| { | |
| "best_metric": 4.711435314310234, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch12-lr6-small/checkpoint-5000", | |
| "epoch": 11.876484560570072, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.4e-08, | |
| "loss": 1.4193, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.4e-08, | |
| "loss": 1.4255, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.44e-07, | |
| "loss": 1.43, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.94e-07, | |
| "loss": 1.2985, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.4399999999999996e-07, | |
| "loss": 1.1832, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.9399999999999996e-07, | |
| "loss": 1.155, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.4399999999999996e-07, | |
| "loss": 1.1072, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.94e-07, | |
| "loss": 0.9823, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.44e-07, | |
| "loss": 0.8871, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.94e-07, | |
| "loss": 0.7735, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.44e-07, | |
| "loss": 0.6804, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.939999999999999e-07, | |
| "loss": 0.6176, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.44e-07, | |
| "loss": 0.5623, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.939999999999999e-07, | |
| "loss": 0.5292, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.44e-07, | |
| "loss": 0.5682, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.94e-07, | |
| "loss": 0.4774, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.439999999999999e-07, | |
| "loss": 0.491, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.939999999999999e-07, | |
| "loss": 0.4744, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.439999999999999e-07, | |
| "loss": 0.4452, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.94e-07, | |
| "loss": 0.4444, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 9.95111111111111e-07, | |
| "loss": 0.398, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.895555555555554e-07, | |
| "loss": 0.4174, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.84e-07, | |
| "loss": 0.4164, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.784444444444444e-07, | |
| "loss": 0.366, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.728888888888888e-07, | |
| "loss": 0.37, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.673333333333332e-07, | |
| "loss": 0.4138, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.617777777777776e-07, | |
| "loss": 0.3824, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.562222222222223e-07, | |
| "loss": 0.3827, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.506666666666667e-07, | |
| "loss": 0.3889, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.451111111111111e-07, | |
| "loss": 0.3985, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.395555555555556e-07, | |
| "loss": 0.4033, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 9.34e-07, | |
| "loss": 0.4152, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.284444444444444e-07, | |
| "loss": 0.4014, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.228888888888888e-07, | |
| "loss": 0.343, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.173333333333333e-07, | |
| "loss": 0.3204, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.117777777777778e-07, | |
| "loss": 0.3402, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 9.062222222222222e-07, | |
| "loss": 0.316, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 9.006666666666666e-07, | |
| "loss": 0.3197, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 8.95111111111111e-07, | |
| "loss": 0.3238, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 8.895555555555555e-07, | |
| "loss": 0.3296, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "eval_loss": 0.30187273025512695, | |
| "eval_runtime": 825.6555, | |
| "eval_samples_per_second": 6.108, | |
| "eval_steps_per_second": 1.019, | |
| "eval_wer": 13.14739326324823, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.839999999999999e-07, | |
| "loss": 0.337, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.784444444444444e-07, | |
| "loss": 0.2967, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 8.728888888888889e-07, | |
| "loss": 0.3238, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 8.673333333333332e-07, | |
| "loss": 0.3102, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 8.617777777777777e-07, | |
| "loss": 0.3003, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 8.562222222222222e-07, | |
| "loss": 0.3089, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 8.506666666666667e-07, | |
| "loss": 0.3021, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 8.451111111111111e-07, | |
| "loss": 0.3172, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 8.395555555555556e-07, | |
| "loss": 0.3249, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 8.34e-07, | |
| "loss": 0.2983, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 8.284444444444444e-07, | |
| "loss": 0.3062, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 8.228888888888889e-07, | |
| "loss": 0.2794, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 8.173333333333333e-07, | |
| "loss": 0.2703, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 8.117777777777778e-07, | |
| "loss": 0.2569, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 8.062222222222221e-07, | |
| "loss": 0.2508, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 8.006666666666666e-07, | |
| "loss": 0.2639, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.95111111111111e-07, | |
| "loss": 0.2583, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 7.895555555555555e-07, | |
| "loss": 0.2641, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 7.84e-07, | |
| "loss": 0.2433, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 7.784444444444444e-07, | |
| "loss": 0.2846, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 7.728888888888888e-07, | |
| "loss": 0.2616, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 7.673333333333332e-07, | |
| "loss": 0.2548, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 7.617777777777778e-07, | |
| "loss": 0.2702, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 7.562222222222222e-07, | |
| "loss": 0.2762, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 7.506666666666667e-07, | |
| "loss": 0.2907, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 7.451111111111111e-07, | |
| "loss": 0.2648, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 7.395555555555555e-07, | |
| "loss": 0.2408, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 7.34e-07, | |
| "loss": 0.2274, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 7.284444444444444e-07, | |
| "loss": 0.2322, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 7.228888888888889e-07, | |
| "loss": 0.2313, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 7.173333333333333e-07, | |
| "loss": 0.2296, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 7.117777777777777e-07, | |
| "loss": 0.2444, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 7.062222222222222e-07, | |
| "loss": 0.2252, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 7.006666666666666e-07, | |
| "loss": 0.2257, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 6.951111111111111e-07, | |
| "loss": 0.2302, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 6.895555555555555e-07, | |
| "loss": 0.2272, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 6.84e-07, | |
| "loss": 0.2377, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 6.784444444444443e-07, | |
| "loss": 0.2111, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 6.728888888888888e-07, | |
| "loss": 0.2443, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 6.673333333333334e-07, | |
| "loss": 0.2247, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_loss": 0.20599021017551422, | |
| "eval_runtime": 1037.1517, | |
| "eval_samples_per_second": 4.862, | |
| "eval_steps_per_second": 0.811, | |
| "eval_wer": 8.704140742329972, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 6.617777777777778e-07, | |
| "loss": 0.2064, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 6.562222222222223e-07, | |
| "loss": 0.2275, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 6.506666666666666e-07, | |
| "loss": 0.2277, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 6.451111111111111e-07, | |
| "loss": 0.2226, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 6.395555555555555e-07, | |
| "loss": 0.2078, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 6.34e-07, | |
| "loss": 0.2107, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 6.284444444444445e-07, | |
| "loss": 0.1824, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 6.228888888888889e-07, | |
| "loss": 0.1923, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 6.173333333333333e-07, | |
| "loss": 0.2209, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 6.117777777777777e-07, | |
| "loss": 0.185, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 6.062222222222222e-07, | |
| "loss": 0.1951, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 6.006666666666666e-07, | |
| "loss": 0.1872, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 5.951111111111111e-07, | |
| "loss": 0.204, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 5.895555555555555e-07, | |
| "loss": 0.2137, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 5.839999999999999e-07, | |
| "loss": 0.2045, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 5.784444444444444e-07, | |
| "loss": 0.2049, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 5.728888888888888e-07, | |
| "loss": 0.2005, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 5.673333333333334e-07, | |
| "loss": 0.1907, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 5.617777777777778e-07, | |
| "loss": 0.193, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 5.562222222222222e-07, | |
| "loss": 0.2018, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 5.506666666666666e-07, | |
| "loss": 0.1988, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 5.451111111111111e-07, | |
| "loss": 0.1784, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 5.395555555555556e-07, | |
| "loss": 0.1668, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 5.34e-07, | |
| "loss": 0.189, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 5.284444444444445e-07, | |
| "loss": 0.1727, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 5.228888888888888e-07, | |
| "loss": 0.1645, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 5.173333333333333e-07, | |
| "loss": 0.1644, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 5.117777777777777e-07, | |
| "loss": 0.1707, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 5.062222222222222e-07, | |
| "loss": 0.1783, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 5.006666666666667e-07, | |
| "loss": 0.1775, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 4.951111111111111e-07, | |
| "loss": 0.1714, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 4.895555555555555e-07, | |
| "loss": 0.173, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 4.839999999999999e-07, | |
| "loss": 0.1759, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 4.784444444444444e-07, | |
| "loss": 0.174, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 4.728888888888889e-07, | |
| "loss": 0.1837, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 4.673333333333333e-07, | |
| "loss": 0.1818, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 4.6177777777777777e-07, | |
| "loss": 0.2009, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 4.5622222222222217e-07, | |
| "loss": 0.1756, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 4.506666666666666e-07, | |
| "loss": 0.1515, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 4.451111111111111e-07, | |
| "loss": 0.1526, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "eval_loss": 0.15617816150188446, | |
| "eval_runtime": 1028.0416, | |
| "eval_samples_per_second": 4.905, | |
| "eval_steps_per_second": 0.818, | |
| "eval_wer": 6.2196953443467065, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 4.3955555555555554e-07, | |
| "loss": 0.1682, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 4.34e-07, | |
| "loss": 0.1561, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 4.2844444444444445e-07, | |
| "loss": 0.1673, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 4.2288888888888886e-07, | |
| "loss": 0.173, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 4.173333333333333e-07, | |
| "loss": 0.1634, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 4.1177777777777777e-07, | |
| "loss": 0.154, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 4.0622222222222217e-07, | |
| "loss": 0.15, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 4.0066666666666663e-07, | |
| "loss": 0.1621, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 3.9511111111111114e-07, | |
| "loss": 0.1502, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 3.8955555555555554e-07, | |
| "loss": 0.1747, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 3.84e-07, | |
| "loss": 0.1662, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 3.7844444444444445e-07, | |
| "loss": 0.1667, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 3.7288888888888886e-07, | |
| "loss": 0.1527, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 3.673333333333333e-07, | |
| "loss": 0.1332, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 3.617777777777777e-07, | |
| "loss": 0.1455, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 3.5622222222222223e-07, | |
| "loss": 0.1437, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 3.506666666666667e-07, | |
| "loss": 0.1326, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 3.451111111111111e-07, | |
| "loss": 0.1668, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 3.3955555555555554e-07, | |
| "loss": 0.1561, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 3.34e-07, | |
| "loss": 0.1428, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 3.284444444444444e-07, | |
| "loss": 0.1512, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 3.2288888888888886e-07, | |
| "loss": 0.1399, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 3.173333333333333e-07, | |
| "loss": 0.1402, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 3.1177777777777777e-07, | |
| "loss": 0.143, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 3.0622222222222223e-07, | |
| "loss": 0.1489, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 3.006666666666667e-07, | |
| "loss": 0.1512, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 2.951111111111111e-07, | |
| "loss": 0.1306, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 2.8955555555555555e-07, | |
| "loss": 0.1402, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 2.8399999999999995e-07, | |
| "loss": 0.1474, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 2.784444444444444e-07, | |
| "loss": 0.1392, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 2.728888888888889e-07, | |
| "loss": 0.1451, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 2.673333333333333e-07, | |
| "loss": 0.1399, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 2.617777777777778e-07, | |
| "loss": 0.1352, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 2.5622222222222223e-07, | |
| "loss": 0.1348, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 2.5066666666666663e-07, | |
| "loss": 0.1272, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 2.451111111111111e-07, | |
| "loss": 0.1309, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 2.3955555555555555e-07, | |
| "loss": 0.1509, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 2.34e-07, | |
| "loss": 0.1321, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.2844444444444443e-07, | |
| "loss": 0.1295, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 2.2288888888888886e-07, | |
| "loss": 0.1476, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "eval_loss": 0.13044601678848267, | |
| "eval_runtime": 1031.7698, | |
| "eval_samples_per_second": 4.888, | |
| "eval_steps_per_second": 0.815, | |
| "eval_wer": 5.035400128727741, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.1733333333333332e-07, | |
| "loss": 0.1407, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 2.1177777777777778e-07, | |
| "loss": 0.1288, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 2.062222222222222e-07, | |
| "loss": 0.1405, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 2.0066666666666666e-07, | |
| "loss": 0.1344, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 1.9511111111111112e-07, | |
| "loss": 0.1394, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 1.8955555555555555e-07, | |
| "loss": 0.126, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 1.8399999999999998e-07, | |
| "loss": 0.127, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 1.7844444444444444e-07, | |
| "loss": 0.1389, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 1.728888888888889e-07, | |
| "loss": 0.1353, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 10.1, | |
| "learning_rate": 1.6733333333333332e-07, | |
| "loss": 0.1323, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 1.6177777777777775e-07, | |
| "loss": 0.1217, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 1.5622222222222224e-07, | |
| "loss": 0.1517, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 10.27, | |
| "learning_rate": 1.5066666666666667e-07, | |
| "loss": 0.1206, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "learning_rate": 1.451111111111111e-07, | |
| "loss": 0.1311, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 1.3955555555555553e-07, | |
| "loss": 0.1289, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "learning_rate": 1.34e-07, | |
| "loss": 0.1241, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "learning_rate": 1.2844444444444444e-07, | |
| "loss": 0.1287, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 1.228888888888889e-07, | |
| "loss": 0.1218, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "learning_rate": 1.1733333333333333e-07, | |
| "loss": 0.1257, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 1.1177777777777778e-07, | |
| "loss": 0.1369, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 1.0622222222222221e-07, | |
| "loss": 0.1204, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 1.0066666666666667e-07, | |
| "loss": 0.1181, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 9.51111111111111e-08, | |
| "loss": 0.1225, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 10.93, | |
| "learning_rate": 8.955555555555555e-08, | |
| "loss": 0.1326, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "learning_rate": 8.4e-08, | |
| "loss": 0.1232, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 7.844444444444444e-08, | |
| "loss": 0.1226, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 11.1, | |
| "learning_rate": 7.288888888888888e-08, | |
| "loss": 0.1188, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 11.16, | |
| "learning_rate": 6.733333333333333e-08, | |
| "loss": 0.1275, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 11.22, | |
| "learning_rate": 6.177777777777777e-08, | |
| "loss": 0.1319, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 5.622222222222222e-08, | |
| "loss": 0.1224, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 5.0666666666666664e-08, | |
| "loss": 0.1191, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 4.511111111111111e-08, | |
| "loss": 0.1189, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "learning_rate": 3.955555555555555e-08, | |
| "loss": 0.1194, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "learning_rate": 3.4e-08, | |
| "loss": 0.1333, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 2.8444444444444443e-08, | |
| "loss": 0.1292, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "learning_rate": 2.2888888888888887e-08, | |
| "loss": 0.1169, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 1.7333333333333333e-08, | |
| "loss": 0.1154, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 1.1777777777777778e-08, | |
| "loss": 0.124, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 11.82, | |
| "learning_rate": 6.222222222222222e-09, | |
| "loss": 0.1362, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 6.666666666666666e-10, | |
| "loss": 0.1173, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "eval_loss": 0.12232152372598648, | |
| "eval_runtime": 1046.7582, | |
| "eval_samples_per_second": 4.818, | |
| "eval_steps_per_second": 0.803, | |
| "eval_wer": 4.711435314310234, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "step": 5000, | |
| "total_flos": 1.728655406456832e+19, | |
| "train_loss": 0.26830010027885437, | |
| "train_runtime": 12204.8548, | |
| "train_samples_per_second": 4.916, | |
| "train_steps_per_second": 0.41 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 12, | |
| "total_flos": 1.728655406456832e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |