Justus Schock

Searching for the latest checkpoint is already handled by the trainer

This is a minor issue, but can we maybe put the sequence number at the end? This seems more intuitive to me

Okay, that's a valid reason too, but this way you get the first run from different days sorted together instead of a "real" development over time. What doe the others...

That's something I wouldn't like. If we create an experiment, the date of the creation should be used for all runs, but if we re-initiate an experiment, I would also...

Ah okay, I would do it like `date_today_1`, `date_today_2` etc. But then I would restart with `date_then1`. Another problem is, that we need some management of adding zeroes. E.g. if...

The failure of the test is caused, because in python 3.5 the ordering of a dict is not preserved. Due to this fact, we have to rewrite this test in...

Yeah something like that. But we don't need to take care of the kfold in this case, because inside the save_path the experiment will also create a separate folder for...

I think we should use randomly created fakedata here (something similar to [this dataset](https://github.com/delira-dev/vision_base/blob/master/deliravision/data/fakedata.py#L6); we might even copy it, but since deliravision is no dependency of delira, I would prefer...

I think so far, there is no-one working on this, so if you want to, we'd appreciate that :)

@iqDF Sorry for the late reply. Of course you can ping me whenever you need :) I recently get spammed with github notifications, so I mussed have missed this one...