<ul><li>Reasoning-enhanced large language models (LLMs) generate intermediate reasoning steps prior to generating final answers, excelling in complex problem-solving.</li><li>Thinking Intervention is a novel paradigm designed to guide the internal reasoning processes of LLMs by strategically inserting or revising specific thinking tokens.</li><li>Comprehensive evaluations show that Thinking Intervention outperforms baseline prompting approaches, achieving significant improvements in instruction following, instruction hierarchy, and safety alignment tasks.</li><li>The research on Thinking Intervention offers a promising new avenue for controlling reasoning LLMs.</li></ul>

Effectively Controlling Reasoning Models through Thinking Intervention

Discover more