Adding Buffer memory to Q&A application
Hi Team,
I'm using Dollyv2 LLM for Q&A application, and using Langchain.
Here, my input data are from pdf files, I have around 50 pdf files and each files are different from each other.
Here, my question is, is it meaningful to add conversational buffer memory to this application (Q&A)? because when I logically think, LLM expects Prompt + Question + relevant chunks. so every time if I ask question to the model it sends different question with different context and those context are different from each other.
when I ask question to them model for first time, it takes/remember the question, context, answer and this question belongs to 1st pdf document. And all of a sudden if I ask a question about 25th pdf document which is completely different the 1st pdf document, here does the conversation buffer memory helps?