🤖 AI Story Generator using GPT-2

This project is part of my Generative AI Engineering Portfolio. It uses GPT-2 via Hugging Face Transformers to generate short creative stories based on user-provided prompts.

---

📌 Project Overview

The project is built in three progressive phases:

1. Phase 1 – Command-line GPT-2 story generator.
2. Phase 2 – FastAPI backend with JSON-based generation.
3. Phase 3 – Gradio-powered interactive UI.

---

✅ Phase 1 – Auto Story Generator (Python Script)

A minimal script that generates a story given a prompt.

🔧 Features

• GPT-2 powered text generation.
• Accepts prompt and optional context.
• Generation tuned for creativity and coherence (temperature, top_p, repetition_penalty).

📁 File

• generate.py – Contains the generate_story() function and CLI interface.

---

🚀 Phase 2 – FastAPI Integration (API Serving)

A FastAPI backend that exposes a POST endpoint /generate.

🔧 Features

• Receives prompt in JSON.
• Returns generated story in JSON.
• Swagger UI auto-enabled for testing.

📁 File

• main.py – FastAPI app with /generate endpoint.

📸 Swagger UI Demo

---

🖼️ Phase 3 – Gradio Interface (Local UI)

A local browser-based UI for interactive story generation.

🔧 Features

• Clean and intuitive layout.
• Accepts prompt via UI.
• Displays story output instantly.

📁 File

• app.py – Launches Gradio interface.

📸 Gradio UI Screenshot

---

🧠 Tech Stack

Tool / Library	Purpose
Python	Core programming language
Hugging Face Transformers	GPT-2 model pipeline
FastAPI	API backend
Uvicorn	ASGI server for FastAPI
Pydantic	Data validation
Gradio	Web UI for LLM interaction

---

⚙️ Getting Started

git clone https://github.com/asifnoushadsharafudeen/ai-story-generator.git
cd ai-story-generator
pip install -r requirements.txt
python gradio_ui.py # or python main.py for API version