Numpy error

[shengzhang90]

Hi, MAISI team

When I run the inference script as the tutorials, I encountered the following issue. Thanks a lot.

Traceback (most recent call last):
File "/home/shengzhang/miniconda3/envs/maisi/lib/python3.10/runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/shengzhang/miniconda3/envs/maisi/lib/python3.10/runpy.py", line 86, in _run_code
exec(code, run_globals)
File "/media/ssd1/zhangsheng/code/MONAI/tutorials/generation/maisi/scripts/inference.py", line 228, in
main()
File "/media/ssd1/zhangsheng/code/MONAI/tutorials/generation/maisi/scripts/inference.py", line 217, in main
output_filenames = ldm_sampler.sample_multiple_images(args.num_output_samples)
File "/media/ssd1/zhangsheng/code/MONAI/tutorials/generation/maisi/scripts/sample.py", line 651, in sample_multiple_images
candidate_mask_files = self.find_closest_masks(num_img)
File "/media/ssd1/zhangsheng/code/MONAI/tutorials/generation/maisi/scripts/sample.py", line 1028, in find_closest_masks
top_region_index, bottom_region_index = get_body_region_index_from_mask(label)
File "/media/ssd1/zhangsheng/code/MONAI/tutorials/generation/maisi/scripts/utils.py", line 411, in get_body_region_index_from_mask
unique_elements = np.lib.arraysetops.unique(nda)
File "/home/shengzhang/miniconda3/envs/maisi/lib/python3.10/site-packages/numpy/lib/init.py", line 77, in getattr
raise AttributeError(
AttributeError: numpy.lib.arraysetops is now private. If you are using a public function, it should be available in the main numpy namespace, otherwise check the NumPy 2.0 migration guide.. Did you mean: 'array_utils'?

[shengzhang90]

ChatGPT's answer

This error occurs because, starting in NumPy 2.0, some internal functions that were previously accessible, like those in numpy.lib.arraysetops, have been moved to private namespaces. To resolve this issue, you'll need to modify your code to use public functions provided directly by NumPy's main namespace or adapt your code based on guidance from the NumPy 2.0 migration guide.

Here are some steps you can try:

1. Replace numpy.lib.arraysetops with Public Functions:

   • If you are using specific functions within numpy.lib.arraysetops (such as set operations on arrays), look for similar functions in the main numpy namespace. For instance:
     • Use numpy.unique instead of numpy.lib.arraysetops.unique.
     • Use numpy.intersect1d, numpy.setdiff1d, or numpy.union1d for set operations directly from numpy.

2. Review the Migration Guide:

   • The migration guide provides insights into specific changes and alternative solutions if a public equivalent isn't directly available.

3. Update the Package or Code:

   • If this error is within an external library, check if there's an updated version compatible with NumPy 2.0 or open an issue in the library's repository.