How to create AzureML environement and add required packages

Question

I want to use add some packages to the environment for AzureML environment. What is the way for that?

Answer 1

There are multiple way for creating the environment for experiment.

from azureml.core.conda_dependencies import CondaDependencies

Create a Python environment for the experiment

my_env = Environment("my-env")

Ensure the required packages are installed

packages = CondaDependencies.create(conda_packages=['pandas','numpy'],
                                    pip_packages=['mlflow','azureml-mlflow'])
my_env.python.conda_dependencies = packages

Then, we can use this env as follows:

Create a script config

script_config = ScriptRunConfig(source_directory=experiment_folder,
                                script='training_script.py',
                                environment=my_env) 

submit the experiment

experiment = Experiment(workspace=ws, name='my_experiment)
run = experiment.submit(config=script_config)
RunDetails(run).show()
run.wait_for_completion()