fix: Handle multiple named chat templates in HuggingFace tokenizers

[JustinTong0323]

Motivation

Fix #17231
SGLang crashes with TypeError: expected string or bytes-like object, got 'dict' when loading models whose HuggingFace tokenizers have multiple named chat templates (e.g., CohereLabs/c4ai-command-r7b-12-2024).

HuggingFace tokenizers can store chat templates in two formats:

• Single template: tokenizer.chat_template = "..."
• Multiple named templates: tokenizer.chat_template = {"default": "...", "tool_use": "...", "rag": "..."}

The current implementation only handles the single template format, causing crashes when encountering the dict format.

Error trace:

File "/sglang/python/sglang/srt/managers/template_manager.py", line 94, in _detect_reasoning_pattern
    has_reasoning = re.search(force_reasoning_pattern, template) is not None
TypeError: expected string or bytes-like object, got 'dict'

Usage

python -m sglang.launch_server \
    --model-path CohereLabs/c4ai-command-r7b-12-2024 \
    --trust-remote-code \
    --hf-chat-template-name tool_use

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review Process

1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
4. After green CI and required approvals, ask Merge Oncalls to merge.

[gemini-code-assist]

Summary of Changes

Hello @JustinTong0323, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the system's compatibility with HuggingFace tokenizers by addressing a critical bug where models using dictionary-based chat templates would cause a TypeError. It introduces robust logic to correctly identify and load these multi-template configurations, providing users with the flexibility to specify a preferred template via a new command-line argument, --hf-chat-template-name. This ensures smoother model loading and operation for a wider range of HuggingFace models.

Highlights

• Bug Fix for Chat Templates: Resolves a TypeError that occurred when HuggingFace tokenizers provided chat templates as a dictionary of named templates instead of a single string.
• Support for Multiple Named Templates: Enables the system to correctly parse and utilize HuggingFace tokenizers that define multiple chat templates (e.g., 'default', 'tool_use', 'rag').
• New CLI Argument: Introduces --hf-chat-template-name to allow users to explicitly select a specific named chat template from those available in the tokenizer.
• Default Template Selection: If --hf-chat-template-name is not specified, the system will now automatically use the first available chat template found in the tokenizer's dictionary.
• Documentation Update: The server_arguments.md documentation has been updated to reflect the new --hf-chat-template-name argument and its usage.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[JustinTong0323]

/tag-and-rerun-ci

[gemini-code-assist]

Code Review

This pull request effectively addresses a crash that occurs when loading HuggingFace tokenizers with multiple named chat templates. The introduction of the --hf-chat-template-name argument is a good solution, and the implementation is clean, including necessary documentation updates. My feedback focuses on enhancing the default behavior to be more predictable by prioritizing the 'default' template and ensuring the documentation is fully consistent with this improvement.

[gemini-code-assist]

The current fallback logic in _select_named_template uses the first available template. However, the help text for --hf-chat-template-name in server_args.py specifies that it should prefer the 'default' template if it exists. It's good practice to align the implementation with the documentation. Prioritizing the 'default' template is a more predictable and user-friendly behavior.

Suggested change

	# Fallback: Use first available template
	first_name = available_names[0]
	logger.info(f"Using first available template: '{first_name}'")
	return templates[first_name]
	# Fallback: Use 'default' if available, otherwise the first available template
	if "default" in templates:
	logger.info("Using 'default' chat template.")
	return templates["default"]
	first_name = available_names[0]
	logger.info(f"No 'default' template found. Using first available template: '{first_name}'")
	return templates[first_name]

[ljw-mc]

Just tested on the original issue, the commands work!

Thank you so much! @JustinTong0323

[JustinTong0323]

/tag-and-rerun-ci