question answering model for text file

[IamExperimenting]

Hi team,

I'm trying to build question answering model using open source LLM, my data source is text file.

I initially approached by creating chunk and pushing those chunks into faiss vectordb, does similarity search and passed top 3 document to LLM model to get answer for the query.

is it possible to directly ask question to raw text file itself? rather creating chunks, can I pass complete text file to LLM and ask question?

can anyone provide me some example?