Reputation
Badges 1
25 × Eureka!And how did you connect your example,yaml?
in the UI the installed packages will be determined through the code via the imports as usual ...
This is only in a case where a user manually executed their code (i.e. without trains-agent), then in the UI after they clone the experiment, they can click on the "Clear" button (hover over the "installed packages" to see it) and remove all the automatically detected packages. This will results in the trains-agent
using the "requirements.txt".
I've seen that the file location of a task is saved
What do you mean by that? is it the execution section "entry point" ?
Try to upload something to the file server ?
None
Okay, let's take a step back and I'll explain how things work.
When running the code (initially) and calling Task.init
A new experiment is created on the server, it automatically stores the git repo link, commit ID, and the local uncommitted changes . these are all stored on the experiment in the server.
Now assume the trains-agent is running on a different machine (which is always the case even if it is actually on the same machine).
The trains-agent will create a new virtual-environmen...
Can you do it manually, i.e. checkout the same commit id, then take the uncommitted changes (you can copy paste it to diff.txt) then call git apply diff.txt ?
If you want to rename it (any pipeline), click on the "Full details" in the "Run Info" (right hand side panel), then in the full detail of the Pipeline Task you will be able to rename the pipeline execution
(Is renaming useful? should we add a right click to rename ?)
Hi @<1533620191232004096:profile|NuttyLobster9>
Hi All, is there a way to clone a pipeline from the web UI like you can with a task?
Right click on the pipeline and select Run (it is basically the same thing as cloning it)
Iβve did saw this βpublishβ option for pipelines, just for models, is this a new feature?
Kind of hidden in the UI (not sure if on purpose), but if you click on the pipeline then go to details, in the new tab (of the pipeline Task) you can publish the Task (aka the pipeline)
In this example:
https://github.com/allegroai/clearml-actions-train-model/blob/7f47f16b438a4b05b91537f88e8813182f39f1fe/train_model.py#L14
replace with something like:
` task = Task.get_tasks(project_name="pipel...
BTW: see if this works:$ CLEARML_API_HOST_VERIFY_CERT=0 clearml-init
Are you referring to Poetry ?
If you set the package_manager to peotry then it will only use the lock files
https://github.com/allegroai/clearml-agent/blob/21c4857795e6392a848b296ceb5480aca5f98e4b/docs/clearml.conf#L53
If you clear the "Installed Packages" section, it will just use the "requirements.txt" in the repository itself.
What's the specific use case, and the problem we are trying to solve?
Hi @<1523702652678967296:profile|DeliciousKoala34>
What's the clearml-server version you are working with?
Can you check with the latest RC?
pip3 install clearml==1.9.2rc2
these are being repeated as well for a single task (this is training a t5_model with transformers):Β (edited)
Seems like someone is storing lots of files with torch.save
that ClearML automatically logs.
You can disable the autolog:task = Task.init(..., auto_connect_frameworks={'pytorch': False})
. I'm thinking it's generically a kernel gateway issue, but I'm not sure if other platforms are using that yet
The odd thing is that you can access the notebook, but it returns zero kernels ..
DisgustedDove53 , TrickySheep9
I'm all for it!
I can think of two options here, (1) use the k8s glue + apply template with ports mode see discussion https://clearml.slack.com/archives/CTK20V944/p1628091020175100
(2) create an interface (queue) to launch arbitrary job on the k8s cluster, with the full pod definition on the Task. This will allow the clearml-session to setup everything from the get go.
How would you interface with the k8s operator, and what exactly will it do?
(BTW: the reas...
is everything on the same network?
@PipelineDecorator.component(repo="..")
The imports are not recognized - they are not on the pythonpath of the task that the agent starts.
RoughTiger69 add the imports inside the functions itself, you cal also specify the, on the component@PipelineDecorator.component(..., package=["pcakge", "package==1.2.3"])
or@PipelineDecorator.component(...): import pandas as pd # noqa ...
And when retrieve just this file? is it working ?
(Maybe for some reason the file is corrupted) ?
Hi SkinnyPanda43
Do you mean the cleaml-agent or the cleaml python (a.k.a the auto package detection) ?
Hi @<1654294828365647872:profile|GorgeousShrimp11>
can you run a pipeline on a
schedule
or are schedules only for Tasks?
I think one tiny details got lost here, Pipelines (the logic driving them) are a type of Task, this means you can clone and enqueue them like other tasta
(Task.enqueue / Task.clone)
Other than that looks good to me, did I miss anything ?
When using the UI with regex to search for experiments, due to the greedy nature of the search, it consistently pops up the "ERROR Fetch Experiments failed" window when starting to use groups in regex (that is, parentheses of any kind).
hmm that is a good point (i.e. only on enter it would actually search)
Could it be updated so that if an invalid regex pattern is given, it simply highlights the search bar in red (or similar) rather than stop us while writing the search pattern?
...
Hi @<1628927672681762816:profile|GreasyKitten62>
Notice that in the github actions example this psecific Task is executed on the GitHub backend, the Task it creates is executed on the clearml-agent.
So basically:
Action -> Git worker -> task_stats_to_comment.py -> Task Pushed to Queue -> Clearml-Agent -> Task execution is here
Does that make sense ?
I changed them to the one exposed to the users (the same I have in my local clearml.conf) and things work.
Nice!
But I can't really figure out why that would be the case...
So the thing is, the link to the files are generated by the clients, which means the actual code generated a link an internal link to the file server (i.e. a link that only works inside the k8s cluster). When you wanted to see the image/plot you were accessing it from outside the cluster, and the link simply ...
Hi SarcasticSquirrel56
But if I then clone the task, and execute it by sending it to a queue, the experiment succeeds,
I'm assuming that on the remote machine the "files_server" is not configured the same way as the local execution. for example it points to an S3 bucket the credentials for the bucket are missing.
(in your specific example I'm assuming that the plot is non-interactive which means this is actually a PNG stored somewhere, usually the file-server configuration). Does tha...
Hmm, I see the jump from 50 to 100, is that consistent with the last iteration on the aborted Task (before continuing )?