17-AI Agent
输入“/”快速插入内容
🕖17-AI Agent
17-AI Agent
2024年8月19日修改
AI Agents represent an exciting development in Generative AI, enabling Large Language Models (LLMs) to evolve from assistants into agents capable of taking actions. AI Agent frameworks enable developers to create applications that give LLMs access to tools and state management. These frameworks also enhance visibility, allowing users and developers to monitor the actions planned by LLMs, thereby improving experience management.
AI 代理代表了生成式 AI 的一个令人兴奋的发展，使大型语言模型 （LLMs） 能够从助手演变为能够采取行动的代理。AI 代理框架使开发人员能够创建应用程序，这些应用程序可以访问LLMs工具和状态管理。这些框架还增强了可见性，允许用户和开发人员监控 LLMs计划的行动，从而改进体验管理。
The lesson will cover the following areas:
本课程将涵盖以下几个方面：
•
Understanding what an AI Agent is - What exactly is an AI Agent?
了解什么是 AI 代理 - AI 代理到底是什么？
•
Exploring four different AI Agent Frameworks - What makes them unique?
探索四种不同的 AI 代理框架 - 是什么让它们与众不同？
•
Applying these AI Agents to different use cases - When should we use AI Agents?
将这些 AI 代理应用于不同的用例 - 我们什么时候应该使用 AI 代理？
After taking this lesson, you'll be able to:
学习完本课程后，您将能够：
•
Explain what AI Agents are and how they can be used.
解释什么是 AI 代理以及如何使用它们。
•
Have an understanding of the differences between some of the popular AI Agent Frameworks, and how they differ.
了解一些流行的 AI 代理框架之间的差异，以及它们之间的区别。
•
Understand how AI Agents function in order to build applications with them.
了解 AI 代理的功能，以便使用它们构建应用程序。
AI Agents are a very exciting field in the world of Generative AI. With this excitement comes sometimes a confusion of terms and their application. To keep things simple and inclusive of most of the tools that refer to AI Agents, we are going to use this definition:
AI 代理是生成式 AI 世界中一个非常令人兴奋的领域。随着这种兴奋，有时会对术语及其应用感到困惑。为了简单起见并包括大多数引用 AI 代理的工具，我们将使用以下定义：
AI Agents allow Large Language Models (LLMs) to perform tasks by giving them access to a state and tools.
AI 代理允许大型语言模型 （LLMs） 通过授予它们对状态和工具的访问权限来执行任务。
Let's define these terms:
让我们定义这些术语：
•
Large Language Models These are the models referred throughout this course such as GPT-3.5, GPT-4, Llama-2, etc.
大型语言模型 - 这些是本课程中引用的模型，例如 GPT-3.5、GPT-4、Llama-2 等。
•
State This refers to the context that the LLM is working in. The LLM uses the context of its past actions and the current context, guiding its decision-making for subsequent actions. AI Agent Frameworks allow developers to maintain this context easier.
状态 - 这是指正在LLM其中工作的上下文。它LLM使用其过去行动和当前环境的背景，指导其后续行动的决策。AI 代理框架使开发人员能够更轻松地维护此上下文。
•
Tools To complete the task that the user has requested and that the LLM has planned out, the LLM needs access to tools. Some examples of tools can be a database, an API, an external application or even another LLM!
工具 - 要完成用户请求并LLM已计划的任务，需要LLM访问工具。工具的一些示例可以是数据库、API、外部应用程序，甚至是另一个LLM！
These definitions will hopefully give you a good grounding going forward as we look at how they are implemented. Let's explore a few different AI Agent frameworks:
这些定义有望为我们今后研究如何实施它们打下良好的基础。让我们来探索几种不同的 AI 代理框架：
LangChain Agents 是我们上面提供的定义的一个实现。
To manage the state , it uses a built-in function called the AgentExecutor. This accepts the defined agent and the tools that are available to it.
为了管理状态，它使用了一个称为的内置函数。 AgentExecutor 它接受定义的 agent 和 tools 可用的。
The Agent Executor also stores the chat history to provide the context of the chat.
还 Agent Executor 存储聊天历史记录以提供聊天的上下文。