The question about “sequence discriminator”.

[lzkzls]

Is the convolution neural network structure in "sequence discriminator" the same as that in "video snippet encoder" in Figure 5? If not, what structure is used? Thank you.