Offline Flashcard Generator using Phi-3 (Ollama)

This is a fully offline Flashcard Generator powered by Phi-3 using the Ollama runtime. It lets you:

• Input text or upload PDFs
• Generate flashcards in Q/A format using phi3
• Export to CSV or directly as .apkg (Anki Deck)
• No API keys, no internet needed — all runs locally on your system

---

Features

• Text or PDF-based input
• Use local LLM (phi3) via Ollama
• Export to .csv for spreadsheets or .apkg for Anki
• Option to choose how many flashcards to generate
• Simple web UI using Streamlit

---

Installation (Mac & Windows)

Prerequisites

Make sure the following are installed:

• Python 3.9+
• Git
• Ollama (for running local phi3 model)

---

Step-by-Step Setup

1. Clone this repo

git clone git@github.com:but-pixelated/flashcard-generator.git
cd flashcard-generator

---

2. Set up virtual environment

Mac/Linux:

python3 -m venv virtualenv
source virtualenv/bin/activate

Windows:

python -m venv virtualenv
virtualenv\Scripts\activate

---

3. Install dependencies

pip install -r requirements.txt

The requirements.txt includes:

• streamlit
• pymupdf
• pandas
• genanki
• ollama

---

4. Install Ollama & Phi-3

Follow the instructions here:
https://ollama.com/download

Then run:

ollama pull phi3

This will download the phi3 model locally (~2GB+)

---

Run the App

python -m streamlit run app.py (recommended, so it does not run the global venv)
or
streamlit run app.py

Visit in browser:
http://localhost:8501

---

Usage Instructions

1. Choose Input Mode: Text or PDF
2. Enter content or upload a PDF
3. Choose number of flashcards to generate
4. Click ** Generate Flashcards**
5. View them in the app
6. Download:
   • as .csv
   • or as Anki .apkg

---

Example Flashcard Format

Q: What is the powerhouse of the cell?  
A: The mitochondria, which generates energy (ATP) through cellular respiration.

---

Export Options

• CSV: Opens in Excel/Google Sheets
• Anki .apkg: Importable directly into the Anki app

---

Troubleshooting

• No flashcards found?
  → Try reducing input size or making input clearer

• App feels slow?
  → Phi3 runs on CPU — be patient!

• Mac permission issues?
  → Try chmod +x or allow Terminal access to documents

---

File Structure

flashcard-generator/
├── app.py                 # main Streamlit app  
├── requirements.txt       # python3 dependencies  
├── README.md              # this file  
├── virtualenv/            # your Python env (not tracked in git)  

---

License

MIT License – free to use, fork, and improve.

---

Credits

Made with 💙 by @but-pixelated
Powered by:

• OpenAI’s Phi-3
• Ollama
• Streamlit
• genanki