View size is not compatible with input tensor's size and stride

#1

i edited .view() to .reshape() but same error still occured…

def accuracy(output, target, topk=(1,)):
  maxk = max(topk)
  batch_size = target.size(0)

  _, pred = output.topk(maxk, 1, True, True)
  pred = pred.t()
  correct = pred.eq(target.view(1, -1).expand_as(pred))

  res = []
  for k in topk:
    correct_k = correct[:k].reshape(-1).float().sum(0)
    res.append(correct_k.mul_(100.0/batch_size))
  return res

full error sentence:
RuntimeError: view size is not compatible with input tensor’s size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(…) instead.

i don’t know why it still cry an error like this…

#2

The posted error is raised, if the tensor it not contiguous in memory and thus view will fail.
You could either use .reshape instead as suggested in the error message or .contiguous().view(...) alternatively.

3 Likes
#3

thank you! ptrblck!
reshape doesn’t work for me but .contiguous() works! :smiley: