yosuke-yasuda/deep_learning_docker_cluster

docker templates for deep learning project

Boilerplates for Deep Learning Project

System Requirements

• Install docker-compose
• Install nvidia-docker (only for gpu usage)

Containers

flask_keras_2_2_4_cpu

flask api container for Keras 2.2.4

$ docker-compose up flask_keras_2_2_4_cpu

In different shell window,

$ curl -F "img=@./data/sample/gu/gu1.jpg" http://localhost:5000/images
{"result":1.0,"success":true}

jupyter_keras_2_2_4_cpu

jupyter container for Keras 2.2.4

$ docker-compose up jupyter_keras_2_2_4_cpu

Access http://localhost:8888/

See keras_2_2_4_cpu_example.ipynb to see an example

jupyter_pytorch_1_0_0_gpu

jupyter container for PyTorch 1.0.0

$ docker-compose up jupyter_pytorch_1_0_0_gpu

Access http://localhost:28888/

See pytorch_1_0_0_gpu_example.ipynb to see an example

train_keras_2_2_4_cpu

training container example for Keras 2.2.4

$ docker-compose up train_keras_2_2_4_cpu

train_keras_2_2_4_gpu

training gpu container example for Keras 2.2.4

$ docker-compose up train_keras_2_2_4_gpu

nginx

nginx container to deploy application to web server

$ docker-compose up nginx

$ curl -F "img=@./data/sample/gu/gu1.jpg" http://localhost/images
{"result":1.0,"success":true}

logger

logging container

$ docker-compose up logger

open http://localhost:6006 for tensorboard

open http://localhost:16006 for omniboard