halffried's picture
Update README.md
c960dce
metadata
license: mit

This is the https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K model converted to a simple CLIPTokenizer model for use with https://huggingface.co/stabilityai/stable-diffusion-2

I produced this tokenizer when trying to update the ckpt converter in Diffusers to work with Stable Diffusion V2, and I personally think it often gives better results than the stock tokenizer.