[BUG] Captum tutorial fails when visualize_image_attr() receives None for original_image

There is the following issue on this page: https://docs.pytorch.org/tutorials/beginner/introyt/captumyt.html

Hi PyTorch Team,

I got an an error when following the tutorial Model Understanding with Captum part. The issue is within Feature Attribution with Integrated Gradients section. The tutorial uses : """ _ = viz.visualize_image_attr( None, np.transpose( transformed_img.squeeze().cpu().detach().numpy(), (1, 2, 0) ), method="original_image", title="Original Image" ) """

When running that cell, I'm getting this error : """ AttributeError: 'NoneType' object has no attribute 'shape' """

The traceback points to captum/attr/_utils/visualization.py, where visualize_image_attr() attempts to access attr.shape. Since the tutorial passes None as attr, this fails before the "original_image" visualization is produced.

======================================================

My utilised env: """ PyTorch version: 2.10.0 TorchVision version: 0.25.0 Python version: 3.14.7 OS: macOS 26.6.2 arm64 CPU: Apple M4 Pro NumPy: 2.4.2 Captum: <0.9.0> """

and python3 -m torch.utils.collect_env output: """ Collecting environment information... PyTorch version: 2.10.0 Is debug build: False CUDA used to build PyTorch: None ROCM used to build PyTorch: N/A

OS: macOS 26.6.2 (arm64) Clang version: 17.0.0

Python version: 3.14.7 Python platform: macOS-26.6.2-arm64-arm-64bit-Mach-O Is CUDA available: False Is XPU available: False Is XNNPACK available: True

CPU: Apple M4 Pro

Versions of relevant libraries: numpy==2.4.2 torch==2.10.0 torchvision==0.25.0 """

Since, the page current verfied date is Nov 05, 2024, it might be an update/compatibility issue.

Best !