Commit f21476f
authored
refactor(grpo): improve code style and add report generator (#37)
- Refactor reward_fn.py in pairwise/pointwise: convert comments to English,
unify code style (double quotes, formatting), remove unused imports
- Refactor chat_rl_dataset.py: improve code quality and formatting
- Add report_generator.py for zero-shot evaluation pipeline1 parent 57cca29 commit f21476f
File tree
4 files changed
+617
-387
lines changed- cookbooks
- training_judge_model/grpo
- pairwise
- pointwise
- zero_shot_evaluation
4 files changed
+617
-387
lines changed
0 commit comments