| { | |
| "best_metric": 59.84047727125349, | |
| "best_model_checkpoint": "./output/small/yt-special-batch4-lr4-small/checkpoint-5000", | |
| "epoch": 0.4294425835265825, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.37, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9e-06, | |
| "loss": 1.6635, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 1.0846, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.0443, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.1125, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9e-05, | |
| "loss": 1.3199, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.1638, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 1.1791, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.2697, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9e-05, | |
| "loss": 1.2493, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 1.5799, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9e-05, | |
| "loss": 1.3674, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 1.3953, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.9e-05, | |
| "loss": 1.4332, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.4e-05, | |
| "loss": 1.5831, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.900000000000001e-05, | |
| "loss": 1.5507, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.4e-05, | |
| "loss": 1.9505, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.900000000000001e-05, | |
| "loss": 1.8439, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.4e-05, | |
| "loss": 1.9936, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.900000000000001e-05, | |
| "loss": 1.8267, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.955555555555556e-05, | |
| "loss": 1.8131, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.900000000000001e-05, | |
| "loss": 2.3006, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.844444444444444e-05, | |
| "loss": 1.9656, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.78888888888889e-05, | |
| "loss": 2.1053, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.733333333333335e-05, | |
| "loss": 1.886, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.677777777777778e-05, | |
| "loss": 2.0105, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.622222222222222e-05, | |
| "loss": 2.1539, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.566666666666667e-05, | |
| "loss": 2.0049, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.511111111111112e-05, | |
| "loss": 2.282, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.455555555555556e-05, | |
| "loss": 2.1915, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.4e-05, | |
| "loss": 2.0148, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.344444444444444e-05, | |
| "loss": 1.9464, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.28888888888889e-05, | |
| "loss": 1.8012, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.233333333333333e-05, | |
| "loss": 1.8871, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.177777777777778e-05, | |
| "loss": 2.084, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.122222222222223e-05, | |
| "loss": 2.0643, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.066666666666667e-05, | |
| "loss": 1.9001, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.011111111111111e-05, | |
| "loss": 2.0156, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.955555555555556e-05, | |
| "loss": 1.9293, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.900000000000001e-05, | |
| "loss": 2.2491, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 1.9142402410507202, | |
| "eval_runtime": 2926.2564, | |
| "eval_samples_per_second": 0.838, | |
| "eval_steps_per_second": 0.419, | |
| "eval_wer": 226.4833668374295, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.844444444444445e-05, | |
| "loss": 1.7898, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.78888888888889e-05, | |
| "loss": 2.0532, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.733333333333333e-05, | |
| "loss": 2.1319, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.677777777777778e-05, | |
| "loss": 1.8704, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.622222222222222e-05, | |
| "loss": 1.9468, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.566666666666667e-05, | |
| "loss": 1.8973, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.511111111111112e-05, | |
| "loss": 2.0607, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.455555555555556e-05, | |
| "loss": 2.1119, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.4e-05, | |
| "loss": 1.9899, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.344444444444445e-05, | |
| "loss": 1.7752, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.28888888888889e-05, | |
| "loss": 1.9727, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.233333333333333e-05, | |
| "loss": 1.7061, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.177777777777778e-05, | |
| "loss": 1.9447, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.122222222222222e-05, | |
| "loss": 1.7837, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.066666666666667e-05, | |
| "loss": 1.8571, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.011111111111111e-05, | |
| "loss": 1.6066, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.955555555555556e-05, | |
| "loss": 1.8066, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.900000000000001e-05, | |
| "loss": 1.855, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.844444444444446e-05, | |
| "loss": 1.8418, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.788888888888888e-05, | |
| "loss": 1.8486, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.733333333333333e-05, | |
| "loss": 1.7621, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.677777777777778e-05, | |
| "loss": 1.5957, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.622222222222223e-05, | |
| "loss": 1.6305, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.566666666666667e-05, | |
| "loss": 1.7105, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.511111111111111e-05, | |
| "loss": 1.7968, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.455555555555556e-05, | |
| "loss": 1.6489, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.4e-05, | |
| "loss": 1.6482, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.344444444444445e-05, | |
| "loss": 1.5142, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.28888888888889e-05, | |
| "loss": 1.7414, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.233333333333335e-05, | |
| "loss": 1.6974, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.177777777777777e-05, | |
| "loss": 1.8226, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 7.122222222222222e-05, | |
| "loss": 1.7511, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.066666666666667e-05, | |
| "loss": 1.8839, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.011111111111112e-05, | |
| "loss": 1.6408, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.955555555555556e-05, | |
| "loss": 1.5931, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.9e-05, | |
| "loss": 1.4517, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.844444444444445e-05, | |
| "loss": 1.5901, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.788888888888888e-05, | |
| "loss": 1.6494, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.733333333333333e-05, | |
| "loss": 1.631, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.677777777777779e-05, | |
| "loss": 1.4702, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 1.6153531074523926, | |
| "eval_runtime": 2351.4276, | |
| "eval_samples_per_second": 1.042, | |
| "eval_steps_per_second": 0.521, | |
| "eval_wer": 115.55022372090006, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.622222222222224e-05, | |
| "loss": 1.6328, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.566666666666666e-05, | |
| "loss": 1.8714, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.511111111111111e-05, | |
| "loss": 1.5653, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.455555555555556e-05, | |
| "loss": 1.5279, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 1.6341, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.344444444444445e-05, | |
| "loss": 1.6861, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.28888888888889e-05, | |
| "loss": 1.5053, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.233333333333334e-05, | |
| "loss": 1.5605, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.177777777777779e-05, | |
| "loss": 1.569, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.122222222222222e-05, | |
| "loss": 1.4575, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.066666666666667e-05, | |
| "loss": 1.605, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.011111111111112e-05, | |
| "loss": 1.8061, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9555555555555554e-05, | |
| "loss": 1.5892, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9e-05, | |
| "loss": 1.8554, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.844444444444445e-05, | |
| "loss": 1.5377, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.788888888888889e-05, | |
| "loss": 1.3947, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.7333333333333336e-05, | |
| "loss": 1.5417, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.6777777777777786e-05, | |
| "loss": 1.4101, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.622222222222222e-05, | |
| "loss": 1.607, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 5.566666666666667e-05, | |
| "loss": 1.5681, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.511111111111111e-05, | |
| "loss": 1.5026, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.455555555555556e-05, | |
| "loss": 1.3117, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 1.3261, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.3444444444444455e-05, | |
| "loss": 1.5585, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.2888888888888885e-05, | |
| "loss": 1.4642, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.2333333333333336e-05, | |
| "loss": 1.3654, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.177777777777778e-05, | |
| "loss": 1.5675, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.122222222222223e-05, | |
| "loss": 1.2993, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.0666666666666674e-05, | |
| "loss": 1.4956, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.011111111111111e-05, | |
| "loss": 1.4068, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.955555555555556e-05, | |
| "loss": 1.3342, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9e-05, | |
| "loss": 1.4048, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.844444444444445e-05, | |
| "loss": 1.2884, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7888888888888886e-05, | |
| "loss": 1.6097, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.7333333333333336e-05, | |
| "loss": 1.4716, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.677777777777778e-05, | |
| "loss": 1.2035, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6222222222222224e-05, | |
| "loss": 1.5005, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.566666666666667e-05, | |
| "loss": 1.3144, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.511111111111112e-05, | |
| "loss": 1.1728, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.4555555555555555e-05, | |
| "loss": 1.609, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 1.3598692417144775, | |
| "eval_runtime": 2433.9904, | |
| "eval_samples_per_second": 1.007, | |
| "eval_steps_per_second": 0.504, | |
| "eval_wer": 113.34543803903767, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.4524, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.344444444444445e-05, | |
| "loss": 1.6242, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.2888888888888886e-05, | |
| "loss": 1.2862, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.233333333333334e-05, | |
| "loss": 1.3138, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.177777777777778e-05, | |
| "loss": 1.3386, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.1222222222222224e-05, | |
| "loss": 1.369, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.066666666666667e-05, | |
| "loss": 1.3546, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.011111111111111e-05, | |
| "loss": 1.3656, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.9555555555555556e-05, | |
| "loss": 1.1057, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 1.2673, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.844444444444444e-05, | |
| "loss": 1.2445, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.7888888888888894e-05, | |
| "loss": 1.2798, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.733333333333334e-05, | |
| "loss": 1.1667, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.677777777777778e-05, | |
| "loss": 1.2804, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.6222222222222225e-05, | |
| "loss": 1.2754, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.566666666666667e-05, | |
| "loss": 1.2082, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.511111111111111e-05, | |
| "loss": 1.3477, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4555555555555556e-05, | |
| "loss": 1.1487, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.2331, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.3444444444444443e-05, | |
| "loss": 1.1283, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.2888888888888894e-05, | |
| "loss": 1.3077, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.233333333333333e-05, | |
| "loss": 1.1942, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.177777777777778e-05, | |
| "loss": 1.2615, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.1222222222222225e-05, | |
| "loss": 1.0541, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.066666666666667e-05, | |
| "loss": 1.114, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.0111111111111113e-05, | |
| "loss": 1.2954, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.955555555555556e-05, | |
| "loss": 1.1905, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.9e-05, | |
| "loss": 1.3065, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.8444444444444447e-05, | |
| "loss": 1.1576, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.788888888888889e-05, | |
| "loss": 1.1185, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.733333333333333e-05, | |
| "loss": 1.1523, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.677777777777778e-05, | |
| "loss": 1.1, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.6222222222222226e-05, | |
| "loss": 1.3244, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.5666666666666666e-05, | |
| "loss": 1.1629, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.5111111111111113e-05, | |
| "loss": 1.2712, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.4555555555555557e-05, | |
| "loss": 1.1737, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.1636, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.3444444444444448e-05, | |
| "loss": 0.9854, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.288888888888889e-05, | |
| "loss": 1.2423, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2333333333333335e-05, | |
| "loss": 1.1817, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 1.125331997871399, | |
| "eval_runtime": 2077.4198, | |
| "eval_samples_per_second": 1.18, | |
| "eval_steps_per_second": 0.59, | |
| "eval_wer": 68.40671811166591, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.177777777777778e-05, | |
| "loss": 1.1133, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1222222222222223e-05, | |
| "loss": 1.0712, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.0666666666666666e-05, | |
| "loss": 1.0364, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.011111111111111e-05, | |
| "loss": 1.364, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.9555555555555557e-05, | |
| "loss": 1.1247, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.1726, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8444444444444445e-05, | |
| "loss": 0.9544, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.788888888888889e-05, | |
| "loss": 1.1842, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7333333333333336e-05, | |
| "loss": 1.0767, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.677777777777778e-05, | |
| "loss": 1.0531, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.6222222222222223e-05, | |
| "loss": 0.9292, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5666666666666667e-05, | |
| "loss": 1.2709, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.5111111111111112e-05, | |
| "loss": 1.0287, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4555555555555556e-05, | |
| "loss": 1.1724, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 1.1525, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3444444444444445e-05, | |
| "loss": 1.0005, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.2888888888888889e-05, | |
| "loss": 1.1709, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.2333333333333334e-05, | |
| "loss": 0.8791, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.1777777777777778e-05, | |
| "loss": 0.9991, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.1222222222222224e-05, | |
| "loss": 1.0882, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0666666666666667e-05, | |
| "loss": 1.0362, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.0111111111111111e-05, | |
| "loss": 0.8298, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 1.1218, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9e-06, | |
| "loss": 0.9947, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 1.0605, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 1.0053, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.333333333333334e-06, | |
| "loss": 0.9818, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 1.1732, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.222222222222222e-06, | |
| "loss": 1.1049, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.9709, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.9238, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.9518, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.9865, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.4444444444444444e-06, | |
| "loss": 1.0085, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9111111111111114e-06, | |
| "loss": 1.1636, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.3555555555555555e-06, | |
| "loss": 0.9237, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8e-06, | |
| "loss": 0.9991, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2444444444444445e-06, | |
| "loss": 0.8775, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.888888888888889e-07, | |
| "loss": 0.9883, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.3333333333333334e-07, | |
| "loss": 0.9678, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_loss": 0.977332353591919, | |
| "eval_runtime": 2057.4118, | |
| "eval_samples_per_second": 1.191, | |
| "eval_steps_per_second": 0.596, | |
| "eval_wer": 59.84047727125349, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "step": 5000, | |
| "total_flos": 5.7717080064e+18, | |
| "train_loss": 1.466285028076172, | |
| "train_runtime": 25651.8701, | |
| "train_samples_per_second": 0.78, | |
| "train_steps_per_second": 0.195 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 1, | |
| "total_flos": 5.7717080064e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |