yqvqy / knowledge-base-question-answering-chatbot

Knowledge-Base-Question-Answering-Chatbot

This is a medical chatbot built using Llama2 and Sentence Transformers for multiple Large PDF files (You can change to any other field PDF files). The chatbot is designed to deliver a seamless conversational experience with its natural language processing capabilities on your own data.

The chatbot is powered by Langchain, Fasiss and Chainlit. LangChain is a framework that makes it easier to build scalable AI/LLM apps and chatbots. Faiss is a vectorstore for storing embeddings and your PDF in text to later retrieve similar docs. Chainlit lets you create ChatGPT-like UIs on top of any Python code.

The chatbot runs on a decent CPU machine with a minimum of 16GB of RAM.

Prerequisites

• Python 3.9.17

• Download llama-2-7b-chat.ggmlv3.q8_0.bin model to local reop folder:

  • https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML/blob/main/llama-2-7b-chat.ggmlv3.q8_0.bin

Install package

pip install -r requirements.txt

Run

Convert your PDFs to embeddings

1. Put your PDF doc under /data folder
2. Run the script python ingest.py to 'ingest' and embed your docs.
3. Verify that the index.pkl and index.faiss files are successfully created in the /vectorstore/db_faiss folder.

Run the app

Once you've verified that the embeddings and content have been successfully added to your faiss store, you can run the app chainlit run model.py -w to launch the local dev environment with Chainlit UI, and then type a question in the chat interface.

Credit:

• AIAnytime
• Chainlit