| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.345794392523365, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002719626168224299, | |
| "loss": 2.3351, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.9539090394973755, | |
| "eval_rouge1": 0.3928238565889721, | |
| "eval_rouge2": 0.24457675199617848, | |
| "eval_rougeL": 0.34443833165269067, | |
| "eval_rougeLsum": 0.35377968284169653, | |
| "eval_runtime": 113.4757, | |
| "eval_samples_per_second": 2.503, | |
| "eval_steps_per_second": 0.626, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0002439252336448598, | |
| "loss": 1.8616, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.8454691171646118, | |
| "eval_rouge1": 0.40135907614173977, | |
| "eval_rouge2": 0.2537731004498688, | |
| "eval_rougeL": 0.3517340305952086, | |
| "eval_rougeLsum": 0.3599553733909089, | |
| "eval_runtime": 90.5265, | |
| "eval_samples_per_second": 3.137, | |
| "eval_steps_per_second": 0.784, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0002158878504672897, | |
| "loss": 1.5474, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.7911098003387451, | |
| "eval_rouge1": 0.40516907769034227, | |
| "eval_rouge2": 0.2571882790410479, | |
| "eval_rougeL": 0.3551430134650718, | |
| "eval_rougeLsum": 0.3650237424768551, | |
| "eval_runtime": 89.8747, | |
| "eval_samples_per_second": 3.16, | |
| "eval_steps_per_second": 0.79, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 0.0001878504672897196, | |
| "loss": 1.3416, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.7877225875854492, | |
| "eval_rouge1": 0.39727138204448276, | |
| "eval_rouge2": 0.2553111741867333, | |
| "eval_rougeL": 0.35344418510391534, | |
| "eval_rougeLsum": 0.3624411592522736, | |
| "eval_runtime": 91.2462, | |
| "eval_samples_per_second": 3.112, | |
| "eval_steps_per_second": 0.778, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 0.0001598130841121495, | |
| "loss": 1.1667, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 1.8065179586410522, | |
| "eval_rouge1": 0.40226039832787347, | |
| "eval_rouge2": 0.25889093132847524, | |
| "eval_rougeL": 0.3547361470543734, | |
| "eval_rougeLsum": 0.3642271784388902, | |
| "eval_runtime": 88.6588, | |
| "eval_samples_per_second": 3.203, | |
| "eval_steps_per_second": 0.801, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 0.0001317757009345794, | |
| "loss": 1.0582, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 1.8586112260818481, | |
| "eval_rouge1": 0.3957092352506892, | |
| "eval_rouge2": 0.2538965597420449, | |
| "eval_rougeL": 0.34916932912187554, | |
| "eval_rougeLsum": 0.3576929636444268, | |
| "eval_runtime": 95.986, | |
| "eval_samples_per_second": 2.959, | |
| "eval_steps_per_second": 0.74, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 0.00010373831775700933, | |
| "loss": 0.9202, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_loss": 1.8837732076644897, | |
| "eval_rouge1": 0.40251597070955947, | |
| "eval_rouge2": 0.2583209858085714, | |
| "eval_rougeL": 0.3540804774162999, | |
| "eval_rougeLsum": 0.36370971781604133, | |
| "eval_runtime": 94.4477, | |
| "eval_samples_per_second": 3.007, | |
| "eval_steps_per_second": 0.752, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 7.570093457943924e-05, | |
| "loss": 0.8575, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 1.9066342115402222, | |
| "eval_rouge1": 0.39449493230029453, | |
| "eval_rouge2": 0.2549659416128175, | |
| "eval_rougeL": 0.348120528239499, | |
| "eval_rougeLsum": 0.3568722790464366, | |
| "eval_runtime": 82.6343, | |
| "eval_samples_per_second": 3.437, | |
| "eval_steps_per_second": 0.859, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 4.766355140186915e-05, | |
| "loss": 0.7691, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_loss": 1.970426321029663, | |
| "eval_rouge1": 0.393735400589476, | |
| "eval_rouge2": 0.25521942254430097, | |
| "eval_rougeL": 0.34772527093463, | |
| "eval_rougeLsum": 0.356885126635194, | |
| "eval_runtime": 88.6133, | |
| "eval_samples_per_second": 3.205, | |
| "eval_steps_per_second": 0.801, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 1.9626168224299062e-05, | |
| "loss": 0.7361, | |
| "step": 1000 | |
| } | |
| ], | |
| "max_steps": 1070, | |
| "num_train_epochs": 10, | |
| "total_flos": 370753309016064.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |