Can you please add a stand alone code snippet that reproduces this?
Just to make sure I understand the flow - you run an experiment and create it inside project 'my_example'
Afterwards you run a pipeline and you specify the controller 'my_example'.
This will make 'my_example' into a hidden project
Am I getting it right?
You can set up username & password, it's in the documentation 🙂
Don't think so, this is something to escalate to the IT
In the HPO application I see the following explanation:
'Maximum iterations per experiment after which it will be stopped. Iterations are based on the experiments' own reporting (for example, if experiments report every epoch, then iterations=epochs)'
Are you sure you migrated all the data correctly?
Hi @<1727497172041076736:profile|TightSheep99> , allegroai package is part of the enterprise and not available publicly, you must be looking at documentation related to the HyperDatasets
Hi @<1618418423996354560:profile|JealousMole49> , why not just use different datasets? Just to make sure I'm understanding correctly - you have a duplication of data on both s3 and local?
ReassuredTiger98 , BitterLeopard33 , I think I've encountered this 4GB http limit before. I think this should be fixed in next SDK release 🙂
Hi @<1544128920209592320:profile|BewilderedLeopard78> , I don't think there is such an option currently. Maybe open a GitHub feature request to track this 🙂
ElatedChimpanzee91 , hi 🙂
I think you can enlarge the graph to see the entire thing OR maybe try adding \n in the title, maybe that would work
Hi @<1547028131527790592:profile|PleasantOtter67> , nothing out of the box. You can however quite easily extract all that information and inject it into a csv programmatically.
I think the bigger question is how would you break it down? Each experiment has several nested properties.
Hi @<1546665634195050496:profile|SolidGoose91> , I think this capability exists when running pipelines. The pipeline controller will detect spot instances that failed and will retry running them.
Are you using the PRO or the open source auto scaler?
Shouldn't be 🙂
Did you notice any difference?
What versions of ClearML/matplotlib are you using?
Can you give an example of a pipeline to play with?
Are you running self deployed?
I don't think so. However you can use the API as well 🙂
I mean when you manually setup your environment, how do you install those packages?
ScaryBluewhale66 , please look in:
https://clear.ml/docs/latest/docs/references/sdk/task#taskinit
The relevant section for you is auto_connect_frameworks
The usage would be along these lines:Task.init(..., auto_connect_frameworks={'matplotlib': False})
@<1524560082761682944:profile|MammothParrot39> , I'm not sure what you mean, if it's in draft, why do you expect it to run?
From the screenshots provided you ticked 'cpu' mode AND I think the machine that you're using n1-standard-1 is a cpu only machine, if I'm not mistaken.
VexedCat68 hi!
What version of clearml , clearml-agent & server are you using?
Hi @<1558624448217616384:profile|ShaggyFrog16> , SSO & LDAP integrations are part of the Scale & Enterprise licenses 🙂
You're running your experiment from pycharm? Are you using the same environment in pycharm for all your experiments and you want the task to take packages from your 'agent' environment?
What errors are you getting?
Hi @<1644147961996775424:profile|HurtStarfish47> , Do you have some basic code snippet that reproduces this behavior?
GiganticTurtle0 , it looks like an issue with the latest RC. We're working on it to fix it 🙂
StickyCoyote36 , if I understand correctly due to your M1 chip limitation you run the script from a different machine and then you use the agent to run on the M1 chip machine and you want the requirements.txt in the repo to override the "installed packages" when running with agent, correct?