Original LSTM Training Details

[Miraç Göksu Öztürk]

Hello, 

I went through the docs, papers given in the documentation, and this group's conversations but couldn't exactly find the answer. I'm sorry if it's already discussed or explained somewhere.

It's basically questions regarding the original training process:

- Was the original data just generated as given in the documentation? Or was there any additional data which was not shared? 

- Do we know the original training process? Is it exactly the same as described in the documentations? Any info on data augmentations, hyperparameters. How long were the shared models trained for?

Any answer or pointer is appreciated.
Thanks.