[Bug] Qwen2.5无法跑通tools call(官方案例代码)

[turkeymz]

Checklist

• 1. I have searched related issues but cannot get the expected help.
• 2. The bug has not been fixed in the latest version.
• 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.

Describe the bug

使用lmdeploy部署Qwen2.5无法跑官方tools call的案例。返回结果的message不会转换到tools中。
但相同的代码，如果使用vllm部署，模型能自动选择tools。

Reproduction

lmdeploy部署命令

lmdeploy serve api_server /data/Qwen/Qwen2.5-32B-Instruct --server-port 8008 --model-name qwen2d5 --enable-prefix-caching

vllm部署命令

python3 -m vllm.entrypoints.openai.api_server  --model /data/Qwen/Qwen2.5-32B-Instruct  --served-model-name qwen2d5  --max_model_len 20000 --enable-auto-tool-choice --tool-call-parser hermes --enable-prefix-caching  --port 8008

Environment

OS: ubuntu
python: 3.10
cuda: 12.4
torch：2.4.0
lmdeploy: 0.6.3
vllm: 0.6.1.post2

Error traceback

No response

[lvhan028]

qwen2.5 tools call 的功能最近才合入，尚未发版。
可以参考 https://lmdeploy.readthedocs.io/en/latest/get_started/installation.html 安装源码后再尝试

[akai-shuuichi]

可以尝试https://github.com/zhyncs/lmdeploy-build/releases/tag/178ec7b 这个版本包含了toolCall