Auto nan check vae precision switch.
Open
Feature
- Dominant language
- Python
- Stars
- 133k
- Forks
- 15.7k
- Avg merge
- 1d 6h
- Merged PRs (30d)
- 155
Description
By default vae would load in fp16 if a tensor with nan is produced it would switch to 32bit float.
Contributor guide
Assessment
This issue has not been assessed yet.