Shortcuts

Environment Configuration

Determined launches trials of experiments and commands in Docker containers. The container configuration is referred to as the environment.

There are two methods of customizing this environment without explicitly specifying a Docker image: environment variables and startup-hook.sh.

Environment Variables

For both trial runners and commands, Determined allows users to configure the environment variables inside the container through the environment.environment_variables configuration field of the experiment config. The format is a list of strings in the format NAME=VALUE:

environment:
  environment_variables:
    - A=hello world
    - B=$A
    - C=${B}
    # `A`, `B`, and `C` will each have the value `hello_world` in the container.

Variables are set sequentially, which affect variables that depend on expansion of other variables.

startup-hook.sh

startup-hook.sh is a script that is called during startup of your Docker container. You can use this script to pip install packages, apt install packages, or practically anything else that you can do with bash. This script should be placed in the base directory of your model definition, (e.g. super-deep-model/startup-hook.sh if the base directory is super-deep-model/). An example startup script that installs the command line utility wget and the Python package pandas is below.

apt-get update && apt-get install -y wget
python3.6 -m pip install pandas

Official Docker Images in Determined

Determined has a set of officially supported Docker images used to launch Docker containers for experiments, commands, and any other workflow in the Determined system.

Default Image

In the current version of Determined, the experiments and commands are executed in containers with the following:

  • Ubuntu 18.04

  • CUDA 10.0

  • Python 3.6.9

  • TensorFlow 1.14.0

  • PyTorch 1.4.0

Determined will automatically select GPU-specific versions of each library when running on agents with GPUs.

In addition to the above settings, all trial runner containers are launched with additional Determined-specific harness code that orchestrates model training and evaluation in the container. Trial runner containers are also loaded with the experiment’s model definition and values of the hyperparameters for the current trial.

Note

The default images are determinedai/environments:cuda-10.0-pytorch-1.4-tf-1.14-gpu-0.3.0 and determinedai/environments:py-3.6.9-pytorch-1.4-tf-1.14-cpu-0.3.0 for GPU and CPU respectively.

TF2 Environment

Determined also supports TensorFlow 2.1 and has a Docker image you can use for experiments and commands containing the following:

  • Ubuntu 18.04

  • CUDA 10.0

  • Python 3.6.9

  • TensorFlow 2.1.0

  • PyTorch 1.4.0

This can be configured in your experiment configuration like below:

environment:
  image:
    gpu: "determinedai/environments:cuda-10.1-pytorch-1.4-tf-2.1-gpu-0.3.0"
    cpu: "determinedai/environments:py-3.6.9-pytorch-1.4-tf-2.1-cpu-0.3.0"

Custom Docker Images in Determined

While our official images contain all the dependencies needed for basic workloads, many workloads have extra dependencies. If those extra dependencies are quick to install, you may want to consider using startup-hook.sh. For situations where installing dependencies via startup-hook.sh would take too long, we suggest building your own Docker image and publishing to a Docker registry like Docker Hub. Below you will find an example of a Dockerfile. For more information on building and publishing Docker images to Docker Hub, consider following https://docs.docker.com/get-started/.

The example Dockerfile below installs conda and pip dependencies based on their respective dependency file format.

Warning

It is important to not install the TensorFlow, PyTorch, Horovod, or Apex packages as this will conflict with the base packages that are installed into Determined’s official environments.

FROM determinedai/environments:cuda-10.0-pytorch-1.4-tf-1.14-gpu-0.3.0
RUN apt-get update && apt-get install -y unzip python-opencv graphviz
COPY environment.yml /tmp/environment.yml
COPY pip_requirements.txt /tmp/pip_requirements.txt
RUN conda env update --name base --file /tmp/environment.yml && \
    conda clean --all --force-pkgs-dirs --yes
RUN eval "$(conda shell.bash hook)" && \
    conda activate base && \
    pip install --requirement /tmp/pip_requirements.txt

Assuming this image is published as det-custom-registry:det-custom-tag then you can configure the environment like:

environment:
  image: "det-custom-registry:det-custom-tag"