| { | |
| "best_metric": 22.336104207729882, | |
| "best_model_checkpoint": "/data/models/checkpoint-5000", | |
| "epoch": 10.038, | |
| "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": 4.973833272194737e-06, | |
| "loss": 0.8775, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.195318418690893e-06, | |
| "loss": 0.559, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.881634451095711e-06, | |
| "loss": 0.4533, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.361221988663844e-06, | |
| "loss": 0.4225, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.730207550743121e-06, | |
| "loss": 0.4111, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.03016458599496e-06, | |
| "loss": 0.4111, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.282894746203441e-06, | |
| "loss": 0.3606, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.501266121799902e-06, | |
| "loss": 0.3653, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.693512601774437e-06, | |
| "loss": 0.3474, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.865222471593567e-06, | |
| "loss": 0.3357, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.020362953730323e-06, | |
| "loss": 0.3576, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.161852281961698e-06, | |
| "loss": 0.3536, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.29189975311636e-06, | |
| "loss": 0.4535, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.412218256259678e-06, | |
| "loss": 0.3593, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.524162683365145e-06, | |
| "loss": 0.3045, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.62882322733502e-06, | |
| "loss": 0.3082, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.727090137141168e-06, | |
| "loss": 0.2835, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.819699807237934e-06, | |
| "loss": 0.2769, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.907268307310855e-06, | |
| "loss": 0.3538, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.990316248055788e-06, | |
| "loss": 0.2965, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.953333333333333e-06, | |
| "loss": 0.2416, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.89777777777778e-06, | |
| "loss": 0.2323, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.2063, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.2012, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.731111111111113e-06, | |
| "loss": 0.2022, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.675555555555555e-06, | |
| "loss": 0.1713, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.620000000000001e-06, | |
| "loss": 0.1754, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.564444444444445e-06, | |
| "loss": 0.1569, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.508888888888889e-06, | |
| "loss": 0.15, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.453333333333335e-06, | |
| "loss": 0.1707, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.397777777777779e-06, | |
| "loss": 0.1728, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.342222222222223e-06, | |
| "loss": 0.2256, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.286666666666667e-06, | |
| "loss": 0.1889, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.231111111111111e-06, | |
| "loss": 0.1431, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.175555555555557e-06, | |
| "loss": 0.1393, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.12e-06, | |
| "loss": 0.1279, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.064444444444447e-06, | |
| "loss": 0.1325, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.008888888888889e-06, | |
| "loss": 0.1593, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.953333333333335e-06, | |
| "loss": 0.1557, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.897777777777779e-06, | |
| "loss": 0.114, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "eval_loss": 0.292724609375, | |
| "eval_runtime": 1889.2151, | |
| "eval_samples_per_second": 4.168, | |
| "eval_steps_per_second": 0.261, | |
| "eval_wer": 24.943011397720454, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.842222222222223e-06, | |
| "loss": 0.1034, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.786666666666668e-06, | |
| "loss": 0.0954, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.73111111111111e-06, | |
| "loss": 0.0877, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.675555555555556e-06, | |
| "loss": 0.0869, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.62e-06, | |
| "loss": 0.0783, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.564444444444445e-06, | |
| "loss": 0.0744, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.50888888888889e-06, | |
| "loss": 0.0751, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.453333333333334e-06, | |
| "loss": 0.0655, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.397777777777778e-06, | |
| "loss": 0.0651, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.342222222222222e-06, | |
| "loss": 0.0773, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.1017, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.0881, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.0663, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.0603, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.0601, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.0541, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.0642, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.0741, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.0497, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.051, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.0402, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.0357, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.0368, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.0353, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.0314, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.0324, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0285, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.0288, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.0339, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.0416, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.0419, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.0277, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.0272, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.0229, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.0237, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0269, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.0348, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.0225, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.0219, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.0198, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "eval_loss": 0.31201171875, | |
| "eval_runtime": 1865.4253, | |
| "eval_samples_per_second": 4.222, | |
| "eval_steps_per_second": 0.264, | |
| "eval_wer": 23.530722426943182, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.0158, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.0167, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.0164, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.015, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0148, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.0138, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.0145, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.0158, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.0179, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.0194, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 6.0711111111111125e-06, | |
| "loss": 0.0151, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 6.015555555555556e-06, | |
| "loss": 0.0114, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.012, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 5.9044444444444446e-06, | |
| "loss": 0.0106, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 5.8488888888888895e-06, | |
| "loss": 0.0116, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 5.793333333333334e-06, | |
| "loss": 0.015, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.737777777777778e-06, | |
| "loss": 0.0129, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.682222222222222e-06, | |
| "loss": 0.0095, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.626666666666667e-06, | |
| "loss": 0.0104, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.571111111111111e-06, | |
| "loss": 0.0077, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.515555555555556e-06, | |
| "loss": 0.0079, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.0075, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.404444444444444e-06, | |
| "loss": 0.007, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 5.348888888888889e-06, | |
| "loss": 0.008, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 5.293333333333334e-06, | |
| "loss": 0.0078, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 5.237777777777778e-06, | |
| "loss": 0.0065, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 5.182222222222223e-06, | |
| "loss": 0.0071, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 5.126666666666668e-06, | |
| "loss": 0.0086, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 5.071111111111111e-06, | |
| "loss": 0.0092, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 5.015555555555556e-06, | |
| "loss": 0.0081, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.0061, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 4.904444444444445e-06, | |
| "loss": 0.0057, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 4.848888888888889e-06, | |
| "loss": 0.0058, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 4.793333333333334e-06, | |
| "loss": 0.0068, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 4.737777777777779e-06, | |
| "loss": 0.0076, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.682222222222223e-06, | |
| "loss": 0.0073, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.626666666666667e-06, | |
| "loss": 0.0055, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.571111111111112e-06, | |
| "loss": 0.0052, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.515555555555556e-06, | |
| "loss": 0.0048, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.0043, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "eval_loss": 0.326171875, | |
| "eval_runtime": 1867.9986, | |
| "eval_samples_per_second": 4.216, | |
| "eval_steps_per_second": 0.264, | |
| "eval_wer": 22.41837346816351, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 4.404444444444445e-06, | |
| "loss": 0.0045, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 4.348888888888889e-06, | |
| "loss": 0.0043, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.2933333333333334e-06, | |
| "loss": 0.0045, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.2377777777777775e-06, | |
| "loss": 0.0039, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 4.182222222222222e-06, | |
| "loss": 0.0039, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 4.126666666666667e-06, | |
| "loss": 0.0041, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 4.071111111111111e-06, | |
| "loss": 0.0045, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 4.015555555555556e-06, | |
| "loss": 0.0049, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.0045, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 3.904444444444444e-06, | |
| "loss": 0.0039, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 3.853333333333334e-06, | |
| "loss": 0.0034, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 3.797777777777778e-06, | |
| "loss": 0.0036, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 3.7422222222222228e-06, | |
| "loss": 0.0032, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 3.686666666666667e-06, | |
| "loss": 0.0043, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.6311111111111113e-06, | |
| "loss": 0.0039, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.575555555555556e-06, | |
| "loss": 0.0037, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.52e-06, | |
| "loss": 0.0035, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.4644444444444446e-06, | |
| "loss": 0.0032, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.4088888888888895e-06, | |
| "loss": 0.0027, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.3533333333333336e-06, | |
| "loss": 0.003, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.297777777777778e-06, | |
| "loss": 0.0027, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.242222222222223e-06, | |
| "loss": 0.0028, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.186666666666667e-06, | |
| "loss": 0.0032, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 3.1311111111111114e-06, | |
| "loss": 0.0023, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 3.0755555555555563e-06, | |
| "loss": 0.0026, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 3.0200000000000003e-06, | |
| "loss": 0.0027, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 2.9644444444444448e-06, | |
| "loss": 0.003, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 2.908888888888889e-06, | |
| "loss": 0.0032, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 2.8533333333333337e-06, | |
| "loss": 0.0027, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.797777777777778e-06, | |
| "loss": 0.0024, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.742222222222222e-06, | |
| "loss": 0.0022, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 2.686666666666667e-06, | |
| "loss": 0.0022, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 2.6311111111111115e-06, | |
| "loss": 0.0023, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.5755555555555556e-06, | |
| "loss": 0.003, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.52e-06, | |
| "loss": 0.0024, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.4644444444444445e-06, | |
| "loss": 0.0022, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.4088888888888894e-06, | |
| "loss": 0.0022, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.3533333333333334e-06, | |
| "loss": 0.0021, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 2.297777777777778e-06, | |
| "loss": 0.0021, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 2.2422222222222223e-06, | |
| "loss": 0.0021, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "eval_loss": 0.342529296875, | |
| "eval_runtime": 1853.0876, | |
| "eval_samples_per_second": 4.25, | |
| "eval_steps_per_second": 0.266, | |
| "eval_wer": 22.346387865284086, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 2.1866666666666668e-06, | |
| "loss": 0.002, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 2.1311111111111112e-06, | |
| "loss": 0.002, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 2.0755555555555557e-06, | |
| "loss": 0.0018, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 2.02e-06, | |
| "loss": 0.002, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 1.9644444444444446e-06, | |
| "loss": 0.0021, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 1.908888888888889e-06, | |
| "loss": 0.0022, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 1.8533333333333333e-06, | |
| "loss": 0.0023, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 1.797777777777778e-06, | |
| "loss": 0.002, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.7422222222222224e-06, | |
| "loss": 0.0019, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.6866666666666667e-06, | |
| "loss": 0.0019, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 1.6355555555555559e-06, | |
| "loss": 0.0018, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 1.5800000000000001e-06, | |
| "loss": 0.0019, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 1.5244444444444446e-06, | |
| "loss": 0.002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1.468888888888889e-06, | |
| "loss": 0.002, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 1.4133333333333335e-06, | |
| "loss": 0.0019, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 1.357777777777778e-06, | |
| "loss": 0.0018, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 1.3022222222222222e-06, | |
| "loss": 0.0017, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 1.2466666666666667e-06, | |
| "loss": 0.0018, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.1911111111111111e-06, | |
| "loss": 0.0017, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.1355555555555558e-06, | |
| "loss": 0.0018, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 1.08e-06, | |
| "loss": 0.0018, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 1.0244444444444445e-06, | |
| "loss": 0.0016, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 9.68888888888889e-07, | |
| "loss": 0.0017, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 9.133333333333334e-07, | |
| "loss": 0.0017, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 8.577777777777778e-07, | |
| "loss": 0.0019, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 8.022222222222223e-07, | |
| "loss": 0.002, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 7.466666666666668e-07, | |
| "loss": 0.0019, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 6.911111111111111e-07, | |
| "loss": 0.0017, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 6.355555555555556e-07, | |
| "loss": 0.0017, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 5.800000000000001e-07, | |
| "loss": 0.0016, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 5.244444444444445e-07, | |
| "loss": 0.0016, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.688888888888889e-07, | |
| "loss": 0.0018, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 4.133333333333334e-07, | |
| "loss": 0.0018, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 3.577777777777778e-07, | |
| "loss": 0.0017, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 3.0222222222222225e-07, | |
| "loss": 0.0016, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 2.466666666666667e-07, | |
| "loss": 0.0016, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 1.911111111111111e-07, | |
| "loss": 0.0016, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 1.3555555555555557e-07, | |
| "loss": 0.0016, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 8e-08, | |
| "loss": 0.0015, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 2.4444444444444447e-08, | |
| "loss": 0.0016, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "eval_loss": 0.346923828125, | |
| "eval_runtime": 1849.4805, | |
| "eval_samples_per_second": 4.258, | |
| "eval_steps_per_second": 0.267, | |
| "eval_wer": 22.336104207729882, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "step": 5000, | |
| "total_flos": 4.6127490743727555e+19, | |
| "train_loss": 0.0703210298538208, | |
| "train_runtime": 40445.5727, | |
| "train_samples_per_second": 3.956, | |
| "train_steps_per_second": 0.124 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 4.6127490743727555e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |