News for hackers. Stuff that matters.
Re:LLMs Corrupt Your Documents When You Delegate by visarga on Saturday May 09, 2026 @01:56PM (#48076840)
You talk as if problem solving is a supervised (imitation) learning problem. No, it is a reinforcement learning problem, models learn by solving problems and getting rated. They generate their own training data. Optimal budget allocation is 1/3 cost pre-training, 1/3 for RL, and 1/3 on inference.

There are no comments on this comment yet.