Can any one help me here with the problem of not building the model with bidirectional function:
forward_lstm <- layer_lstm(units = 64, return_sequences = T , input_shape = c(look_back, 2))
backward_lstm <- layer_lstm(units = 64, activation = "relu",
return_sequences = T, go_backwards = T)
b_model <- keras_model_sequential() %>%
bidirectional (layer = forward_lstm, backward_layer = backward_lstm, merge_mode = "concat") %>%
layer_dense(units = 16, activation = "relu") %>%
layer_dense(units = 1)
I am gettin this error
summary(b_model)
Model: <no summary available, model was not built>
thanks