exdysa/

mitsua-vae-SAFETENSORS

name: mitsua-vae-SAFETENSORS
base_model: Mitsua/mitsua-diffusion-one
license: other
pipeline_tag: feature-extraction
tasks :
 - feature-extraction
 - image-to-image
language :
 - ja
 - en
 datasets :

The Float16 VAE model from Mitsua diffusion converted into safetensors format. The diffusion model was trained from scratch using 11 million images public domain, Creative Commons Zero, or permission-first copyrighted images obtained in compliance with relevant terms and conditions. Unfortunately the VAE seems to be an altered version of Stable Diffusion 2's, so much of this information is likely irrelevant.