Does Anyone Have A Good Idea On How To Set The Correct Python Version For A Task? Apparently I Can Set It On A Task Level Or Via The Web-Interface.

Answered

Does anyone have a good idea on how to set the correct Python version for a task?
Apparently I can set it on a task level or via the web-interface.

from clearml import Task

task = Task.init(
    project_name='Your Project Name',
    task_name='Your Task Name',
    python='3.9'  # Specify the desired Python version
)

But when running it on a vanilla -docker i get this:

docker_cmd = nvidia/cuda:12.6.3-runtime-ubuntu24.04
[...]
Python executable with version '3.9' requested by the Task, not found in path, using '/usr/bin/python3' (v3.12.3) instead
2025-02-27 15:59:51
created virtual environment CPython3.12.3.final.0-64 in 492ms
  creator CPython3Posix(dest=/root/.clearml/venvs-builds/3.12, clear=False, no_vcs_ignore=False, global=True)
  seeder FromAppData(download=False, pip=bundle, via=copy, app_data_dir=/root/.local/share/virtualenv)
    added seed packages: pip==25.0.1
  activators BashActivator,CShellActivator,FishActivator,NushellActivator,PowerShellActivator,PythonActivator

It looks like i t creates a venv anyhow so I wonder how to set the Python version then?

  				
Posted 
	one month ago

					More  		
  Report
		
					GrumpyDog7
				
					0
					 × 1

Votes Newest

Answers 3

Hi GrumpyDog7 , it shows the reason in the log:

Python executable with version '3.9' requested by the Task, not found in path, using '/usr/bin/python3' (v3.12.3) instead

You either need a container with the relevant python version available or have it installed using the bash script section.

Makes sense?

  				
Posted 
	one month ago

					More  		
  Report
		
					CostlyOstrich36
				
					0

Yes you can set everything on the task level and of course you can also use different docker images for different python versions

  				
Posted 
	one month ago

					More  		
  Report
		
					CostlyOstrich36
				
					0

Hi John. Thanks, this makes sense. What I ideally want to achieve is one GPU queue with the autoscaler that can handle tasks with varying requirements (i.e. python version and requirements). I figure as extra_docker_shell_script is agent-level and not task level I would need to go the container route? Looks like i can set this on a task level?

  				
Posted 
	one month ago

					More  		
  Report
		
					GrumpyDog7
				
					0
					 × 1

Write your answer

163 Views

3 Answers

one month ago