RAG Chatbot

Introduction

We are aiming do make a chatbot that has access to our local files and can answer question about them. This aims to overcome the knowladge cutoff and halucination problems in chatbots.

We are using Gradio to create a web interface for the chatbot and LangChain to interact with different chat and embedding models.

Getting Started

Clone the repository and install the requirements:

$ git clone https://github.com/ComponentSoftTeam/rag-chatbot-demo.git
$ cd rag-chatbot-demo
$ pip install -r requirements.txt

If you have conflicting dependencies, you can create a new environment and install the requirements there:

$ pythoh -m venv .venv
$ source .venv/bin/activate
$ pip install -r requirements.txt

Copy the .env.example file to .env and fill in the required values

Running the application

You can run the application from within the src directory.

$ cd src
$ python app.py

Then open your browser and navigate the link provided in the terminal.

Note

The application is using python 3.10^

Warning

For the mistral embedding to work corretly it needs to load a tokenizer, for this the HF_TOKEN is set in the .env file, but you also need to visit the following link: Mistral Huggingface and agree to the terms and conditions. If you choose to not go with this option a heuristic solution will be used instead.

You can experiment with the jupyter notebook version of the application by loading it into a jupyter notebook environment.

Development

Here are some documents that could be helpful for developers of a similar project

RAG

• Chroma persistent client
• Adding sources
• Optimal Chunk size
• Hybrid Search
• Prompt condensation

Resources

• https://python.langchain.com/docs/use_cases/question_answering/quickstart
• https://www.aporia.com/learn/build-a-rag-chatbot/
• https://github.com/QuivrHQ/quivr