Find answers from the community

Updated 2 months ago

Hello, I was developing a chatbot

Hello, I was developing a chatbot application with an LLM locally with a focus on acquiring data from PDF documents, I was wondering if it was possible through llama-index to somehow configure this chat model so that the responses provide the location or the name of the document from which the answer to a certain question was obtained, or if, on the contrary, I would need some other framework, I ask for your advice.
n
S
3 comments
It is possible. I think you would just have to parse the pdf (which is probably something llama-hub has a tool for. But even if you didn't use llama-hub tools for it so as long as you parsed it to extract the text, images, and other objects from the pdf and could store that info in a vector database for retrieval you can do a lot with llama index.
Thank you very much, I didn't know that with llama-hub you could implement something similar to what I was looking for
Its basically already done for you. I believe the PDF parser is a tool you can load into an Agent. The tool itself is already made you just need to give it to an agent.

https://llamahub.ai/l/file-pdf
Add a reply
Sign up and join the conversation on Discord