🧠 AI Flashcard Generator

📘 A lightweight, LLM-powered tool to convert educational content into flashcards.

---

🎥 Demo Video

Flashcard.Demo.1.mp4

---

🖼️ UI Screenshots

---

📑 Table of Contents

• Project Overview
• Features
• Tech Stack
• Setup and Installation
• How to Run
• Sample Execution
• Design Decisions & Prompt Engineering
• Project Structure
• Future Work

---

🚀 Project Overview

This tool streamlines the study process by automatically generating flashcards from educational content such as textbook chapters, lecture notes, or articles.

Built with Streamlit and powered by Zephyr-7B-β (via Hugging Face), it offers an intuitive UI and delivers clean, structured Q&A flashcards that can be exported to Quizlet or CSV.

---

✅ Features

🔹 Core Features

• ✅ Multiple Content Inputs – Paste text or upload .txt / .pdf files.
• ✅ LLM-Powered Generation – Generate a custom number of flashcards using Zephyr-7B.
• ✅ Interactive UI – Streamlit-based frontend for simplicity and clarity.
• ✅ Subject-Specific Context – Optional subject selection improves relevance.

🔸 Bonus Features

• ✅ Quizlet Export – Download in .txt format (tab-separated).
• ✅ CSV Export – Download a .csv file of generated flashcards.
• ✅ Robust JSON Parsing – Handles minor LLM output errors for stable performance.

---

🛠️ Tech Stack

Layer	Tools / Libraries
Language	Python
UI Framework	Streamlit
Model	Zephyr-7B-β via Hugging Face Inference API
Text Processing	PyMuPDF (fitz)
Data Handling	pandas
JSON Parsing	demjson3
HTTP Requests	requests

---

🧰 Setup and Installation

1. Clone the Repository

git clone https://github.com/Deepender25/FlashCard-Generator.git
cd FlashCard-Generator

---

2. Create and Activate Virtual Environment

For Unix/macOS:

python3 -m venv venv
source venv/bin/activate

For Windows:

python -m venv venv
venv\Scripts\activate

---

3. Install Dependencies

pip install -r requirements.txt

---

4. Get Your Hugging Face API Token

• Visit huggingface.co/settings/tokens and generate a User Access Token (with write permissions).
• You’ll be prompted to paste this token into the Streamlit sidebar during app usage.

---

▶️ How to Run

streamlit run app.py

• Visit the local URL shown (usually http://localhost:8501) to access the app.

---

🧪 Sample Execution

🔹 Input Text

Photosynthesis is a process used by plants, algae, and certain bacteria to convert light energy into chemical energy...

🔹 Sample Flashcards (JSON)

[
  {
    "question": "What are the two stages of photosynthesis?",
    "answer": "The two stages are the light-dependent reactions and the light-independent reactions (Calvin cycle)."
  },
  {
    "question": "Where do the light-dependent reactions occur?",
    "answer": "They occur in the thylakoid membranes of chloroplasts."
  }
]

---

🔹 Export Files

quizlet_import.txt

Where do the light-dependent reactions occur?    They occur in the thylakoid membranes of chloroplasts.

flashcards.csv

question,answer
Where do the light-dependent reactions occur?,They occur in the thylakoid membranes of chloroplasts.

🧠 Design Decisions & Prompt Engineering

🔸 Why Zephyr-7B-β?

• Instruction-tuned for Q&A tasks
• Efficient and open-source
• Easy integration with Hugging Face Inference API

---

🔸 Prompt Engineering Strategy

Prompt used in llm_handler_hf.py:

prompt = f"""<|system|>
You are an expert flashcard creator. Your task is to generate exactly {num_cards} question-answer flashcards based on the provided text.
Your entire response MUST be a single, valid JSON list of objects. Each object must have a "question" key and an "answer" key.
Do not add any introduction, explanation, or any text outside of the main JSON list.
Your response must start with '[' and end with ']'.</s>
<|user|>
Generate flashcards for the following text on the subject of '{subject}':
---
{text}
---</s>
<|assistant|>
"""

---

🔸 JSON Cleanup Handling

• fix_json_format() strips out extra text or missing brackets.
• Uses demjson3 for lenient parsing.
• Robust error handling for API failures, timeouts, and decoding issues.

---

📁 Project Structure

.
├── app.py              # Main Streamlit UI
├── llm_handler_hf.py   # Hugging Face API & prompt logic
├── requirements.txt    # Dependency list
├── .gitignore          # Files to ignore in Git
├── .env.example        # Example for API token handling
└── README.md           # You're reading it!

---

🔮 Future Work

• 📝 Flashcard Editor – Let users edit flashcards before exporting.
• 📦 Anki Export – Add .apkg support via genanki.
• 🧠 Topic Clustering – Group questions by sub-topic or difficulty.
• 🚀 Public Deployment – Host on Streamlit Cloud or Hugging Face Spaces.
• 🌐 Multi-language Support – Generate flashcards in different languages.

---

📄 License

This project is licensed under the MIT License.