[theano-users] [theano.gradient.DisconnectedInputError]

[Priyank Pathak]

Hey,

So I'm trying to implement a VAE on keras, and the following is a snippet
of the code I've written.

z_mean = Dense( self.z_dim , init=initialization , activation='linear')(H2)
z_mean = LeakyReLU(alpha=.001)(z_mean)


z_log_var = Dense( self.z_dim,init=initialization , activation='linear')(H2)
z_log_var = LeakyReLU(alpha=.001)(z_log_var)


z = Lambda(self.sampling , output_shape=K.int_shape(z_mean) )([z_mean,
z_log_var])

*H3 = Dense(input_dim - 1, init=initialization , activation='linear')(z)*

Now when I later compile the model:

model = Model(input=data_input, output=[xh , z_mean , z_log_var ] )
grads = K.gradients(cost, trainable_vars)

it gives me the error:
*theano.gradient.DisconnectedInputError: *

*Backtrace when that variable is created:*

* H3 = Dense(input_dim - 1, init=initialization , activation='linear')(z)*


Does anyone have any idea why this error is there?
My guess is that the gradient fail to apply from the z to z_mean and
z_log_var

Regards
Priyank Pathak