0.1.3 Patch release

This patch releases refactors the model architecture of VQModel or AutoencoderKL including the weight naming. Therefore the official weights of the CompVis organization have been re-uploaded, see:

• https://huggingface.co/CompVis/ldm-celebahq-256/commit/63b33cf3bbdd833de32080a8ba55ba4d0b111859
• https://huggingface.co/CompVis/ldm-celebahq-256/commit/03978f22272a3c2502da709c3940e227c9714bdd
• https://huggingface.co/CompVis/ldm-text2im-large-256/commit/31ff4edafd3ee09656d2068d05a4d5338129d592
• https://huggingface.co/CompVis/ldm-text2im-large-256/commit/9bd2b48d2d45e6deb6fb5a03eb2a601e4b95bd91

Corresponding PR: #137

Please make sure to upgrade diffusers to have those models running correctly: pip install --upgrade diffusers

Bug fixes

• Fix FileNotFoundError: 'model_card_template.md' #136