CODE # It next stage I am able to clone this task from my clearML server and everything works fine. First problem is that the variable task is None when the run is cloned from dashboard. I would like to have a feature for training where you just do not use clearML. It means that I do not pass pname and tname at all. However i would like to treat this issue before task.init ` is called. Is there any option I can recognize that I am running it from clearML dashboard as a clone against the local run?

Hi ExasperatedCrocodile76 , what do you mean by "I would like to have a feature for training where you just do not use clearML."

Can you please elaborate?

It means that I do not pass

pname

and

tname

at all. However i would like to treat this issue before

task.init

is called.

Do you mean that after you use this code after you've cloned a task into a specific project with a specific name when run in the agent it loses it's name and project association?

Hi ExasperatedCrocodile76 , you can check the clearml function running_remotely()

Hello ! I would like to just skip whole clearML process (e.g. someone do not wants to log everything for this time). The problem I have is that if you do not provide task name and project name <- it should skip clearml process. However, when the task is cloned also project name and task name are whole Task is None before task init. I am asking if there is any way to know we are running a clone before task.init after we call task = Task.current_task()

ExasperatedCrocodile76 running_remotely() is a utility function you can import from the clearml package and does not require you to call Task.init() or initialize any clearml component. Use it as follows:
` from clearml.config import running_remotely

if running_remotely():
...
else:
... `

Thanks ! That is exactly what I meant 🙂
Have a nice day!