Vectors are over, hashes are the future of AI

Re:Vectors are over, hashes are the future of AI by sayonaraman on Monday September 27, 2021 @08:27PM (#28677873)

Whoever wrote the article must have done a cursory search at best, I'm surprised they didn't mention semantic hashing by Salakhutdinov & Hinton (2007) https://www.cs.utoronto.ca/~rsalakhu/papers/semantic_final.p...

Edit: also, talking about LSH, must check out FAISS library https://github.com/facebookresearch/faiss and the current SOTA http://ann-benchmarks.com/

[ Parent | on HN ]

Re:Vectors are over, hashes are the future of AI by LuisMondragon on Tuesday September 28, 2021 @02:02PM (#28685612)

I'm intrigued by FAISS. Do you know if I can use BERT embeddings with FAISS?

[ Parent | on HN ]

Re:Vectors are over, hashes are the future of AI by gk1 on Tuesday September 28, 2021 @02:04PM (#28685643)

Yes. Check out this example with Faiss + Bert: https://www.pinecone.io/learn/faiss-tutorial/

[ Parent | on HN ]

Re:Vectors are over, hashes are the future of AI by LuisMondragon on Wednesday September 29, 2021 @10:37AM (#28694226)

This tutorial was excellent. I got everything working in a couple of hours. On CPU alone the similarity search is mindbogglingly fast. Thank you!

[ Parent | on HN ]

Re:Vectors are over, hashes are the future of AI by mish15 on Wednesday September 29, 2021 @01:09PM (#28696672)

We will add neural hash based ANN to that repo when we get time. I expect HNSW to get pushed out in time for many reasons.

[ Parent | on HN ]

Re:Vectors are over, hashes are the future of AI by mish15 on Monday September 27, 2021 @09:56PM (#28678450)

Interesting. Hadn’t seen it actually. BOW is problematic as a starting point, but this is neat and a long time ago too.

[ Parent | on HN ]