Meaningless Tokens, Meaningful Gains: How Activation Shifts Enhance LLM Reasoning

Open in new window