Mitigating Cognitive Inertia in Large Reasoning Models via Latent Spike Steering
通过潜在尖峰引导缓解大推理模型中的认知惯性
机构 * Department of Artificial Intelligence, Chung-Ang University, Seoul, Korea(人工智能系, Chung-Ang 大学,首尔,韩国)
专题命中 测试时计算 :reasoning(title,abstract);test-time compute(abstract);self-correction(abstract);分类 cs.LG
AI总结 STARS通过监测潜在动态,实时引导大推理模型缓解认知惯性问题,提升推理准确性和效率。
Comments 21 pages, 6 figures