ExplainGPT

ExplainGPT is a project that allows you to interact with language models, visualize token probabilities, and understand model behavior through an intuitive interface.

Features

• Generate text using language models
• Visualize token probabilities for generated text
• Interactive playground for model exploration
• Model selection capability
• Token-by-token explanation of generated output

Project Structure

• Frontend: Next.js application with React 19 and Tailwind CSS
• Backend: FastAPI Python server that handles model loading and text generation

Setup and Installation

Backend Setup

1. Navigate to the backend directory:

   cd backend

2. Sync dependencies using uv:

   uv sync

3. Activate the virtual environment:

   # If using standard venv
   source .venv/bin/activate  # On Unix/macOS
   # OR
   .venv\Scripts\activate     # On Windows

4. Run the FastAPI server:

   fastapi dev main.py

Frontend Setup

1. Navigate to the frontend directory:

   cd frontend

2. Install dependencies:

   bun install

3. Start the development server:

   bun run dev

4. Access the application at http://localhost:3000

Technologies

• Frontend: Next.js, React, Tailwind CSS, Radix UI
• Backend: FastAPI, PyTorch, Transformers
• Development: Bun, uv

License

This project is licensed under the MIT License - see the LICENSE file for details.

Copyright (c) 2025 Anand Rajaram