| { | |
| "best_metric": 6.049679487179487, | |
| "best_model_checkpoint": "./exp/whisper-small-taiwanese-asr-v2/checkpoint-7000", | |
| "epoch": 22.675736961451246, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05668934240362812, | |
| "grad_norm": 39.12731170654297, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 3.459, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11337868480725624, | |
| "grad_norm": 27.20041847229004, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 3.0795, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17006802721088435, | |
| "grad_norm": 23.331146240234375, | |
| "learning_rate": 1.42e-06, | |
| "loss": 2.4576, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.22675736961451248, | |
| "grad_norm": 18.951128005981445, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.8809, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2834467120181406, | |
| "grad_norm": 18.83504295349121, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.3831, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3401360544217687, | |
| "grad_norm": 16.38509750366211, | |
| "learning_rate": 2.92e-06, | |
| "loss": 1.049, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3968253968253968, | |
| "grad_norm": 17.442344665527344, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.8571, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.45351473922902497, | |
| "grad_norm": 15.140103340148926, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.6812, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 16.342971801757812, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.5756, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5668934240362812, | |
| "grad_norm": 18.02117347717285, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.5135, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6235827664399093, | |
| "grad_norm": 19.521883010864258, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.4338, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6802721088435374, | |
| "grad_norm": 14.71985149383545, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.3755, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7369614512471655, | |
| "grad_norm": 6.9398722648620605, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.3662, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7936507936507936, | |
| "grad_norm": 8.429058074951172, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.3196, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8503401360544217, | |
| "grad_norm": 7.899880409240723, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.3186, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.9070294784580499, | |
| "grad_norm": 9.854070663452148, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.2849, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.963718820861678, | |
| "grad_norm": 9.135157585144043, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.3387, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.0204081632653061, | |
| "grad_norm": 7.550724983215332, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.2661, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0770975056689343, | |
| "grad_norm": 6.407596111297607, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.1618, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.1337868480725624, | |
| "grad_norm": 8.470088958740234, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.2531, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.1904761904761905, | |
| "grad_norm": 14.254158020019531, | |
| "learning_rate": 9.977894736842106e-06, | |
| "loss": 0.2415, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.2471655328798186, | |
| "grad_norm": 5.5717339515686035, | |
| "learning_rate": 9.951578947368423e-06, | |
| "loss": 0.1863, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.3038548752834467, | |
| "grad_norm": 9.035225868225098, | |
| "learning_rate": 9.925263157894738e-06, | |
| "loss": 0.1772, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.3605442176870748, | |
| "grad_norm": 12.706698417663574, | |
| "learning_rate": 9.898947368421054e-06, | |
| "loss": 0.1589, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.417233560090703, | |
| "grad_norm": 8.393030166625977, | |
| "learning_rate": 9.87263157894737e-06, | |
| "loss": 0.172, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.473922902494331, | |
| "grad_norm": 5.842218399047852, | |
| "learning_rate": 9.846315789473684e-06, | |
| "loss": 0.1656, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.5306122448979593, | |
| "grad_norm": 9.525617599487305, | |
| "learning_rate": 9.820000000000001e-06, | |
| "loss": 0.186, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.5873015873015874, | |
| "grad_norm": 19.124101638793945, | |
| "learning_rate": 9.793684210526316e-06, | |
| "loss": 0.1943, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.6439909297052155, | |
| "grad_norm": 9.63739013671875, | |
| "learning_rate": 9.767368421052632e-06, | |
| "loss": 0.1514, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.7006802721088436, | |
| "grad_norm": 14.47154712677002, | |
| "learning_rate": 9.741052631578947e-06, | |
| "loss": 0.142, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.7573696145124718, | |
| "grad_norm": 6.553714275360107, | |
| "learning_rate": 9.714736842105264e-06, | |
| "loss": 0.1563, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.8140589569160999, | |
| "grad_norm": 6.272864818572998, | |
| "learning_rate": 9.68842105263158e-06, | |
| "loss": 0.1454, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.870748299319728, | |
| "grad_norm": 6.724349021911621, | |
| "learning_rate": 9.662105263157896e-06, | |
| "loss": 0.3289, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.927437641723356, | |
| "grad_norm": 6.75054931640625, | |
| "learning_rate": 9.635789473684212e-06, | |
| "loss": 0.1455, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.9841269841269842, | |
| "grad_norm": 7.212646961212158, | |
| "learning_rate": 9.609473684210527e-06, | |
| "loss": 0.2098, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.0408163265306123, | |
| "grad_norm": 4.189349174499512, | |
| "learning_rate": 9.583157894736842e-06, | |
| "loss": 0.1016, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.0975056689342404, | |
| "grad_norm": 9.265031814575195, | |
| "learning_rate": 9.556842105263159e-06, | |
| "loss": 0.1116, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.1541950113378685, | |
| "grad_norm": 4.884438991546631, | |
| "learning_rate": 9.530526315789474e-06, | |
| "loss": 0.1209, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.2108843537414966, | |
| "grad_norm": 4.258934497833252, | |
| "learning_rate": 9.50421052631579e-06, | |
| "loss": 0.1331, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.2675736961451247, | |
| "grad_norm": 5.421435832977295, | |
| "learning_rate": 9.477894736842106e-06, | |
| "loss": 0.083, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.2675736961451247, | |
| "eval_loss": 0.1953069269657135, | |
| "eval_runtime": 158.2725, | |
| "eval_samples_per_second": 2.957, | |
| "eval_steps_per_second": 0.493, | |
| "eval_wer": 8.173076923076923, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.324263038548753, | |
| "grad_norm": 15.091805458068848, | |
| "learning_rate": 9.451578947368422e-06, | |
| "loss": 0.0908, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 9.11117172241211, | |
| "learning_rate": 9.425263157894737e-06, | |
| "loss": 0.1528, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.437641723356009, | |
| "grad_norm": 8.174238204956055, | |
| "learning_rate": 9.398947368421052e-06, | |
| "loss": 0.059, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.494331065759637, | |
| "grad_norm": 12.375757217407227, | |
| "learning_rate": 9.372631578947369e-06, | |
| "loss": 0.0773, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.5510204081632653, | |
| "grad_norm": 8.518278121948242, | |
| "learning_rate": 9.346315789473684e-06, | |
| "loss": 0.0714, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.6077097505668934, | |
| "grad_norm": 7.095608234405518, | |
| "learning_rate": 9.32e-06, | |
| "loss": 0.0778, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.6643990929705215, | |
| "grad_norm": 5.508459568023682, | |
| "learning_rate": 9.293684210526317e-06, | |
| "loss": 0.0696, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.7210884353741496, | |
| "grad_norm": 1.7915226221084595, | |
| "learning_rate": 9.267368421052632e-06, | |
| "loss": 0.0504, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 4.419779300689697, | |
| "learning_rate": 9.241052631578949e-06, | |
| "loss": 0.0667, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.834467120181406, | |
| "grad_norm": 4.396119594573975, | |
| "learning_rate": 9.214736842105264e-06, | |
| "loss": 0.1037, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.891156462585034, | |
| "grad_norm": 4.604401111602783, | |
| "learning_rate": 9.18842105263158e-06, | |
| "loss": 0.1276, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.947845804988662, | |
| "grad_norm": 6.504410743713379, | |
| "learning_rate": 9.162105263157895e-06, | |
| "loss": 0.0683, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.00453514739229, | |
| "grad_norm": 3.4603614807128906, | |
| "learning_rate": 9.13578947368421e-06, | |
| "loss": 0.0548, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 3.061224489795918, | |
| "grad_norm": 8.05783748626709, | |
| "learning_rate": 9.109473684210527e-06, | |
| "loss": 0.0423, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.1179138321995463, | |
| "grad_norm": 1.4570063352584839, | |
| "learning_rate": 9.083157894736842e-06, | |
| "loss": 0.0388, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.1746031746031744, | |
| "grad_norm": 2.493945360183716, | |
| "learning_rate": 9.056842105263159e-06, | |
| "loss": 0.0704, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.2312925170068025, | |
| "grad_norm": 3.1329710483551025, | |
| "learning_rate": 9.030526315789474e-06, | |
| "loss": 0.0418, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.287981859410431, | |
| "grad_norm": 0.9819092154502869, | |
| "learning_rate": 9.00421052631579e-06, | |
| "loss": 0.0572, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.3446712018140587, | |
| "grad_norm": 3.2691588401794434, | |
| "learning_rate": 8.977894736842107e-06, | |
| "loss": 0.0391, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.4013605442176873, | |
| "grad_norm": 0.9585368037223816, | |
| "learning_rate": 8.951578947368422e-06, | |
| "loss": 0.0468, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.458049886621315, | |
| "grad_norm": 5.486790657043457, | |
| "learning_rate": 8.925263157894739e-06, | |
| "loss": 0.0423, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.5147392290249435, | |
| "grad_norm": 7.715363502502441, | |
| "learning_rate": 8.898947368421054e-06, | |
| "loss": 0.119, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.571428571428571, | |
| "grad_norm": 0.7970800995826721, | |
| "learning_rate": 8.872631578947369e-06, | |
| "loss": 0.0264, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.6281179138321997, | |
| "grad_norm": 9.640973091125488, | |
| "learning_rate": 8.846315789473685e-06, | |
| "loss": 0.0305, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.6848072562358274, | |
| "grad_norm": 5.621988296508789, | |
| "learning_rate": 8.82e-06, | |
| "loss": 0.0329, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.741496598639456, | |
| "grad_norm": 2.357621908187866, | |
| "learning_rate": 8.793684210526317e-06, | |
| "loss": 0.0657, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.798185941043084, | |
| "grad_norm": 3.6409835815429688, | |
| "learning_rate": 8.767368421052632e-06, | |
| "loss": 0.0311, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.854875283446712, | |
| "grad_norm": 2.0802805423736572, | |
| "learning_rate": 8.741052631578949e-06, | |
| "loss": 0.0763, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.9115646258503403, | |
| "grad_norm": 14.157549858093262, | |
| "learning_rate": 8.714736842105264e-06, | |
| "loss": 0.029, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.9682539682539684, | |
| "grad_norm": 3.303739309310913, | |
| "learning_rate": 8.688421052631579e-06, | |
| "loss": 0.0429, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.024943310657596, | |
| "grad_norm": 3.3203935623168945, | |
| "learning_rate": 8.662105263157895e-06, | |
| "loss": 0.0212, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 4.081632653061225, | |
| "grad_norm": 0.7093961834907532, | |
| "learning_rate": 8.63578947368421e-06, | |
| "loss": 0.012, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.138321995464852, | |
| "grad_norm": 2.5663654804229736, | |
| "learning_rate": 8.609473684210527e-06, | |
| "loss": 0.0232, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 4.195011337868481, | |
| "grad_norm": 27.419864654541016, | |
| "learning_rate": 8.583157894736843e-06, | |
| "loss": 0.0192, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.2517006802721085, | |
| "grad_norm": 0.45838263630867004, | |
| "learning_rate": 8.556842105263158e-06, | |
| "loss": 0.0577, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 4.308390022675737, | |
| "grad_norm": 1.192967176437378, | |
| "learning_rate": 8.530526315789475e-06, | |
| "loss": 0.0622, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.365079365079365, | |
| "grad_norm": 5.143068790435791, | |
| "learning_rate": 8.50421052631579e-06, | |
| "loss": 0.0217, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 4.421768707482993, | |
| "grad_norm": 3.8326940536499023, | |
| "learning_rate": 8.477894736842107e-06, | |
| "loss": 0.0212, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.478458049886621, | |
| "grad_norm": 1.7538135051727295, | |
| "learning_rate": 8.451578947368422e-06, | |
| "loss": 0.0499, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.535147392290249, | |
| "grad_norm": 0.5768720507621765, | |
| "learning_rate": 8.425263157894737e-06, | |
| "loss": 0.0444, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.535147392290249, | |
| "eval_loss": 0.19730134308338165, | |
| "eval_runtime": 158.9638, | |
| "eval_samples_per_second": 2.944, | |
| "eval_steps_per_second": 0.491, | |
| "eval_wer": 6.944444444444445, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.591836734693878, | |
| "grad_norm": 4.980081558227539, | |
| "learning_rate": 8.398947368421053e-06, | |
| "loss": 0.0205, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.648526077097506, | |
| "grad_norm": 2.384981155395508, | |
| "learning_rate": 8.372631578947368e-06, | |
| "loss": 0.027, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.705215419501133, | |
| "grad_norm": 1.0443973541259766, | |
| "learning_rate": 8.346315789473685e-06, | |
| "loss": 0.0537, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.761904761904762, | |
| "grad_norm": 1.0461288690567017, | |
| "learning_rate": 8.32e-06, | |
| "loss": 0.0146, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.81859410430839, | |
| "grad_norm": 6.465195655822754, | |
| "learning_rate": 8.293684210526317e-06, | |
| "loss": 0.0349, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.875283446712018, | |
| "grad_norm": 4.054196834564209, | |
| "learning_rate": 8.267368421052632e-06, | |
| "loss": 0.0186, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.931972789115647, | |
| "grad_norm": 1.7794383764266968, | |
| "learning_rate": 8.241052631578948e-06, | |
| "loss": 0.0224, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.988662131519274, | |
| "grad_norm": 5.188144683837891, | |
| "learning_rate": 8.214736842105265e-06, | |
| "loss": 0.047, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.045351473922903, | |
| "grad_norm": 0.08940722048282623, | |
| "learning_rate": 8.18842105263158e-06, | |
| "loss": 0.0186, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 5.1020408163265305, | |
| "grad_norm": 3.670670747756958, | |
| "learning_rate": 8.162105263157895e-06, | |
| "loss": 0.0183, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 5.158730158730159, | |
| "grad_norm": 2.3486833572387695, | |
| "learning_rate": 8.135789473684212e-06, | |
| "loss": 0.015, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 5.215419501133787, | |
| "grad_norm": 4.572961807250977, | |
| "learning_rate": 8.109473684210527e-06, | |
| "loss": 0.0239, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.272108843537415, | |
| "grad_norm": 1.114425539970398, | |
| "learning_rate": 8.083157894736843e-06, | |
| "loss": 0.0396, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 5.328798185941043, | |
| "grad_norm": 0.35853487253189087, | |
| "learning_rate": 8.056842105263158e-06, | |
| "loss": 0.0185, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.3854875283446715, | |
| "grad_norm": 5.747114181518555, | |
| "learning_rate": 8.030526315789475e-06, | |
| "loss": 0.0136, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 5.442176870748299, | |
| "grad_norm": 0.33728256821632385, | |
| "learning_rate": 8.00421052631579e-06, | |
| "loss": 0.0366, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 5.498866213151928, | |
| "grad_norm": 13.636420249938965, | |
| "learning_rate": 7.977894736842105e-06, | |
| "loss": 0.0441, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 2.6928389072418213, | |
| "learning_rate": 7.951578947368421e-06, | |
| "loss": 0.014, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.612244897959184, | |
| "grad_norm": 0.18423891067504883, | |
| "learning_rate": 7.925263157894736e-06, | |
| "loss": 0.007, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.668934240362812, | |
| "grad_norm": 9.745643615722656, | |
| "learning_rate": 7.898947368421053e-06, | |
| "loss": 0.0065, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.72562358276644, | |
| "grad_norm": 1.0876612663269043, | |
| "learning_rate": 7.87263157894737e-06, | |
| "loss": 0.0464, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 5.782312925170068, | |
| "grad_norm": 3.2040324211120605, | |
| "learning_rate": 7.846315789473685e-06, | |
| "loss": 0.0085, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.839002267573696, | |
| "grad_norm": 0.07621826976537704, | |
| "learning_rate": 7.820000000000001e-06, | |
| "loss": 0.0417, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.895691609977324, | |
| "grad_norm": 0.11689532548189163, | |
| "learning_rate": 7.793684210526316e-06, | |
| "loss": 0.0113, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.9523809523809526, | |
| "grad_norm": 1.2615737915039062, | |
| "learning_rate": 7.767368421052633e-06, | |
| "loss": 0.017, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 6.00907029478458, | |
| "grad_norm": 6.465045928955078, | |
| "learning_rate": 7.741052631578948e-06, | |
| "loss": 0.0134, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 6.065759637188209, | |
| "grad_norm": 0.6422730088233948, | |
| "learning_rate": 7.714736842105263e-06, | |
| "loss": 0.0407, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 6.122448979591836, | |
| "grad_norm": 0.07282353192567825, | |
| "learning_rate": 7.68842105263158e-06, | |
| "loss": 0.0066, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.179138321995465, | |
| "grad_norm": 0.13422216475009918, | |
| "learning_rate": 7.662105263157895e-06, | |
| "loss": 0.0345, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 6.235827664399093, | |
| "grad_norm": 4.29299783706665, | |
| "learning_rate": 7.635789473684211e-06, | |
| "loss": 0.0047, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.292517006802721, | |
| "grad_norm": 0.05262218415737152, | |
| "learning_rate": 7.609473684210526e-06, | |
| "loss": 0.0023, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 6.349206349206349, | |
| "grad_norm": 6.577340602874756, | |
| "learning_rate": 7.583157894736842e-06, | |
| "loss": 0.005, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 6.405895691609977, | |
| "grad_norm": 0.2815414071083069, | |
| "learning_rate": 7.556842105263158e-06, | |
| "loss": 0.0155, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 6.462585034013605, | |
| "grad_norm": 0.05367182940244675, | |
| "learning_rate": 7.5305263157894745e-06, | |
| "loss": 0.04, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 6.519274376417234, | |
| "grad_norm": 0.14756247401237488, | |
| "learning_rate": 7.50421052631579e-06, | |
| "loss": 0.0049, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 6.575963718820862, | |
| "grad_norm": 2.2667605876922607, | |
| "learning_rate": 7.477894736842106e-06, | |
| "loss": 0.0028, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 6.63265306122449, | |
| "grad_norm": 0.13220186531543732, | |
| "learning_rate": 7.451578947368422e-06, | |
| "loss": 0.0114, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 6.6893424036281175, | |
| "grad_norm": 2.413485288619995, | |
| "learning_rate": 7.425263157894738e-06, | |
| "loss": 0.0028, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 6.746031746031746, | |
| "grad_norm": 1.9141589403152466, | |
| "learning_rate": 7.398947368421054e-06, | |
| "loss": 0.0079, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 6.802721088435375, | |
| "grad_norm": 5.837501525878906, | |
| "learning_rate": 7.3726315789473694e-06, | |
| "loss": 0.0294, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.802721088435375, | |
| "eval_loss": 0.19840003550052643, | |
| "eval_runtime": 156.6444, | |
| "eval_samples_per_second": 2.988, | |
| "eval_steps_per_second": 0.498, | |
| "eval_wer": 6.517094017094018, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.859410430839002, | |
| "grad_norm": 0.0996256172657013, | |
| "learning_rate": 7.346315789473684e-06, | |
| "loss": 0.0027, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 6.91609977324263, | |
| "grad_norm": 0.028051115572452545, | |
| "learning_rate": 7.32e-06, | |
| "loss": 0.008, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.9727891156462585, | |
| "grad_norm": 0.14238622784614563, | |
| "learning_rate": 7.293684210526316e-06, | |
| "loss": 0.0263, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 7.029478458049887, | |
| "grad_norm": 0.112076535820961, | |
| "learning_rate": 7.267368421052632e-06, | |
| "loss": 0.0192, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 7.086167800453515, | |
| "grad_norm": 0.14218856394290924, | |
| "learning_rate": 7.241052631578948e-06, | |
| "loss": 0.0099, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 7.142857142857143, | |
| "grad_norm": 0.08735861629247665, | |
| "learning_rate": 7.2147368421052635e-06, | |
| "loss": 0.0013, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 7.199546485260771, | |
| "grad_norm": 0.07858515530824661, | |
| "learning_rate": 7.18842105263158e-06, | |
| "loss": 0.043, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 7.2562358276643995, | |
| "grad_norm": 0.04395943507552147, | |
| "learning_rate": 7.162105263157896e-06, | |
| "loss": 0.0196, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 7.312925170068027, | |
| "grad_norm": 0.13118867576122284, | |
| "learning_rate": 7.135789473684212e-06, | |
| "loss": 0.0032, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 7.369614512471656, | |
| "grad_norm": 0.5723868608474731, | |
| "learning_rate": 7.109473684210528e-06, | |
| "loss": 0.0193, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 7.426303854875283, | |
| "grad_norm": 0.17946386337280273, | |
| "learning_rate": 7.08421052631579e-06, | |
| "loss": 0.0066, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 7.482993197278912, | |
| "grad_norm": 0.31329983472824097, | |
| "learning_rate": 7.057894736842106e-06, | |
| "loss": 0.004, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 7.5396825396825395, | |
| "grad_norm": 0.08119330555200577, | |
| "learning_rate": 7.031578947368422e-06, | |
| "loss": 0.0079, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 7.596371882086168, | |
| "grad_norm": 0.2911977767944336, | |
| "learning_rate": 7.005263157894738e-06, | |
| "loss": 0.0053, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 7.653061224489796, | |
| "grad_norm": 0.024690864607691765, | |
| "learning_rate": 6.9789473684210525e-06, | |
| "loss": 0.0054, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 7.709750566893424, | |
| "grad_norm": 0.09139782190322876, | |
| "learning_rate": 6.953684210526316e-06, | |
| "loss": 0.0043, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 7.766439909297052, | |
| "grad_norm": 0.3346998393535614, | |
| "learning_rate": 6.9273684210526326e-06, | |
| "loss": 0.0063, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 7.8231292517006805, | |
| "grad_norm": 2.5169262886047363, | |
| "learning_rate": 6.901052631578948e-06, | |
| "loss": 0.0036, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 7.879818594104308, | |
| "grad_norm": 0.05118921771645546, | |
| "learning_rate": 6.874736842105264e-06, | |
| "loss": 0.0152, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 7.936507936507937, | |
| "grad_norm": 0.141609787940979, | |
| "learning_rate": 6.84842105263158e-06, | |
| "loss": 0.0029, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.993197278911564, | |
| "grad_norm": 0.059452034533023834, | |
| "learning_rate": 6.822105263157896e-06, | |
| "loss": 0.0084, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 8.049886621315192, | |
| "grad_norm": 9.327384948730469, | |
| "learning_rate": 6.795789473684211e-06, | |
| "loss": 0.0056, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 8.106575963718821, | |
| "grad_norm": 0.031346723437309265, | |
| "learning_rate": 6.769473684210527e-06, | |
| "loss": 0.0095, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 8.16326530612245, | |
| "grad_norm": 0.03407048434019089, | |
| "learning_rate": 6.7431578947368425e-06, | |
| "loss": 0.031, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 8.219954648526077, | |
| "grad_norm": 0.023022985085844994, | |
| "learning_rate": 6.716842105263158e-06, | |
| "loss": 0.0061, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 8.276643990929704, | |
| "grad_norm": 0.1935151070356369, | |
| "learning_rate": 6.690526315789474e-06, | |
| "loss": 0.0049, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 1.2079312801361084, | |
| "learning_rate": 6.66421052631579e-06, | |
| "loss": 0.0107, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 8.390022675736962, | |
| "grad_norm": 0.20077985525131226, | |
| "learning_rate": 6.637894736842106e-06, | |
| "loss": 0.0021, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 8.44671201814059, | |
| "grad_norm": 18.450538635253906, | |
| "learning_rate": 6.611578947368421e-06, | |
| "loss": 0.0159, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 8.503401360544217, | |
| "grad_norm": 0.03869379311800003, | |
| "learning_rate": 6.585263157894738e-06, | |
| "loss": 0.0052, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 8.560090702947846, | |
| "grad_norm": 0.026166923344135284, | |
| "learning_rate": 6.558947368421054e-06, | |
| "loss": 0.0007, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 8.616780045351474, | |
| "grad_norm": 0.02891625091433525, | |
| "learning_rate": 6.532631578947369e-06, | |
| "loss": 0.0041, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 8.673469387755102, | |
| "grad_norm": 0.08599188923835754, | |
| "learning_rate": 6.506315789473685e-06, | |
| "loss": 0.0009, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 8.73015873015873, | |
| "grad_norm": 0.017651915550231934, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 0.0211, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 8.786848072562359, | |
| "grad_norm": 14.024144172668457, | |
| "learning_rate": 6.4536842105263165e-06, | |
| "loss": 0.021, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 8.843537414965986, | |
| "grad_norm": 0.057578567415475845, | |
| "learning_rate": 6.427368421052632e-06, | |
| "loss": 0.0021, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 8.900226757369614, | |
| "grad_norm": 0.04060237482190132, | |
| "learning_rate": 6.401052631578948e-06, | |
| "loss": 0.0025, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 8.956916099773242, | |
| "grad_norm": 0.04346761852502823, | |
| "learning_rate": 6.374736842105264e-06, | |
| "loss": 0.0055, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 9.013605442176871, | |
| "grad_norm": 0.014886971563100815, | |
| "learning_rate": 6.348421052631579e-06, | |
| "loss": 0.0067, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 9.070294784580499, | |
| "grad_norm": 2.309483289718628, | |
| "learning_rate": 6.322105263157895e-06, | |
| "loss": 0.0334, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.070294784580499, | |
| "eval_loss": 0.20987384021282196, | |
| "eval_runtime": 157.5775, | |
| "eval_samples_per_second": 2.97, | |
| "eval_steps_per_second": 0.495, | |
| "eval_wer": 6.303418803418803, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.126984126984127, | |
| "grad_norm": 0.2992580235004425, | |
| "learning_rate": 6.2957894736842105e-06, | |
| "loss": 0.0048, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 9.183673469387756, | |
| "grad_norm": 0.1327856183052063, | |
| "learning_rate": 6.269473684210526e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 9.240362811791384, | |
| "grad_norm": 0.034492090344429016, | |
| "learning_rate": 6.243157894736842e-06, | |
| "loss": 0.0007, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 9.297052154195011, | |
| "grad_norm": 0.021785929799079895, | |
| "learning_rate": 6.216842105263159e-06, | |
| "loss": 0.0013, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 9.353741496598639, | |
| "grad_norm": 0.17917415499687195, | |
| "learning_rate": 6.190526315789475e-06, | |
| "loss": 0.0186, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 9.410430839002268, | |
| "grad_norm": 0.04263261333107948, | |
| "learning_rate": 6.1642105263157905e-06, | |
| "loss": 0.0011, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 9.467120181405896, | |
| "grad_norm": 0.06537426263093948, | |
| "learning_rate": 6.137894736842106e-06, | |
| "loss": 0.0007, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 9.523809523809524, | |
| "grad_norm": 0.05984114482998848, | |
| "learning_rate": 6.111578947368422e-06, | |
| "loss": 0.0031, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 9.580498866213151, | |
| "grad_norm": 0.0186479389667511, | |
| "learning_rate": 6.085263157894737e-06, | |
| "loss": 0.0067, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 9.63718820861678, | |
| "grad_norm": 0.01856757327914238, | |
| "learning_rate": 6.058947368421053e-06, | |
| "loss": 0.0012, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 9.693877551020408, | |
| "grad_norm": 0.01589180715382099, | |
| "learning_rate": 6.032631578947369e-06, | |
| "loss": 0.0033, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 9.750566893424036, | |
| "grad_norm": 0.013659857213497162, | |
| "learning_rate": 6.0063157894736845e-06, | |
| "loss": 0.0136, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 9.807256235827664, | |
| "grad_norm": 0.07311205565929413, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.0114, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 9.863945578231293, | |
| "grad_norm": 0.00827726535499096, | |
| "learning_rate": 5.953684210526316e-06, | |
| "loss": 0.0006, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 9.920634920634921, | |
| "grad_norm": 0.02168506383895874, | |
| "learning_rate": 5.927368421052632e-06, | |
| "loss": 0.0141, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 9.977324263038549, | |
| "grad_norm": 0.09996296465396881, | |
| "learning_rate": 5.901052631578947e-06, | |
| "loss": 0.0022, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 10.034013605442176, | |
| "grad_norm": 0.016953065991401672, | |
| "learning_rate": 5.8747368421052645e-06, | |
| "loss": 0.0024, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 10.090702947845806, | |
| "grad_norm": 0.0506548210978508, | |
| "learning_rate": 5.84842105263158e-06, | |
| "loss": 0.0049, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 10.147392290249433, | |
| "grad_norm": 0.017473401501774788, | |
| "learning_rate": 5.822105263157895e-06, | |
| "loss": 0.0009, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 10.204081632653061, | |
| "grad_norm": 0.016834545880556107, | |
| "learning_rate": 5.795789473684211e-06, | |
| "loss": 0.0016, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 10.260770975056689, | |
| "grad_norm": 0.07962112873792648, | |
| "learning_rate": 5.769473684210527e-06, | |
| "loss": 0.0005, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 10.317460317460318, | |
| "grad_norm": 0.020868808031082153, | |
| "learning_rate": 5.743157894736843e-06, | |
| "loss": 0.0162, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 10.374149659863946, | |
| "grad_norm": 0.0351821593940258, | |
| "learning_rate": 5.7168421052631585e-06, | |
| "loss": 0.0217, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 10.430839002267573, | |
| "grad_norm": 0.036179013550281525, | |
| "learning_rate": 5.690526315789474e-06, | |
| "loss": 0.0022, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 10.487528344671201, | |
| "grad_norm": 0.04735976830124855, | |
| "learning_rate": 5.66421052631579e-06, | |
| "loss": 0.0061, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 10.54421768707483, | |
| "grad_norm": 0.015306883491575718, | |
| "learning_rate": 5.637894736842105e-06, | |
| "loss": 0.0009, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 10.600907029478458, | |
| "grad_norm": 0.015260276384651661, | |
| "learning_rate": 5.611578947368421e-06, | |
| "loss": 0.0067, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 10.657596371882086, | |
| "grad_norm": 0.029503723606467247, | |
| "learning_rate": 5.585263157894737e-06, | |
| "loss": 0.0006, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 10.714285714285714, | |
| "grad_norm": 0.017955463379621506, | |
| "learning_rate": 5.558947368421053e-06, | |
| "loss": 0.0005, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 10.770975056689343, | |
| "grad_norm": 0.11942701041698456, | |
| "learning_rate": 5.532631578947368e-06, | |
| "loss": 0.003, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 10.82766439909297, | |
| "grad_norm": 0.02495400980114937, | |
| "learning_rate": 5.506315789473685e-06, | |
| "loss": 0.0017, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 10.884353741496598, | |
| "grad_norm": 0.1065245270729065, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 0.0047, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 10.941043083900226, | |
| "grad_norm": 0.10682205855846405, | |
| "learning_rate": 5.453684210526317e-06, | |
| "loss": 0.0045, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 10.997732426303855, | |
| "grad_norm": 0.016077643260359764, | |
| "learning_rate": 5.4273684210526325e-06, | |
| "loss": 0.0007, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 11.054421768707483, | |
| "grad_norm": 0.020081788301467896, | |
| "learning_rate": 5.401052631578948e-06, | |
| "loss": 0.0004, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 11.11111111111111, | |
| "grad_norm": 0.017382116988301277, | |
| "learning_rate": 5.374736842105263e-06, | |
| "loss": 0.0012, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 11.167800453514738, | |
| "grad_norm": 0.5074162483215332, | |
| "learning_rate": 5.348421052631579e-06, | |
| "loss": 0.0038, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 11.224489795918368, | |
| "grad_norm": 0.011525845155119896, | |
| "learning_rate": 5.322105263157895e-06, | |
| "loss": 0.0011, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 11.281179138321995, | |
| "grad_norm": 0.016790462657809258, | |
| "learning_rate": 5.295789473684211e-06, | |
| "loss": 0.0013, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 11.337868480725623, | |
| "grad_norm": 2.887037754058838, | |
| "learning_rate": 5.269473684210527e-06, | |
| "loss": 0.0011, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.337868480725623, | |
| "eval_loss": 0.2228717803955078, | |
| "eval_runtime": 177.4582, | |
| "eval_samples_per_second": 2.637, | |
| "eval_steps_per_second": 0.44, | |
| "eval_wer": 6.3835470085470085, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.39455782312925, | |
| "grad_norm": 0.02366207167506218, | |
| "learning_rate": 5.243157894736842e-06, | |
| "loss": 0.0002, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 11.45124716553288, | |
| "grad_norm": 1.1653227806091309, | |
| "learning_rate": 5.216842105263158e-06, | |
| "loss": 0.0052, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 11.507936507936508, | |
| "grad_norm": 0.2194329798221588, | |
| "learning_rate": 5.190526315789474e-06, | |
| "loss": 0.0014, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 11.564625850340136, | |
| "grad_norm": 0.03989162668585777, | |
| "learning_rate": 5.164210526315791e-06, | |
| "loss": 0.0006, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 11.621315192743765, | |
| "grad_norm": 0.02930096909403801, | |
| "learning_rate": 5.1378947368421065e-06, | |
| "loss": 0.003, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 11.678004535147393, | |
| "grad_norm": 0.011350632645189762, | |
| "learning_rate": 5.1115789473684215e-06, | |
| "loss": 0.0039, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 11.73469387755102, | |
| "grad_norm": 0.010673941113054752, | |
| "learning_rate": 5.085263157894737e-06, | |
| "loss": 0.0004, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 11.791383219954648, | |
| "grad_norm": 0.05096409469842911, | |
| "learning_rate": 5.058947368421053e-06, | |
| "loss": 0.0048, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 11.848072562358277, | |
| "grad_norm": 0.011835623532533646, | |
| "learning_rate": 5.032631578947369e-06, | |
| "loss": 0.0093, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 11.904761904761905, | |
| "grad_norm": 0.013359226286411285, | |
| "learning_rate": 5.006315789473685e-06, | |
| "loss": 0.0062, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 11.961451247165533, | |
| "grad_norm": 0.03088083118200302, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.0002, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 12.01814058956916, | |
| "grad_norm": 0.02855735644698143, | |
| "learning_rate": 4.953684210526316e-06, | |
| "loss": 0.0002, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 12.07482993197279, | |
| "grad_norm": 0.04174978658556938, | |
| "learning_rate": 4.927368421052631e-06, | |
| "loss": 0.0002, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 12.131519274376418, | |
| "grad_norm": 0.013824643567204475, | |
| "learning_rate": 4.901052631578947e-06, | |
| "loss": 0.0002, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 12.188208616780045, | |
| "grad_norm": 0.010323552414774895, | |
| "learning_rate": 4.874736842105264e-06, | |
| "loss": 0.0002, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 12.244897959183673, | |
| "grad_norm": 7.9211883544921875, | |
| "learning_rate": 4.84842105263158e-06, | |
| "loss": 0.0041, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 12.301587301587302, | |
| "grad_norm": 0.011090376414358616, | |
| "learning_rate": 4.8221052631578955e-06, | |
| "loss": 0.0046, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 12.35827664399093, | |
| "grad_norm": 0.993200957775116, | |
| "learning_rate": 4.7957894736842105e-06, | |
| "loss": 0.0077, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 12.414965986394558, | |
| "grad_norm": 0.5387348532676697, | |
| "learning_rate": 4.769473684210526e-06, | |
| "loss": 0.0014, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 12.471655328798185, | |
| "grad_norm": 0.030508503317832947, | |
| "learning_rate": 4.743157894736842e-06, | |
| "loss": 0.0029, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 12.528344671201815, | |
| "grad_norm": 0.015367632731795311, | |
| "learning_rate": 4.716842105263159e-06, | |
| "loss": 0.002, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 12.585034013605442, | |
| "grad_norm": 0.01937568373978138, | |
| "learning_rate": 4.690526315789475e-06, | |
| "loss": 0.0016, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 12.64172335600907, | |
| "grad_norm": 0.06024911627173424, | |
| "learning_rate": 4.6642105263157896e-06, | |
| "loss": 0.0016, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 12.698412698412698, | |
| "grad_norm": 0.005301471799612045, | |
| "learning_rate": 4.637894736842105e-06, | |
| "loss": 0.0014, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 12.755102040816327, | |
| "grad_norm": 0.03389279916882515, | |
| "learning_rate": 4.611578947368421e-06, | |
| "loss": 0.0017, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 12.811791383219955, | |
| "grad_norm": 0.13554659485816956, | |
| "learning_rate": 4.585263157894737e-06, | |
| "loss": 0.0105, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 12.868480725623582, | |
| "grad_norm": 0.018741684034466743, | |
| "learning_rate": 4.558947368421053e-06, | |
| "loss": 0.004, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 12.92517006802721, | |
| "grad_norm": 0.1560622751712799, | |
| "learning_rate": 4.532631578947369e-06, | |
| "loss": 0.0005, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 12.98185941043084, | |
| "grad_norm": 0.1737220734357834, | |
| "learning_rate": 4.5063157894736845e-06, | |
| "loss": 0.0019, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 13.038548752834467, | |
| "grad_norm": 0.02937311679124832, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.002, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 13.095238095238095, | |
| "grad_norm": 0.006793774198740721, | |
| "learning_rate": 4.453684210526316e-06, | |
| "loss": 0.0005, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 13.151927437641723, | |
| "grad_norm": 0.014649259857833385, | |
| "learning_rate": 4.427368421052632e-06, | |
| "loss": 0.0025, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 13.208616780045352, | |
| "grad_norm": 0.0212300606071949, | |
| "learning_rate": 4.401052631578948e-06, | |
| "loss": 0.0014, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 13.26530612244898, | |
| "grad_norm": 0.013082647696137428, | |
| "learning_rate": 4.374736842105264e-06, | |
| "loss": 0.0039, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 13.321995464852607, | |
| "grad_norm": 0.05612126737833023, | |
| "learning_rate": 4.348421052631579e-06, | |
| "loss": 0.0007, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 13.378684807256235, | |
| "grad_norm": 0.37419337034225464, | |
| "learning_rate": 4.322105263157895e-06, | |
| "loss": 0.0012, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 13.435374149659864, | |
| "grad_norm": 0.03296487405896187, | |
| "learning_rate": 4.295789473684211e-06, | |
| "loss": 0.0043, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 13.492063492063492, | |
| "grad_norm": 0.009840169921517372, | |
| "learning_rate": 4.269473684210527e-06, | |
| "loss": 0.0039, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 13.54875283446712, | |
| "grad_norm": 0.015135574154555798, | |
| "learning_rate": 4.243157894736843e-06, | |
| "loss": 0.0014, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 13.60544217687075, | |
| "grad_norm": 0.022306112572550774, | |
| "learning_rate": 4.2168421052631585e-06, | |
| "loss": 0.0001, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 13.60544217687075, | |
| "eval_loss": 0.22002862393856049, | |
| "eval_runtime": 147.8825, | |
| "eval_samples_per_second": 3.165, | |
| "eval_steps_per_second": 0.527, | |
| "eval_wer": 6.209935897435898, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 13.662131519274377, | |
| "grad_norm": 0.0066378237679600716, | |
| "learning_rate": 4.1905263157894735e-06, | |
| "loss": 0.0007, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 13.718820861678005, | |
| "grad_norm": 0.010166754946112633, | |
| "learning_rate": 4.16421052631579e-06, | |
| "loss": 0.0008, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 13.775510204081632, | |
| "grad_norm": 0.0136796273291111, | |
| "learning_rate": 4.137894736842106e-06, | |
| "loss": 0.0009, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 13.83219954648526, | |
| "grad_norm": 0.017127549275755882, | |
| "learning_rate": 4.111578947368422e-06, | |
| "loss": 0.0015, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 13.88888888888889, | |
| "grad_norm": 0.024442024528980255, | |
| "learning_rate": 4.085263157894737e-06, | |
| "loss": 0.0032, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 13.945578231292517, | |
| "grad_norm": 0.17017020285129547, | |
| "learning_rate": 4.0589473684210526e-06, | |
| "loss": 0.0015, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 14.002267573696145, | |
| "grad_norm": 0.00828185211867094, | |
| "learning_rate": 4.032631578947368e-06, | |
| "loss": 0.0026, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 14.058956916099774, | |
| "grad_norm": 0.007860764861106873, | |
| "learning_rate": 4.006315789473684e-06, | |
| "loss": 0.0034, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 14.115646258503402, | |
| "grad_norm": 0.011614521034061909, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.0044, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 14.17233560090703, | |
| "grad_norm": 0.018276942893862724, | |
| "learning_rate": 3.953684210526316e-06, | |
| "loss": 0.0004, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 14.229024943310657, | |
| "grad_norm": 0.012453128583729267, | |
| "learning_rate": 3.927368421052632e-06, | |
| "loss": 0.0002, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 14.285714285714286, | |
| "grad_norm": 0.015684612095355988, | |
| "learning_rate": 3.9010526315789475e-06, | |
| "loss": 0.0003, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 14.342403628117914, | |
| "grad_norm": 0.006171511020511389, | |
| "learning_rate": 3.874736842105263e-06, | |
| "loss": 0.0001, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 14.399092970521542, | |
| "grad_norm": 0.006511132698506117, | |
| "learning_rate": 3.848421052631579e-06, | |
| "loss": 0.0004, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 14.45578231292517, | |
| "grad_norm": 0.050404928624629974, | |
| "learning_rate": 3.822105263157895e-06, | |
| "loss": 0.0002, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 14.512471655328799, | |
| "grad_norm": 10.834954261779785, | |
| "learning_rate": 3.795789473684211e-06, | |
| "loss": 0.0012, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 14.569160997732427, | |
| "grad_norm": 0.13081848621368408, | |
| "learning_rate": 3.7694736842105266e-06, | |
| "loss": 0.0016, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 14.625850340136054, | |
| "grad_norm": 0.011676596477627754, | |
| "learning_rate": 3.7431578947368424e-06, | |
| "loss": 0.0044, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 14.682539682539682, | |
| "grad_norm": 0.014040385372936726, | |
| "learning_rate": 3.716842105263158e-06, | |
| "loss": 0.0001, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 14.739229024943311, | |
| "grad_norm": 0.011808693408966064, | |
| "learning_rate": 3.690526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 14.795918367346939, | |
| "grad_norm": 0.012787124142050743, | |
| "learning_rate": 3.6642105263157894e-06, | |
| "loss": 0.0045, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 14.852607709750567, | |
| "grad_norm": 0.005284798797219992, | |
| "learning_rate": 3.6378947368421057e-06, | |
| "loss": 0.0001, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 14.909297052154194, | |
| "grad_norm": 0.036842990666627884, | |
| "learning_rate": 3.6115789473684215e-06, | |
| "loss": 0.0004, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 14.965986394557824, | |
| "grad_norm": 0.024015046656131744, | |
| "learning_rate": 3.5852631578947373e-06, | |
| "loss": 0.0003, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 15.022675736961451, | |
| "grad_norm": 0.012714927084743977, | |
| "learning_rate": 3.558947368421053e-06, | |
| "loss": 0.0025, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 15.079365079365079, | |
| "grad_norm": 0.017115842550992966, | |
| "learning_rate": 3.5326315789473685e-06, | |
| "loss": 0.0012, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 15.136054421768707, | |
| "grad_norm": 0.008900342509150505, | |
| "learning_rate": 3.5063157894736843e-06, | |
| "loss": 0.0001, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 15.192743764172336, | |
| "grad_norm": 0.007803457789123058, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.0004, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 15.249433106575964, | |
| "grad_norm": 0.013770255260169506, | |
| "learning_rate": 3.4536842105263164e-06, | |
| "loss": 0.0011, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 15.306122448979592, | |
| "grad_norm": 0.06877677142620087, | |
| "learning_rate": 3.427368421052632e-06, | |
| "loss": 0.0018, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 15.36281179138322, | |
| "grad_norm": 0.011991630308330059, | |
| "learning_rate": 3.4010526315789476e-06, | |
| "loss": 0.0001, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 15.419501133786849, | |
| "grad_norm": 0.013049086555838585, | |
| "learning_rate": 3.3747368421052634e-06, | |
| "loss": 0.0002, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 15.476190476190476, | |
| "grad_norm": 0.012979848310351372, | |
| "learning_rate": 3.3484210526315792e-06, | |
| "loss": 0.0002, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 15.532879818594104, | |
| "grad_norm": 0.006456771399825811, | |
| "learning_rate": 3.3221052631578946e-06, | |
| "loss": 0.0001, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 15.589569160997732, | |
| "grad_norm": 0.012274966575205326, | |
| "learning_rate": 3.2957894736842104e-06, | |
| "loss": 0.0001, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 15.646258503401361, | |
| "grad_norm": 0.010561280883848667, | |
| "learning_rate": 3.2694736842105267e-06, | |
| "loss": 0.0004, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 15.702947845804989, | |
| "grad_norm": 0.009232975542545319, | |
| "learning_rate": 3.2431578947368425e-06, | |
| "loss": 0.0002, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 15.759637188208616, | |
| "grad_norm": 0.015166404657065868, | |
| "learning_rate": 3.2168421052631583e-06, | |
| "loss": 0.0001, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 15.816326530612244, | |
| "grad_norm": 0.0094530014321208, | |
| "learning_rate": 3.1905263157894737e-06, | |
| "loss": 0.0001, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 15.873015873015873, | |
| "grad_norm": 0.017429711297154427, | |
| "learning_rate": 3.1642105263157895e-06, | |
| "loss": 0.0001, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 15.873015873015873, | |
| "eval_loss": 0.2297230064868927, | |
| "eval_runtime": 188.711, | |
| "eval_samples_per_second": 2.48, | |
| "eval_steps_per_second": 0.413, | |
| "eval_wer": 6.049679487179487, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 15.929705215419501, | |
| "grad_norm": 0.005318532232195139, | |
| "learning_rate": 3.1378947368421054e-06, | |
| "loss": 0.001, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 15.986394557823129, | |
| "grad_norm": 0.011030340567231178, | |
| "learning_rate": 3.111578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 16.04308390022676, | |
| "grad_norm": 0.004018599167466164, | |
| "learning_rate": 3.0852631578947374e-06, | |
| "loss": 0.0001, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 16.099773242630384, | |
| "grad_norm": 0.006292372010648251, | |
| "learning_rate": 3.058947368421053e-06, | |
| "loss": 0.0063, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 16.156462585034014, | |
| "grad_norm": 0.008721155114471912, | |
| "learning_rate": 3.0326315789473686e-06, | |
| "loss": 0.0001, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 16.213151927437643, | |
| "grad_norm": 0.007462701760232449, | |
| "learning_rate": 3.0063157894736844e-06, | |
| "loss": 0.0013, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 16.26984126984127, | |
| "grad_norm": 0.0060442672111094, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.0002, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 16.3265306122449, | |
| "grad_norm": 0.007179939653724432, | |
| "learning_rate": 2.9536842105263157e-06, | |
| "loss": 0.0001, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 16.383219954648528, | |
| "grad_norm": 0.006970668211579323, | |
| "learning_rate": 2.927368421052632e-06, | |
| "loss": 0.0025, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 16.439909297052154, | |
| "grad_norm": 0.0061109112575650215, | |
| "learning_rate": 2.9010526315789477e-06, | |
| "loss": 0.0003, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 16.496598639455783, | |
| "grad_norm": 0.02214565873146057, | |
| "learning_rate": 2.8747368421052635e-06, | |
| "loss": 0.0001, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 16.55328798185941, | |
| "grad_norm": 0.009677527472376823, | |
| "learning_rate": 2.8484210526315794e-06, | |
| "loss": 0.0001, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 16.60997732426304, | |
| "grad_norm": 3.2391059398651123, | |
| "learning_rate": 2.8221052631578948e-06, | |
| "loss": 0.0013, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 16.666666666666668, | |
| "grad_norm": 0.005941161885857582, | |
| "learning_rate": 2.7957894736842106e-06, | |
| "loss": 0.0001, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 16.723356009070294, | |
| "grad_norm": 0.005041074473410845, | |
| "learning_rate": 2.7694736842105264e-06, | |
| "loss": 0.0001, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 16.780045351473923, | |
| "grad_norm": 0.010110282339155674, | |
| "learning_rate": 2.7431578947368426e-06, | |
| "loss": 0.0001, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 16.836734693877553, | |
| "grad_norm": 0.0052981507033109665, | |
| "learning_rate": 2.7168421052631585e-06, | |
| "loss": 0.0001, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 16.89342403628118, | |
| "grad_norm": 0.007062564603984356, | |
| "learning_rate": 2.690526315789474e-06, | |
| "loss": 0.0004, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 16.950113378684808, | |
| "grad_norm": 0.005766382906585932, | |
| "learning_rate": 2.6642105263157897e-06, | |
| "loss": 0.0001, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 17.006802721088434, | |
| "grad_norm": 8.800177574157715, | |
| "learning_rate": 2.6378947368421055e-06, | |
| "loss": 0.0022, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 17.063492063492063, | |
| "grad_norm": 0.003864010563120246, | |
| "learning_rate": 2.6115789473684213e-06, | |
| "loss": 0.0001, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 17.120181405895693, | |
| "grad_norm": 0.003043045522645116, | |
| "learning_rate": 2.5852631578947367e-06, | |
| "loss": 0.0001, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 17.17687074829932, | |
| "grad_norm": 0.0026129058096557856, | |
| "learning_rate": 2.558947368421053e-06, | |
| "loss": 0.0001, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 17.233560090702948, | |
| "grad_norm": 0.003994261380285025, | |
| "learning_rate": 2.5326315789473688e-06, | |
| "loss": 0.0001, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 17.290249433106577, | |
| "grad_norm": 0.006509356200695038, | |
| "learning_rate": 2.5063157894736846e-06, | |
| "loss": 0.0001, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 17.346938775510203, | |
| "grad_norm": 0.006231856532394886, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.0001, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 17.403628117913833, | |
| "grad_norm": 0.00826491229236126, | |
| "learning_rate": 2.453684210526316e-06, | |
| "loss": 0.0027, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 17.46031746031746, | |
| "grad_norm": 0.007504597306251526, | |
| "learning_rate": 2.427368421052632e-06, | |
| "loss": 0.0001, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 17.517006802721088, | |
| "grad_norm": 0.0033706706017255783, | |
| "learning_rate": 2.4010526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 17.573696145124718, | |
| "grad_norm": 0.004383792169392109, | |
| "learning_rate": 2.3747368421052632e-06, | |
| "loss": 0.0018, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 17.630385487528343, | |
| "grad_norm": 0.004007370211184025, | |
| "learning_rate": 2.348421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 17.687074829931973, | |
| "grad_norm": 0.006295809056609869, | |
| "learning_rate": 2.322105263157895e-06, | |
| "loss": 0.0001, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 17.743764172335602, | |
| "grad_norm": 0.006831544451415539, | |
| "learning_rate": 2.2957894736842107e-06, | |
| "loss": 0.0003, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 17.800453514739228, | |
| "grad_norm": 0.0033715348690748215, | |
| "learning_rate": 2.2694736842105265e-06, | |
| "loss": 0.0001, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 17.857142857142858, | |
| "grad_norm": 0.006168752908706665, | |
| "learning_rate": 2.2431578947368423e-06, | |
| "loss": 0.0012, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 17.913832199546484, | |
| "grad_norm": 0.006377949379384518, | |
| "learning_rate": 2.216842105263158e-06, | |
| "loss": 0.0052, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 17.970521541950113, | |
| "grad_norm": 1.9607151746749878, | |
| "learning_rate": 2.190526315789474e-06, | |
| "loss": 0.0004, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 18.027210884353742, | |
| "grad_norm": 0.0046304683201014996, | |
| "learning_rate": 2.16421052631579e-06, | |
| "loss": 0.001, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 18.08390022675737, | |
| "grad_norm": 0.008269163779914379, | |
| "learning_rate": 2.1378947368421056e-06, | |
| "loss": 0.0001, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 18.140589569160998, | |
| "grad_norm": 0.0044001140631735325, | |
| "learning_rate": 2.111578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 18.140589569160998, | |
| "eval_loss": 0.2317376732826233, | |
| "eval_runtime": 158.0944, | |
| "eval_samples_per_second": 2.96, | |
| "eval_steps_per_second": 0.493, | |
| "eval_wer": 6.076388888888888, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 18.197278911564627, | |
| "grad_norm": 0.006642814259976149, | |
| "learning_rate": 2.085263157894737e-06, | |
| "loss": 0.0001, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 18.253968253968253, | |
| "grad_norm": 0.00599477905780077, | |
| "learning_rate": 2.058947368421053e-06, | |
| "loss": 0.0001, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 18.310657596371883, | |
| "grad_norm": 0.0045234388671815395, | |
| "learning_rate": 2.0326315789473685e-06, | |
| "loss": 0.0001, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 18.367346938775512, | |
| "grad_norm": 0.002616587560623884, | |
| "learning_rate": 2.0063157894736843e-06, | |
| "loss": 0.0001, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 18.424036281179138, | |
| "grad_norm": 0.0050145648419857025, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.0001, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 18.480725623582767, | |
| "grad_norm": 0.0029045080300420523, | |
| "learning_rate": 1.953684210526316e-06, | |
| "loss": 0.0002, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 18.537414965986393, | |
| "grad_norm": 0.0041219014674425125, | |
| "learning_rate": 1.9273684210526317e-06, | |
| "loss": 0.0009, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 18.594104308390023, | |
| "grad_norm": 0.006918082479387522, | |
| "learning_rate": 1.9010526315789476e-06, | |
| "loss": 0.0001, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 18.650793650793652, | |
| "grad_norm": 0.0045529440976679325, | |
| "learning_rate": 1.8747368421052634e-06, | |
| "loss": 0.0005, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 18.707482993197278, | |
| "grad_norm": 0.004559030756354332, | |
| "learning_rate": 1.848421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 18.764172335600907, | |
| "grad_norm": 0.003188680624589324, | |
| "learning_rate": 1.8221052631578948e-06, | |
| "loss": 0.0001, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 18.820861678004537, | |
| "grad_norm": 0.0051582190208137035, | |
| "learning_rate": 1.7957894736842108e-06, | |
| "loss": 0.0001, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 18.877551020408163, | |
| "grad_norm": 0.003925441298633814, | |
| "learning_rate": 1.7694736842105264e-06, | |
| "loss": 0.0001, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 18.934240362811792, | |
| "grad_norm": 0.005711190402507782, | |
| "learning_rate": 1.7431578947368423e-06, | |
| "loss": 0.0, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 18.990929705215418, | |
| "grad_norm": 0.006396492477506399, | |
| "learning_rate": 1.716842105263158e-06, | |
| "loss": 0.0002, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 19.047619047619047, | |
| "grad_norm": 0.0038605357985943556, | |
| "learning_rate": 1.6905263157894739e-06, | |
| "loss": 0.0001, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 19.104308390022677, | |
| "grad_norm": 0.003054672619327903, | |
| "learning_rate": 1.6642105263157895e-06, | |
| "loss": 0.0, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 19.160997732426303, | |
| "grad_norm": 0.0045293658040463924, | |
| "learning_rate": 1.6378947368421053e-06, | |
| "loss": 0.0, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 19.217687074829932, | |
| "grad_norm": 0.005090142600238323, | |
| "learning_rate": 1.6115789473684211e-06, | |
| "loss": 0.0001, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 19.27437641723356, | |
| "grad_norm": 0.003649334190413356, | |
| "learning_rate": 1.585263157894737e-06, | |
| "loss": 0.0, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 19.331065759637188, | |
| "grad_norm": 0.014431001618504524, | |
| "learning_rate": 1.5589473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 19.387755102040817, | |
| "grad_norm": 0.005074130836874247, | |
| "learning_rate": 1.5326315789473686e-06, | |
| "loss": 0.0, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 19.444444444444443, | |
| "grad_norm": 0.0036257512401789427, | |
| "learning_rate": 1.5063157894736844e-06, | |
| "loss": 0.0001, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 19.501133786848072, | |
| "grad_norm": 0.0060266111977398396, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.0001, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 19.5578231292517, | |
| "grad_norm": 0.004775646608322859, | |
| "learning_rate": 1.453684210526316e-06, | |
| "loss": 0.0, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 19.614512471655328, | |
| "grad_norm": 0.006195446942001581, | |
| "learning_rate": 1.4273684210526317e-06, | |
| "loss": 0.0003, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 19.671201814058957, | |
| "grad_norm": 0.0044461763463914394, | |
| "learning_rate": 1.4010526315789475e-06, | |
| "loss": 0.0001, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 19.727891156462587, | |
| "grad_norm": 0.0022899750620126724, | |
| "learning_rate": 1.374736842105263e-06, | |
| "loss": 0.0, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 19.784580498866212, | |
| "grad_norm": 0.005077675450593233, | |
| "learning_rate": 1.3484210526315791e-06, | |
| "loss": 0.0001, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 19.841269841269842, | |
| "grad_norm": 0.005482817534357309, | |
| "learning_rate": 1.322105263157895e-06, | |
| "loss": 0.0, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 19.897959183673468, | |
| "grad_norm": 0.00202305824495852, | |
| "learning_rate": 1.2957894736842105e-06, | |
| "loss": 0.0001, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 19.954648526077097, | |
| "grad_norm": 0.005788388196378946, | |
| "learning_rate": 1.2694736842105266e-06, | |
| "loss": 0.0001, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 20.011337868480727, | |
| "grad_norm": 0.004344166722148657, | |
| "learning_rate": 1.2431578947368422e-06, | |
| "loss": 0.0014, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 20.068027210884352, | |
| "grad_norm": 0.004923074971884489, | |
| "learning_rate": 1.216842105263158e-06, | |
| "loss": 0.0, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 20.124716553287982, | |
| "grad_norm": 0.0024910017382353544, | |
| "learning_rate": 1.1905263157894738e-06, | |
| "loss": 0.0, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 20.18140589569161, | |
| "grad_norm": 0.0026297103613615036, | |
| "learning_rate": 1.1642105263157896e-06, | |
| "loss": 0.0, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 20.238095238095237, | |
| "grad_norm": 0.005076109431684017, | |
| "learning_rate": 1.1378947368421052e-06, | |
| "loss": 0.0001, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 20.294784580498867, | |
| "grad_norm": 0.003189537674188614, | |
| "learning_rate": 1.1115789473684213e-06, | |
| "loss": 0.0, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 20.351473922902493, | |
| "grad_norm": 0.0030386645812541246, | |
| "learning_rate": 1.0852631578947369e-06, | |
| "loss": 0.0, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 20.408163265306122, | |
| "grad_norm": 0.001944896299391985, | |
| "learning_rate": 1.0589473684210527e-06, | |
| "loss": 0.0001, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 20.408163265306122, | |
| "eval_loss": 0.23750941455364227, | |
| "eval_runtime": 160.3655, | |
| "eval_samples_per_second": 2.918, | |
| "eval_steps_per_second": 0.486, | |
| "eval_wer": 6.396901709401709, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 20.46485260770975, | |
| "grad_norm": 0.0040268674492836, | |
| "learning_rate": 1.0326315789473685e-06, | |
| "loss": 0.0, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 20.521541950113377, | |
| "grad_norm": 2.8923232555389404, | |
| "learning_rate": 1.0063157894736843e-06, | |
| "loss": 0.0002, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 20.578231292517007, | |
| "grad_norm": 0.002543982584029436, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 0.0, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 20.634920634920636, | |
| "grad_norm": 0.002120724180713296, | |
| "learning_rate": 9.536842105263158e-07, | |
| "loss": 0.0002, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 20.691609977324262, | |
| "grad_norm": 0.0036805281415581703, | |
| "learning_rate": 9.273684210526317e-07, | |
| "loss": 0.0, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 20.74829931972789, | |
| "grad_norm": 0.002348339883610606, | |
| "learning_rate": 9.010526315789474e-07, | |
| "loss": 0.0001, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 20.80498866213152, | |
| "grad_norm": 0.0036750957369804382, | |
| "learning_rate": 8.747368421052632e-07, | |
| "loss": 0.0, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 20.861678004535147, | |
| "grad_norm": 0.004974485840648413, | |
| "learning_rate": 8.48421052631579e-07, | |
| "loss": 0.0, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 20.918367346938776, | |
| "grad_norm": 0.004362870939075947, | |
| "learning_rate": 8.221052631578947e-07, | |
| "loss": 0.0001, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 20.975056689342402, | |
| "grad_norm": 0.002504055853933096, | |
| "learning_rate": 7.957894736842107e-07, | |
| "loss": 0.0001, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 21.03174603174603, | |
| "grad_norm": 0.004000427667051554, | |
| "learning_rate": 7.694736842105263e-07, | |
| "loss": 0.0001, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 21.08843537414966, | |
| "grad_norm": 0.003886349266394973, | |
| "learning_rate": 7.431578947368422e-07, | |
| "loss": 0.0, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 21.145124716553287, | |
| "grad_norm": 0.003481630701571703, | |
| "learning_rate": 7.16842105263158e-07, | |
| "loss": 0.0, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 21.201814058956916, | |
| "grad_norm": 0.01767110824584961, | |
| "learning_rate": 6.905263157894737e-07, | |
| "loss": 0.0, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 21.258503401360546, | |
| "grad_norm": 0.005323050078004599, | |
| "learning_rate": 6.642105263157895e-07, | |
| "loss": 0.0, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 21.31519274376417, | |
| "grad_norm": 0.05196991562843323, | |
| "learning_rate": 6.378947368421053e-07, | |
| "loss": 0.0001, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 21.3718820861678, | |
| "grad_norm": 0.003023393452167511, | |
| "learning_rate": 6.115789473684211e-07, | |
| "loss": 0.0, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 21.428571428571427, | |
| "grad_norm": 0.0037847934290766716, | |
| "learning_rate": 5.852631578947369e-07, | |
| "loss": 0.0, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 21.485260770975056, | |
| "grad_norm": 0.0039050974883139133, | |
| "learning_rate": 5.589473684210526e-07, | |
| "loss": 0.0001, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 21.541950113378686, | |
| "grad_norm": 0.0036291517317295074, | |
| "learning_rate": 5.326315789473684e-07, | |
| "loss": 0.0001, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 21.598639455782312, | |
| "grad_norm": 0.003725625341758132, | |
| "learning_rate": 5.063157894736842e-07, | |
| "loss": 0.0001, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 21.65532879818594, | |
| "grad_norm": 0.0034233913756906986, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 0.0, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 21.71201814058957, | |
| "grad_norm": 0.005571336485445499, | |
| "learning_rate": 4.5368421052631583e-07, | |
| "loss": 0.0, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 21.768707482993197, | |
| "grad_norm": 0.0034070161636918783, | |
| "learning_rate": 4.273684210526316e-07, | |
| "loss": 0.0001, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 21.825396825396826, | |
| "grad_norm": 0.0027184481732547283, | |
| "learning_rate": 4.0105263157894736e-07, | |
| "loss": 0.0, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 21.882086167800452, | |
| "grad_norm": 0.001929171965457499, | |
| "learning_rate": 3.7473684210526323e-07, | |
| "loss": 0.0, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 21.93877551020408, | |
| "grad_norm": 0.003997990861535072, | |
| "learning_rate": 3.48421052631579e-07, | |
| "loss": 0.0, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 21.99546485260771, | |
| "grad_norm": 0.0038647083565592766, | |
| "learning_rate": 3.2210526315789476e-07, | |
| "loss": 0.0, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 22.052154195011337, | |
| "grad_norm": 0.00401474442332983, | |
| "learning_rate": 2.9578947368421053e-07, | |
| "loss": 0.0, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 22.108843537414966, | |
| "grad_norm": 0.0030088857747614384, | |
| "learning_rate": 2.6947368421052635e-07, | |
| "loss": 0.0001, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 22.165532879818596, | |
| "grad_norm": 0.003003130666911602, | |
| "learning_rate": 2.431578947368421e-07, | |
| "loss": 0.0, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 22.22222222222222, | |
| "grad_norm": 0.004516700282692909, | |
| "learning_rate": 2.168421052631579e-07, | |
| "loss": 0.0, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 22.27891156462585, | |
| "grad_norm": 0.002650737063959241, | |
| "learning_rate": 1.9052631578947372e-07, | |
| "loss": 0.0, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 22.335600907029477, | |
| "grad_norm": 0.00888384971767664, | |
| "learning_rate": 1.642105263157895e-07, | |
| "loss": 0.0, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 22.392290249433106, | |
| "grad_norm": 0.0020597188267856836, | |
| "learning_rate": 1.3789473684210528e-07, | |
| "loss": 0.0, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 22.448979591836736, | |
| "grad_norm": 0.004189135041087866, | |
| "learning_rate": 1.1157894736842106e-07, | |
| "loss": 0.0, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 22.50566893424036, | |
| "grad_norm": 0.003128908108919859, | |
| "learning_rate": 8.526315789473685e-08, | |
| "loss": 0.0, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 22.56235827664399, | |
| "grad_norm": 0.0031592377927154303, | |
| "learning_rate": 5.8947368421052637e-08, | |
| "loss": 0.0, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 22.61904761904762, | |
| "grad_norm": 0.004137367941439152, | |
| "learning_rate": 3.263157894736842e-08, | |
| "loss": 0.0, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 22.675736961451246, | |
| "grad_norm": 0.0022842560429126024, | |
| "learning_rate": 6.315789473684211e-09, | |
| "loss": 0.0, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 22.675736961451246, | |
| "eval_loss": 0.23789168894290924, | |
| "eval_runtime": 182.0655, | |
| "eval_samples_per_second": 2.571, | |
| "eval_steps_per_second": 0.428, | |
| "eval_wer": 6.490384615384616, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 22.675736961451246, | |
| "step": 10000, | |
| "total_flos": 1.73151240192e+19, | |
| "train_loss": 0.06552563527043676, | |
| "train_runtime": 7527.9049, | |
| "train_samples_per_second": 7.97, | |
| "train_steps_per_second": 1.328 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 23, | |
| "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": 1.73151240192e+19, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |