Progress notes from Hayate, a sub-millisecond web-page categorizer distilled from an LLM.https://henriqueokuti.github.io/hayate/https://henriqueokuti.github.io/hayate/posts/03-who-teaches-hayate/https://henriqueokuti.github.io/hayate/posts/03-who-teaches-hayate/Picking the LLM that labels my training data: three prompts, four models, a fix that went too far, and a consistency number that was too good to be true.Sun, 18 Oct 2026 00:00:00 GMThttps://henriqueokuti.github.io/hayate/posts/02-twenty-thousand-pages/https://henriqueokuti.github.io/hayate/posts/02-twenty-thousand-pages/Sampling Portuguese and English pages fairly from a 150 GB index without downloading it, and getting blocked once along the way.Sat, 10 Oct 2026 00:00:00 GMThttps://henriqueokuti.github.io/hayate/posts/01-pinning-the-labels/https://henriqueokuti.github.io/hayate/posts/01-pinning-the-labels/Setting up the repo, downloading a taxonomy that keeps changing under its version number, and deciding how deep the labels go.Sat, 10 Oct 2026 00:00:00 GMT