| { | |
| "best_metric": 33.995690687096, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 21.0088, | |
| "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": 4.6000000000000004e-07, | |
| "loss": 1.867, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.5157, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.46e-06, | |
| "loss": 1.2678, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 1.008, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.8678, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.7157, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.6309, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.5991, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.6767, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.7499, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.6989, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.6123, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.5472, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.4821, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.4683, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.4069, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.3646, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.4116, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.4857, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.5278, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.94888888888889e-06, | |
| "loss": 0.4212, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.893333333333334e-06, | |
| "loss": 0.3934, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.837777777777778e-06, | |
| "loss": 0.3402, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.782222222222222e-06, | |
| "loss": 0.3308, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.726666666666668e-06, | |
| "loss": 0.2823, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.671111111111112e-06, | |
| "loss": 0.2477, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.615555555555558e-06, | |
| "loss": 0.2467, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.56e-06, | |
| "loss": 0.3069, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.504444444444446e-06, | |
| "loss": 0.3619, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.44888888888889e-06, | |
| "loss": 0.315, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.393333333333334e-06, | |
| "loss": 0.2796, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.33777777777778e-06, | |
| "loss": 0.2473, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.282222222222222e-06, | |
| "loss": 0.2243, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.226666666666668e-06, | |
| "loss": 0.2061, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.171111111111112e-06, | |
| "loss": 0.1796, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.115555555555556e-06, | |
| "loss": 0.1591, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.1936, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.004444444444445e-06, | |
| "loss": 0.2415, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.94888888888889e-06, | |
| "loss": 0.2493, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.893333333333333e-06, | |
| "loss": 0.206, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "eval_loss": 0.4355769455432892, | |
| "eval_runtime": 2396.1457, | |
| "eval_samples_per_second": 3.219, | |
| "eval_steps_per_second": 0.101, | |
| "eval_wer": 36.244268061361666, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.83777777777778e-06, | |
| "loss": 0.1852, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.782222222222223e-06, | |
| "loss": 0.1598, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.726666666666667e-06, | |
| "loss": 0.1505, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.671111111111113e-06, | |
| "loss": 0.1267, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.615555555555555e-06, | |
| "loss": 0.1149, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 0.1204, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 8.504444444444445e-06, | |
| "loss": 0.153, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 8.448888888888889e-06, | |
| "loss": 0.1857, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.393333333333335e-06, | |
| "loss": 0.1538, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.337777777777777e-06, | |
| "loss": 0.1386, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 8.282222222222223e-06, | |
| "loss": 0.1193, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 8.226666666666667e-06, | |
| "loss": 0.1103, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 8.171111111111113e-06, | |
| "loss": 0.0939, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.083, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.0746, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.0962, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.1232, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.1268, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.1002, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.0879, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.0764, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.0716, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.058, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 0.0507, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.504444444444445e-06, | |
| "loss": 0.0564, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 7.44888888888889e-06, | |
| "loss": 0.0745, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.393333333333333e-06, | |
| "loss": 0.0971, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.337777777777778e-06, | |
| "loss": 0.0731, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.282222222222222e-06, | |
| "loss": 0.0651, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.226666666666667e-06, | |
| "loss": 0.0553, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.171111111111112e-06, | |
| "loss": 0.0496, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 7.115555555555557e-06, | |
| "loss": 0.0404, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.0361, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 7.004444444444445e-06, | |
| "loss": 0.0342, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 6.948888888888889e-06, | |
| "loss": 0.0461, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.0595, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.837777777777779e-06, | |
| "loss": 0.0597, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.782222222222222e-06, | |
| "loss": 0.0453, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.726666666666667e-06, | |
| "loss": 0.0405, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.671111111111112e-06, | |
| "loss": 0.0332, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "eval_loss": 0.4582775831222534, | |
| "eval_runtime": 2393.6386, | |
| "eval_samples_per_second": 3.223, | |
| "eval_steps_per_second": 0.101, | |
| "eval_wer": 34.05093828842931, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.615555555555556e-06, | |
| "loss": 0.0301, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.0247, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 6.504444444444446e-06, | |
| "loss": 0.0219, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 6.448888888888889e-06, | |
| "loss": 0.0252, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 6.393333333333334e-06, | |
| "loss": 0.0348, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 6.3377777777777786e-06, | |
| "loss": 0.0455, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 6.282222222222223e-06, | |
| "loss": 0.0324, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 6.2266666666666675e-06, | |
| "loss": 0.028, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 6.171111111111112e-06, | |
| "loss": 0.0242, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.0211, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.0178, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.0158, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 5.948888888888889e-06, | |
| "loss": 0.0155, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 5.893333333333334e-06, | |
| "loss": 0.0206, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.837777777777777e-06, | |
| "loss": 0.027, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.782222222222222e-06, | |
| "loss": 0.025, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.726666666666667e-06, | |
| "loss": 0.0194, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.671111111111112e-06, | |
| "loss": 0.0174, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.615555555555556e-06, | |
| "loss": 0.015, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 0.0139, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.504444444444444e-06, | |
| "loss": 0.0119, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 5.448888888888889e-06, | |
| "loss": 0.0112, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 5.393333333333334e-06, | |
| "loss": 0.013, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 5.337777777777779e-06, | |
| "loss": 0.0158, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 5.282222222222223e-06, | |
| "loss": 0.0192, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 5.226666666666667e-06, | |
| "loss": 0.0144, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.171111111111111e-06, | |
| "loss": 0.013, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.115555555555556e-06, | |
| "loss": 0.0114, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.011, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 5.004444444444445e-06, | |
| "loss": 0.0096, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.94888888888889e-06, | |
| "loss": 0.0089, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.893333333333334e-06, | |
| "loss": 0.0091, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 4.837777777777778e-06, | |
| "loss": 0.011, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.7822222222222226e-06, | |
| "loss": 0.0134, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.7266666666666674e-06, | |
| "loss": 0.012, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.6711111111111115e-06, | |
| "loss": 0.0104, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.6155555555555555e-06, | |
| "loss": 0.0094, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.56e-06, | |
| "loss": 0.0089, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.504444444444444e-06, | |
| "loss": 0.0082, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.448888888888889e-06, | |
| "loss": 0.0074, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "eval_loss": 0.5119307637214661, | |
| "eval_runtime": 2393.0367, | |
| "eval_samples_per_second": 3.224, | |
| "eval_steps_per_second": 0.101, | |
| "eval_wer": 34.43951308447358, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 4.393333333333334e-06, | |
| "loss": 0.0073, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 4.337777777777778e-06, | |
| "loss": 0.0083, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 4.282222222222222e-06, | |
| "loss": 0.0096, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 4.226666666666667e-06, | |
| "loss": 0.0102, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 4.171111111111111e-06, | |
| "loss": 0.0086, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 4.115555555555556e-06, | |
| "loss": 0.008, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.0075, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 4.004444444444445e-06, | |
| "loss": 0.0073, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.948888888888889e-06, | |
| "loss": 0.0064, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.893333333333333e-06, | |
| "loss": 0.0062, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.837777777777778e-06, | |
| "loss": 0.0064, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 3.782222222222223e-06, | |
| "loss": 0.0077, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 3.726666666666667e-06, | |
| "loss": 0.0085, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.6711111111111113e-06, | |
| "loss": 0.0075, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.615555555555556e-06, | |
| "loss": 0.007, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 0.0065, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.5044444444444447e-06, | |
| "loss": 0.0063, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 3.4488888888888896e-06, | |
| "loss": 0.0058, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 3.3933333333333336e-06, | |
| "loss": 0.0055, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.337777777777778e-06, | |
| "loss": 0.0054, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.282222222222223e-06, | |
| "loss": 0.0061, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 3.226666666666667e-06, | |
| "loss": 0.0071, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 3.1711111111111114e-06, | |
| "loss": 0.007, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 3.1155555555555555e-06, | |
| "loss": 0.0063, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.0059, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 3.004444444444445e-06, | |
| "loss": 0.0056, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.948888888888889e-06, | |
| "loss": 0.0054, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.8933333333333337e-06, | |
| "loss": 0.0049, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.837777777777778e-06, | |
| "loss": 0.0047, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.7822222222222222e-06, | |
| "loss": 0.0051, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 2.726666666666667e-06, | |
| "loss": 0.0058, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.6711111111111116e-06, | |
| "loss": 0.0065, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.6155555555555556e-06, | |
| "loss": 0.0058, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.0054, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.504444444444445e-06, | |
| "loss": 0.0051, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.448888888888889e-06, | |
| "loss": 0.005, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 2.3933333333333334e-06, | |
| "loss": 0.0046, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 2.337777777777778e-06, | |
| "loss": 0.0044, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.2822222222222223e-06, | |
| "loss": 0.0045, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.226666666666667e-06, | |
| "loss": 0.005, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "eval_loss": 0.5394303202629089, | |
| "eval_runtime": 2392.476, | |
| "eval_samples_per_second": 3.224, | |
| "eval_steps_per_second": 0.101, | |
| "eval_wer": 33.995690687096, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 2.1711111111111113e-06, | |
| "loss": 0.0056, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 2.1155555555555557e-06, | |
| "loss": 0.0056, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 2.06e-06, | |
| "loss": 0.005, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 2.0044444444444446e-06, | |
| "loss": 0.0048, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 1.948888888888889e-06, | |
| "loss": 0.0045, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.8933333333333333e-06, | |
| "loss": 0.0045, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.837777777777778e-06, | |
| "loss": 0.0041, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.7822222222222225e-06, | |
| "loss": 0.004, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.7266666666666667e-06, | |
| "loss": 0.0043, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 1.6711111111111112e-06, | |
| "loss": 0.0049, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.6155555555555559e-06, | |
| "loss": 0.0054, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.56e-06, | |
| "loss": 0.0047, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.5044444444444446e-06, | |
| "loss": 0.0045, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.4488888888888892e-06, | |
| "loss": 0.0043, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 1.3933333333333335e-06, | |
| "loss": 0.0042, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 1.337777777777778e-06, | |
| "loss": 0.0039, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 1.2822222222222222e-06, | |
| "loss": 0.0038, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 1.2266666666666666e-06, | |
| "loss": 0.0039, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 1.171111111111111e-06, | |
| "loss": 0.0043, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 1.1155555555555558e-06, | |
| "loss": 0.005, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 1.06e-06, | |
| "loss": 0.0047, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 1.0044444444444445e-06, | |
| "loss": 0.0044, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 9.488888888888889e-07, | |
| "loss": 0.0041, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 8.933333333333334e-07, | |
| "loss": 0.004, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 8.37777777777778e-07, | |
| "loss": 0.0039, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 7.822222222222223e-07, | |
| "loss": 0.0037, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 7.266666666666668e-07, | |
| "loss": 0.0036, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 6.711111111111111e-07, | |
| "loss": 0.004, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 6.155555555555556e-07, | |
| "loss": 0.0045, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 5.6e-07, | |
| "loss": 0.0048, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 5.044444444444445e-07, | |
| "loss": 0.0043, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 4.488888888888889e-07, | |
| "loss": 0.0041, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.9333333333333336e-07, | |
| "loss": 0.0039, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 3.3777777777777777e-07, | |
| "loss": 0.004, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 2.822222222222222e-07, | |
| "loss": 0.0036, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 2.266666666666667e-07, | |
| "loss": 0.0035, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 1.7111111111111114e-07, | |
| "loss": 0.0037, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 20.05, | |
| "learning_rate": 1.1555555555555556e-07, | |
| "loss": 0.0041, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 0.0047, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 4.444444444444445e-09, | |
| "loss": 0.0045, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "eval_loss": 0.5461024045944214, | |
| "eval_runtime": 2375.6312, | |
| "eval_samples_per_second": 3.247, | |
| "eval_steps_per_second": 0.102, | |
| "eval_wer": 34.102502716340396, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "step": 5000, | |
| "total_flos": 2.073068169265152e+19, | |
| "train_loss": 0.12548666781038045, | |
| "train_runtime": 85032.7636, | |
| "train_samples_per_second": 3.763, | |
| "train_steps_per_second": 0.059 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 2.073068169265152e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |