如何在GCP上跑深度学习模型

直播视频: [video width=”1440” height=”900” mp4=”https://hacking-linux.oss-cn-hongkong.aliyuncs.com/blog/uploads/2018/06/1.mp4"\]\[/video\] [video width=”1440” height=”900” mp4=”https://hacking-linux.oss-cn-hongkong.aliyuncs.com/blog/uploads/2018/06/2.mp4"\]\[/video\] 文字材料: 1.下载install.sh wget –no-check-certificate -c https://oss.hacking-linux.com/cuda/install.sh 2.下载lubcudnn相关包: wget –no-check-certificate -c https://oss.hacking-linux.com/cuda/libcudnn7-dev\_7.1.4.18-1%2Bcuda9.0\_amd64.deb wget –no-check-certificate -c https://oss.hacking-linux.com/cuda/libcudnn7\_7.1.4.18-1%2Bcuda9.0\_amd64.deb 3.安装cuda: bash ./install.sh 4.安装libcudnn: dpkg -i libcudnn7* 5.检查是否安装成功: nvidia-smi 6.安装python3-pip apt install python3-pip -y 7.安装tensorflow-gpu pip install tensorflow-gpu 8.安装notebook pip install jupyter notebook 9.为notebook配置密码 mkdir ~/.jupyter jupyter-notebook password 10.配置jupyter-notebook监听ip与启动端口并启动(注意, 如果是root用户, 需要在后面加上–allow-root参数) jupyter-notebook –ip 0.0.0.0 –port 90