Billy Cao
Gradio has merged https://github.com/gradio-app/gradio/pull/2846 which also fixes this. When they make a stable release, users of this repo need to update gradio. For those want to workaround their issue asap,...
Gradio has released 3.15.0 which fixes this. However SD webui seems to be only compatiable up to 3.9.0..
Are you running your script using pycharm?
They have merged a fix at pillow, you can pip install their latest master branch
Have you tried adding `--disable-safe-unpickle`? Your model weights are not loaded at all, possibly due to you using torch 1.13 which requires `--disable-safe-unpickle` to load existing checkpoints.
Did you add an image before you interrogate?
I think its more about the issues. Many of the issues I saw are either due to user's fault or a completely different downstream project that happen to use this...
This should not be a bug, it is perfectly normal to OOM when training TI on 6GB. Batch 4 is way too big for you, even on my 12GB 3080Ti...
What makes you think saving image does not need VRAM? It is not as simple as 'oh copy the data out'. When you save, it has to be converted to...