Reproduction of TurboQuant
There have been many TurboQuant implementations recently in llama.cpp, mlx, vllm, and sglang, but a lot of the discussion and code around them feels pretty noisy and looks to be AI-generated. I’m trying to understand which claims from the…