It can quantize vectors to significantly reduce storage requirements for high-dimensional embedding models while maintaining near state-of-the-art retrieval quality, with only a small drop in recall accuracy.
MIT lic.
https://github.com/RyanCodrai/turbovec
It can quantize vectors to significantly reduce storage requirements for high-dimensional embedding models while maintaining near state-of-the-art retrieval quality, with only a small drop in recall accuracy.
MIT lic.
https://github.com/RyanCodrai/turbovec