| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.1952, | |
| "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": 1.7586, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.4846, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.1469, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.7319, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.5317, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.4885, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.4334, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.4209, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.3878, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.4018, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.4037, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.4106, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.4308, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.3829, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.3652, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.3763, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.3654, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.3733, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.3664, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.3919, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.4172, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.3857, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.3675, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.3521, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.3639, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.3409, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.3521, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.3279, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.3124, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.3198, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.3142, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.3647, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.355, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.3795, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3667, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.367, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.338, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.3548, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.3412, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.3656, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.3898088037967682, | |
| "eval_runtime": 6251.8367, | |
| "eval_samples_per_second": 2.573, | |
| "eval_steps_per_second": 0.161, | |
| "eval_wer": 17.649710811211627, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.3364, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.3291, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.3053, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.3307, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.3177, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.3162, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.2874, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.2828, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.2852, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.3226, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.2664, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.2809, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.2735, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.2788, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.2689, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.2495, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.2417, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.2261, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.2767, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.2638, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.2404, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.2956, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.2617, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.2336, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.2276, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.2415, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.228, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.2622, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.249, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.2735, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.2569, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.256, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.2859, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.2648, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.267, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.253, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.2631, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.2227, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.2763, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.2745, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_loss": 0.3617454171180725, | |
| "eval_runtime": 4328.2378, | |
| "eval_samples_per_second": 3.717, | |
| "eval_steps_per_second": 0.232, | |
| "eval_wer": 15.774581047011715, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.2507, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.357, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.3746, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.3452, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.2697, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.2273, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.1872, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.1862, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.186, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.1947, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.1911, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.2035, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.2066, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.197, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.2201, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.2109, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.2019, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.1912, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.2127, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.1835, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.1751, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.1747, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.1782, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.1462, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.1501, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.1659, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.1776, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.1863, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.1644, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.191, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.2022, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.201, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.1926, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.1954, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.1994, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.1812, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.1835, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.1709, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.1609, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.1659, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_loss": 0.3525247275829315, | |
| "eval_runtime": 4364.8046, | |
| "eval_samples_per_second": 3.686, | |
| "eval_steps_per_second": 0.23, | |
| "eval_wer": 15.30831973898858, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.1907, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.1899, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.2117, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.2008, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.2143, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.1747, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.1787, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.1657, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.1703, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.2, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.1718, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.2225, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.2147, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.191, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.1873, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.1736, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.1414, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.1714, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.1542, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.1589, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.1727, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.1922, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.1807, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.1818, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.1754, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.1744, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.1672, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.1753, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.1753, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.1654, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.1634, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.1658, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.2002, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.1807, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.1619, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.1754, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.1766, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.1816, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.1528, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.1847, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "eval_loss": 0.3426906168460846, | |
| "eval_runtime": 4209.633, | |
| "eval_samples_per_second": 3.822, | |
| "eval_steps_per_second": 0.239, | |
| "eval_wer": 14.786890108260417, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.1781, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.1762, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.1858, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.1921, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.1904, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.1715, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.1713, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.1857, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.1892, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.1752, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.1701, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.2339, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.1955, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.2246, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.225, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.2224, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.2056, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.2059, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.2196, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.1952, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.0622222222222222e-06, | |
| "loss": 0.1747, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 1.0066666666666668e-06, | |
| "loss": 0.1765, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.1363, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.1058, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.1297, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 0.1274, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 7.28888888888889e-07, | |
| "loss": 0.1115, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.733333333333334e-07, | |
| "loss": 0.1092, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.177777777777778e-07, | |
| "loss": 0.1117, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.622222222222223e-07, | |
| "loss": 0.1311, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.1696, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.1669, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.9555555555555557e-07, | |
| "loss": 0.1568, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.1888, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 2.844444444444445e-07, | |
| "loss": 0.1556, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 2.2888888888888892e-07, | |
| "loss": 0.1872, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 1.7333333333333335e-07, | |
| "loss": 0.1823, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.1777777777777778e-07, | |
| "loss": 0.1647, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 6.222222222222223e-08, | |
| "loss": 0.1548, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 6.666666666666667e-09, | |
| "loss": 0.1593, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "eval_loss": 0.3397696614265442, | |
| "eval_runtime": 4150.7623, | |
| "eval_samples_per_second": 3.876, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 14.908497701319886, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "step": 5000, | |
| "total_flos": 4.615750326878208e+19, | |
| "train_loss": 0.25935648283958435, | |
| "train_runtime": 72928.9053, | |
| "train_samples_per_second": 2.194, | |
| "train_steps_per_second": 0.069 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 4.615750326878208e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |