David Khristepher Santos

I got this problem too after setting up my environment with CUDA "properly" so that torch detected it. I have a GTX 1650 with 4GB VRAM. I was able to...

Thanks for reporting, I'll look into this.

So I just tried loading a checkpoint with a file `diffusion.json` in the root of the zip and it looks like torch itself refuses to load the model weights if...

With safetensors, the proposed method of hashing the CRCs no longer works, because safetensors aren't zip files. This leads to the question, how can we hash safetensors properly? @JustMaier I've...

Hi! The problem with computing hashes is that Automatic1111 does it in real time, and that's probably why such shortcut method was used. If we're going to add a new...

Yea, and we could extend this concept to embeddings, hypernetworks. Different extension perhaps, but having a header there. I'm the author of Diffusion Toolkit, https://github.com/RupertAvery/DiffusionToolkit and I see that as...

> I wonder if there is a way to include the metadata without making the files not work in tools that don’t support the metadata so that adoption doesn’t have...

See this repository for more information and as a place for discussion on the proposed container format: https://github.com/RupertAvery/DiffusionFormat

I will add an appropriate return value. * 0 = success * -1 = error * -2 = cancelled

So I found out that in linux the number returned is an unsigned int from 0-255, so the above gets returned as 0, 255, 254. I read that error values...