Reputation
Badges 1
39 × Eureka!The installed packages of the task say this:
# Python 3.11.2 (main, Mar 13 2023, 12:18:29) [GCC 12.2.0]
PyYAML == 6.0.1
clearml == 1.15.1
google google_api_core
google_cloud_storage == 2.16.0
ultralytics == 8.2.2
I do not know where the google_api_core comes from and I'd like to remove it.
Figured it out, I installed clearml[gs] but since I don't need that I removed it. it's gone now.
This is the full log of the task.
I am trying to run HPO.
Sure can do
It happens on all of my pipeline run attempts and there's nothing more that gives insight.
As an example:
python src/train.py
ClearML Task: created new task id=102a4f25c5ac4972abd41f1d0b6b9708
ClearML results page:
<unknown>:1: SyntaxWarning:
invalid decimal literal
<unknown>:1: SyntaxWarning:
invalid decimal literal
<unknown>:1: SyntaxWarning:
invalid decimal literal
<unknown>:1: SyntaxWarning:
invalid decimal literal
<unknown>:1: SyntaxWarning:
invalid decimal...
Here are the codefiles for my pipelines.
They do not work yet, I am struggling with the pipeline stuff quite a bit.
But both pipelines always give these warnings.
I am getting the same when starting regular tasks.
I think it has something to do with my paramaters, which contain an environment variable which contains a list of datasets
It comes from the PipelineDecorator.pipeline I assume or from PipelineDecorator.component
I noticed that it's actually independent of the pipelines
This function shows the same behaviour once the task gets initialized:
# Training helper functions
def prepare_training(env: dict, model_variant: str, dataset_id: str, args: dict, project: str = "LVGL UI Detector"):
from clearml import Task, Dataset
import os
print(f"Training {model_variant} on dataset: {dataset_id}")
# Fetch dataset YAML
env['FILES'][dataset_id] = Dataset.get(dataset_id).list_files("*.yaml")
# Download & modify dataset
env['DIRS']['target'] ...
Yup.
I really don't know what it's about.
It doesn't affect the process. Everything seems to run fine.
If the warnings would provide a bit more info I could maybe pinpoint it better, but it's really all I got
Is there some verbose mode I could run it with?
Not that I would know of..
I attached the possible problematic argument.
The strings are just regular string, nothing fancy there.
args
:{'epochs': 3, 'imgsz': 480, 'data': '/home/rini-debian/git-stash/lvgl-ui-detector/datasets/ui_randoms.yaml'}
model_variant
:yolov8n
dataset_id
:50e10f640d7548458d9c38ab9aac571b
I have the strong suspicion it is somewhat related to my parameters of the function or generally the hyperparameters gathered by the task automatically.
Maybe it has something to do with my general environment? I am running on WSL2 in debian
Yea, I get that.. But it's really hard to tell what's causing it due to the "<unknown>"
There has been a restart of my machine in the mean time :man-shrugging:
So only the matrix knows now I guess..
When developing I use the poetry environment, but in the queues I let clearML handle the installation via pip.
There is no need to use poetry if the task is a one-time thing
Hi @<1523701087100473344:profile|SuccessfulKoala55>
I am using 1.8.0 for the clearml-agent.
Attached is the logfile.
I cleared the vcs cache manually already, it results in the same behaviour illustrated above
(allthough the logs show that it used the cache, I had another run without cache - but don't have the logs from that)
Hey. I should have closed this..
The thing that I was looking for is called set_parameter
on the task.
The HPO uses a task I created previously and I had trouble with that, since it contained a path, which wasn't available on the colab instance.
I fixed my code, so it always updates this parameter depending on the environment.
It was less of an HPO issue, more of a programming failure on the function, which didn't properly update the parameter, even though I thought it should.
Back when I wrote this, I thought HPO does something magical for overwriting the general args of the task when cloning.
Turns out it just was my code that was missing a more explicit set_parameter
for this environment path.
No idea what's going on now, but I cannot reproduce the behaviour either.. also tried my old code posted here, but the warning doesn't pop up anymore.
I will inform once it pops again and will use the provided traceback function then.
I have a slight suspicion that it was indeed environment based on my local machine, but I have no idea what is the trigger for that.
It may or may not be related to this
2024-04-29 23:38:25,932 - clearml.Task - WARNING - Parameters must be of builtin ty...
Alright cool!
I will check it out and let you know what it was.
I noticed poetry can be a problem in my run.
Not specifically due to the cache, but due to the installation of much more packages than the runtime might need.
When using regular pip, it will use the requirements list determined by ClearML to install necessary packages, which usually already excludes all dev-tools and similar.
I am not sure if poetry uses the cache properly, but I can't verify either atm.
On another attempt with a cleaned repository (no dirty commits) I get the same result, even though it states that it got a new commit id, so I'm at a loss at what is actually going wrong here:
`Using cached repository in "/root/.clearml/vcs-cache/lvgl-ui-detector.git.7c8ae2688810ceed26c1ebcc1e911cf2/lvgl-ui-detector.git"
remote: Enumerating objects: 11, done.
remote: Counting objects: 100% (11/11), done.
remote: Compressing objects: 100% (5/5), done.
remote: Total 8 (delta 4), reused 7 ...
If it were possible to override the checkout behaviour I would ignore all submodules anyways, but in the documentation of clearml.conf as well as the pipeline decorator I couldn't find anything that would allow me doing that.
According to None I am supposed to install
libgl1
I changed my clearml.conf
to include that installation for the task container started by the agent.
Will see if it helps in a minute