Rag

TurboQuant compresses embeddings aggressively without corpus-specific training. This post covers the algorithm, the turboquant-embed implementation, and the retrieval benchmarks that hold up on BeIR.