TranslateProject/published/201812/20181102 Create a containerized machine learning model.md

192 lines
5.5 KiB
Markdown
Raw Normal View History

2018-12-10 09:43:03 +08:00
[#]: collector: (lujun9972)
[#]: translator: (geekpi)
[#]: reviewer: (wxy)
[#]: publisher: (wxy)
2018-12-10 09:43:03 +08:00
[#]: subject: (Create a containerized machine learning model)
[#]: via: (https://fedoramagazine.org/create-containerized-machine-learning-model/)
[#]: author: (Sven Bösiger)
[#]: url: (https://linux.cn/article-10349-1.html)
2018-12-10 09:43:03 +08:00
2018-12-13 08:54:41 +08:00
创建一个容器化的机器学习模型
======
![](https://fedoramagazine.org/wp-content/uploads/2018/10/machinelearning-816x345.jpg)
2018-12-13 08:54:41 +08:00
数据科学家在创建机器学习模型后,必须将其部署到生产中。要在不同的基础架构上运行它,使用容器并通过 REST API 公开模型是部署机器学习模型的常用方法。本文演示了如何在 [Podman][3] 容器中使用 [Connexion][2] 推出使用 REST API 的 [TensorFlow][1] 机器学习模型。
2018-12-13 08:54:41 +08:00
### 准备
2018-12-13 08:54:41 +08:00
首先,使用以下命令安装 Podman
```
sudo dnf -y install podman
```
2018-12-13 08:54:41 +08:00
接下来,为容器创建一个新文件夹并切换到该目录。
```
mkdir deployment_container && cd deployment_container
```
2018-12-13 08:54:41 +08:00
### TensorFlow 模型的 REST API
2018-12-13 08:54:41 +08:00
下一步是为机器学习模型创建 REST API。这个 [github 仓库][4]包含一个预训练模型,以及能让 REST API 工作的设置。
使用以下命令在 `deployment_container` 目录中克隆它:
```
git clone https://github.com/svenboesiger/titanic_tf_ml_model.git
```
2018-12-13 08:54:41 +08:00
#### prediction.py 和 ml_model/
2018-12-13 08:54:41 +08:00
[prediction.py][5] 能进行 Tensorflow 预测,而 20x20x20 神经网络的权重位于文件夹 [ml_model/][6] 中。
#### swagger.yaml
[swagger.yaml][12] 使用 [Swagger规范][7] 定义 Connexion 库的 API。此文件包含让你的服务器提供输入参数验证、输出响应数据验证、URL 端点定义所需的所有信息。
2018-12-13 08:54:41 +08:00
额外地Connexion 还将给你提供一个简单但有用的单页 Web 应用,它演示了如何使用 Javascript 调用 API 和更新 DOM。
```
swagger: "2.0"
info:
description: This is the swagger file that goes with our server code
version: "1.0.0"
title: Tensorflow Podman Article
consumes:
- "application/json"
produces:
- "application/json"
basePath: "/"
paths:
/survival_probability:
post:
operationId: "prediction.post"
tags:
- "Prediction"
summary: "The prediction data structure provided by the server application"
description: "Retrieve the chance of surviving the titanic disaster"
parameters:
- in: body
name: passenger
required: true
schema:
$ref: '#/definitions/PredictionPost'
responses:
'201':
description: 'Survival probability of an individual Titanic passenger'
definitions:
PredictionPost:
type: object
```
2018-12-13 08:54:41 +08:00
#### server.py 和 requirements.txt
2018-12-13 08:54:41 +08:00
[server.py][8] 定义了启动 Connexion 服务器的入口点。
```
import connexion
app = connexion.App(__name__, specification_dir='./')
app.add_api('swagger.yaml')
if __name__ == '__main__':
app.run(debug=True)
```
2018-12-13 08:54:41 +08:00
[requirements.txt][9] 定义了运行程序所需的 python 包。
```
connexion
tensorflow
pandas
```
2018-12-13 08:54:41 +08:00
### 容器化!
为了让 Podman 构建映像,请在上面的准备步骤中创建的 `deployment_container` 目录中创建一个名为 `Dockerfile` 的新文件:
```
FROM fedora:28
# File Author / Maintainer
MAINTAINER Sven Boesiger <donotspam@ujelang.com>
# Update the sources
RUN dnf -y update --refresh
# Install additional dependencies
RUN dnf -y install libstdc++
RUN dnf -y autoremove
# Copy the application folder inside the container
ADD /titanic_tf_ml_model /titanic_tf_ml_model
# Get pip to download and install requirements:
RUN pip3 install -r /titanic_tf_ml_model/requirements.txt
# Expose ports
EXPOSE 5000
# Set the default directory where CMD will execute
WORKDIR /titanic_tf_ml_model
# Set the default command to execute
# when creating a new container
CMD python3 server.py
```
2018-12-13 08:54:41 +08:00
接下来,使用以下命令构建容器镜像:
```
podman build -t ml_deployment .
```
2018-12-13 08:54:41 +08:00
### 运行容器
2018-12-13 08:54:41 +08:00
随着容器镜像的构建和准备就绪,你可以使用以下命令在本地运行它:
```
podman run -p 5000:5000 ml_deployment
```
2018-12-13 08:54:41 +08:00
在 Web 浏览器中输入 [http://0.0.0.0:5000/ui][10] 访问 Swagger/Connexion UI 并测试模型:
![][11]
2018-12-13 08:54:41 +08:00
当然,你现在也可以在应用中通过 REST API 访问模型。
--------------------------------------------------------------------------------
via: https://fedoramagazine.org/create-containerized-machine-learning-model/
作者:[Sven Bösiger][a]
选题:[lujun9972][b]
2018-12-13 08:54:41 +08:00
译者:[geekpi](https://github.com/geekpi)
校对:[wxy](https://github.com/wxy)
本文由 [LCTT](https://github.com/LCTT/TranslateProject) 原创编译,[Linux中国](https://linux.cn/) 荣誉推出
[a]: https://fedoramagazine.org/author/r00nz/
[b]: https://github.com/lujun9972
[1]: https://www.tensorflow.org
[2]: https://connexion.readthedocs.io/en/latest/
[3]: https://fedoramagazine.org/running-containers-with-podman/
[4]: https://github.com/svenboesiger/titanic_tf_ml_model
[5]: https://github.com/svenboesiger/titanic_tf_ml_model/blob/master/prediction.py
[6]: https://github.com/svenboesiger/titanic_tf_ml_model/tree/master/ml_model/titanic
[7]: https://github.com/OAI/OpenAPI-Specification/blob/master/versions/2.0.md
[8]: https://github.com/svenboesiger/titanic_tf_ml_model/blob/master/server.py
[9]: https://github.com/svenboesiger/titanic_tf_ml_model/blob/master/requirements.txt
[10]: http://0.0.0.0:5000/
[11]: https://fedoramagazine.org/wp-content/uploads/2018/10/Screenshot-from-2018-10-27-14-46-56-682x1024.png
[12]: https://github.com/svenboesiger/titanic_tf_ml_model/blob/master/swagger.yaml