Find answers from the community

Updated 3 months ago

Has anyone experienced an issue with

Has anyone experienced an issue with LlamaParse where the LLM only responds based on the first page of a PDF? Any tips on what to check to identify the problem?
L
A
6 comments
Can you clarify what you mean? Or give some code sample?
for an example , a job contract , the salary is mentioned in page 9
when i ask a question about it to the llm, on terminal output i only see it retrived nodes of page 1
and when i ask it how many pages you have access to in a given pdf , it also says 1
while its a 9 page pdf
Well, its only retrievieing the topk, so its only every going to retrieve the top k nodes. Asking how many pages there are will never work

Sounds like you might just need to increase your top k,

index.as_query_engine(similarity_top_k=4) for example
Add a reply
Sign up and join the conversation on Discord