[Bug]: 使用vllm进行推理时，设置parallel_tool_calls似乎不生效。我想实现单工具调用，应该怎么设置呢

[RayneSun]

Model Series

Qwen2.5

What are the models used?

Qwen2.5-72B-Instruction

What is the scenario where the problem happened?

vllm

Is this a known issue?

• I have followed the GitHub README.
• I have checked the Qwen documentation and cannot find an answer there.
• I have checked the documentation of the related framework and cannot find useful information.
• I have searched the issues and there is not a similar one.

Information about environment

vllm>0.0.0

Log output

curl --location --request POST '***' \
--header 'User-Agent: Apifox/1.0.0 (https://apifox.com)' \
--header 'Content-Type: application/json' \
--data-raw '{
  "model": "Qwen2.5",
  "stream": true,
  "parallel_function_calls":false,
  "messages": [
    {
      "role": "user",
      "content": "查一下西安和北京的天气"
    }
  ],
  "stream_options":{"include_usage": true},
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "查天气",
        "description": "根据城市名查询天气",
        "parameters": {
          "properties": {
            "city": {
              "type": "string",
              "description": "城市名"
            }
          },
          "type": "object"
        }
      }
    }
  ]
}'

I got 2 call rather than 1.

Description

Steps to reproduce

This happens to Qwen2.5-72B-Instruct
The problem can be reproduced with the following steps:
curl --location --request POST '*****'
--header 'User-Agent: Apifox/1.0.0 (https://apifox.com)'
--header 'Content-Type: application/json'
--data-raw '{
"model": "Qwen2.5",
"stream": true,
"parallel_function_calls":false,
"messages": [
{
"role": "user",
"content": "查一下西安和北京的天气"
}
],
"stream_options":{"include_usage": true},
"tools": [
{
"type": "function",
"function": {
"name": "查天气",
"description": "根据城市名查询天气",
"parameters": {
"properties": {
"city": {
"type": "string",
"description": "城市名"
}
},
"type": "object"
}
}
}
]
}'

Expected results

The results are expected to be call one tools

Attempts to fix

I have tried several ways to fix this, including:

1. make parallel_tool_calls usable

[RayneSun]

我想关闭模型并行调用，因为会出现严重的幻觉，请问怎么能关闭并行调用呢

[jklj077]

Not supported by vllm; feature request at vllm-project/vllm#9451

[RayneSun]

How can I solve it through myself, just like give model a system message. I think 'parallel_tool_calls' might change the input of the model, maybe I could change it to solve this promblem!

[endNone]

@RayneSun Which tool parser is used for deploying Qwen with vLLM?

[RayneSun]

Hermes