Synthetic Data Forge

AI-powered synthetic data generation tool using Qwen-2.5-7B-Instruct.

Description

Synthetic Data Forge is a tool designed to generate high-quality synthetic data for various business niches. It leverages the power of the Qwen/Qwen2.5-7B-Instruct Large Language Model (LLM) to create realistic datasets. The tool features a user-friendly Gradio interface, allowing users to specify a business niche and the number of rows to generate. The generated data can be viewed directly in the browser and downloaded as a CSV file.

Features

• AI-Powered Generation: Uses Qwen-2.5-7B-Instruct for intelligent and context-aware data generation.
• Gradio Interface: Simple web-based UI for easy interaction.
• Customizable: Generate data for any business niche (e.g., Hospital, E-commerce, Finance) by providing a prompt.
• Realistic Data: Integrates Faker library for realistic names, addresses, and dates, combined with LLM-generated logic for consistency.
• Export: Download generated datasets as CSV files.
• GPU Accelerated: Optimized for running on GPU (CUDA) for faster generation.

Prerequisites

To run this notebook, you need the following Python libraries:

• torch
• transformers
• accelerate
• bitsandbytes (for 4-bit quantization)
• gradio
• faker
• pandas
• huggingface_hub

You can install them using pip:

pip install torch transformers accelerate bitsandbytes gradio faker pandas huggingface_hub

Usage

1. Open the Notebook: Open DATAGEN.ipynb in Google Colab or a Jupyter environment with GPU support.
2. Install Dependencies: Run the first few cells to install the necessary libraries.
3. Hugging Face Login: You will need a Hugging Face token to access the model. authentiate using notebook_login() or set the HF_TOKEN secret in Colab.
4. Run the Generator: Execute the cells to load the model and define the generation functions.
5. Launch Interface: Run the final cell to launch the Gradio interface.
6. Generate Data:
   • Enter a business niche prompt (e.g., "Retail store inventory").
   • Specify the number of rows.
   • Click "Submit".
   • Download the generated CSV.

Files

• DATAGEN.ipynb: The main Jupyter Notebook containing the code.
• README.md: This file.