🧠 Text2SQL App — Natural Language to SQL using LangChain, HuggingFace & Ollama

This project allows users to ask natural language questions and get answers from a MySQL database by converting the questions into SQL queries using:

• LangChain
• HuggingFace Embeddings
• Ollama LLMs (Qwen2)

It uses few-shot learning, semantic similarity, and local LLM inference for a fast, secure, and intelligent querying experience.

---

📦 Key Python Packages Used

• langchain-huggingface
• langchain-ollama
• langchain_community
• watchdog
• cryptography
• streamlit
• python-dotenv

---

📁 Dataset & Embeddings

• The dataset (SQL schema + data) is located in ./dataset/db_creation_atliq_tshirts.sql.
• We embed both the schema and its content using HuggingFace’s "sentence-transformers/all-MiniLM-L6-v2".
• We have performed Few-shot training on LLM to train it on edge case queries.
• LLM used: Qwen2 (running via Ollama).

---

🐬 Setting Up MySQL with Docker

Step 1: Run MySQL Docker Container

docker run --name atliq-mysql \
  -e MYSQL_ROOT_PASSWORD=<password> \
  -e MYSQL_DATABASE=atliq_tshirts \
  -p 3306:3306 \
  -d mysql:8.0

Step 2: Load Schema & Data into MySQL

docker cp ./dataset/db_creation_atliq_tshirts.sql atliq-mysql:/db_creation.sql
docker exec -it atliq-mysql bash

Inside the container:

mysql -u root -p
# Enter: root

USE atliq_tshirts;
SOURCE /db_creation.sql;

⚙️ Environment Setup

Step 1: Clone the Repo

git clone https://github.com/your-username/text2sql.git
cd text2sql

Step 2: Create & Activate Virtual Environment

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

---

🔐 Create .env File

touch .env

Add the following:

DB_USER=<db user name>
DB_PASSWORD=<db password>
DB_HOST=<host name>
DB_PORT=3306
DB_NAME=atliq_tshirts

LLM_MODEL_NAME=<model name>
LLM_MODEL_TEMP=<model temp>
LLM_BASE_URL=<llm base url>

EMBEDDING_MODEL_NAME=sentence-transformers/all-MiniLM-L6-v2