| { | |
| "best_metric": 6.472859901308732, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch8-lr6-small/checkpoint-5000", | |
| "epoch": 7.9239302694136295, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4e-08, | |
| "loss": 1.4328, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.4e-08, | |
| "loss": 1.4148, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.44e-07, | |
| "loss": 1.4051, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.94e-07, | |
| "loss": 1.4043, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.4399999999999996e-07, | |
| "loss": 1.2898, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.9399999999999996e-07, | |
| "loss": 1.1444, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.4399999999999996e-07, | |
| "loss": 1.0854, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.94e-07, | |
| "loss": 0.9595, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.44e-07, | |
| "loss": 0.9625, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.94e-07, | |
| "loss": 0.9066, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.44e-07, | |
| "loss": 0.6918, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.939999999999999e-07, | |
| "loss": 0.6384, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.44e-07, | |
| "loss": 0.589, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 6.939999999999999e-07, | |
| "loss": 0.572, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.44e-07, | |
| "loss": 0.509, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.94e-07, | |
| "loss": 0.5418, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.439999999999999e-07, | |
| "loss": 0.5221, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.939999999999999e-07, | |
| "loss": 0.5158, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.439999999999999e-07, | |
| "loss": 0.5084, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.94e-07, | |
| "loss": 0.4492, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.95111111111111e-07, | |
| "loss": 0.465, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.895555555555554e-07, | |
| "loss": 0.4877, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.84e-07, | |
| "loss": 0.4555, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.784444444444444e-07, | |
| "loss": 0.4367, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.728888888888888e-07, | |
| "loss": 0.4637, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.673333333333332e-07, | |
| "loss": 0.4346, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.617777777777776e-07, | |
| "loss": 0.413, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.562222222222223e-07, | |
| "loss": 0.3905, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.506666666666667e-07, | |
| "loss": 0.3906, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.451111111111111e-07, | |
| "loss": 0.3564, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.395555555555556e-07, | |
| "loss": 0.3979, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.34e-07, | |
| "loss": 0.3028, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.284444444444444e-07, | |
| "loss": 0.3383, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.228888888888888e-07, | |
| "loss": 0.3794, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.173333333333333e-07, | |
| "loss": 0.364, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.117777777777778e-07, | |
| "loss": 0.3377, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.062222222222222e-07, | |
| "loss": 0.36, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.006666666666666e-07, | |
| "loss": 0.3508, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 8.95111111111111e-07, | |
| "loss": 0.3646, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.895555555555555e-07, | |
| "loss": 0.3812, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_loss": 0.3284018039703369, | |
| "eval_runtime": 1137.8075, | |
| "eval_samples_per_second": 4.432, | |
| "eval_steps_per_second": 1.108, | |
| "eval_wer": 14.348852177644284, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.839999999999999e-07, | |
| "loss": 0.3607, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.784444444444444e-07, | |
| "loss": 0.3375, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.728888888888889e-07, | |
| "loss": 0.3931, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.673333333333332e-07, | |
| "loss": 0.3583, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.617777777777777e-07, | |
| "loss": 0.3617, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 8.562222222222222e-07, | |
| "loss": 0.3749, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.506666666666667e-07, | |
| "loss": 0.373, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.451111111111111e-07, | |
| "loss": 0.3757, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.395555555555556e-07, | |
| "loss": 0.387, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.34e-07, | |
| "loss": 0.3225, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.284444444444444e-07, | |
| "loss": 0.3071, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.228888888888889e-07, | |
| "loss": 0.3026, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 8.173333333333333e-07, | |
| "loss": 0.3326, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.117777777777778e-07, | |
| "loss": 0.3171, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.062222222222221e-07, | |
| "loss": 0.2751, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.006666666666666e-07, | |
| "loss": 0.3204, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.95111111111111e-07, | |
| "loss": 0.2981, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.895555555555555e-07, | |
| "loss": 0.3096, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.84e-07, | |
| "loss": 0.2902, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.784444444444444e-07, | |
| "loss": 0.2813, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.728888888888888e-07, | |
| "loss": 0.3082, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.673333333333332e-07, | |
| "loss": 0.3051, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.617777777777778e-07, | |
| "loss": 0.2862, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.562222222222222e-07, | |
| "loss": 0.2478, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.506666666666667e-07, | |
| "loss": 0.2556, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.451111111111111e-07, | |
| "loss": 0.2488, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.395555555555555e-07, | |
| "loss": 0.2963, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.34e-07, | |
| "loss": 0.2866, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 7.284444444444444e-07, | |
| "loss": 0.2856, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 7.228888888888889e-07, | |
| "loss": 0.2732, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.173333333333333e-07, | |
| "loss": 0.306, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.117777777777777e-07, | |
| "loss": 0.2954, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.062222222222222e-07, | |
| "loss": 0.264, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.006666666666666e-07, | |
| "loss": 0.2844, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 6.951111111111111e-07, | |
| "loss": 0.2856, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.895555555555555e-07, | |
| "loss": 0.2715, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.84e-07, | |
| "loss": 0.2429, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.784444444444443e-07, | |
| "loss": 0.2367, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.728888888888888e-07, | |
| "loss": 0.2324, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 6.673333333333334e-07, | |
| "loss": 0.243, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "eval_loss": 0.2384428232908249, | |
| "eval_runtime": 1063.3104, | |
| "eval_samples_per_second": 4.743, | |
| "eval_steps_per_second": 1.186, | |
| "eval_wer": 10.152327826646642, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 6.617777777777778e-07, | |
| "loss": 0.2703, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 6.562222222222223e-07, | |
| "loss": 0.295, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.506666666666666e-07, | |
| "loss": 0.2683, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 6.451111111111111e-07, | |
| "loss": 0.2618, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.395555555555555e-07, | |
| "loss": 0.2385, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.34e-07, | |
| "loss": 0.2422, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.284444444444445e-07, | |
| "loss": 0.2486, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 6.228888888888889e-07, | |
| "loss": 0.2198, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 6.173333333333333e-07, | |
| "loss": 0.23, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.117777777777777e-07, | |
| "loss": 0.2443, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 6.062222222222222e-07, | |
| "loss": 0.2439, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.006666666666666e-07, | |
| "loss": 0.2579, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 5.951111111111111e-07, | |
| "loss": 0.2318, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 5.895555555555555e-07, | |
| "loss": 0.2471, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 5.839999999999999e-07, | |
| "loss": 0.2224, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 5.784444444444444e-07, | |
| "loss": 0.2422, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 5.728888888888888e-07, | |
| "loss": 0.2285, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.673333333333334e-07, | |
| "loss": 0.2241, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 5.617777777777778e-07, | |
| "loss": 0.2263, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.562222222222222e-07, | |
| "loss": 0.2113, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.506666666666666e-07, | |
| "loss": 0.2461, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.451111111111111e-07, | |
| "loss": 0.2179, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.395555555555556e-07, | |
| "loss": 0.1978, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 5.34e-07, | |
| "loss": 0.2208, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 5.284444444444445e-07, | |
| "loss": 0.1959, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 5.228888888888888e-07, | |
| "loss": 0.2211, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 5.173333333333333e-07, | |
| "loss": 0.1886, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 5.117777777777777e-07, | |
| "loss": 0.2161, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 5.062222222222222e-07, | |
| "loss": 0.1965, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 5.006666666666667e-07, | |
| "loss": 0.2322, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 4.951111111111111e-07, | |
| "loss": 0.221, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 4.895555555555555e-07, | |
| "loss": 0.1939, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 4.839999999999999e-07, | |
| "loss": 0.2081, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 4.784444444444444e-07, | |
| "loss": 0.2016, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 4.728888888888889e-07, | |
| "loss": 0.2085, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.673333333333333e-07, | |
| "loss": 0.2066, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 4.6177777777777777e-07, | |
| "loss": 0.2011, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 4.5622222222222217e-07, | |
| "loss": 0.2275, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 4.506666666666666e-07, | |
| "loss": 0.2265, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 4.451111111111111e-07, | |
| "loss": 0.2088, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_loss": 0.19160950183868408, | |
| "eval_runtime": 1003.8357, | |
| "eval_samples_per_second": 5.024, | |
| "eval_steps_per_second": 1.256, | |
| "eval_wer": 7.944647071443896, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.3955555555555554e-07, | |
| "loss": 0.2188, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 4.34e-07, | |
| "loss": 0.2171, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 4.2844444444444445e-07, | |
| "loss": 0.2175, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 4.2288888888888886e-07, | |
| "loss": 0.1954, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 4.173333333333333e-07, | |
| "loss": 0.204, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 4.1177777777777777e-07, | |
| "loss": 0.1982, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.0622222222222217e-07, | |
| "loss": 0.178, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.0066666666666663e-07, | |
| "loss": 0.1816, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.9511111111111114e-07, | |
| "loss": 0.1868, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.8955555555555554e-07, | |
| "loss": 0.1831, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.84e-07, | |
| "loss": 0.2125, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.7844444444444445e-07, | |
| "loss": 0.1779, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 3.7288888888888886e-07, | |
| "loss": 0.1762, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.673333333333333e-07, | |
| "loss": 0.1805, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.617777777777777e-07, | |
| "loss": 0.2095, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.5622222222222223e-07, | |
| "loss": 0.1855, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.506666666666667e-07, | |
| "loss": 0.184, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.451111111111111e-07, | |
| "loss": 0.1778, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.3955555555555554e-07, | |
| "loss": 0.1979, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 3.34e-07, | |
| "loss": 0.1684, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.284444444444444e-07, | |
| "loss": 0.2078, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 3.2288888888888886e-07, | |
| "loss": 0.1852, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.173333333333333e-07, | |
| "loss": 0.188, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 3.1177777777777777e-07, | |
| "loss": 0.178, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 3.0622222222222223e-07, | |
| "loss": 0.1843, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.006666666666667e-07, | |
| "loss": 0.2204, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.951111111111111e-07, | |
| "loss": 0.2049, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.8955555555555555e-07, | |
| "loss": 0.1702, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 2.8399999999999995e-07, | |
| "loss": 0.1929, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 2.784444444444444e-07, | |
| "loss": 0.1808, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 2.728888888888889e-07, | |
| "loss": 0.1835, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.673333333333333e-07, | |
| "loss": 0.156, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.617777777777778e-07, | |
| "loss": 0.1728, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.5622222222222223e-07, | |
| "loss": 0.1886, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 2.5066666666666663e-07, | |
| "loss": 0.1515, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 2.451111111111111e-07, | |
| "loss": 0.1548, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 2.3955555555555555e-07, | |
| "loss": 0.1775, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 2.34e-07, | |
| "loss": 0.1601, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 2.2844444444444443e-07, | |
| "loss": 0.1879, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 2.2288888888888886e-07, | |
| "loss": 0.1798, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "eval_loss": 0.1675407886505127, | |
| "eval_runtime": 1057.188, | |
| "eval_samples_per_second": 4.77, | |
| "eval_steps_per_second": 1.193, | |
| "eval_wer": 6.848315812057498, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 2.1733333333333332e-07, | |
| "loss": 0.172, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 2.1177777777777778e-07, | |
| "loss": 0.1642, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 2.062222222222222e-07, | |
| "loss": 0.1672, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 2.0066666666666666e-07, | |
| "loss": 0.1864, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 1.9511111111111112e-07, | |
| "loss": 0.1633, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 1.8955555555555555e-07, | |
| "loss": 0.1621, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 1.8399999999999998e-07, | |
| "loss": 0.1781, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 1.7844444444444444e-07, | |
| "loss": 0.1659, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.728888888888889e-07, | |
| "loss": 0.1782, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 1.6733333333333332e-07, | |
| "loss": 0.1735, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 1.6177777777777775e-07, | |
| "loss": 0.1771, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.5622222222222224e-07, | |
| "loss": 0.1822, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 1.5066666666666667e-07, | |
| "loss": 0.1537, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 1.451111111111111e-07, | |
| "loss": 0.1823, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 1.3955555555555553e-07, | |
| "loss": 0.1811, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 1.34e-07, | |
| "loss": 0.1792, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.2844444444444444e-07, | |
| "loss": 0.1369, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.228888888888889e-07, | |
| "loss": 0.197, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 1.1733333333333333e-07, | |
| "loss": 0.1807, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 1.1177777777777778e-07, | |
| "loss": 0.1656, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 1.0622222222222221e-07, | |
| "loss": 0.1539, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 1.0066666666666667e-07, | |
| "loss": 0.1563, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 9.51111111111111e-08, | |
| "loss": 0.1651, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 8.955555555555555e-08, | |
| "loss": 0.1486, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 8.4e-08, | |
| "loss": 0.167, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 7.844444444444444e-08, | |
| "loss": 0.1484, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 7.288888888888888e-08, | |
| "loss": 0.1605, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 6.733333333333333e-08, | |
| "loss": 0.1729, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 6.177777777777777e-08, | |
| "loss": 0.1555, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 5.622222222222222e-08, | |
| "loss": 0.1583, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 5.0666666666666664e-08, | |
| "loss": 0.1577, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 4.511111111111111e-08, | |
| "loss": 0.1596, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 3.955555555555555e-08, | |
| "loss": 0.1739, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 3.422222222222222e-08, | |
| "loss": 0.1583, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 2.8666666666666665e-08, | |
| "loss": 0.1556, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 2.311111111111111e-08, | |
| "loss": 0.164, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.7555555555555555e-08, | |
| "loss": 0.1564, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 1.2e-08, | |
| "loss": 0.1689, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 6.444444444444444e-09, | |
| "loss": 0.1571, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 8.888888888888889e-10, | |
| "loss": 0.1671, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "eval_loss": 0.15959425270557404, | |
| "eval_runtime": 1033.6723, | |
| "eval_samples_per_second": 4.879, | |
| "eval_steps_per_second": 1.22, | |
| "eval_wer": 6.472859901308732, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "step": 5000, | |
| "total_flos": 1.15333155237888e+19, | |
| "train_loss": 0.30382301959991453, | |
| "train_runtime": 10304.0521, | |
| "train_samples_per_second": 3.882, | |
| "train_steps_per_second": 0.485 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 8, | |
| "total_flos": 1.15333155237888e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |