| { | |
| "best_metric": 19.772043934466467, | |
| "best_model_checkpoint": "./checkpoint-9000", | |
| "epoch": 4.880429477794046, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.25e-06, | |
| "loss": 3.0853, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.75e-06, | |
| "loss": 1.2521, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.25e-06, | |
| "loss": 0.9253, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.75e-06, | |
| "loss": 0.6905, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.954081632653062e-06, | |
| "loss": 0.3063, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.903061224489798e-06, | |
| "loss": 0.1729, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.852040816326532e-06, | |
| "loss": 0.1352, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.801020408163266e-06, | |
| "loss": 0.1091, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.75e-06, | |
| "loss": 0.1015, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.698979591836736e-06, | |
| "loss": 0.086, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.64795918367347e-06, | |
| "loss": 0.0787, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.596938775510205e-06, | |
| "loss": 0.072, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.54591836734694e-06, | |
| "loss": 0.066, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.494897959183675e-06, | |
| "loss": 0.0615, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.44387755102041e-06, | |
| "loss": 0.0562, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.392857142857143e-06, | |
| "loss": 0.0506, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.341836734693879e-06, | |
| "loss": 0.0501, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.290816326530613e-06, | |
| "loss": 0.0466, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.239795918367349e-06, | |
| "loss": 0.0436, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.188775510204083e-06, | |
| "loss": 0.0384, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_loss": 0.13486750423908234, | |
| "eval_runtime": 653.9606, | |
| "eval_samples_per_second": 5.308, | |
| "eval_steps_per_second": 0.167, | |
| "eval_wer": 40.37399521384304, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.137755102040817e-06, | |
| "loss": 0.0386, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.08673469387755e-06, | |
| "loss": 0.0341, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.035714285714287e-06, | |
| "loss": 0.0335, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.98469387755102e-06, | |
| "loss": 0.0321, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.933673469387756e-06, | |
| "loss": 0.0311, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.88265306122449e-06, | |
| "loss": 0.0273, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.831632653061224e-06, | |
| "loss": 0.0277, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.78061224489796e-06, | |
| "loss": 0.0281, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.729591836734694e-06, | |
| "loss": 0.0257, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.67857142857143e-06, | |
| "loss": 0.0254, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.627551020408164e-06, | |
| "loss": 0.024, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.576530612244898e-06, | |
| "loss": 0.0229, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.525510204081632e-06, | |
| "loss": 0.0217, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.474489795918368e-06, | |
| "loss": 0.0207, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.423469387755104e-06, | |
| "loss": 0.0204, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.372448979591838e-06, | |
| "loss": 0.0203, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.321428571428573e-06, | |
| "loss": 0.0185, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.270408163265307e-06, | |
| "loss": 0.0191, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.219387755102041e-06, | |
| "loss": 0.0186, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.168367346938775e-06, | |
| "loss": 0.0175, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_loss": 0.1600993424654007, | |
| "eval_runtime": 629.294, | |
| "eval_samples_per_second": 5.516, | |
| "eval_steps_per_second": 0.173, | |
| "eval_wer": 22.646806160643063, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.117346938775511e-06, | |
| "loss": 0.0159, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.066326530612245e-06, | |
| "loss": 0.0128, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.015306122448981e-06, | |
| "loss": 0.012, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.964285714285715e-06, | |
| "loss": 0.011, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.913265306122449e-06, | |
| "loss": 0.0121, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.862244897959185e-06, | |
| "loss": 0.0126, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.811224489795919e-06, | |
| "loss": 0.0131, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.760204081632655e-06, | |
| "loss": 0.0124, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.709183673469389e-06, | |
| "loss": 0.0109, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.658163265306123e-06, | |
| "loss": 0.011, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.6071428571428575e-06, | |
| "loss": 0.0104, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.556122448979592e-06, | |
| "loss": 0.0104, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.5051020408163264e-06, | |
| "loss": 0.0109, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.454081632653062e-06, | |
| "loss": 0.0109, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.403061224489797e-06, | |
| "loss": 0.0103, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.352040816326531e-06, | |
| "loss": 0.0106, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.301020408163266e-06, | |
| "loss": 0.0093, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 7.25e-06, | |
| "loss": 0.0104, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.198979591836736e-06, | |
| "loss": 0.0092, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.14795918367347e-06, | |
| "loss": 0.0091, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "eval_loss": 0.1817348748445511, | |
| "eval_runtime": 622.4071, | |
| "eval_samples_per_second": 5.577, | |
| "eval_steps_per_second": 0.175, | |
| "eval_wer": 23.15150027612444, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.096938775510205e-06, | |
| "loss": 0.0097, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.045918367346939e-06, | |
| "loss": 0.0095, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.9948979591836745e-06, | |
| "loss": 0.01, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.943877551020409e-06, | |
| "loss": 0.0096, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.892857142857143e-06, | |
| "loss": 0.0096, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.841836734693878e-06, | |
| "loss": 0.0095, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.790816326530612e-06, | |
| "loss": 0.0089, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.739795918367348e-06, | |
| "loss": 0.0077, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 6.688775510204082e-06, | |
| "loss": 0.0084, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 6.637755102040817e-06, | |
| "loss": 0.0091, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.586734693877551e-06, | |
| "loss": 0.0084, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 6.535714285714286e-06, | |
| "loss": 0.0099, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.484693877551021e-06, | |
| "loss": 0.0087, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 6.433673469387756e-06, | |
| "loss": 0.0102, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.382653061224491e-06, | |
| "loss": 0.0083, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.331632653061225e-06, | |
| "loss": 0.0068, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.28061224489796e-06, | |
| "loss": 0.0082, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.2295918367346944e-06, | |
| "loss": 0.0081, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.178571428571429e-06, | |
| "loss": 0.0075, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.127551020408163e-06, | |
| "loss": 0.0082, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "eval_loss": 0.21248996257781982, | |
| "eval_runtime": 634.8932, | |
| "eval_samples_per_second": 5.467, | |
| "eval_steps_per_second": 0.172, | |
| "eval_wer": 23.913910535681413, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.076530612244898e-06, | |
| "loss": 0.0081, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.025510204081632e-06, | |
| "loss": 0.0073, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.974489795918368e-06, | |
| "loss": 0.0047, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.923469387755103e-06, | |
| "loss": 0.004, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.872448979591837e-06, | |
| "loss": 0.004, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.821428571428573e-06, | |
| "loss": 0.0046, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.770408163265307e-06, | |
| "loss": 0.0049, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.719387755102042e-06, | |
| "loss": 0.0047, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.668367346938776e-06, | |
| "loss": 0.0035, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.6173469387755105e-06, | |
| "loss": 0.004, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.566326530612245e-06, | |
| "loss": 0.0041, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.51530612244898e-06, | |
| "loss": 0.0049, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.464285714285714e-06, | |
| "loss": 0.0043, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.413265306122449e-06, | |
| "loss": 0.0041, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5.362244897959185e-06, | |
| "loss": 0.004, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5.311224489795919e-06, | |
| "loss": 0.0045, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.260204081632654e-06, | |
| "loss": 0.0044, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5.209183673469388e-06, | |
| "loss": 0.0044, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.158163265306123e-06, | |
| "loss": 0.0043, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.107142857142857e-06, | |
| "loss": 0.0048, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "eval_loss": 0.21101494133472443, | |
| "eval_runtime": 625.1988, | |
| "eval_samples_per_second": 5.552, | |
| "eval_steps_per_second": 0.174, | |
| "eval_wer": 20.252193655273977, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.056122448979593e-06, | |
| "loss": 0.0039, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.005102040816327e-06, | |
| "loss": 0.004, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 4.9540816326530616e-06, | |
| "loss": 0.0049, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.9030612244897964e-06, | |
| "loss": 0.0037, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.8520408163265305e-06, | |
| "loss": 0.004, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.801020408163265e-06, | |
| "loss": 0.0041, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 4.75e-06, | |
| "loss": 0.0035, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 4.698979591836735e-06, | |
| "loss": 0.0047, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.64795918367347e-06, | |
| "loss": 0.0044, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.596938775510205e-06, | |
| "loss": 0.0037, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.545918367346939e-06, | |
| "loss": 0.0041, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.494897959183674e-06, | |
| "loss": 0.0042, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.443877551020409e-06, | |
| "loss": 0.0034, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.392857142857143e-06, | |
| "loss": 0.0043, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.341836734693878e-06, | |
| "loss": 0.004, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.2908163265306126e-06, | |
| "loss": 0.0043, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.2397959183673475e-06, | |
| "loss": 0.0041, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.188775510204082e-06, | |
| "loss": 0.0039, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.137755102040817e-06, | |
| "loss": 0.0034, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.086734693877551e-06, | |
| "loss": 0.0037, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "eval_loss": 0.22696612775325775, | |
| "eval_runtime": 628.453, | |
| "eval_samples_per_second": 5.523, | |
| "eval_steps_per_second": 0.173, | |
| "eval_wer": 21.485549487635762, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.035714285714286e-06, | |
| "loss": 0.0039, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.984693877551021e-06, | |
| "loss": 0.0039, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.933673469387755e-06, | |
| "loss": 0.0035, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.88265306122449e-06, | |
| "loss": 0.0019, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.831632653061225e-06, | |
| "loss": 0.0018, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.7806122448979593e-06, | |
| "loss": 0.0023, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.729591836734694e-06, | |
| "loss": 0.0023, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.678571428571429e-06, | |
| "loss": 0.0021, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.6275510204081636e-06, | |
| "loss": 0.0018, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.5765306122448985e-06, | |
| "loss": 0.0025, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.525510204081633e-06, | |
| "loss": 0.0022, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.474489795918368e-06, | |
| "loss": 0.002, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.4234693877551023e-06, | |
| "loss": 0.0025, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.3724489795918367e-06, | |
| "loss": 0.002, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.3214285714285716e-06, | |
| "loss": 0.0022, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.270408163265306e-06, | |
| "loss": 0.0019, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.219387755102041e-06, | |
| "loss": 0.0017, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.168367346938776e-06, | |
| "loss": 0.0022, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.1173469387755108e-06, | |
| "loss": 0.002, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.0663265306122452e-06, | |
| "loss": 0.0017, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "eval_loss": 0.2534159719944, | |
| "eval_runtime": 625.513, | |
| "eval_samples_per_second": 5.549, | |
| "eval_steps_per_second": 0.174, | |
| "eval_wer": 20.239921457937044, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.01530612244898e-06, | |
| "loss": 0.002, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 2.9642857142857146e-06, | |
| "loss": 0.0022, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 2.913265306122449e-06, | |
| "loss": 0.0021, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 2.862244897959184e-06, | |
| "loss": 0.002, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 2.8112244897959184e-06, | |
| "loss": 0.0021, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.760204081632653e-06, | |
| "loss": 0.0022, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.7091836734693878e-06, | |
| "loss": 0.0021, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.658163265306123e-06, | |
| "loss": 0.002, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 2.6071428571428575e-06, | |
| "loss": 0.0022, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 2.556122448979592e-06, | |
| "loss": 0.0019, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 2.505102040816327e-06, | |
| "loss": 0.0017, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.4540816326530614e-06, | |
| "loss": 0.0015, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.4030612244897962e-06, | |
| "loss": 0.0018, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 2.3520408163265307e-06, | |
| "loss": 0.0018, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 2.3010204081632656e-06, | |
| "loss": 0.002, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 2.25e-06, | |
| "loss": 0.0017, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.198979591836735e-06, | |
| "loss": 0.0016, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 2.1479591836734694e-06, | |
| "loss": 0.0015, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 2.0969387755102043e-06, | |
| "loss": 0.0017, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.045918367346939e-06, | |
| "loss": 0.0018, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "eval_loss": 0.27059340476989746, | |
| "eval_runtime": 624.7551, | |
| "eval_samples_per_second": 5.556, | |
| "eval_steps_per_second": 0.174, | |
| "eval_wer": 20.727741302080137, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.9948979591836737e-06, | |
| "loss": 0.0017, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.943877551020408e-06, | |
| "loss": 0.0018, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.892857142857143e-06, | |
| "loss": 0.0014, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.8418367346938777e-06, | |
| "loss": 0.0017, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.7908163265306126e-06, | |
| "loss": 0.001, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.739795918367347e-06, | |
| "loss": 0.001, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.6887755102040817e-06, | |
| "loss": 0.0009, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.6377551020408164e-06, | |
| "loss": 0.0009, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 1.586734693877551e-06, | |
| "loss": 0.0009, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.535714285714286e-06, | |
| "loss": 0.0009, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 1.4846938775510206e-06, | |
| "loss": 0.0009, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.433673469387755e-06, | |
| "loss": 0.001, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 1.3826530612244898e-06, | |
| "loss": 0.0009, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 1.3316326530612245e-06, | |
| "loss": 0.0008, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 1.2806122448979593e-06, | |
| "loss": 0.0007, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 1.229591836734694e-06, | |
| "loss": 0.0009, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 1.1785714285714287e-06, | |
| "loss": 0.0009, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 1.1275510204081634e-06, | |
| "loss": 0.0007, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 1.076530612244898e-06, | |
| "loss": 0.0009, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 1.0255102040816327e-06, | |
| "loss": 0.0005, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "eval_loss": 0.2806273102760315, | |
| "eval_runtime": 621.569, | |
| "eval_samples_per_second": 5.584, | |
| "eval_steps_per_second": 0.175, | |
| "eval_wer": 19.772043934466467, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 9.744897959183674e-07, | |
| "loss": 0.0011, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 9.234693877551021e-07, | |
| "loss": 0.0009, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 8.724489795918369e-07, | |
| "loss": 0.0008, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 8.214285714285715e-07, | |
| "loss": 0.0009, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 7.704081632653061e-07, | |
| "loss": 0.0007, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 7.193877551020409e-07, | |
| "loss": 0.0006, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 6.683673469387756e-07, | |
| "loss": 0.0008, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 6.173469387755102e-07, | |
| "loss": 0.0008, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 5.663265306122449e-07, | |
| "loss": 0.001, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 5.153061224489796e-07, | |
| "loss": 0.0008, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 4.642857142857143e-07, | |
| "loss": 0.0006, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 4.1326530612244905e-07, | |
| "loss": 0.001, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 3.622448979591837e-07, | |
| "loss": 0.001, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.1122448979591836e-07, | |
| "loss": 0.0007, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.602040816326531e-07, | |
| "loss": 0.0008, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.091836734693878e-07, | |
| "loss": 0.0007, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.5816326530612244e-07, | |
| "loss": 0.0008, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 1.0714285714285716e-07, | |
| "loss": 0.0006, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.612244897959184e-08, | |
| "loss": 0.0007, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 5.102040816326532e-09, | |
| "loss": 0.0006, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "eval_loss": 0.29291146993637085, | |
| "eval_runtime": 621.652, | |
| "eval_samples_per_second": 5.584, | |
| "eval_steps_per_second": 0.175, | |
| "eval_wer": 20.974719273485917, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "step": 10000, | |
| "total_flos": 9.231154351276032e+19, | |
| "train_loss": 0.043158661832101645, | |
| "train_runtime": 45207.8362, | |
| "train_samples_per_second": 7.078, | |
| "train_steps_per_second": 0.221 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 5, | |
| "total_flos": 9.231154351276032e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |