| { | |
| "best_metric": 90.98360655737704, | |
| "best_model_checkpoint": "./whripser-small-thai/checkpoint-5000", | |
| "epoch": 10.004, | |
| "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-09, | |
| "loss": 0.7192, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.4e-09, | |
| "loss": 0.7311, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4399999999999998e-08, | |
| "loss": 0.7026, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.94e-08, | |
| "loss": 0.6412, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4399999999999997e-08, | |
| "loss": 0.7782, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.94e-08, | |
| "loss": 0.8013, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.44e-08, | |
| "loss": 0.7156, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.94e-08, | |
| "loss": 0.8338, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.44e-08, | |
| "loss": 1.011, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.94e-08, | |
| "loss": 1.0262, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.4400000000000004e-08, | |
| "loss": 0.7159, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.9399999999999996e-08, | |
| "loss": 0.7569, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.44e-08, | |
| "loss": 0.8146, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.939999999999999e-08, | |
| "loss": 0.6515, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.439999999999999e-08, | |
| "loss": 0.6634, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.94e-08, | |
| "loss": 0.5949, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.439999999999999e-08, | |
| "loss": 0.5729, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.939999999999999e-08, | |
| "loss": 0.5238, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.439999999999999e-08, | |
| "loss": 0.5218, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.939999999999999e-08, | |
| "loss": 0.514, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.95111111111111e-08, | |
| "loss": 0.5364, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.895555555555555e-08, | |
| "loss": 0.544, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.84e-08, | |
| "loss": 0.4954, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.784444444444444e-08, | |
| "loss": 0.4859, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.728888888888889e-08, | |
| "loss": 0.5973, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.673333333333334e-08, | |
| "loss": 0.5621, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.617777777777777e-08, | |
| "loss": 0.5139, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.562222222222221e-08, | |
| "loss": 0.6064, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.506666666666666e-08, | |
| "loss": 0.7408, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.45111111111111e-08, | |
| "loss": 0.7071, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.395555555555555e-08, | |
| "loss": 0.5247, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.34e-08, | |
| "loss": 0.56, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.284444444444444e-08, | |
| "loss": 0.6259, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.228888888888889e-08, | |
| "loss": 0.5078, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.173333333333333e-08, | |
| "loss": 0.5037, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.117777777777777e-08, | |
| "loss": 0.4575, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.062222222222222e-08, | |
| "loss": 0.4597, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.006666666666665e-08, | |
| "loss": 0.4307, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.95111111111111e-08, | |
| "loss": 0.4338, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.895555555555555e-08, | |
| "loss": 0.4319, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.42025384306907654, | |
| "eval_runtime": 77.7625, | |
| "eval_samples_per_second": 2.572, | |
| "eval_steps_per_second": 0.09, | |
| "eval_wer": 97.54098360655738, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.84e-08, | |
| "loss": 0.4914, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.784444444444445e-08, | |
| "loss": 0.4791, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.72888888888889e-08, | |
| "loss": 0.4491, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.673333333333333e-08, | |
| "loss": 0.446, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.617777777777778e-08, | |
| "loss": 0.5707, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.562222222222221e-08, | |
| "loss": 0.5088, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.506666666666666e-08, | |
| "loss": 0.4841, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.451111111111111e-08, | |
| "loss": 0.5717, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.395555555555556e-08, | |
| "loss": 0.6943, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 8.339999999999999e-08, | |
| "loss": 0.6298, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.284444444444444e-08, | |
| "loss": 0.4782, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.228888888888889e-08, | |
| "loss": 0.5367, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.173333333333334e-08, | |
| "loss": 0.5775, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.117777777777777e-08, | |
| "loss": 0.4723, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.062222222222221e-08, | |
| "loss": 0.4586, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.006666666666666e-08, | |
| "loss": 0.4325, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.95111111111111e-08, | |
| "loss": 0.4157, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.895555555555555e-08, | |
| "loss": 0.3988, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.84e-08, | |
| "loss": 0.402, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.784444444444445e-08, | |
| "loss": 0.4143, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.728888888888888e-08, | |
| "loss": 0.4761, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.673333333333333e-08, | |
| "loss": 0.4544, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.617777777777778e-08, | |
| "loss": 0.4262, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.562222222222222e-08, | |
| "loss": 0.423, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.506666666666666e-08, | |
| "loss": 0.5421, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.45111111111111e-08, | |
| "loss": 0.4873, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.395555555555555e-08, | |
| "loss": 0.4706, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.34e-08, | |
| "loss": 0.558, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.284444444444444e-08, | |
| "loss": 0.675, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.228888888888889e-08, | |
| "loss": 0.5827, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.173333333333334e-08, | |
| "loss": 0.4601, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.117777777777778e-08, | |
| "loss": 0.5262, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.062222222222221e-08, | |
| "loss": 0.5461, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.006666666666666e-08, | |
| "loss": 0.453, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.951111111111111e-08, | |
| "loss": 0.4287, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.895555555555556e-08, | |
| "loss": 0.403, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.84e-08, | |
| "loss": 0.3984, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.784444444444444e-08, | |
| "loss": 0.3806, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.728888888888889e-08, | |
| "loss": 0.3906, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.673333333333334e-08, | |
| "loss": 0.4003, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.3932681679725647, | |
| "eval_runtime": 80.2975, | |
| "eval_samples_per_second": 2.491, | |
| "eval_steps_per_second": 0.087, | |
| "eval_wer": 95.08196721311475, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.617777777777777e-08, | |
| "loss": 0.4578, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.562222222222222e-08, | |
| "loss": 0.4328, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.506666666666665e-08, | |
| "loss": 0.4132, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.45111111111111e-08, | |
| "loss": 0.4059, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 6.395555555555555e-08, | |
| "loss": 0.545, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 6.34e-08, | |
| "loss": 0.4708, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 6.284444444444445e-08, | |
| "loss": 0.4639, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 6.22888888888889e-08, | |
| "loss": 0.5434, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.173333333333333e-08, | |
| "loss": 0.6705, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.117777777777778e-08, | |
| "loss": 0.5391, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 6.062222222222221e-08, | |
| "loss": 0.4407, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 6.006666666666666e-08, | |
| "loss": 0.5235, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 5.951111111111111e-08, | |
| "loss": 0.5225, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 5.895555555555556e-08, | |
| "loss": 0.4377, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.8399999999999994e-08, | |
| "loss": 0.4083, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.784444444444444e-08, | |
| "loss": 0.3959, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 5.7288888888888884e-08, | |
| "loss": 0.3816, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 5.673333333333333e-08, | |
| "loss": 0.3584, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 5.617777777777778e-08, | |
| "loss": 0.3807, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.5622222222222215e-08, | |
| "loss": 0.3939, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.5066666666666663e-08, | |
| "loss": 0.4502, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.4511111111111105e-08, | |
| "loss": 0.4195, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.3955555555555553e-08, | |
| "loss": 0.4009, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.34e-08, | |
| "loss": 0.4026, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.284444444444445e-08, | |
| "loss": 0.5409, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.2288888888888885e-08, | |
| "loss": 0.4502, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 5.173333333333333e-08, | |
| "loss": 0.4439, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 5.1177777777777775e-08, | |
| "loss": 0.5446, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 5.0622222222222223e-08, | |
| "loss": 0.6696, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 5.0066666666666665e-08, | |
| "loss": 0.5091, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 4.9511111111111107e-08, | |
| "loss": 0.4341, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 4.8955555555555555e-08, | |
| "loss": 0.5252, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.8399999999999997e-08, | |
| "loss": 0.5005, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.7844444444444445e-08, | |
| "loss": 0.4282, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 4.7288888888888887e-08, | |
| "loss": 0.3882, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 4.673333333333333e-08, | |
| "loss": 0.3797, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 4.6177777777777777e-08, | |
| "loss": 0.3738, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 4.562222222222222e-08, | |
| "loss": 0.3625, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 4.5066666666666667e-08, | |
| "loss": 0.3668, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.451111111111111e-08, | |
| "loss": 0.3844, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 0.3800288140773773, | |
| "eval_runtime": 77.9431, | |
| "eval_samples_per_second": 2.566, | |
| "eval_steps_per_second": 0.09, | |
| "eval_wer": 93.0327868852459, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.395555555555555e-08, | |
| "loss": 0.4484, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.34e-08, | |
| "loss": 0.4026, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.2844444444444446e-08, | |
| "loss": 0.3964, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.228888888888889e-08, | |
| "loss": 0.3929, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 4.173333333333333e-08, | |
| "loss": 0.5369, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 4.117777777777778e-08, | |
| "loss": 0.4424, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.062222222222222e-08, | |
| "loss": 0.4513, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.006666666666667e-08, | |
| "loss": 0.5267, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 3.951111111111111e-08, | |
| "loss": 0.6717, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 3.895555555555555e-08, | |
| "loss": 0.4795, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 3.84e-08, | |
| "loss": 0.4338, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 3.784444444444445e-08, | |
| "loss": 0.5366, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 3.728888888888888e-08, | |
| "loss": 0.4661, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 3.673333333333333e-08, | |
| "loss": 0.4188, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 3.617777777777777e-08, | |
| "loss": 0.3841, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 3.562222222222222e-08, | |
| "loss": 0.3564, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 3.506666666666667e-08, | |
| "loss": 0.3743, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 3.4511111111111105e-08, | |
| "loss": 0.3548, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 3.395555555555555e-08, | |
| "loss": 0.3544, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.34e-08, | |
| "loss": 0.3892, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.284444444444444e-08, | |
| "loss": 0.4471, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.2288888888888885e-08, | |
| "loss": 0.3999, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.173333333333333e-08, | |
| "loss": 0.3776, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.1177777777777775e-08, | |
| "loss": 0.3994, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.062222222222222e-08, | |
| "loss": 0.5304, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.006666666666667e-08, | |
| "loss": 0.4276, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 2.951111111111111e-08, | |
| "loss": 0.4601, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 2.8955555555555555e-08, | |
| "loss": 0.5235, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.8399999999999996e-08, | |
| "loss": 0.665, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.784444444444444e-08, | |
| "loss": 0.4652, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 2.728888888888889e-08, | |
| "loss": 0.4232, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 2.673333333333333e-08, | |
| "loss": 0.5323, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 2.6177777777777776e-08, | |
| "loss": 0.4561, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 2.5622222222222224e-08, | |
| "loss": 0.4178, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.5066666666666663e-08, | |
| "loss": 0.368, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.451111111111111e-08, | |
| "loss": 0.3619, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 2.3955555555555553e-08, | |
| "loss": 0.3577, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 2.34e-08, | |
| "loss": 0.3538, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 2.2844444444444443e-08, | |
| "loss": 0.3573, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.2288888888888888e-08, | |
| "loss": 0.3876, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 0.37291276454925537, | |
| "eval_runtime": 79.2057, | |
| "eval_samples_per_second": 2.525, | |
| "eval_steps_per_second": 0.088, | |
| "eval_wer": 91.80327868852459, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.1733333333333333e-08, | |
| "loss": 0.437, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 2.1177777777777778e-08, | |
| "loss": 0.3826, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.0622222222222223e-08, | |
| "loss": 0.3923, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.0066666666666664e-08, | |
| "loss": 0.4029, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 1.9511111111111113e-08, | |
| "loss": 0.5237, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 1.8955555555555554e-08, | |
| "loss": 0.4191, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 1.84e-08, | |
| "loss": 0.4611, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 1.784444444444444e-08, | |
| "loss": 0.5215, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 1.728888888888889e-08, | |
| "loss": 0.655, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 1.6733333333333334e-08, | |
| "loss": 0.4492, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 1.6177777777777776e-08, | |
| "loss": 0.4362, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 1.5622222222222224e-08, | |
| "loss": 0.5311, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 1.5066666666666666e-08, | |
| "loss": 0.4447, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 1.4511111111111111e-08, | |
| "loss": 0.4059, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.3955555555555554e-08, | |
| "loss": 0.3611, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.3400000000000001e-08, | |
| "loss": 0.3626, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 1.2844444444444444e-08, | |
| "loss": 0.3532, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 1.228888888888889e-08, | |
| "loss": 0.3445, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 1.1733333333333333e-08, | |
| "loss": 0.3562, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1.1177777777777778e-08, | |
| "loss": 0.3941, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 1.0622222222222221e-08, | |
| "loss": 0.4311, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 1.0066666666666666e-08, | |
| "loss": 0.3903, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 9.51111111111111e-09, | |
| "loss": 0.3799, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 8.955555555555556e-09, | |
| "loss": 0.4114, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 8.4e-09, | |
| "loss": 0.5189, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 7.844444444444444e-09, | |
| "loss": 0.4149, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 7.288888888888889e-09, | |
| "loss": 0.4635, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 6.7333333333333324e-09, | |
| "loss": 0.5325, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 6.177777777777777e-09, | |
| "loss": 0.6428, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 5.6222222222222216e-09, | |
| "loss": 0.4416, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 5.0666666666666665e-09, | |
| "loss": 0.4396, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 4.511111111111111e-09, | |
| "loss": 0.5301, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 3.955555555555555e-09, | |
| "loss": 0.4305, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 3.4000000000000003e-09, | |
| "loss": 0.4067, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 2.8444444444444444e-09, | |
| "loss": 0.3553, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 2.288888888888889e-09, | |
| "loss": 0.3609, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 1.7333333333333331e-09, | |
| "loss": 0.3544, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 1.1777777777777777e-09, | |
| "loss": 0.3452, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 6.222222222222222e-10, | |
| "loss": 0.3595, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 6.666666666666666e-11, | |
| "loss": 0.3899, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 0.37025901675224304, | |
| "eval_runtime": 78.3836, | |
| "eval_samples_per_second": 2.552, | |
| "eval_steps_per_second": 0.089, | |
| "eval_wer": 90.98360655737704, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 5000, | |
| "total_flos": 9.22809534603264e+19, | |
| "train_loss": 0.48666157550811767, | |
| "train_runtime": 34620.2227, | |
| "train_samples_per_second": 9.243, | |
| "train_steps_per_second": 0.144 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 9.22809534603264e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |