JohnAlcatraz

Since I have been looking at doing this a while ago, there's a few things I can comment, if I remember correctly: For RMCs with only one _section_ that needs...

That's unfortunate, since shipping anything Python in my native app would really not be practical. I need to use https://github.com/sisong/HDiffPatch then instead.

> Hi, The workaround is to use a World to Tangent transform instead (and a Tangent to World in other places). The tangent transform is slightly less accurate, but not...

Would it be possible to make this a command line argument? Something like `--40xx`. So that it can be easily enabled by anyone who knows they don't plan to do...

I see the same issue with flash attention. I followed the steps described in the readme, but with flash attention enabled, I always get this error when trying to generate...

> @JohnAlcatraz If you really do so, you will get a very, very bad quality, according to my experiments before. I also see very bad quality with flash attention, but...

With Falcon-40B being significantly better than LLaMA-65B, and actually being fully open source under Apache 2.0, it's definitely the new king of open source LLMs. It would be great to...

> Currently, im2col is being used for convolutions, which consumes a very high amount of RAM during the VAE phase. But I did my comparison with TAESD instead of the...

But you say there is still a difference of ~5 seconds here vs 1.43 seconds in diffusers? I think in that case, that is a very interesting benchmark and it...

koboldcpp also uses stablediffusion.cpp: https://github.com/LostRuins/koboldcpp