// packing at all, but makes SRU usage compatible with nn.RNN usage
orig_input = input
if isinstance(orig_input, PackedSequence):
input, batch_sizes, sorted_indices, unsorted_indices = input
length = input.size(0)
batch_size = input.size(1)
mask_pad = torch.arange(batch_size,
device=batch_sizes.device).expand(length, batch_size)
mask_pad = (mask_pad >= batch_sizes.view(length, 1)).contiguous()
else:
length = input.size(0)
batch_size = input.size(1)
batch_sizes = None
sorted_indices = None
unsorted_indices = None
After Change
input, lengths = nn.utils.rnn.pad_packed_sequence(input)
max_length = lengths.max().item()
mask_pad = torch.ByteTensor([[0] * l + [1] * (max_length - l) for l in lengths.tolist()])
mask_pad = mask_pad.to(input.device).transpose(0, 1).contiguous()
// The dimensions of `input` should be: `(sequence_length, batch_size, input_size)`.
if input.dim() != 3:
raise ValueError("There must be 3 dimensions for (length, batch_size, input_size)")
