Implementing a masking layer with ConvLSTM2D, Dense, Dropout layers in keras

Question

I am trying to implement a Masking layer in my model but it is not working and do not understand why...

I want to predict a continuous time series (predictand variabl