{ "best_metric": 11.648690292758088, "best_model_checkpoint": "whisper-small-pt-cv11-v7/checkpoint-1500", "epoch": 5.524861878453039, "global_step": 3000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "learning_rate": 1.0500000000000001e-07, "loss": 1.3312, "step": 25 }, { "epoch": 0.09, "learning_rate": 2.3000000000000002e-07, "loss": 1.2895, "step": 50 }, { "epoch": 0.14, "learning_rate": 3.55e-07, "loss": 1.2314, "step": 75 }, { "epoch": 0.18, "learning_rate": 4.800000000000001e-07, "loss": 1.1129, "step": 100 }, { "epoch": 0.23, "learning_rate": 6.05e-07, "loss": 0.9069, "step": 125 }, { "epoch": 0.28, "learning_rate": 7.3e-07, "loss": 0.6229, "step": 150 }, { "epoch": 0.32, "learning_rate": 8.550000000000002e-07, "loss": 0.4441, "step": 175 }, { "epoch": 0.37, "learning_rate": 9.800000000000001e-07, "loss": 0.394, "step": 200 }, { "epoch": 0.41, "learning_rate": 1.105e-06, "loss": 0.3244, "step": 225 }, { "epoch": 0.46, "learning_rate": 1.23e-06, "loss": 0.2862, "step": 250 }, { "epoch": 0.51, "learning_rate": 1.3550000000000002e-06, "loss": 0.3068, "step": 275 }, { "epoch": 0.55, "learning_rate": 1.48e-06, "loss": 0.3021, "step": 300 }, { "epoch": 0.6, "learning_rate": 1.605e-06, "loss": 0.2899, "step": 325 }, { "epoch": 0.64, "learning_rate": 1.73e-06, "loss": 0.2605, "step": 350 }, { "epoch": 0.69, "learning_rate": 1.8550000000000002e-06, "loss": 0.2597, "step": 375 }, { "epoch": 0.74, "learning_rate": 1.98e-06, "loss": 0.2537, "step": 400 }, { "epoch": 0.78, "learning_rate": 2.105e-06, "loss": 0.2524, "step": 425 }, { "epoch": 0.83, "learning_rate": 2.2300000000000002e-06, "loss": 0.2765, "step": 450 }, { "epoch": 0.87, "learning_rate": 2.355e-06, "loss": 0.2555, "step": 475 }, { "epoch": 0.92, "learning_rate": 2.4800000000000004e-06, "loss": 0.2476, "step": 500 }, { "epoch": 0.92, "eval_cer": 4.976512848853274, "eval_loss": 0.2900051772594452, "eval_runtime": 195.0851, "eval_samples_per_second": 5.126, "eval_steps_per_second": 0.323, "eval_wer": 13.204930662557782, "step": 500 }, { "epoch": 0.97, "learning_rate": 2.6050000000000005e-06, "loss": 0.2508, "step": 525 }, { "epoch": 1.01, "learning_rate": 2.7300000000000005e-06, "loss": 0.2186, "step": 550 }, { "epoch": 1.06, "learning_rate": 2.855e-06, "loss": 0.2019, "step": 575 }, { "epoch": 1.1, "learning_rate": 2.9800000000000003e-06, "loss": 0.2164, "step": 600 }, { "epoch": 1.15, "learning_rate": 3.1050000000000003e-06, "loss": 0.1846, "step": 625 }, { "epoch": 1.2, "learning_rate": 3.2300000000000004e-06, "loss": 0.1943, "step": 650 }, { "epoch": 1.24, "learning_rate": 3.3550000000000005e-06, "loss": 0.2007, "step": 675 }, { "epoch": 1.29, "learning_rate": 3.48e-06, "loss": 0.2049, "step": 700 }, { "epoch": 1.34, "learning_rate": 3.6050000000000002e-06, "loss": 0.1862, "step": 725 }, { "epoch": 1.38, "learning_rate": 3.7300000000000003e-06, "loss": 0.1946, "step": 750 }, { "epoch": 1.43, "learning_rate": 3.855e-06, "loss": 0.1858, "step": 775 }, { "epoch": 1.47, "learning_rate": 3.980000000000001e-06, "loss": 0.1958, "step": 800 }, { "epoch": 1.52, "learning_rate": 4.1050000000000005e-06, "loss": 0.1903, "step": 825 }, { "epoch": 1.57, "learning_rate": 4.23e-06, "loss": 0.1948, "step": 850 }, { "epoch": 1.61, "learning_rate": 4.355000000000001e-06, "loss": 0.1759, "step": 875 }, { "epoch": 1.66, "learning_rate": 4.48e-06, "loss": 0.1954, "step": 900 }, { "epoch": 1.7, "learning_rate": 4.605000000000001e-06, "loss": 0.191, "step": 925 }, { "epoch": 1.75, "learning_rate": 4.7300000000000005e-06, "loss": 0.1792, "step": 950 }, { "epoch": 1.8, "learning_rate": 4.855e-06, "loss": 0.1822, "step": 975 }, { "epoch": 1.84, "learning_rate": 4.980000000000001e-06, "loss": 0.1886, "step": 1000 }, { "epoch": 1.84, "eval_cer": 4.617297596021, "eval_loss": 0.26112794876098633, "eval_runtime": 194.6573, "eval_samples_per_second": 5.137, "eval_steps_per_second": 0.324, "eval_wer": 12.280431432973806, "step": 1000 }, { "epoch": 1.89, "learning_rate": 4.9883333333333336e-06, "loss": 0.1909, "step": 1025 }, { "epoch": 1.93, "learning_rate": 4.974444444444445e-06, "loss": 0.1772, "step": 1050 }, { "epoch": 1.98, "learning_rate": 4.9605555555555564e-06, "loss": 0.1828, "step": 1075 }, { "epoch": 2.03, "learning_rate": 4.946666666666667e-06, "loss": 0.129, "step": 1100 }, { "epoch": 2.07, "learning_rate": 4.932777777777778e-06, "loss": 0.1184, "step": 1125 }, { "epoch": 2.12, "learning_rate": 4.918888888888889e-06, "loss": 0.1085, "step": 1150 }, { "epoch": 2.16, "learning_rate": 4.9050000000000005e-06, "loss": 0.1149, "step": 1175 }, { "epoch": 2.21, "learning_rate": 4.891111111111111e-06, "loss": 0.1065, "step": 1200 }, { "epoch": 2.26, "learning_rate": 4.8772222222222225e-06, "loss": 0.1157, "step": 1225 }, { "epoch": 2.3, "learning_rate": 4.863333333333334e-06, "loss": 0.1141, "step": 1250 }, { "epoch": 2.35, "learning_rate": 4.849444444444445e-06, "loss": 0.1165, "step": 1275 }, { "epoch": 2.39, "learning_rate": 4.835555555555556e-06, "loss": 0.1212, "step": 1300 }, { "epoch": 2.44, "learning_rate": 4.821666666666667e-06, "loss": 0.1165, "step": 1325 }, { "epoch": 2.49, "learning_rate": 4.807777777777779e-06, "loss": 0.1097, "step": 1350 }, { "epoch": 2.53, "learning_rate": 4.793888888888889e-06, "loss": 0.1081, "step": 1375 }, { "epoch": 2.58, "learning_rate": 4.78e-06, "loss": 0.1055, "step": 1400 }, { "epoch": 2.62, "learning_rate": 4.766111111111111e-06, "loss": 0.1174, "step": 1425 }, { "epoch": 2.67, "learning_rate": 4.752222222222223e-06, "loss": 0.1177, "step": 1450 }, { "epoch": 2.72, "learning_rate": 4.7383333333333334e-06, "loss": 0.1213, "step": 1475 }, { "epoch": 2.76, "learning_rate": 4.724444444444445e-06, "loss": 0.1066, "step": 1500 }, { "epoch": 2.76, "eval_cer": 4.476374689140647, "eval_loss": 0.256792277097702, "eval_runtime": 195.1022, "eval_samples_per_second": 5.126, "eval_steps_per_second": 0.323, "eval_wer": 11.648690292758088, "step": 1500 }, { "epoch": 2.81, "learning_rate": 4.710555555555556e-06, "loss": 0.1207, "step": 1525 }, { "epoch": 2.85, "learning_rate": 4.696666666666667e-06, "loss": 0.1167, "step": 1550 }, { "epoch": 2.9, "learning_rate": 4.682777777777778e-06, "loss": 0.1063, "step": 1575 }, { "epoch": 2.95, "learning_rate": 4.66888888888889e-06, "loss": 0.1134, "step": 1600 }, { "epoch": 2.99, "learning_rate": 4.655e-06, "loss": 0.1126, "step": 1625 }, { "epoch": 3.04, "learning_rate": 4.641111111111111e-06, "loss": 0.076, "step": 1650 }, { "epoch": 3.08, "learning_rate": 4.627222222222222e-06, "loss": 0.0629, "step": 1675 }, { "epoch": 3.13, "learning_rate": 4.613333333333334e-06, "loss": 0.0588, "step": 1700 }, { "epoch": 3.18, "learning_rate": 4.599444444444444e-06, "loss": 0.0719, "step": 1725 }, { "epoch": 3.22, "learning_rate": 4.585555555555556e-06, "loss": 0.0711, "step": 1750 }, { "epoch": 3.27, "learning_rate": 4.571666666666667e-06, "loss": 0.0623, "step": 1775 }, { "epoch": 3.31, "learning_rate": 4.557777777777778e-06, "loss": 0.0643, "step": 1800 }, { "epoch": 3.36, "learning_rate": 4.543888888888889e-06, "loss": 0.0578, "step": 1825 }, { "epoch": 3.41, "learning_rate": 4.530000000000001e-06, "loss": 0.0691, "step": 1850 }, { "epoch": 3.45, "learning_rate": 4.516111111111111e-06, "loss": 0.0651, "step": 1875 }, { "epoch": 3.5, "learning_rate": 4.502222222222223e-06, "loss": 0.0653, "step": 1900 }, { "epoch": 3.55, "learning_rate": 4.488333333333333e-06, "loss": 0.0658, "step": 1925 }, { "epoch": 3.59, "learning_rate": 4.474444444444445e-06, "loss": 0.0711, "step": 1950 }, { "epoch": 3.64, "learning_rate": 4.460555555555556e-06, "loss": 0.0658, "step": 1975 }, { "epoch": 3.68, "learning_rate": 4.446666666666667e-06, "loss": 0.0698, "step": 2000 }, { "epoch": 3.68, "eval_cer": 4.614534401768444, "eval_loss": 0.27006372809410095, "eval_runtime": 193.7972, "eval_samples_per_second": 5.16, "eval_steps_per_second": 0.325, "eval_wer": 11.879815100154083, "step": 2000 }, { "epoch": 3.73, "learning_rate": 4.432777777777778e-06, "loss": 0.066, "step": 2025 }, { "epoch": 3.78, "learning_rate": 4.41888888888889e-06, "loss": 0.0669, "step": 2050 }, { "epoch": 3.82, "learning_rate": 4.405e-06, "loss": 0.0678, "step": 2075 }, { "epoch": 3.87, "learning_rate": 4.391111111111112e-06, "loss": 0.0666, "step": 2100 }, { "epoch": 3.91, "learning_rate": 4.377222222222223e-06, "loss": 0.0635, "step": 2125 }, { "epoch": 3.96, "learning_rate": 4.363333333333334e-06, "loss": 0.0649, "step": 2150 }, { "epoch": 4.01, "learning_rate": 4.349444444444445e-06, "loss": 0.0687, "step": 2175 }, { "epoch": 4.05, "learning_rate": 4.3355555555555565e-06, "loss": 0.038, "step": 2200 }, { "epoch": 4.1, "learning_rate": 4.321666666666667e-06, "loss": 0.0348, "step": 2225 }, { "epoch": 4.14, "learning_rate": 4.307777777777778e-06, "loss": 0.0403, "step": 2250 }, { "epoch": 4.19, "learning_rate": 4.293888888888889e-06, "loss": 0.0419, "step": 2275 }, { "epoch": 4.24, "learning_rate": 4.2800000000000005e-06, "loss": 0.0382, "step": 2300 }, { "epoch": 4.28, "learning_rate": 4.266111111111111e-06, "loss": 0.0368, "step": 2325 }, { "epoch": 4.33, "learning_rate": 4.2522222222222225e-06, "loss": 0.0352, "step": 2350 }, { "epoch": 4.37, "learning_rate": 4.238333333333334e-06, "loss": 0.0374, "step": 2375 }, { "epoch": 4.42, "learning_rate": 4.2244444444444446e-06, "loss": 0.0365, "step": 2400 }, { "epoch": 4.47, "learning_rate": 4.210555555555556e-06, "loss": 0.0421, "step": 2425 }, { "epoch": 4.51, "learning_rate": 4.1966666666666674e-06, "loss": 0.037, "step": 2450 }, { "epoch": 4.56, "learning_rate": 4.182777777777778e-06, "loss": 0.0384, "step": 2475 }, { "epoch": 4.6, "learning_rate": 4.168888888888889e-06, "loss": 0.0403, "step": 2500 }, { "epoch": 4.6, "eval_cer": 4.440453163857419, "eval_loss": 0.28314679861068726, "eval_runtime": 193.6946, "eval_samples_per_second": 5.163, "eval_steps_per_second": 0.325, "eval_wer": 11.864406779661017, "step": 2500 }, { "epoch": 4.65, "learning_rate": 4.155e-06, "loss": 0.0369, "step": 2525 }, { "epoch": 4.7, "learning_rate": 4.1411111111111115e-06, "loss": 0.0358, "step": 2550 }, { "epoch": 4.74, "learning_rate": 4.127222222222222e-06, "loss": 0.0395, "step": 2575 }, { "epoch": 4.79, "learning_rate": 4.1133333333333335e-06, "loss": 0.0389, "step": 2600 }, { "epoch": 4.83, "learning_rate": 4.099444444444445e-06, "loss": 0.0391, "step": 2625 }, { "epoch": 4.88, "learning_rate": 4.085555555555556e-06, "loss": 0.0406, "step": 2650 }, { "epoch": 4.93, "learning_rate": 4.071666666666667e-06, "loss": 0.0389, "step": 2675 }, { "epoch": 4.97, "learning_rate": 4.057777777777778e-06, "loss": 0.0372, "step": 2700 }, { "epoch": 5.02, "learning_rate": 4.04388888888889e-06, "loss": 0.0321, "step": 2725 }, { "epoch": 5.06, "learning_rate": 4.03e-06, "loss": 0.0193, "step": 2750 }, { "epoch": 5.11, "learning_rate": 4.016111111111111e-06, "loss": 0.0187, "step": 2775 }, { "epoch": 5.16, "learning_rate": 4.002222222222222e-06, "loss": 0.0208, "step": 2800 }, { "epoch": 5.2, "learning_rate": 3.988333333333334e-06, "loss": 0.0201, "step": 2825 }, { "epoch": 5.25, "learning_rate": 3.974444444444444e-06, "loss": 0.0177, "step": 2850 }, { "epoch": 5.29, "learning_rate": 3.960555555555556e-06, "loss": 0.0191, "step": 2875 }, { "epoch": 5.34, "learning_rate": 3.946666666666667e-06, "loss": 0.0222, "step": 2900 }, { "epoch": 5.39, "learning_rate": 3.932777777777778e-06, "loss": 0.0185, "step": 2925 }, { "epoch": 5.43, "learning_rate": 3.918888888888889e-06, "loss": 0.021, "step": 2950 }, { "epoch": 5.48, "learning_rate": 3.905000000000001e-06, "loss": 0.0203, "step": 2975 }, { "epoch": 5.52, "learning_rate": 3.891111111111111e-06, "loss": 0.019, "step": 3000 }, { "epoch": 5.52, "eval_cer": 4.465321912130422, "eval_loss": 0.31483325362205505, "eval_runtime": 194.2855, "eval_samples_per_second": 5.147, "eval_steps_per_second": 0.324, "eval_wer": 11.756548536209554, "step": 3000 }, { "epoch": 5.52, "step": 3000, "total_flos": 2.770419843072e+19, "train_loss": 0.1680191035370032, "train_runtime": 9745.7814, "train_samples_per_second": 32.835, "train_steps_per_second": 1.026 } ], "max_steps": 10000, "num_train_epochs": 19, "total_flos": 2.770419843072e+19, "trial_name": null, "trial_params": null }