| { |
| "best_metric": 38.89501329356072, |
| "best_model_checkpoint": "./output/small/yt-special-batch12-lr4-augment-small/checkpoint-5000", |
| "epoch": 0.6441638752898737, |
| "global_step": 5000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 4.2000000000000004e-06, |
| "loss": 2.437, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 9.2e-06, |
| "loss": 1.5643, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 1.42e-05, |
| "loss": 1.1751, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 1.9200000000000003e-05, |
| "loss": 1.1066, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 2.4200000000000002e-05, |
| "loss": 1.1131, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 2.9199999999999998e-05, |
| "loss": 1.1213, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 3.4200000000000005e-05, |
| "loss": 1.1921, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 3.9200000000000004e-05, |
| "loss": 1.202, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.4200000000000004e-05, |
| "loss": 1.2534, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.92e-05, |
| "loss": 1.1287, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 5.420000000000001e-05, |
| "loss": 1.3, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 5.92e-05, |
| "loss": 1.2233, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 6.42e-05, |
| "loss": 1.4307, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 6.92e-05, |
| "loss": 1.3264, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 7.42e-05, |
| "loss": 1.3162, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 7.920000000000001e-05, |
| "loss": 1.3683, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 8.42e-05, |
| "loss": 1.4094, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 8.92e-05, |
| "loss": 1.4662, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 9.42e-05, |
| "loss": 1.4977, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 9.92e-05, |
| "loss": 1.5349, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 9.955555555555556e-05, |
| "loss": 1.8817, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 9.900000000000001e-05, |
| "loss": 1.6916, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 9.844444444444444e-05, |
| "loss": 1.5593, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 9.78888888888889e-05, |
| "loss": 1.5015, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 9.733333333333335e-05, |
| "loss": 1.6808, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 9.677777777777778e-05, |
| "loss": 1.5124, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 9.622222222222222e-05, |
| "loss": 1.5804, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 9.566666666666667e-05, |
| "loss": 1.4944, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 9.511111111111112e-05, |
| "loss": 1.5928, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 9.455555555555556e-05, |
| "loss": 1.558, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 9.4e-05, |
| "loss": 1.4307, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 9.344444444444444e-05, |
| "loss": 1.4087, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 9.28888888888889e-05, |
| "loss": 1.4799, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 9.233333333333333e-05, |
| "loss": 1.4866, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 9.177777777777778e-05, |
| "loss": 1.4138, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.122222222222223e-05, |
| "loss": 1.4391, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.066666666666667e-05, |
| "loss": 1.4554, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.011111111111111e-05, |
| "loss": 1.4142, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 8.955555555555556e-05, |
| "loss": 1.4077, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 8.900000000000001e-05, |
| "loss": 1.296, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.13, |
| "eval_loss": 1.3286159038543701, |
| "eval_runtime": 503.8235, |
| "eval_samples_per_second": 4.865, |
| "eval_steps_per_second": 0.812, |
| "eval_wer": 110.01232086116335, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 8.844444444444445e-05, |
| "loss": 1.4934, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 8.78888888888889e-05, |
| "loss": 1.3412, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 8.733333333333333e-05, |
| "loss": 1.3166, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 8.677777777777778e-05, |
| "loss": 1.2458, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 8.622222222222222e-05, |
| "loss": 1.3324, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 8.566666666666667e-05, |
| "loss": 1.3044, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 8.511111111111112e-05, |
| "loss": 1.4607, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 8.455555555555556e-05, |
| "loss": 1.3496, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 8.4e-05, |
| "loss": 1.2821, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 8.344444444444445e-05, |
| "loss": 1.381, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 8.28888888888889e-05, |
| "loss": 1.302, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 8.233333333333333e-05, |
| "loss": 1.3004, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 8.177777777777778e-05, |
| "loss": 1.1992, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 8.122222222222222e-05, |
| "loss": 1.3097, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 8.066666666666667e-05, |
| "loss": 1.1473, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 8.011111111111111e-05, |
| "loss": 1.1803, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 7.955555555555556e-05, |
| "loss": 1.277, |
| "step": 1425 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 7.900000000000001e-05, |
| "loss": 1.2602, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 7.844444444444446e-05, |
| "loss": 1.3802, |
| "step": 1475 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 7.788888888888888e-05, |
| "loss": 1.202, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 7.733333333333333e-05, |
| "loss": 1.261, |
| "step": 1525 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 7.677777777777778e-05, |
| "loss": 1.1906, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 7.622222222222223e-05, |
| "loss": 1.2868, |
| "step": 1575 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 7.566666666666667e-05, |
| "loss": 1.285, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 7.511111111111111e-05, |
| "loss": 1.2064, |
| "step": 1625 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 7.455555555555556e-05, |
| "loss": 1.1396, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 7.4e-05, |
| "loss": 1.1513, |
| "step": 1675 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 7.344444444444445e-05, |
| "loss": 1.2145, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 7.28888888888889e-05, |
| "loss": 1.1844, |
| "step": 1725 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 7.233333333333335e-05, |
| "loss": 1.1214, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 7.177777777777777e-05, |
| "loss": 1.2056, |
| "step": 1775 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 7.122222222222222e-05, |
| "loss": 1.1613, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 7.066666666666667e-05, |
| "loss": 1.2531, |
| "step": 1825 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 7.011111111111112e-05, |
| "loss": 1.1234, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 6.955555555555556e-05, |
| "loss": 1.1991, |
| "step": 1875 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 6.9e-05, |
| "loss": 1.1337, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 6.844444444444445e-05, |
| "loss": 1.0764, |
| "step": 1925 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 6.788888888888888e-05, |
| "loss": 1.1944, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 6.733333333333333e-05, |
| "loss": 1.1, |
| "step": 1975 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 6.677777777777779e-05, |
| "loss": 1.1522, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.26, |
| "eval_loss": 1.091708779335022, |
| "eval_runtime": 434.7606, |
| "eval_samples_per_second": 5.638, |
| "eval_steps_per_second": 0.941, |
| "eval_wer": 81.63543220284029, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 6.622222222222224e-05, |
| "loss": 1.0742, |
| "step": 2025 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 6.566666666666666e-05, |
| "loss": 1.1638, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 6.511111111111111e-05, |
| "loss": 1.0376, |
| "step": 2075 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 6.455555555555556e-05, |
| "loss": 1.0214, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 6.400000000000001e-05, |
| "loss": 1.0725, |
| "step": 2125 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 6.344444444444445e-05, |
| "loss": 1.0547, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 6.28888888888889e-05, |
| "loss": 1.0442, |
| "step": 2175 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 6.233333333333334e-05, |
| "loss": 1.1098, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 6.177777777777779e-05, |
| "loss": 1.1119, |
| "step": 2225 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 6.122222222222222e-05, |
| "loss": 0.9511, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 6.066666666666667e-05, |
| "loss": 0.9848, |
| "step": 2275 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 6.011111111111112e-05, |
| "loss": 1.0339, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 5.9555555555555554e-05, |
| "loss": 1.0054, |
| "step": 2325 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 5.9e-05, |
| "loss": 1.0645, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 5.844444444444445e-05, |
| "loss": 1.0975, |
| "step": 2375 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 5.788888888888889e-05, |
| "loss": 0.8899, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 5.7333333333333336e-05, |
| "loss": 1.0181, |
| "step": 2425 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 5.6777777777777786e-05, |
| "loss": 0.9479, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 5.622222222222222e-05, |
| "loss": 1.0964, |
| "step": 2475 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 5.566666666666667e-05, |
| "loss": 1.0784, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 5.511111111111111e-05, |
| "loss": 0.8982, |
| "step": 2525 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 5.455555555555556e-05, |
| "loss": 1.0107, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 5.4000000000000005e-05, |
| "loss": 0.8865, |
| "step": 2575 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 5.3444444444444455e-05, |
| "loss": 1.0161, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 5.2888888888888885e-05, |
| "loss": 1.0056, |
| "step": 2625 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 5.2333333333333336e-05, |
| "loss": 0.9969, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 5.177777777777778e-05, |
| "loss": 0.9812, |
| "step": 2675 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 5.122222222222223e-05, |
| "loss": 0.9782, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 5.0666666666666674e-05, |
| "loss": 0.9906, |
| "step": 2725 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 5.011111111111111e-05, |
| "loss": 0.9422, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.955555555555556e-05, |
| "loss": 0.997, |
| "step": 2775 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.9e-05, |
| "loss": 0.9982, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.844444444444445e-05, |
| "loss": 0.9779, |
| "step": 2825 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.7888888888888886e-05, |
| "loss": 0.8814, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.7333333333333336e-05, |
| "loss": 0.92, |
| "step": 2875 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.677777777777778e-05, |
| "loss": 0.8371, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.6222222222222224e-05, |
| "loss": 0.931, |
| "step": 2925 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.566666666666667e-05, |
| "loss": 0.8743, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.511111111111112e-05, |
| "loss": 0.8781, |
| "step": 2975 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.4555555555555555e-05, |
| "loss": 1.0175, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.39, |
| "eval_loss": 0.9398985505104065, |
| "eval_runtime": 408.4907, |
| "eval_samples_per_second": 6.0, |
| "eval_steps_per_second": 1.001, |
| "eval_wer": 60.12580247714156, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.4000000000000006e-05, |
| "loss": 0.8422, |
| "step": 3025 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.344444444444445e-05, |
| "loss": 0.846, |
| "step": 3050 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.2888888888888886e-05, |
| "loss": 0.9857, |
| "step": 3075 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.233333333333334e-05, |
| "loss": 1.0341, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.177777777777778e-05, |
| "loss": 0.823, |
| "step": 3125 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.1222222222222224e-05, |
| "loss": 0.8259, |
| "step": 3150 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.066666666666667e-05, |
| "loss": 0.899, |
| "step": 3175 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.011111111111111e-05, |
| "loss": 0.8578, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.9555555555555556e-05, |
| "loss": 0.8267, |
| "step": 3225 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.9000000000000006e-05, |
| "loss": 0.8756, |
| "step": 3250 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 3.844444444444444e-05, |
| "loss": 0.785, |
| "step": 3275 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.7888888888888894e-05, |
| "loss": 0.7485, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.733333333333334e-05, |
| "loss": 0.7924, |
| "step": 3325 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.677777777777778e-05, |
| "loss": 0.7986, |
| "step": 3350 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 3.6222222222222225e-05, |
| "loss": 0.8421, |
| "step": 3375 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.566666666666667e-05, |
| "loss": 0.8089, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.511111111111111e-05, |
| "loss": 0.8736, |
| "step": 3425 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 3.4555555555555556e-05, |
| "loss": 0.7106, |
| "step": 3450 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.4000000000000007e-05, |
| "loss": 0.8124, |
| "step": 3475 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.3444444444444443e-05, |
| "loss": 0.7908, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 3.2888888888888894e-05, |
| "loss": 0.8274, |
| "step": 3525 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.233333333333333e-05, |
| "loss": 0.7894, |
| "step": 3550 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.177777777777778e-05, |
| "loss": 0.8306, |
| "step": 3575 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 3.1222222222222225e-05, |
| "loss": 0.8281, |
| "step": 3600 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 3.066666666666667e-05, |
| "loss": 0.7585, |
| "step": 3625 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 3.0111111111111113e-05, |
| "loss": 0.8072, |
| "step": 3650 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 2.955555555555556e-05, |
| "loss": 0.7316, |
| "step": 3675 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 2.9e-05, |
| "loss": 0.7521, |
| "step": 3700 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 2.8444444444444447e-05, |
| "loss": 0.8323, |
| "step": 3725 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 2.788888888888889e-05, |
| "loss": 0.804, |
| "step": 3750 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 2.733333333333333e-05, |
| "loss": 0.719, |
| "step": 3775 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 2.677777777777778e-05, |
| "loss": 0.7169, |
| "step": 3800 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 2.6222222222222226e-05, |
| "loss": 0.7348, |
| "step": 3825 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 2.5666666666666666e-05, |
| "loss": 0.7335, |
| "step": 3850 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 2.5111111111111113e-05, |
| "loss": 0.7399, |
| "step": 3875 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 2.4555555555555557e-05, |
| "loss": 0.7904, |
| "step": 3900 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 2.4e-05, |
| "loss": 0.7552, |
| "step": 3925 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 2.3444444444444448e-05, |
| "loss": 0.7508, |
| "step": 3950 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 2.288888888888889e-05, |
| "loss": 0.7625, |
| "step": 3975 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 2.2333333333333335e-05, |
| "loss": 0.7844, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.52, |
| "eval_loss": 0.8174441456794739, |
| "eval_runtime": 335.0331, |
| "eval_samples_per_second": 7.316, |
| "eval_steps_per_second": 1.221, |
| "eval_wer": 43.48615524285066, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 2.177777777777778e-05, |
| "loss": 0.7431, |
| "step": 4025 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 2.1222222222222223e-05, |
| "loss": 0.6498, |
| "step": 4050 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 2.0666666666666666e-05, |
| "loss": 0.7371, |
| "step": 4075 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 2.011111111111111e-05, |
| "loss": 0.6258, |
| "step": 4100 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 1.9555555555555557e-05, |
| "loss": 0.6992, |
| "step": 4125 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 1.9e-05, |
| "loss": 0.7135, |
| "step": 4150 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 1.8444444444444445e-05, |
| "loss": 0.625, |
| "step": 4175 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 1.788888888888889e-05, |
| "loss": 0.7359, |
| "step": 4200 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 1.7333333333333336e-05, |
| "loss": 0.7254, |
| "step": 4225 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 1.677777777777778e-05, |
| "loss": 0.7082, |
| "step": 4250 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 1.6222222222222223e-05, |
| "loss": 0.6745, |
| "step": 4275 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 1.5666666666666667e-05, |
| "loss": 0.6164, |
| "step": 4300 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 1.5111111111111112e-05, |
| "loss": 0.6287, |
| "step": 4325 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 1.4555555555555556e-05, |
| "loss": 0.7443, |
| "step": 4350 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 1.4000000000000001e-05, |
| "loss": 0.7288, |
| "step": 4375 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 1.3444444444444445e-05, |
| "loss": 0.6984, |
| "step": 4400 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 1.2888888888888889e-05, |
| "loss": 0.7351, |
| "step": 4425 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 1.2333333333333334e-05, |
| "loss": 0.64, |
| "step": 4450 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 1.1777777777777778e-05, |
| "loss": 0.5714, |
| "step": 4475 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 1.1222222222222224e-05, |
| "loss": 0.6156, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 1.0666666666666667e-05, |
| "loss": 0.6056, |
| "step": 4525 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 1.0111111111111111e-05, |
| "loss": 0.6628, |
| "step": 4550 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 9.555555555555556e-06, |
| "loss": 0.6848, |
| "step": 4575 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 9e-06, |
| "loss": 0.5885, |
| "step": 4600 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 8.444444444444446e-06, |
| "loss": 0.5594, |
| "step": 4625 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.88888888888889e-06, |
| "loss": 0.632, |
| "step": 4650 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.333333333333334e-06, |
| "loss": 0.566, |
| "step": 4675 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.777777777777779e-06, |
| "loss": 0.6581, |
| "step": 4700 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.222222222222222e-06, |
| "loss": 0.5415, |
| "step": 4725 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.666666666666667e-06, |
| "loss": 0.5902, |
| "step": 4750 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.1111111111111115e-06, |
| "loss": 0.6086, |
| "step": 4775 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.555555555555556e-06, |
| "loss": 0.6157, |
| "step": 4800 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 0.5811, |
| "step": 4825 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 3.4444444444444444e-06, |
| "loss": 0.5793, |
| "step": 4850 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 2.888888888888889e-06, |
| "loss": 0.6283, |
| "step": 4875 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 2.3333333333333336e-06, |
| "loss": 0.5631, |
| "step": 4900 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 1.777777777777778e-06, |
| "loss": 0.5213, |
| "step": 4925 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 1.2222222222222223e-06, |
| "loss": 0.6272, |
| "step": 4950 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.666666666666667e-07, |
| "loss": 0.5401, |
| "step": 4975 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 1.1111111111111112e-07, |
| "loss": 0.5919, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.64, |
| "eval_loss": 0.7352390289306641, |
| "eval_runtime": 329.899, |
| "eval_samples_per_second": 7.43, |
| "eval_steps_per_second": 1.24, |
| "eval_wer": 38.89501329356072, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.64, |
| "step": 5000, |
| "total_flos": 1.73151240192e+19, |
| "train_loss": 1.0241181322097779, |
| "train_runtime": 8795.9818, |
| "train_samples_per_second": 6.821, |
| "train_steps_per_second": 0.568 |
| } |
| ], |
| "max_steps": 5000, |
| "num_train_epochs": 1, |
| "total_flos": 1.73151240192e+19, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
