How were you saving the model with pytorch?

Hi @<1751777178984386560:profile|ConfusedGoat3> , I think you might need to run some migration script on the database, basically changing the paths of the artifacts registered to the new IP

I think you'd have to write it yourself. Basically, the artifact paths in experiments are saved in mongo. You would need to write a script that would modify those values in Mongo directly

Hmmmm you can automate the cleanup. Iterate through folders, if such an experiment exists, skip, if no experiment exists, delete folder

Hi @<1523704157695905792:profile|VivaciousBadger56> , you can read about the differences here:
None
None

ClearML should be backwards compatible - any combination will work. However it's always suggested to use the latest versions 🙂

Python 2 is no longer supported, I'd suggest finding an AMI that already has python3 built in (Or install it using the init script, not suggested though) and also CUDA enabled to avoid that installation to support cuda images

@<1577468638728818688:profile|DelightfulArcticwolf22> , in addition to @<1523701087100473344:profile|SuccessfulKoala55> ’s answer, you have data management, orchestration (Integration with SLURM), pipelines, reports and much more.

As I mentioned, provisioning resources according to different groups - i.e. role based access controls are an enterprise feature.

I suggest you watch the onboarding videos on the ClearML Youtube channel - None

Hi @<1787653555927126016:profile|SoggyDuck67> , can you please provide the full log of the run? Also, can you please add a screenshot of the 'execution' tab of the experiment? I assume the original experiment was ran on python 3.10?

Can you please open a github issue so the issue can be followed?

How do you currently save artifacts now?

Then you'd need to change the image in the docker compose or to spin up the webserver individually and make the necessary changes in the docker compose. Either way, you need a backend to work with the web ui

Did anything change in your configurations? In the previous version there was no such issue? Is the agent version the only change?

Did you wait 10-15~ mins for it to time out?

VexedCat68 , what errors are you getting? What exactly is not working, the webserver or apiserver? Are you trying to access the server from the machine you set it up on or remotely?

VexedCat68 , It appears to be a bug of sorts, we'll sort it out 🙂

Hi AbruptWorm50 ,

You can use a stand alone file, this way the file will be saved to the backend and used every time without needing to clone the repo. What do you think?

Try to set agent.enable_git_ask_pass: true for the agent running inside the container, perhaps that will help

ProudElephant77 , I think you might need to finalize the dataset for it to appear

I think you can force the script diff to be empty with Task.set_script(diff="") or maybe Task.set_script(diff=None)
https://clear.ml/docs/latest/docs/references/sdk/task#set_script

Where did you get this autoscaler? I don't think a GCP autoscaler was released in the open source

Hi @<1547028074090991616:profile|ShaggySwan64> , can you please provide minimal sample code that reproduces this? The local imports - are they from the private repo?

Hi @<1533257278776414208:profile|SuperiorCockroach75> , what do you mean? ClearML logs automatically scikit learn

Let me play with it a bit and see if I can find more 🙂

But the issue still reproduces?

Can you please provide a snippet of how the debug images are saved, also an example url would be useful :)

Oh LOL 😛

All artifact links are saved in mongodb, all debug samples are saved in ElasticSearch. I think you would need to read up on how to change values inside those dbs. I would assume server would need to be down when such a script would be running

Can you add here the configuration of the autoscaler?

VexedCat68 , what do you mean by trigger? You want some indication that a dataset whats published so you can move to the next step in your pipeline?