[tests] add qwen2_5_vl batch_infer test

[Jintao-Huang]

No description provided.

[gemini-code-assist]

Summary of Changes

Hello @Jintao-Huang, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the testing framework by replacing a single-instance test for the Qwen2.5-VL-7B-Instruct model with a comprehensive batch inference test. The new test validates the model's ability to efficiently process multiple inputs, including both images and videos, in a single batch, thereby confirming its robust multi-modal capabilities and improving test coverage for performance-critical scenarios.

Highlights

• Batch Inference Test for Qwen2.5-VL: A new test, test_qwen2_5_vl_batch_infer, has been introduced to specifically validate the batch inference capabilities of the Qwen2.5-VL-7B-Instruct model.
• Multi-modal Batch Processing: The updated test now includes scenarios for batch inference involving both image and video inputs, demonstrating the model's multi-modal processing efficiency.
• Test Suite Activation: The newly implemented batch inference test for Qwen2.5-VL is now actively enabled within the test suite, ensuring its regular execution.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request adds a new batch inference test for qwen2_5_vl. The new test is more comprehensive than the one it replaces. My review focuses on improving the readability and maintainability of the new test code by refactoring duplicated data and hardcoded values.

[gemini-code-assist]

The test function test_qwen2_5_vl_batch_infer can be refactored to improve readability and reduce code duplication.

• Hardcoded URLs and queries are used multiple times. They can be extracted into constants.
• The construction of infer_data for pt_engine.infer and messages for manual inference is repetitive and can be simplified by defining the test data once and reusing it.

Applying these changes will make the test easier to understand and maintain.

def test_qwen2_5_vl_batch_infer():
    from qwen_vl_utils import process_vision_info
    pt_engine = PtEngine('Qwen/Qwen2.5-VL-7B-Instruct', max_batch_size=2)
    request_config = RequestConfig(max_tokens=128, temperature=0)

    image_url = 'https://qianwen-res.oss-accelerate-overseas.aliyuncs.com/Qwen2-VL/demo_small.jpg'
    video_url = 'https://modelscope-open.oss-cn-hangzhou.aliyuncs.com/images/baby.mp4'
    image_query = 'What kind of dog is this?'
    video_query = 'describe the video.'

    infer_data = [{
        'messages': [{
            'role': 'user',
            'content': f'<image>{image_query}'
        }],
        'images': [image_url]
    }, {
        'messages': [{
            'role': 'user',
            'content': f'<video>{video_query}'
        }],
        'videos': [video_url]
    }]
    resp = pt_engine.infer(infer_data, request_config=request_config)
    response_list = [r.choices[0].message.content for r in resp]

    model = pt_engine.model
    template = pt_engine.default_template
    processor = template.processor

    manual_inference_data = [
        {'type': 'image', 'media': image_url, 'text': image_query},
        {'type': 'video', 'media': video_url, 'text': video_query},
    ]

    messages = [[{
        'role': 'user',
        'content': [
            {'type': item['type'], item['type']: item['media']},
            {'type': 'text', 'text': item['text']},
        ],
    }] for item in manual_inference_data]

    texts = [processor.apply_chat_template(msg, tokenize=False, add_generation_prompt=True) for msg in messages]
    image_inputs, video_inputs = process_vision_info(messages)
    inputs = processor(
        text=texts,
        images=image_inputs,
        videos=video_inputs,
        padding=True,
        return_tensors='pt',
        padding_side='left',
    )
    inputs = inputs.to('cuda')

    # Batch Inference
    generated_ids = model.generate(**inputs, max_new_tokens=128)
    generated_ids_trimmed = [out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)]
    output_texts = processor.batch_decode(
        generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False)
    assert output_texts == response_list