repr of device is broken

[rgommers]

🐛 Bug

The repr of device instances is broken - the index for GPUs is missing, as is a closing bracket. A repr is supposed to be able to be eval-ed (not that that's a good idea typically, but there's no reason for it not being true here).

To Reproduce

In [1]: import torch                                                         

In [2]: t = torch.tensor([1, 2, 3, 4], device='cuda:0')                      

In [3]: t                                                                    
Out[3]: tensor([1, 2, 3, 4], device='cuda:0')

In [4]: t.device                                                             
Out[4]: device(type='cuda', index=

In [5]: repr(t.device)                                                      
Out[5]: "device(type='cuda', index="

In [6]: str(t.device)                                                       
Out[6]: 'cuda:0'

Expected behavior

Repr should be, in the above case, "device(type='cuda', index=0)"

Environment

$ python torch/utils/collect_env.py 
Collecting environment information...
PyTorch version: 1.8.0a0+2fe382e
Is debug build: False
CUDA used to build PyTorch: 11.1
ROCM used to build PyTorch: N/A

OS: Antergos Linux (x86_64)
GCC version: (GCC) 8.4.0
Clang version: 11.0.0
CMake version: version 3.14.0

Python version: 3.7 (64-bit runtime)
Is CUDA available: True
CUDA runtime version: Could not collect
GPU models and configuration: GPU 0: GeForce GTX 1080 Ti
Nvidia driver version: 455.38
cuDNN version: Probably one of the following:
/usr/lib/libcudnn.so.8.0.5
/usr/lib/libcudnn_adv_infer.so.8.0.5
/usr/lib/libcudnn_adv_train.so.8.0.5
/usr/lib/libcudnn_cnn_infer.so.8.0.5
/usr/lib/libcudnn_cnn_train.so.8.0.5
/usr/lib/libcudnn_ops_infer.so.8.0.5
/usr/lib/libcudnn_ops_train.so.8.0.5
HIP runtime version: N/A
MIOpen runtime version: N/A

Versions of relevant libraries:
[pip3] numpy==1.18.1
[pip3] pytorch-sphinx-theme==0.0.24
[pip3] torch==1.8.0a0
[conda] blas                      1.0                         mkl  
[conda] mkl                       2020.0                      166  
[conda] mkl-include               2019.4                      243  
[conda] mkl-service               2.3.0            py37he904b0f_0  
[conda] mkl_fft                   1.1.0            py37h23d657b_0  
[conda] mkl_random                1.1.0            py37hd6b4f25_0  
[conda] numpy                     1.18.1           py37h4f9e942_0  
[conda] numpy-base                1.18.1           py37hde5b4d6_1  
[conda] pytorch-sphinx-theme      0.0.24                    dev_0    <develop>
[conda] torch                     1.8.0a0                   dev_0    <develop>

cc @ezyang @gchanan @zou3519 @bdhirsh

[gchanan]

confirmed on master.

[gchanan]

also a regression from 1.7.0.

[gchanan]

I can't reproduce it in a local build but the nightlies do show the problem.

[sparkingdark]

Can I work on it and try to give a fix?

[rgommers]

@sparkingdark please feel free to work on any open issue that's unassigned. In this case it's a high-prio regression, so there's some time pressure, which means someone else may jump in and fix it. If you're new to PyTorch development, I'd usually recommend to pick issues where it's clear what to do and that haven't moved in say a week or two.

[sparkingdark]

okay I try to do my best and I am new to pytorch development can you tell me how to getting started actually I was looking for good first issue but all of them are taken that's why i am now searching above that. @rgommers

[sparkingdark]

hey i need some help @rgommers can you tell me how to compile and install the pytorch 1.8.1.8.0a0 is it on the developer branch.
and one thing can I use google colab for testing purpose.

[kshitij12345]

@sparkingdark Have a look at https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md#developing-pytorch
That has all the necessary instructions.

[sparkingdark]

Yeah I am new to pytorch development and try to do it,is their any community channels available for pytorch I can join ?

…

On Tue, 1 Dec 2020, 20:19 Ralf Gommers, ***@***.***> wrote: @sparkingdark <https://github.com/sparkingdark> please feel free to work on any open issue that's unassigned. In this case it's a high-prio regression, so there's some time pressure, which means someone else may jump in and fix it. If you're new to PyTorch development, I'd usually recommend to pick issues where it's clear what to do and that haven't moved in say a week or two. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#48585 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AL7XHCDX5RENFMP6BL6W5PTSST7BFANCNFSM4UHWC4SA> .

[rgommers]

Yeah I am new to pytorch development and try to do it,is their any community channels available for pytorch I can join ?

https://discuss.pytorch.org/. A word of advice though: please do your homework first, like searching for the available information on a topic, and then ask a specific question when you do get stuck. That makes it much more likely someone will answer your question.

[sparkingdark]

Yeah will do. Thanks for sharing.

…

On Wed, 2 Dec 2020, 17:44 Ralf Gommers, ***@***.***> wrote: Yeah I am new to pytorch development and try to do it,is their any community channels available for pytorch I can join ? https://discuss.pytorch.org/. A word of advice though: please do your homework first, like searching for the available information on a topic, and then ask a specific question when you do get stuck. That makes it much more likely someone will answer your question. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#48585 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AL7XHCATOOVJJAOMFIFIM7TSSYVRBANCNFSM4UHWC4SA> .