| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9715578539107952, | |
| "eval_steps": 500, | |
| "global_step": 6100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8383968972204266e-05, | |
| "loss": 0.144, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6767937944408533e-05, | |
| "loss": 0.137, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.51519069166128e-05, | |
| "loss": 0.1307, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.3535875888817068e-05, | |
| "loss": 0.1263, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1919844861021333e-05, | |
| "loss": 0.1241, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0303813833225597e-05, | |
| "loss": 0.12, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_gen_len": 18.2165, | |
| "eval_loss": 0.06346726417541504, | |
| "eval_rouge1": 0.1191, | |
| "eval_rouge2": 0.0373, | |
| "eval_rougeL": 0.1197, | |
| "eval_rougeLsum": 0.1193, | |
| "eval_runtime": 34.2167, | |
| "eval_samples_per_second": 58.451, | |
| "eval_steps_per_second": 0.935, | |
| "step": 3094 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.687782805429864e-06, | |
| "loss": 0.1159, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.071751777634132e-06, | |
| "loss": 0.1147, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.4557207498383965e-06, | |
| "loss": 0.1122, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.839689722042664e-06, | |
| "loss": 0.111, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.2236586942469294e-06, | |
| "loss": 0.1104, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.076276664511959e-07, | |
| "loss": 0.1089, | |
| "step": 6000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 6188, | |
| "num_train_epochs": 2, | |
| "save_steps": 100, | |
| "total_flos": 1.1413435354310246e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |