SKILLER：面向小型语言模型的可复用技能提取的语言级强化学习框架
原标题：SKILLER: Language-Level Reinforcement Learning for Reusable Skill Extraction in Small Language Models
AI 摘要
SKILLER 是一个针对小型开源模型的强化学习框架，通过自然语言驱动的强化学习自动生成定制技能，以降低推理成本并保持高性能。实验表明，在五个基准上，SKILLER 使用 Qwen3.5-9B 和 Qwen3.5-4B 模型，相比其他方法取得了 4.3 到 20.4 个百分点（9B）和 1.8 到 13.3 个百分点（4B）的绝对提升，并在 SkillsBench 的单技能任务上匹配了强闭源模型的性能。项目已开源。
正文节选
SKILLER: Language-Level Reinforcement Learning for Reusable Skill Extraction in Small Language Models Abstract SKILLER is a reinforcement learning framework that automatically generates tailored skills for small open-source models to reduce inference costs while maintaining high task performance. Agent skills represent a standardized format for packaging procedural knowledge and domain expertise, serving within agent harness systems as an essential mechanism to continually constrain a language m