| accelerator_kwargs/logging_dir: /content/results/R-best-fine-tuned-bart-base-full-ft-reward_short_sentences_and_words-2023-07-13T06-49-08/runs | |
| adap_kl_ctrl: true | |
| batch_size: 64 | |
| cliprange: 0.2 | |
| cliprange_value: 0.2 | |
| compare_steps: 100 | |
| early_stopping: false | |
| forward_batch_size: null | |
| gamma: 1 | |
| gradient_accumulation_steps: 1 | |
| horizon: 10000 | |
| init_kl_coef: 0.2 | |
| lam: 0.95 | |
| learning_rate: 1.0e-06 | |
| log_with: tensorboard | |
| max_grad_norm: null | |
| mini_batch_size: 1 | |
| model_name: R-best-fine-tuned-bart-base-full-ft-reward_short_sentences_and_words-2023-07-13T06-49-08 | |
| optimize_cuda_cache: false | |
| ppo_epochs: 4 | |
| push_to_hub_if_best_kwargs/commit_message: 'add: new best model' | |
| push_to_hub_if_best_kwargs/private: true | |
| push_to_hub_if_best_kwargs/repo_id: nlp-lab-2023-seq2seq/R-best-fine-tuned-bart-base-full-ft-reward_short_sentences_and_words-2023-07-13T06-49-08 | |
| remove_unused_columns: true | |
| seed: 0 | |
| steps: 10000 | |
| target: 6 | |
| target_kl: 0.1 | |
| total_ppo_epochs: 157 | |
| tracker_project_name: trl | |
| vf_coef: 0.1 | |