BITSFUSION: 1.99 BIT WEIGHT QUANTIZATION OF DIFFUSION MODELS

Summary from the TLDR AI newsletter

The team at Snap Research was able to reduce the size of the Stable Diffusion UNet model from 1.72 GB down to 219MB while increasing performance with their new quantization scheme. The quantization method is somewhat complex, but paints a strong path forward in running generative models on consumer hardware.