Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Answered
Hey Guys! I'M Having Some Issues With Pytorch And Clearml. I Am Starting A New Task Using Task.Create And Setting Pytorch As A Requirement Under `Packages`. For Some Reason Pytorch With Cuda 12 Is Being Installed, But I Need Cuda 11. Do You Know How To Se

Hey guys! I'm having some issues with pytorch and clearml. I am starting a new task using task.create and setting pytorch as a requirement under packages. For some reason pytorch with CUDA 12 is being installed, but I need CUDA 11. Do you know how to set it to install CUDA 11?

  
  
Posted 2 months ago
Votes Newest

Answers 41


Just to make sure, run the code on the machine itself to verify that python can actually detect the driver

  
  
Posted 2 months ago

Collecting pip<20.2
Using cached pip-20.1.1-py2.py3-none-any.whl (1.5 MB)
Installing collected packages: pip
Attempting uninstall: pip
Found existing installation: pip 20.0.2
Not uninstalling pip at /usr/lib/python3/dist-packages, outside environment /usr
Can't uninstall 'pip'. No files were found to uninstall.

  
  
Posted 2 months ago

This one seems to be compatible: [nvcr.io/nvidia/pytorch:22.04-py3](http://nvcr.io/nvidia/pytorch:22.04-py3)

  
  
Posted 2 months ago

agent.cuda_version="11.2"

  
  
Posted 2 months ago

@<1523701070390366208:profile|CostlyOstrich36> I'm now running the agent with --docker , and I'm using task.create(docker="nvidia/cuda:11.0.3-cudnn8-runtime-ubuntu20.04")

  
  
Posted 2 months ago

to achieve running both the agent and the deployment on the same machine, adding --network=host to the run arguments solved it!

  
  
Posted 2 months ago

It means that there is an issue with the drivers. I suggest trying this docker image - nvcr.io/nvidia/pytorch:23.04-py3

  
  
Posted 2 months ago

Hi @<1734020162731905024:profile|RattyBluewhale45> , what version of pytorch are you specifying?

  
  
Posted 2 months ago

@<1523701070390366208:profile|CostlyOstrich36> do you have any ideas?

  
  
Posted 2 months ago

OK, then just try the docker image I suggested 🙂

  
  
Posted 2 months ago

Thank you

  
  
Posted 2 months ago

Hi @<1523701070390366208:profile|CostlyOstrich36> I am not specifying a version 🙂

  
  
Posted 2 months ago

I suggest running it in docker mode with a docker image that already has cuda installed

  
  
Posted 2 months ago

In the config file it should be something like this: agent.cuda_version="11.2" I think

  
  
Posted 2 months ago

I can install the correct torch version with this command:
pip install --pre torchvision --force-reinstall --index-url ` None ```

  
  
Posted 2 months ago

pip install --pre torchvision --force-reinstall --index-url None

  
  
Posted 2 months ago

This has been resolved now! Thank you for your help @<1523701070390366208:profile|CostlyOstrich36>

  
  
Posted 2 months ago

ERROR: This container was built for NVIDIA Driver Release 530.30 or later, but
       version 460.32.03 was detected and compatibility mode is UNAVAILABLE.

       [[System has unsupported display driver / cuda driver combination (CUDA_ERROR_SYSTEM_DRIVER_MISMATCH) cuInit()=803]]
  
  
Posted 2 months ago

I am running the agent with clearml-agent daemon --queue training

  
  
Posted 2 months ago

docker="nvidia/cuda:11.8.0-base-ubuntu20.04"

  
  
Posted 2 months ago

Isn't the problem that CUDA 12 is being installed?

  
  
Posted 2 months ago

Thank you I will try that

  
  
Posted 2 months ago

or cu11x

  
  
Posted 2 months ago

Thank you for getting back to me

  
  
Posted 2 months ago

But the process is still hanging, and not proceeding to actually running the clearml task

  
  
Posted 2 months ago

unrelated to the agent itself

  
  
Posted 2 months ago

I think it tries to get the latest one. Are you using the agent in docker mode? you can also control this via clearml.conf with agent.cuda_version

  
  
Posted 2 months ago

I have set agent{cuda_version: 11.2}

  
  
Posted 2 months ago

CUDA is the driver itself. The agent doesn't install CUDA but installs a compatible torch assuming that CUDA is properly installed.

  
  
Posted 2 months ago

What I dont understand is how to tell clearml to install this version of pytorch and torchvision, with cu118

  
  
Posted 2 months ago
6K Views
41 Answers
2 months ago
2 months ago
Tags