[P] GPU friendly lossless 12-bit BF16 format with 0.03% escape rate and 1 integer ADD decode works for AMD & NVIDIA

About this article

Hi everyone : ) I just released a new research prototype It’s a lossless BF16 compression format that stores weights in 12 bits by replacing the 8-bit exponent with a 4-bit group code. For 99.97% of weights, decoding is just one integer ADD. https://preview.redd.it/qbx94xeeo2tg1.png?width=1536&format=png&auto=webp&s=831da49f6b1729bd0a0e2d1f075786274e5a7398 Yes 12 bit not 11 bit !! The main idea was not just “compress weights more”, but to make the format GPU-friendly enough to use...