How to process multi-images?

by Yiyiyi - opened Feb 6

Feb 6

Would you mind share how to process multi-images?

how to change this ' img_input=img_path,'??? it seems it only accept single-image?

Thank U!

generated_ids = model.generate(
**inputs,
temperature=0.1,
top_p=0.001,
repetition_penalty=1.05,
do_sample=True,
max_new_tokens=32768,
img_input=img_path,
)

efdsf324

Feb 10

Thank you for your interest to Youtu-VL!

img_input is designed for CV tasks (segmentation, detection, depth, etc.) and currently supports single-image input only.
For VL tasks (VQA, multimodal reasoning, etc.), multi-image input is supported by adding multiple images inside messages. You can omit img_input in generate().

Example:

messages = [{
  "role": "user",
  "content": [
    {"type": "image", "image": "/path/to/image-A"},
    {"type": "image", "image": "/path/to/image-B"},
    {"type": "text",  "text": "Compare these two images."}
  ]
}]

If you have any further questions, please feel free to let me know.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment