After I update the keras and seq2seq, I have a problem with the network running and get the error as below:
Input 0 of the graph (indices start from 0), used to compute Shape(<TensorType(float32, 3D)>), was not provided and not given a value
My network is a input followed by some dense layer, and followed by the attention model. After that, 3 output put dense layers are connected. The parameter table is shown as below:
theSage21
commented
7 years ago
Hi, dear all,
After I update the keras and seq2seq, I have a problem with the network running and get the error as below:
Input 0 of the graph (indices start from 0), used to compute Shape(<TensorType(float32, 3D)>), was not provided and not given a value
My network is a input followed by some dense layer, and followed by the attention model. After that, 3 output put dense layers are connected. The parameter table is shown as below:
Layer (type) Output Shape Param # Connected to
input_1 (InputLayer) (None, 4670, 13) 0
masking_1 (Masking) (None, 4670, 13) 0
time_distributed_1 (TimeDistribu (None, 4670, 128) 1792
activation_1 (Activation) (None, 4670, 128) 0
dropout_1 (Dropout) (None, 4670, 128) 0
model_10 (Model) multiple 280449
dropout_3 (Dropout) (None, 1, 128) 0
lstm_1 (LSTM) (None, 128) 131584
dropout_4 (Dropout) (None, 128) 0
dense_24 (Dense) (None, 8) 1032
dense_25 (Dense) (None, 2028) 261612
dense_26 (Dense) (None, 2) 258
activation_42 (Activation) (None, 8) 0
activation_43 (Activation) (None, 2028) 0
activation_44 (Activation) (None, 2) 0
Total params: 676,727.0 Trainable params: 676,727.0 Non-trainable params: 0.0
Does anyone know how to fix it?
Thank you.