News/Beyond Entropy: Learning from Token-Level Distri…
Beyond Entropy: Learning from Token-Level Distributional Deviations for LLM Reasoning