Skip to content

<think> tags for thinking models #513

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
JoelNiklaus opened this issue Jan 24, 2025 · 2 comments · May be fixed by #617
Open

<think> tags for thinking models #513

JoelNiklaus opened this issue Jan 24, 2025 · 2 comments · May be fixed by #617
Labels
feature/enhancement New feature/request

Comments

@JoelNiklaus
Copy link
Contributor

Thinking models like DeepSeek-R1 emit thinking tags in the output. Is there a way to filter these out easily? Currently they make it directly into the output and so mess up metrics.
@clefourrier
Copy link
Member

Nope, not at the moment! We used to have regex parsers but they were underused so we removed them.

@mapmeld
Copy link
Contributor

mapmeld commented Mar 14, 2025

I had the same question and got something working. For now it's more of a hack, but hopefully this is a starting point to get it working generally. Here is my branch and demo notebook.

Notes:

  • could add a flag (similar to use_chat_template). I saw some code uses add_reasoning_prompt
  • <think> is already in tokenizer_config.json's chat_template field, so maybe instead of hardcoding, possible to detect common tokens there, and then insert it in the chat template again?
  • I needed to change my answer options from ["A", "B"...] to ["The answer is A", ...]
  • 2048 tokens seems like the right size
  • didn't test few-shot
  • this isn't maximally efficient because it works on one doc at a time, but I don't know how well large batches would fit

Here's the key section in prompt_manager:

        elif use_chat_template:
            chat_preview = self.model.tokenizer.apply_chat_template(
                output, tokenize=False, add_generation_prompt=True
            )
            tokenized = self.model.tokenizer(chat_preview, return_tensors="pt").to(self.model.device)
            prepared_batch = Batch(
                input_ids=tokenized["input_ids"],
                input_mask=tokenized["attention_mask"],
                input_lengths=[len(tokenized["input_ids"][0])],
                truncated=[False],
                padded=[False],
            )
            response = self.model._generate(
                batch=prepared_batch,
                max_new_tokens=2048,
                stop_tokens=["</think>"],
            )
            all_start = chat_preview + response[0].result[0] + "</think>"
            return all_start, num_effective_fewshots

@mapmeld mapmeld linked a pull request Mar 17, 2025 that will close this issue
Open
@NathanHB NathanHB added the feature/enhancement New feature/request label May 15, 2025
@NathanHB NathanHB linked a pull request May 15, 2025 that will close this issue
Add CLI arg generate_until_token to support reasoning and CoT models #617
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
feature/enhancement New feature/request
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Add CLI arg generate_until_token to support reasoning and CoT models mapmeld/lighteval-tasks
4 participants
@mapmeld @JoelNiklaus @clefourrier @NathanHB