Wikipedia search-by-vibes through millions of pages offline (via) Really cool demo by Lee Butterman, who built embeddings of 2 million Wikipedia pages and figured out how to serve them directly to the browser, where they are used to implement “vibes based” similarity search returning results in 250ms. Lots of interesting details about how he pulled this off, using Arrow as the file format and ONNX to run the model in the browser.
Recent articles
- Reverse engineering some updates to Claude - 31st July 2025
- Trying out Qwen3 Coder Flash using LM Studio and Open WebUI and LLM - 31st July 2025
- My 2.5 year old laptop can write Space Invaders in JavaScript now, using GLM-4.5 Air and MLX - 29th July 2025