vector-index-embedding 0.1.0

HNSW based output embeddings for LLM's

vector-index-embedding

Faster hnswlib

Our faster implementation can be found here: https://github.com/martinloretzzz/hnswlib

Warning: This implementation might not work on all systems as it was only tested on the one where we'Re running the bnechmarks and the SIMD implementation was only adapted for that architecture.

For benchmarking we use our own fork of hnswlib, that includes 2 improvements for fast inner product distances on high dimensional data:

• We calculate all the inner products in paralell, that way we reduce memory accesses in half (we load one element of the query and compare it to N other vectors at the same time)
• We removed a heuristic that restricted multi-threading, as as our data is extremly high dimesional and always benifit from using all cores.