Ternlight – 7 MB embedding model that runs in browser (WASM)
a month ago
- Embedding model runs locally in browser without API calls.
- Compact size: 7 MB for full version, 5 MB for mini variant.
- Fast embeddings processed in ~5 ms on CPU only, no GPU required.
- Easy integration via npm package with no model download or server setup.
- Includes functions like embed and similar for tasks like semantic search.
- Example application: searching React's docs powered by the 5 MB mini variant.