Qwen 2.5-vl support？

[Gusha-nye]

Hi guys, recently I tried to run Qwen2.5-vl-3b on intel gpu with inference using ipex framework, but it failed and reported the following error:

Can you tell me how this should be solved, or is the Qwen2.5-vl model not currently supported?

[cyita]

Hi @Gusha-nye , Qwen2.5-vl is not supported.

[xiangyuT]

Hi @Gusha-nye, we have supported IPEX-LLM-optimized Qwen2.5-VL models with vLLM. You could try with our docs for vLLM to get started.

[buffliu]

Hi @xiangyuT , could you tell how to send a video to Qwen2.5 VL as recognition?

[hzjane]

@buffliu You can add --allowed-local-media-path /llm/models/media on starting vllm service, and then you can send a vedio like:

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen2.5-VL-7B-Instruct",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
	    "text": "what is in the vedio?"
          },
          {
            "type": "video_url",
            "video_url": {
              "url": "file:/llm/models/media/test.mp4"
            }
          }
        ]
      }
    ],
    "max_tokens": 512
  }'

[buffliu]

@hzjane，thank you so much for your help.

[lalalapotter]

Issue resolved, close.