In the 'sample' directory, we include some 'gif' examples for the Moving MNIST dataset.

For example:
    '1-input.gif' is the input (10 frames)
    '1-truth.gif' is the ground truth (10 frames)
    '1-pred-FCLSTM.gif' is the prediction of FC-LSTM
    '1-pred-ConvLSTM-1layer.gif' is the prediction of the 1-layer ConvLSTM
    '1-pred-ConvLSTM-2layer.gif' is the prediction of the 2-layer ConvLSTM
    '1-pred-ConvLSTM-3layer.gif' is the prediction of the 3-layer ConvLSTM
    '1-pred-ConvLSTM-2layer-noconv.gif' is the prediction of the 2-layer ConvLSTM without state-to-state convolution
    '1-pred-ConvLSTM-3layer-noconv.gif' is the prediction of the 3-layer ConvLSTM without state-to-state convolution

In the 'sample-out-of-domain' directory, we include some 'gif' examples for an 'out-of-domain' run of the Moving MNIST dataset with 3 digits.

For example:
    '1-input.gif' is the input (10 frames)
    '1-truth.gif' is the ground truth (10 frames)
    '1-pred-ConvLSTM-3layer.gif' is the prediction of the 3-layer ConvLSTM
