Chatbot-answer-law-documents

figure 1: Example 1: figure 2: Example 2: figure 3: Example 3:

Table of Contents

Overview

A simple chatbot project, using GPT model to answer VietNamese 's VAT taxation. In this project, I used

  • Embedding model: "bkai-foundation-models/vietnamese-bi-encoder" I did many experiments to choose embedding model, and generated answers using vietnamese-bi-encoder embedding model is more revelant to the queries than other vietnamese ones.
  • Chunk size/chunk overlap: 512/40
  • Search method: FAISS
  • LLM: GPT4-Turbo

Getting Started

To install chatbot, follow these steps:

  1. Install uv - Python Package manager link: https://github.com/astral-sh/uv

  2. Clone the my repository

Expand
git clone git@github.com:BaHuy15/Chatbot-answer-law-documents.git
  1. Create a virtual environment and install the required dependencies (you can use any virtual environment manager):
uv venv

# On macOS and Linux.
source .venv/bin/activate

# On Windows.
.venv\Scripts\activate

uv pip install -r requirements.txt
  1. Setup API keys:
    Put your key in config.toml file
[API_KEYS]
OPENAI = "<YOUR_OPENAI_KEY>"    

Run application

python main.py  

Visit this link to see result: http://localhost:8080/
FastAPI Endpoint: http://localhost:8080/docs

Result

This is the result of all my experiments image