| { | |
| "best_metric": 15.36559705418201, | |
| "best_model_checkpoint": "./checkpoint-3000", | |
| "epoch": 37.878787878787875, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.3000000000000002e-07, | |
| "loss": 1.3949, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.2698, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.3e-07, | |
| "loss": 0.9741, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 0.6225, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.23e-06, | |
| "loss": 0.4511, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.396, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.73e-06, | |
| "loss": 0.3595, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.3365, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.2300000000000002e-06, | |
| "loss": 0.3351, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.316, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.7300000000000005e-06, | |
| "loss": 0.2856, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.214, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.2300000000000004e-06, | |
| "loss": 0.2106, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.2302, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.7300000000000003e-06, | |
| "loss": 0.2099, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.1893, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 4.23e-06, | |
| "loss": 0.1249, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.128, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 4.7300000000000005e-06, | |
| "loss": 0.12, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.1272, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 4.974444444444445e-06, | |
| "loss": 0.1241, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 4.946666666666667e-06, | |
| "loss": 0.0786, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 4.918888888888889e-06, | |
| "loss": 0.0713, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 4.891111111111111e-06, | |
| "loss": 0.0708, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 4.863333333333334e-06, | |
| "loss": 0.0754, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 4.835555555555556e-06, | |
| "loss": 0.0721, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 4.807777777777779e-06, | |
| "loss": 0.0526, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.041, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 4.752222222222223e-06, | |
| "loss": 0.0426, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 4.724444444444445e-06, | |
| "loss": 0.0405, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 4.696666666666667e-06, | |
| "loss": 0.0453, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 4.66888888888889e-06, | |
| "loss": 0.0372, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 4.641111111111111e-06, | |
| "loss": 0.0229, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 4.613333333333334e-06, | |
| "loss": 0.0229, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 4.585555555555556e-06, | |
| "loss": 0.0257, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 4.557777777777778e-06, | |
| "loss": 0.0247, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 4.530000000000001e-06, | |
| "loss": 0.0247, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 4.502222222222223e-06, | |
| "loss": 0.0147, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 4.474444444444445e-06, | |
| "loss": 0.0135, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 4.446666666666667e-06, | |
| "loss": 0.016, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "eval_loss": 0.4492044448852539, | |
| "eval_runtime": 1228.0958, | |
| "eval_samples_per_second": 1.737, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 15.739084692267227, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 4.41888888888889e-06, | |
| "loss": 0.0176, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 4.391111111111112e-06, | |
| "loss": 0.0155, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 4.363333333333334e-06, | |
| "loss": 0.013, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 4.3355555555555565e-06, | |
| "loss": 0.0079, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 4.307777777777778e-06, | |
| "loss": 0.0115, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.0112, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 4.2522222222222225e-06, | |
| "loss": 0.0133, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.2244444444444446e-06, | |
| "loss": 0.0093, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 4.1966666666666674e-06, | |
| "loss": 0.0076, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 4.168888888888889e-06, | |
| "loss": 0.0082, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 4.1411111111111115e-06, | |
| "loss": 0.0065, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 4.1133333333333335e-06, | |
| "loss": 0.0062, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 4.085555555555556e-06, | |
| "loss": 0.008, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 4.057777777777778e-06, | |
| "loss": 0.0052, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 4.03e-06, | |
| "loss": 0.0062, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 4.002222222222222e-06, | |
| "loss": 0.0068, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 3.974444444444444e-06, | |
| "loss": 0.0051, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 3.946666666666667e-06, | |
| "loss": 0.006, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 3.918888888888889e-06, | |
| "loss": 0.0047, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "learning_rate": 3.891111111111111e-06, | |
| "loss": 0.0034, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 3.863333333333333e-06, | |
| "loss": 0.0037, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 11.74, | |
| "learning_rate": 3.835555555555555e-06, | |
| "loss": 0.0051, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 3.8077777777777782e-06, | |
| "loss": 0.0043, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 12.12, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.0031, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 3.7522222222222227e-06, | |
| "loss": 0.0018, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 3.724444444444445e-06, | |
| "loss": 0.0038, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 12.69, | |
| "learning_rate": 3.6966666666666667e-06, | |
| "loss": 0.0029, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 12.88, | |
| "learning_rate": 3.668888888888889e-06, | |
| "loss": 0.0032, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 3.641111111111111e-06, | |
| "loss": 0.0026, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 13.26, | |
| "learning_rate": 3.6133333333333336e-06, | |
| "loss": 0.0018, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "learning_rate": 3.585555555555556e-06, | |
| "loss": 0.0024, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 13.64, | |
| "learning_rate": 3.5577777777777785e-06, | |
| "loss": 0.0031, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 13.83, | |
| "learning_rate": 3.53e-06, | |
| "loss": 0.0038, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.5022222222222225e-06, | |
| "loss": 0.0028, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "learning_rate": 3.4744444444444445e-06, | |
| "loss": 0.0015, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 14.39, | |
| "learning_rate": 3.446666666666667e-06, | |
| "loss": 0.0017, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 14.58, | |
| "learning_rate": 3.4188888888888894e-06, | |
| "loss": 0.0036, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 3.391111111111111e-06, | |
| "loss": 0.0036, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 14.96, | |
| "learning_rate": 3.3633333333333335e-06, | |
| "loss": 0.0015, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "learning_rate": 3.335555555555556e-06, | |
| "loss": 0.0014, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "eval_loss": 0.5305609703063965, | |
| "eval_runtime": 1235.8408, | |
| "eval_samples_per_second": 1.726, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 15.45502367175171, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 15.34, | |
| "learning_rate": 3.307777777777778e-06, | |
| "loss": 0.0013, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 15.53, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.0018, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 15.72, | |
| "learning_rate": 3.252222222222223e-06, | |
| "loss": 0.0012, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 15.91, | |
| "learning_rate": 3.2244444444444444e-06, | |
| "loss": 0.0011, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 16.1, | |
| "learning_rate": 3.196666666666667e-06, | |
| "loss": 0.0009, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 16.29, | |
| "learning_rate": 3.1688888888888893e-06, | |
| "loss": 0.0013, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 16.48, | |
| "learning_rate": 3.1411111111111113e-06, | |
| "loss": 0.0008, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 3.1133333333333337e-06, | |
| "loss": 0.0009, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 16.86, | |
| "learning_rate": 3.085555555555556e-06, | |
| "loss": 0.0008, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 3.0577777777777778e-06, | |
| "loss": 0.0007, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 3.0300000000000002e-06, | |
| "loss": 0.0007, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 17.42, | |
| "learning_rate": 3.0022222222222227e-06, | |
| "loss": 0.0006, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 2.9744444444444447e-06, | |
| "loss": 0.0006, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "learning_rate": 2.946666666666667e-06, | |
| "loss": 0.0006, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 2.9188888888888887e-06, | |
| "loss": 0.0006, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "learning_rate": 2.891111111111111e-06, | |
| "loss": 0.0006, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 18.37, | |
| "learning_rate": 2.8633333333333336e-06, | |
| "loss": 0.0006, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 18.56, | |
| "learning_rate": 2.835555555555556e-06, | |
| "loss": 0.0005, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 2.807777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 18.94, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0006, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 19.13, | |
| "learning_rate": 2.752222222222222e-06, | |
| "loss": 0.0005, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 19.32, | |
| "learning_rate": 2.7244444444444445e-06, | |
| "loss": 0.0005, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 19.51, | |
| "learning_rate": 2.696666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 19.7, | |
| "learning_rate": 2.6688888888888894e-06, | |
| "loss": 0.0005, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 2.6411111111111114e-06, | |
| "loss": 0.0005, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 20.08, | |
| "learning_rate": 2.6133333333333334e-06, | |
| "loss": 0.0005, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 20.27, | |
| "learning_rate": 2.5855555555555555e-06, | |
| "loss": 0.0005, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 20.45, | |
| "learning_rate": 2.557777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 20.64, | |
| "learning_rate": 2.5300000000000003e-06, | |
| "loss": 0.0005, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 20.83, | |
| "learning_rate": 2.5022222222222224e-06, | |
| "loss": 0.0005, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 2.474444444444445e-06, | |
| "loss": 0.0005, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 21.21, | |
| "learning_rate": 2.446666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 21.4, | |
| "learning_rate": 2.418888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 21.59, | |
| "learning_rate": 2.3911111111111113e-06, | |
| "loss": 0.0004, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 21.78, | |
| "learning_rate": 2.3633333333333337e-06, | |
| "loss": 0.0004, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 21.97, | |
| "learning_rate": 2.3355555555555557e-06, | |
| "loss": 0.0004, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 22.16, | |
| "learning_rate": 2.3077777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 22.35, | |
| "learning_rate": 2.28e-06, | |
| "loss": 0.0004, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 22.54, | |
| "learning_rate": 2.252222222222222e-06, | |
| "loss": 0.0004, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 22.73, | |
| "learning_rate": 2.2244444444444447e-06, | |
| "loss": 0.0004, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 22.73, | |
| "eval_loss": 0.5758941769599915, | |
| "eval_runtime": 1232.4016, | |
| "eval_samples_per_second": 1.731, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 15.36559705418201, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 22.92, | |
| "learning_rate": 2.196666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 23.11, | |
| "learning_rate": 2.168888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 23.3, | |
| "learning_rate": 2.141111111111111e-06, | |
| "loss": 0.0004, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 23.48, | |
| "learning_rate": 2.1133333333333336e-06, | |
| "loss": 0.0004, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 23.67, | |
| "learning_rate": 2.0855555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 23.86, | |
| "learning_rate": 2.057777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 24.05, | |
| "learning_rate": 2.0300000000000005e-06, | |
| "loss": 0.0004, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 24.24, | |
| "learning_rate": 2.0022222222222225e-06, | |
| "loss": 0.0003, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 24.43, | |
| "learning_rate": 1.9744444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 24.62, | |
| "learning_rate": 1.9466666666666665e-06, | |
| "loss": 0.0004, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 24.81, | |
| "learning_rate": 1.918888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 1.8911111111111114e-06, | |
| "loss": 0.0003, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 25.19, | |
| "learning_rate": 1.8633333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 25.38, | |
| "learning_rate": 1.8355555555555557e-06, | |
| "loss": 0.0003, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 25.57, | |
| "learning_rate": 1.807777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 25.76, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 0.0003, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 25.95, | |
| "learning_rate": 1.7522222222222223e-06, | |
| "loss": 0.0003, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 26.14, | |
| "learning_rate": 1.7244444444444448e-06, | |
| "loss": 0.0003, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 26.33, | |
| "learning_rate": 1.6966666666666668e-06, | |
| "loss": 0.0003, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 26.52, | |
| "learning_rate": 1.668888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 26.7, | |
| "learning_rate": 1.6411111111111115e-06, | |
| "loss": 0.0003, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 26.89, | |
| "learning_rate": 1.6133333333333335e-06, | |
| "loss": 0.0003, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 27.08, | |
| "learning_rate": 1.5855555555555557e-06, | |
| "loss": 0.0003, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 27.27, | |
| "learning_rate": 1.5577777777777777e-06, | |
| "loss": 0.0003, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 27.46, | |
| "learning_rate": 1.5300000000000002e-06, | |
| "loss": 0.0003, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 27.65, | |
| "learning_rate": 1.5022222222222224e-06, | |
| "loss": 0.0003, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 27.84, | |
| "learning_rate": 1.4744444444444444e-06, | |
| "loss": 0.0003, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 28.03, | |
| "learning_rate": 1.4466666666666669e-06, | |
| "loss": 0.0003, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 28.22, | |
| "learning_rate": 1.418888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 28.41, | |
| "learning_rate": 1.3911111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 28.6, | |
| "learning_rate": 1.3633333333333336e-06, | |
| "loss": 0.0003, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 28.79, | |
| "learning_rate": 1.3355555555555558e-06, | |
| "loss": 0.0003, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 28.98, | |
| "learning_rate": 1.3077777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 29.17, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.0003, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 29.36, | |
| "learning_rate": 1.2522222222222225e-06, | |
| "loss": 0.0003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 29.55, | |
| "learning_rate": 1.2244444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 29.73, | |
| "learning_rate": 1.1966666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 29.92, | |
| "learning_rate": 1.168888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 30.11, | |
| "learning_rate": 1.1411111111111112e-06, | |
| "loss": 0.0003, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 30.3, | |
| "learning_rate": 1.1133333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 30.3, | |
| "eval_loss": 0.5981439352035522, | |
| "eval_runtime": 1232.471, | |
| "eval_samples_per_second": 1.731, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 15.46554445028932, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 30.49, | |
| "learning_rate": 1.0855555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 30.68, | |
| "learning_rate": 1.0577777777777779e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 30.87, | |
| "learning_rate": 1.03e-06, | |
| "loss": 0.0003, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 31.06, | |
| "learning_rate": 1.0022222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 9.744444444444445e-07, | |
| "loss": 0.0003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 31.44, | |
| "learning_rate": 9.466666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 31.63, | |
| "learning_rate": 9.18888888888889e-07, | |
| "loss": 0.0003, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 31.82, | |
| "learning_rate": 8.911111111111112e-07, | |
| "loss": 0.0002, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 32.01, | |
| "learning_rate": 8.633333333333334e-07, | |
| "loss": 0.0002, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 32.2, | |
| "learning_rate": 8.355555555555556e-07, | |
| "loss": 0.0002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 32.39, | |
| "learning_rate": 8.077777777777779e-07, | |
| "loss": 0.0002, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 32.58, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.0002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 32.77, | |
| "learning_rate": 7.522222222222223e-07, | |
| "loss": 0.0003, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 32.95, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.0002, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 33.14, | |
| "learning_rate": 6.966666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 33.33, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 33.52, | |
| "learning_rate": 6.411111111111111e-07, | |
| "loss": 0.0002, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 33.71, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.0002, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 33.9, | |
| "learning_rate": 5.855555555555555e-07, | |
| "loss": 0.0002, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 34.09, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.0002, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 34.28, | |
| "learning_rate": 5.3e-07, | |
| "loss": 0.0002, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 34.47, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0002, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 34.66, | |
| "learning_rate": 4.7444444444444447e-07, | |
| "loss": 0.0002, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 34.85, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 35.04, | |
| "learning_rate": 4.18888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 35.23, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.0002, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 35.42, | |
| "learning_rate": 3.633333333333334e-07, | |
| "loss": 0.0002, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 35.61, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.0002, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 35.8, | |
| "learning_rate": 3.077777777777778e-07, | |
| "loss": 0.0002, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 35.98, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0002, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 36.17, | |
| "learning_rate": 2.5222222222222225e-07, | |
| "loss": 0.0002, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 36.36, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.0002, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 36.55, | |
| "learning_rate": 1.9666666666666668e-07, | |
| "loss": 0.0002, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 36.74, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.0002, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 36.93, | |
| "learning_rate": 1.411111111111111e-07, | |
| "loss": 0.0002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 37.12, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.0002, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 37.31, | |
| "learning_rate": 8.555555555555557e-08, | |
| "loss": 0.0002, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 37.5, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.0002, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 37.69, | |
| "learning_rate": 3.0000000000000004e-08, | |
| "loss": 0.0002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 37.88, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.0002, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 37.88, | |
| "eval_loss": 0.6072476506233215, | |
| "eval_runtime": 1230.7659, | |
| "eval_samples_per_second": 1.733, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 15.507627564439769, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 37.88, | |
| "step": 5000, | |
| "total_flos": 1.6276811186110464e+20, | |
| "train_loss": 0.0474502604149282, | |
| "train_runtime": 26751.0899, | |
| "train_samples_per_second": 5.981, | |
| "train_steps_per_second": 0.187 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 38, | |
| "total_flos": 1.6276811186110464e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |