Auto3dSeg hyperparameters missing explanation

[che85]

Hello,

I am looking for detailed information about the hyperparameters used for Auto3dSeg training.

To give examples, what are the auto_scale_batch, auto_scale_filters, quick used for and what could be the effect of changing these parameters. Generally, it would be great if there was some more detailed documentation for all hyperparameters.

Also, what's the difference between batch_size and num_images_per_batch? I would use these interchangeably. Or would those be regarding mini batches?

Or did I miss it somewhere?

Thank you in advance!