| { | |
| "best_metric": 42.9052871954476, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 21.0088, | |
| "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": 4.5999999999999995e-08, | |
| "loss": 1.8761, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.6e-08, | |
| "loss": 1.6558, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4599999999999998e-07, | |
| "loss": 1.606, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.96e-07, | |
| "loss": 1.5309, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.46e-07, | |
| "loss": 1.458, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9599999999999995e-07, | |
| "loss": 1.2538, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.4599999999999995e-07, | |
| "loss": 1.1572, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.96e-07, | |
| "loss": 1.115, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.46e-07, | |
| "loss": 1.1885, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.96e-07, | |
| "loss": 1.2714, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.46e-07, | |
| "loss": 1.1721, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.96e-07, | |
| "loss": 1.023, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.46e-07, | |
| "loss": 0.9135, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.959999999999999e-07, | |
| "loss": 0.8338, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.459999999999999e-07, | |
| "loss": 0.8033, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.96e-07, | |
| "loss": 0.7277, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.459999999999999e-07, | |
| "loss": 0.6727, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.96e-07, | |
| "loss": 0.7331, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.459999999999999e-07, | |
| "loss": 0.8357, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.959999999999999e-07, | |
| "loss": 0.9048, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.948888888888888e-07, | |
| "loss": 0.7653, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.893333333333332e-07, | |
| "loss": 0.7293, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.837777777777778e-07, | |
| "loss": 0.6546, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.782222222222222e-07, | |
| "loss": 0.6598, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.726666666666666e-07, | |
| "loss": 0.597, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.67111111111111e-07, | |
| "loss": 0.5606, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.615555555555556e-07, | |
| "loss": 0.5619, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.559999999999998e-07, | |
| "loss": 0.6677, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.504444444444443e-07, | |
| "loss": 0.7758, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.448888888888888e-07, | |
| "loss": 0.6975, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.393333333333334e-07, | |
| "loss": 0.6585, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.337777777777778e-07, | |
| "loss": 0.5934, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.282222222222222e-07, | |
| "loss": 0.5724, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.226666666666666e-07, | |
| "loss": 0.5531, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9.171111111111111e-07, | |
| "loss": 0.5142, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.115555555555555e-07, | |
| "loss": 0.4758, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.06e-07, | |
| "loss": 0.5603, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 9.004444444444445e-07, | |
| "loss": 0.6599, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.948888888888888e-07, | |
| "loss": 0.692, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.893333333333333e-07, | |
| "loss": 0.6046, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "eval_loss": 0.6535120606422424, | |
| "eval_runtime": 2254.6535, | |
| "eval_samples_per_second": 3.421, | |
| "eval_steps_per_second": 0.107, | |
| "eval_wer": 52.30842894237675, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.837777777777777e-07, | |
| "loss": 0.5742, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.782222222222222e-07, | |
| "loss": 0.5186, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.726666666666666e-07, | |
| "loss": 0.5179, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.671111111111111e-07, | |
| "loss": 0.4702, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.615555555555555e-07, | |
| "loss": 0.4501, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.559999999999999e-07, | |
| "loss": 0.4748, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 8.504444444444444e-07, | |
| "loss": 0.5663, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 8.448888888888888e-07, | |
| "loss": 0.6569, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.393333333333334e-07, | |
| "loss": 0.5742, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.337777777777777e-07, | |
| "loss": 0.5533, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 8.282222222222222e-07, | |
| "loss": 0.4962, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 8.226666666666666e-07, | |
| "loss": 0.4929, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 8.171111111111111e-07, | |
| "loss": 0.458, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 8.115555555555556e-07, | |
| "loss": 0.4306, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.06e-07, | |
| "loss": 0.4087, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.004444444444444e-07, | |
| "loss": 0.496, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 7.948888888888888e-07, | |
| "loss": 0.5851, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.893333333333333e-07, | |
| "loss": 0.5989, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.837777777777777e-07, | |
| "loss": 0.5241, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.782222222222222e-07, | |
| "loss": 0.4943, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.726666666666666e-07, | |
| "loss": 0.4531, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.67111111111111e-07, | |
| "loss": 0.4567, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.615555555555555e-07, | |
| "loss": 0.4075, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.559999999999999e-07, | |
| "loss": 0.3831, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.504444444444444e-07, | |
| "loss": 0.427, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 7.448888888888889e-07, | |
| "loss": 0.5049, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.393333333333333e-07, | |
| "loss": 0.605, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.337777777777778e-07, | |
| "loss": 0.5045, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.282222222222222e-07, | |
| "loss": 0.4905, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.226666666666667e-07, | |
| "loss": 0.4406, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.171111111111111e-07, | |
| "loss": 0.4399, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 7.115555555555556e-07, | |
| "loss": 0.4001, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 7.059999999999999e-07, | |
| "loss": 0.3802, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 7.004444444444444e-07, | |
| "loss": 0.3682, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 6.948888888888889e-07, | |
| "loss": 0.4575, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 6.893333333333333e-07, | |
| "loss": 0.5417, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.837777777777778e-07, | |
| "loss": 0.5345, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.782222222222221e-07, | |
| "loss": 0.4683, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.726666666666666e-07, | |
| "loss": 0.4498, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.67111111111111e-07, | |
| "loss": 0.4037, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "eval_loss": 0.5706056952476501, | |
| "eval_runtime": 2239.3323, | |
| "eval_samples_per_second": 3.445, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 46.687906300068136, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.615555555555555e-07, | |
| "loss": 0.4119, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.56e-07, | |
| "loss": 0.3656, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 6.504444444444445e-07, | |
| "loss": 0.3423, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 6.448888888888889e-07, | |
| "loss": 0.3894, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 6.393333333333333e-07, | |
| "loss": 0.4784, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 6.337777777777778e-07, | |
| "loss": 0.556, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 6.282222222222222e-07, | |
| "loss": 0.4571, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 6.226666666666667e-07, | |
| "loss": 0.444, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 6.171111111111111e-07, | |
| "loss": 0.3993, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 6.115555555555555e-07, | |
| "loss": 0.4016, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 6.06e-07, | |
| "loss": 0.3666, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 6.004444444444444e-07, | |
| "loss": 0.3405, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 5.948888888888889e-07, | |
| "loss": 0.3427, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 5.893333333333333e-07, | |
| "loss": 0.4299, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.837777777777777e-07, | |
| "loss": 0.5153, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.782222222222221e-07, | |
| "loss": 0.4744, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.726666666666666e-07, | |
| "loss": 0.4376, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.671111111111111e-07, | |
| "loss": 0.4021, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.615555555555555e-07, | |
| "loss": 0.3848, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.560000000000001e-07, | |
| "loss": 0.3765, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.504444444444444e-07, | |
| "loss": 0.3383, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 5.448888888888889e-07, | |
| "loss": 0.3122, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 5.393333333333333e-07, | |
| "loss": 0.376, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 5.337777777777778e-07, | |
| "loss": 0.4585, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 5.282222222222223e-07, | |
| "loss": 0.5083, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 5.226666666666666e-07, | |
| "loss": 0.4252, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.171111111111111e-07, | |
| "loss": 0.4106, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 5.115555555555555e-07, | |
| "loss": 0.3676, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 5.06e-07, | |
| "loss": 0.3751, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 5.004444444444444e-07, | |
| "loss": 0.3376, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.948888888888888e-07, | |
| "loss": 0.3141, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.893333333333333e-07, | |
| "loss": 0.3248, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 4.837777777777777e-07, | |
| "loss": 0.4109, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.782222222222222e-07, | |
| "loss": 0.4945, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.7266666666666664e-07, | |
| "loss": 0.4353, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.6711111111111104e-07, | |
| "loss": 0.4115, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.6155555555555555e-07, | |
| "loss": 0.3743, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.56e-07, | |
| "loss": 0.3584, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.504444444444444e-07, | |
| "loss": 0.3458, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.4488888888888887e-07, | |
| "loss": 0.3172, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "eval_loss": 0.5368593335151672, | |
| "eval_runtime": 2255.6569, | |
| "eval_samples_per_second": 3.42, | |
| "eval_steps_per_second": 0.107, | |
| "eval_wer": 44.1041601443804, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 4.393333333333333e-07, | |
| "loss": 0.2999, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 4.3377777777777773e-07, | |
| "loss": 0.3642, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 4.282222222222222e-07, | |
| "loss": 0.4459, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 4.226666666666667e-07, | |
| "loss": 0.4701, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 4.171111111111111e-07, | |
| "loss": 0.4027, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 4.1155555555555555e-07, | |
| "loss": 0.3833, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 4.06e-07, | |
| "loss": 0.3552, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 4.004444444444444e-07, | |
| "loss": 0.3504, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.9488888888888887e-07, | |
| "loss": 0.3143, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.8933333333333327e-07, | |
| "loss": 0.2925, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.837777777777778e-07, | |
| "loss": 0.3127, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 3.7822222222222224e-07, | |
| "loss": 0.4049, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 3.7266666666666664e-07, | |
| "loss": 0.4758, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.671111111111111e-07, | |
| "loss": 0.404, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.6155555555555555e-07, | |
| "loss": 0.3941, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.5599999999999996e-07, | |
| "loss": 0.3497, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.504444444444444e-07, | |
| "loss": 0.3466, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 3.4488888888888887e-07, | |
| "loss": 0.3217, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 3.3933333333333333e-07, | |
| "loss": 0.3017, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.337777777777778e-07, | |
| "loss": 0.2846, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.2822222222222224e-07, | |
| "loss": 0.3554, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 3.2266666666666664e-07, | |
| "loss": 0.4441, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 3.171111111111111e-07, | |
| "loss": 0.4407, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 3.115555555555555e-07, | |
| "loss": 0.384, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 3.0599999999999996e-07, | |
| "loss": 0.3663, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 3.0044444444444447e-07, | |
| "loss": 0.3374, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.948888888888889e-07, | |
| "loss": 0.3356, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.8933333333333333e-07, | |
| "loss": 0.2994, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.837777777777778e-07, | |
| "loss": 0.2773, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.782222222222222e-07, | |
| "loss": 0.3088, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 2.7266666666666665e-07, | |
| "loss": 0.3914, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.671111111111111e-07, | |
| "loss": 0.4613, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.615555555555555e-07, | |
| "loss": 0.3927, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.56e-07, | |
| "loss": 0.3701, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.5044444444444447e-07, | |
| "loss": 0.3378, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.448888888888889e-07, | |
| "loss": 0.3389, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 2.3933333333333333e-07, | |
| "loss": 0.3054, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 2.3377777777777776e-07, | |
| "loss": 0.2843, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.282222222222222e-07, | |
| "loss": 0.2818, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.2266666666666668e-07, | |
| "loss": 0.3606, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "eval_loss": 0.5218492746353149, | |
| "eval_runtime": 2234.5827, | |
| "eval_samples_per_second": 3.452, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 43.07655475958085, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 2.171111111111111e-07, | |
| "loss": 0.4345, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 2.1155555555555554e-07, | |
| "loss": 0.4174, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 2.06e-07, | |
| "loss": 0.3716, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 2.0044444444444445e-07, | |
| "loss": 0.3474, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 1.9488888888888888e-07, | |
| "loss": 0.3285, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.893333333333333e-07, | |
| "loss": 0.3215, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.837777777777778e-07, | |
| "loss": 0.2915, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.7822222222222222e-07, | |
| "loss": 0.2677, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.7266666666666665e-07, | |
| "loss": 0.3108, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 1.6711111111111108e-07, | |
| "loss": 0.3961, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.6155555555555556e-07, | |
| "loss": 0.4526, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.56e-07, | |
| "loss": 0.3725, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.5044444444444442e-07, | |
| "loss": 0.3579, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.448888888888889e-07, | |
| "loss": 0.3286, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 1.3933333333333334e-07, | |
| "loss": 0.3296, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 1.3377777777777777e-07, | |
| "loss": 0.2964, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 1.282222222222222e-07, | |
| "loss": 0.2744, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 1.2266666666666665e-07, | |
| "loss": 0.2787, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 1.1711111111111111e-07, | |
| "loss": 0.36, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 1.1155555555555555e-07, | |
| "loss": 0.4377, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 1.06e-07, | |
| "loss": 0.3993, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 1.0044444444444444e-07, | |
| "loss": 0.3686, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 9.488888888888888e-08, | |
| "loss": 0.3361, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 8.933333333333333e-08, | |
| "loss": 0.3169, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 8.377777777777778e-08, | |
| "loss": 0.3172, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 7.822222222222221e-08, | |
| "loss": 0.2805, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 7.266666666666667e-08, | |
| "loss": 0.2652, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 6.71111111111111e-08, | |
| "loss": 0.3207, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 6.155555555555556e-08, | |
| "loss": 0.4025, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 5.6e-08, | |
| "loss": 0.427, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 5.044444444444444e-08, | |
| "loss": 0.3717, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 4.4888888888888885e-08, | |
| "loss": 0.3525, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.933333333333333e-08, | |
| "loss": 0.3217, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 3.377777777777777e-08, | |
| "loss": 0.3287, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 2.822222222222222e-08, | |
| "loss": 0.2816, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 2.2666666666666668e-08, | |
| "loss": 0.2723, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 20.04, | |
| "learning_rate": 1.711111111111111e-08, | |
| "loss": 0.2837, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 20.05, | |
| "learning_rate": 1.1555555555555554e-08, | |
| "loss": 0.3654, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 6e-09, | |
| "loss": 0.4403, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 4.4444444444444443e-10, | |
| "loss": 0.3792, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "eval_loss": 0.5171416997909546, | |
| "eval_runtime": 2140.854, | |
| "eval_samples_per_second": 3.603, | |
| "eval_steps_per_second": 0.113, | |
| "eval_wer": 42.9052871954476, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "step": 5000, | |
| "total_flos": 2.073068169265152e+19, | |
| "train_loss": 0.4945518872261047, | |
| "train_runtime": 85569.1464, | |
| "train_samples_per_second": 3.74, | |
| "train_steps_per_second": 0.058 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 2.073068169265152e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |