| { | |
| "best_metric": 32.55118472509777, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch4-tiny/checkpoint-5000", | |
| "epoch": 2.420135527589545, | |
| "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": 4.2000000000000006e-07, | |
| "loss": 1.9606, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 1.7727, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.42e-06, | |
| "loss": 1.7317, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.5526, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.375, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.92e-06, | |
| "loss": 1.2417, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 1.1807, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 1.1283, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.42e-06, | |
| "loss": 1.0607, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.92e-06, | |
| "loss": 1.0918, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.9749, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.9035, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.9489, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.9285, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 1.0048, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.9067, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.8485, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.8338, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.8439, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.814, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.8382, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.8951, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.8564, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.8215, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.827, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.7769, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.7514, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.8001, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.8244, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.7981, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.7814, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.7988, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.7532, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.7538, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.7372, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.7757, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.7212, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.6626, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.762, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.7055, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_loss": 0.6328660845756531, | |
| "eval_runtime": 2032.5756, | |
| "eval_samples_per_second": 1.782, | |
| "eval_steps_per_second": 0.891, | |
| "eval_wer": 42.10720036806993, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.7525, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.7208, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.6369, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.699, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.7602, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.6352, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.7883, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.7229, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.6715, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.7042, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.286666666666668e-06, | |
| "loss": 0.6413, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.231111111111112e-06, | |
| "loss": 0.6097, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.175555555555556e-06, | |
| "loss": 0.693, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.120000000000002e-06, | |
| "loss": 0.7185, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.064444444444444e-06, | |
| "loss": 0.7088, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.00888888888889e-06, | |
| "loss": 0.6692, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.953333333333334e-06, | |
| "loss": 0.7743, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.897777777777778e-06, | |
| "loss": 0.755, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.842222222222224e-06, | |
| "loss": 0.5714, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.786666666666666e-06, | |
| "loss": 0.6912, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.731111111111112e-06, | |
| "loss": 0.6729, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.675555555555556e-06, | |
| "loss": 0.7024, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.620000000000001e-06, | |
| "loss": 0.6035, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.564444444444446e-06, | |
| "loss": 0.6226, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.50888888888889e-06, | |
| "loss": 0.6312, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.453333333333334e-06, | |
| "loss": 0.6303, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.3977777777777786e-06, | |
| "loss": 0.6512, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.342222222222223e-06, | |
| "loss": 0.6884, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.583, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.231111111111112e-06, | |
| "loss": 0.6249, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.1755555555555556e-06, | |
| "loss": 0.6507, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.1200000000000004e-06, | |
| "loss": 0.6226, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.0644444444444445e-06, | |
| "loss": 0.6722, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.008888888888889e-06, | |
| "loss": 0.6407, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.953333333333334e-06, | |
| "loss": 0.6078, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.897777777777779e-06, | |
| "loss": 0.6461, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.842222222222222e-06, | |
| "loss": 0.5932, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.786666666666667e-06, | |
| "loss": 0.7005, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.731111111111111e-06, | |
| "loss": 0.6109, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.675555555555556e-06, | |
| "loss": 0.5685, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 0.5514832139015198, | |
| "eval_runtime": 1912.9967, | |
| "eval_samples_per_second": 1.893, | |
| "eval_steps_per_second": 0.947, | |
| "eval_wer": 35.86381412468369, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.620000000000001e-06, | |
| "loss": 0.7237, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 0.6418, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 0.614, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.3931, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.4267, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.4434, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 0.4, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 0.4754, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 0.3614, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 0.4202, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 0.4019, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.4194, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.4121, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.3952, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.4661, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.4139, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.4611, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.3758, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.4486, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.4285, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.3983, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.4224, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.4263, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.3711, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.3749, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.3959, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.4137, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.4599, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.4779, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.4144, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.4375, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.4275, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.4606, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.3835, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.4885, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.4206, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.4837, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.4156, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.4653, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.3807, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_loss": 0.5231785178184509, | |
| "eval_runtime": 1911.8055, | |
| "eval_samples_per_second": 1.895, | |
| "eval_steps_per_second": 0.947, | |
| "eval_wer": 34.018863584080975, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.4096, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.4196, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.396, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.4276, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.4102, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.3876, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.4102, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.3557, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.3546, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.4502, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.3704, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.4407, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.4302, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.3638, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.3679, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.41, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.4003, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.4139, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.3804, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.4575, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.4337, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.4123, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.4012, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.4396, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.4405, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.4514, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.409, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.3581, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.4054, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.4343, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.3944, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.3944, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.4115, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.4076, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.4259, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.3621, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.4118, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.3843, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.3846, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.3766, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 0.49930620193481445, | |
| "eval_runtime": 1866.5046, | |
| "eval_samples_per_second": 1.941, | |
| "eval_steps_per_second": 0.97, | |
| "eval_wer": 32.67080745341615, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.3843, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.43, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.4177, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.4174, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.3761, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.3039, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.3078, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.3087, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.294, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.2603, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.277, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.2479, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.2899, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.3152, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.2907, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.316, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.2618, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.336, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.2937, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.2706, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.27, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.3287, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.2978, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.279, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.2883, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.2912, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.3138, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.2951, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.2676, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.3002, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.2944, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.2854, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.2903, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.2774, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.2592, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.2947, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.2789, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.2984, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.2825, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.3567, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_loss": 0.4997063875198364, | |
| "eval_runtime": 1867.4461, | |
| "eval_samples_per_second": 1.94, | |
| "eval_steps_per_second": 0.97, | |
| "eval_wer": 32.55118472509777, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "step": 5000, | |
| "total_flos": 4.9222998908928e+17, | |
| "train_loss": 0.557375843334198, | |
| "train_runtime": 15732.1346, | |
| "train_samples_per_second": 1.271, | |
| "train_steps_per_second": 0.318 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.9222998908928e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |