OpenGVLab / OpenGVLab/VideoChat-Flash

[Q] InternVideo 2.5,, TPO

Open
#61 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
530
Forks
20
PR merge metrics
No merged PRs in 30d

Description

Hello, thank you for the excellent research and for sharing the code.

I understand that TPO (Task Preference Optimization) has been applied to InternVideo 2.5, and as mentioned in the related paper, it includes three task-specific heads: region, temporal, and mask.

I have two questions regarding this:

  1. Are these three heads already implemented and integrated into the current InternVideo 2.5 codebase?
  2. The paper describes a detailed multi-stage training process, but the repository currently provides only inference scripts. Will the training scripts for these heads be released in the future? Alternatively, is there any guidance or reference available to perform supervised fine-tuning (sFT) with these task heads?

Any support or clarification would be greatly appreciated. Thank you again for your valuable contribution!

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No specific file, test, or entry point is named. Review the repository's existing inference scripts and the paper's TPO description to determine whether the three task heads and training process are present; done would require maintainer clarification or documented guidance.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Documentation
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.