Hello 🙂
I am having issues persisting output images to a self-hosted clearml server. When I use the matplotlib automagical hooks (with plt.show() ) I get empty images. I get the same empty images if I try to disable the hooks and use the manual logger.report_matplotlib_figure methods. I can persist other plot types though; logger.report_confusion_matrix saves with no issues.

Here is a script that reproduces my issue:

import logging
import joblib
import pandas as pd
import matplotlib.pyplot as plt
from clearml import Task, TaskTypes
from sklearn import tree
from sklearn import datasets
from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix
from sklearn.model_selection import train_test_split


def preprocess(df: pd.DataFrame) -> tuple[any]:
    X = df.loc[
        :,
        [
            "sepal length (cm)",
            "sepal width (cm)",
            "petal length (cm)",
            "petal width (cm)",
        ],
    ].to_numpy()
    y = df.target
    return (X, y)


def main():
    task = Task.init(
        project_name="Demo-Data",
        task_name="iris_training",
        task_type=TaskTypes.training,
        tags=["demo", "iris"],
        reuse_last_task_id=False,
        output_uri=True,
        auto_connect_frameworks={"matplotlib": False},
    )
    logger = task.get_logger()
    iris_set = datasets.load_iris()
    iris_df = pd.DataFrame(data=iris_set.data, columns=iris_set.feature_names).assign(
        target=iris_set.target
    )
    train, test = train_test_split(iris_df, test_size=0.2)
    model = tree.DecisionTreeClassifier().fit(*preprocess(train))
    joblib.dump(model, "model.pkl")  # persist model artifact using hooks
    test_features, test_labels = preprocess(test)
    predicted_labels = model.predict(test_features)

    ####persisting outputs####
    fig, ax = plt.subplots()
    cm = confusion_matrix(test_labels, predicted_labels)
    ConfusionMatrixDisplay(cm).plot(ax=ax)
    logger.report_text(
        "trying to save an image", level=logging.DEBUG, print_console=False
    )
    logger.report_matplotlib_figure(
        title="confusion_matrix",
        series="mpl",
        iteration=0,
        figure=fig,
        report_interactive=False,
    )
    logger.report_text(
        "trying to save a confusion matrix", level=logging.DEBUG, print_console=False
    )
    logger.report_confusion_matrix(
        title="confusion_matrix", series="direct_logging", iteration=0, matrix=cm
    )


if __name__ == "__main__":
    main()