| { | |
| "best_metric": 68.84339815762537, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 33.0232, | |
| "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": 4.0000000000000003e-07, | |
| "loss": 1.6358, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 1.4032, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 1.0334, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.7593, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.6845, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.6317, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.5755, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.5429, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.515, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.4814, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.4958, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.4818, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.4399, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.4275, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.4058, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.3816, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.4072, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.3761, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.3563, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.3294, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.955555555555556e-06, | |
| "loss": 0.3277, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.2944, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 0.3243, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 0.2984, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.2677, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 0.2548, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 0.24, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 0.2186, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 0.2487, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 0.2271, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.2028, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 0.1863, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 0.1826, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 0.1673, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 0.1884, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 0.1535, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.1524, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 0.1371, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 0.133, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.1221, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "eval_loss": 0.4879000782966614, | |
| "eval_runtime": 557.2741, | |
| "eval_samples_per_second": 8.45, | |
| "eval_steps_per_second": 0.266, | |
| "eval_wer": 68.84339815762537, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 0.1443, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 0.1199, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 0.1099, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 0.099, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 0.0963, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 0.0903, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 0.1016, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 0.0832, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.076, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 0.0685, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.0683, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.0659, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.076, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.0539, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.0545, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.0476, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.0481, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.047, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.0522, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.0394, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.0367, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.0308, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.0328, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.0327, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.0368, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.0258, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0248, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.0214, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.0222, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.0227, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.025, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.0178, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.0168, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.0149, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.0146, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.016, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.0165, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.0125, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.0122, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.0119, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "eval_loss": 0.5566783547401428, | |
| "eval_runtime": 543.8133, | |
| "eval_samples_per_second": 8.659, | |
| "eval_steps_per_second": 0.272, | |
| "eval_wer": 70.23541453428864, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.011, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.0123, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.0116, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.0096, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0093, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.0088, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.0085, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.0094, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.0087, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.0077, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.0076, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 0.0071, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 0.0069, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.0076, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 0.0071, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.0065, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.0064, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 0.006, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 0.0058, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.0066, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 0.0061, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.0056, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.0055, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 0.0053, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 0.0051, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.0057, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 0.0053, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.005, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.0049, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 0.0047, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 0.0046, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.005, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.0047, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 0.0044, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.0044, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.0041, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.0042, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.0045, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.0041, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.004, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "eval_loss": 0.5889936089515686, | |
| "eval_runtime": 542.9965, | |
| "eval_samples_per_second": 8.672, | |
| "eval_steps_per_second": 0.273, | |
| "eval_wer": 70.62436028659161, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.004, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 0.0037, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.0039, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 0.0041, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 0.0037, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 0.0037, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 0.0036, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 0.0034, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.0035, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.0038, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 0.0034, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 0.0034, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 0.0033, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 0.0032, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.0033, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 22.03, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.0034, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 0.0032, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.0031, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.0031, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.0029, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 0.0031, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 0.0032, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 0.0029, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.0029, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 0.0029, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 0.0027, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 24.02, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 0.0029, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.0029, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.0028, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 25.01, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.0027, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 0.0027, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 0.0025, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 25.02, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 0.0028, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 0.0028, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 0.0026, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 26.01, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 0.0026, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0026, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.0024, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 26.03, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 0.0026, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 0.0027, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_loss": 0.6128411889076233, | |
| "eval_runtime": 680.8053, | |
| "eval_samples_per_second": 6.917, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 72.40532241555783, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 0.0024, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 27.01, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.0025, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.0024, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 27.02, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.0023, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 27.03, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.0025, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.0024, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.0023, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 28.01, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.0024, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.0023, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 28.02, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.0022, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 28.03, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.0024, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.0024, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.0023, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 29.01, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.0023, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0022, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.0022, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 29.03, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.0024, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.0022, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.0022, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 30.01, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.0022, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.0021, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 30.02, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.0021, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 30.03, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.0023, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.0022, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.0022, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 31.01, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.0021, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.0021, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 31.02, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.0021, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 31.03, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.0022, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.0021, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 32.01, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.0021, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 32.01, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.0021, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 32.02, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.002, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 32.02, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.0021, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 32.03, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.0022, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.002, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 33.01, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.0021, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 33.01, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.0021, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 33.02, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 33.02, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.0021, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 33.02, | |
| "eval_loss": 0.6176665425300598, | |
| "eval_runtime": 557.1499, | |
| "eval_samples_per_second": 8.452, | |
| "eval_steps_per_second": 0.266, | |
| "eval_wer": 71.91402251791197, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 33.02, | |
| "step": 5000, | |
| "total_flos": 7.83417238511616e+18, | |
| "train_loss": 0.09576444422230125, | |
| "train_runtime": 25987.0924, | |
| "train_samples_per_second": 12.314, | |
| "train_steps_per_second": 0.192 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 7.83417238511616e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |