What happens if I do blah/dataset_url ?

Yes using clearml-data.

Can I pass a s3 path to ds.add_files() essentially so that I can directly store a dataset without having to get the files to local and then upload again. Makes sense?

AgitatedDove14 - apologies for late reply. So to give context this in a Sagemaker notebook which has conda envs.

I use a lifecycle like this to pip install a package (a .tar.gz downloaded from s3) in a conda env- https://github.com/aws-samples/amazon-sagemaker-notebook-instance-lifecycle-config-samples/blob/master/scripts/install-pip-package-single-environment/on-start.sh

In the notebook I can do things like create experiments and so on. Now the problem is in running the cloned experimen...

pipeline code itself is pretty standard

So didn’t want to update the defaults

Will try it out. A weird one this.

Anything I am missing

If i were to push the private package to, say artifactory, is it possible to use that do the install?

Can I switch off git diff (change detection?)

AgitatedDove14 - i had not used the autoscaler since it asks for access key. Mainly looking for GPU use cases - with sagemaker one can choose any instance they want and use it, autoscaler would need set instance configured right? need to revisit. Also I want to use the k8s glue if not for this. Suggestions?

For different workloads, I need to habe different cluster scaler rules and account for different gpu needs

Would like to get to the Maturity Level 2 here

https://cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning#mlops_level_2_cicd_pipeline_automation

Running multiple k8s_daemon rightt? k8s_daemon("1xGPU") and k8s_daemon('cpu') right?

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

If i publish a keras_mnist model and experiment on, each of it gets pushed as a separate Model entity right? But there’s only one unique model with multiple different version of it

Updating to 1.1.0 gives this error:

ERROR: Could not push back task [e55e0f0ea228407a921e004f0d8f7901] to k8s pending queue [c288c73b8c434a6c8c55ebb709684b28], error: Invalid task status (Task already in requested status): current_status=queued, new_status=queued

Ah, just saw from the example that even that is doing the config pbtxt stuff - https://github.com/allegroai/clearml-serving/blob/main/examples/keras/keras_mnist.py#L51

Sorry, don’t get it

Also going off this 🙂

The GCP image and Helm chart for ClearML Server maybe slightly delayed for purely man-power reasons.

Also the pipeline ran as per this example - https://github.com/allegroai/clearml/blob/master/examples/pipeline/pipeline_controller.py

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

The image to run is empty essentially

So packages have to be installed and not just be mentioned in requirements / imported?

Ok i did a pip install -r requirements.txt and NOW it picks them up correctly

Why do you think?

Get the imports, yes.

I am doing something like this with a yaml based pipelines DSL

PipelineController with 1 task. That 1 task passed but the pipeline says running

With an hash of course 🙂

I am on 1.0.0