Embedding layer does not check input range

[PavelOstyakov]

🐛 Bug

Embedding layer does not check input range when running on cuda

To Reproduce

Code to reproduce the behavior:

import torch
import torch.nn as nn

emb = nn.Embedding(10, 20).cuda()
x = torch.zeros(5, 1).long() + 10
x = x.cuda()

y = emb(x)

z = torch.zeros(10).cuda()

RuntimeError                              Traceback (most recent call last)
<ipython-input-1-758fb390ae81> in <module>()
      8 y = emb(x)
      9 
---> 10 z = torch.zeros(10).cuda()

RuntimeError: CUDA error: device-side assert triggered

Expected behavior

It should raise a correct error

Environment

PyTorch version: 1.0.1.post2
Is debug build: No
CUDA used to build PyTorch: 9.0.176

OS: Ubuntu 18.04.1 LTS
GCC version: (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0
CMake version: Could not collect

Python version: 3.6
Is CUDA available: Yes
CUDA runtime version: 9.1.85
GPU models and configuration:
GPU 0: GeForce GTX 1080 Ti
GPU 1: GeForce GTX 1080 Ti

Nvidia driver version: 390.116
cuDNN version: Could not collect

Versions of relevant libraries:
[pip3] numpy==1.16.2
[pip3] torch==1.0.1.post2
[pip3] torchvision==0.2.2.post3
[conda] Could not collect

[colesbury]

We're unlikely to be able to check the index bounds as a recoverable Python error due to performance implications. But we can do a few things:

1. Improve the error message (improved assert message in the case of "CUDA error: device-side assert triggered" #17425)
2. Add "index out of bounds" to the assertion like in Tensor indexing
3. Document the behavior in https://pytorch.org/docs/stable/nn.html?highlight=embedding#torch.nn.Embedding