some problems while running gen_lama_prior

[Starryskyswt]

Thanks to your assistance,I have overcome all the problems I have mentioned. Now when I run gen_lama_prior,I find some mistakes and I am confused about the reason ,could you help me figure out the problem?

[cuteyyt]

I haven't seen such mistakes before, and could you provide more details, like which step in the script causes such an error?

And when I run lama with multi GPUs, setting environment variables like CUDA_VISIBLE_DEVICES=2 may cause problems. You may try the script without specifying the GPU.

[Starryskyswt]

I have deleted CUDA_VISIBLE_DEVICES=2 ,but it seems that it doesn't work. And I send the feedlback to the Chatgpt,it gives me some explanations as follow.

It seems like you're encountering an error related to a Python script that uses various libraries and modules such as TensorFlow, skimage, and albumentations. The error message you provided indicates an issue with the usage of the np.bool attribute from the NumPy library, which has been deprecated and replaced with the built-in bool type.

Here's a breakdown of the error message:

The error occurs in the script /data0/swt/or-nerf-master/prior/lama/bin/predict.py.
It seems to be using TensorFlow and TensorFlow-related warnings are shown.
There's a warning related to TensorRT (TF-TRT Warning: Could not find TensorRT).
The script then tries to import modules from the albumentations library for image augmentations.
The error traceback shows that the issue originates from the skimage library when it tries to use the deprecated np.bool attribute.
To address this issue, you might need to make some modifications to the code to replace the deprecated np.bool with the built-in bool type. You would need to find the specific line in the skimage library or the related code where np.bool is being used and change it to bool. It's possible that the library has already been updated to use bool, so you might want to consider updating the skimage library to its latest version.

Keep in mind that working with code from multiple libraries that have dependencies on each other can sometimes be complex, and updates to one library might require updates in others. If you're not the developer of this code, you might want to check for updates in the relevant libraries or reach out to the maintainers for support.

And that's what make me feel confused. It seems that the code should be corrected somewhere,and I will spend some time check it again.

[Starryskyswt]

Oh it's amazing,After reviewing the information, I found that the name np.bool has been deprecated in Numpy's 1.24 version and replaced with np.bool_ .Now,it works!