parthsarthi03 / raptor

The official implementation of RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval
https://arxiv.org/abs/2401.18059
MIT License
688 stars 98 forks source link

multi doc #32

Open icemoon-creative opened 2 months ago

icemoon-creative commented 2 months ago

小白求问,这个支持add多个document,然后进行rag吗?非常感谢非常感谢

icemoon-creative commented 2 months ago

Hello, I'm new to this and would like to ask if it supports adding multiple documents and then performing RAG (Retrieval-Augmented Generation)? Thank you so much!

medxiaorudan commented 1 month ago

@icemoon-creative Currently, you may need to combine all the documentation together first and then build the tree, if you want to continue building on the existing tree, the project is still being worked on and you can find details in issue #17, issue #20 and issue #27.

isConic commented 1 month ago

@icemoon-creative multidoc does not work, like @medxiaorudan mentioned , you can hack it to build a giant document. But this approach is not very scalable. I've also seen document chunks that include details from multiple documents.

For multidoc, I would recommend checking out alternative implementations.

There are others that have replicated RAPTOR. For example: