1
Towards Understanding Steering Strength
ICML 2026论文深入解析模型引导技术中的强度量化问题，为AI可解释性提供新视角。
arXiv:2602.02712v2 Announce Type: replace Abstract: A popular approach to post-training control of large language models (LLMs) is the steering of int…
2
DenseSteer: Steering Small Language Models towards Dense Math Reasoning
新方法让小型语言模型实现密集数学推理，小模型也能有大智慧。
arXiv:2605.29247v1 Announce Type: new Abstract: Large language models (LLMs) demonstrate strong chain-of-thought (CoT) reasoning abilities, while smal…
3
Rethinking Visual Neglect: Steering via Context-Preference for MLLM Hallucination Mitigation
从视觉忽视视角切入，提出上下文偏好引导新方法，精准缓解多模态大模型幻觉问题
arXiv:2605.27993v1 Announce Type: new Abstract: Object hallucination remains a primary obstacle to the reliable deployment of Multimodal Large Languag…
4
How to Train Your Advisor: Steering Black-Box LLMs with Advisor Models
用小型开源模型生成动态建议，巧妙引导黑盒大模型提升输出质量，无需修改权重。
arXiv:2510.02453v3 Announce Type: replace-cross Abstract: Frontier language models are deployed as black-box services, where model weights cannot be m…