Try updating gpt_index to latest version! If it's still happening ping here again

Interesting article talking through some of the same concepts of dividing data into chunks and embedding them. Seems like a DB is used instead of indices: https://simonwillison.net/2023/Jan/13/semantic-search-answers/

Ah cool didn't see that. I'll merge my stuff in and PR

Wonderful project! A question I had from reading the docs -- how do the indices determine how to create the nodes? i.e. how does it know when to split the chunks? Is it just after a set amount of tokens?