o
TÃi£ ã @ s2 d dl Z deeeeef dee fdd„ZdS )é NÚ completionsÚreturnc s2 d‰ dd„ | D ƒ}‡ fdd„|D ƒ}dd„ |D ƒS )a¢
Reward function that checks if the reasoning process is enclosed within `""` and `""` tags. The
function returns a reward of 1.0 if the format is correct, otherwise 0.0.
Args:
completions (`list[list[dict[str, str]]]`):
List of completions to be evaluated. Each completion must be a list of one message, i.e. a dictionary
containing the key `"content"` with the value being the text of the completion.
**kwargs:
Additional keyword arguments. This function does not use them, but they are required in the function
signature to ensure compatibility with trainers like [`GRPOTrainer`].
Returns:
`list[float]`:
A list of rewards, where each reward is 1.0 if the completion matches the expected format, otherwise 0.0.
Example:
```python
>>> from trl.rewards import think_format_reward
>>> completions = [
... [{"content": "\nThis is my reasoning.\n\nThis is my answer."}],
... [{"content": "\nThis is my reasoning.\nThis is my answer."}],
... ]
>>> think_format_reward(completions)
[1.0, 0.0]
```
z%^(?!.*)(.*?).*$c S s g | ]}|d d ‘qS )r Úcontent© )Ú.0Ú
completionr r úN/home/ubuntu/.local/lib/python3.10/site-packages/trl/rewards/format_rewards.pyÚ
0 ó z'think_format_reward..c s" g | ]
}t ˆ |t jt jB ¡‘qS r )ÚreÚmatchÚDOTALLÚ MULTILINE)r r ©Úpatternr r r 1 s " c S s g | ]}|rd nd‘qS )g ð?g r )r r r r r r 2 r
r )r ÚkwargsÚcompletion_contentsÚmatchesr r r Úthink_format_reward s r )r ÚlistÚdictÚstrÚfloatr r r r r Ú s *