How to create AzureML environment for an experiment from conda specification file

Question

I want to use my existing specification file of conda environment for creating AzureML experiment environment. How could we do that?

Answer 1

Create the conda specification file and upload it to any AzureML workspace directory.

Then, use the following command

from azureml.core import Experiment
myenv = Environment.from_pip_requirements(name="myenv", file_path="path-to-pip-requirements-file")

Then, you can use the environment for your experiment.

For example:

src = ScriptRunConfig(source_directory=".",
                      script="example.py",
                      compute_target="local",
                      environment=myenv)

run = myexp.submit(config=src)

run.wait_for_completion(show_output=True)

source_directory is the directory that store the python script of the experiment(training script).