| { | |
| "best_metric": 11.408629675328264, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 21.1864406779661, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 0.9539, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 0.6773, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.3302, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.2437, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.2103, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.1938, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.1946, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.980000000000001e-06, | |
| "loss": 0.1788, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.48e-06, | |
| "loss": 0.1669, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.980000000000001e-06, | |
| "loss": 0.1427, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.480000000000001e-06, | |
| "loss": 0.118, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.98e-06, | |
| "loss": 0.1184, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.480000000000001e-06, | |
| "loss": 0.1103, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.98e-06, | |
| "loss": 0.1143, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 7.48e-06, | |
| "loss": 0.1102, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.980000000000002e-06, | |
| "loss": 0.116, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.48e-06, | |
| "loss": 0.103, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.98e-06, | |
| "loss": 0.1067, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.48e-06, | |
| "loss": 0.1014, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.980000000000001e-06, | |
| "loss": 0.045, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 9.946666666666667e-06, | |
| "loss": 0.0502, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 9.891111111111113e-06, | |
| "loss": 0.051, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.835555555555556e-06, | |
| "loss": 0.0467, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 0.0495, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 9.724444444444445e-06, | |
| "loss": 0.0465, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 9.66888888888889e-06, | |
| "loss": 0.0489, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.613333333333335e-06, | |
| "loss": 0.0525, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 9.557777777777777e-06, | |
| "loss": 0.0506, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.502222222222223e-06, | |
| "loss": 0.0278, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 9.446666666666667e-06, | |
| "loss": 0.0201, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 9.391111111111111e-06, | |
| "loss": 0.0201, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 9.335555555555557e-06, | |
| "loss": 0.0203, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 9.280000000000001e-06, | |
| "loss": 0.021, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 9.224444444444445e-06, | |
| "loss": 0.0214, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 9.168888888888889e-06, | |
| "loss": 0.0239, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 9.113333333333335e-06, | |
| "loss": 0.0223, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 9.057777777777779e-06, | |
| "loss": 0.0243, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.002222222222223e-06, | |
| "loss": 0.0199, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 8.946666666666669e-06, | |
| "loss": 0.0097, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 8.891111111111111e-06, | |
| "loss": 0.0105, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "eval_loss": 0.1972593367099762, | |
| "eval_runtime": 8895.0887, | |
| "eval_samples_per_second": 0.867, | |
| "eval_steps_per_second": 0.027, | |
| "eval_wer": 12.613027384394393, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 8.835555555555557e-06, | |
| "loss": 0.0112, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 8.78e-06, | |
| "loss": 0.0108, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 8.724444444444445e-06, | |
| "loss": 0.0116, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 8.66888888888889e-06, | |
| "loss": 0.0114, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 8.613333333333333e-06, | |
| "loss": 0.011, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 8.557777777777778e-06, | |
| "loss": 0.0096, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 8.502222222222223e-06, | |
| "loss": 0.0126, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 8.446666666666668e-06, | |
| "loss": 0.0078, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 8.391111111111112e-06, | |
| "loss": 0.006, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 8.335555555555556e-06, | |
| "loss": 0.0053, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 8.28e-06, | |
| "loss": 0.0067, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 8.224444444444444e-06, | |
| "loss": 0.0055, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 8.16888888888889e-06, | |
| "loss": 0.0059, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 8.113333333333334e-06, | |
| "loss": 0.0061, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 8.057777777777778e-06, | |
| "loss": 0.0057, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 8.002222222222222e-06, | |
| "loss": 0.0071, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.946666666666666e-06, | |
| "loss": 0.0061, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 7.891111111111112e-06, | |
| "loss": 0.0047, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 7.835555555555556e-06, | |
| "loss": 0.004, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 7.78e-06, | |
| "loss": 0.005, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 7.724444444444446e-06, | |
| "loss": 0.0049, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 7.66888888888889e-06, | |
| "loss": 0.0036, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 7.613333333333334e-06, | |
| "loss": 0.0035, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 7.557777777777779e-06, | |
| "loss": 0.0046, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 7.502222222222223e-06, | |
| "loss": 0.0045, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 7.446666666666668e-06, | |
| "loss": 0.0044, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 7.3911111111111125e-06, | |
| "loss": 0.0022, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 7.335555555555556e-06, | |
| "loss": 0.0019, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 0.0019, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 7.224444444444445e-06, | |
| "loss": 0.0024, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 7.1688888888888895e-06, | |
| "loss": 0.0022, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 7.113333333333334e-06, | |
| "loss": 0.0029, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 7.057777777777778e-06, | |
| "loss": 0.0026, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 7.0022222222222225e-06, | |
| "loss": 0.0033, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 6.946666666666667e-06, | |
| "loss": 0.0026, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 6.891111111111111e-06, | |
| "loss": 0.0025, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.16, | |
| "learning_rate": 6.835555555555556e-06, | |
| "loss": 0.0019, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 0.0022, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 6.724444444444444e-06, | |
| "loss": 0.0015, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 6.668888888888889e-06, | |
| "loss": 0.0016, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "eval_loss": 0.2197847217321396, | |
| "eval_runtime": 9649.3249, | |
| "eval_samples_per_second": 0.799, | |
| "eval_steps_per_second": 0.025, | |
| "eval_wer": 11.898491740483601, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 6.613333333333334e-06, | |
| "loss": 0.0023, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 6.557777777777778e-06, | |
| "loss": 0.0022, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 6.502222222222223e-06, | |
| "loss": 0.002, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 6.446666666666668e-06, | |
| "loss": 0.0019, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 6.391111111111111e-06, | |
| "loss": 0.0019, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 6.335555555555556e-06, | |
| "loss": 0.0014, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 6.280000000000001e-06, | |
| "loss": 0.002, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 6.224444444444445e-06, | |
| "loss": 0.0014, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 6.16888888888889e-06, | |
| "loss": 0.0018, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 6.113333333333333e-06, | |
| "loss": 0.0018, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 6.057777777777778e-06, | |
| "loss": 0.0016, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 6.002222222222223e-06, | |
| "loss": 0.0017, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 5.946666666666668e-06, | |
| "loss": 0.0014, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 5.891111111111112e-06, | |
| "loss": 0.0018, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 5.8355555555555565e-06, | |
| "loss": 0.0014, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "learning_rate": 5.78e-06, | |
| "loss": 0.0011, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 5.724444444444445e-06, | |
| "loss": 0.0008, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 5.6688888888888895e-06, | |
| "loss": 0.0009, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 10.49, | |
| "learning_rate": 5.613333333333334e-06, | |
| "loss": 0.0008, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 5.557777777777778e-06, | |
| "loss": 0.0008, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.7, | |
| "learning_rate": 5.5022222222222224e-06, | |
| "loss": 0.0008, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 5.4466666666666665e-06, | |
| "loss": 0.0012, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 5.391111111111111e-06, | |
| "loss": 0.0006, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.335555555555556e-06, | |
| "loss": 0.0006, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 5.28e-06, | |
| "loss": 0.0008, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 5.224444444444445e-06, | |
| "loss": 0.0005, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 5.168888888888889e-06, | |
| "loss": 0.0005, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 5.113333333333333e-06, | |
| "loss": 0.0005, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 5.057777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 5.002222222222223e-06, | |
| "loss": 0.0005, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 4.946666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.86, | |
| "learning_rate": 4.891111111111111e-06, | |
| "loss": 0.0006, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 11.97, | |
| "learning_rate": 4.835555555555556e-06, | |
| "loss": 0.0005, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 12.08, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.0004, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 12.18, | |
| "learning_rate": 4.724444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 12.29, | |
| "learning_rate": 4.66888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 12.39, | |
| "learning_rate": 4.613333333333334e-06, | |
| "loss": 0.0004, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 4.557777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 12.61, | |
| "learning_rate": 4.502222222222223e-06, | |
| "loss": 0.0004, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 4.446666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "eval_loss": 0.230974942445755, | |
| "eval_runtime": 9689.0706, | |
| "eval_samples_per_second": 0.796, | |
| "eval_steps_per_second": 0.025, | |
| "eval_wer": 11.454669343106021, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.82, | |
| "learning_rate": 4.391111111111112e-06, | |
| "loss": 0.0005, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 12.92, | |
| "learning_rate": 4.3355555555555565e-06, | |
| "loss": 0.0004, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.0004, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 13.14, | |
| "learning_rate": 4.2244444444444446e-06, | |
| "loss": 0.0004, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "learning_rate": 4.168888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 4.1133333333333335e-06, | |
| "loss": 0.0003, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "learning_rate": 4.057777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 4.002222222222222e-06, | |
| "loss": 0.0004, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 13.67, | |
| "learning_rate": 3.946666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 13.77, | |
| "learning_rate": 3.891111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "learning_rate": 3.835555555555555e-06, | |
| "loss": 0.0003, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 13.98, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.0003, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "learning_rate": 3.724444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 3.668888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "learning_rate": 3.6133333333333336e-06, | |
| "loss": 0.0003, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "learning_rate": 3.5577777777777785e-06, | |
| "loss": 0.0003, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 14.51, | |
| "learning_rate": 3.5022222222222225e-06, | |
| "loss": 0.0003, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 3.446666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 14.72, | |
| "learning_rate": 3.391111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "learning_rate": 3.335555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.0003, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 3.2244444444444444e-06, | |
| "loss": 0.0003, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "learning_rate": 3.1688888888888893e-06, | |
| "loss": 0.0003, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 15.25, | |
| "learning_rate": 3.1133333333333337e-06, | |
| "loss": 0.0003, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 15.36, | |
| "learning_rate": 3.0577777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 15.47, | |
| "learning_rate": 3.0022222222222227e-06, | |
| "loss": 0.0003, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 15.57, | |
| "learning_rate": 2.946666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 15.68, | |
| "learning_rate": 2.891111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 15.78, | |
| "learning_rate": 2.835555555555556e-06, | |
| "loss": 0.0003, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0003, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.7244444444444445e-06, | |
| "loss": 0.0003, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 16.1, | |
| "learning_rate": 2.6688888888888894e-06, | |
| "loss": 0.0003, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 16.21, | |
| "learning_rate": 2.6133333333333334e-06, | |
| "loss": 0.0003, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 16.31, | |
| "learning_rate": 2.557777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 2.5022222222222224e-06, | |
| "loss": 0.0003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 16.53, | |
| "learning_rate": 2.446666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 16.63, | |
| "learning_rate": 2.3911111111111113e-06, | |
| "loss": 0.0002, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 16.74, | |
| "learning_rate": 2.3355555555555557e-06, | |
| "loss": 0.0002, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 16.84, | |
| "learning_rate": 2.28e-06, | |
| "loss": 0.0003, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "learning_rate": 2.2244444444444447e-06, | |
| "loss": 0.0003, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "eval_loss": 0.23803874850273132, | |
| "eval_runtime": 8967.4245, | |
| "eval_samples_per_second": 0.86, | |
| "eval_steps_per_second": 0.027, | |
| "eval_wer": 11.427045542439366, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.06, | |
| "learning_rate": 2.168888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 17.16, | |
| "learning_rate": 2.1133333333333336e-06, | |
| "loss": 0.0002, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 17.27, | |
| "learning_rate": 2.057777777777778e-06, | |
| "loss": 0.0002, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "learning_rate": 2.0022222222222225e-06, | |
| "loss": 0.0002, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 1.9466666666666665e-06, | |
| "loss": 0.0002, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "learning_rate": 1.8911111111111114e-06, | |
| "loss": 0.0002, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 1.8355555555555557e-06, | |
| "loss": 0.0002, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 0.0002, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 17.9, | |
| "learning_rate": 1.7244444444444448e-06, | |
| "loss": 0.0002, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.668888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 18.11, | |
| "learning_rate": 1.6133333333333335e-06, | |
| "loss": 0.0002, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 1.5577777777777777e-06, | |
| "loss": 0.0002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 18.33, | |
| "learning_rate": 1.5022222222222224e-06, | |
| "loss": 0.0002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 18.43, | |
| "learning_rate": 1.4466666666666669e-06, | |
| "loss": 0.0002, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 18.54, | |
| "learning_rate": 1.3911111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 18.64, | |
| "learning_rate": 1.3355555555555558e-06, | |
| "loss": 0.0002, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.0002, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "learning_rate": 1.2244444444444445e-06, | |
| "loss": 0.0002, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "learning_rate": 1.168888888888889e-06, | |
| "loss": 0.0002, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 19.07, | |
| "learning_rate": 1.1133333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 19.17, | |
| "learning_rate": 1.0577777777777779e-06, | |
| "loss": 0.0002, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 19.28, | |
| "learning_rate": 1.0022222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 19.39, | |
| "learning_rate": 9.466666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 19.49, | |
| "learning_rate": 8.911111111111112e-07, | |
| "loss": 0.0002, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 19.6, | |
| "learning_rate": 8.355555555555556e-07, | |
| "loss": 0.0002, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 19.7, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.0002, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 19.81, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.0002, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 19.92, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.0002, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 20.13, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.0002, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 20.23, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0002, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 20.34, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0002, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 20.44, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.0002, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 20.55, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.0002, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 20.66, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 20.76, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.0002, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 20.87, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.0002, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 20.97, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.0002, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 21.08, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.0002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.0002, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "eval_loss": 0.24174600839614868, | |
| "eval_runtime": 9721.5421, | |
| "eval_samples_per_second": 0.793, | |
| "eval_steps_per_second": 0.025, | |
| "eval_wer": 11.408629675328264, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "step": 5000, | |
| "total_flos": 3.2620778175135744e+20, | |
| "train_loss": 0.026235109101515264, | |
| "train_runtime": 142520.4283, | |
| "train_samples_per_second": 2.245, | |
| "train_steps_per_second": 0.035 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 22, | |
| "total_flos": 3.2620778175135744e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |