@<1523701205467926528:profile|AgitatedDove14> - any thoughts on this. Would like to use profile / iam roles as well.

Was asking about using iam roles without keys

Interesting. How do you do PVC? by using the yaml template optionn?

If you don’t mind, can you point me at the code where this happens?

Do people generally update the same model “entry”? That feels so wrong to me…how do you reproduce a older model version or do a rollback etc?

Any reason to not have those as two datasets?

Yeah was planning to use nested projects for that

Can makes some PRs around this. Playing with some changes already for my setup

(I need this because I refer to datasets in the same project but without specifying the project name)

I saw that the debug param wasn’t adding anything additional for this?

Ok it’s here? https://github.com/allegroai/clearml/blob/master/clearml/backend_interface/task/task.py#L1352

Now do Task.current_task() and it creates a new task

To confirm, if i have fresh venv with no dependency installed except clearml

I have a requirements.txt file in root, and a script at scripts/script1.py

The script1.py does task.init(), execute_remotely and then imports few dependenceies

Now I run python scripts/script1.py

And it should pick up the installed packages correctly?

Think I will have to fork and play around with it 🙂

I am not running in Docker mode

forking and using the latest code fixes the boto issue at least

I am doing Task.init but it’s not adding expected libraries imported in the script or from requirements.txt

CynicalBee90 - on the platform agnostic aspect - dvc does it with the CLI right? It that what made you give a green checkmark for it?

Do we support GPUs in a) docker mode b) k8s glue?

AgitatedDove14 is it possible to get the pipeline task running a step in a step? Is task.parent something that could help?

Yeah, when doing:

task.set_base_docker( "nvidia/cuda:11.4.1-cudnn8-runtime-ubuntu20.04", docker_setup_bash_script=script, )

The helm chart installs a agentservice, how is that related if at all?

Gitlab has support for S3 based cache btw.

Yeah please if you can share some general active ones to discuss both algos and engineering side

With the human activity being a step where some manual validations, annotations, feedback might be required

Is Task.current_task() creating a task?

As in run a training experiment, then a test/validation experiment to choose best model etc etc and also have a human validate sample results via annotations all as part of a pipeline

I would prefer controlled behavior than some available version being used. Here triggered a bunch of jobs that all went fine and even evaluations were fine and then when we triggered a inference deploy it failed

Thanks

Looks like Task.current_task() is indeed None in this case. Bit of log below where I print(Task.current_task()) as first step in the script

Environment setup completed successfully Starting Task Execution: None