Maybe we should rename it?! it actually creates a Task but will not auto connect it...

Task.init should be called before pytorch distribution is called, then on each instance you need to call Task.current_task() to get the instance (and make sure the logs are tracked).

server

LOL 😊

still the same 😞

Hi ExcitedFish86

In Pytorch-Lightning I use DDP

I think a fix for pytorch multi-node / process distribution was commited to 1.0.4rc1, could you verify it solves the issue ? (rc1 should fix this specific issue)
BTW: no problem working with cleaml-server < 1

ExcitedFish86 regarding the <1 version - are you talking about ClearML Server or ClearML SDK?

Is this you case?

I can't find it in PyPi

got it

An easier fix for now will probably be some kind of warning to the user that a task is created but not connected

ohh that's why 🙂

sounds great.
BTW the code is working now out-of-the-box. Just 2 magic line - import + Task.init

oops. I used create instead of init 😳

An easier fix for now will probably be some kind of warning to the user that a task is created but not connected

That is a good point, maybe if you do not have a "main" Task, then we print the warning (with some flag to disable the warning) ?

ExcitedFish86 You came to ClearML because it's free, you stayed because of the magic 🎊 🎉

I think so. IMHO all API calls should maybe reside in a different module since they usually happen inside some control code

pip install clearml==1.0.4rc1