| { |
| "best_metric": 71.27942416185721, |
| "best_model_checkpoint": "./output/small/yt-special-batch12-tiny/checkpoint-5000", |
| "epoch": 1.2883277505797475, |
| "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": 4.2000000000000006e-07, |
| "loss": 2.9751, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 9.000000000000001e-07, |
| "loss": 3.0967, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.4000000000000001e-06, |
| "loss": 2.6146, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.9000000000000002e-06, |
| "loss": 2.3808, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 2.4000000000000003e-06, |
| "loss": 2.0314, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 2.9e-06, |
| "loss": 2.0284, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 3.4000000000000005e-06, |
| "loss": 1.8475, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 3.900000000000001e-06, |
| "loss": 1.92, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.4e-06, |
| "loss": 1.7619, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.9000000000000005e-06, |
| "loss": 1.803, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 5.400000000000001e-06, |
| "loss": 1.6079, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 5.9e-06, |
| "loss": 1.7026, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 6.4000000000000006e-06, |
| "loss": 1.6243, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 6.9e-06, |
| "loss": 1.6926, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 7.4e-06, |
| "loss": 1.6141, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 7.9e-06, |
| "loss": 1.6821, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 8.400000000000001e-06, |
| "loss": 1.5946, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 8.900000000000001e-06, |
| "loss": 1.5995, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 9.4e-06, |
| "loss": 1.5004, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 9.9e-06, |
| "loss": 1.5515, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 9.955555555555556e-06, |
| "loss": 1.4719, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 9.9e-06, |
| "loss": 1.4326, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 9.844444444444446e-06, |
| "loss": 1.4214, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 9.78888888888889e-06, |
| "loss": 1.4867, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 9.733333333333334e-06, |
| "loss": 1.4801, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 9.677777777777778e-06, |
| "loss": 1.409, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 9.622222222222222e-06, |
| "loss": 1.4713, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 9.566666666666668e-06, |
| "loss": 1.4711, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 9.511111111111112e-06, |
| "loss": 1.4745, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 9.455555555555557e-06, |
| "loss": 1.3874, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 9.4e-06, |
| "loss": 1.5176, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 9.344444444444446e-06, |
| "loss": 1.4939, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 9.28888888888889e-06, |
| "loss": 1.4472, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 9.233333333333334e-06, |
| "loss": 1.3727, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 9.17777777777778e-06, |
| "loss": 1.4422, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 9.122222222222223e-06, |
| "loss": 1.3619, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 9.066666666666667e-06, |
| "loss": 1.4105, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 9.011111111111111e-06, |
| "loss": 1.4199, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 8.955555555555555e-06, |
| "loss": 1.3523, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 8.900000000000001e-06, |
| "loss": 1.375, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.26, |
| "eval_loss": 1.3638936281204224, |
| "eval_runtime": 238.9212, |
| "eval_samples_per_second": 10.259, |
| "eval_steps_per_second": 1.712, |
| "eval_wer": 103.19693923870048, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 8.844444444444445e-06, |
| "loss": 1.4456, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 8.788888888888891e-06, |
| "loss": 1.4109, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 8.733333333333333e-06, |
| "loss": 1.2913, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 8.677777777777779e-06, |
| "loss": 1.369, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 8.622222222222223e-06, |
| "loss": 1.3553, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 8.566666666666667e-06, |
| "loss": 1.2961, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 8.511111111111113e-06, |
| "loss": 1.3585, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 8.455555555555555e-06, |
| "loss": 1.2917, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 8.400000000000001e-06, |
| "loss": 1.3555, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 8.344444444444445e-06, |
| "loss": 1.3, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 8.288888888888889e-06, |
| "loss": 1.3759, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 8.233333333333335e-06, |
| "loss": 1.3636, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 8.177777777777779e-06, |
| "loss": 1.3149, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.122222222222223e-06, |
| "loss": 1.2901, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 8.066666666666667e-06, |
| "loss": 1.3365, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 8.011111111111113e-06, |
| "loss": 1.2778, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 7.955555555555557e-06, |
| "loss": 1.2149, |
| "step": 1425 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 7.9e-06, |
| "loss": 1.4009, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 7.844444444444446e-06, |
| "loss": 1.3407, |
| "step": 1475 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 7.788888888888889e-06, |
| "loss": 1.2332, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 7.733333333333334e-06, |
| "loss": 1.2813, |
| "step": 1525 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 7.677777777777778e-06, |
| "loss": 1.3404, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 7.622222222222223e-06, |
| "loss": 1.3895, |
| "step": 1575 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 7.566666666666667e-06, |
| "loss": 1.2544, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 7.511111111111111e-06, |
| "loss": 1.2909, |
| "step": 1625 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.455555555555556e-06, |
| "loss": 1.2545, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 7.4e-06, |
| "loss": 1.2419, |
| "step": 1675 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.344444444444445e-06, |
| "loss": 1.3259, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 7.28888888888889e-06, |
| "loss": 1.2864, |
| "step": 1725 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 7.233333333333334e-06, |
| "loss": 1.2436, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.177777777777778e-06, |
| "loss": 1.2894, |
| "step": 1775 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 7.122222222222222e-06, |
| "loss": 1.3266, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 7.066666666666667e-06, |
| "loss": 1.2784, |
| "step": 1825 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.011111111111112e-06, |
| "loss": 1.3272, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 6.955555555555557e-06, |
| "loss": 1.1696, |
| "step": 1875 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 6.9e-06, |
| "loss": 1.2616, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.844444444444445e-06, |
| "loss": 1.2555, |
| "step": 1925 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 6.788888888888889e-06, |
| "loss": 1.2985, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 6.733333333333334e-06, |
| "loss": 1.356, |
| "step": 1975 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.677777777777779e-06, |
| "loss": 1.2229, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.52, |
| "eval_loss": 1.2348008155822754, |
| "eval_runtime": 180.4487, |
| "eval_samples_per_second": 13.583, |
| "eval_steps_per_second": 2.267, |
| "eval_wer": 81.97911938266, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 6.6222222222222236e-06, |
| "loss": 1.2086, |
| "step": 2025 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.566666666666667e-06, |
| "loss": 1.1407, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 6.511111111111112e-06, |
| "loss": 1.2543, |
| "step": 2075 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 6.455555555555556e-06, |
| "loss": 1.2885, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.4000000000000006e-06, |
| "loss": 1.21, |
| "step": 2125 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 6.3444444444444454e-06, |
| "loss": 1.2844, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 6.28888888888889e-06, |
| "loss": 1.2011, |
| "step": 2175 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 6.2333333333333335e-06, |
| "loss": 1.175, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 6.177777777777778e-06, |
| "loss": 1.1657, |
| "step": 2225 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 6.1222222222222224e-06, |
| "loss": 1.1706, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 6.066666666666667e-06, |
| "loss": 1.1911, |
| "step": 2275 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 6.011111111111112e-06, |
| "loss": 1.2894, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 5.955555555555555e-06, |
| "loss": 1.3254, |
| "step": 2325 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.9e-06, |
| "loss": 1.2341, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 5.844444444444445e-06, |
| "loss": 1.1495, |
| "step": 2375 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.788888888888889e-06, |
| "loss": 1.1695, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 5.733333333333334e-06, |
| "loss": 1.2141, |
| "step": 2425 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 5.677777777777779e-06, |
| "loss": 1.2616, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 5.622222222222222e-06, |
| "loss": 1.2569, |
| "step": 2475 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 5.566666666666667e-06, |
| "loss": 1.1775, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.511111111111112e-06, |
| "loss": 1.2098, |
| "step": 2525 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.455555555555556e-06, |
| "loss": 1.1982, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.400000000000001e-06, |
| "loss": 1.1951, |
| "step": 2575 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.344444444444446e-06, |
| "loss": 1.2728, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.288888888888889e-06, |
| "loss": 1.126, |
| "step": 2625 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.233333333333334e-06, |
| "loss": 1.2071, |
| "step": 2650 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 5.177777777777779e-06, |
| "loss": 1.1721, |
| "step": 2675 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 5.122222222222223e-06, |
| "loss": 1.2271, |
| "step": 2700 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 5.0666666666666676e-06, |
| "loss": 1.1708, |
| "step": 2725 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 5.011111111111111e-06, |
| "loss": 1.1825, |
| "step": 2750 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 4.9555555555555565e-06, |
| "loss": 1.1668, |
| "step": 2775 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 4.9000000000000005e-06, |
| "loss": 1.2259, |
| "step": 2800 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 4.8444444444444446e-06, |
| "loss": 1.186, |
| "step": 2825 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 4.7888888888888894e-06, |
| "loss": 1.3066, |
| "step": 2850 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 4.7333333333333335e-06, |
| "loss": 1.1446, |
| "step": 2875 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 4.677777777777778e-06, |
| "loss": 1.2538, |
| "step": 2900 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 4.622222222222222e-06, |
| "loss": 1.195, |
| "step": 2925 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 4.566666666666667e-06, |
| "loss": 1.2484, |
| "step": 2950 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 4.511111111111111e-06, |
| "loss": 1.227, |
| "step": 2975 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 4.455555555555555e-06, |
| "loss": 1.2384, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.77, |
| "eval_loss": 1.1718696355819702, |
| "eval_runtime": 180.5419, |
| "eval_samples_per_second": 13.576, |
| "eval_steps_per_second": 2.265, |
| "eval_wer": 87.50405291485637, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 4.4e-06, |
| "loss": 1.083, |
| "step": 3025 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 4.344444444444445e-06, |
| "loss": 1.2478, |
| "step": 3050 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 4.288888888888889e-06, |
| "loss": 1.1939, |
| "step": 3075 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 4.233333333333334e-06, |
| "loss": 1.2112, |
| "step": 3100 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 4.177777777777778e-06, |
| "loss": 1.192, |
| "step": 3125 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 4.122222222222222e-06, |
| "loss": 1.2728, |
| "step": 3150 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 4.066666666666667e-06, |
| "loss": 1.0322, |
| "step": 3175 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 4.011111111111111e-06, |
| "loss": 1.2554, |
| "step": 3200 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.955555555555556e-06, |
| "loss": 1.0806, |
| "step": 3225 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.900000000000001e-06, |
| "loss": 1.1314, |
| "step": 3250 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.844444444444445e-06, |
| "loss": 1.1556, |
| "step": 3275 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.7888888888888893e-06, |
| "loss": 1.1304, |
| "step": 3300 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.7333333333333337e-06, |
| "loss": 1.0426, |
| "step": 3325 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.6777777777777778e-06, |
| "loss": 1.1634, |
| "step": 3350 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.6222222222222226e-06, |
| "loss": 1.1665, |
| "step": 3375 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.566666666666667e-06, |
| "loss": 1.2005, |
| "step": 3400 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.511111111111111e-06, |
| "loss": 1.1668, |
| "step": 3425 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.455555555555556e-06, |
| "loss": 1.1735, |
| "step": 3450 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.4000000000000005e-06, |
| "loss": 1.1669, |
| "step": 3475 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.3444444444444445e-06, |
| "loss": 1.1397, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.2888888888888894e-06, |
| "loss": 1.1569, |
| "step": 3525 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.2333333333333334e-06, |
| "loss": 1.1175, |
| "step": 3550 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.177777777777778e-06, |
| "loss": 1.1542, |
| "step": 3575 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.1222222222222228e-06, |
| "loss": 1.1476, |
| "step": 3600 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.066666666666667e-06, |
| "loss": 1.2121, |
| "step": 3625 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.0111111111111113e-06, |
| "loss": 1.15, |
| "step": 3650 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 2.955555555555556e-06, |
| "loss": 1.1461, |
| "step": 3675 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 2.9e-06, |
| "loss": 0.9944, |
| "step": 3700 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 2.8444444444444446e-06, |
| "loss": 1.1293, |
| "step": 3725 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 2.788888888888889e-06, |
| "loss": 1.208, |
| "step": 3750 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 2.7333333333333336e-06, |
| "loss": 1.1021, |
| "step": 3775 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.677777777777778e-06, |
| "loss": 1.1093, |
| "step": 3800 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 2.6222222222222225e-06, |
| "loss": 1.1853, |
| "step": 3825 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 2.566666666666667e-06, |
| "loss": 1.1247, |
| "step": 3850 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 2.5111111111111114e-06, |
| "loss": 1.2292, |
| "step": 3875 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 2.455555555555556e-06, |
| "loss": 1.0086, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 2.4000000000000003e-06, |
| "loss": 0.949, |
| "step": 3925 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 2.3444444444444448e-06, |
| "loss": 1.0815, |
| "step": 3950 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 2.2888888888888892e-06, |
| "loss": 1.032, |
| "step": 3975 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 2.2333333333333333e-06, |
| "loss": 0.9738, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.03, |
| "eval_loss": 1.1388514041900635, |
| "eval_runtime": 190.8907, |
| "eval_samples_per_second": 12.84, |
| "eval_steps_per_second": 2.143, |
| "eval_wer": 71.38317878218015, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 2.1777777777777777e-06, |
| "loss": 1.0481, |
| "step": 4025 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 2.1222222222222226e-06, |
| "loss": 0.965, |
| "step": 4050 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 2.0666666666666666e-06, |
| "loss": 0.9748, |
| "step": 4075 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 2.011111111111111e-06, |
| "loss": 0.9869, |
| "step": 4100 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 1.955555555555556e-06, |
| "loss": 0.9268, |
| "step": 4125 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 1.9000000000000002e-06, |
| "loss": 1.0397, |
| "step": 4150 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 1.8444444444444445e-06, |
| "loss": 0.962, |
| "step": 4175 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 1.788888888888889e-06, |
| "loss": 0.9856, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 1.7333333333333336e-06, |
| "loss": 1.008, |
| "step": 4225 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 1.6777777777777779e-06, |
| "loss": 0.9956, |
| "step": 4250 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 1.6222222222222223e-06, |
| "loss": 1.0277, |
| "step": 4275 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 1.566666666666667e-06, |
| "loss": 1.0549, |
| "step": 4300 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 1.5111111111111112e-06, |
| "loss": 1.0795, |
| "step": 4325 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 1.4555555555555557e-06, |
| "loss": 1.0079, |
| "step": 4350 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 1.4000000000000001e-06, |
| "loss": 0.9753, |
| "step": 4375 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 1.3444444444444446e-06, |
| "loss": 1.0847, |
| "step": 4400 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 1.288888888888889e-06, |
| "loss": 0.9879, |
| "step": 4425 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 1.2333333333333335e-06, |
| "loss": 0.9383, |
| "step": 4450 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 1.1777777777777778e-06, |
| "loss": 1.0121, |
| "step": 4475 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 1.1222222222222222e-06, |
| "loss": 0.9703, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 1.066666666666667e-06, |
| "loss": 0.9821, |
| "step": 4525 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 1.0111111111111111e-06, |
| "loss": 1.0396, |
| "step": 4550 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 9.555555555555556e-07, |
| "loss": 0.9878, |
| "step": 4575 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 9.000000000000001e-07, |
| "loss": 0.9714, |
| "step": 4600 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 8.444444444444445e-07, |
| "loss": 1.0082, |
| "step": 4625 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 7.888888888888889e-07, |
| "loss": 0.9743, |
| "step": 4650 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 7.333333333333334e-07, |
| "loss": 1.0269, |
| "step": 4675 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 6.777777777777779e-07, |
| "loss": 0.9287, |
| "step": 4700 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 6.222222222222223e-07, |
| "loss": 1.0011, |
| "step": 4725 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 5.666666666666667e-07, |
| "loss": 0.9719, |
| "step": 4750 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 5.111111111111112e-07, |
| "loss": 0.9388, |
| "step": 4775 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 4.5555555555555563e-07, |
| "loss": 0.9516, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 4.0000000000000003e-07, |
| "loss": 0.9687, |
| "step": 4825 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 3.444444444444445e-07, |
| "loss": 0.8859, |
| "step": 4850 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.888888888888889e-07, |
| "loss": 0.9387, |
| "step": 4875 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.3333333333333336e-07, |
| "loss": 1.0103, |
| "step": 4900 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 1.777777777777778e-07, |
| "loss": 0.9238, |
| "step": 4925 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 1.2222222222222225e-07, |
| "loss": 1.0117, |
| "step": 4950 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 6.666666666666668e-08, |
| "loss": 0.9691, |
| "step": 4975 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 1.1111111111111112e-08, |
| "loss": 0.9485, |
| "step": 5000 |
| }, |
| { |
| "epoch": 1.29, |
| "eval_loss": 1.1267486810684204, |
| "eval_runtime": 147.6915, |
| "eval_samples_per_second": 16.595, |
| "eval_steps_per_second": 2.769, |
| "eval_wer": 71.27942416185721, |
| "step": 5000 |
| }, |
| { |
| "epoch": 1.29, |
| "step": 5000, |
| "total_flos": 1.47708386942976e+18, |
| "train_loss": 1.2652734420776368, |
| "train_runtime": 3115.0084, |
| "train_samples_per_second": 19.262, |
| "train_steps_per_second": 1.605 |
| } |
| ], |
| "max_steps": 5000, |
| "num_train_epochs": 2, |
| "total_flos": 1.47708386942976e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
