You can build Docker image by following:
# PROJ_DIR=/path/to/capsnet_pytorch
$ cd $PROJ_DIR/docker
$ bash build.sh
You can launch a container from the Docker image by following:
$ cd $PROJ_DIR/docker
$ bash run.sh
# Now you will be inside the container
Root of this repository is mounted to /workspace of the container.
You can start training inside the container by following.
$(docker) cd /workspace
$(docker) python main.py
You can run a command in the container already running.
This is useful to execute tensorboard to monitor training status:
$ cd $PROJ_DIR/docker
$ bash exec.sh
# Now you should be inside the container already running
$(docker) cd /workspace
$(docker) tensorboard --logdir runs
# Then, open "http://localhost:6006" from your browser.