当前位置:首页 > 报告详情

The Future is Here, A Deep Dive into Autonomous Agent.pdf

上传人: 张** 编号:175734 2024-09-09 44页 9.67MB

1、The Future is Here,The Future is Here,A Deep Dive into Autonomous AgentA Deep Dive into Autonomous Agent演讲人:宋恺涛微软亚洲研究院 高级研究员CONTENTS目 录01AI智能体的崛起02如何构建智能体03如何评估智能体04总结AI智能体的崛起The Rise of Autonomous AgentWhat is Autonomous Agent?What is Autonomous Agent?Target Target Autonomously accomplish any compl

2、ex user instructions from real-world scenarios.DefinitionDefinition A system can work as human to fulfill user targets.It can perform different actions like human behaviors to solve complex tasks.Besides,it can also simulate social behaviors like human to sense,interact and give feedback in the real

3、-world environment.An autonomous agent is a system situated within and a part of an environment that senses that environment and acts on it,over time,in pursuit of its own agenda and so as to effect what it senses in the future.-Franklin and GraesserWhy Autonomous Agent Rises?Why Autonomous Agent Ri

4、ses?The rise of LLMsIs it possible to extend the capability of LLMs to the real-world scenario?The emergent ability in LLMs allow them to complete any generative tasks in language.Agent,empowered by LLMIf LLMs are really powerful at language capability,they should utilize language like human to addr

5、ess other tasks.The Rise of LLMsThe Rise of LLMsTimelineTimelineThe Wave of Autonomous Agent Generative Agent SuperAGI ChatDev MetaGPT 2023.8ChatGPT-LLMThe revolutionary technology of large language model.2022-ChatGPTThe Rise of Agent and LLM HuggingGPT AutoGPT Visual ChatGPT AgentGPT BabyAGI LLAMA

6、2023.3GPT,BERT,GPT-2,The Fashion of pre-trained model.2018From LLMs to AgentsFrom LLMs to AgentsRoadmapRoadmapLLMLLMM-LLMM-LLMAgentAgentMulti-AgentMulti-AgentLanguageLanguageMulti-ModalMulti-ModalReal-worldReal-worldReal-worldReal-world(Society)TaskTaskFrom this roadmap,we observe that Agent is buil

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
本文主要探讨了自主智能体(Autonomous Agent)的崛起、构建和评估。宋恺涛,微软亚洲研究院高级研究员,详细阐述了自主智能体的定义、发展历程以及其在现实世界中的应用。自主智能体的崛起源于大型语言模型(LLM)的兴起,它们能够自主完成复杂的用户指令。此外,自主智能体还能模拟人类社会行为,实现与现实环境的交互。文章提出了自主智能体的设计框架,包括基础模型、功能组件和环境交互。同时,文章还介绍了EvoAgent,一种通过进化算法自动扩展智能体功能的通用方法。在评估方面,文章强调了任务自动化的重要性,并提出了TaskBench作为评估工具。总的来说,自主智能体是AGI的一种预览形式,其发展依赖于LLM的能力。未来,自主智能体的挑战在于支持持续进化、构建AI智能体社区以及部署可控的智能体。
"AI智能体如何评估其性能?" "AI智能体在未来有哪些挑战和机遇?" "如何设计一个高效的AI智能体?"
客服
商务合作
小程序
服务号
折叠