Skip to main content

YiGraph Windows Local Deployment Guide

This guide provides detailed instructions on how to deploy the YiGraph system locally on Windows (CPU-only environment). The official YiGraph documentation primarily targets Linux environments. This guide highlights Windows-specific issues and provides verified solutions.

Table of Contents​

System Requirements​

Operating System​

  • Windows 10 / Windows 11 (This guide was tested on Windows 11)

Hardware​

  • CPU-only Environment: This guide is written for systems without NVIDIA GPU
  • Memory: Minimum 16GB available RAM recommended
  • Disk Space: Minimum 20GB available disk space recommended

Software​

  • Python: Version 3.11 (conda environment recommended)
  • Java: OpenJDK 11 (required by Neo4j 3.5.25)
  • Docker Desktop: Required for running Milvus vector database on Windows
  • Git: For cloning source code

Install Prerequisites​

1. Install Python Environment (using conda)​

If you don't have conda installed, download and install Miniforge3.

# Create virtual environment
conda create -n AAG python=3.11

# Activate environment
conda activate AAG

2. Install Java 11​

Windows-specific issue: Java is not included by default on Windows and must be installed manually.

Neo4j 3.5.25 requires Java 11 to run. We recommend installing Adoptium OpenJDK 11:

  1. Visit Adoptium OpenJDK 11 Downloads
  2. Download the Windows x64 installer
  3. Run the installer and follow the installation wizard
  4. Verify installation:
java -version

You should see output similar to:

openjdk version "11.0.xx"
OpenJDK Runtime Environment Temurin-11.0.x
OpenJDK 64-Bit Server VM Temurin-11.0.x

Clone Source Code​

git clone https://github.com/iDC-NEU/YiGraph.git
cd YiGraph

Modify Dependencies List​

Windows-specific issue: requirements.txt contains several Linux-only packages that must be commented out for installation to succeed.

Open the requirements.txt file, find and comment out the following lines by adding # at the beginning of each line:

Starting from nvidia-cuda-cupti-cu12 to triton, total 14 lines:

# nvidia-cuda-cupti-cu12==12.1.105
# nvidia-cuda-nvrtc-cu12==12.1.105
# nvidia-cuda-runtime-cu12==12.1.105
# nvidia-cudnn-cu12==8.9.2.26
# nvidia-cublas-cu12==12.1.3.1
# torch-scatter==2.1.1
# torch-sparse==0.6.18
# torch-cluster==1.6.3
# torch-spline-conv==1.2.2
# torch_geometric==2.4.0
# pyg-lib==0.3.0
# fa2-cython==0.3.1
# pynvml==11.5.0
# triton==2.1.0

Then, find the line milvus-lite==2.5.1 and comment it out as well:

# milvus-lite==2.5.1

Note: All NVIDIA CUDA acceleration packages and milvus-lite only support Linux and do not work on Windows. For CPU-only environments, these packages are not required. Milvus must be run via Docker on Windows.

Install Python Dependencies​

pip install -r requirements.txt

Fix pkg_resources Module Not Found Error​

Windows-specific issue: Newer versions of setuptools have removed pkg_resources, which causes startup errors.

If you encounter ModuleNotFoundError: No module named 'pkg_resources', downgrade setuptools:

pip install setuptools==65.6.0

Install CPU Version of PyTorch​

Windows-specific issue: The default installed PyTorch is CUDA version. If you don't have an NVIDIA GPU, you need to reinstall the CPU version.

# Uninstall existing torch
pip uninstall torch -y

# Install CPU version
pip install torch --index-url https://download.pytorch.org/whl/cpu

Configure Neo4j Graph Database​

1. Download Neo4j 3.5.25​

YiGraph requires Neo4j version 3.5.25. Older versions are not easily found on the official website. You can download directly using this link:

After downloading, extract to your desired location, for example: D:\neo4j\neo4j-community-3.5.25

2. Configure Neo4j​

Navigate to the conf directory, open neo4j.conf, and modify/add the following configuration:

dbms.connectors.default_listen_address=0.0.0.0
dbms.connectors.default_advertised_address=localhost
dbms.connector.bolt.listen_address=0.0.0.0:7687
dbms.connector.http.listen_address=0.0.0.0:7474
dbms.connector.https.enabled=true

3. Start Neo4j​

Open PowerShell as Administrator, navigate to Neo4j's bin directory:

cd D:\neo4j\neo4j-community-3.5.25\bin

# Start Neo4j (console mode, you can see logs)
.\neo4j.bat console

Alternatively, install as a Windows service to run in the background:

.\neo4j.bat install-service
.\neo4j.bat start-service

4. Change Default Password​

  1. Open your browser and go to http://localhost:7474
  2. Default username: neo4j, default password: neo4j
  3. After login, follow the prompts to change your password. Remember your new password
  4. Use this username and password in your YiGraph configuration

Configure Milvus Vector Database​

Windows-specific issue: milvus-lite does not support Windows. Milvus must be run using Docker.

1. Install Docker Desktop​

Download and install Docker Desktop for Windows. After installation, start Docker.

2. Create Milvus Configuration File​

Create a directory for Milvus configuration, for example D:\milvus, and create a docker-compose.yml file in that directory:

version: '3.5'

services:
  etcd:
    container_name: milvus-etcd
    image: quay.io/coreos/etcd:v3.5.0
    environment:
      - ETCD_AUTO_COMPACTION_MODE=revision
      - ETCD_AUTO_COMPACTION_RETENTION=1000
      - ETCD_QUOTA_BACKEND_BYTES=4294967296
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/volumes/etcd:/etcd
    command: etcd -advertise-client-urls=http://0.0.0.0:2379 -listen-client-urls http://0.0.0.0:2379 --data-dir /etcd
    healthcheck:
      test: ["CMD", "etcdctl", "endpoint", "health"]
      interval: 30s
      timeout: 20s
      retries: 3

  minio:
    container_name: milvus-minio
    image: minio/minio:RELEASE.2023-03-20T20-16-18Z
    environment:
      MINIO_ACCESS_KEY: minioadmin
      MINIO_SECRET_KEY: minioadmin
    ports:
      - "9001:9001"
      - "9000:9000"
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/volumes/minio:/minio_data
    command: minio server /minio_data --console-address ":9001"
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9000/minio/health/live"]
      interval: 30s
      timeout: 20s
      retries: 3

  standalone:
    container_name: milvus-standalone
    image: milvusdb/milvus:v2.5.1
    command: ["milvus", "run", "standalone"]
    security_opt:
    - seccomp:unconfined
    environment:
      ETCD_ENDPOINTS: etcd:2379
      MINIO_ADDRESS: minio:9000
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/volumes/milvus:/var/lib/milvus
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9091/healthz"]
      interval: 30s
      start_period: 90s
      timeout: 20s
      retries: 3
    ports:
      - "19530:19530"
      - "9091:9091"
    depends_on:
      - "etcd"
      - "minio"

networks:
  default:
    name: milvus

3. Start Milvus​

In PowerShell, navigate to D:\milvus and execute:

docker-compose up -d

First startup will download images, which takes a few minutes. After startup completes, you can check the status with:

docker ps

You should see three containers running: milvus-standalone, milvus-etcd, and milvus-minio.

Configure LLM API​

Edit the configuration file config/engine_config.yaml to configure your LLM API information.

Example Configuration for Volcano Engine Coding Plan​

If you use Volcano Engine's Coding Plan service, configure as follows:

reasoner:
  llm:
    provider: "openai"
    openai:
      base_url: "https://ark.cn-beijing.volces.com/api/coding/v3/"
      api_key: "Your-API-Key"
      model: "ark-code-latest" # or other available model names

Important: base_url must end with a trailing slash /, otherwise the request will fail.

Example Configuration for OpenAI​

reasoner:
  llm:
    provider: "openai"
    openai:
      base_url: "https://api.openai.com/v1/"
      api_key: "Your-API-Key"
      model: "gpt-4o-mini"

Configure Vector Database Connection​

In the same configuration file, confirm the Milvus connection configuration:

retrieval:
  database:
    vector:
      host: "localhost"
      port: 19530

This configuration generally doesn't need to be modified unless you've changed Milvus's default port.

Configure Embedding Device​

Change embedding device to cpu:

retrieval:
  embedding:
    model_name: "BAAI/bge-large-zh-v1.5"
    device: "cpu"  # Change here to cpu

Start and Verify​

1. Confirm All Services Are Running​

  • Neo4j service is started (http://localhost:7474 is accessible)
  • Milvus Docker containers are running (port 19530 is listening)

2. Activate conda environment​

conda activate AAG

3. Start YiGraph Web Service​

cd YiGraph
python web/frontend/run.py

After successful startup, the terminal will output something like:

* Running on http://0.0.0.0:5000

Open your browser and go to http://localhost:5000. You should see the YiGraph Web interface.

4. Test Analysis Functionality​

  1. Select dataset AMLSim1K
  2. Enter a test question, for example: "Please help me analyze the top 10 most important accounts in this transaction network"
  3. Click send and observe the system execution flow

If the system can complete the analysis, generate DAG visualization, and output analysis results, then deployment is successful!🎉

Successful Deployment Validation Criteria​

These are the indicators of a successful deployment:

✅ All dependency services running normally

  • Neo4j graph database - Running normally
  • Milvus vector database (Docker) - Running normally
  • MCP NetworkX computation engine - Started successfully
  • LLM API - Call successful

✅ Core analysis pipeline connected

  • Question routing and classification works
  • DAG built successfully, frontend visualization normal
  • Algorithm selection normal
  • Dependency analysis normal
  • Step execution in order normal
  • Can finally output analysis results

YiGraph successfully deployed and running on Windows CPU-only local environment 🎉

Acknowledgments​

This deployment guide is compiled based on actual troubleshooting experience. Thank you to community contributors for testing and feedback.