How to use BPR/JPQ in production?

[nickchomey]

Hi there, I have a dumb question that I hope you don't mind answering.

Your papers and work related to INCOME seem extremely promising given how much it can reduce storage costs.

My question is how do you implement this approach? You mention faiss in the paper - does this mean I could use it with Opensearch's faiss backend? Is there any way to use it with Elasticsearch or Solr's lucene hnsw implementation?

Thanks!