Week 1 :: Jazz with LSTM :: djmodel


In the implementation of djmodel, we are creating the LSTM layer one cell at a time, then manually linking each output unit.

Is this done for pedagogical purposes or is there are better way to implement this many-to-many network via higher level keras APIs?

From reading the documentation and researching online, I believe that it is possible to create the entire LSTM layer at once.


Do you still need a reply for this question?

Yes that would be great, thank you