| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.1952, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 1.7953, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.6464, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.46e-06, | |
| "loss": 1.521, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 1.3862, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.46e-06, | |
| "loss": 1.114, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.96e-06, | |
| "loss": 1.0264, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.9577, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.9321, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.8744, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.8655, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.8481, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.8753, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.8619, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.8021, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.7985, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.787, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.766, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.7584, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.7492, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.7711, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.94888888888889e-06, | |
| "loss": 0.8271, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.893333333333334e-06, | |
| "loss": 0.7849, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.837777777777778e-06, | |
| "loss": 0.74, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.782222222222222e-06, | |
| "loss": 0.7151, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.726666666666668e-06, | |
| "loss": 0.7309, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.671111111111112e-06, | |
| "loss": 0.7087, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.615555555555558e-06, | |
| "loss": 0.7169, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.56e-06, | |
| "loss": 0.6725, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.504444444444446e-06, | |
| "loss": 0.6581, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.44888888888889e-06, | |
| "loss": 0.673, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.393333333333334e-06, | |
| "loss": 0.6402, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.33777777777778e-06, | |
| "loss": 0.728, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.282222222222222e-06, | |
| "loss": 0.7284, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.226666666666668e-06, | |
| "loss": 0.7507, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.171111111111112e-06, | |
| "loss": 0.7309, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.115555555555556e-06, | |
| "loss": 0.713, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.6999, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.004444444444445e-06, | |
| "loss": 0.7197, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.94888888888889e-06, | |
| "loss": 0.6895, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.893333333333333e-06, | |
| "loss": 0.7185, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.7607626914978027, | |
| "eval_runtime": 2486.7575, | |
| "eval_samples_per_second": 6.47, | |
| "eval_steps_per_second": 0.405, | |
| "eval_wer": 38.16357512215949, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.83777777777778e-06, | |
| "loss": 0.7057, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.782222222222223e-06, | |
| "loss": 0.6478, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.726666666666667e-06, | |
| "loss": 0.6414, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.671111111111113e-06, | |
| "loss": 0.6536, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.615555555555555e-06, | |
| "loss": 0.6539, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 0.6492, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.504444444444445e-06, | |
| "loss": 0.6283, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.448888888888889e-06, | |
| "loss": 0.6183, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.393333333333335e-06, | |
| "loss": 0.6249, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.337777777777777e-06, | |
| "loss": 0.6717, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.282222222222223e-06, | |
| "loss": 0.5978, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.226666666666667e-06, | |
| "loss": 0.6155, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.171111111111113e-06, | |
| "loss": 0.6114, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.611, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.5953, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.5647, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.5527, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.5348, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.6021, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.5874, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.558, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.6495, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.6047, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 0.5399, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.504444444444445e-06, | |
| "loss": 0.5474, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.44888888888889e-06, | |
| "loss": 0.5658, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.393333333333333e-06, | |
| "loss": 0.5646, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.337777777777778e-06, | |
| "loss": 0.5939, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.282222222222222e-06, | |
| "loss": 0.584, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.226666666666667e-06, | |
| "loss": 0.5925, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.171111111111112e-06, | |
| "loss": 0.5813, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.115555555555557e-06, | |
| "loss": 0.5994, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.6274, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.004444444444445e-06, | |
| "loss": 0.6248, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.948888888888889e-06, | |
| "loss": 0.606, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.5844, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.837777777777779e-06, | |
| "loss": 0.6259, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.782222222222222e-06, | |
| "loss": 0.5586, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.726666666666667e-06, | |
| "loss": 0.6175, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.671111111111112e-06, | |
| "loss": 0.6052, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_loss": 0.6948559284210205, | |
| "eval_runtime": 2462.093, | |
| "eval_samples_per_second": 6.535, | |
| "eval_steps_per_second": 0.409, | |
| "eval_wer": 34.95125480335879, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.615555555555556e-06, | |
| "loss": 0.5687, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.7425, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.504444444444446e-06, | |
| "loss": 0.7846, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.448888888888889e-06, | |
| "loss": 0.719, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.393333333333334e-06, | |
| "loss": 0.6106, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.3377777777777786e-06, | |
| "loss": 0.5368, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.282222222222223e-06, | |
| "loss": 0.481, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.2266666666666675e-06, | |
| "loss": 0.4964, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.171111111111112e-06, | |
| "loss": 0.4791, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.4903, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.4848, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.5104, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.948888888888889e-06, | |
| "loss": 0.5144, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.893333333333334e-06, | |
| "loss": 0.5078, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.837777777777777e-06, | |
| "loss": 0.5206, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.782222222222222e-06, | |
| "loss": 0.5087, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.726666666666667e-06, | |
| "loss": 0.4925, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.671111111111112e-06, | |
| "loss": 0.4732, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.615555555555556e-06, | |
| "loss": 0.5057, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 0.4803, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.504444444444444e-06, | |
| "loss": 0.4691, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.448888888888889e-06, | |
| "loss": 0.4661, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.393333333333334e-06, | |
| "loss": 0.4578, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.337777777777779e-06, | |
| "loss": 0.4164, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.282222222222223e-06, | |
| "loss": 0.43, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.226666666666667e-06, | |
| "loss": 0.4556, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.171111111111111e-06, | |
| "loss": 0.4852, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.115555555555556e-06, | |
| "loss": 0.4722, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.4281, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.004444444444445e-06, | |
| "loss": 0.4696, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.94888888888889e-06, | |
| "loss": 0.5112, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.893333333333334e-06, | |
| "loss": 0.4951, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.837777777777778e-06, | |
| "loss": 0.4962, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.7822222222222226e-06, | |
| "loss": 0.5156, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.7266666666666674e-06, | |
| "loss": 0.4957, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.6711111111111115e-06, | |
| "loss": 0.4678, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.6155555555555555e-06, | |
| "loss": 0.5, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.56e-06, | |
| "loss": 0.4721, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.504444444444444e-06, | |
| "loss": 0.4345, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.448888888888889e-06, | |
| "loss": 0.4467, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_loss": 0.6708351969718933, | |
| "eval_runtime": 2295.5806, | |
| "eval_samples_per_second": 7.009, | |
| "eval_steps_per_second": 0.438, | |
| "eval_wer": 34.33927131268087, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.393333333333334e-06, | |
| "loss": 0.4815, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.337777777777778e-06, | |
| "loss": 0.488, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.282222222222222e-06, | |
| "loss": 0.5183, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.226666666666667e-06, | |
| "loss": 0.5115, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.171111111111111e-06, | |
| "loss": 0.5123, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.115555555555556e-06, | |
| "loss": 0.4831, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.4807, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.004444444444445e-06, | |
| "loss": 0.4665, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.948888888888889e-06, | |
| "loss": 0.4621, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.893333333333333e-06, | |
| "loss": 0.494, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.837777777777778e-06, | |
| "loss": 0.4437, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.782222222222223e-06, | |
| "loss": 0.5556, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.726666666666667e-06, | |
| "loss": 0.5229, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.6711111111111113e-06, | |
| "loss": 0.4869, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.615555555555556e-06, | |
| "loss": 0.4771, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 0.4483, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.5044444444444447e-06, | |
| "loss": 0.3973, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.4488888888888896e-06, | |
| "loss": 0.4525, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.3933333333333336e-06, | |
| "loss": 0.4284, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.337777777777778e-06, | |
| "loss": 0.4275, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.282222222222223e-06, | |
| "loss": 0.4534, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.226666666666667e-06, | |
| "loss": 0.4828, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.1711111111111114e-06, | |
| "loss": 0.4538, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.1155555555555555e-06, | |
| "loss": 0.461, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.4601, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.004444444444445e-06, | |
| "loss": 0.4651, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.948888888888889e-06, | |
| "loss": 0.4528, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.8933333333333337e-06, | |
| "loss": 0.4504, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.837777777777778e-06, | |
| "loss": 0.4585, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.7822222222222222e-06, | |
| "loss": 0.4413, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.726666666666667e-06, | |
| "loss": 0.4486, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.6711111111111116e-06, | |
| "loss": 0.4194, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.6155555555555556e-06, | |
| "loss": 0.4928, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.4946, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.504444444444445e-06, | |
| "loss": 0.4639, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.448888888888889e-06, | |
| "loss": 0.4771, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.3933333333333334e-06, | |
| "loss": 0.4876, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.337777777777778e-06, | |
| "loss": 0.4912, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.2822222222222223e-06, | |
| "loss": 0.4738, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.226666666666667e-06, | |
| "loss": 0.4773, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "eval_loss": 0.6536200046539307, | |
| "eval_runtime": 2257.5997, | |
| "eval_samples_per_second": 7.127, | |
| "eval_steps_per_second": 0.446, | |
| "eval_wer": 33.21018549267043, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2.1711111111111113e-06, | |
| "loss": 0.4798, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.1155555555555557e-06, | |
| "loss": 0.4668, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.06e-06, | |
| "loss": 0.4737, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.5164, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.5055, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.4588, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.4409, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.4718, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.4616, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.4453, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.4419, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.5293, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.5006, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.541, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.5413, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.5523, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.515, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.4911, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.527, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.5076, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.4598, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.4625, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.4092, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.3465, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.3815, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.3879, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.3723, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.3688, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.3604, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.3971, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.4186, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.4418, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.4241, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.4739, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.4342, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.4776, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.477, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.4767, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.4393, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.4479, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "eval_loss": 0.6509412527084351, | |
| "eval_runtime": 2321.7687, | |
| "eval_samples_per_second": 6.93, | |
| "eval_steps_per_second": 0.433, | |
| "eval_wer": 33.06549172161867, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "step": 5000, | |
| "total_flos": 3.93764296163328e+18, | |
| "train_loss": 0.5807094034194946, | |
| "train_runtime": 31914.6141, | |
| "train_samples_per_second": 5.013, | |
| "train_steps_per_second": 0.157 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 3.93764296163328e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |