Base models of SD 1.5 are taken from here and here, but fp16/cleaned - smaller size, same result. Not for generation directly, but for merges/training.
*all rights belong to the companies that created and have rights to this intellectual property, this files are for informational purposes only
*my Telegram / Ko-fi / Hugging Face / Tensor.art - feel free to use files, contact or support me
Description
FAQ
Comments (4)
What would happen if you did use this for simple text-2-image generation? The example seems to show that it's similar enough that I can't tell the difference. Just curious.
Hi, general question, will there ever be SDXL inpainting models or is that no longer possible as there is no base SDXL inpainting model?
How to do this "Cleanly" I had some big models that wanted to prune down and were kinda of bit confuse with the parameters "Force CLIP position_id to int64 before convert,
Fix clip, Delete known junk data". xd
Is this the EMA or non-EMA version of the weights?