| { | |
| "best_metric": 19.76386244886387, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 24.0112, | |
| "eval_steps": 1000, | |
| "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": 4.4e-07, | |
| "loss": 1.789, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.3621, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.1102, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.9283, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.8089, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.6961, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.6281, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.5614, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.5622, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.4946, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.5023, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.4393, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.4225, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.3609, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.3447, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.3013, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.3174, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.293, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.2814, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.265, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.2534, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.2124, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.1995, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.1732, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.178, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.1611, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.1534, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.1499, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.1326, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.1173, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.1069, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.0927, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.0884, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.0878, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.0804, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.0791, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.0673, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.0621, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.0531, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.047, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "eval_loss": 0.338570773601532, | |
| "eval_runtime": 835.5606, | |
| "eval_samples_per_second": 7.888, | |
| "eval_steps_per_second": 0.247, | |
| "eval_wer": 21.74247640649682, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.0427, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.0433, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.0386, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.0398, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.0336, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.0307, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.0272, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.0245, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.0202, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.022, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.0215, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.0193, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.0173, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.0162, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.0146, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.0131, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.0116, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.0128, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.012, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.0126, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.0109, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.0108, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.0093, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.0088, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.0069, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.0079, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.0075, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.0072, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.007, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.0064, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.0067, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.0066, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.0057, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.005, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.0051, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.005, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.0047, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.004, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.0044, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.0043, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "eval_loss": 0.3903014361858368, | |
| "eval_runtime": 823.8638, | |
| "eval_samples_per_second": 8.0, | |
| "eval_steps_per_second": 0.25, | |
| "eval_wer": 20.177001903681802, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.004, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.0043, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.0035, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.0033, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.0031, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.0031, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.003, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.0026, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.0023, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.0024, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.0022, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.0022, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.0022, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.0021, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.0022, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.0018, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.0017, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.0018, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.0017, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.0017, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.0017, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.0019, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.0019, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.0015, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.0014, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.0014, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.0014, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.0014, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.0013, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.0015, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.0013, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.0013, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.0012, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.0012, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.0012, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.0012, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.0012, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.0012, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.0012, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.0011, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "eval_loss": 0.4209349751472473, | |
| "eval_runtime": 823.6051, | |
| "eval_samples_per_second": 8.003, | |
| "eval_steps_per_second": 0.25, | |
| "eval_wer": 19.78006399611163, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.0011, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.001, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.001, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.0011, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.001, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.0011, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.001, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.001, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.001, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.0009, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.0009, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.001, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.0009, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.001, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.0009, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.0009, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.0009, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.0008, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.0008, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.0009, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.0008, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.0009, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.0008, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.0008, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.0008, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.0008, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.0008, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.0008, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.0008, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.0008, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.0008, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.0008, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.0007, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.0007, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.0007, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.0007, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.0007, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.0008, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.0008, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "eval_loss": 0.43805626034736633, | |
| "eval_runtime": 809.976, | |
| "eval_samples_per_second": 8.137, | |
| "eval_steps_per_second": 0.254, | |
| "eval_wer": 20.017011624610152, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.0007, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.0007, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.0007, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.0007, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.0007, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.0007, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.0007, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.0007, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.0007, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.0007, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.0007, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.0006, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.0006, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.0007, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.0006, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.0007, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.0007, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.0006, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 21.03, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.0006, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 21.04, | |
| "learning_rate": 1.0622222222222222e-06, | |
| "loss": 0.0006, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 1.0066666666666668e-06, | |
| "loss": 0.0006, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.0006, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.0007, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.0006, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 0.0006, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 22.03, | |
| "learning_rate": 7.28888888888889e-07, | |
| "loss": 0.0006, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 22.03, | |
| "learning_rate": 6.733333333333334e-07, | |
| "loss": 0.0006, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 22.04, | |
| "learning_rate": 6.177777777777778e-07, | |
| "loss": 0.0006, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 5.622222222222223e-07, | |
| "loss": 0.0006, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.0006, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.0006, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 3.9555555555555557e-07, | |
| "loss": 0.0006, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.0006, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 2.844444444444445e-07, | |
| "loss": 0.0006, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 23.03, | |
| "learning_rate": 2.2888888888888892e-07, | |
| "loss": 0.0006, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 23.04, | |
| "learning_rate": 1.7333333333333335e-07, | |
| "loss": 0.0006, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 1.1777777777777778e-07, | |
| "loss": 0.0006, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 6.222222222222223e-08, | |
| "loss": 0.0006, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "learning_rate": 6.666666666666667e-09, | |
| "loss": 0.0006, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "eval_loss": 0.4463981091976166, | |
| "eval_runtime": 822.0952, | |
| "eval_samples_per_second": 8.017, | |
| "eval_steps_per_second": 0.251, | |
| "eval_wer": 19.76386244886387, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 24.01, | |
| "step": 5000, | |
| "total_flos": 9.233347600318464e+19, | |
| "train_loss": 0.07856089599095285, | |
| "train_runtime": 39537.7318, | |
| "train_samples_per_second": 8.094, | |
| "train_steps_per_second": 0.126 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 9.233347600318464e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |