Skip to main content
Open In ColabOpen on GitHub

上下文智能重排序器

Contextual AI 的指令跟随重排序器是全球首个能够根据特定标准（如时效性、来源和元数据）遵循自定义指令来优先处理文档的重排序器。它在 BEIR 基准测试中表现出色（得分为 61.2，显著超越竞争对手），为企业 RAG 应用提供了前所未有的控制力和准确性。

关键能力​

  • 指令跟随: 通过自然语言命令动态控制文档排名
  • 冲突解决: 智能处理来自多个知识源的矛盾信息
  • 卓越的准确性: 在行业基准测试中实现最先进的性能
  • 无缝集成: 替换现有RAG管道中的重排序器，轻松实现即插即用

重排序器在解决企业知识库中的现实问题方面表现出色，例如优先选择最新文档而非过时文档，或倾向于内部文档而非外部来源。

要了解更多关于我们指令跟随重排序器的信息并查看其实际应用示例，请访问我们的产品概述。

有关Contextual AI产品全面的文档，请访问我们的开发者门户。

此集成需要 contextual-client Python SDK。了解更多相关信息，请点击这里。

概览​

此集成调用了Contextual AI的Grounded语言模型。

集成详情​

类包本地可序列化的JS 支持软件包下载最新包裹
ContextualReranklangchain-contextual❌beta❌PyPI - DownloadsPyPI - Version

设置​

要访问Contextual的重排序模型，您需要创建一个Contextual AI帐户，获取API密钥，并安装langchain-contextual集成包。

凭据​

前往 app.contextual.ai 注册 Contextual 并生成 API 密钥。完成后，请设置 CONTEXTUAL_AI_API_KEY 环境变量：

import getpass
import os

if not os.getenv("CONTEXTUAL_AI_API_KEY"):
    os.environ["CONTEXTUAL_AI_API_KEY"] = getpass.getpass(
        "Enter your Contextual API key: "
    )

安装​

LangChain 上下文集成位于 langchain-contextual 包中：

%pip install -qU langchain-contextual

实例化​

上下文重排序器的参数是：

参数类型描述
documentslist[Document]A sequence of documents to rerank. Any metadata contained in the documents will also be used for reranking.
querystrThe query to use for reranking.
modelstrThe version of the reranker to use. Currently, we just have "ctxl-rerank-en-v1-instruct".
top_nOptional[int]The number of results to return. If None returns all results. Defaults to self.top_n.
instructionOptional[str]The instruction to be used for the reranker.
callbacksOptional[Callbacks]Callbacks to run during the compression process.
from langchain_contextual import ContextualRerank

api_key = ""
model = "ctxl-rerank-en-v1-instruct"

compressor = ContextualRerank(
    model=model,
    api_key=api_key,
)

使用​

首先，我们将设置全局变量和示例，并实例化我们的重排序客户端。

from langchain_core.documents import Document

query = "What is the current enterprise pricing for the RTX 5090 GPU for bulk orders?"
instruction = "Prioritize internal sales documents over market analysis reports. More recent documents should be weighted higher. Enterprise portal content supersedes distributor communications."

document_contents = [
    "Following detailed cost analysis and market research, we have implemented the following changes: AI training clusters will see a 15% uplift in raw compute performance, enterprise support packages are being restructured, and bulk procurement programs (100+ units) for the RTX 5090 Enterprise series will operate on a $2,899 baseline.",
    "Enterprise pricing for the RTX 5090 GPU bulk orders (100+ units) is currently set at $3,100-$3,300 per unit. This pricing for RTX 5090 enterprise bulk orders has been confirmed across all major distribution channels.",
    "RTX 5090 Enterprise GPU requires 450W TDP and 20% cooling overhead.",
]

metadata = [
    {
        "Date": "January 15, 2025",
        "Source": "NVIDIA Enterprise Sales Portal",
        "Classification": "Internal Use Only",
    },
    {"Date": "11/30/2023", "Source": "TechAnalytics Research Group"},
    {
        "Date": "January 25, 2025",
        "Source": "NVIDIA Enterprise Sales Portal",
        "Classification": "Internal Use Only",
    },
]

documents = [
    Document(page_content=content, metadata=metadata[i])
    for i, content in enumerate(document_contents)
]
reranked_documents = compressor.compress_documents(
    query=query,
    instruction=instruction,
    documents=documents,
)
API 参考：文档

在链中使用​

示例即将推出。

API 参考​

有关ChatContextual所有功能和配置的详细文档，请访问Github页面： https://github.com/ContextualAI//langchain-contextual