600% memory footprint reduction on Vector Embeddings Quantizationgithub.com/JaneaSystems4 pointsbettoa year ago