Shaul Shvimmer

Here's a possible solution for ver 7.3 mat files, similar to scipy's wosmat: f = h5py.File(filepath, 'r') var_list = list(f.keys())

I had the same issue, this solved it for me: pip install opencv-contrib-python --upgrade

Solution: I've made it work, here's the main steps: (some more workarounds are needed to make it run, in the inference as well, but this is the main idea): 1....

@chrieke Hi, I'm not sure what you mean by "second image".

@nikooc2s @chrieke No, there is no need to do anything else, using the `expand_dims` this way, adds a third dimension with size 1, so the argument in the for loop...

Same issue in Kubuntu 20 with torch 2.0.1 and cuda 11.8 python 3.9 / python 3.10 and flash-attn versions 0.2.8 / 1.0.4 / 1.0.5 / 1.0.6 / 1.0.7 with and...

Hi @nekoshadow1 I have adjusted K-diffusion to support img2img [()](https://github.com/crowsonkb/k-diffusion/issues/70) So now I want to move on to distill this img2img model to fit fast inference using this Consistency model...