1
votes

I chose to use Python 3.8.1 Azure ML in Azure Machine learning studio, but when i run the command !python train.py, it uses python Anconda 3.6.9, when i downloaded python 3.8 and run the command !python38 train.py in the same dir as before, the response was python3.8: can't open file . Any idea? Also Python 3 in azure, is always busy, without anything running from my side. Thank you.

2
is the the new Azure Machine Learning? Are you using a compute instance? I assume you're using a notebook? Have you tried opening a terminal, this has been an option for a few months now. - Anders Swanson
Yes, but i would like to use the notebook option as my primary. Terminal was the last choice. - Konstantinos Zacharakis

2 Answers

1
votes

You should try adding a new Python 3.8 Kernel. Here and instructions how to add a new Kernel: https://docs.microsoft.com/en-us/azure/machine-learning/how-to-access-terminal#add-new-kernels

0
votes

Yeah I understand your pain point, and I agree that calling bash commands in a notebook cell should execute in the same conda environment as the one associated with the selected kernel of the notebook. I think this is bug, I'll flag it to the notebook feature team, but I encourage you to open a priority support ticket if you want to ensure that your problem is addressed!