I commented the upload_artifact at the end of the code and it finishes correctly now

sure 🙂

just this one...it marks as completed when executed locally

is there a way to prevent from creating a new setup in my worker each time?

clearml == 0.17.5rc5
google_cloud_storage == 1.36.1
joblib == 1.0.1
matplotlib == 3.3.4
numpy == 1.20.0
object_detection == 0.1
opencv_python_headless == 4.5.1.48
pandas == 1.2.3
scikit_learn == 0.24.1
tensorflow == 2.4.0

yes, that works..but wasn't the issue with logging tensors?

not sure, I'm using GCS not S3. Is download_folder doing something different than downloading all files inside the folder?

and in the script I removed the output_uri= , in the task initialization

whereas this is what is being logged in your toy example: tf.Tensor(1742.0144, shape=(), dtype=float32)

if I launch the same script in GCP, (I don't run it as a clearml-agent), then everything works fine

it's my error: I have tensorflow==2.2 in my venv, and added Task.add_requirements('tensorflow') which forces tensorflow==2.4:

Storing stdout and stderr log into [/tmp/.clearml_agent_out.kmqde7st.txt]
Traceback (most recent call last):
 File "aicalibration/generate_tfrecord_pipeline.py", line 15, in <module>
  task = Task.init(project_name='AI Calibration', task_name='Pipeline step 1 dataset artifact')
 File "/home/username/.clearml/venvs-builds/3.7/lib/python3.7/site-packages/clearm...

since it didn't change (the packages installed)

where can I find more info about why it failed?

I see, thanks!

How would you like me to share it?
So far I have this:
tensorflow_object_detection_autoinstall.sh
Before running:
You need to set your venv
install numpy
export TF_DIR=$HOME/tensorflow mkdir $TF_DIR cd $TF_DIR echopwdwget unzip protoc-3.14.0-linux-x86_64.zip -d protoc export PATH=$PATH:pwd`/protoc/bin

git clone
cd models
git checkout 8a06433
cd $TF_DIR/models/research
protoc object_detection/protos/*.proto --python_out=.

git clone
cd cocoapi/PythonAPI
make
cp -r py...

and can the agent when running locally with no base docker, inherit as well system wide packages?

ah great! I'll try that then

Hi! I'm tryin to find a workaround for this: can't do pip install <name_of_package>
I executed the task, and it created a cache venv.
But when running the code, it couldn't import the package because it wasn't listed.
I then sourced the venv, and manually installed the package
so if I do python -c 'import object_detection' if works

so does the container install anything, or just runs the script? how is the setup done there?

but then I run again the task, it uses the same cache venv, but fails to find object_detection

so I need to run a sed command to replace some lines in one of the tensorflow files..do you know if I can do this as part of the setup.py install?

no, only in the clearml.conf file
Now I removed the output_uri in the conf file of the machine that started the task, and when I run it as agent in GCP it works.
Is this a bug?

if I put ~/clearml in the default_output_uri key, and start the task, when run as agent in GCP I get clearml.Task - INFO - Completed model upload to file:///$github_proj_directory/~/clearml/$proj_name/$experiment_name

so, the thing is that to install the object_detection package you need to manually run some commands, copy a setup.py file, etc. There is no git repo that does that for you...is that more clear?

yes

but it still saves using the output_uri from the server that created the task

I'll need to see how to extract only the part that we care about

the issue is with StorageManager.download_folder

FYI, in case it is useful for someone else:
` import tensorflow as tf
import numpy as np
from plotly.subplots import make_subplots
import plotly.graph_objects as go
import itertools
from clearml import Task

def get_trace(z, series, classes, colorscale='blues', showscale=True, verbose=False):
if verbose:
print(z)
ind = '1' if series=='train' else '2'
trace = dict(
type="heatmap",
z=z[::-1],
x=classes,
y=classes[::-1],
...

no prob! 🙂