search

Element-Research / rnn

Recurrent Neural Network library for Torch7's nn
BSD 3-Clause "New" or "Revised" License
939 stars 313 forks source link

Using GRU inside Recurrent #309

Open ethanabrooks opened 8 years ago

ethanabrooks commented 8 years ago

I am using an nn.GRU inside the transfer module of a Recurrent module (I'm doing this because a sequence-to-sequence model needs to redirect its output back into itself during testing -- that's what I'm using Recurrent for). My questions are:

  1. Will the GRUs feed their previous output back into themselves, in addition to the feedback that occurs as part of Recurrent?
  2. Right now the GRUs are causing a memory leak somehow (my memory usage increases until I get an out of memory error). How do I prevent this?

I guess I should also ask: is there a better way of going about this whole business? Thanks.