Sherjil Ozair

`pip3 install --user .` is the right command.

Lots of examples here: https://github.com/fchollet/deep-learning-models/releases/

How about Mouse.Position is basically a getter, which returns the mouse position for the window for which MouseOver is true. And similarly for MouseRender. Maybe another variable could also expose...

Possibly the only library that comes close to handling it right now is: https://github.com/nicholas-leonard/dp/. Looking forward to your results.

The benchmarks are really cool! Thanks! How do you interpolate with the KN-5 model, though? And from where is the KN-5 model obtained?

Thanks for your comment. I'll try to switch in a GPU implementation of the FFT functions and benchmark it.

BPTT is much more popularly used than RTRL, so might be better to prioritize BPTT over RTRL.

Pretty much all other RNN implementations are slow, and not suitable for production. My own reason for betting on arrayfire is that this might yield production-ready implementations for deep learning...

@mohammadpz, this is a known bug in my wrapper. If you take a look at https://github.com/sherjilozair/ctc/blob/master/python/ctc.py#L91, you will notice that I completely ignore the output_grads, because of which, the cost...

The quickest way would be to compute the grads, and then divide it by the batch size. That should work.