5 ms·

Nope... It IS using a vector database / faiss embedings for the semantic search. Check the indexer: https://github.com/Olow304/memvid/blob/main/memvid/index.py

Nope... It IS using a vector database / faiss embedings for the semantic search. Check the indexer: https://github.com/Olow304/memvid/blob/main/memvid/index.py https://github.com/Olow304/memvid/blob/main/memvid/index.py

kgeist 1y ago
Indeed, they use FAISS for vector search, and the actual innovation is to store PDFs in a MP4 file. I think they should compare this to just compressing PDFs using ZIP, 7z, or RAR to see if it actually makes things better.