Research notes by Jacob Mapel.https://jacobmapel.com/https://jacobmapel.com/research/lemario-flywheel/https://jacobmapel.com/research/lemario-flywheel/We trained a self-supervised JEPA world model on NES Super Mario Bros and ran a closed-loop flywheel — collect, retrain the world model, train a policy in imagination, repeat — for fifteen iterations. The headline iter looked like a breakthrough; a five-seed re-run revealed it was a draw from a much wider distribution than we'd been measuring. This is the project so far, and the redesign that comes next.Mon, 25 May 2026 00:00:00 GMThttps://jacobmapel.com/research/sport-transfer-ssl-followup/https://jacobmapel.com/research/sport-transfer-ssl-followup/After the first writeup, three follow-up questions kept nagging. With more compute we tested them all — and one downstream task we hadn't measured before flipped the story.Mon, 18 May 2026 00:00:00 GMThttps://jacobmapel.com/research/sport-transfer-ssl/https://jacobmapel.com/research/sport-transfer-ssl/A one-week investigation into whether self-supervised pretraining on in-domain video improves transfer to new sports tasks — and what those experiments taught us about how to actually measure SSL progress.Fri, 15 May 2026 00:00:00 GMT