Simon Kern

I'm impressed you were able to get the training going! **different weights size:** This can have various reasons: The `keras` and `tensorflow` versions you are using are muchnewer than the...

Code to train the model is stored at https://github.com/skjerns/AutoSleepScorerDev Emphasis on _stored_ as I had no time yet to clean up the repository to make is usable, but it's on...

@shqing I created a small script `run_sample.py` that should help you to get started: https://github.com/skjerns/AutoSleepScorerDev/blob/master/run_sample.py just clone the https://github.com/skjerns/AutoSleepScorerDev/ repo. Hope it works. edit: just checked it: I think I...

Yes, this was discussed in https://github.com/scipy/scipy/issues/11351 Main points: * would induce optional dependency on h5py * would have to be backwards-compatible, which would need a lot of work * currently...

nice to see that there is progress w.t.r! However, I think it's still quite cumbersome having to program all the backwards compatibility and kwarg combinations that the mess that is...

That is quite unfortunate :-/ I don't think there is a lot I can do without a test case. What you can try: 1. remove all data/variables that are not...

I don't quite understand what exactly you mean by that? To indicate whether squeezing should happen or not? Feel free to open a PR, including some test cases if possible,...

I was a bit optistic when proposing that it would be easy to implement. Which data types do you generally need? The problem is a bit, that datatypes don't map...

Yes, I think for this purpose it is much faster to hard code something and use hdf5 to prevent reading the entire file. However, that might also be a bit...

This would be perfect for easy blocking of certain results "rightclick->don't show this app"