bytedance / bytedance/InfiniteYou
How to handle the new parameter GUIDANCE during training？
- Dominant language
- Python
- Stars
- 2.7k
- Forks
- 285
- PR merge metrics
- No merged PRs in 30d
Description
What a great job!
I noticed that the model used was FLUX1.0-dev, and this model underwent Guidance distillation. Therefore, a new parameter, guidance, was added in the forward process.
May I ask how you set up this guidance throughout the entire training process?
Set it to 0 and then treat it as a model without the influence of this parameter? Is there any better solution?
Looking forward to your reply
Contributor guide
No contributing guide indexed for this repository
Research direction
The issue names no files, tests, or entry points. Start by locating the FLUX1.0-dev training code and its forward-process handling of guidance; done means documenting the configured value and the rationale for that training setup.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Documentation
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100