arXiv：cs.AI· Bohan Lin, Hejia Geng, Xinyi Xie, Heng Zhou, Qinghua Xing, Chen Zhang, Yudong Zhang·· 21 小时前
Emotion2Skill：利用模型内部情感信号实现技能选择与进化
Emotion2Skill: Model-Internal Emotion Signals for Adaptive Skill Selection and Evolution
AI 导读
Emotion2Skill 框架提取 LLM 内部情感向量，用于技能选择与技能进化。在 Qwen3-8B 上，其 WebShop 成功率超 Trace2Skill 3.1 个百分点，ALFWorld 平均成功率高 5.7 个百分点，优于七个基线；相同情感向量无需重新提取即可迁移至 MATH 和 MBPP。
正文
Abstract:Skill-based LLM agents select reusable procedures from external libraries to solve complex tasks, yet their routing decisions rely primarily on task descriptions, verbal reflections, and experience-derived rules, without explicitly observing the model's internal decision state. Recent interpretability work has identified emotion representations that influence model behavior, but their use has largely focused on post-hoc analysis and output steering. Their potential to guide agent-level decisions remains underexplored. We propose Emotion2Skill, a framework that extracts LLM-internal emotion vectors and incorporates them into both skill selection and skill evolution. The framework uses these representations to inform skill choice and tracks their changes to identify problematic invocations, enabling targeted skill refinement beyond coarse outcome feedback. Across two backbones, Emotion2Skill outperforms seven baselines. On Qwen3-8B, it exceeds Trace2Skill by 3.1 percentage points in WebShop success and 5.7 in ALFWorld average success. Ablations and specificity controls support the utility of context-aligned emotion signals, while co-activation analysis reveals semantically coherent emotion--skill associations. The same emotion vectors also transfer to MATH and MBPP without re-extraction. These results establish LLM-internal emotion representations as an effective decision-level signal for orchestrating agent skill systems, extending their utility beyond interpretability and output steering. Code is available at this https URL
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2608.09248 [cs.AI] |
| (or arXiv:2608.09248v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2608.09248 arXiv-issued DOI via DataCite |
Submission history
From: Bohan Lin [view email]
[v1]
Mon, 10 Aug 2026 08:09:24 UTC (1,114 KB)
[v2]
Tue, 11 Aug 2026 02:32:19 UTC (1,114 KB)
[v3]
Tue, 13 Oct 2026 08:05:10 UTC (1,120 KB)
来源：arXiv：cs.AI · arxiv.org