is it possible in eg Keras to do bp through time but truncated to some fixed time horizon? @fchollet
-
drat yeah I was worried about that :(
you can do truncated backprop through time in Keras by using a stateful RNN (see docs)
-
-
I last used Keras before your TensorFlow integration; I look forward to playing with it again!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I'm new to LSTMs so this question might sound stupid. In trying to perform truncated backprop through time in Keras by using a stateful RNN, do we just have to set stateful=True and reshape appropiately or is there anything else we need to consider?
-
I am not sure since I don't use Keras much but this article seems to give a nice overview and discussion of the issues at least http://philipperemy.github.io/keras-stateful-lstm/ … , hope it can help you.
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.