I just test this module and it throws an exception when I call summary with input shape (3, 96, 96) by the following statement

summary(model, input_size = (3, 96, 96))

class TestNet(nn.Module):
    def __init__(self):
        super(TestNet, self).__init__()
        self.cdc_3t = CDC_T(3, 32, kernel_size=3, padding=1, theta=0.6)

    def forward(self, x):
        x = self.cdc_3t(x)
        
        return x

RuntimeError: Given groups=1, weight of size [32, 3, 3, 3, 3], expected input[1, 2, 96, 96, 3] to have 3 channels, but got 2 channels instead

The annotation files of IsoGD dataset (for instance "rgb_train_lst.txt") used in this paper not found!
The official annotation file of IsoGD dataset has three columns and looks like this:

valid/001/M_00001.avi valid/001/K_00001.avi 109
valid/001/M_00002.avi valid/001/K_00002.avi 241
...

However, in your dataset parser codes like Videodatasets, it seems that the columns are different from official annotation file.

def get_data_list_and_label(data_df, typ):
      T = 0  # if typ == 'M' else 1
      return [(lambda arr: ('/'.join(arr[T].split('/')[1:]), int(arr[1]), int(arr[2])))(i[:-1].split(' '))
              for i in open(data_df).readlines()]

Would you like to share your annotation files or provide a detailed explanation.
Looking forward to your reply! Thanks!