Lot of us are averse to using git repo directly

Nothing. nothing apparent at least

Is there a good reference to get started with k8s glue?

Great, thanks

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

Like it said, it works, but goes into the error loop

The console output in the UI says done but the pipeline is still “running”

Found the custom backend aspect of Triton - https://github.com/triton-inference-server/python_backend
Is that the right way?

Ahhhh thanks!

I am seeing that it still picks up nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04

Thanks!

Sorry, don’t get it

Yeah got it. Was mainly wondering if k8s glue was meant for this as well or not

AgitatedDove14 - thoughts on this? I remember that it was Draft before, but maybe because it was in a notebook vs now I am running a script?

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

AgitatedDove14 - on a similar note, using this is it possible to add to requirements of task with task_overrides?

Yes I have multiple lines

Ah thanks for the info.

Yeah mostly. With k8s glue going, want to finally look at clearml-session and how people are using it.

create_task_from_functionI was looking at options to implement this just today, as part of the same remote debugging that I was talking of in this thread

no task.close()

+1 to this question

Here’s an example - https://github.com/allegroai/clearml-serving/blob/b5f5d72046f878bd09505606ca1147d93a5df069/clearml_serving/serving_service.py#L58

Only one param, just playing around

Very bad at python packaging terms - but I use pyproject.toml and python -m build that generates a wheel and a tar and I install the tar

So General would have created a General instead of Args?

Maybe related to doing in notebook. Doing a task.close() finished it as expected

Ah ok

Tried context provider for Task?

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