[簡易教學]快速建立深度學習的環境 CUDA + Tensorflow + virtualenv

前言

隨著電腦運算能力的增強, 特別是GPU的出現, 讓深度學習這個領域變得非常火紅, 進而成為目前人工智慧的主流, 由於他實在很強大, 現在不管是Google, Amazon, 或是Microsoft都在積極地推出深度學習的相關應用,


難道深度學習只有像Google這種大公司才玩得起嗎?  (答案是: No)
建立一套深度學習網路並沒有想像中那麼難, 以下就教各位如何快速地建立一個深度學習的環境

Step 1. 安裝NVidia驅動程式

一開始可以先用ubuntu-drivers devices列出你的顯卡需要裝什麼驅動, 如果沒有特別的偏好, 直接用 autoinstall來裝就可以了

ubuntu-drivers devices
sudo ubuntu-drivers autoinstall

安裝完後建議重開機

Step 2. 檢查是否安裝成功

nvidia-smi 可以用來監測GPU的使用率

nvidia-smi

如果驅動有安裝成功, 基本上可以到目前GPU的使用狀況

Step 3. 安裝 CUDA & CUDNN

# Add NVIDIA package repository

sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub

wget http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_9.1.85-1_amd64.deb

sudo apt install ./cuda-repo-ubuntu1604_9.1.85-1_amd64.deb

wget http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64/nvidia-machine-learning-repo-ubuntu1604_1.0.0-1_amd64.deb

sudo apt install ./nvidia-machine-learning-repo-ubuntu1604_1.0.0-1_amd64.deb

sudo apt update





# Install CUDA and tools. Include optional NCCL 2.x

sudo apt install cuda9.0 cuda-cublas-9-0 cuda-cufft-9-0 cuda-curand-9-0 \

    cuda-cusolver-9-0 cuda-cusparse-9-0 libcudnn7=7.2.1.38-1+cuda9.0 \

    libnccl2=2.2.13-1+cuda9.0 cuda-command-line-tools-9-0





# Optional: Install the TensorRT runtime (must be after CUDA install)

sudo apt update

sudo apt install libnvinfer4=4.1.2-1+cuda9.0

Step 4.  安裝virtualenv

sudo apt update

sudo apt install python3-dev python3-pip

sudo pip3 install -U virtualenv  # system-wide install

建立虛擬環境

virtualevn --system-site-packages -p python3 ./venv

source ./venv/bin/activate

pip install --upgrade pip

Step 5. 安裝tensorflow-gpu & keras

pip install --upgrade tensorflow

pip install keras

Step 6. 跑範例測試環境有沒有建起來

建立test.py, 將以下的範例程式碼複製貼到test.py裡面, 然後執行python3 test.py

from keras.datasets import mnist

from keras.utils import np_utils



def load_data():

        (x_train, y_train), (x_test, y_test) = mnist.load_data()

        x_train = x_train.reshape(x_train.shape[0], 28*28)

        x_test = x_test.reshape(x_test.shape[0], 28*28)

        x_train = x_train.astype('float32')

        x_test = x_test.astype('float32')

        x_train = x_train/255

        x_test = x_test/255

        y_train = np_utils.to_categorical(y_train, 10)

        y_test = np_utils.to_categorical(y_test, 10)

        return (x_train, y_train), (x_test, y_test)



import numpy as np

from keras.models import Sequential

from keras.layers.core import Dense,Activation

from keras.optimizers import  Adam





def build_model():

        model = Sequential()

        model.add(Dense(input_dim=28*28,units=500,activation='relu'))

        model.add(Dense(units=500,activation='relu'))

        model.add(Dense(units=500,activation='relu'))

        model.add(Dense(units=10,activation='softmax'))

        model.summary()

        return model





(x_train,y_train),(x_test,y_test)=load_data()

model = build_model()

model.fit(x_train,y_train,batch_size=100,epochs=20)

score = model.evaluate(x_train,y_train)

print ('\nTrain Acc:', score[1])

score = model.evaluate(x_test,y_test)

print ('\nTest Acc:', score[1])