| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 20.080321285140563, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.10040160642570281, | |
| "grad_norm": 5.931593418121338, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.0153, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.20080321285140562, | |
| "grad_norm": 4.555520534515381, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7001, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.30120481927710846, | |
| "grad_norm": 3.582505941390991, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.4808, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.40160642570281124, | |
| "grad_norm": 4.382739067077637, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.4384, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5020080321285141, | |
| "grad_norm": 4.169747352600098, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.4109, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.6024096385542169, | |
| "grad_norm": 3.8426809310913086, | |
| "learning_rate": 3e-06, | |
| "loss": 0.3934, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7028112449799196, | |
| "grad_norm": 3.972766399383545, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.3788, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.8032128514056225, | |
| "grad_norm": 3.828845739364624, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.372, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9036144578313253, | |
| "grad_norm": 4.017838954925537, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.362, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.0040160642570282, | |
| "grad_norm": 2.762294054031372, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3486, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.104417670682731, | |
| "grad_norm": 3.553866386413574, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.2512, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.2048192771084336, | |
| "grad_norm": 3.850424289703369, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2581, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.3052208835341366, | |
| "grad_norm": 3.749298572540283, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.2579, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.4056224899598393, | |
| "grad_norm": 3.389807939529419, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2721, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.5060240963855422, | |
| "grad_norm": 3.3439886569976807, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.2516, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.606425702811245, | |
| "grad_norm": 3.6428940296173096, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2577, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.7068273092369477, | |
| "grad_norm": 3.5750339031219482, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.2484, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.8072289156626506, | |
| "grad_norm": 3.3576269149780273, | |
| "learning_rate": 9e-06, | |
| "loss": 0.2418, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.9076305220883534, | |
| "grad_norm": 3.252156972885132, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.247, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.0080321285140563, | |
| "grad_norm": 2.6496663093566895, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2256, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.108433734939759, | |
| "grad_norm": 3.2420847415924072, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.1409, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.208835341365462, | |
| "grad_norm": 2.1582603454589844, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.1367, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.3092369477911645, | |
| "grad_norm": 2.708833932876587, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.1284, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.4096385542168672, | |
| "grad_norm": 2.957946300506592, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.1352, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.5100401606425704, | |
| "grad_norm": 2.999298334121704, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.1346, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.610441767068273, | |
| "grad_norm": 2.4836103916168213, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.1257, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.710843373493976, | |
| "grad_norm": 2.5539424419403076, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.1359, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.8112449799196786, | |
| "grad_norm": 2.5347061157226562, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.1363, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.9116465863453813, | |
| "grad_norm": 2.7126588821411133, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.137, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.0120481927710845, | |
| "grad_norm": 1.696603775024414, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.1269, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.112449799196787, | |
| "grad_norm": 2.1536343097686768, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.0626, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.21285140562249, | |
| "grad_norm": 2.163684844970703, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.0625, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.3132530120481927, | |
| "grad_norm": 1.543696641921997, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.0633, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.4136546184738954, | |
| "grad_norm": 2.221062660217285, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.0619, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.5140562248995986, | |
| "grad_norm": 2.847231149673462, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.065, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.6144578313253013, | |
| "grad_norm": 2.0118203163146973, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.0671, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.714859437751004, | |
| "grad_norm": 1.9212677478790283, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.0656, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.8152610441767068, | |
| "grad_norm": 2.1838502883911133, | |
| "learning_rate": 9e-06, | |
| "loss": 0.0693, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.9156626506024095, | |
| "grad_norm": 2.106473445892334, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.0658, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.016064257028113, | |
| "grad_norm": 1.0204880237579346, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0619, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.016064257028113, | |
| "eval_loss": 0.3070617914199829, | |
| "eval_runtime": 2205.2781, | |
| "eval_samples_per_second": 2.44, | |
| "eval_steps_per_second": 0.153, | |
| "eval_wer": 0.21380571792118971, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.116465863453815, | |
| "grad_norm": 1.0052565336227417, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.0339, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.216867469879518, | |
| "grad_norm": 1.4552688598632812, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.033, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.317269076305221, | |
| "grad_norm": 1.3099792003631592, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.0346, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.417670682730924, | |
| "grad_norm": 1.9144798517227173, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.0343, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.518072289156627, | |
| "grad_norm": 1.7304869890213013, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.0353, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.618473895582329, | |
| "grad_norm": 1.9012507200241089, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.0338, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.718875502008032, | |
| "grad_norm": 1.9311727285385132, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.0377, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.8192771084337345, | |
| "grad_norm": 2.47721266746521, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.0376, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.919678714859438, | |
| "grad_norm": 2.12300181388855, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.0407, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.020080321285141, | |
| "grad_norm": 1.4431580305099487, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0345, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.120481927710843, | |
| "grad_norm": 1.7874999046325684, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.0204, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.220883534136546, | |
| "grad_norm": 0.6316199898719788, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.02, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.321285140562249, | |
| "grad_norm": 1.5528149604797363, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.019, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.421686746987952, | |
| "grad_norm": 0.8070461750030518, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.019, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.522088353413655, | |
| "grad_norm": 1.8495326042175293, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.0248, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.622489959839357, | |
| "grad_norm": 0.8098943829536438, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.022, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 5.72289156626506, | |
| "grad_norm": 1.6946097612380981, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.0202, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 5.823293172690763, | |
| "grad_norm": 3.2773447036743164, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.0199, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 5.923694779116466, | |
| "grad_norm": 1.4831568002700806, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.0216, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.024096385542169, | |
| "grad_norm": 0.6036717891693115, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0226, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.124497991967871, | |
| "grad_norm": 0.7774052619934082, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0138, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.224899598393574, | |
| "grad_norm": 0.9542578458786011, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.0116, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.325301204819277, | |
| "grad_norm": 0.5289311408996582, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0117, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.42570281124498, | |
| "grad_norm": 0.8694401383399963, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0126, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.526104417670683, | |
| "grad_norm": 1.1825799942016602, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0116, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 6.626506024096385, | |
| "grad_norm": 1.5035139322280884, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.0141, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 6.7269076305220885, | |
| "grad_norm": 0.9342186450958252, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.0128, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 6.827309236947791, | |
| "grad_norm": 1.0788260698318481, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.0126, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 6.927710843373494, | |
| "grad_norm": 1.3436901569366455, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0137, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.028112449799197, | |
| "grad_norm": 0.681151807308197, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0124, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.128514056224899, | |
| "grad_norm": 0.9661208987236023, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.008, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.228915662650603, | |
| "grad_norm": 1.0090680122375488, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0074, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.329317269076305, | |
| "grad_norm": 1.4430378675460815, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0079, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.429718875502008, | |
| "grad_norm": 1.111413836479187, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0085, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.530120481927711, | |
| "grad_norm": 0.9819089770317078, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0083, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 7.6305220883534135, | |
| "grad_norm": 0.962616503238678, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.008, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 7.730923694779117, | |
| "grad_norm": 0.8837612271308899, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.01, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 7.831325301204819, | |
| "grad_norm": 0.9247878789901733, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.0095, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 7.931726907630522, | |
| "grad_norm": 0.6101934909820557, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0092, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.032128514056225, | |
| "grad_norm": 0.9556750655174255, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0069, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.032128514056225, | |
| "eval_loss": 0.36700183153152466, | |
| "eval_runtime": 2231.9147, | |
| "eval_samples_per_second": 2.411, | |
| "eval_steps_per_second": 0.151, | |
| "eval_wer": 0.20452105451509006, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.132530120481928, | |
| "grad_norm": 0.511352002620697, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0051, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.23293172690763, | |
| "grad_norm": 0.762175977230072, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0058, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 0.09607477486133575, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0038, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.433734939759036, | |
| "grad_norm": 0.6300207376480103, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.0048, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 8.534136546184738, | |
| "grad_norm": 0.21270623803138733, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0091, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 8.634538152610443, | |
| "grad_norm": 0.6666727066040039, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.0058, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 8.734939759036145, | |
| "grad_norm": 1.9148590564727783, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0063, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 8.835341365461847, | |
| "grad_norm": 0.272132009267807, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.0056, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 8.93574297188755, | |
| "grad_norm": 1.3808581829071045, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0062, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.036144578313253, | |
| "grad_norm": 0.6878290772438049, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0034, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.136546184738956, | |
| "grad_norm": 0.1612684726715088, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.0034, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.236947791164658, | |
| "grad_norm": 0.4624500572681427, | |
| "learning_rate": 6e-06, | |
| "loss": 0.003, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.337349397590362, | |
| "grad_norm": 0.5339928865432739, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0024, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 9.437751004016064, | |
| "grad_norm": 0.4839150905609131, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0033, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 9.538152610441767, | |
| "grad_norm": 0.11668159067630768, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.0033, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 9.638554216867469, | |
| "grad_norm": 0.2360651195049286, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0031, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 9.738955823293173, | |
| "grad_norm": 0.4457962214946747, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0025, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 9.839357429718875, | |
| "grad_norm": 0.41899001598358154, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0031, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 9.939759036144578, | |
| "grad_norm": 0.4257371425628662, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0052, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.040160642570282, | |
| "grad_norm": 0.4485076367855072, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0025, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.140562248995984, | |
| "grad_norm": 0.30146104097366333, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0021, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.240963855421686, | |
| "grad_norm": 0.13998575508594513, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.0025, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.34136546184739, | |
| "grad_norm": 1.0045596361160278, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0023, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 10.441767068273093, | |
| "grad_norm": 0.07078930735588074, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0017, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 10.542168674698795, | |
| "grad_norm": 0.042963068932294846, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.0019, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 10.642570281124499, | |
| "grad_norm": 0.13671617209911346, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.0018, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 10.742971887550201, | |
| "grad_norm": 0.6172053217887878, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0048, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 10.843373493975903, | |
| "grad_norm": 2.069375991821289, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0022, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 10.943775100401606, | |
| "grad_norm": 0.2148715853691101, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0021, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.04417670682731, | |
| "grad_norm": 0.045111846178770065, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0012, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.144578313253012, | |
| "grad_norm": 0.03575390577316284, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.244979919678714, | |
| "grad_norm": 0.036444906145334244, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0012, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 11.345381526104418, | |
| "grad_norm": 0.18367384374141693, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0015, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 11.44578313253012, | |
| "grad_norm": 0.03629644960165024, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0011, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 11.546184738955823, | |
| "grad_norm": 0.08484747260808945, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0011, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 11.646586345381525, | |
| "grad_norm": 0.2961013913154602, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0011, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 11.74698795180723, | |
| "grad_norm": 0.04454226791858673, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0019, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 11.847389558232932, | |
| "grad_norm": 0.06762082129716873, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.001, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 11.947791164658634, | |
| "grad_norm": 0.051877710968256, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0019, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.048192771084338, | |
| "grad_norm": 0.04771376773715019, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0013, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.048192771084338, | |
| "eval_loss": 0.3645249009132385, | |
| "eval_runtime": 2193.5952, | |
| "eval_samples_per_second": 2.453, | |
| "eval_steps_per_second": 0.154, | |
| "eval_wer": 0.19776134239930018, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.14859437751004, | |
| "grad_norm": 0.020401790738105774, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.0007, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 12.248995983935743, | |
| "grad_norm": 0.037684116512537, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0006, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 12.349397590361447, | |
| "grad_norm": 0.025216449052095413, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 12.449799196787149, | |
| "grad_norm": 0.021326890215277672, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0005, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 12.550200803212851, | |
| "grad_norm": 0.02904532290995121, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0009, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 12.650602409638553, | |
| "grad_norm": 0.15470072627067566, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0006, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 12.751004016064257, | |
| "grad_norm": 0.5353085398674011, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0008, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 12.85140562248996, | |
| "grad_norm": 0.1263090819120407, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0005, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 12.951807228915662, | |
| "grad_norm": 0.018346522003412247, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.0005, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 13.052208835341366, | |
| "grad_norm": 0.012502867728471756, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0005, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 13.152610441767068, | |
| "grad_norm": 0.035849809646606445, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.0005, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 13.25301204819277, | |
| "grad_norm": 0.013340278528630733, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 13.353413654618475, | |
| "grad_norm": 0.01812613196671009, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.0004, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 13.453815261044177, | |
| "grad_norm": 0.016993574798107147, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.0004, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 13.55421686746988, | |
| "grad_norm": 0.012604492716491222, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0004, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 13.654618473895582, | |
| "grad_norm": 0.013149112462997437, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0007, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 13.755020080321286, | |
| "grad_norm": 0.01593812368810177, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0004, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 13.855421686746988, | |
| "grad_norm": 0.01589050143957138, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 13.95582329317269, | |
| "grad_norm": 0.014971195720136166, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0004, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 14.056224899598394, | |
| "grad_norm": 0.01241573691368103, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0003, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 14.156626506024097, | |
| "grad_norm": 0.011940378695726395, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 14.257028112449799, | |
| "grad_norm": 0.01354218740016222, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0003, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 14.357429718875501, | |
| "grad_norm": 0.011458562687039375, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 14.457831325301205, | |
| "grad_norm": 0.010744108818471432, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0003, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 14.558232931726907, | |
| "grad_norm": 0.01174489688128233, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 14.65863453815261, | |
| "grad_norm": 0.01333660539239645, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0003, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 14.759036144578314, | |
| "grad_norm": 0.012421938590705395, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 14.859437751004016, | |
| "grad_norm": 0.01224998664110899, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 14.959839357429718, | |
| "grad_norm": 0.015473966486752033, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0003, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 15.060240963855422, | |
| "grad_norm": 0.012373683042824268, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0003, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 15.160642570281125, | |
| "grad_norm": 0.010334338992834091, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0004, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 15.261044176706827, | |
| "grad_norm": 0.015428266488015652, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 15.36144578313253, | |
| "grad_norm": 0.011229559779167175, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0003, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 15.461847389558233, | |
| "grad_norm": 0.009039835073053837, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0003, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 15.562248995983936, | |
| "grad_norm": 0.0124340346083045, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 15.662650602409638, | |
| "grad_norm": 0.009659750387072563, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 15.763052208835342, | |
| "grad_norm": 0.013714014552533627, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 15.863453815261044, | |
| "grad_norm": 0.00784530583769083, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0003, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 15.963855421686747, | |
| "grad_norm": 0.010499561205506325, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 16.06425702811245, | |
| "grad_norm": 0.01295757107436657, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.06425702811245, | |
| "eval_loss": 0.4106931984424591, | |
| "eval_runtime": 2194.2451, | |
| "eval_samples_per_second": 2.452, | |
| "eval_steps_per_second": 0.154, | |
| "eval_wer": 0.19143902342041433, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.164658634538153, | |
| "grad_norm": 0.010303654707968235, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 16.265060240963855, | |
| "grad_norm": 0.008325839415192604, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 16.365461847389557, | |
| "grad_norm": 0.00958819966763258, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0003, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 16.46586345381526, | |
| "grad_norm": 0.008696039207279682, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0003, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 16.566265060240966, | |
| "grad_norm": 0.010361140593886375, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 16.666666666666668, | |
| "grad_norm": 0.008520281873643398, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 16.76706827309237, | |
| "grad_norm": 0.012094419449567795, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0004, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 16.867469879518072, | |
| "grad_norm": 0.00969509407877922, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 16.967871485943775, | |
| "grad_norm": 0.010107293725013733, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0003, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 17.068273092369477, | |
| "grad_norm": 0.008444724604487419, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 17.16867469879518, | |
| "grad_norm": 0.008527095429599285, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0003, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 17.269076305220885, | |
| "grad_norm": 0.007767422124743462, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0003, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 17.369477911646587, | |
| "grad_norm": 0.008298359811306, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 17.46987951807229, | |
| "grad_norm": 0.00869645643979311, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 17.570281124497992, | |
| "grad_norm": 0.009767497889697552, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0002, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 17.670682730923694, | |
| "grad_norm": 0.009676006622612476, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 17.771084337349397, | |
| "grad_norm": 0.010664808563888073, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0003, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 17.8714859437751, | |
| "grad_norm": 0.011917660012841225, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 17.971887550200805, | |
| "grad_norm": 0.008082253858447075, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0002, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 18.072289156626507, | |
| "grad_norm": 0.008449643850326538, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 18.17269076305221, | |
| "grad_norm": 0.00893787294626236, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0002, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 18.27309236947791, | |
| "grad_norm": 0.007984068244695663, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0003, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 18.373493975903614, | |
| "grad_norm": 0.008762707002460957, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0002, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 18.473895582329316, | |
| "grad_norm": 0.008354073390364647, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 18.57429718875502, | |
| "grad_norm": 0.010145510546863079, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0002, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 18.674698795180724, | |
| "grad_norm": 0.009271888993680477, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0002, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 18.775100401606426, | |
| "grad_norm": 0.007404220290482044, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0002, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 18.87550200803213, | |
| "grad_norm": 0.00828209612518549, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 18.97590361445783, | |
| "grad_norm": 0.008047865703701973, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0002, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 19.076305220883533, | |
| "grad_norm": 0.00727940583601594, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0002, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 19.176706827309236, | |
| "grad_norm": 0.008462085388600826, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0002, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 19.27710843373494, | |
| "grad_norm": 0.008368249051272869, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0002, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 19.377510040160644, | |
| "grad_norm": 0.007355119101703167, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0002, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 19.477911646586346, | |
| "grad_norm": 0.00911016296595335, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0002, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 19.57831325301205, | |
| "grad_norm": 0.00813527312129736, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 19.67871485943775, | |
| "grad_norm": 0.007565053179860115, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0002, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 19.779116465863453, | |
| "grad_norm": 0.008976846002042294, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0002, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 19.879518072289155, | |
| "grad_norm": 0.007831977680325508, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0002, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 19.97991967871486, | |
| "grad_norm": 0.0071573760360479355, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 20.080321285140563, | |
| "grad_norm": 0.008769778534770012, | |
| "learning_rate": 0.0, | |
| "loss": 0.0002, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 20.080321285140563, | |
| "eval_loss": 0.41853219270706177, | |
| "eval_runtime": 2165.4621, | |
| "eval_samples_per_second": 2.485, | |
| "eval_steps_per_second": 0.156, | |
| "eval_wer": 0.19108115630840192, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 20.080321285140563, | |
| "step": 5000, | |
| "total_flos": 5.435997290496e+20, | |
| "train_loss": 0.05159526972509921, | |
| "train_runtime": 59448.8268, | |
| "train_samples_per_second": 2.691, | |
| "train_steps_per_second": 0.084 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 21, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.435997290496e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |