vllm icon indicating copy to clipboard operation
vllm copied to clipboard

[ToolParser] Add Qwen2.5 models tool parser

Open xiayouran opened this issue 1 year ago • 1 comments

Add Qwen2.5 models tool parser, support model: Qwen2.5-1.5B-Instruct, Qwen2.5-7B-Instruct, Qwen2.5-32B-Instruct-GPTQ-Int4, etc.

xiayouran avatar Feb 18 '25 15:02 xiayouran

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

github-actions[bot] avatar Feb 18 '25 15:02 github-actions[bot]

@ywang96 @DarkLight1337 @youkaichao Bros, take a look at this PR.

xiayouran avatar Feb 26 '25 06:02 xiayouran

@K-Mistele can you help review this? cc @mgoin

DarkLight1337 avatar Feb 26 '25 06:02 DarkLight1337

Qwen 2.5 uses the Hermes tool-calling standard - this format is already supported as hermes, more details on it in the vLLM tool calling docs.

Per the Qwen 2.5 release blog post, the Qwen team recommends using the existing parser for hermes that I wrote:

vllm serve Qwen/Qwen2.5-7B-Instruct --enable-auto-tool-choice --tool-call-parser hermes

So, I don't think we need to add a separate parser for qwen @DarkLight1337

K-Mistele avatar Feb 26 '25 16:02 K-Mistele