Hidden State of RNN (not averaged ?)

Hello guys ,
Thank you very much for this work 👍 ! I have a quick question about the code :
Why in the cell method you output `output, h[self.args.num_blocks - 1]`. It looks like the h is only one loose end , whereas in the paper it looks like h = output = average over the loose ends.

Start by reading the cell method, especially the `output, h[self.args.num_blocks - 1]` return expression, and compare it with the paper's description of averaging over loose ends. Determine whether the implementation and paper differ; done means the behavior is explained or the issue is clarified with a reproducible conclusion.

Written by the indexing model from the issue text.