Fu-Chun2 Abstract Thanks to the rapid development of large-scale pre-training models possessing world knowledge in recent years, embodied intelligence based on large models has achieved good results in RoboGen[42], Mimicgen[43], Scaling up and distilling down[44], DrEureka[45], Omnigrasp[46], Meta-World[47], BEHAVIOR-1K[48] ALOHA[37], UMI[38], HumanPlus[39], GELLO[40], GC-DA[41] MineDojo[31], VRB[32] Liu Y, Chen W X, Bai Y J, Liang X D, Li G B, Gao W, et al. Aligning cyber space with physical world: A comprehensive survey on embodied AI. arXiv preprint arXiv: 2407.06886, 2024. 28 Yang

网址：https://www.iyiou.com/research 邮箱：hezuo@iyiou.com 电话：010-57293241 9 A 10 701 2707-2708 C 6 4 World Trade Center, 29th Floor-Office 67, 150 Greenwich St, New York, NY 10006 ©亿欧智库-陈皓 (79125) ©亿欧智库-陈皓

The effectiveness of LogPrompt is evaluated mainly using the logHub datasets,which contains real-world logs from nine different domains, including supercomputers, distributed systems, operating systems

new risks created, as companies are developing autonomous AI: systems that can plan, act in the world, and pursue goals[1] • 学术界、开源社区和工业界影响力： • 图灵奖获得者 Bengio 和 Hinton 一作论文引用： HuggingGPT 智能时代一直没有出现像

体现了分类模型对样本的识别 能力。Recall 越高，说明模型对样本的识别能力越强。从数据看出纯 AutoGen 模型的表 现略高于其他组合。3）ALFWorld（Adaptive Learning Framework World）研究： ALFWorld 是专门用于研究和开发智能体的仿真环境，在其虚拟世界中，包含了各种场 景、任务和智能体需要完成的挑战。ALFWorld 实验发现 3 个智能体的成功率高于 2 个 智能体