Running RAG with a (small) language model in the browser

My blog does semantic search in your browser. Now it also runs a (small) language model in your browser.