1. 安装torch
下载
git clone https://github.com/torch/distro.git ~/torch --recursive
安装
cd ~/torch
bash install-deps
./install.sh
source ~/.bashrc
验证
$th
3. 安装显卡驱动
手头有一个Quadro K2000显卡，安装：
打开Software&Updates/Additional Drivers/NVIDIA Corporation: GK107GL[Quadro K2000]
选Using NVIDIA binary driver - version 367.57 from nvidia-367 (proprietary,tested)
验证
nvidia-settings
nvidia-smi
4. 安装CUDA-8.0
官网下载https://developer.nvidia.com/cuda-downloads，选择Linux/x6_64/Ubuntu/16.04/deb(local)/cuda-repo-ubuntu1604-8-0-local_8.0.44-1_amd64.deb，较慢，建议多线程下载。
安装
sudo dpkg -i cuda-repo-ubuntu1604-8-0-local_8.0.44-1_amd64.deb
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub
su
cat 7fa2af80.pub | apt-key add -
exit
sudo apt-get update
sudo apt-get install cuda
vi ~/.bashrc
export PATH=/usr/local/cuda-8.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
sudo reboot
验证
cd /usr/local/cuda-8.0/samples/
sudo make
cd /usr/local/cuda-8.0/samples/bin/x86_64/linux/release
./deviceQuery
网上注册下载Nvidia Deep Learning Toolkit的cuDNN
https://developer.nvidia.com/rdp/cudnn-download
安装
tar -zxf cudnn-8.0-linux-x64-v5.1.tgz
cd cuda
sudo cp include/* /usr/local/cuda/include/
sudo cp lib64/* /usr/local/cuda/lib64/
5. 安装fast-neural-style
git clone https://github.com/jcjohnson/fast-neural-style.git
cd fast_neural_style
bash models/download_style_transfer_models.sh
把某图片改变风格
th fast_neural_style.lua \
-model models/eccv16/starry_night.t7 \
-input_image images/content/chicago.jpg \
-output_image out.png \
查看图片
eog out.png
