Hi all - I've gotten this code to work in a RAG, but it is prohibitively slow even for a small number of documents (I have access to multiple A100s). There seem to be some proposed solutions floating around on the internet. Do you anticipate addressing this problem in the near future?