Ryuketsukami

This skill implements, uses, or explains TurboQuant, Google's data-oblivious vector quantization algorithm for LLM KV cache compression. It is applicable for topics such as KV cache compression, TurboQuant, and reducing LLM memory usage.