Distill Your Docs: an 8B Teacher, a 0.6B Student, and What Actually Transfers

An 8B reasoning model summarizes and labels documents for a retrieval index very well, at 10 seconds per document. A 0.6B model answers in under a second but classifies documents at chance level. So I made two different 8B teachers each write a textbook, and trained a 0.6B student on each, on a Mac, in about twelve minutes per student. One student became a perfect classifier. The other became a faithful writer. Neither became both, and that is the finding. The headline numbers: the student runs…