Fetching the paper…

Think, Prune, Train, Improve: Scaling Reasoning without Scaling Models · Around