Pytorch
# conda 介绍
conda 是一个开源的软件包管理系统和环境管理软件,用于安装多个版本的软件包及其依赖关系,并在它们之间轻松切换。conda 是为 Python 程序创建的,类似于 Linux、MacOS、Windows,也可以打包和分发其他软件。 注!必须在 cmd 里面才可以,在 powershell 里面输入命令有些是无效的
https://www.anaconda.com/download 通过该地址,填写邮件,然后通过邮件去下载安装包,安装完成后。
conda 分为 anaconda 和 miniconda,anaconda 是一个包含了许多常用库的集合版本,miniconda 是精简版本(只包含 conda、pip、zlib、python 以及它们所需的包),剩余的通过 conda install command 命令自行安装即可;
# 安装
下载后需要查看环境变量,如果没有配置,需要自己配置
D:\tools\conda\Library\mingw-w64\bin
D:\tools\conda\Library\bin
D:\tools\conda\Scripts
D:\tools\conda
2
3
4
如果在下载的 miniconda 没找到 mingw-w64 目录,那么需要自己安装,通过下面命令就会安装一个 mingw-w64 目录
conda install m2w64-toolchain
# conda 命令
检查版本
conda --version
升级 conda
conda update conda
添加镜像源
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/bioconda/
conda config --add channels https://mirrors.bfsu.edu.cn/anaconda/cloud/bioconda/
conda config --add channels https://mirrors.bfsu.edu.cn/anaconda/cloud/conda-forge/
conda config --add channels https://mirrors.bfsu.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.bfsu.edu.cn/anaconda/pkgs/main/
下面这个我没用过, 可以添加一下试试看.
conda config --add channels https://mirrors.bfsu.edu.cn/anaconda/pkgs/r/
目前国内提供conda镜像的大学
清华大学: https://mirrors.tuna.tsinghua.edu.cn/help/anaconda/
北京外国语大学: https://mirrors.bfsu.edu.cn/help/anaconda/
南京邮电大学: https://mirrors.njupt.edu.cn/
南京大学: http://mirrors.nju.edu.cn/
重庆邮电大学: http://mirror.cqupt.edu.cn/
上海交通大学: https://mirror.sjtu.edu.cn/
哈尔滨工业大学: http://mirrors.hit.edu.cn/#/home
(目测哈工大的镜像同步的是最勤最新的)
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
查看已经添加的渠道
conda config --get channels
恢复默认镜像源
conda config --remove-key channels
# scikit-learn
Python 语言中专门针对机器学习应用而发展起来的一款开源框架 (算法库),可以实现 数据预处理
、 分类
、 回归
、 降维
、 模型选择
等常用的机器学习算法。其特点是集成各类机器学习算法,安装使用便捷,案例和文档丰富,但仅支持 Python 语言,不支持深度学习和强化学习。通过官网可查看其能解决的问题及对应案例,如线性回归属于回归问题,官网有相关模型介绍和调用方法。
下载速度过慢 我们也可以使用清华镜像源来下载安装
pip install scikit-learn -i https://pypi.tuna.tsinghua.edu.cn/simple
创建一个虚拟环境
conda create -n pytorch python=3.8
- -n 指定虚拟环境名称
- python=3.8 指定 python 版本
切换虚拟环境
conda activate pytorch
查看虚拟环境有哪些包
pip list
# Pytorch
官网地址:https://pytorch.org/ 进入官网后选择合适的 pytorch 版本,如果你没有 GPU,直接选择 CPU,如果你有 GPU,需要确认你的 GPU 是否支持 CUDA
查看 GPU 是否支持 CUDA,选择帮助 -> 系统信息
使用 nvidia-smi 检查 CUDA 版本是否兼容,CUDA9.2 以上只支持 Driver Version: 396.26 以上的
# matplotlib
matplotlib 是一个绘图库,它提供了许多绘图工具,如折线图、饼图、散点图、直方图等,并且可以生成多种格式的图表,如 PDF、SVG、PNG 等。学习地址:https://www.runoob.com/matplotlib/matplotlib-tutorial.html
安装 matplotlib
pip install -U matplotlib
案例
import matplotlib.pyplot as plt
if __name__ == '__main__':
# 表示做一个 2行1列的表格,在表格第1行作图
fig1 = plt.subplot(2,1,1)
plt.scatter([1,2,3,4],[5,6,7,8]) # 线型图
# 表示做一个 2行1列的表格,在表格第2行作图
fig2 = plt.subplot(2,1,1)
plt.scatter([1,2,3,4],[5,6,7,8]) # 线型图
2
3
4
5
6
7
8
9
10