8G or Lowvram machines encountered random step reset latent noise issues when using the fill dev model.
- Dominant language
- Python
- Stars
- 133k
- Forks
- 15.7k
- Avg merge
- 1d 6h
- Merged PRs (30d)
- 155
Description
### Expected Behavior
Fix the issue of noise being reset
### Actual Behavior
Random step reset latent noise
### Steps to Reproduce
torch2.4.1+cu124,2070 8GB,flux_fill_outpaint，ksamples\ksamplerAdvanced\samplercustomadvanced,same issues, any sampler&scheduler
### Debug Logs
```powershell
There are no issues with the logs
```
### Other
_No response_
Contributor guide
Research direction
Start by reproducing the report with torch 2.4.1+cu124 on an 8 GB RTX 2070 using flux_fill_outpaint with KSampler, KSamplerAdvanced, and SamplerCustomAdvanced across the reported samplers and schedulers. Trace when latent noise changes between steps; done means the latent noise is not randomly reset during fill-model sampling.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100