(average case)
So you can see how it’s important to pick the right centroid count for your dataset sizeHNSW for FAISS, SPTAG for SPANN
It pisses me off that some projects are advising people to compress their vectors when they don’t support reranking
Talk track: with 3x overquery we can beat uncompressed recall (and speed!) while compressing 64xFor full context, see https://thenewstack.io/why-vector-size-matters/
Less useful because even with overquery you can’t make up the recall loss (except for ada002)
Openai-v3 works fine with BQ too but I’d rather compress it 64x with PQ
Not quantitative!
LVQ (Locally-adaptive Vector Quantization) is a new compression design that is accurate enough to be used in reranking. We can replace the full-resolution vectors on disk with LVQ-compressed, reducing index size by a factor of ~4 and speeding up queries about 20%
My intro slide was titled “modern” vector search
ADC is well knownQuick ADC and Quicker ADC are more obscure and only used in slower, partitioned index designsFused ADC is new in JVector and the first application to graph indexes