Describe the Bug

When the dimension of the input tensor is 3 or more, the functions all and any can't work normally.

To Reproduce

Expected Behavior

Code Snippets

System Details (please complete the following information):

• OS: [e.g. Ubuntu 18.04]
• CUDA version: [e.g CUDA 11.4]
• g++ version: [e.g. 9.3.0]

Additional Context