Enable qwen2-vl multimodal input on v0.6.1

[hzjane]

Refer to this issue. We need to remove self.rope_scaling["type"] = "default". "fp8" is recommended to use.

pip install transformers==4.45.1
# vim /usr/local/lib/python3.11/dist-packages/transformers/models/qwen2_vl/configuration_qwen2_vl.py line 241
if self.rope_scaling["type"] == "mrope":
    #self.rope_scaling["type"] = "default"
    pass

online test

curl http://localhost:7999/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen2-VL-7B-Instruct",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "What is in the image?"
          },
          {
            "type": "image_url",
            "image_url": {
              "url": "http://farm6.staticflickr.com/5268/5602445367_3504763978_z.jpg"
            }
          }
        ]
      }
    ],
    "max_tokens": 128,
    "temperature": 0.1,
    "top_p": 0.001,
    "repetition_penalty": 1.05
  }'

If encounter duplicate output problems, try to increaserepetition_penalty to 1.99.

[xiangyuT]

LGTM

[guang11644331]

it's work ^_^!