@<1523701070390366208:profile|CostlyOstrich36> any thoughts? Are the model files themselves easier to serve?

but is model files easier to serve?

Sorry for the late reply @<1722061389024989184:profile|ResponsiveKoala38> . So this is the diff between my local version (hosted together on a single server with docker-compose). Does anything spring to mind?

It's running v7.17.18 @<1722061389024989184:profile|ResponsiveKoala38>

Hi Martin,
It doesn't seem to work with dev.azure though:

Using user/pass credentials - replacing ssh url 'git@ssh.dev.azure.com:v3/ORG/TEAM/PROJECT' with https url '

'
fatal: repository '

' not found

The expected format for the https protocol is None .
Thoughts @<1523701205467926528:profile|AgitatedDove14> ?

Just wanted to share a workaround for using a TriggerScheduler to execute a script using the latest commit of a given branch, without relying on cloning a Task. Don't know if it has been shown before in here 🙂

from clearml import Model, Task
from clearml.automation import TriggerScheduler

def trigger_model_func(model_id: str):
    model = Model(model_id)

    print(f"Triggered model export for model '{model.name}' ({model_id})")

    # NOTE: To execute from the branch of
    # task...

Hi CostlyOstrich36
I have created a base task on which I'm optimizing hyperparameters. With clearml-param-search I could use --params-override to set a static parameter, which should not be optimized, e.g. changing the number of epochs for all experiments. It seems to me that this capability is not present in HyperParameterOptimizer . Does that make sense?

From the example on https://clear.ml/docs/latest/docs/apps/clearml_param_search/ :
` clearml-param-search {...} --p...

Hi CurvedHedgehog15 , thanks for replying!
I guess that one could modify the config with variable interpolation (similar to how it's done in YAML, e.g. ${encoder.layers} ) - however, it seems to be quite invasive to specify that in our trainer script 😞

Well, one solution could be to say that models can only be exported from main/master and then have devops start a new trigger on PR completion. That would require some logic for stopping the existing TriggerScheduler, but that shouldn't be too difficult.

However, the most flexible solution would be to have some way of triggering the execution of a script in the parent task environment, something along the lines of clearml-agent build ... . I just can't wrap my head around triggering that ...

Yeah, that makes sense. The only drawback is that you'll get a single point that all lines will go through in the Parallel Coordinates plot when the optimization finishes 🙂

Hi CostlyOstrich36
What I'm seeing is expected behavior:

In my toy example, I have a VAE which is defined by a YAML config file and parsed with PytorchLightning CLI. Part of the config defines the latent dimension (n_latents) and the number of input channels of the decoder (in_channels). These two values needs to be the same. When I just use the Lightning CLI, I can use variable interpolation with OmegaConf like this:
` class_path: mymodel.VAE
init_args:
{...}
bottleneck:
class_pat...

Hi @<1523701087100473344:profile|SuccessfulKoala55> , thanks for responding. I've found out that my first error came from cloning a super old version of the clean up task in the web UI 😄
I don't know about the other error, to me it looks like the task gets deleted before handling errors, but since an error occurred (some 404 stuff, maybe the files actually aren't there) when deleting some artifacts on the task, clearml tries to reload the task and fails, with the 400/201 or 400/101. ...

I've tried setting the output_uri on Task.init, but that seems to only affect model checkpoints and artifacts

@<1590514584836378624:profile|AmiableSeaturtle81> that’s the service we are using :-)

How much RAM have you assigned to your elastic service?

@<1576381444509405184:profile|ManiacalLizard2> what happens when ES hits the limit? Does it go OOM, or does the scalars loading just take a long time in the web-ui? And what about tasks putting scalars in the index?

Sure. Really, I'm just using the default client:
# ClearML SDK configuration file
api {
web_server: http://server.azure.com:8080
api_server: http://server.azure.com:8008
files_server: http://server.azure.com:8081
credentials {
"access_key" = "..."
"secret_key" = "..."
}

}
sdk {
# ClearML - default SDK configuration

storage {
    cache {
        # Defaults to system temp folder / cache
        default_base_dir: "~/.clearml/c...

Hey SweetBadger76 , thanks for answering. I'll check it out! Does that correspond to filling out azure.storage in the clearml.conf file?

And how do I ensure that the server can access the files from the blob storage?

How does it look in the Web UI?

I just had a look, and they are visible under debug samples, but not under plots, as I had expected.
I thought that by using report_matplotlib_figure it would get grouped under plots? 🙂

Do you mean to the Web UI?

Yes that's what I meant, sorry I'm still coming to terms with ClearML terminology 😅 . Is it possible to store the web app cloud access token serverside so we don't have to input it in the Web UI? 🙂

The lightning folks won't include new loggers anymore (since mid-2022, see None ) 🙂

Specifically, this is what I get in the console log when the agent spins up a task:

Poetry Enabled: Ignoring requested python packages, using repository poetry lock file!
Creating virtualenv latent-features in /data/clearml/venvs-builds/3.9/task_repository/our-repo/.venv
Installing dependencies from lock file

Hi again CostlyOstrich36 ,

I just wanted to share what ended up working for me. Basically I worked it out both for Hydra (thanks CurvedHedgehog15 ) and for PytorchLightningCLI.

So, for PL-CLI, I used this construct so we don't have to modify our training scripts based on our experiment tracker

` from pytorch_lightning.utilities.cli import LightningCLI
from clearml import Task

class MyCLI(LightningCLI):
def before_instantiate_classes(self) -> None:
# init the task
tas...

Well, consider the case where you start the trigger scheduler on commit A, then you do some work that defines a new model and commit as commit B, train some model and now you want to export/deploy the model by publishing it and tagging it with some tag that triggers the export, as in your example. The scheduler will then fail, because the model is not implemented at commit A.

Anyways, I think I've solved it, I'll post the workaround when I get around to it 🙂
You can create a task in the t...

Hi @<1523701070390366208:profile|CostlyOstrich36>
Is 87G a lot for an index? Enough that you would consider adding more RAM?

And also, how can I check that we are not storing scalars for deleted tasks? ClearML used to write a lot of errors in the cleanup script, although that seems to have been fixed in recent updates

Which version of the server are you running?

Any tips on how to check if we are storing data on deleted tasks? Maybe @<1722061389024989184:profile|ResponsiveKoala38> knows? Is there a field on each scalar that I can cross check with ClearML?

@<1722061389024989184:profile|ResponsiveKoala38> cool, thanks! I guess it will then be straightforward to script then.

What is your gut feeling regarding the size of the index? Is 87G a lot for an elastisearch index?

This is an example of the console output of a task aborted via the webUI:

Epoch 1/29 ━╸━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 699/16945 0:04:53 • 1:55:25 2.35it/s v_num: 0.000
2024-09-16 12:52:57,263 - clearml.Task - WARNING - ### TASK STOPPING - USER ABORTED - LAUNCHING CALLBACK (timeout 30.0 sec) ###
[2024-09-16 12:52:57,284][core.callbacks.model_checkpoint][INFO] - Marking task as `in_progress`
[2024-09-16 12:52:57,309][core.callbacks.model_checkpoint][INFO] - Saving last checkpoint...

I just tried and the result is the same. The other method only triggers on exceptions

This is on clearml v1.16.4