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

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

AgitatedDove14 sounds almost what might be needed, will give it a shot. Thanks, as always 🙂

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

As we can’t create keys in our AWS due to infosec requirements

ClearDocs? 🙂

Thanks you! Does this go as a root logging {} element in the main conf? outside SDK right?

yeah meant this, within clearml.conf:

logging {} sdk {}

Ok code suggests so. Looking for more powerful pipeline scheduling like on datasets publish, actions on model publish etc

AgitatedDove14 - any doc yet for scheduler? Is it essentially for just time based scheduling?

🙏

Was asking about using iam roles without keys

Something like this - https://aws.amazon.com/blogs/machine-learning/processing-pdf-documents-with-a-human-loop-using-amazon-textract-and-amazon-augmented-ai/

Seems like usual? https://github.com/thoughtworks/mlops-platforms#contributions

Yeah, when doing:

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

Ok, but doesn't work for me though. Can you or AgitatedDove14 help me in linking to relevant code so that I can see what's wrong?

Latest version was released 11 hours ago - https://github.com/jpadilla/pyjwt/releases/tag/2.2.0

I guess the question is - I want to use services queue for running services, and I want to do it on k8s

I guess it won’t due to the nature of services?

I think I mean if it supports the first

😕

Chance of recording being available?

AgitatedDove14 - yeah wanted to see what’s happening before disabling as I wasn’t sure if this is what’s expected.

AgitatedDove14 - i am disabling pytorch like above but still see auto models . I even see model import when running evaluation from a loaded model

Does a pipeline step behave differently?

Also is there a way to disable pytorch like this from clearml.conf?

` if project_name is None and Task.current_task() is not None:
project_name = Task.current_task().get_project_name()

if project_name is None and not Task.running_locally():
task = Task.init()
project_name = task.get_project_name() `

OK i found what’s happening:

I had an additional Task.init() - just the blank one, to get the project name. Adding the disable to that as well fixed the issue

Yeah concerns make sense.

The underlying root issue is unnecessary models being added or at least what I think are unnecessary and even happening when you load a model to test.

Do people use ClearML with huggingface transformers? The code is std transformers code.

Will create an issue.

Maybe two thing here:

If Task.init() is called in an already running task, don’t reset auto_connect_frameworks? (if i am understanding the behaviour right) Option to disable these in the clearml.conf