openai/consistencydecoder
OpenAI's consistency decoder is an improved VAE decoder for Stable Diffusion that produces higher quality images than standard GAN-based decoders.

Velocity · 7d
+2.3
★ / day
Trend
→steady
star history
This repository provides OpenAI’s consistency decoder, a distilled diffusion-based variational autoencoder decoder designed to replace the standard GAN decoder in Stable Diffusion pipelines. It takes latents from Stable Diffusion’s VAE encoder and decodes them into higher quality images with better fidelity. The model is approximately 2.49 GB and runs on CUDA devices.