Please provide model_index.json and all the additionnal things as in V1?
Hello there!
Basically I'm trying to convert model to a AMD gpu compatible format, and repo is missing things to do that atm. Basically need to be able to run StableDiffusionPipeline.from_pretrained(model_path, torch_dtype=dtype).to(device) which does not work on repositories only containg .cpkt files.
Because of that, I would really appreciate if you could commit to repo all the diffusers related files, as you did for prompthero/openjourney, which I'll definitely do a conversion of and use in the meantime.
Conversion script available here, should you want to take a look at it. Also going to mention that I did test V2 through Artroom in cpu mode, and while real slow results are really good.
Thanks for reading me, and please keep releasing such awesome content.
+1
