New LLM compression algorithm by Google
A set of advanced theoretically grounded quantization algorithms that enable massive compression for large language models and vector search engines.