Reinforcement learning with verifiable rewards (RLVR) has improved the reasoning capabilities of large language models (LLMs), yet their predictions remain sensitive to task-irrelevant prompt features.
← Back to all developments · Signals are a development reported by more than one source. See the methodology.