
Reputation
Badges 1
5 × Eureka!ZanyPig66 maybe this example can help?
https://github.com/allegroai/clearml/blob/master/examples/frameworks/pytorch/tensorboard_toy_pytorch.py
You could also try to upload an image or directory:
https://clear.ml/docs/latest/docs/guides/reporting/artifacts/#image-files
In this particular case using a naming convention is probably the best answer.
If you already have a naming convention for projects, you could just reuse that to prevent confusion :)
Wait... Will it limit memory usage but not show it in overview or just plain stop the container if it happens to use more memory?
Can you try again with "-m 16g" ?
I'm not exactly sure but it seems this is an Airflow error when a library isn't working.
Can you tryos.environ["no_proxy"]="*"
I've found this both here: https://github.com/apache/airflow/discussions/24463#discussioncomment-4211269
and here: https://stackoverflow.com/a/73983599
Do you get any error when uploading?
It looks like it can upload but can't download afterwards.
You can use https://clear.ml/docs/latest/docs/references/sdk/task/#taskget_project_id to get the id of the last updated project with that name
Could it be multiple metrics that were combined into a single metric later on? Before the optimizer?
Can you try: '${pipeline.docker_image}'
?
You can also use https://clear.ml/docs/latest/docs/references/sdk/task/#taskget_task since task.clone also accepts a task object
To use a specific binary you can set in in the config: https://clear.ml/docs/latest/docs/configs/clearml_conf/#:~:text=python%20version%20(default)-,agent.python_binary,-(string)
But if you're trying to cache virtual environments you might be more interested in: https://clear.ml/docs/latest/docs/clearml_agent#environment-caching
Could you test the following:
Without reusing the virtual environment you made manually:
Can you run a task twice and see if the second run is at least reusing the virtual environment of the first run?
So could you just setup your virtual environment with a task?
Just checking, are you just trying to use a different docker image in a task? Because then you might want to use this: https://clear.ml/docs/latest/docs/apps/clearml_task/#docker
https://clear.ml/docs/latest/docs/clearml_agent#docker-mode
You can set where to store it via this config file: https://clear.ml/docs/latest/docs/fundamentals/artifacts/#setting-upload-destination
Do you mean what's visible in the UI, projects -> Execution: Installed Packages?
We're sorry about that, this seem like a bug indeed. Could you open a github issue?
Or you can just load a config file or object: https://clear.ml/docs/latest/docs/references/sdk/task/#connect_configuration
Could you elaborate on S3 checkpoint name?
I'm assuming it's a filename?
Possibly post those few lines of code?
We checked in the UI and if the model description is edited with double spaces, they remain, so the problem is likely somewhere in the SDK.
Can you elaborate on question #2?
Do you want to reuse a task or something else?
Both server and agent can be configured with different ports. Which is it you`re looking for?
I'm not sure if you can delete it when using pipelines but I would say try it on a new project?
It looks like there is this option under Settings->Configuration->Show Hidden Projects
Can you elaborate on the hidden project? Is this part of a task you created and something is not showing up?