MachineLearning-docker

Doc

GPU docker

• cuda 10.0
• cudnn v7
• python 3.6
• anaconda 5.2.0
• Jupyter Notebook @:8888
• Xgboost latest(gpu)
• lightgbm latest(gpu)
• tensorflow 1.13.0rc0(pip)
• tensorboard latest (pip) @:6006
• pytorch latest(pip)
• torchvision latest (pip)
• keras latest (pip)

CPU docker

• python 3.6
• anaconda 5.2.0
• Jupyter Notebook @:8888
• AwS api boto3(pip)
• Xgboost latest(pip)
• lightgbm latest

• install docker env

sudo bash ./docker_install.sh

• pull docker from AWS ecr

aws configure
aws ecr get-login --no-include-email > login.sh
sudo bash ./login.sh
sudo rm login.sh
sudo docker pull url:tag

• run gpu docker

sudo docker run --runtime=nvidia --rm -d -p 8888:8888 -p 6006:6006 --name ml-gpu -e PASSWORD= -v /home/ubuntu/notebook:/notebook <image-name>

• run cpu docker

sudo docker run --rm -d -p 8888:8888 --name ml-cpu -e PASSWORD= -v /home/ubuntu/notebook:/notebook <image-name>