issues with the "DataParallelCriterion"

I employ the DataParallelCriterion like below,

    model = DataParallelModel(deeplab)
    model.train()     
    model.float()
    model.cuda()    
    # .......
    criterion = DataParallelCriterion(criterion)
    criterion.cuda()
    cudnn.benchmark = True

Then I got such errors inform me an error related the line 163 in your parallel.py

Traceback (most recent call last):
  File "train.py", line 160, in <module>
  File "train.py", line 137, in main
    loss = criterion(preds, labels)
  File "/root/miniconda3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 491, in __call__
    result = self.forward(*input, **kwargs)
  File "/teamscratch/msravcshare/yuyua/OCNet/utils/parallel.py", line 134, in forward
    outputs = _criterion_parallel_apply(replicas, inputs, targets, kwargs)
  File "/teamscratch/msravcshare/yuyua/OCNet/utils/parallel.py", line 188, in _criterion_parallel_apply
    raise output
  File "/teamscratch/msravcshare/yuyua/OCNet/utils/parallel.py", line 163, in _worker
    output = module(*(input + target), **kwargs)
TypeError: can only concatenate list (not "tuple") to list

Start with utils/parallel.py at _worker line 163 and follow the call from _criterion_parallel_apply; compare the input and target values passed from train.py at criterion(preds, labels). Reproduce the DataParallelCriterion call and confirm that training no longer raises the reported list/tuple concatenation TypeError.