| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 49.9937106918239, | |
| "global_step": 3950, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.849999999999999e-06, | |
| "loss": 18.5969, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 9.849999999999999e-06, | |
| "loss": 4.2811, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.485e-05, | |
| "loss": 3.9192, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 1.9849999999999998e-05, | |
| "loss": 2.7895, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 2.4849999999999998e-05, | |
| "loss": 2.142, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "learning_rate": 2.985e-05, | |
| "loss": 1.9161, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 3.485e-05, | |
| "loss": 1.7888, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 10.13, | |
| "learning_rate": 3.984999999999999e-05, | |
| "loss": 1.7494, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 11.39, | |
| "learning_rate": 4.484999999999999e-05, | |
| "loss": 1.7129, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "learning_rate": 4.984999999999999e-05, | |
| "loss": 1.7019, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "eval_cer": 0.2589387701435627, | |
| "eval_loss": 1.051009178161621, | |
| "eval_runtime": 302.233, | |
| "eval_samples_per_second": 14.777, | |
| "eval_steps_per_second": 1.85, | |
| "eval_wer": 0.9832177220854778, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 13.92, | |
| "learning_rate": 5.484999999999999e-05, | |
| "loss": 1.6797, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 15.19, | |
| "learning_rate": 5.985e-05, | |
| "loss": 1.6992, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 16.45, | |
| "learning_rate": 6.484999999999999e-05, | |
| "loss": 1.703, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 17.72, | |
| "learning_rate": 6.984999999999999e-05, | |
| "loss": 1.7304, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 18.98, | |
| "learning_rate": 7.484999999999999e-05, | |
| "loss": 1.7436, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 20.25, | |
| "learning_rate": 7.203061224489795e-05, | |
| "loss": 1.7564, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 21.52, | |
| "learning_rate": 6.896938775510203e-05, | |
| "loss": 1.6982, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 22.78, | |
| "learning_rate": 6.590816326530612e-05, | |
| "loss": 1.7184, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 24.05, | |
| "learning_rate": 6.28469387755102e-05, | |
| "loss": 1.692, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 25.31, | |
| "learning_rate": 5.978571428571428e-05, | |
| "loss": 1.6385, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 25.31, | |
| "eval_cer": 0.18507013781031975, | |
| "eval_loss": 0.6670215129852295, | |
| "eval_runtime": 300.7027, | |
| "eval_samples_per_second": 14.852, | |
| "eval_steps_per_second": 1.859, | |
| "eval_wer": 0.9914969791899754, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 26.58, | |
| "learning_rate": 5.6724489795918356e-05, | |
| "loss": 1.6092, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 27.84, | |
| "learning_rate": 5.3663265306122446e-05, | |
| "loss": 1.6209, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 29.11, | |
| "learning_rate": 5.060204081632652e-05, | |
| "loss": 1.5933, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 30.38, | |
| "learning_rate": 4.754081632653061e-05, | |
| "loss": 1.5673, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 31.64, | |
| "learning_rate": 4.447959183673469e-05, | |
| "loss": 1.5518, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 32.91, | |
| "learning_rate": 4.141836734693877e-05, | |
| "loss": 1.5398, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 34.18, | |
| "learning_rate": 3.8357142857142855e-05, | |
| "loss": 1.5095, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 35.44, | |
| "learning_rate": 3.529591836734693e-05, | |
| "loss": 1.508, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 36.7, | |
| "learning_rate": 3.2234693877551015e-05, | |
| "loss": 1.4769, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 37.97, | |
| "learning_rate": 2.9173469387755098e-05, | |
| "loss": 1.4344, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 37.97, | |
| "eval_cer": 0.17974508788797672, | |
| "eval_loss": 0.6182843446731567, | |
| "eval_runtime": 301.8085, | |
| "eval_samples_per_second": 14.797, | |
| "eval_steps_per_second": 1.852, | |
| "eval_wer": 1.0212575520250615, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 39.24, | |
| "learning_rate": 2.611224489795918e-05, | |
| "loss": 1.4331, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 40.5, | |
| "learning_rate": 2.3051020408163264e-05, | |
| "loss": 1.4042, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 41.77, | |
| "learning_rate": 2.0020408163265303e-05, | |
| "loss": 1.3822, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 43.04, | |
| "learning_rate": 1.6959183673469386e-05, | |
| "loss": 1.3768, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 44.3, | |
| "learning_rate": 1.3897959183673468e-05, | |
| "loss": 1.3447, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 45.57, | |
| "learning_rate": 1.0836734693877551e-05, | |
| "loss": 1.3359, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 46.83, | |
| "learning_rate": 7.77551020408163e-06, | |
| "loss": 1.3113, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 48.1, | |
| "learning_rate": 4.714285714285714e-06, | |
| "loss": 1.3053, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 49.36, | |
| "learning_rate": 1.6530612244897958e-06, | |
| "loss": 1.2802, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 49.99, | |
| "step": 3950, | |
| "total_flos": 2.372869475210465e+20, | |
| "train_loss": 2.1743111564539657, | |
| "train_runtime": 44274.9554, | |
| "train_samples_per_second": 11.465, | |
| "train_steps_per_second": 0.089 | |
| } | |
| ], | |
| "max_steps": 3950, | |
| "num_train_epochs": 50, | |
| "total_flos": 2.372869475210465e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |