|
{ |
|
"best_metric": 62.69218363616815, |
|
"best_model_checkpoint": "./checkpoint-5000", |
|
"epoch": 31.0204, |
|
"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": 4.0000000000000003e-07, |
|
"loss": 10.8663, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.000000000000001e-07, |
|
"loss": 10.2008, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.4000000000000001e-06, |
|
"loss": 9.1731, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.9000000000000002e-06, |
|
"loss": 8.0232, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 6.7324, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.9e-06, |
|
"loss": 6.1548, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3.4000000000000005e-06, |
|
"loss": 5.2367, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 3.900000000000001e-06, |
|
"loss": 4.7892, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.4e-06, |
|
"loss": 4.8815, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.9000000000000005e-06, |
|
"loss": 4.484, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 5.400000000000001e-06, |
|
"loss": 4.2984, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 5.9e-06, |
|
"loss": 4.4209, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 6.4000000000000006e-06, |
|
"loss": 4.1719, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 6.9e-06, |
|
"loss": 3.8703, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 7.4e-06, |
|
"loss": 4.0647, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 7.9e-06, |
|
"loss": 3.9076, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 8.400000000000001e-06, |
|
"loss": 3.7608, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 8.900000000000001e-06, |
|
"loss": 3.6414, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 9.4e-06, |
|
"loss": 3.7097, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"learning_rate": 9.9e-06, |
|
"loss": 3.3424, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"learning_rate": 9.955555555555556e-06, |
|
"loss": 3.4226, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 3.02, |
|
"learning_rate": 9.9e-06, |
|
"loss": 3.5549, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 3.02, |
|
"learning_rate": 9.844444444444446e-06, |
|
"loss": 3.2538, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 9.78888888888889e-06, |
|
"loss": 3.0186, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 9.733333333333334e-06, |
|
"loss": 3.3356, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 9.677777777777778e-06, |
|
"loss": 2.9369, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 4.01, |
|
"learning_rate": 9.622222222222222e-06, |
|
"loss": 2.8598, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 4.01, |
|
"learning_rate": 9.566666666666668e-06, |
|
"loss": 3.123, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 9.511111111111112e-06, |
|
"loss": 2.7978, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 9.455555555555557e-06, |
|
"loss": 2.6693, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 4.03, |
|
"learning_rate": 9.4e-06, |
|
"loss": 2.8226, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 9.344444444444446e-06, |
|
"loss": 2.5958, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 9.28888888888889e-06, |
|
"loss": 2.3949, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 9.233333333333334e-06, |
|
"loss": 2.5876, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 5.02, |
|
"learning_rate": 9.17777777777778e-06, |
|
"loss": 2.4048, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 5.02, |
|
"learning_rate": 9.122222222222223e-06, |
|
"loss": 2.3453, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 5.03, |
|
"learning_rate": 9.066666666666667e-06, |
|
"loss": 2.2082, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"learning_rate": 9.011111111111111e-06, |
|
"loss": 2.2576, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"learning_rate": 8.955555555555555e-06, |
|
"loss": 1.9449, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"learning_rate": 8.900000000000001e-06, |
|
"loss": 2.0512, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"eval_loss": 2.5996744632720947, |
|
"eval_runtime": 1025.6953, |
|
"eval_samples_per_second": 4.942, |
|
"eval_steps_per_second": 0.155, |
|
"eval_wer": 87.19492527685195, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 6.02, |
|
"learning_rate": 8.844444444444445e-06, |
|
"loss": 2.1762, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 6.02, |
|
"learning_rate": 8.788888888888891e-06, |
|
"loss": 1.8816, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 6.03, |
|
"learning_rate": 8.733333333333333e-06, |
|
"loss": 1.7013, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 6.03, |
|
"learning_rate": 8.677777777777779e-06, |
|
"loss": 1.9135, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"learning_rate": 8.622222222222223e-06, |
|
"loss": 1.6226, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 7.01, |
|
"learning_rate": 8.566666666666667e-06, |
|
"loss": 1.5529, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 7.01, |
|
"learning_rate": 8.511111111111113e-06, |
|
"loss": 1.7884, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 7.02, |
|
"learning_rate": 8.455555555555555e-06, |
|
"loss": 1.5288, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 7.02, |
|
"learning_rate": 8.400000000000001e-06, |
|
"loss": 1.3715, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 7.03, |
|
"learning_rate": 8.344444444444445e-06, |
|
"loss": 1.5344, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"learning_rate": 8.288888888888889e-06, |
|
"loss": 1.3419, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 8.01, |
|
"learning_rate": 8.233333333333335e-06, |
|
"loss": 1.1691, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 8.01, |
|
"learning_rate": 8.177777777777779e-06, |
|
"loss": 1.3838, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 8.02, |
|
"learning_rate": 8.122222222222223e-06, |
|
"loss": 1.2271, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 8.02, |
|
"learning_rate": 8.066666666666667e-06, |
|
"loss": 1.1797, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 8.03, |
|
"learning_rate": 8.011111111111113e-06, |
|
"loss": 1.1166, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"learning_rate": 7.955555555555557e-06, |
|
"loss": 1.1374, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 9.01, |
|
"learning_rate": 7.9e-06, |
|
"loss": 0.8853, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 9.01, |
|
"learning_rate": 7.844444444444446e-06, |
|
"loss": 1.011, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 9.02, |
|
"learning_rate": 7.788888888888889e-06, |
|
"loss": 1.0932, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 9.02, |
|
"learning_rate": 7.733333333333334e-06, |
|
"loss": 0.9271, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 9.03, |
|
"learning_rate": 7.677777777777778e-06, |
|
"loss": 0.7826, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 9.03, |
|
"learning_rate": 7.622222222222223e-06, |
|
"loss": 0.9417, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"learning_rate": 7.566666666666667e-06, |
|
"loss": 0.7138, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 10.01, |
|
"learning_rate": 7.511111111111111e-06, |
|
"loss": 0.6907, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 10.01, |
|
"learning_rate": 7.455555555555556e-06, |
|
"loss": 0.8956, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 10.02, |
|
"learning_rate": 7.4e-06, |
|
"loss": 0.7546, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 10.02, |
|
"learning_rate": 7.344444444444445e-06, |
|
"loss": 0.6118, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 10.03, |
|
"learning_rate": 7.28888888888889e-06, |
|
"loss": 0.7483, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 11.0, |
|
"learning_rate": 7.233333333333334e-06, |
|
"loss": 0.5868, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 11.01, |
|
"learning_rate": 7.177777777777778e-06, |
|
"loss": 0.4992, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 11.01, |
|
"learning_rate": 7.122222222222222e-06, |
|
"loss": 0.678, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 11.02, |
|
"learning_rate": 7.066666666666667e-06, |
|
"loss": 0.5908, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 11.02, |
|
"learning_rate": 7.011111111111112e-06, |
|
"loss": 0.5346, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 11.03, |
|
"learning_rate": 6.955555555555557e-06, |
|
"loss": 0.5069, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 12.0, |
|
"learning_rate": 6.9e-06, |
|
"loss": 0.5083, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 12.01, |
|
"learning_rate": 6.844444444444445e-06, |
|
"loss": 0.3572, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 12.01, |
|
"learning_rate": 6.788888888888889e-06, |
|
"loss": 0.479, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 12.02, |
|
"learning_rate": 6.733333333333334e-06, |
|
"loss": 0.5096, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 12.02, |
|
"learning_rate": 6.677777777777779e-06, |
|
"loss": 0.4367, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 12.02, |
|
"eval_loss": 1.8089243173599243, |
|
"eval_runtime": 1029.0604, |
|
"eval_samples_per_second": 4.926, |
|
"eval_steps_per_second": 0.155, |
|
"eval_wer": 68.12708310934308, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 12.03, |
|
"learning_rate": 6.6222222222222236e-06, |
|
"loss": 0.3293, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 12.03, |
|
"learning_rate": 6.566666666666667e-06, |
|
"loss": 0.4292, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 13.0, |
|
"learning_rate": 6.511111111111112e-06, |
|
"loss": 0.2875, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 13.01, |
|
"learning_rate": 6.455555555555556e-06, |
|
"loss": 0.2965, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 13.01, |
|
"learning_rate": 6.4000000000000006e-06, |
|
"loss": 0.4354, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 13.02, |
|
"learning_rate": 6.3444444444444454e-06, |
|
"loss": 0.3249, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 13.02, |
|
"learning_rate": 6.28888888888889e-06, |
|
"loss": 0.2625, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 13.03, |
|
"learning_rate": 6.2333333333333335e-06, |
|
"loss": 0.3368, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 14.0, |
|
"learning_rate": 6.177777777777778e-06, |
|
"loss": 0.2394, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 14.01, |
|
"learning_rate": 6.1222222222222224e-06, |
|
"loss": 0.1974, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 14.01, |
|
"learning_rate": 6.066666666666667e-06, |
|
"loss": 0.3194, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 14.02, |
|
"learning_rate": 6.011111111111112e-06, |
|
"loss": 0.2597, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 14.02, |
|
"learning_rate": 5.955555555555555e-06, |
|
"loss": 0.2463, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 14.03, |
|
"learning_rate": 5.9e-06, |
|
"loss": 0.2206, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 15.0, |
|
"learning_rate": 5.844444444444445e-06, |
|
"loss": 0.2137, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 15.01, |
|
"learning_rate": 5.788888888888889e-06, |
|
"loss": 0.1401, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 15.01, |
|
"learning_rate": 5.733333333333334e-06, |
|
"loss": 0.2169, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 15.02, |
|
"learning_rate": 5.677777777777779e-06, |
|
"loss": 0.2412, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 15.02, |
|
"learning_rate": 5.622222222222222e-06, |
|
"loss": 0.1928, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 15.03, |
|
"learning_rate": 5.566666666666667e-06, |
|
"loss": 0.1467, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 15.03, |
|
"learning_rate": 5.511111111111112e-06, |
|
"loss": 0.1832, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 16.0, |
|
"learning_rate": 5.455555555555556e-06, |
|
"loss": 0.1158, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 16.01, |
|
"learning_rate": 5.400000000000001e-06, |
|
"loss": 0.125, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 16.01, |
|
"learning_rate": 5.344444444444446e-06, |
|
"loss": 0.2077, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 16.02, |
|
"learning_rate": 5.288888888888889e-06, |
|
"loss": 0.1517, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 16.02, |
|
"learning_rate": 5.233333333333334e-06, |
|
"loss": 0.1097, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 16.03, |
|
"learning_rate": 5.177777777777779e-06, |
|
"loss": 0.1484, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 17.0, |
|
"learning_rate": 5.122222222222223e-06, |
|
"loss": 0.0974, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 17.01, |
|
"learning_rate": 5.0666666666666676e-06, |
|
"loss": 0.088, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 17.01, |
|
"learning_rate": 5.011111111111111e-06, |
|
"loss": 0.1453, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 17.02, |
|
"learning_rate": 4.9555555555555565e-06, |
|
"loss": 0.1197, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 17.02, |
|
"learning_rate": 4.9000000000000005e-06, |
|
"loss": 0.1048, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 17.03, |
|
"learning_rate": 4.8444444444444446e-06, |
|
"loss": 0.1001, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 18.0, |
|
"learning_rate": 4.7888888888888894e-06, |
|
"loss": 0.0868, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 18.01, |
|
"learning_rate": 4.7333333333333335e-06, |
|
"loss": 0.0596, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 18.01, |
|
"learning_rate": 4.677777777777778e-06, |
|
"loss": 0.1002, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 18.02, |
|
"learning_rate": 4.622222222222222e-06, |
|
"loss": 0.1092, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 18.02, |
|
"learning_rate": 4.566666666666667e-06, |
|
"loss": 0.0855, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 18.03, |
|
"learning_rate": 4.511111111111111e-06, |
|
"loss": 0.0664, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 18.03, |
|
"learning_rate": 4.455555555555555e-06, |
|
"loss": 0.0806, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 18.03, |
|
"eval_loss": 1.7969127893447876, |
|
"eval_runtime": 1014.7752, |
|
"eval_samples_per_second": 4.995, |
|
"eval_steps_per_second": 0.157, |
|
"eval_wer": 63.57112138479734, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 19.0, |
|
"learning_rate": 4.4e-06, |
|
"loss": 0.0552, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 19.01, |
|
"learning_rate": 4.344444444444445e-06, |
|
"loss": 0.0579, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 19.01, |
|
"learning_rate": 4.288888888888889e-06, |
|
"loss": 0.0955, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 19.02, |
|
"learning_rate": 4.233333333333334e-06, |
|
"loss": 0.0687, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 19.02, |
|
"learning_rate": 4.177777777777778e-06, |
|
"loss": 0.0509, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 19.03, |
|
"learning_rate": 4.122222222222222e-06, |
|
"loss": 0.0686, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 20.0, |
|
"learning_rate": 4.066666666666667e-06, |
|
"loss": 0.0439, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 20.01, |
|
"learning_rate": 4.011111111111111e-06, |
|
"loss": 0.0434, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 20.01, |
|
"learning_rate": 3.955555555555556e-06, |
|
"loss": 0.0715, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 20.02, |
|
"learning_rate": 3.900000000000001e-06, |
|
"loss": 0.0565, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 20.02, |
|
"learning_rate": 3.844444444444445e-06, |
|
"loss": 0.0496, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 20.03, |
|
"learning_rate": 3.7888888888888893e-06, |
|
"loss": 0.0484, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 21.0, |
|
"learning_rate": 3.7333333333333337e-06, |
|
"loss": 0.043, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 21.01, |
|
"learning_rate": 3.6777777777777778e-06, |
|
"loss": 0.0307, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 21.01, |
|
"learning_rate": 3.6222222222222226e-06, |
|
"loss": 0.0509, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 21.02, |
|
"learning_rate": 3.566666666666667e-06, |
|
"loss": 0.0526, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 21.02, |
|
"learning_rate": 3.511111111111111e-06, |
|
"loss": 0.0427, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 21.03, |
|
"learning_rate": 3.455555555555556e-06, |
|
"loss": 0.0342, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 21.03, |
|
"learning_rate": 3.4000000000000005e-06, |
|
"loss": 0.04, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 22.0, |
|
"learning_rate": 3.3444444444444445e-06, |
|
"loss": 0.0267, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 22.01, |
|
"learning_rate": 3.2888888888888894e-06, |
|
"loss": 0.0321, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 22.01, |
|
"learning_rate": 3.2333333333333334e-06, |
|
"loss": 0.0503, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 22.02, |
|
"learning_rate": 3.177777777777778e-06, |
|
"loss": 0.0353, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 22.02, |
|
"learning_rate": 3.1222222222222228e-06, |
|
"loss": 0.0261, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 22.03, |
|
"learning_rate": 3.066666666666667e-06, |
|
"loss": 0.0346, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 23.0, |
|
"learning_rate": 3.0111111111111113e-06, |
|
"loss": 0.0252, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 23.01, |
|
"learning_rate": 2.955555555555556e-06, |
|
"loss": 0.0239, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 23.01, |
|
"learning_rate": 2.9e-06, |
|
"loss": 0.0377, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 23.02, |
|
"learning_rate": 2.8444444444444446e-06, |
|
"loss": 0.0306, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 23.02, |
|
"learning_rate": 2.788888888888889e-06, |
|
"loss": 0.0257, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 23.03, |
|
"learning_rate": 2.7333333333333336e-06, |
|
"loss": 0.0276, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 24.0, |
|
"learning_rate": 2.677777777777778e-06, |
|
"loss": 0.0236, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 24.01, |
|
"learning_rate": 2.6222222222222225e-06, |
|
"loss": 0.0179, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 24.01, |
|
"learning_rate": 2.566666666666667e-06, |
|
"loss": 0.029, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 24.02, |
|
"learning_rate": 2.5111111111111114e-06, |
|
"loss": 0.0282, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 24.02, |
|
"learning_rate": 2.455555555555556e-06, |
|
"loss": 0.0233, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 24.03, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 0.0194, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 24.03, |
|
"learning_rate": 2.3444444444444448e-06, |
|
"loss": 0.0228, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 25.0, |
|
"learning_rate": 2.2888888888888892e-06, |
|
"loss": 0.0161, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 25.01, |
|
"learning_rate": 2.2333333333333333e-06, |
|
"loss": 0.0194, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 25.01, |
|
"eval_loss": 1.8434560298919678, |
|
"eval_runtime": 895.0413, |
|
"eval_samples_per_second": 5.663, |
|
"eval_steps_per_second": 0.178, |
|
"eval_wer": 63.46629394688743, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 25.02, |
|
"learning_rate": 2.1777777777777777e-06, |
|
"loss": 0.0281, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 25.02, |
|
"learning_rate": 2.1222222222222226e-06, |
|
"loss": 0.0193, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 25.02, |
|
"learning_rate": 2.0666666666666666e-06, |
|
"loss": 0.0155, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 25.03, |
|
"learning_rate": 2.011111111111111e-06, |
|
"loss": 0.022, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 26.0, |
|
"learning_rate": 1.955555555555556e-06, |
|
"loss": 0.0152, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 26.01, |
|
"learning_rate": 1.9000000000000002e-06, |
|
"loss": 0.0139, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 26.01, |
|
"learning_rate": 1.8444444444444445e-06, |
|
"loss": 0.0233, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 26.02, |
|
"learning_rate": 1.788888888888889e-06, |
|
"loss": 0.0181, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 26.02, |
|
"learning_rate": 1.7333333333333336e-06, |
|
"loss": 0.0153, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 26.03, |
|
"learning_rate": 1.6777777777777779e-06, |
|
"loss": 0.0177, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 27.0, |
|
"learning_rate": 1.6222222222222223e-06, |
|
"loss": 0.0149, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 27.01, |
|
"learning_rate": 1.566666666666667e-06, |
|
"loss": 0.0118, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 27.01, |
|
"learning_rate": 1.5111111111111112e-06, |
|
"loss": 0.0187, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 27.02, |
|
"learning_rate": 1.4555555555555557e-06, |
|
"loss": 0.0178, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 27.02, |
|
"learning_rate": 1.4000000000000001e-06, |
|
"loss": 0.015, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 27.03, |
|
"learning_rate": 1.3444444444444446e-06, |
|
"loss": 0.0137, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 28.0, |
|
"learning_rate": 1.288888888888889e-06, |
|
"loss": 0.0152, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 28.01, |
|
"learning_rate": 1.2333333333333335e-06, |
|
"loss": 0.0109, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 28.01, |
|
"learning_rate": 1.1777777777777778e-06, |
|
"loss": 0.0141, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 28.02, |
|
"learning_rate": 1.1222222222222222e-06, |
|
"loss": 0.0184, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 28.02, |
|
"learning_rate": 1.066666666666667e-06, |
|
"loss": 0.0141, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 28.03, |
|
"learning_rate": 1.0111111111111111e-06, |
|
"loss": 0.011, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 28.03, |
|
"learning_rate": 9.555555555555556e-07, |
|
"loss": 0.0153, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 29.0, |
|
"learning_rate": 9.000000000000001e-07, |
|
"loss": 0.0112, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 29.01, |
|
"learning_rate": 8.444444444444445e-07, |
|
"loss": 0.0108, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 29.01, |
|
"learning_rate": 7.888888888888889e-07, |
|
"loss": 0.0171, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 29.02, |
|
"learning_rate": 7.333333333333334e-07, |
|
"loss": 0.0138, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 29.02, |
|
"learning_rate": 6.777777777777779e-07, |
|
"loss": 0.0113, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 29.03, |
|
"learning_rate": 6.222222222222223e-07, |
|
"loss": 0.014, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 30.0, |
|
"learning_rate": 5.666666666666667e-07, |
|
"loss": 0.0119, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 30.01, |
|
"learning_rate": 5.111111111111112e-07, |
|
"loss": 0.0094, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 30.01, |
|
"learning_rate": 4.5555555555555563e-07, |
|
"loss": 0.0149, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 30.02, |
|
"learning_rate": 4.0000000000000003e-07, |
|
"loss": 0.0138, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 30.02, |
|
"learning_rate": 3.444444444444445e-07, |
|
"loss": 0.0122, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 30.03, |
|
"learning_rate": 2.888888888888889e-07, |
|
"loss": 0.0115, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 31.0, |
|
"learning_rate": 2.3333333333333336e-07, |
|
"loss": 0.0127, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 31.01, |
|
"learning_rate": 1.777777777777778e-07, |
|
"loss": 0.0093, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 31.01, |
|
"learning_rate": 1.2222222222222225e-07, |
|
"loss": 0.0122, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 31.02, |
|
"learning_rate": 6.666666666666668e-08, |
|
"loss": 0.0156, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 31.02, |
|
"learning_rate": 1.1111111111111112e-08, |
|
"loss": 0.0121, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 31.02, |
|
"eval_loss": 1.8737508058547974, |
|
"eval_runtime": 942.5122, |
|
"eval_samples_per_second": 5.378, |
|
"eval_steps_per_second": 0.169, |
|
"eval_wer": 62.69218363616815, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 31.02, |
|
"step": 5000, |
|
"total_flos": 4.876280170020864e+19, |
|
"train_loss": 1.0773965574353934, |
|
"train_runtime": 41952.9149, |
|
"train_samples_per_second": 7.628, |
|
"train_steps_per_second": 0.119 |
|
} |
|
], |
|
"max_steps": 5000, |
|
"num_train_epochs": 9223372036854775807, |
|
"total_flos": 4.876280170020864e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|