Hi @<1649946171692552192:profile|EnchantingDolphin84> , what about this example?
None
Add argparser to change the configuration of the HyperParameterOptimizer class.

What do you think?

No need, you can set multiple compute resources per single autoscaler instance

You can do pip show clearml to see the clearml version

Does ClearML support running the experiments on any "serverless" environments

Can you please elaborate by what you mean "serverless"?

such that GPU resources are allocated on demand?

You can define various queues for resources according to whatever structure you want. Does that make sense?

Alternatively, is there a story for auto-scaling GPU machines based on experiments waiting in the queue and some policy?

Do you mean an autoscaler for AWS for example?

Interesting, how long ago do you figure?

Hi @<1696331935023894528:profile|BoredBee87> , role based access controls are available only in the Scale & Enterprise licenses. These licenses do support full on premise deployments. Besides role based access controls there are many other features available. I'd suggest contacting ClearML directly to hear about other options 🙂

What's the docker image that you're using?

I'm checking if I can find a way to circumvent this 🙂

To use the SDK see here:
https://clear.ml/docs/latest/docs/references/sdk/task#taskget_all

Hi @<1648134232087728128:profile|AlertFrog99> , I don't think there is anything specifically built in for that. You can fetch a list of all children and then see the latest.

Hi @<1600661428556009472:profile|HighCoyote66> , I think you need to set it in the docker compose.

Hi RattyLouse61 ,
I think packages are detected in runtime and it only shows the packages used by the script directly. When you run with ClearML-Agent, it will log all packages including dependencies that were used.

Via the SDK I assume?

I'm not sure. Maybe @<1523701087100473344:profile|SuccessfulKoala55> can help 🙂

WittyOwl57 , when creating credentials, the credentials are associated with your user. So even if you give others those credentials, the experiments in the system will show up under the user who's credentials were being used when running the experiment 🙂

Hope this helps

You mean only show children when there is nothing inside the parent?

I would suggest pipeline with decorators. But all of them would achieve your goal

Hi @<1590152201068613632:profile|StaleLeopard22> , you can simply add the extra index url as part of the agent requirements as such:

agent.package_manager.extra_index_url=["<extra_index_url>",...]

Yep, although I'm quite sure you could build some logic on top of that to manage proper queueing

Hi @<1578555761724755968:profile|GrievingKoala83> , from my understanding this is a feature to be added to the imminent release of clearml-serving

Hi @<1577468611524562944:profile|MagnificentBear85> , what version of clearml are you using?

Hi @<1558986867771183104:profile|ShakyKangaroo32> , can you please open a GitHub issue to follow up on this? I think a fix should be issued shortly afterwards

Hello CurvedHedgehog15 , I don't think there is such an option. You can however add metrics over a completed task.

Hi @<1546665634195050496:profile|SolidGoose91> , when configuring a new autoscaler you can click on '+ Add item' under compute resources and this will allow you to have another resource that is listening to another queue.

You need to set up all the resources to listen to the appropriate queues to enable this allocation of jobs according to resources.

Also in general - I wouldn't suggest having multiple autoscalers/resources listen to the same queue. 1 resource per queue. A good way to mana...

Is the pipelinecontroller also working on preemptible instances?

Hi @<1658281112104865792:profile|ExasperatedDove89> , I would suggest going over this doc 🙂
None

Yep, the setup should be very similar to minio

AgitatedDove41 Hi!

If I understand correctly you would like to run the training on AWS?

Regarding workspaces - workspaces aren't supported in the self hosted version. I think that's feature on http://app.clear.ml

Hi WickedCat12 ,

During Task.init() you can specify auto_connect_frameworks=False for the framework you're working with. However please note that this will stop auto reporting scalars etc

https://clear.ml/docs/latest/docs/references/sdk/task#taskinit