create_docker_2.py

# Create conda environment for docker image
from azureml.core.conda_dependencies import CondaDependencies 

# DEFINE CONDA DEPENDENCIES 
myenv = CondaDependencies.create(pip_packages=ast.literal_eval(config['docker']['pip_packages']),conda_packages=ast.literal_eval(config['train']['conda_packages']))
myenv.add_pip_package("pynacl==1.2.1")

# CREATE CONDA ENVIRONMENT FILE
with open(config['docker']['conda_env_file'],"w") as f:
    f.write(myenv.serialize_to_string())