Welcome to this project. The main objective of this chatbot application is to enable its user to upload the url of any webpage and then able to ask questions in form of chat to get expected responses to the questions asked by the user.
- Website Interaction: The chatbot uses the latest version of LangChain to interact with and extract information from various websites.
- Large Language Model Integration: Compatibility with models. In this code I am using GPT-4, but you can change it to any other model like Llama2.
- Streamlit GUI: A clean and intuitive user interface built with Streamlit, making it accessible for users with varying levels of technical expertise.
- Python-based: Entirely coded in Python.
A RAG bot, or Retrieval-Augmented Generation bot, enhances the capabilities of our LLM by incorporating additional information provided in the prompt. Initially, we vectorize the text we intend to use as "augmented knowledge" and compare it with the prompt. The system identifies the most relevant text based on similarity scores and uses it to enrich the conversation. This selected text acts as a prefix, providing context and guiding the responses generated by our LLM. We have used chroma as our vector storage and OpenAI for embedding our text chunks into vectors.
pip install streamlit langchain langchain-openai validators python-dotenvOPENAI_API_KEY = your api key goes herestreamlit run app.py