Error when batch size = 1

line 135 in skip-thoughts.torch/pytorch/skipthoughts/skipthoughts.py ```lengths = list(max_length - input.data.eq(0).sum(1).squeeze())``` will raise an error if the batch size is 1, saying ``typeError: iteration over a 0-d tensor''

Start in skip-thoughts.torch/pytorch/skipthoughts/skipthoughts.py at line 135 and reproduce the failure with a batch size of 1. Done means that case no longer raises the reported TypeError while the existing batch-length behavior remains intact.

Written by the indexing model from the issue text.