| { | |
| "best_metric": 77.96530707923112, | |
| "best_model_checkpoint": "model_output/whisper-small-cn/checkpoint-2000", | |
| "epoch": 7.561436672967864, | |
| "eval_steps": 1000, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 4.7143, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 3.4488, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.42e-06, | |
| "loss": 2.2293, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.5856, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.1524, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.92e-06, | |
| "loss": 1.1127, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.9826, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.8764, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.8712, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.7879, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.761, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.7329, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.6688, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.5491, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.583, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.5337, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.5773, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.4585, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.536, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.4673, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.5347, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.86857142857143e-06, | |
| "loss": 0.4144, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.797142857142858e-06, | |
| "loss": 0.4115, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.725714285714287e-06, | |
| "loss": 0.4049, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 9.654285714285716e-06, | |
| "loss": 0.3807, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.582857142857143e-06, | |
| "loss": 0.3634, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.511428571428572e-06, | |
| "loss": 0.3757, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.3804, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.368571428571428e-06, | |
| "loss": 0.394, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.297142857142857e-06, | |
| "loss": 0.3877, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.225714285714286e-06, | |
| "loss": 0.4812, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.154285714285715e-06, | |
| "loss": 0.3287, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.082857142857143e-06, | |
| "loss": 0.3446, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.011428571428572e-06, | |
| "loss": 0.3504, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.3594, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.86857142857143e-06, | |
| "loss": 0.3475, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.797142857142857e-06, | |
| "loss": 0.4678, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.725714285714286e-06, | |
| "loss": 0.3885, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 8.654285714285715e-06, | |
| "loss": 0.383, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 8.582857142857144e-06, | |
| "loss": 0.4142, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "eval_loss": 0.47647178173065186, | |
| "eval_runtime": 1887.3717, | |
| "eval_samples_per_second": 1.122, | |
| "eval_steps_per_second": 0.14, | |
| "eval_wer": 78.15283638068449, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.511428571428571e-06, | |
| "loss": 0.314, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.3837, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.36857142857143e-06, | |
| "loss": 0.3422, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.297142857142859e-06, | |
| "loss": 0.2169, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.225714285714288e-06, | |
| "loss": 0.2195, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 8.154285714285715e-06, | |
| "loss": 0.2107, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.082857142857144e-06, | |
| "loss": 0.2584, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.011428571428573e-06, | |
| "loss": 0.2343, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.1806, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.86857142857143e-06, | |
| "loss": 0.2412, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.797142857142858e-06, | |
| "loss": 0.2238, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.725714285714286e-06, | |
| "loss": 0.1992, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.654285714285715e-06, | |
| "loss": 0.2242, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.5828571428571444e-06, | |
| "loss": 0.2042, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 7.511428571428572e-06, | |
| "loss": 0.2437, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.2523, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.36857142857143e-06, | |
| "loss": 0.2403, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 7.297142857142858e-06, | |
| "loss": 0.2081, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.225714285714286e-06, | |
| "loss": 0.1916, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.154285714285715e-06, | |
| "loss": 0.2182, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.082857142857143e-06, | |
| "loss": 0.2109, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.011428571428572e-06, | |
| "loss": 0.2337, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.1998, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.868571428571429e-06, | |
| "loss": 0.1653, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 6.797142857142858e-06, | |
| "loss": 0.1294, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.725714285714287e-06, | |
| "loss": 0.1017, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 6.654285714285716e-06, | |
| "loss": 0.1391, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 6.582857142857143e-06, | |
| "loss": 0.1095, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 6.511428571428572e-06, | |
| "loss": 0.1499, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.14, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 6.368571428571429e-06, | |
| "loss": 0.1308, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 6.297142857142857e-06, | |
| "loss": 0.1137, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.225714285714286e-06, | |
| "loss": 0.1415, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 6.1542857142857145e-06, | |
| "loss": 0.0994, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 6.0828571428571435e-06, | |
| "loss": 0.1537, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 6.011428571428572e-06, | |
| "loss": 0.1867, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.1214, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 5.868571428571429e-06, | |
| "loss": 0.1255, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 5.797142857142858e-06, | |
| "loss": 0.1131, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 5.725714285714287e-06, | |
| "loss": 0.1914, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "eval_loss": 0.504170835018158, | |
| "eval_runtime": 2281.9102, | |
| "eval_samples_per_second": 0.928, | |
| "eval_steps_per_second": 0.116, | |
| "eval_wer": 77.96530707923112, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 5.654285714285714e-06, | |
| "loss": 0.0949, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.582857142857143e-06, | |
| "loss": 0.0872, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 5.511428571428572e-06, | |
| "loss": 0.1385, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.1132, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.368571428571429e-06, | |
| "loss": 0.1307, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 5.297142857142858e-06, | |
| "loss": 0.0654, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 5.225714285714286e-06, | |
| "loss": 0.0713, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 5.154285714285715e-06, | |
| "loss": 0.0728, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 5.082857142857144e-06, | |
| "loss": 0.0746, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 5.011428571428571e-06, | |
| "loss": 0.0985, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.1006, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 4.868571428571429e-06, | |
| "loss": 0.0536, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 4.797142857142857e-06, | |
| "loss": 0.0673, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 4.725714285714286e-06, | |
| "loss": 0.0816, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 4.6542857142857145e-06, | |
| "loss": 0.0753, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 4.5828571428571435e-06, | |
| "loss": 0.0757, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 4.511428571428572e-06, | |
| "loss": 0.0523, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.0722, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 4.368571428571429e-06, | |
| "loss": 0.0798, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 4.297142857142858e-06, | |
| "loss": 0.0848, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 4.225714285714286e-06, | |
| "loss": 0.0705, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 4.154285714285714e-06, | |
| "loss": 0.0829, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 4.082857142857143e-06, | |
| "loss": 0.0604, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 4.011428571428571e-06, | |
| "loss": 0.076, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.0872, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.8685714285714286e-06, | |
| "loss": 0.0627, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.7971428571428576e-06, | |
| "loss": 0.047, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 3.7257142857142857e-06, | |
| "loss": 0.0333, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.6542857142857148e-06, | |
| "loss": 0.0416, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 3.582857142857143e-06, | |
| "loss": 0.0428, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 3.511428571428572e-06, | |
| "loss": 0.0415, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.0334, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 3.3685714285714287e-06, | |
| "loss": 0.0478, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.2971428571428577e-06, | |
| "loss": 0.0565, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.225714285714286e-06, | |
| "loss": 0.0491, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 3.154285714285715e-06, | |
| "loss": 0.0461, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 3.082857142857143e-06, | |
| "loss": 0.0323, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 3.0114285714285716e-06, | |
| "loss": 0.0442, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.0339, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 2.868571428571429e-06, | |
| "loss": 0.0578, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "eval_loss": 0.5367042422294617, | |
| "eval_runtime": 2308.2603, | |
| "eval_samples_per_second": 0.917, | |
| "eval_steps_per_second": 0.115, | |
| "eval_wer": 81.24706985466479, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 2.797142857142857e-06, | |
| "loss": 0.0467, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 2.725714285714286e-06, | |
| "loss": 0.0328, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 2.654285714285714e-06, | |
| "loss": 0.0404, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.582857142857143e-06, | |
| "loss": 0.056, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.5114285714285718e-06, | |
| "loss": 0.0448, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.0401, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.3685714285714285e-06, | |
| "loss": 0.0439, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 2.297142857142857e-06, | |
| "loss": 0.03, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.2257142857142857e-06, | |
| "loss": 0.0171, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 2.1542857142857147e-06, | |
| "loss": 0.0223, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 2.0828571428571433e-06, | |
| "loss": 0.0413, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 2.0114285714285715e-06, | |
| "loss": 0.022, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.0284, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 1.8685714285714289e-06, | |
| "loss": 0.0174, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 1.7971428571428572e-06, | |
| "loss": 0.0298, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 1.7257142857142858e-06, | |
| "loss": 0.0235, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 1.6542857142857144e-06, | |
| "loss": 0.0218, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 1.582857142857143e-06, | |
| "loss": 0.0187, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 1.5114285714285714e-06, | |
| "loss": 0.0286, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.0194, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 1.3685714285714286e-06, | |
| "loss": 0.0266, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 1.2971428571428574e-06, | |
| "loss": 0.0304, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 1.2257142857142857e-06, | |
| "loss": 0.031, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.1542857142857143e-06, | |
| "loss": 0.0246, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 1.082857142857143e-06, | |
| "loss": 0.0167, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 1.0114285714285715e-06, | |
| "loss": 0.0359, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.0247, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 8.685714285714286e-07, | |
| "loss": 0.0229, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 7.971428571428572e-07, | |
| "loss": 0.0217, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 7.257142857142857e-07, | |
| "loss": 0.0166, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 6.542857142857144e-07, | |
| "loss": 0.0117, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 5.82857142857143e-07, | |
| "loss": 0.0265, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 5.114285714285714e-07, | |
| "loss": 0.0187, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 4.4e-07, | |
| "loss": 0.0134, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 3.685714285714286e-07, | |
| "loss": 0.0181, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 2.9714285714285715e-07, | |
| "loss": 0.0165, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 2.2571428571428574e-07, | |
| "loss": 0.0165, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 7.47, | |
| "learning_rate": 1.542857142857143e-07, | |
| "loss": 0.0155, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 8.285714285714285e-08, | |
| "loss": 0.0218, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 1.142857142857143e-08, | |
| "loss": 0.0213, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "eval_loss": 0.5499102473258972, | |
| "eval_runtime": 2355.5425, | |
| "eval_samples_per_second": 0.899, | |
| "eval_steps_per_second": 0.113, | |
| "eval_wer": 80.49695264885138, | |
| "step": 4000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 4000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 8, | |
| "save_steps": 1000, | |
| "total_flos": 1.846946562048e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |