Hello Everybody, I Have Been Testing Clearml As An All In One Solution For Mlops In Our Team. I Deployed The Clearml Server And Everything Worked Fine

Answered

Hello everybody,
I have been testing ClearML as an all in one solution for MLOPS in our team.
I deployed the CLearML server and everything worked fine 🙂 ! The issue comes frome remote experiments . When deploying a clearml-agent in PIP or Conda mode ( we don't have access to docker on our servers). I see the following issue with pytorch package:
Exception when trying to resolve python wheel: Could not find pytorch wheel URL for: torch==1.12.1 with cuda 102 support

I found a similar issue in github : None and log file in Slack in October 20th. Using non containarized remote agent is a must in our situation 😕 do you have any solution to this problem ?
Thank for your help, I've been stuck for several days on this issue.

  				
Posted 
	one year ago

					More
				  		
  Report
		
					EagerGiraffe33
				
					0
					 × 1

Votes Newest

Answers 3

Hi @<1534496192929468416:profile|EagerGiraffe33> , what if you try to put a specific version of pytorch you've tested on your remote environment in the requirements section of the cloned task?

  				
Posted 
	one year ago

					More
				  		
  Report
		
					CostlyOstrich36
				
					0

Thanks for the reponse.
I have try this exactly and still the same issue 😕 . I think that the clearml-agent adds a specific search for the pytorch wheels withe ther PytorchRequirements class and then raises this issue .
I don't quite understand why clearml tries to solve specificaly for the pytorch package 🤔 , You could just add the repos url in pip extra and let it solve the wheels for itself ?

  				
Posted 
	one year ago

					More
				  		
  Report
		
					EagerGiraffe33
				
					0
					 × 1

Here is the full log of the experiment

  				
Posted 
	one year ago

					More
				  		
  Report
		
					EagerGiraffe33
				
					0
					 × 1

Write your answer

721 Views

3 Answers

one year ago