WordHunt

WordHunt is a Contexto-style semantic word guessing game built with Streamlit.

Setup

pip install -r requirements.txt
python build_index.py
streamlit run app.py

Initial embedding/index generation is time-consuming because it creates embeddings for 50,000 words.
Vocabulary generation uses the most frequent English words (daily-life friendly) and stores 50,000 lowercase alphabetic terms.
After data/embeddings.npy and data/faiss.index are built, gameplay is fast and runs offline.
The app computes embeddings only for user guesses during runtime.
The data/ artifacts are tracked so Streamlit deployments can run app.py without rebuilding the index.