InternVL icon indicating copy to clipboard operation
InternVL copied to clipboard

[Bug] The video processor of InternVL-3.5-1B-HF is inconsistent with image processor

Open yeliudev opened this issue 2 months ago • 1 comments

Checklist

  • [x] 1. I have searched related issues but cannot get the expected help.
  • [x] 2. The bug has not been fixed in the latest version.
  • [x] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.

Describe the bug

Hi @Weiyun1025, I notice that InternVL-3.5-1B-HF introduces an extra video_preprocessor_config.json compared to InternVL-3-1B-HF, in which some key parameters like image_mean, image_std, and size are inconsistent with the numbers in preprocessor_config.json. I was wondering whether this is normal, as to my understanding the images and videos share the same encoder and shall have the same sizes and mean/std parameters.

yeliudev avatar Oct 30 '25 17:10 yeliudev

Related issue: #1184

yeliudev avatar Oct 30 '25 17:10 yeliudev