Stop execution of workflow but continue batch when Ksampler outputs NaNs
Open
Feature
- Dominant language
- Python
- Stars
- 133k
- Forks
- 15.7k
- Avg merge
- 1d 6h
- Merged PRs (30d)
- 155
Description
### Feature Idea
Is there anyway to stop a run but continue on with the batch when the Ksampler decides to output a large amount of NaNs? This would stop the output of black images.
### Existing Solutions
There is none.
### Other
_No response_
Contributor guide
Research direction
Start by tracing the Ksampler entry point and how workflow batches handle sampler outputs. Define how a large NaN result should stop only the current run while allowing the remaining batch items to continue, and verify that black images are no longer produced.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- ai, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100