from sklearn.datasets import load_iris
import tensorflow as tf
import numpy as np
from clearml import Task, Logger
import argparse

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--epochs', metavar='N', default=64, type=int)
    args = parser.parse_args()
    parsed_args = vars(args)
    task = Task.init(project_name="My Workshop Examples", task_name="scikit-learn joblib example")
    iris = load_iris()
    data = iris.data
    target = iris.target
    labels = np.unique(target)
    epochs = parsed_args['epochs']
    task.connect(args)
    model = tf.keras.models.Sequential([
        tf.keras.layers.Dense(64, input_dim=4, activation='relu'),
        tf.keras.layers.Dense(len(labels), activation='softmax')
    ])
    model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
    model.fit(data, target, epochs=epochs)
    print(model.evaluate(data, target))

if name == "main":
    main()

Here's the script I'm testing with.
Here's the script I run it with.
clearml-task --project ClearML-Learn --name EpochsConnectReturns --requirements requirements.txt --script demo.py