| { | |
| "best_metric": 54.70462356526814, | |
| "best_model_checkpoint": "./output/small/yt-special-batch4-lr6-small/checkpoint-5000", | |
| "epoch": 0.4294425835265825, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4e-08, | |
| "loss": 2.4854, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9e-08, | |
| "loss": 2.3311, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4e-07, | |
| "loss": 2.217, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.8999999999999998e-07, | |
| "loss": 2.3064, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4e-07, | |
| "loss": 2.4084, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9e-07, | |
| "loss": 2.299, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 1.8911, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8999999999999997e-07, | |
| "loss": 2.0025, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.3999999999999997e-07, | |
| "loss": 1.917, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9e-07, | |
| "loss": 1.8306, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.4e-07, | |
| "loss": 1.9323, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9e-07, | |
| "loss": 1.5776, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.4e-07, | |
| "loss": 1.4313, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.9e-07, | |
| "loss": 1.3474, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.4e-07, | |
| "loss": 1.29, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.9e-07, | |
| "loss": 1.1974, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.399999999999999e-07, | |
| "loss": 1.299, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.9e-07, | |
| "loss": 1.1418, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.399999999999999e-07, | |
| "loss": 1.205, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.9e-07, | |
| "loss": 0.9866, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.955555555555554e-07, | |
| "loss": 0.9506, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.9e-07, | |
| "loss": 1.2328, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.844444444444445e-07, | |
| "loss": 1.1429, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.788888888888889e-07, | |
| "loss": 1.005, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.733333333333333e-07, | |
| "loss": 0.9108, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.677777777777777e-07, | |
| "loss": 0.9505, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.62222222222222e-07, | |
| "loss": 1.1187, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.566666666666667e-07, | |
| "loss": 0.9913, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 1.132, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.455555555555556e-07, | |
| "loss": 1.1477, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.399999999999999e-07, | |
| "loss": 0.9871, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.344444444444444e-07, | |
| "loss": 1.024, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.288888888888888e-07, | |
| "loss": 0.8419, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.233333333333333e-07, | |
| "loss": 0.896, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.177777777777777e-07, | |
| "loss": 1.0737, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.122222222222222e-07, | |
| "loss": 1.057, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.066666666666665e-07, | |
| "loss": 0.9556, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.01111111111111e-07, | |
| "loss": 1.0581, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.955555555555555e-07, | |
| "loss": 0.8865, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.9e-07, | |
| "loss": 1.1374, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 0.9853608012199402, | |
| "eval_runtime": 596.2346, | |
| "eval_samples_per_second": 4.111, | |
| "eval_steps_per_second": 2.056, | |
| "eval_wer": 64.96336164969846, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.844444444444445e-07, | |
| "loss": 0.9606, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.788888888888889e-07, | |
| "loss": 0.9526, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.733333333333333e-07, | |
| "loss": 1.0831, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.677777777777777e-07, | |
| "loss": 0.8729, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.622222222222222e-07, | |
| "loss": 0.977, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.566666666666667e-07, | |
| "loss": 0.9041, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.511111111111111e-07, | |
| "loss": 1.0707, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.455555555555555e-07, | |
| "loss": 1.093, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.399999999999999e-07, | |
| "loss": 1.102, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.344444444444444e-07, | |
| "loss": 0.8571, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.288888888888888e-07, | |
| "loss": 1.0215, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.233333333333333e-07, | |
| "loss": 0.9254, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.177777777777778e-07, | |
| "loss": 1.0406, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.122222222222221e-07, | |
| "loss": 0.9516, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.066666666666666e-07, | |
| "loss": 0.9585, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.01111111111111e-07, | |
| "loss": 0.7759, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.955555555555556e-07, | |
| "loss": 0.919, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.9e-07, | |
| "loss": 0.9281, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 1.0186, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.788888888888889e-07, | |
| "loss": 0.8863, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.733333333333333e-07, | |
| "loss": 0.96, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.677777777777778e-07, | |
| "loss": 0.7853, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.622222222222222e-07, | |
| "loss": 0.9674, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.566666666666667e-07, | |
| "loss": 0.8892, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.51111111111111e-07, | |
| "loss": 0.8938, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.455555555555555e-07, | |
| "loss": 0.8152, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.4e-07, | |
| "loss": 0.8955, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.344444444444444e-07, | |
| "loss": 0.8278, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.288888888888889e-07, | |
| "loss": 0.9115, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.233333333333333e-07, | |
| "loss": 0.9066, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.177777777777777e-07, | |
| "loss": 1.0192, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.122222222222221e-07, | |
| "loss": 0.9415, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.066666666666666e-07, | |
| "loss": 1.0481, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.01111111111111e-07, | |
| "loss": 0.8349, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.955555555555556e-07, | |
| "loss": 0.8967, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.9e-07, | |
| "loss": 0.8016, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.844444444444444e-07, | |
| "loss": 0.9607, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.788888888888889e-07, | |
| "loss": 0.8532, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.733333333333333e-07, | |
| "loss": 0.9257, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.677777777777778e-07, | |
| "loss": 0.8775, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 0.9139208793640137, | |
| "eval_runtime": 610.1359, | |
| "eval_samples_per_second": 4.017, | |
| "eval_steps_per_second": 2.009, | |
| "eval_wer": 66.46131898061086, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.622222222222222e-07, | |
| "loss": 0.8847, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.566666666666666e-07, | |
| "loss": 1.0419, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.511111111111111e-07, | |
| "loss": 0.8739, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.455555555555555e-07, | |
| "loss": 0.8261, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.4e-07, | |
| "loss": 0.9817, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.344444444444444e-07, | |
| "loss": 1.0232, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.288888888888889e-07, | |
| "loss": 0.8482, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.233333333333332e-07, | |
| "loss": 0.8584, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.177777777777777e-07, | |
| "loss": 0.8966, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.122222222222222e-07, | |
| "loss": 0.8923, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.066666666666666e-07, | |
| "loss": 0.9748, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.011111111111112e-07, | |
| "loss": 1.1212, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.955555555555555e-07, | |
| "loss": 0.9538, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9e-07, | |
| "loss": 1.1401, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.844444444444444e-07, | |
| "loss": 0.9019, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.788888888888889e-07, | |
| "loss": 0.8402, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.733333333333334e-07, | |
| "loss": 0.8792, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.677777777777778e-07, | |
| "loss": 0.9592, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.622222222222222e-07, | |
| "loss": 0.9085, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.566666666666666e-07, | |
| "loss": 1.0189, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.511111111111111e-07, | |
| "loss": 0.8341, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.455555555555555e-07, | |
| "loss": 0.7564, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.4e-07, | |
| "loss": 0.7948, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.344444444444445e-07, | |
| "loss": 1.0225, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.288888888888888e-07, | |
| "loss": 0.9563, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.233333333333333e-07, | |
| "loss": 0.7843, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.177777777777777e-07, | |
| "loss": 1.0565, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.122222222222222e-07, | |
| "loss": 0.7467, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.9035, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.011111111111111e-07, | |
| "loss": 0.8723, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.955555555555556e-07, | |
| "loss": 0.8477, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9e-07, | |
| "loss": 0.8342, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.844444444444445e-07, | |
| "loss": 0.7835, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.788888888888889e-07, | |
| "loss": 1.0936, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.733333333333333e-07, | |
| "loss": 0.9851, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.677777777777778e-07, | |
| "loss": 0.7425, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.622222222222222e-07, | |
| "loss": 0.9534, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5666666666666665e-07, | |
| "loss": 0.9309, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.7555, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.455555555555555e-07, | |
| "loss": 0.9735, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 0.8845030665397644, | |
| "eval_runtime": 580.1372, | |
| "eval_samples_per_second": 4.225, | |
| "eval_steps_per_second": 2.113, | |
| "eval_wer": 58.66675312885027, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.3999999999999997e-07, | |
| "loss": 0.9725, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.344444444444444e-07, | |
| "loss": 1.0319, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.2888888888888883e-07, | |
| "loss": 0.8079, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.2333333333333334e-07, | |
| "loss": 0.8567, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.177777777777778e-07, | |
| "loss": 0.9613, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.122222222222222e-07, | |
| "loss": 0.9328, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.0666666666666666e-07, | |
| "loss": 0.8802, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.0111111111111106e-07, | |
| "loss": 0.8767, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.955555555555555e-07, | |
| "loss": 0.7442, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.8999999999999997e-07, | |
| "loss": 0.8729, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.8444444444444443e-07, | |
| "loss": 0.9029, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.788888888888889e-07, | |
| "loss": 0.8988, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.7333333333333334e-07, | |
| "loss": 0.8658, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.6777777777777774e-07, | |
| "loss": 0.9137, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.622222222222222e-07, | |
| "loss": 0.8178, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.5666666666666666e-07, | |
| "loss": 0.8481, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.5111111111111106e-07, | |
| "loss": 0.8868, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4555555555555557e-07, | |
| "loss": 0.7846, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.8963, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.3444444444444443e-07, | |
| "loss": 0.7407, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.288888888888889e-07, | |
| "loss": 0.9731, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.233333333333333e-07, | |
| "loss": 0.8753, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.1777777777777775e-07, | |
| "loss": 0.885, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.122222222222222e-07, | |
| "loss": 0.7584, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.066666666666666e-07, | |
| "loss": 0.8814, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.011111111111111e-07, | |
| "loss": 0.9843, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.9555555555555557e-07, | |
| "loss": 0.9026, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.9e-07, | |
| "loss": 0.9259, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.8444444444444443e-07, | |
| "loss": 0.8303, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.788888888888889e-07, | |
| "loss": 0.7775, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.733333333333333e-07, | |
| "loss": 0.9112, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.6777777777777775e-07, | |
| "loss": 0.8211, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.6222222222222226e-07, | |
| "loss": 0.9493, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.5666666666666666e-07, | |
| "loss": 0.9079, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.511111111111111e-07, | |
| "loss": 0.8974, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.455555555555555e-07, | |
| "loss": 0.9766, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.4e-07, | |
| "loss": 0.9226, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.3444444444444444e-07, | |
| "loss": 0.7136, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.288888888888889e-07, | |
| "loss": 0.9056, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2333333333333332e-07, | |
| "loss": 0.8359, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 0.8695711493492126, | |
| "eval_runtime": 584.3999, | |
| "eval_samples_per_second": 4.194, | |
| "eval_steps_per_second": 2.098, | |
| "eval_wer": 59.58757538421633, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1777777777777775e-07, | |
| "loss": 0.9031, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.122222222222222e-07, | |
| "loss": 0.8199, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.0666666666666666e-07, | |
| "loss": 0.8729, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.011111111111111e-07, | |
| "loss": 1.0134, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.9555555555555555e-07, | |
| "loss": 0.8379, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8999999999999998e-07, | |
| "loss": 0.954, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8444444444444444e-07, | |
| "loss": 0.7528, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7888888888888887e-07, | |
| "loss": 0.8616, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7333333333333332e-07, | |
| "loss": 0.8091, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6777777777777778e-07, | |
| "loss": 0.8108, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.622222222222222e-07, | |
| "loss": 0.799, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5666666666666667e-07, | |
| "loss": 1.0137, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.511111111111111e-07, | |
| "loss": 0.8332, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4555555555555555e-07, | |
| "loss": 0.9052, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4e-07, | |
| "loss": 0.8797, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3466666666666665e-07, | |
| "loss": 0.8401, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.291111111111111e-07, | |
| "loss": 0.9753, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.2355555555555554e-07, | |
| "loss": 0.7506, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.1799999999999998e-07, | |
| "loss": 0.7586, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.1244444444444444e-07, | |
| "loss": 0.7905, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0688888888888888e-07, | |
| "loss": 0.8575, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0133333333333333e-07, | |
| "loss": 0.6719, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.577777777777777e-08, | |
| "loss": 0.9679, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.022222222222221e-08, | |
| "loss": 0.884, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.466666666666667e-08, | |
| "loss": 0.8952, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.91111111111111e-08, | |
| "loss": 0.9371, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.355555555555556e-08, | |
| "loss": 0.9077, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.8e-08, | |
| "loss": 1.0169, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.244444444444444e-08, | |
| "loss": 0.9629, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.6888888888888887e-08, | |
| "loss": 0.8887, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.133333333333333e-08, | |
| "loss": 0.8295, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.577777777777777e-08, | |
| "loss": 0.848, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.022222222222222e-08, | |
| "loss": 0.7996, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.4666666666666666e-08, | |
| "loss": 0.9296, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.911111111111111e-08, | |
| "loss": 0.928, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.3555555555555556e-08, | |
| "loss": 0.7704, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8e-08, | |
| "loss": 0.917, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2444444444444444e-08, | |
| "loss": 0.8176, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.888888888888889e-09, | |
| "loss": 0.8245, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3333333333333333e-09, | |
| "loss": 0.9089, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 0.8638723492622375, | |
| "eval_runtime": 557.1668, | |
| "eval_samples_per_second": 4.399, | |
| "eval_steps_per_second": 2.2, | |
| "eval_wer": 54.70462356526814, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "step": 5000, | |
| "total_flos": 5.7717080064e+18, | |
| "train_loss": 0.996151273727417, | |
| "train_runtime": 5789.5442, | |
| "train_samples_per_second": 3.455, | |
| "train_steps_per_second": 0.864 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 1, | |
| "total_flos": 5.7717080064e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |