跳到主要内容

GraphRAG快速部署与调用方法详解

GraphRAG快速部署与调用方法详解

一、微软GraphRAG项目介绍与GraphRAG流程回顾​

项目地址：https://github.com/microsoft/graphrag/

检索增强生成（RAG） 是一种通过结合真实世界的信息来提升大型语言模型（LLM）输出质量的技术。RAG 技术是大多数基于 LLM 的工具中的一个重要组成部分。大多数 RAG 方法使用 向量相似性 作为检索技术，我们将其称为 基线 RAG（Baseline RAG）。

GraphRAG 则使用 知识图谱 来在推理复杂信息时显著提升问答性能。当需要对复杂数据进行推理时，GraphRAG 展示了优于基线 RAG 的性能，特别是在 知识图谱 的帮助下。

RAG 技术在帮助 LLM 推理私有数据集方面显示了很大的潜力——例如，LLM 没有在训练时接触过的、企业的专有研究、业务文档或通信数据。基线 RAG 技术最初是为了解决这个问题而提出的，但我们观察到，在某些情况下，基线 RAG 的表现并不理想。以下是几个典型的场景：

  1. 基线 RAG 很难将信息串联起来：当一个问题的答案需要通过多个不同的信息片段，并通过它们共享的属性来连接，进而提供新的综合见解时，基线 RAG 表现得很差。
  2. 例如，在回答类似“如何通过现有的数据推断出新结论”这种问题时，基线 RAG 无法很好地处理这些散布在不同文档中的相关信息，它可能会遗漏一些关键联系点。
  3. 基线 RAG 无法有效理解大型数据集或单一大文档的整体语义概念：当被要求在大量数据或复杂文档中进行总结、提炼和理解时，基线 RAG 往往表现不佳。
  4. 例如，如果问题要求对整个文档或多篇文档的主题进行总结和理解，基线 RAG 的简单向量检索方法可能无法处理文档间的复杂关系，导致对全局语义的理解不完整。

为了应对这些挑战，技术社区正在努力开发扩展和增强 RAG 的方法。微软研究院（Microsoft Research）提出的 GraphRAG 方法，使用 LLM 基于输入语料库构建 知识图谱。这个图谱与社区总结和图谱机器学习输出结合，能够在查询时增强提示（prompt）。GraphRAG 在回答以上两类问题时，展示了 显著的改进，尤其是在 复杂信息的推理能力 和 智能性 上，超越了基线 RAG 之前应用于私有数据集的其他方法。

1.GraphRAG项目简介​

GraphRAG 是微软研究院开发的一种先进的增强检索生成（RAG）框架，旨在提升语言模型（LLM）在处理复杂数据时的性能。与传统的 RAG 方法依赖向量相似性检索不同，GraphRAG 利用 知识图谱 来显著增强语言模型的问答能力，特别是在处理私有数据集或大型、复杂数据集时表现尤为出色。

2.GraphRAG核心特点​

传统的 Baseline RAG 方法在某些情况下表现不佳，尤其是当查询需要在不同信息片段之间建立联系时，或是当需要对大规模数据集进行整体理解时。GraphRAG 通过以下方式克服了这些问题：

  • 更好的连接信息点：GraphRAG 能够处理那些需要从多个数据点合成新见解的任务。

  • 更全面的理解能力：GraphRAG 更擅长对大型数据集进行全面理解，能够更好地处理复杂的抽象问题。

    而借助微软开源的GeaphRAG项目，我们可以快速做到以下事项：

  • 基于图的检索：传统的 RAG 方法使用向量相似性进行检索，而 GraphRAG 引入了知识图谱来捕捉实体、关系及其他重要元数据，从而更有效地进行推理。

  • 层次聚类：GraphRAG 使用 Leiden 技术进行层次聚类，将实体及其关系进行组织，提供更丰富的上下文信息来处理复杂的查询。

  • 多模式查询：支持多种查询模式：

    • 全局搜索：通过利用社区总结来进行全局性推理。
    • 局部搜索：通过扩展相关实体的邻居和关联概念来进行具体实体的推理。
    • DRIFT 搜索：结合局部搜索和社区信息，提供更准确和相关的答案。

  • 图机器学习：集成了图机器学习技术，提升查询响应质量，并提供来自结构化和非结构化数据的深度洞察。

  • Prompt 调优：提供调优工具，帮助根据特定数据和需求调整查询提示，从而提高结果质量。

3.GraphRAG运行流程​

索引（Indexing）过程

  1. 文本单元切分：将输入文本分割成 TextUnits，每个 TextUnit 是一个可分析的单元，用于提取关键信息。
  2. 实体和关系提取：使用 LLM 从 TextUnits 中提取实体、关系和关键声明。
  3. 图构建：构建知识图谱，使用 Leiden 算法进行实体的层次聚类。每个实体用节点表示，节点的大小和颜色分别代表实体的度数和所属社区。
  4. 社区总结：从下到上生成每个社区及其成员的总结，帮助全局理解数据集。

查询（Query）过程
索引完成后，用户可以通过不同的搜索模式进行查询：

  • 全局搜索：当我们想了解整个语料库或数据集的整体概况时，GraphRAG 可以利用 社区总结 来快速推理和获取信息。这种方式适用于大范围问题，如某个主题的总体理解。
  • 局部搜索：如果问题关注于某个特定的实体，GraphRAG 会向该实体的 邻居（即相关实体）扩展搜索，以获得更详细和精准的答案。
  • DRIFT 搜索：这是对局部搜索的增强，除了获取邻居和相关概念，还引入了 社区信息 的上下文，从而提供更深入的推理和连接。

Prompt 调优
为了获得最佳性能，GraphRAG 强烈建议进行 Prompt 调优，确保模型可以根据你的特定数据和查询需求进行优化，从而提供更准确和相关的答案。

📍更多大模型技术内容学习

扫码添加助理英英，回复“大模型”，了解更多大模型技术详情哦👇

此外，扫码回复“入群”，即可加入大模型技术社群：海量硬核独家技术干货内容+无门槛技术交流！

4.GraphRAG核心原理回顾​

二、GraphRAG安装与Indexing检索流程实现​

注，以下实验环境均为Ubuntu系统，以更好的模拟真实企业应用场景，其中大多数方法也可以直接迁移至Windows操作系统中。

【补充阅读】在线算力租赁平台AutoDL使用指南与虚拟环境创建方法​

from IPython.display import Markdown, display
with open('【补充材料】在线算力租赁平台AutoDL使用指南与虚拟环境创建方法.md', 'r') as file:
    md_content = file.read()
display(Markdown(md_content))

【补充材料】在线算力租赁平台AutoDL使用指南与虚拟环境创建方法​

Step 1.AutoDL算力租赁流程​

  • AutoDL简介

      AutoDL是一个稳定高效的算力租赁平台，AutoDL支持一键部署基础Linux基础环境、镜像环境，同时支持镜像文件保存与迁移、跨实力读取文件等，且平台显卡较为充裕，并支持多种不同计费方式，对于大模型初学者来说是非常不错的算力租赁渠道。

但与此同时，AutoDL也有一定的局限，如没有公网IP、不支持Docker镜像等。不过对于本次公开课，对于不具备本地硬件环境的同学来说，AutoDL肯定是最佳算力租赁渠道，公开课所有的演示流程也将是基于AutoDL环境的操作，便于各位同学跟着公开课内容逐步手动实现。

  • AutoDL注册与充值

      点击AutoDL官网即可注册与登录👉https://www.autodl.com/login

除了常规登录流程外，AutoDL还要求绑定微信和进行实名认证。实名认证其实是目前所有算力租赁平台的基本要求，大家按照AutoDL官网指引👉https://www.autodl.com/console/center/account/auth，一步步操作即可：

  • 账户充值

      AutoDL采用的是预付费模式，即需要先在AutoDL平台上储值，然后再租赁相关算力。AutoDL充值地址👉https://www.autodl.com/recharge。需要注意的是，若需要全程完成公开课的模型训练流程，则需要约50元左右算力成本，而若只想简单测试最终模型效果，则算力成本在10元以内，大家可以根据自己实际需求进行充值。

  • 算力租赁

      接下来就需要进行GPU租赁了，本次公开课实验需要28G-30G左右显存，同时为了给大家介绍单机多卡的模型训练方法，推荐租赁双卡3090服务器，相对来说会较为划算。此外，也可以考虑租赁更高配置的服务器，如双卡4090、双卡L20、双卡A100等，可根据实际需求进行租赁。AutoDL算力市场👉：https://www.autodl.com/market/list

这里以租赁双卡3090为例进行介绍，计费方式可选根据实际需求进行选择，然后点击3090专区，并GPU数量选择2，然后任意选择一个有闲置GPU的服务器，点击2卡可租即可：

然后再次确认计费方式，以及账户剩余费用是否足够支付，在基础镜像中，选择一个镜像：

这里我们选择Pytorch 2.3.0（即最高版本）+Python 3.12+CUDA 12.1

选择完毕后，点击立即创建即可：

注，这里的镜像配置和计费方式，都可以随时修改。

  • 查看当前服务器

      创建完成后，页面会自动跳转到当前账号的在线服务器管理页面，这里可以看到刚刚租赁的服务器正在开机中：

稍作等待，待显示运行中时，代表远程服务器已经顺利运行，接下来则需要考虑如何连接远程服务器。

Step 2.使用FinalShell连接远程服务器​

    连接远程服务器的方式有很多种，首先服务器已预安装了JupyterLab，我们可以直接通过JupyterLab中的Terminal来连接服务器，并通过命令行的方式操作远程服务器。但由于JupyterLab中Termianl界面较为原始，对新人用户操作并不友好。因此推荐使用专门的远程终端连接软件连接远程服务器，这里推荐使用FinalShel

其中finalshell软件就在本次公开课的课件中，下载后按照普通软件安装流程进行安装即可：

安装完成后，即可使用FinalShell连接远程服务器。由于接下来FinalShell连接远程服务器需要使用远程服务器的地址和端口，因此我们需要先查阅此时远程服务器的基本信息。回到AutoDL控制台页面，分别复制登陆指令和密码到任意文本编辑器中：

例如此时改服务器的登录指令为（需要找到你的主机的登录指令和密码）：

ssh -p 24503 root@conxxx

密码为：

wxxxx

记录好了之后即可打开FincalShell，使用该信息连接远程服务器。接下来打开FinalShell：

首次使用时点击左上方文件夹，进入连接管理器：

再点击创建新的连接，并在弹出的选项中选择SSH连接（Linux）：

此处名称可以随意填写，主机为conxxx，也就是登陆指令root@conxxx中的conxxx部分，端口为24503，也就是登录指令中ssh -p 24503 root@conxxx的24503部分，用户名为root，也就是root@conxxx中的conxxx中的root，而密码则是wxxxx，也就是此前复制的密码：

创建完成后回到连接管理器页面，双击刚刚创建的连接：

即可进入到连接页面，此时点击接受并保存即可：

当进入到如下页面，则说明已经连接成功：

接下来即可借助FinalShell的命令行来操作远程服务器，在远程服务器上进行模型训练相关操作。

Step 3.创建虚拟环境​

  • 打开finalshell，创建和服务器的shell连接

  • 创建MateConv虚拟环境（虚拟环境名称可自定义）
conda create --name MateConv python=3.10
conda init
source ~/.bashrc
conda activate MateConv

  • 创建Jupyter Kernel
conda install jupyterlab
conda install ipykernel
python -m ipykernel install --user --name MateConv --display-name "Python (MateConv)"
  • 创建项目主目录
 cd ~/autodl-tmp/
 mkdir MateConv

  • 打开Jupyter
cd ~/autodl-tmp/MateConv
juputer lab --allow-root

  • 本地连接Jupyter并查看虚拟环境kernel

      当服务器开启Jupyter时，Windows环境下，本地可通过AutoDL-SSH-Tools进行访问，**下图扫码添加英英助教，回复LLM**，即可领取本次公开课课件：

其中AutoDL-SSH-Tools软件就在本次公开课的课件中：

下载完成后解压缩后双击打开AutoDL.exe

在弹出的栏页中，SSH指令和密码就是AutoDL官网上给每个实例提供的密钥：

代理端口填写8889（也就是Juypter运行端口），点击开始代理，然后点击访问即可：

然后任意打开一个Kernel，选择找到刚才安装的虚拟环境即可

Mac登陆指令，打开终端并输入：（以下要填入自己的URL和port端口号）

ssh -CNg -L 8889:127.0.0.1:8889 root@YOUR_URL -p port

然后根据提示输入密码

接下来再次点击finalshell中+号，再次创建一个和服务器的连接：

然后再次进入到虚拟环境，准备后续的操作

conda activate MateConv

1.GraphRAG安装与项目创建​

  • Step 1.使用pip安装graphrag
pip install graphrag
!pip install graphrag
Looking in indexes: http://mirrors.aliyun.com/pypi/simple
Requirement already satisfied: graphrag in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (0.5.0)
Requirement already satisfied: aiofiles<25.0.0,>=24.1.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (24.1.0)
Requirement already satisfied: aiolimiter<2.0.0,>=1.1.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.1.0)
Requirement already satisfied: azure-identity<2.0.0,>=1.17.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.19.0)
Requirement already satisfied: azure-search-documents<12.0.0,>=11.4.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (11.5.2)
Requirement already satisfied: azure-storage-blob<13.0.0,>=12.22.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (12.24.0)
Requirement already satisfied: datashaper<0.0.50,>=0.0.49 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.0.49)
Requirement already satisfied: devtools<0.13.0,>=0.12.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.12.2)
Requirement already satisfied: environs<12.0.0,>=11.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (11.2.1)
Requirement already satisfied: future<2.0.0,>=1.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.0.0)
Requirement already satisfied: graspologic<4.0.0,>=3.4.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (3.4.1)
Requirement already satisfied: json-repair<0.31.0,>=0.30.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.30.2)
Requirement already satisfied: lancedb<0.14.0,>=0.13.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.13.0)
Requirement already satisfied: matplotlib<4.0.0,>=3.9.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (3.9.2)
Requirement already satisfied: networkx<4,>=3 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (3.4.2)
Requirement already satisfied: nltk==3.9.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (3.9.1)
Requirement already satisfied: numpy<2.0.0,>=1.25.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.26.4)
Requirement already satisfied: openai<2.0.0,>=1.51.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.55.1)
Requirement already satisfied: pandas<3.0.0,>=2.2.3 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (2.2.3)
Requirement already satisfied: pyaml-env<2.0.0,>=1.2.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.2.1)
Requirement already satisfied: pyarrow<16.0.0,>=15.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (15.0.2)
Requirement already satisfied: pydantic<3.0.0,>=2.9.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (2.10.1)
Requirement already satisfied: python-dotenv<2.0.0,>=1.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (1.0.1)
Requirement already satisfied: pyyaml<7.0.0,>=6.0.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (6.0.2)
Requirement already satisfied: rich<14.0.0,>=13.6.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (13.9.4)
Requirement already satisfied: tenacity<10.0.0,>=9.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (9.0.0)
Requirement already satisfied: tiktoken<0.8.0,>=0.7.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.7.0)
Requirement already satisfied: typer<0.13.0,>=0.12.5 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.12.5)
Requirement already satisfied: typing-extensions<5.0.0,>=4.12.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (4.12.2)
Requirement already satisfied: umap-learn<0.6.0,>=0.5.6 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graphrag) (0.5.7)
Requirement already satisfied: click in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from nltk==3.9.1->graphrag) (8.1.7)
Requirement already satisfied: joblib in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from nltk==3.9.1->graphrag) (1.4.2)
Requirement already satisfied: regex>=2021.8.3 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from nltk==3.9.1->graphrag) (2024.11.6)
Requirement already satisfied: tqdm in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from nltk==3.9.1->graphrag) (4.67.1)
Requirement already satisfied: azure-core>=1.31.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from azure-identity<2.0.0,>=1.17.1->graphrag) (1.32.0)
Requirement already satisfied: cryptography>=2.5 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from azure-identity<2.0.0,>=1.17.1->graphrag) (43.0.3)
Requirement already satisfied: msal>=1.30.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from azure-identity<2.0.0,>=1.17.1->graphrag) (1.31.1)
Requirement already satisfied: msal-extensions>=1.2.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from azure-identity<2.0.0,>=1.17.1->graphrag) (1.2.0)
Requirement already satisfied: azure-common>=1.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from azure-search-documents<12.0.0,>=11.4.0->graphrag) (1.1.28)
Requirement already satisfied: isodate>=0.6.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from azure-search-documents<12.0.0,>=11.4.0->graphrag) (0.7.2)
Requirement already satisfied: diskcache<6.0.0,>=5.6.3 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from datashaper<0.0.50,>=0.0.49->graphrag) (5.6.3)
Requirement already satisfied: jsonschema<5.0.0,>=4.21.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from datashaper<0.0.50,>=0.0.49->graphrag) (4.23.0)
Requirement already satisfied: asttokens<3.0.0,>=2.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from devtools<0.13.0,>=0.12.2->graphrag) (2.0.5)
Requirement already satisfied: executing>=1.1.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from devtools<0.13.0,>=0.12.2->graphrag) (2.1.0)
Requirement already satisfied: pygments>=2.15.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from devtools<0.13.0,>=0.12.2->graphrag) (2.15.1)
Requirement already satisfied: marshmallow>=3.13.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from environs<12.0.0,>=11.0.0->graphrag) (3.23.1)
Requirement already satisfied: POT<0.10,>=0.9 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (0.9.5)
Requirement already satisfied: anytree<3.0.0,>=2.12.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (2.12.1)
Requirement already satisfied: beartype<0.19.0,>=0.18.5 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (0.18.5)
Requirement already satisfied: gensim<5.0.0,>=4.3.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (4.3.3)
Requirement already satisfied: graspologic-native<2.0.0,>=1.2.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (1.2.1)
Requirement already satisfied: hyppo<0.5.0,>=0.4.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (0.4.0)
Requirement already satisfied: scikit-learn<2.0.0,>=1.4.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (1.5.2)
Requirement already satisfied: scipy==1.12.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (1.12.0)
Requirement already satisfied: seaborn<0.14.0,>=0.13.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (0.13.2)
Requirement already satisfied: statsmodels<0.15.0,>=0.14.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from graspologic<4.0.0,>=3.4.1->graphrag) (0.14.4)
Requirement already satisfied: deprecation in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (2.1.0)
Requirement already satisfied: pylance==0.17.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (0.17.0)
Requirement already satisfied: requests>=2.31.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (2.32.3)
Requirement already satisfied: retry>=0.9.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (0.9.2)
Requirement already satisfied: attrs>=21.3.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (24.2.0)
Requirement already satisfied: packaging in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (24.1)
Requirement already satisfied: cachetools in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (5.5.0)
Requirement already satisfied: overrides>=0.7 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from lancedb<0.14.0,>=0.13.0->graphrag) (7.4.0)
Requirement already satisfied: contourpy>=1.0.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (1.3.1)
Requirement already satisfied: cycler>=0.10 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (0.12.1)
Requirement already satisfied: fonttools>=4.22.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (4.55.0)
Requirement already satisfied: kiwisolver>=1.3.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (1.4.7)
Requirement already satisfied: pillow>=8 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (11.0.0)
Requirement already satisfied: pyparsing>=2.3.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (3.2.0)
Requirement already satisfied: python-dateutil>=2.7 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from matplotlib<4.0.0,>=3.9.0->graphrag) (2.9.0.post0)
Requirement already satisfied: anyio<5,>=3.5.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from openai<2.0.0,>=1.51.2->graphrag) (4.6.2)
Requirement already satisfied: distro<2,>=1.7.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from openai<2.0.0,>=1.51.2->graphrag) (1.9.0)
Requirement already satisfied: httpx<1,>=0.23.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from openai<2.0.0,>=1.51.2->graphrag) (0.27.0)
Requirement already satisfied: jiter<1,>=0.4.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from openai<2.0.0,>=1.51.2->graphrag) (0.7.1)
Requirement already satisfied: sniffio in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from openai<2.0.0,>=1.51.2->graphrag) (1.3.0)
Requirement already satisfied: pytz>=2020.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from pandas<3.0.0,>=2.2.3->graphrag) (2024.1)
Requirement already satisfied: tzdata>=2022.7 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from pandas<3.0.0,>=2.2.3->graphrag) (2024.2)
Requirement already satisfied: annotated-types>=0.6.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from pydantic<3.0.0,>=2.9.2->graphrag) (0.7.0)
Requirement already satisfied: pydantic-core==2.27.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from pydantic<3.0.0,>=2.9.2->graphrag) (2.27.1)
Requirement already satisfied: markdown-it-py>=2.2.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from rich<14.0.0,>=13.6.0->graphrag) (3.0.0)
Requirement already satisfied: shellingham>=1.3.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from typer<0.13.0,>=0.12.5->graphrag) (1.5.4)
Requirement already satisfied: numba>=0.51.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from umap-learn<0.6.0,>=0.5.6->graphrag) (0.60.0)
Requirement already satisfied: pynndescent>=0.5 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from umap-learn<0.6.0,>=0.5.6->graphrag) (0.5.13)
Requirement already satisfied: idna>=2.8 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from anyio<5,>=3.5.0->openai<2.0.0,>=1.51.2->graphrag) (3.7)
Requirement already satisfied: six in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from anytree<3.0.0,>=2.12.1->graspologic<4.0.0,>=3.4.1->graphrag) (1.16.0)
Requirement already satisfied: cffi>=1.12 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from cryptography>=2.5->azure-identity<2.0.0,>=1.17.1->graphrag) (1.17.1)
Requirement already satisfied: smart-open>=1.8.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from gensim<5.0.0,>=4.3.2->graspologic<4.0.0,>=3.4.1->graphrag) (7.0.5)
Requirement already satisfied: certifi in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from httpx<1,>=0.23.0->openai<2.0.0,>=1.51.2->graphrag) (2024.8.30)
Requirement already satisfied: httpcore==1.* in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from httpx<1,>=0.23.0->openai<2.0.0,>=1.51.2->graphrag) (1.0.2)
Requirement already satisfied: h11<0.15,>=0.13 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from httpcore==1.*->httpx<1,>=0.23.0->openai<2.0.0,>=1.51.2->graphrag) (0.14.0)
Requirement already satisfied: autograd>=1.3 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from hyppo<0.5.0,>=0.4.0->graspologic<4.0.0,>=3.4.1->graphrag) (1.7.0)
Requirement already satisfied: jsonschema-specifications>=2023.03.6 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from jsonschema<5.0.0,>=4.21.1->datashaper<0.0.50,>=0.0.49->graphrag) (2023.7.1)
Requirement already satisfied: referencing>=0.28.4 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from jsonschema<5.0.0,>=4.21.1->datashaper<0.0.50,>=0.0.49->graphrag) (0.30.2)
Requirement already satisfied: rpds-py>=0.7.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from jsonschema<5.0.0,>=4.21.1->datashaper<0.0.50,>=0.0.49->graphrag) (0.10.6)
Requirement already satisfied: mdurl~=0.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from markdown-it-py>=2.2.0->rich<14.0.0,>=13.6.0->graphrag) (0.1.2)
Requirement already satisfied: PyJWT<3,>=1.0.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from PyJWT[crypto]<3,>=1.0.0->msal>=1.30.0->azure-identity<2.0.0,>=1.17.1->graphrag) (2.10.0)
Requirement already satisfied: portalocker<3,>=1.4 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from msal-extensions>=1.2.0->azure-identity<2.0.0,>=1.17.1->graphrag) (2.10.1)
Requirement already satisfied: llvmlite<0.44,>=0.43.0dev0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from numba>=0.51.2->umap-learn<0.6.0,>=0.5.6->graphrag) (0.43.0)
Requirement already satisfied: charset-normalizer<4,>=2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from requests>=2.31.0->lancedb<0.14.0,>=0.13.0->graphrag) (3.3.2)
Requirement already satisfied: urllib3<3,>=1.21.1 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from requests>=2.31.0->lancedb<0.14.0,>=0.13.0->graphrag) (2.2.3)
Requirement already satisfied: decorator>=3.4.2 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from retry>=0.9.2->lancedb<0.14.0,>=0.13.0->graphrag) (5.1.1)
Requirement already satisfied: py<2.0.0,>=1.4.26 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from retry>=0.9.2->lancedb<0.14.0,>=0.13.0->graphrag) (1.11.0)
Requirement already satisfied: threadpoolctl>=3.1.0 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from scikit-learn<2.0.0,>=1.4.2->graspologic<4.0.0,>=3.4.1->graphrag) (3.5.0)
Requirement already satisfied: patsy>=0.5.6 in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from statsmodels<0.15.0,>=0.14.2->graspologic<4.0.0,>=3.4.1->graphrag) (1.0.1)
Requirement already satisfied: pycparser in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from cffi>=1.12->cryptography>=2.5->azure-identity<2.0.0,>=1.17.1->graphrag) (2.21)
Requirement already satisfied: wrapt in /root/miniconda3/envs/graphrag/lib/python3.11/site-packages (from smart-open>=1.8.1->gensim<5.0.0,>=4.3.2->graspologic<4.0.0,>=3.4.1->graphrag) (1.17.0)

← 上一篇
GraphRAG快速入门与原理详解
下一篇 →
GraphRAG+Ollama运行本地大模型