bstollnitz / aml_online_endpoint Goto Github PK

Hi Bea,

Thank you for these tutorials, they are extremely helpful and pretty much the only ones out there that comprehensively explain how to model with AzureML.

I'm going through your online endpoints tutorial and there's something off with the environment creation.

When I run
conda env create -n aml_endpoint -f environment.yml
it just hangs on the 'Solving Environment' step indefinitely.

I typically use pipenv for my environment manager, so also tried to use that. I was able to build the models running everything in a pipenv environment, but when I ran the endpoint_1 deployment:
mlflow models predict --model-uri model --input-path "../test_data/images.csv" --content-type csv

I got the following error:
mlflow.exceptions.MlflowException: Could not find the pyenv binary. See https://github.com/pyenv-win/pyenv-win#installation for installation instructions.

I guess it requires a pyenv which I could create but would rather just duplicate your env with conda.