Re:Haystack DB – 10x faster than FAISS with binary embedding...
by babelfish on Monday April 29, 2024 @12:42AM
(#40194514)
Very impressive! Do you have similar scale tests at higher numbers? I’m curious what the numbers are at 1B/1T vectors, and amounts larger than what can fit in memory
[ Parent
| on HN ]
Re:Haystack DB – 10x faster than FAISS with binary embedding...
by carsonpoole on Monday April 29, 2024 @01:04AM
(#40194654)
not yet, but it's roughly linear at scaling, since it's a brute force algorithm. so with the current version it'd probably be about 22 seconds for a 1B vector search. the whole point of having metadata queries are to prevent those kind of searches from being necessary, and hopefully with some FTS interspersed it can reduce the number of similarity comparisons required even more