HuggingFace Daily Papers· Rui Wang, Zishen Lai, Jiayu Liu, Cheng Qian, Jeonghwan Kim, Heng Wang, Xuejun Zhang, Jiateng Liu, Bingxuan Li, Shujin Wu, Qing Zong, Tianshi Zheng, Chunyang Li, Qihan Lin, Chen Li, Dadi Guo, Xiusi Chen, Yangqiu Song·· 7 天前AI 评分55
AdaExploreBench: 基准测试LLM智能体在动态环境中的主动探索与测试时适应能力
AdaExploreBench: Benchmarking LLM Agents for Proactive Exploration and Test-time Adaptation in Evolving Environments
AI 导读
研究团队提出AdaExploreBench基准，测试LLM智能体在动态环境中主动探索和适应的能力。实验显示，即使是最先进的GPT-6 Sol模型也仅达到61.9%的校准准确率，而提供足够证据或移除规则漂移可提高近10%的性能。
来源：HuggingFace Daily Papers · arxiv.org