For just 100K items why don't you simply load the embeds into numpy and use cosine similarity directly? It's like 2 lines of code and works well for "small" num

For just 100K items why don't you simply load the embeds into numpy and use cosine similarity directly? It's like 2 lines of code and works well for "small" number of documents. This would be exact NN search. Use approximate NN search when you have high volume of searches over millions of vectors.