MixedPrecisionOps Documentation
- Dominant language
- Python
- Stars
- 133k
- Forks
- 15.7k
- Avg merge
- 1d 7h
- Merged PRs (30d)
- 158
Description
### Feature Idea
It would be really useful with a proper writeup and documentation of the changes to loading fp8 model.
Because as it is now, commits like this #11000 one completely broke loaders (such as my [ComfyUI_Hybrid-Scaled_fp8-Loader](https://github.com/silveroxides/ComfyUI_Hybrid-Scaled_fp8-Loader)) that was trying to adhere to using ComfyUI Core loading while using mixed precision before the new ops. These breaking changes was done rather abruptly and there have been several [others affected](https://github.com/comfyanonymous/ComfyUI/commit/43071e3de3780f984a46549e90935a0bf405e9df#commitcomment-172090695). Related to #10912 but what I request is a proper writeup and documentation as there is none and if the format is to be adopted there needs to be proper documentation for it.
### Existing Solutions
_No response_
### Other
_No response_
Contributor guide
Assessment
This issue has not been assessed yet.