Embeddings Conversation | Handle Neutral User Responses


We have been working on a chatbot that helps in answering questions for the PDF document added as context.

We are using pinecone as vector db and using openai embeddings.

Problem Statement: While in conversation if the user writes some neutral words like “Ok”, “Thanks”, “I understand” then bot responds some random responses.

Expected result: Looking a way if openai can handle such. inputs from users and respond using standard NLP flow.


1 Like

This feels like something that can be fixed in your prompt template. Are you using a framework like langchain? It might be helpful to share a code snippet of the prompt you’re using and potentially the details about your semantic search retrieval pipeline.