modelscope / modelscope/DiffSynth-Studio

Wan2.1 finetuning not considering unconditional training

Open
#627 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
13.1k
Forks
1.3k
Avg merge
13h 12m
Merged PRs (30d)
45

Description

I'm given to understand that when training a conditional FM/DM the condition (i.e. prompts and referrence images) should be droped with a certain probability (typically 0.5). However I found the Wan finetuning precedure lacks the random dropping part, is there any motivations for this? Any suggestion is appreaciated.

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

The issue names no file or test; begin by locating the Wan finetuning procedure and reading how prompts and reference images are handled. Compare that behavior with the proposed unconditional-training condition dropping, then establish whether the omission is intentional or define a testable expected behavior before making changes.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.