Comfy-Org / Comfy-Org/ComfyUI

'Save Checkpoint' node automatically prunes all training weights

Open
#6,329 0 comments 0 reactions 0 assignees View on GitHub
Potential Bug
Dominant language
Python
Stars
133k
Forks
15.7k
Avg merge
1d 6h
Merged PRs (30d)
155

Description

### Expected Behavior

When using the node by default, or when using `--force-fp32`, it should keep the training weights intact or at least provide you the option.

### Actual Behavior

It automatically removes all training weights as well as any other data it deems as junk. This brings an SD 1.5 model down to 3.5gb from 7.5gb when in fp32 precision.

### Steps to Reproduce

Load Checkpoint -> Save Checkpoint

### Debug Logs

```powershell
Nothing relevant in the logs
```

### Other

_No response_

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.