.1035, | |
| "step": 21568 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.795191081858341e-07, | |
| "loss": 0.1898, | |
| "step": 21632 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.785703697115241e-07, | |
| "loss": 0.2901, | |
| "step": 21696 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.776216312372143e-07, | |
| "loss": 0.1986, | |
| "step": 21760 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.766728927629043e-07, | |
| "loss": 0.1774, | |
| "step": 21824 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.757241542885943e-07, | |
| "loss": 0.1945, | |
| "step": 21888 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.747754158142845e-07, | |
| "loss": 0.1453, | |
| "step": 21952 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.738266773399745e-07, | |
| "loss": 0.1882, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.728779388656646e-07, | |
| "loss": 0.2266, | |
| "step": 22080 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.719292003913545e-07, | |
| "loss": 0.2192, | |
| "step": 22144 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.709804619170446e-07, | |
| "loss": 0.1191, | |
| "step": 22208 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.700317234427347e-07, | |
| "loss": 0.1323, | |
| "step": 22272 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.690829849684247e-07, | |
| "loss": 0.2345, | |
| "step": 22336 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.681342464941149e-07, | |
| "loss": 0.122, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.671855080198049e-07, | |
| "loss": 0.1643, | |
| "step": 22464 | |
| } | |
| ], | |
| "logging_steps": 64, | |
| "max_steps": 67458, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 22486, | |
| "total_flos": 4.77416235737088e+19, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |