Welcome to the Corrective RAG Model GitHub repository, designed to answer questions related to the Paris 2024 Olympics. This project leverages the GPT-3.5-turbo model via API calls to OpenAI and uses a Corrective Retrieval-Augmented Generation (CRAG) approach to enhance the accuracy and relevance of responses.
The model is designed to provide precise and up-to-date information by retrieving data from relevant sources before generating a response.
Here are the sources:
If any of the documents retrieved are irrelevant to the question, the model performs a web search using Tavily Search.
The implementation of the Corrective RAG model in this project is based on the example provided by LangChain. You can find the reference code here.