| { | |
| "best_metric": 22.84622791429533, | |
| "best_model_checkpoint": "models/whisper-small-et-ERR2020/checkpoint-8000", | |
| "epoch": 2.232, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.3000000000000002e-07, | |
| "loss": 1.8519, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.7228, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.3e-07, | |
| "loss": 1.4752, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 1.3247, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.23e-06, | |
| "loss": 1.2788, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.48e-06, | |
| "loss": 1.5968, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.72e-06, | |
| "loss": 1.7944, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.97e-06, | |
| "loss": 2.0695, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.2200000000000003e-06, | |
| "loss": 2.0161, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.47e-06, | |
| "loss": 1.8474, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.7200000000000002e-06, | |
| "loss": 1.6321, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.97e-06, | |
| "loss": 1.5135, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.2200000000000005e-06, | |
| "loss": 1.3337, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.4700000000000002e-06, | |
| "loss": 1.2364, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.7200000000000004e-06, | |
| "loss": 1.1471, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.97e-06, | |
| "loss": 1.0805, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.22e-06, | |
| "loss": 0.9798, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.47e-06, | |
| "loss": 0.9293, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.7200000000000005e-06, | |
| "loss": 0.8488, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.970000000000001e-06, | |
| "loss": 0.8375, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.220000000000001e-06, | |
| "loss": 0.7664, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.470000000000001e-06, | |
| "loss": 0.721, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.72e-06, | |
| "loss": 0.6851, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.9700000000000004e-06, | |
| "loss": 0.6628, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.220000000000001e-06, | |
| "loss": 0.643, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.470000000000001e-06, | |
| "loss": 0.5972, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.720000000000001e-06, | |
| "loss": 0.5608, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.97e-06, | |
| "loss": 0.5628, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.22e-06, | |
| "loss": 0.5326, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.4700000000000005e-06, | |
| "loss": 0.51, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.72e-06, | |
| "loss": 0.4723, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.970000000000002e-06, | |
| "loss": 0.4806, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 0.4442, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.47e-06, | |
| "loss": 0.4265, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.720000000000001e-06, | |
| "loss": 0.4037, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.97e-06, | |
| "loss": 0.3984, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.220000000000002e-06, | |
| "loss": 0.3758, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.47e-06, | |
| "loss": 0.3603, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.72e-06, | |
| "loss": 0.3532, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.970000000000001e-06, | |
| "loss": 0.3387, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.5383420586585999, | |
| "eval_runtime": 7376.8423, | |
| "eval_samples_per_second": 1.073, | |
| "eval_steps_per_second": 0.034, | |
| "eval_wer": 33.82156081446051, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.975555555555556e-06, | |
| "loss": 0.3371, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.947777777777779e-06, | |
| "loss": 0.3173, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.3067, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.892222222222223e-06, | |
| "loss": 0.2879, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.864444444444446e-06, | |
| "loss": 0.2784, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.836666666666668e-06, | |
| "loss": 0.2731, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.80888888888889e-06, | |
| "loss": 0.2817, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.781111111111113e-06, | |
| "loss": 0.2626, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.753333333333335e-06, | |
| "loss": 0.2544, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.725555555555557e-06, | |
| "loss": 0.2287, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.697777777777778e-06, | |
| "loss": 0.2388, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.67e-06, | |
| "loss": 0.2166, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.642222222222223e-06, | |
| "loss": 0.2207, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.614444444444445e-06, | |
| "loss": 0.2136, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.586666666666667e-06, | |
| "loss": 0.2155, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.55888888888889e-06, | |
| "loss": 0.2104, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.531111111111112e-06, | |
| "loss": 0.1918, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.503333333333334e-06, | |
| "loss": 0.1945, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.475555555555557e-06, | |
| "loss": 0.1932, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.447777777777778e-06, | |
| "loss": 0.1777, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.1847, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.392222222222222e-06, | |
| "loss": 0.1763, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.364444444444445e-06, | |
| "loss": 0.1842, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.336666666666666e-06, | |
| "loss": 0.1671, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.30888888888889e-06, | |
| "loss": 0.1743, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.281111111111112e-06, | |
| "loss": 0.1705, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.253333333333333e-06, | |
| "loss": 0.1734, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.225555555555556e-06, | |
| "loss": 0.1567, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.197777777777779e-06, | |
| "loss": 0.1564, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.17e-06, | |
| "loss": 0.1542, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.142222222222223e-06, | |
| "loss": 0.1567, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.114444444444444e-06, | |
| "loss": 0.1575, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.086666666666667e-06, | |
| "loss": 0.1501, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.05888888888889e-06, | |
| "loss": 0.1536, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.031111111111111e-06, | |
| "loss": 0.145, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.003333333333334e-06, | |
| "loss": 0.1349, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.975555555555557e-06, | |
| "loss": 0.1414, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.947777777777778e-06, | |
| "loss": 0.1371, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.1403, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.892222222222224e-06, | |
| "loss": 0.1393, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.48974087834358215, | |
| "eval_runtime": 6006.6233, | |
| "eval_samples_per_second": 1.317, | |
| "eval_steps_per_second": 0.041, | |
| "eval_wer": 27.75459719275722, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.864444444444445e-06, | |
| "loss": 0.1399, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.836666666666668e-06, | |
| "loss": 0.1366, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.80888888888889e-06, | |
| "loss": 0.1368, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.781111111111112e-06, | |
| "loss": 0.1276, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.753333333333333e-06, | |
| "loss": 0.1334, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.725555555555556e-06, | |
| "loss": 0.1331, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.697777777777779e-06, | |
| "loss": 0.1299, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.67e-06, | |
| "loss": 0.128, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.642222222222223e-06, | |
| "loss": 0.1313, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.614444444444446e-06, | |
| "loss": 0.1235, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.586666666666667e-06, | |
| "loss": 0.1284, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.55888888888889e-06, | |
| "loss": 0.1209, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.531111111111113e-06, | |
| "loss": 0.1159, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.503333333333334e-06, | |
| "loss": 0.1185, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.475555555555555e-06, | |
| "loss": 0.1169, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.447777777777778e-06, | |
| "loss": 0.1237, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.108, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.392222222222222e-06, | |
| "loss": 0.1186, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.364444444444445e-06, | |
| "loss": 0.1275, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.336666666666668e-06, | |
| "loss": 0.1051, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.30888888888889e-06, | |
| "loss": 0.1169, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.281111111111112e-06, | |
| "loss": 0.1035, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.253333333333334e-06, | |
| "loss": 0.114, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.225555555555557e-06, | |
| "loss": 0.1182, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.197777777777779e-06, | |
| "loss": 0.1183, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.17e-06, | |
| "loss": 0.1051, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.142222222222223e-06, | |
| "loss": 0.1112, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.114444444444445e-06, | |
| "loss": 0.12, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.086666666666667e-06, | |
| "loss": 0.1067, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.05888888888889e-06, | |
| "loss": 0.1042, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.031111111111112e-06, | |
| "loss": 0.106, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.003333333333334e-06, | |
| "loss": 0.1064, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.975555555555556e-06, | |
| "loss": 0.1117, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.94777777777778e-06, | |
| "loss": 0.1005, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.1094, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.892222222222223e-06, | |
| "loss": 0.1048, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.864444444444444e-06, | |
| "loss": 0.1069, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.836666666666667e-06, | |
| "loss": 0.1064, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.808888888888888e-06, | |
| "loss": 0.1078, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.781111111111111e-06, | |
| "loss": 0.0982, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.5477123260498047, | |
| "eval_runtime": 7493.8564, | |
| "eval_samples_per_second": 1.056, | |
| "eval_steps_per_second": 0.033, | |
| "eval_wer": 26.78153451123758, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.753333333333334e-06, | |
| "loss": 0.0977, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.725555555555555e-06, | |
| "loss": 0.1119, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.697777777777778e-06, | |
| "loss": 0.1119, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.670000000000001e-06, | |
| "loss": 0.1072, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.642222222222222e-06, | |
| "loss": 0.1098, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.614444444444445e-06, | |
| "loss": 0.0993, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.586666666666668e-06, | |
| "loss": 0.108, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.55888888888889e-06, | |
| "loss": 0.1077, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.531111111111111e-06, | |
| "loss": 0.1002, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.503333333333333e-06, | |
| "loss": 0.0987, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.475555555555556e-06, | |
| "loss": 0.1072, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.447777777777778e-06, | |
| "loss": 0.1042, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.102, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.392222222222223e-06, | |
| "loss": 0.1056, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.364444444444445e-06, | |
| "loss": 0.099, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.336666666666668e-06, | |
| "loss": 0.1012, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.30888888888889e-06, | |
| "loss": 0.1013, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.281111111111112e-06, | |
| "loss": 0.098, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.253333333333335e-06, | |
| "loss": 0.1059, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.225555555555556e-06, | |
| "loss": 0.0998, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.197777777777778e-06, | |
| "loss": 0.0969, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.17e-06, | |
| "loss": 0.0883, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.142222222222223e-06, | |
| "loss": 0.1033, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.114444444444445e-06, | |
| "loss": 0.0947, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.086666666666667e-06, | |
| "loss": 0.1035, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.05888888888889e-06, | |
| "loss": 0.0988, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.031111111111112e-06, | |
| "loss": 0.0946, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.003333333333335e-06, | |
| "loss": 0.1006, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.975555555555557e-06, | |
| "loss": 0.0951, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.947777777777779e-06, | |
| "loss": 0.0961, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.0995, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.892222222222223e-06, | |
| "loss": 0.096, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.864444444444445e-06, | |
| "loss": 0.0969, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.836666666666667e-06, | |
| "loss": 0.0944, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.80888888888889e-06, | |
| "loss": 0.0984, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.781111111111112e-06, | |
| "loss": 0.1003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.753333333333334e-06, | |
| "loss": 0.0887, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.7255555555555565e-06, | |
| "loss": 0.0969, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.6977777777777785e-06, | |
| "loss": 0.0992, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.6700000000000005e-06, | |
| "loss": 0.0912, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "eval_loss": 0.5195096135139465, | |
| "eval_runtime": 6463.6033, | |
| "eval_samples_per_second": 1.224, | |
| "eval_steps_per_second": 0.038, | |
| "eval_wer": 24.881596355124824, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.642222222222223e-06, | |
| "loss": 0.0986, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.6144444444444445e-06, | |
| "loss": 0.0929, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.5866666666666666e-06, | |
| "loss": 0.0912, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.5588888888888894e-06, | |
| "loss": 0.0895, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.5311111111111114e-06, | |
| "loss": 0.1007, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.5033333333333335e-06, | |
| "loss": 0.083, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.475555555555556e-06, | |
| "loss": 0.0845, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.447777777777778e-06, | |
| "loss": 0.0945, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.0897, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.392222222222223e-06, | |
| "loss": 0.088, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.364444444444445e-06, | |
| "loss": 0.0881, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.336666666666667e-06, | |
| "loss": 0.09, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.30888888888889e-06, | |
| "loss": 0.0901, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.281111111111111e-06, | |
| "loss": 0.0801, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.253333333333333e-06, | |
| "loss": 0.0861, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.225555555555556e-06, | |
| "loss": 0.0825, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.197777777777778e-06, | |
| "loss": 0.0877, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.17e-06, | |
| "loss": 0.0817, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.142222222222223e-06, | |
| "loss": 0.084, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.114444444444445e-06, | |
| "loss": 0.0901, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.086666666666667e-06, | |
| "loss": 0.0849, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.05888888888889e-06, | |
| "loss": 0.0838, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.031111111111112e-06, | |
| "loss": 0.088, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.003333333333334e-06, | |
| "loss": 0.0839, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.975555555555555e-06, | |
| "loss": 0.0875, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.947777777777778e-06, | |
| "loss": 0.0761, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.0755, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.892222222222222e-06, | |
| "loss": 0.0742, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.864444444444445e-06, | |
| "loss": 0.081, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.836666666666667e-06, | |
| "loss": 0.0798, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.80888888888889e-06, | |
| "loss": 0.0793, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.781111111111112e-06, | |
| "loss": 0.0745, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.753333333333334e-06, | |
| "loss": 0.0734, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.725555555555557e-06, | |
| "loss": 0.0746, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.697777777777779e-06, | |
| "loss": 0.0791, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.67e-06, | |
| "loss": 0.0776, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.642222222222222e-06, | |
| "loss": 0.0688, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.614444444444445e-06, | |
| "loss": 0.0761, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.586666666666667e-06, | |
| "loss": 0.0686, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.558888888888889e-06, | |
| "loss": 0.0811, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "eval_loss": 0.5373284816741943, | |
| "eval_runtime": 7326.6021, | |
| "eval_samples_per_second": 1.08, | |
| "eval_steps_per_second": 0.034, | |
| "eval_wer": 25.928245434120605, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.531111111111112e-06, | |
| "loss": 0.0781, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.503333333333334e-06, | |
| "loss": 0.075, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.475555555555557e-06, | |
| "loss": 0.0728, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.447777777777779e-06, | |
| "loss": 0.0702, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.0679, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.3922222222222235e-06, | |
| "loss": 0.0693, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.364444444444445e-06, | |
| "loss": 0.0707, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.336666666666667e-06, | |
| "loss": 0.0704, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.308888888888889e-06, | |
| "loss": 0.0668, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.2811111111111116e-06, | |
| "loss": 0.0724, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.2533333333333336e-06, | |
| "loss": 0.0678, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.225555555555556e-06, | |
| "loss": 0.072, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.1977777777777785e-06, | |
| "loss": 0.0681, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.1700000000000005e-06, | |
| "loss": 0.0685, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.1422222222222225e-06, | |
| "loss": 0.0695, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.114444444444445e-06, | |
| "loss": 0.0719, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.086666666666667e-06, | |
| "loss": 0.0662, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.05888888888889e-06, | |
| "loss": 0.0687, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.031111111111111e-06, | |
| "loss": 0.0711, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.0033333333333334e-06, | |
| "loss": 0.0682, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.9755555555555554e-06, | |
| "loss": 0.0714, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.947777777777778e-06, | |
| "loss": 0.0693, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.064, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.892222222222222e-06, | |
| "loss": 0.0728, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.864444444444445e-06, | |
| "loss": 0.0688, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.836666666666667e-06, | |
| "loss": 0.0685, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.808888888888889e-06, | |
| "loss": 0.0667, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.781111111111111e-06, | |
| "loss": 0.0706, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.753333333333333e-06, | |
| "loss": 0.0664, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.725555555555556e-06, | |
| "loss": 0.0606, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.697777777777778e-06, | |
| "loss": 0.0688, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.670000000000001e-06, | |
| "loss": 0.0636, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.642222222222222e-06, | |
| "loss": 0.0634, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.614444444444445e-06, | |
| "loss": 0.0692, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.586666666666667e-06, | |
| "loss": 0.0693, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.558888888888889e-06, | |
| "loss": 0.0649, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.531111111111112e-06, | |
| "loss": 0.0687, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.503333333333333e-06, | |
| "loss": 0.0675, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.475555555555556e-06, | |
| "loss": 0.0645, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.447777777777778e-06, | |
| "loss": 0.0649, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "eval_loss": 0.5421532392501831, | |
| "eval_runtime": 6372.6624, | |
| "eval_samples_per_second": 1.242, | |
| "eval_steps_per_second": 0.039, | |
| "eval_wer": 23.728481849993347, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.072, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.392222222222223e-06, | |
| "loss": 0.0692, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.364444444444445e-06, | |
| "loss": 0.0661, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.336666666666667e-06, | |
| "loss": 0.0648, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.308888888888889e-06, | |
| "loss": 0.0665, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.281111111111112e-06, | |
| "loss": 0.0644, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.253333333333334e-06, | |
| "loss": 0.0625, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.225555555555556e-06, | |
| "loss": 0.0674, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.197777777777779e-06, | |
| "loss": 0.0645, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.17e-06, | |
| "loss": 0.0638, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.142222222222223e-06, | |
| "loss": 0.0624, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.114444444444445e-06, | |
| "loss": 0.0691, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.086666666666667e-06, | |
| "loss": 0.0647, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.05888888888889e-06, | |
| "loss": 0.0602, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.031111111111111e-06, | |
| "loss": 0.0598, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.003333333333334e-06, | |
| "loss": 0.0587, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.975555555555556e-06, | |
| "loss": 0.0707, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.9477777777777786e-06, | |
| "loss": 0.0618, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.0637, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.892222222222223e-06, | |
| "loss": 0.0591, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.864444444444445e-06, | |
| "loss": 0.064, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.836666666666667e-06, | |
| "loss": 0.0683, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.808888888888889e-06, | |
| "loss": 0.0597, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.7811111111111115e-06, | |
| "loss": 0.0622, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.753333333333334e-06, | |
| "loss": 0.0578, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.7255555555555556e-06, | |
| "loss": 0.0652, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.697777777777778e-06, | |
| "loss": 0.0649, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.6700000000000004e-06, | |
| "loss": 0.0615, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.6422222222222225e-06, | |
| "loss": 0.0591, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.614444444444445e-06, | |
| "loss": 0.0661, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.5866666666666673e-06, | |
| "loss": 0.0615, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.558888888888889e-06, | |
| "loss": 0.0626, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.5311111111111114e-06, | |
| "loss": 0.0644, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.5033333333333334e-06, | |
| "loss": 0.056, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.475555555555556e-06, | |
| "loss": 0.0627, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.4477777777777783e-06, | |
| "loss": 0.0663, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.0685, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.3922222222222223e-06, | |
| "loss": 0.0639, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.3644444444444447e-06, | |
| "loss": 0.0612, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.3366666666666668e-06, | |
| "loss": 0.0618, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_loss": 0.5503928065299988, | |
| "eval_runtime": 6509.6002, | |
| "eval_samples_per_second": 1.215, | |
| "eval_steps_per_second": 0.038, | |
| "eval_wer": 23.51789949976906, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.308888888888889e-06, | |
| "loss": 0.063, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.2811111111111116e-06, | |
| "loss": 0.0634, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.2533333333333332e-06, | |
| "loss": 0.0598, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.2255555555555557e-06, | |
| "loss": 0.064, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.197777777777778e-06, | |
| "loss": 0.065, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.17e-06, | |
| "loss": 0.0636, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.1422222222222226e-06, | |
| "loss": 0.0658, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.114444444444445e-06, | |
| "loss": 0.0647, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.0866666666666666e-06, | |
| "loss": 0.0631, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.058888888888889e-06, | |
| "loss": 0.0639, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.0311111111111115e-06, | |
| "loss": 0.058, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.0033333333333335e-06, | |
| "loss": 0.0642, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.975555555555556e-06, | |
| "loss": 0.0612, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.9477777777777784e-06, | |
| "loss": 0.0611, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.0586, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.8922222222222224e-06, | |
| "loss": 0.0584, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.864444444444445e-06, | |
| "loss": 0.0575, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.836666666666667e-06, | |
| "loss": 0.0637, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.8088888888888893e-06, | |
| "loss": 0.0669, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.781111111111111e-06, | |
| "loss": 0.0579, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7533333333333334e-06, | |
| "loss": 0.0629, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.725555555555556e-06, | |
| "loss": 0.0593, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.6977777777777783e-06, | |
| "loss": 0.0581, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6700000000000003e-06, | |
| "loss": 0.0609, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6422222222222227e-06, | |
| "loss": 0.0626, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6144444444444443e-06, | |
| "loss": 0.0616, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.5866666666666667e-06, | |
| "loss": 0.0584, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.558888888888889e-06, | |
| "loss": 0.0662, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.531111111111111e-06, | |
| "loss": 0.0632, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.5033333333333336e-06, | |
| "loss": 0.0614, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.4755555555555557e-06, | |
| "loss": 0.0538, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.447777777777778e-06, | |
| "loss": 0.06, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.0627, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.3922222222222226e-06, | |
| "loss": 0.0644, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.3644444444444446e-06, | |
| "loss": 0.0612, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.3366666666666666e-06, | |
| "loss": 0.056, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.308888888888889e-06, | |
| "loss": 0.0592, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.281111111111111e-06, | |
| "loss": 0.0577, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.2533333333333335e-06, | |
| "loss": 0.06, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.225555555555556e-06, | |
| "loss": 0.0558, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_loss": 0.5321349501609802, | |
| "eval_runtime": 6266.1492, | |
| "eval_samples_per_second": 1.263, | |
| "eval_steps_per_second": 0.04, | |
| "eval_wer": 22.84622791429533, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.197777777777778e-06, | |
| "loss": 0.0555, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.17e-06, | |
| "loss": 0.0573, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.1422222222222224e-06, | |
| "loss": 0.057, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.1144444444444444e-06, | |
| "loss": 0.0625, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.086666666666667e-06, | |
| "loss": 0.0528, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.0588888888888893e-06, | |
| "loss": 0.0584, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.0311111111111113e-06, | |
| "loss": 0.0558, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.0033333333333334e-06, | |
| "loss": 0.0534, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.975555555555556e-06, | |
| "loss": 0.0556, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.947777777777778e-06, | |
| "loss": 0.061, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.0522, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.8933333333333333e-06, | |
| "loss": 0.0551, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.8655555555555558e-06, | |
| "loss": 0.0583, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.837777777777778e-06, | |
| "loss": 0.0591, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.81e-06, | |
| "loss": 0.0515, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.7822222222222225e-06, | |
| "loss": 0.0571, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.7544444444444445e-06, | |
| "loss": 0.0583, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.7266666666666667e-06, | |
| "loss": 0.0569, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.6988888888888892e-06, | |
| "loss": 0.0529, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.6711111111111112e-06, | |
| "loss": 0.0493, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.6433333333333334e-06, | |
| "loss": 0.0547, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.6155555555555559e-06, | |
| "loss": 0.0482, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5877777777777779e-06, | |
| "loss": 0.051, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.56e-06, | |
| "loss": 0.0543, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5322222222222225e-06, | |
| "loss": 0.0507, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5044444444444446e-06, | |
| "loss": 0.0501, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4766666666666668e-06, | |
| "loss": 0.0473, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4488888888888892e-06, | |
| "loss": 0.0494, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4211111111111112e-06, | |
| "loss": 0.0517, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.3933333333333335e-06, | |
| "loss": 0.0524, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.3655555555555557e-06, | |
| "loss": 0.0445, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.337777777777778e-06, | |
| "loss": 0.0502, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.3100000000000002e-06, | |
| "loss": 0.0512, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.2822222222222222e-06, | |
| "loss": 0.0474, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.2544444444444446e-06, | |
| "loss": 0.0548, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.2266666666666666e-06, | |
| "loss": 0.0543, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.198888888888889e-06, | |
| "loss": 0.0462, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.171111111111111e-06, | |
| "loss": 0.0465, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.1433333333333333e-06, | |
| "loss": 0.0439, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.1155555555555558e-06, | |
| "loss": 0.0452, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_loss": 0.5542896389961243, | |
| "eval_runtime": 6394.219, | |
| "eval_samples_per_second": 1.237, | |
| "eval_steps_per_second": 0.039, | |
| "eval_wer": 23.58130905504106, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.0877777777777778e-06, | |
| "loss": 0.0491, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.06e-06, | |
| "loss": 0.0458, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.0322222222222225e-06, | |
| "loss": 0.0483, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.0044444444444445e-06, | |
| "loss": 0.0475, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.766666666666667e-07, | |
| "loss": 0.0476, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.488888888888889e-07, | |
| "loss": 0.0471, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.211111111111113e-07, | |
| "loss": 0.0446, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 8.933333333333334e-07, | |
| "loss": 0.0504, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 8.655555555555556e-07, | |
| "loss": 0.0458, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.37777777777778e-07, | |
| "loss": 0.0475, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.100000000000001e-07, | |
| "loss": 0.0477, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 7.822222222222223e-07, | |
| "loss": 0.0435, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 7.544444444444444e-07, | |
| "loss": 0.0465, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 7.266666666666668e-07, | |
| "loss": 0.0475, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 6.98888888888889e-07, | |
| "loss": 0.0497, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 6.711111111111111e-07, | |
| "loss": 0.0436, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 6.433333333333335e-07, | |
| "loss": 0.0468, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 6.155555555555556e-07, | |
| "loss": 0.0487, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.877777777777778e-07, | |
| "loss": 0.0446, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.6e-07, | |
| "loss": 0.0488, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.322222222222223e-07, | |
| "loss": 0.0464, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.044444444444445e-07, | |
| "loss": 0.0463, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.766666666666667e-07, | |
| "loss": 0.0413, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.488888888888889e-07, | |
| "loss": 0.0446, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.211111111111112e-07, | |
| "loss": 0.0429, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.9333333333333336e-07, | |
| "loss": 0.0436, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.655555555555556e-07, | |
| "loss": 0.0452, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.3777777777777777e-07, | |
| "loss": 0.0488, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.1000000000000005e-07, | |
| "loss": 0.0485, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.822222222222222e-07, | |
| "loss": 0.0449, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.5444444444444446e-07, | |
| "loss": 0.0457, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.266666666666667e-07, | |
| "loss": 0.0487, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.9888888888888891e-07, | |
| "loss": 0.0436, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.7111111111111114e-07, | |
| "loss": 0.0469, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.4333333333333335e-07, | |
| "loss": 0.0486, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.1555555555555556e-07, | |
| "loss": 0.0486, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.777777777777779e-08, | |
| "loss": 0.0461, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 0.0498, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.2222222222222224e-08, | |
| "loss": 0.0418, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.444444444444445e-09, | |
| "loss": 0.0462, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "eval_loss": 0.5424047112464905, | |
| "eval_runtime": 5941.7602, | |
| "eval_samples_per_second": 1.332, | |
| "eval_steps_per_second": 0.042, | |
| "eval_wer": 22.883021113033404, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "step": 10000, | |
| "total_flos": 1.8469061600919552e+20, | |
| "train_loss": 0.009933160030841827, | |
| "train_runtime": 126590.0181, | |
| "train_samples_per_second": 5.056, | |
| "train_steps_per_second": 0.079 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.8469061600919552e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |