Keras-retinanet: What does the steps argument do in training?

[sv-youn]

I understand how batch-size and epochs work. But I have no idea about the steps argument.
The following is from bin/train.py.
parser.add_argument('--steps', help='Number of steps per epoch.', type=int, default=10000)
Can anyone explain about this argument?

[Mastemace]

You should google it;) See the explanation, e.g., on https://datascience.stackexchange.com/questions/47405/what-to-set-in-steps-per-epoch-in-keras-fit-generator

[sv-youn]

Thanks, Mastemace. That's very helpful. I will close the issue.