| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 9365, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.932194340630005e-05, | |
| "loss": 1.2388, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_gen_len": 48.53731343283582, | |
| "eval_loss": 1.1303935050964355, | |
| "eval_rouge1": 66.5164, | |
| "eval_rouge2": 54.1964, | |
| "eval_rougeL": 60.1255, | |
| "eval_rougeLsum": 62.6329, | |
| "eval_runtime": 248.0848, | |
| "eval_samples_per_second": 5.401, | |
| "eval_steps_per_second": 0.27, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8643886812600106e-05, | |
| "loss": 0.9911, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "eval_gen_len": 48.21940298507463, | |
| "eval_loss": 1.0996825695037842, | |
| "eval_rouge1": 66.7923, | |
| "eval_rouge2": 54.4921, | |
| "eval_rougeL": 60.3864, | |
| "eval_rougeLsum": 63.0007, | |
| "eval_runtime": 245.1086, | |
| "eval_samples_per_second": 5.467, | |
| "eval_steps_per_second": 0.273, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.7965830218900163e-05, | |
| "loss": 0.8473, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "eval_gen_len": 51.417164179104475, | |
| "eval_loss": 1.1019301414489746, | |
| "eval_rouge1": 68.1867, | |
| "eval_rouge2": 55.4038, | |
| "eval_rougeL": 61.5061, | |
| "eval_rougeLsum": 64.2051, | |
| "eval_runtime": 255.7993, | |
| "eval_samples_per_second": 5.238, | |
| "eval_steps_per_second": 0.262, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 7.287773625200214e-06, | |
| "loss": 0.7315, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "eval_gen_len": 50.42686567164179, | |
| "eval_loss": 1.111816167831421, | |
| "eval_rouge1": 67.961, | |
| "eval_rouge2": 55.3863, | |
| "eval_rougeL": 61.4022, | |
| "eval_rougeLsum": 64.1305, | |
| "eval_runtime": 261.9465, | |
| "eval_samples_per_second": 5.116, | |
| "eval_steps_per_second": 0.256, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 9365, | |
| "total_flos": 1.1949135422717952e+17, | |
| "train_loss": 0.9121758847458122, | |
| "train_runtime": 3599.8478, | |
| "train_samples_per_second": 52.03, | |
| "train_steps_per_second": 2.601 | |
| } | |
| ], | |
| "max_steps": 9365, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.1949135422717952e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |