I am trying to tweak this ConvLSTM model for another image sequence problem. I keep running into this error
Tensor-typed variable initializers must either be