0, deep learning 21 days learning challenge 】 【 set up learning environment

​

活动地址：CSDN21天学习挑战赛

写在前面：Participate in event opportunities

I am an ordinary office worker,There are also some work related to processing data on a daily basis,I also have a strong interest in machine learning and deep learning,自学了一段时间,But progress is slow,总结原因：
1、Lack of a teacher who can solve problems,Usually the problem is only on your own搜索+摸索,效率可想而知.
2、The learning process lacks architecture,Usually it's where you want to learn,Or pick the easy ones.
3、There is no communication platform for learning together
…
This time there is a chance to participateCSDN21天学习挑战赛 ,This learning mode solved all the pain points on my self-learning journey,Must be out of trouble,Follow the teacher to learn and improve！

1、learning environment requirements

The first lesson posted by the teacher：深度学习100例-卷积神经网络（CNN）实现mnist手写数字识别 | 第1天

K同学啊The teacher introduced the learning environment in the article：python3.6.5 + tensorflow2,编辑器推荐的:jupyter notebook,所以,工欲善其事,必先利其器,Before the formal learning and hands-on practice with the teacher,Set up the environment first.

Because of the other version installed on the computerpythonand a bunch of libraries,所以,Here I want to pretendAnaconda,来创建和管理虚拟环境.

Anaconda是一个安装、管理python相关包的软件,还自带python、Jupyter Notebook、Spyder,There are management packsconda工具,非常有用.

简单的说,它就是一个python的包、环境管理工具.可以用于在同一个机器上安装不同版本的软件包及其依赖,And able to move between different environments切换,It also comes with a large collection of popular data science packages.

如果你和我一样,Other versions existpythonOr hope that multiple versions coexist,建议安装Anaconda.

2、下载、安装Anaconda

2.1 下载Anaconda

推荐大家去官网下载

下载完成,双击安装即可

2.2 安装Anaconda

The installation process is basically all the way to the next step,唯一要注意的是,这一步：

这两个地方一定要勾选,Below is for youanaconda设置为Python3,The above is to add environment variables,This saves us the trouble of adding environment variables.

Why add to environment variables？Because of future use,We sometimes operate and configure the environment through the command line.

一路下一步,直到完成.

2.3 验证是否安装成功

使用win+r键,打开运行窗口,输入：cmd ,确定,进入命令行窗口

在命令行窗口输入：conda -V 按回车执行

返回版本号,说明安装成功.

Via a shortcut on the desktop,The main program can be opened,也可以说明安装成功.

3、配置开发环境

配置环境,可以通过：界面 和 命令行 两种方式,为了简单方便,We operate with the interface,Friends who are interested in the command line,可以自行学习.

3.1 创建开发环境

1、Via a shortcut on the desktop,可以anaconda,在左侧菜单选择Environments(虚拟环境)

2、选择下方菜单中的Create(创建)

3、Fill in the environment name in the pop-up windowname,这里我填写的是：tf21day
Then select from the drop-down boxpython版本,这里我没有找到python3.6.5,I chose the one with the same major version numberpython3.6.13

4、点击Create(创建),After a little while the environment is created,We can see the environment we just created

3.2 安装TensorFlow

anaconda默认的安装源,速度很慢,Let's go through the command line window first,配置国内镜像地址：
conda config --show channels 查看当前镜像地址
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/ Configured as a Tsinghua mirror

1、Make sure the current environment is active,That is, the green button next to the environment name is lit,如果没有亮,点击一下,切换到当前环境

2、The corresponding list on the right,It is the various modules of your current environment,下拉框选择 Not installed(未安装的),然后,在右侧搜索框输入：tensorflow,回车键 搜索：

3、found in the list of search resultstensorflow

Tick ​​and click belowapply

显示TensorFlowInstall and its related dependencies,选择apply

The installation process can be seen in the lower right corner：

3.3 Associate and usejupyter

1、选择左侧菜单中的home

2、Switch to our environment,找到jupyter,选择install安装

3、安装完毕,jupyter下方显示Launch运行

4、验证环境

运行jupyter,新建python文件

输入代码：

import tensorflow as tf
print(tf.__version__) #打印TensorFlow版本号

运行结果如下：

恭喜你,The installation configuration environment was successful！

学习日记

**
1,学习知识点

Install the configuration from scratchTensorFlowrelevant learning environment

2,学习遇到的问题

a、Mainly because the virtual environment may be a little confusing at first,But just do it,There is a need for multi-environment switching later,自然就明白了
b、jupyter是Anaconda自带安装,If your environment doesn't have the install option,You can try the current environmentjupyterLab,可以关联到root环境下的jupyter