arXiv cs.CL· Mikey Watts (Independent Researcher), Yuchen Cui (University of California, Los Angeles)·· 4 小时前AI 评分32
改写后再行动：Vision-Language-Action 模型的语言敏感性表征与缓解
Rephrase Before You Act: Characterizing and Mitigating Language Sensitivity in Vision-Language-Action Models
AI 导读
Vision-Language-Action 模型对指令措辞高度敏感，一词之差可致成功率相差数十个百分点：π₀.₅ 对"switch on the stove"成功率为 100%，对"switch on the hot plate"仅 2%。
来源：arXiv cs.CL · arxiv.org