RuntimeError: No such operator image::read_file

[datumbox]

🐛 Describe the bug

The CI job unittest_windows_gpu_py3.8 stared failing with:

______ test_decode_jpeg_cuda[False-ImageReadMode.GRAY-gray_pytorch.jpg] _______
Traceback (most recent call last):
  File "C:\Users\circleci\project\env\lib\site-packages\torch\_ops.py", line 501, in __getattr__
    op, overload_names = torch._C._jit_get_operation(qualified_op_name)
RuntimeError: No such operator image::read_file

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "C:\Users\circleci\project\test\test_image.py", line 363, in test_decode_jpeg_cuda
    data = read_file(img_path)
  File "c:\users\circleci\project\torchvision\io\image.py", line 47, in read_file
    data = torch.ops.image.read_file(path)
  File "C:\Users\circleci\project\env\lib\site-packages\torch\_ops.py", line 505, in __getattr__
    raise AttributeError(
AttributeError: '_OpNamespace' 'image' object has no attribute 'read_file'

It seems like the kernel is not registered properly. Might be related to a change on Core.

Versions

The issue started appearing on torch-1.13.0.dev20221004 py3.8_cuda11.6_cudnn8_0

[atalman]

This issue is caused by inclusion of cudart dll in windows torch/lib folder. The builder change #1157 is in place to resolve this issue

[weiwangmeta]

pytorch/builder#1157 :)

[dbl001]

I got the exception: "AttributeError: '_OpNamespace' 'image' object has no attribute 'read_file'"
with Pytorch Version: 2.0.0a0+gitf8b2879 on MacOS Ventura 13.1

Shared Interest project:
https://github.com/mitvis/shared-interest

 % python test.py
/Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torchvision-0.14.0a0+a61e6ef-py3.10-macosx-10.9-x86_64.egg/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension: dlopen(/Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torchvision-0.14.0a0+a61e6ef-py3.10-macosx-10.9-x86_64.egg/torchvision/image.so, 0x0006): Symbol not found: __ZN2at4_ops19empty_memory_format4callEN3c108ArrayRefIxEENS2_8optionalINS2_10ScalarTypeEEENS5_INS2_6LayoutEEENS5_INS2_6DeviceEEENS5_IbEENS5_INS2_12MemoryFormatEEE
  Referenced from: <B53D2070-C006-35E7-821C-04385AFECC3B> /Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torchvision-0.14.0a0+a61e6ef-py3.10-macosx-10.9-x86_64.egg/torchvision/image.so
  Expected in:     <EFA33186-A337-3933-9E32-73FE6E861466> /Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torch/lib/libtorch_cpu.dylib
  warn(f"Failed to load image Python extension: {e}")
Traceback (most recent call last):
  File "/Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torch/_ops.py", line 562, in __getattr__
    op, overload_names = torch._C._jit_get_operation(qualified_op_name)
RuntimeError: No such operator image::read_file

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/davidlaxer/shared-interest/test.py", line 10, in <module>
    img = read_image("/Users/davidlaxer/ImageNet/ILSVRC/Data/CLS-LOC/train/n01494475/n01494475_6.JPEG")
  File "/Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torchvision-0.14.0a0+a61e6ef-py3.10-macosx-10.9-x86_64.egg/torchvision/io/image.py", line 251, in read_image
    data = read_file(path)
  File "/Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torchvision-0.14.0a0+a61e6ef-py3.10-macosx-10.9-x86_64.egg/torchvision/io/image.py", line 47, in read_file
    data = torch.ops.image.read_file(path)
  File "/Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torch/_ops.py", line 566, in __getattr__
    raise AttributeError(
AttributeError: '_OpNamespace' 'image' object has no attribute 'read_file'

% pip show torch
Name: torch
Version: 2.0.0a0+gitf8b2879
Summary: Tensors and Dynamic neural networks in Python with strong GPU acceleration
Home-page: https://pytorch.org/
Author: PyTorch Team
Author-email: packages@pytorch.org
License: BSD-3
Location: /Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages
Requires: networkx, sympy, typing_extensions
Required-by: accelerate, aifeynman, audiolm-pytorch, benepar, captum, parallelformers, pytorch-transformers, sentence-transformers, torch-struct, torchaudio, torchtraining-nightly, torchvision, vector-quantize-pytorch, whisper, xformers

(AI-Feynman) davidlaxer@x86_64-apple-darwin13 shared-interest % pip show torchvision
Name: torchvision
Version: 0.14.0a0+a61e6ef
Summary: image and video datasets and models for torch deep learning
Home-page: https://github.com/pytorch/vision
Author: PyTorch Core Team
Author-email: soumith@pytorch.org
License: BSD
Location: /Users/davidlaxer/anaconda3/envs/AI-Feynman/lib/python3.10/site-packages/torchvision-0.14.0a0+a61e6ef-py3.10-macosx-10.9-x86_64.egg
Requires: numpy, pillow, requests, torch, typing_extensions
Required-by: aifeynman, sentence-transformers