Skip to content

Latest commit

 

History

History
113 lines (85 loc) · 3.66 KB

README.md

File metadata and controls

113 lines (85 loc) · 3.66 KB
Raw

Run experiments in docker

Docker is a container solution that makes it easy to build and deploy software in a virtual environment. The examples in this folder use docker to easily deploy an experiment with all its dependencies, either on a local machine or on the cloud.

Prerequisites

Install docker on your local machine by following the installation instructions for Windows, Linux, MacOS.

Prepare a docker machine on Azure, follow the local installation steps above, then run:

docker-machine create --driver azure --azure-size Standard_D12 --azure-subscription-id <subscription-id> <machine-name>

Replace <subscription-id> with your Azure subsciption id - you can find this on the Azure dashboard after logging on to https://portal.azure.com. The <machine-name> is arbitrary.

Additional docker-machine options are listed here: https://docs.docker.com/machine/drivers/azure/ Azure machine sizes are detailed on: https://docs.microsoft.com/en-us/azure/virtual-machines/virtual-machines-linux-sizes (we recommend to use at least size Standard_D12)

Configure docker to deploy to <machine-name>. Run:

docker-machine env <machine-name>

This will provide a script / instructions on how to prepare your environment to work with .

Build the docker images

Build the required docker images:

cd docker
docker build malmo -t malmo:latest
docker build malmopy-cntk-cpu-py27 -t malmopy-cntk-cpu-py27:latest

Check to make sure that the images have been compiled:

docker images

You should see a list that includes the compiled images, e.g.,

REPOSITORY              TAG                          IMAGE ID            CREATED             SIZE
malmopy-cntk-cpu-py27   latest                       0161af81632d        29 minutes ago      5.62 GB
malmo                   latest                       1b67b8e2cfa8        41 minutes ago      1.04 GB
...

Run the experiment

Run the challenge task with an example agent:

cd malmopy-ai-challenge
docker-compose up

The experiment is set up to start a tensorboard process alongside the experiment. You can view it by pointing your browser to http://127.0.0.1:6006.

Write your own

The provided docker files load malmopy and sample code directly from the malmo-challenge git repository. To include your own code, create a file called Dockerfile with the following content:

FROM malmopy-cntk-cpu-py27:latest

# add your own experiment code here
# ADD copies content from your local machine into the docker image
ADD ai_challenge/pig_chase /local/malmo-challenge/ai_challenge/pig_chase

Build this new image using:

docker build . -t my_malmo_experiment:latest

Point the agents service in docker-compose.py to the new image by replacing image: malmopy-cntk-cpu-py27:latest with the name of the image you have just built (e.g., image:my_malmo_experiment:latest). Also check if the working directory or command need to be changed.

Then run the new experiment:

docker-compose up

Cleaning up

If you are using a docker machine on Azure, make sure to shutdown and decomission the machine when your experiments have completed, to avoid incurring costs.

To shut a machine down:

docker-machine stop <machine-name>

To remove (decomission) a machine:

docker-machine rm <machine-name>

Further reading