I need help on RNN input shape for both Keras SimpleRNN and gwtaylor's rnn.py. According to the two links below, both input shapes have to be (nb_samples, timesteps, input_dim).
Keras SimpleRNN: http://keras.io/layers/recurrent/ gwtaylor's rnn: https://github.com/gwtaylor/theano-rnn/blob/master/rnn.py My question is if I have data, say 100 samples with 10 features and window size 5, can I just simply make the input shape (100, 5, 10)? Or do I need any data preprocessing as stated this (http://stackoverflow.com/questions/36136562/python-keras-simplernn-wrong-number-of-dimensions-on-model-fit)? Any help is much appreciated. Thanks, Hiroki -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
