使用 LangGraph 构建自定义 RAG 代理

概述

在本教程中，我们将使用 LangGraph 构建一个 检索 代理。

LangChain 提供了内置的 代理 实现，这些实现是使用 LangGraph 原语构建的。如果需要更深入的定制，可以直接在 LangGraph 中实现代理。本指南演示了检索代理的一个示例实现。检索 代理在您希望 LLM 决定是否从向量存储中检索上下文或直接响应用户时非常有用。

在本教程结束时，我们将完成以下操作：

1. 获取并预处理将用于检索的文档。
2. 索引这些文档以进行语义搜索并为代理创建检索器工具。
3. 构建一个可以决定何时使用检索器工具的代理式 RAG 系统。

概念

我们将涵盖以下概念：

• 使用 文档加载器、文本分割器、嵌入 和 向量存储 进行 检索
• LangGraph 图 API，包括状态、节点、边和条件边。

设置

让我们下载所需的包并设置我们的 API 密钥：

pip install -U langgraph "langchain[openai]" langchain-community langchain-text-splitters bs4

import getpass
import os


def _set_env(key: str):
    if key not in os.environ:
        os.environ[key] = getpass.getpass(f"{key}:")


_set_env("OPENAI_API_KEY")

注册 LangSmith 以快速发现问题并提高 LangGraph 项目的性能。LangSmith 允许您使用跟踪数据来调试、测试和监控使用 LangGraph 构建的 LLM 应用程序。

1. 预处理文档

1. 获取将在我们的 RAG 系统中使用的文档。我们将使用 Lilian Weng 的优秀博客 上最新的三个页面。我们首先使用 WebBaseLoader 实用程序获取页面的内容：

from langchain_community.document_loaders import WebBaseLoader

urls = [
    "https://lilianweng.github.io/posts/2024-11-28-reward-hacking/",
    "https://lilianweng.github.io/posts/2024-07-07-hallucination/",
    "https://lilianweng.github.io/posts/2024-04-12-diffusion-video/",
]

docs = [WebBaseLoader(url).load() for url in urls]

docs[0][0].page_content.strip()[:1000]

2. 将获取的文档拆分为较小的块以便索引到我们的向量存储中：

from langchain_text_splitters import RecursiveCharacterTextSplitter

docs_list = [item for sublist in docs for item in sublist]

text_splitter = RecursiveCharacterTextSplitter.from_tiktoken_encoder(
    chunk_size=100, chunk_overlap=50
)
doc_splits = text_splitter.split_documents(docs_list)

doc_splits[0].page_content.strip()

2. 创建检索器工具

现在我们已经有了拆分好的文档，我们可以将它们索引到一个向量存储中，我们将用它来进行语义搜索。

1. 使用内存中的向量存储和 OpenAI 嵌入：

from langchain_core.vectorstores import InMemoryVectorStore
from langchain_openai import OpenAIEmbeddings

vectorstore = InMemoryVectorStore.from_documents(
    documents=doc_splits, embedding=OpenAIEmbeddings()
)
retriever = vectorstore.as_retriever()

2. 使用 @tool 装饰器创建检索器工具：

from langchain.tools import tool

@tool
def retrieve_blog_posts(query: str) -> str:
    """Search and return information about Lilian Weng blog posts."""
    docs = retriever.invoke(query)
    return "\n\n".join([doc.page_content for doc in docs])

retriever_tool = retrieve_blog_posts

3. 测试该工具：

retriever_tool.invoke({"query": "types of reward hacking"})

3. 生成查询

现在我们将开始为我们的代理式 RAG 图构建组件（节点 和 边）。

请注意，组件将操作 MessagesState——包含带有聊天消息列表的 messages 键的图状态。

1. 构建 generate_query_or_respond 节点。它将调用 LLM 基于当前图状态（消息列表）生成响应。给定输入消息，它将决定使用检索器工具进行检索，或直接响应用户。请注意，我们通过 .bind_tools 给予聊天模型访问之前创建的 retriever_tool 的权限：

from langgraph.graph import MessagesState
from langchain.chat_models import init_chat_model

response_model = init_chat_model("gpt-4.1", temperature=0)


def generate_query_or_respond(state: MessagesState):
    """Call the model to generate a response based on the current state. Given
    the question, it will decide to retrieve using the retriever tool, or simply respond to the user.
    """
    response = (
        response_model
        .bind_tools([retriever_tool]).invoke(state["messages"])
    )
    return {"messages": [response]}

2. 尝试随机输入：

input = {"messages": [{"role": "user", "content": "hello!"}]}
generate_query_or_respond(input)["messages"][-1].pretty_print()

输出：

================================== Ai Message ==================================

Hello! How can I help you today?

3. 询问需要语义搜索的问题：

input = {
    "messages": [
        {
            "role": "user",
            "content": "What does Lilian Weng say about types of reward hacking?",
        }
    ]
}
generate_query_or_respond(input)["messages"][-1].pretty_print()

输出：

================================== Ai Message ==================================
Tool Calls:
retrieve_blog_posts (call_tYQxgfIlnQUDMdtAhdbXNwIM)
Call ID: call_tYQxgfIlnQUDMdtAhdbXNwIM
Args:
    query: types of reward hacking

4. 评估文档

1. 添加一个 条件边—grade_documents—以确定检索到的文档是否与问题相关。我们将使用具有结构化输出模式 GradeDocuments 的模型进行文档评估。grade_documents 函数将根据评估决策（generate_answer 或 rewrite_question）返回要前往的节点名称：

from pydantic import BaseModel, Field
from typing import Literal

GRADE_PROMPT = (
    "You are a grader assessing relevance of a retrieved document to a user question. \n "
    "Here is the retrieved document: \n\n {context} \n\n"
    "Here is the user question: {question} \n"
    "If the document contains keyword(s) or semantic meaning related to the user question, grade it as relevant. \n"
    "Give a binary score 'yes' or 'no' score to indicate whether the document is relevant to the question."
)


class GradeDocuments(BaseModel):
    """Grade documents using a binary score for relevance check."""

    binary_score: str = Field(
        description="Relevance score: 'yes' if relevant, or 'no' if not relevant"
    )


grader_model = init_chat_model("gpt-4.1", temperature=0)


def grade_documents(
    state: MessagesState,
) -> Literal["generate_answer", "rewrite_question"]:
    """Determine whether the retrieved documents are relevant to the question."""
    question = state["messages"][0].content
    context = state["messages"][-1].content

    prompt = GRADE_PROMPT.format(question=question, context=context)
    response = (
        grader_model
        .with_structured_output(GradeDocuments).invoke(
            [{"role": "user", "content": prompt}]
        )
    )
    score = response.binary_score

    if score == "yes":
        return "generate_answer"
    else:
        return "rewrite_question"

2. 在工具响应中使用不相关的文档运行此代码：

from langchain_core.messages import convert_to_messages

input = {
    "messages": convert_to_messages(
        [
            {
                "role": "user",
                "content": "What does Lilian Weng say about types of reward hacking?",
            },
            {
                "role": "assistant",
                "content": "",
                "tool_calls": [
                    {
                        "id": "1",
                        "name": "retrieve_blog_posts",
                        "args": {"query": "types of reward hacking"},
                    }
                ],
            },
            {"role": "tool", "content": "meow", "tool_call_id": "1"},
        ]
    )
}
grade_documents(input)

3. 确认相关文档被分类为相关：

input = {
    "messages": convert_to_messages(
        [
            {
                "role": "user",
                "content": "What does Lilian Weng say about types of reward hacking?",
            },
            {
                "role": "assistant",
                "content": "",
                "tool_calls": [
                    {
                        "id": "1",
                        "name": "retrieve_blog_posts",
                        "args": {"query": "types of reward hacking"},
                    }
                ],
            },
            {
                "role": "tool",
                "content": "reward hacking can be categorized into two types: environment or goal misspecification, and reward tampering",
                "tool_call_id": "1",
            },
        ]
    )
}
grade_documents(input)

5. 重写问题

1. 构建 rewrite_question 节点。检索器工具可能会返回可能不相关的文档，这表明需要改进原始用户问题。为此，我们将调用 rewrite_question 节点：

from langchain.messages import HumanMessage

REWRITE_PROMPT = (
    "Look at the input and try to reason about the underlying semantic intent / meaning.\n"
    "Here is the initial question:"
    "\n ------- \n"
    "{question}"
    "\n ------- \n"
    "Formulate an improved question:"
)


def rewrite_question(state: MessagesState):
    """Rewrite the original user question."""
    messages = state["messages"]
    question = messages[0].content
    prompt = REWRITE_PROMPT.format(question=question)
    response = response_model.invoke([{"role": "user", "content": prompt}])
    return {"messages": [HumanMessage(content=response.content)]}

2. 尝试一下：

input = {
    "messages": convert_to_messages(
        [
            {
                "role": "user",
                "content": "What does Lilian Weng say about types of reward hacking?",
            },
            {
                "role": "assistant",
                "content": "",
                "tool_calls": [
                    {
                        "id": "1",
                        "name": "retrieve_blog_posts",
                        "args": {"query": "types of reward hacking"},
                    }
                ],
            },
            {"role": "tool", "content": "meow", "tool_call_id": "1"},
        ]
    )
}

response = rewrite_question(input)
print(response["messages"][-1].content)

输出：

What are the different types of reward hacking described by Lilian Weng, and how does she explain them?

6. 生成答案

1. 构建 generate_answer 节点：如果我们通过了评估器的检查，我们可以基于原始问题和检索到的上下文生成最终答案：

GENERATE_PROMPT = (
    "You are an assistant for question-answering tasks. "
    "Use the following pieces of retrieved context to answer the question. "
    "If you don't know the answer, just say that you don't know. "
    "Use three sentences maximum and keep the answer concise.\n"
    "Question: {question} \n"
    "Context: {context}"
)


def generate_answer(state: MessagesState):
    """Generate an answer."""
    question = state["messages"][0].content
    context = state["messages"][-1].content
    prompt = GENERATE_PROMPT.format(question=question, context=context)
    response = response_model.invoke([{"role": "user", "content": prompt}])
    return {"messages": [response]}

2. 尝试一下：

input = {
    "messages": convert_to_messages(
        [
            {
                "role": "user",
                "content": "What does Lilian Weng say about types of reward hacking?",
            },
            {
                "role": "assistant",
                "content": "",
                "tool_calls": [
                    {
                        "id": "1",
                        "name": "retrieve_blog_posts",
                        "args": {"query": "types of reward hacking"},
                    }
                ],
            },
            {
                "role": "tool",
                "content": "reward hacking can be categorized into two types: environment or goal misspecification, and reward tampering",
                "tool_call_id": "1",
            },
        ]
    )
}

response = generate_answer(input)
response["messages"][-1].pretty_print()

输出：

================================== Ai Message ==================================

Lilian Weng categorizes reward hacking into two types: environment or goal misspecification, and reward tampering. She considers reward hacking as a broad concept that includes both of these categories. Reward hacking occurs when an agent exploits flaws or ambiguities in the reward function to achieve high rewards without performing the intended behaviors.

7. 组装图

现在我们将所有节点和边组装成一个完整的图：

• 从 generate_query_or_respond 开始，确定是否需要调用 retriever_tool
• 使用 tools_condition 路由到下一步：
  • 如果 generate_query_or_respond 返回了 tool_calls，调用 retriever_tool 检索上下文
  • 否则，直接响应用户
• 评估检索到的文档内容与问题的相关性（grade_documents）并路由到下一步：
  • 如果不相关，使用 rewrite_question 重写问题，然后再次调用 generate_query_or_respond
  • 如果相关，继续执行 generate_answer 并使用带检索文档上下文的 ToolMessage 生成最终响应

from langgraph.graph import StateGraph, START, END
from langgraph.prebuilt import ToolNode, tools_condition

workflow = StateGraph(MessagesState)

# Define the nodes we will cycle between
workflow.add_node(generate_query_or_respond)
workflow.add_node("retrieve", ToolNode([retriever_tool]))
workflow.add_node(rewrite_question)
workflow.add_node(generate_answer)

workflow.add_edge(START, "generate_query_or_respond")

# Decide whether to retrieve
workflow.add_conditional_edges(
    "generate_query_or_respond",
    # Assess LLM decision (call `retriever_tool` tool or respond to the user)
    tools_condition,
    {
        # Translate the condition outputs to nodes in our graph
        "tools": "retrieve",
        END: END,
    },
)

# Edges taken after the `action` node is called.
workflow.add_conditional_edges(
    "retrieve",
    # Assess agent decision
    grade_documents,
)
workflow.add_edge("generate_answer", END)
workflow.add_edge("rewrite_question", "generate_query_or_respond")

# Compile
graph = workflow.compile()

可视化该图：

from IPython.display import Image, display

display(Image(graph.get_graph().draw_mermaid_png()))

8. 运行代理式 RAG

现在让我们通过用问题运行它来测试完整的图：

for chunk in graph.stream(
    {
        "messages": [
            {
                "role": "user",
                "content": "What does Lilian Weng say about types of reward hacking?",
            }
        ]
    }
):
    for node, update in chunk.items():
        print("Update from node", node)
        update["messages"][-1].pretty_print()
        print("\n\n")

输出：

Update from node generate_query_or_respond
================================== Ai Message ==================================
Tool Calls:
  retrieve_blog_posts (call_NYu2vq4km9nNNEFqJwefWKu1)
 Call ID: call_NYu2vq4km9nNNEFqJwefWKu1
  Args:
    query: types of reward hacking



Update from node retrieve
================================= Tool Message ==================================
Name: retrieve_blog_posts

(Note: Some work defines reward tampering as a distinct category of misalignment behavior from reward hacking. But I consider reward hacking as a broader concept here.)
At a high level, reward hacking can be categorized into two types: environment or goal misspecification, and reward tampering.

Why does Reward Hacking Exist?#

Pan et al. (2022) investigated reward hacking as a function of agent capabilities, including (1) model size, (2) action space resolution, (3) observation space noise, and (4) training time. They also proposed a taxonomy of three types of misspecified proxy rewards:

Let's Define Reward Hacking#
Reward shaping in RL is challenging. Reward hacking occurs when an RL agent exploits flaws or ambiguities in the reward function to obtain high rewards without genuinely learning the intended behaviors or completing the task as designed. In recent years, several related concepts have been proposed, all referring to some form of reward hacking:



Update from node generate_answer
================================== Ai Message ==================================

Lilian Weng categorizes reward hacking into two types: environment or goal misspecification, and reward tampering. She considers reward hacking as a broad concept that includes both of these categories. Reward hacking occurs when an agent exploits flaws or ambiguities in the reward function to achieve high rewards without performing the intended behaviors.

---

Edit this page on GitHub or file an issue.

Connect these docs to Claude, VSCode, and more via MCP for real-time answers.