🐛 Bug

python crashes (with SIGFPE) when dropout is called on an empty cuda tensor.

To Reproduce

Steps to reproduce the behavior:

import torch
torch.nn.functional.dropout(torch.Tensor([]).to('cuda'))
Floating point exception (core dumped)

Expected behavior

Either a python exception should be raised or it should be a no-op.

Environment

• PyTorch Version (e.g., 1.0): master
• OS (e.g., Linux): centos
• How you installed PyTorch (conda, pip, source): source
• Build command you used (if compiling from source):
• Python version: 3.6
• CUDA/cuDNN version: cuda 9.2
• GPU models and configuration:
• Any other relevant information:

Additional context