Hendrik Kleikamp

The POD method in pyMOR actually returns the high-fidelity states and not the reduced coefficients, i.e. the left singular vectors of the snapshot matrix and not the right ones. So...

A first version using rich is implemented [here](https://github.com/pymor/pymor/tree/progress-bar-neural-network-training) for the neural network training. You can check it out by running one of the neural network demos.

A small question from some with only a vague idea about what is happening in ERA: In the case of a single input and zero initial state, isn't it possible...

> edit: I just realized I forgot the docstring for `feedthrough` sweat_smile I hope its clear now, thanks! Ah, great, now it's clear!

If @sdrave has nothing to add, let's wait for the CI and merge this then.

> What's going on with the docs build for 3.9 ? This seems to be the problem: ``` WARNING: Pygments lexer name 'ipython3' is not known ``` Occurred elsewhere lately...

Well, thank you very much @sdrave for your review, the last changes were a complete game changer. I was able to completely get rid of the dedicated models for the...

> Well, thank you very much @sdrave for your review, the last changes were a complete game changer. I was able to completely get rid of the dedicated models for...

Now this looks much better, at least from my point of view. Maybe you can take another look @sdrave? @TiKeil, do you have anything to add here?

Currently, torch version 1.8.1 gets installed during the CI pipelines. However, the LSTM implementation changed from 1.10 to 1.11 and is much more flexible now. Is there any chance to...