Bringing Nunchaku 4-bit Diffusion Inference to Diffusers
Hugging Face integrates Nunchaku 4-bit quantization into Diffusers for faster diffusion inference
Hugging Face is bringing Nunchaku, a 4-bit quantization framework for diffusion models, into the Diffusers library. This lowers the hardware barrier for running high-quality image generation models by reducing memory requirements and accelerating inference. The integration matters because Diffusers is the dominant open-source library for diffusion pipelines, so native 4-bit support could meaningfully expand who can run these models locally.