| { | |
| "best_metric": 12.275132275132275, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch8-augment-small/checkpoint-4000", | |
| "epoch": 2.420135527589545, | |
| "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": 1.6773, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.4789, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.1944, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.0194, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.669, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.5553, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.5677, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.538, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.5272, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.5007, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.4593, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.5243, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.4019, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.4513, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.4546, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.4499, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.4464, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.4948, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.4327, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.4434, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.3866, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.3965, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.3821, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.4119, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.356, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.3782, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.3433, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.3923, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.4081, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.3447, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.3507, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.3063, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.3646, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.3626, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3559, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.3281, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.352, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.3588, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.3067, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.3166, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_loss": 0.26330870389938354, | |
| "eval_runtime": 2497.298, | |
| "eval_samples_per_second": 1.45, | |
| "eval_steps_per_second": 0.363, | |
| "eval_wer": 14.837819185645273, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.3061, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.3326, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.2932, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.3438, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.3046, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.2693, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.2822, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.3549, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.2952, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.2414, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.2915, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.2694, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.2898, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.2787, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.2545, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.2684, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.2544, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.2684, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.2452, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.2519, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.256, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.2345, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.233, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.2978, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.2354, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.2101, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.2496, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.2548, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.261, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.2869, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.2205, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.2465, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.2349, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.2333, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.1858, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.2296, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.185, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.2273, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.2353, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.2021, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 0.24563950300216675, | |
| "eval_runtime": 2497.1423, | |
| "eval_samples_per_second": 1.45, | |
| "eval_steps_per_second": 0.363, | |
| "eval_wer": 13.6783988957902, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.2266, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.2181, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.1639, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.0941, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.0983, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.1122, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.0997, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.0963, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.1157, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.0936, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.0892, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.0909, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.0962, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.0878, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.084, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.0856, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.0934, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.1065, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.0922, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.0924, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.0944, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.105, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.1078, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.0791, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.0827, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.0851, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.0922, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.09, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.1035, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.0903, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.0861, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.0944, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.0768, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.076, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.0887, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.0825, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.0685, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.0732, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.0809, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.0834, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_loss": 0.24526779353618622, | |
| "eval_runtime": 2493.6237, | |
| "eval_samples_per_second": 1.453, | |
| "eval_steps_per_second": 0.363, | |
| "eval_wer": 12.675408327582241, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.0864, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.074, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.068, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.0783, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.0654, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.066, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.068, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.0694, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.0916, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.0932, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.0825, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.071, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.0601, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.076, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.08, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.0602, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.0725, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.0813, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.0765, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.0645, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.0769, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.0756, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.0958, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.0595, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.0743, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.0753, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.0488, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.0718, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.0543, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.0751, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.0726, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.0648, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.062, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.064, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.0592, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.0797, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.0932, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.0495, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.0707, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.0572, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 0.24054929614067078, | |
| "eval_runtime": 2493.7592, | |
| "eval_samples_per_second": 1.452, | |
| "eval_steps_per_second": 0.363, | |
| "eval_wer": 12.275132275132275, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.0455, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.0549, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.0568, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.0864, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.0553, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.0279, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.0223, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.0267, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.0276, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.0249, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.0237, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.0184, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.0216, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.024, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.0194, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.0242, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.022, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.024, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.0188, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.0212, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.0241, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0209, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.0205, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.0326, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.0185, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.0288, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.0167, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.0171, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.0197, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.0229, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.0144, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.0198, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.0228, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.0227, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0222, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.0246, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.0281, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.0194, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.0193, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.017, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_loss": 0.25745365023612976, | |
| "eval_runtime": 2494.9891, | |
| "eval_samples_per_second": 1.452, | |
| "eval_steps_per_second": 0.363, | |
| "eval_wer": 12.28893489763055, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "step": 5000, | |
| "total_flos": 1.153995298799616e+19, | |
| "train_loss": 0.1941181287109852, | |
| "train_runtime": 37219.4225, | |
| "train_samples_per_second": 1.075, | |
| "train_steps_per_second": 0.134 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.153995298799616e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |