Vector Quantization
A data compression technique that replaces massive, continuous floating-point vectors with integer pointers mapped to a discrete, limited codebook.
Think of It Like This
Like mapping every exact shade of paint in the world to the closest available option in a standard 64-color crayon box.
By clustering millions of vectors into a few thousand centroids (the codebook), models only need to store the ID of the nearest centroid. This drastically reduces the memory footprint of massive embedding indexes and enables highly optimized hardware execution. It is a foundational component of systems like FAISS and VQ-VAEs.