/ sandbox / embeddings
In-browser embeddings · 384-d MiniLM
Index novels client-side. Search by meaning.4 files · monaco editor↗ Open on GitHub
semantic search
// ready to embed
Three steps. One browser.
- 1Load the model (~25MB, cached forever in IndexedDB)
- 2Index a few books (streams ~paragraph-sized chunks through MiniLM)
- 3Search by meaning (cosine similarity against every chunk in RAM)