Project Overview
In this project, users have the ability to engage with any CSV file by querying its contents. The application responds to these inquiries by leveraging a Language Learning Model (LLM), which extracts and formulates answers directly from the information within the file.
Working of the Project:
As soon as the user uploads a CSV file, the OpenAI embedding model is utilized to generate embeddings of the file’s contents. These embeddings are then stored in a vector database called FAISS, which is maintained locally. When a user poses a question, it is converted into embeddings and compared with those in the database. Subsequently, the most relevant embedding is decoded and processed by the Google Palm2 LLM, which I am employing in this project due to its free accessibility. The final answer is then retrieved and displayed on the screen.
Technologies Leveraged:
- Embedding by OPENAI
- LLM by Google (PaLM 2)
- Langchain
- Streamlit for UI.