人工智慧課程

(fork from https://github.com/hunkim/PyTorchZeroToAll)

下載本課程的程式碼

mkdir ~/workspace
cd ~/workspace
git clone https://github.com/joshhu/chtpytorch.git

00

安裝必要的東西

先更新Linux!

sudo apt-get update
sudo apt-get upgrade
sudo apt-get install git

檢查Nvidia驅動程式

nvidia-smi
/usr/local/cuda/bin/nvcc -V

下載anaconda

或是

cd ~
wget https://repo.anaconda.com/archive/Anaconda3-5.2.0-Linux-x86_64.sh
chmod +x Anaconda3-5.2.0-Linux-x86_64.sh
./Anaconda3-5.2.0-Linux-x86_64.sh

建立兩個python virtual envs

conda create -n pytorch anaconda nb_conda
conda create -n py27 python=2.7 anaconda nb_conda

先切換到Pytorch的環境，並安裝pytorch

source activate pytorch
conda install pytorch-cpu torchvision-cpu -c pytorch

建立一個工作目錄

mkdir ~/workspace

檢查pytorch是否安裝成功

import torch
print(torch.__version__)

Graident Descent

線上解方程式

線上繪圖

Data Augmentation

Data Aug

Learning Rate Decay

Learning Rate Decay