| { | |
| "best_metric": 40.85702163767501, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 8.0352, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4e-07, | |
| "loss": 0.4645, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 0.2443, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.1375, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.1666, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.3541, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.4968, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.8646, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.5315, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.38e-06, | |
| "loss": 0.3975, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 0.3329, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.380000000000001e-06, | |
| "loss": 0.4836, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 0.4685, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.380000000000001e-06, | |
| "loss": 0.5611, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.88e-06, | |
| "loss": 0.4805, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 0.7273, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.88e-06, | |
| "loss": 0.6222, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.380000000000001e-06, | |
| "loss": 0.8831, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.880000000000001e-06, | |
| "loss": 0.8603, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.38e-06, | |
| "loss": 0.698, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.88e-06, | |
| "loss": 0.7289, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.6182, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.904444444444445e-06, | |
| "loss": 0.676, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.84888888888889e-06, | |
| "loss": 0.6483, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.793333333333333e-06, | |
| "loss": 0.5832, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.737777777777779e-06, | |
| "loss": 0.7001, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.682222222222223e-06, | |
| "loss": 0.5131, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.626666666666667e-06, | |
| "loss": 0.2972, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.571111111111113e-06, | |
| "loss": 0.2305, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.515555555555557e-06, | |
| "loss": 0.359, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.5019, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.404444444444445e-06, | |
| "loss": 0.3049, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.348888888888889e-06, | |
| "loss": 0.3046, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.293333333333335e-06, | |
| "loss": 0.3144, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.237777777777779e-06, | |
| "loss": 0.314, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.182222222222223e-06, | |
| "loss": 0.323, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.126666666666667e-06, | |
| "loss": 0.3468, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.07111111111111e-06, | |
| "loss": 0.3956, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.015555555555557e-06, | |
| "loss": 0.3349, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.4008, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.904444444444446e-06, | |
| "loss": 0.5167, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "eval_loss": 0.7032740116119385, | |
| "eval_runtime": 1467.9388, | |
| "eval_samples_per_second": 0.362, | |
| "eval_steps_per_second": 0.091, | |
| "eval_wer": 67.24649978786593, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.848888888888889e-06, | |
| "loss": 0.3753, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.795555555555555e-06, | |
| "loss": 0.2899, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.740000000000001e-06, | |
| "loss": 0.2059, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.684444444444445e-06, | |
| "loss": 0.1371, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.62888888888889e-06, | |
| "loss": 0.1252, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.573333333333333e-06, | |
| "loss": 0.1091, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.517777777777779e-06, | |
| "loss": 0.112, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.462222222222223e-06, | |
| "loss": 0.0843, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.406666666666667e-06, | |
| "loss": 0.3613, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.351111111111113e-06, | |
| "loss": 0.3208, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.295555555555557e-06, | |
| "loss": 0.2015, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.24e-06, | |
| "loss": 0.1637, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.184444444444445e-06, | |
| "loss": 0.2277, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.128888888888889e-06, | |
| "loss": 0.1954, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.073333333333335e-06, | |
| "loss": 0.3214, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.017777777777779e-06, | |
| "loss": 0.2646, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.962222222222223e-06, | |
| "loss": 0.2254, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.906666666666667e-06, | |
| "loss": 0.1218, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.851111111111112e-06, | |
| "loss": 0.1675, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.795555555555556e-06, | |
| "loss": 0.183, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.74e-06, | |
| "loss": 0.1912, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.684444444444446e-06, | |
| "loss": 0.1781, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.628888888888889e-06, | |
| "loss": 0.2308, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.573333333333333e-06, | |
| "loss": 0.2426, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.517777777777778e-06, | |
| "loss": 0.2141, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.462222222222223e-06, | |
| "loss": 0.1197, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.406666666666667e-06, | |
| "loss": 0.0915, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.351111111111112e-06, | |
| "loss": 0.0917, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.295555555555556e-06, | |
| "loss": 0.0822, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.24e-06, | |
| "loss": 0.0586, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.184444444444445e-06, | |
| "loss": 0.0526, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 7.12888888888889e-06, | |
| "loss": 0.0731, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.073333333333334e-06, | |
| "loss": 0.0431, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.017777777777778e-06, | |
| "loss": 0.0174, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.962222222222223e-06, | |
| "loss": 0.0255, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.906666666666667e-06, | |
| "loss": 0.0118, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.851111111111112e-06, | |
| "loss": 0.0441, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.795555555555557e-06, | |
| "loss": 0.086, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.740000000000001e-06, | |
| "loss": 0.0855, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.684444444444445e-06, | |
| "loss": 0.0886, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "eval_loss": 0.773045539855957, | |
| "eval_runtime": 1251.3368, | |
| "eval_samples_per_second": 0.425, | |
| "eval_steps_per_second": 0.106, | |
| "eval_wer": 51.18795078489605, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.628888888888889e-06, | |
| "loss": 0.123, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.573333333333334e-06, | |
| "loss": 0.14, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.5177777777777785e-06, | |
| "loss": 0.1362, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.462222222222223e-06, | |
| "loss": 0.1278, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.4066666666666674e-06, | |
| "loss": 0.1878, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.3511111111111115e-06, | |
| "loss": 0.1521, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.2955555555555555e-06, | |
| "loss": 0.1131, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.24e-06, | |
| "loss": 0.0933, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.184444444444445e-06, | |
| "loss": 0.1119, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.12888888888889e-06, | |
| "loss": 0.1697, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.073333333333333e-06, | |
| "loss": 0.1741, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.017777777777778e-06, | |
| "loss": 0.1292, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.962222222222222e-06, | |
| "loss": 0.0865, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.906666666666667e-06, | |
| "loss": 0.1044, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.851111111111112e-06, | |
| "loss": 0.107, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.795555555555557e-06, | |
| "loss": 0.0951, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.74e-06, | |
| "loss": 0.0824, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.684444444444445e-06, | |
| "loss": 0.1014, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.628888888888889e-06, | |
| "loss": 0.1119, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.573333333333334e-06, | |
| "loss": 0.0778, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.517777777777779e-06, | |
| "loss": 0.0549, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.462222222222223e-06, | |
| "loss": 0.0522, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.406666666666667e-06, | |
| "loss": 0.0778, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.351111111111112e-06, | |
| "loss": 0.0847, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.295555555555556e-06, | |
| "loss": 0.0583, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5.240000000000001e-06, | |
| "loss": 0.0371, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 5.1844444444444455e-06, | |
| "loss": 0.0333, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 5.128888888888889e-06, | |
| "loss": 0.0331, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 5.073333333333334e-06, | |
| "loss": 0.0185, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 5.0177777777777785e-06, | |
| "loss": 0.0175, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.9622222222222225e-06, | |
| "loss": 0.0143, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.9066666666666666e-06, | |
| "loss": 0.0082, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.8511111111111114e-06, | |
| "loss": 0.0266, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.795555555555556e-06, | |
| "loss": 0.0531, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.74e-06, | |
| "loss": 0.0502, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.684444444444444e-06, | |
| "loss": 0.051, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.628888888888889e-06, | |
| "loss": 0.0594, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.573333333333333e-06, | |
| "loss": 0.0964, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.517777777777778e-06, | |
| "loss": 0.0876, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 4.462222222222223e-06, | |
| "loss": 0.0808, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "eval_loss": 0.7812442779541016, | |
| "eval_runtime": 1080.011, | |
| "eval_samples_per_second": 0.493, | |
| "eval_steps_per_second": 0.123, | |
| "eval_wer": 52.58803563852354, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 4.406666666666667e-06, | |
| "loss": 0.0675, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 4.351111111111111e-06, | |
| "loss": 0.0416, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 4.295555555555556e-06, | |
| "loss": 0.0266, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.0215, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 4.184444444444445e-06, | |
| "loss": 0.0222, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.12888888888889e-06, | |
| "loss": 0.0392, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.073333333333334e-06, | |
| "loss": 0.0306, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 4.017777777777778e-06, | |
| "loss": 0.0447, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.962222222222222e-06, | |
| "loss": 0.0361, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.906666666666667e-06, | |
| "loss": 0.0304, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.851111111111112e-06, | |
| "loss": 0.0239, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.7955555555555557e-06, | |
| "loss": 0.018, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.74e-06, | |
| "loss": 0.0228, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.684444444444445e-06, | |
| "loss": 0.0183, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.628888888888889e-06, | |
| "loss": 0.008, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.5733333333333336e-06, | |
| "loss": 0.007, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.5177777777777784e-06, | |
| "loss": 0.0157, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.4622222222222225e-06, | |
| "loss": 0.0229, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.406666666666667e-06, | |
| "loss": 0.0286, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.351111111111112e-06, | |
| "loss": 0.0284, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.295555555555556e-06, | |
| "loss": 0.0285, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.2400000000000003e-06, | |
| "loss": 0.0386, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.1844444444444444e-06, | |
| "loss": 0.0379, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.1288888888888892e-06, | |
| "loss": 0.0303, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3.0733333333333337e-06, | |
| "loss": 0.0275, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 3.0177777777777777e-06, | |
| "loss": 0.0185, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.9622222222222226e-06, | |
| "loss": 0.0166, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.906666666666667e-06, | |
| "loss": 0.0079, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.851111111111111e-06, | |
| "loss": 0.0122, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.795555555555556e-06, | |
| "loss": 0.019, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.7400000000000004e-06, | |
| "loss": 0.0175, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.6844444444444445e-06, | |
| "loss": 0.0196, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.6288888888888894e-06, | |
| "loss": 0.0171, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.573333333333334e-06, | |
| "loss": 0.0143, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.517777777777778e-06, | |
| "loss": 0.0239, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.4622222222222223e-06, | |
| "loss": 0.0158, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.4066666666666668e-06, | |
| "loss": 0.0183, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.3511111111111112e-06, | |
| "loss": 0.013, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.2955555555555557e-06, | |
| "loss": 0.0149, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 2.24e-06, | |
| "loss": 0.0232, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "eval_loss": 0.8797702789306641, | |
| "eval_runtime": 952.1019, | |
| "eval_samples_per_second": 0.559, | |
| "eval_steps_per_second": 0.14, | |
| "eval_wer": 40.85702163767501, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 2.1844444444444446e-06, | |
| "loss": 0.0119, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 2.128888888888889e-06, | |
| "loss": 0.0093, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 2.0733333333333335e-06, | |
| "loss": 0.0111, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.017777777777778e-06, | |
| "loss": 0.0047, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.9622222222222224e-06, | |
| "loss": 0.0037, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.906666666666667e-06, | |
| "loss": 0.0024, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.8511111111111114e-06, | |
| "loss": 0.0025, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.7955555555555556e-06, | |
| "loss": 0.0012, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.74e-06, | |
| "loss": 0.0035, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.6844444444444447e-06, | |
| "loss": 0.0006, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.628888888888889e-06, | |
| "loss": 0.0005, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.5733333333333334e-06, | |
| "loss": 0.0004, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.5177777777777781e-06, | |
| "loss": 0.0005, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.4622222222222224e-06, | |
| "loss": 0.0055, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.4066666666666668e-06, | |
| "loss": 0.0045, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.351111111111111e-06, | |
| "loss": 0.0077, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.2955555555555557e-06, | |
| "loss": 0.0059, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.2400000000000002e-06, | |
| "loss": 0.0078, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.1844444444444447e-06, | |
| "loss": 0.0085, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.128888888888889e-06, | |
| "loss": 0.0128, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.0733333333333334e-06, | |
| "loss": 0.0106, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 1.0177777777777778e-06, | |
| "loss": 0.01, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 9.622222222222223e-07, | |
| "loss": 0.0082, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 9.066666666666668e-07, | |
| "loss": 0.0124, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 8.511111111111112e-07, | |
| "loss": 0.0127, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 7.955555555555557e-07, | |
| "loss": 0.0086, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 7.4e-07, | |
| "loss": 0.0051, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 6.844444444444446e-07, | |
| "loss": 0.009, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 6.288888888888889e-07, | |
| "loss": 0.0057, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 5.733333333333334e-07, | |
| "loss": 0.0023, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 5.177777777777778e-07, | |
| "loss": 0.0041, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 4.6222222222222225e-07, | |
| "loss": 0.0029, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 4.0666666666666666e-07, | |
| "loss": 0.0031, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 3.5111111111111117e-07, | |
| "loss": 0.0044, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.9555555555555557e-07, | |
| "loss": 0.0048, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 0.0049, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 1.8444444444444446e-07, | |
| "loss": 0.0053, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 1.288888888888889e-07, | |
| "loss": 0.002, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 7.333333333333334e-08, | |
| "loss": 0.0018, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 1.777777777777778e-08, | |
| "loss": 0.001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "eval_loss": 0.9316898584365845, | |
| "eval_runtime": 888.7134, | |
| "eval_samples_per_second": 0.599, | |
| "eval_steps_per_second": 0.15, | |
| "eval_wer": 41.02672889266017, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "step": 5000, | |
| "total_flos": 8.4926840832e+19, | |
| "train_loss": 0.15010352901350707, | |
| "train_runtime": 20688.2827, | |
| "train_samples_per_second": 1.933, | |
| "train_steps_per_second": 0.242 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 8.4926840832e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |