隐语隐私计算 | 第4讲上手Secretflow的安装和部署

mocco 2024-03-23 12:58:53

1. miniconda3安装

https://mirrors.tuna.tsinghua.edu.cn/anaconda/miniconda/Miniconda3-py38_23.11.0-1-Linux-x86_64.sh

执行完成后，将$HOME/miniconda3/bin加入到.bash_profile的PATH中

2. Pip方式安装secretflow

conda create -n sf python==3.8
conda activate sf

pip install -U secretflow  -i https://pypi.tuna.tsinghua.edu.cn/simple

花了1个多小时，才下载安装完成

3. 尝试第一个SecretFlow程序

结果执行 `import secretflow as sf ` 这句就报错

解决方法：

pip uninstall scikit-learn

pip install scikit-learn

结果又报错

安装pip install chardet

终于顺利执行第一个测试程序

import secretflow as sf
sf.init(['alice', 'bob', 'carol'], address='local')
dev = sf.PYU('alice')
import numpy as np
data = dev(np.random.rand)(3, 4)
data

4. docker的方式运行secretflow

docker run -it secretflow/secretflow-anolis8:latest

import secretflow as sf
sf.init(parties=['alice', 'bob', 'carol'], address='local')
alice_device = sf.PYU('alice')
message_from_alice = alice_device(lambda x:x)("Hello World!")
print(sf.reveal(message_from_alice))

5. 安装SecretNote

 帮助文档  secretflow/secretnote (github.com)
 新建文件夹，并新建文件 docker-compose.yml

services:
  alice:
    image: 'secretflow/secretnote:unstable-amd64'
    platform: linux/amd64
    environment:
      - SELF_PARTY=alice
      - ALL_PARTIES=alice,bob
    ports:
      - 8090:8888
    entrypoint: /root/scripts/start.sh
    volumes:
      - /root/scripts

  bob:
    image: 'secretflow/secretnote:unstable-amd64'
    platform: linux/amd64
    environment:
      - SELF_PARTY=bob
      - ALL_PARTIES=alice,bob
    ports:
      - 8092:8888
    entrypoint: /root/scripts/start.sh
    volumes:
      - /root/scripts

然后在新建的文件夹中执行以下命令：