Query

I think typically only your leaf nodes would be in the vector store

The other nodes are pulled from the docstore when merging needs to happen

Oh, I see. What if the whole thing happens in an IngestionPipeline?

It's a bit more complicated

Yea definitely more complicated -- not sure I would recommend that 😅 you could have a custom transformation step to sort out the leaf nodes vs. Everything else to avoid indexing non-leaf nodes (and would also give you a chance to throw the nodes into a docstore?)

Okay.. but I'm thinking it might not even be worth it to use this approach 😄
I mean it loads all the node into memory, right?

Depends if you use a remote docstore and vector store or not