ModuleNotFoundError: No module named 'numpy._core' when using residual encoder

[redundancer]

I try to run the predictions, but it does not work for my model with residual encoder. The one without residuals works as expected:

(matto) PS C:\Users\anonymous\local_projects\MATTO\nnunetv2> nnUNetv2_predict -d 101 -i .\nnUNet_raw\Dataset101_MATTO-GTV-PET\imagesTs\ -o .\inference\Dataset101_MATTO-GTV-PET\default\not_preprocessed -f  0 1 2 3 4 -tr nnUNetTrainer -c 3d_fullres -p nnUNetPlans

#######################################################################
Please cite the following paper when using nnU-Net:
Isensee, F., Jaeger, P. F., Kohl, S. A., Petersen, J., & Maier-Hein, K. H. (2021). nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature methods, 18(2), 203-211.
#######################################################################

There are 28 cases in the source folder
I am process 0 out of 1 (max process ID is 0, we start counting with 0!)
There are 28 cases that I would like to predict

Predicting 232:
perform_everything_on_device: True
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:05<00:00,  2.82s/it]
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 10.72it/s]
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00,  9.90it/s]
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 10.47it/s]
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00,  9.90it/s]
sending off prediction to background worker for resampling and export
done with 232

I can also find the best configuration for the residual model:

(matto) PS C:\Users\anonymous\local_projects\MATTO\nnunetv2> nnUNetv2_find_best_configuration 101 -c 3d_fullres -p nnUNetResEncUNetLPlans

***All results:***
nnUNetTrainer__nnUNetResEncUNetLPlans__3d_fullres: 0.6783556306654125

*Best*: nnUNetTrainer__nnUNetResEncUNetLPlans__3d_fullres: 0.6783556306654125

***Determining postprocessing for best model/ensemble***
Removing all but the largest foreground region did not improve results!

***Run inference like this:***

nnUNetv2_predict -d Dataset101_MATTO-GTV-PET -i INPUT_FOLDER -o OUTPUT_FOLDER -f  0 1 2 3 4 -tr nnUNetTrainer -c 3d_fullres -p nnUNetResEncUNetLPlans

***Once inference is completed, run postprocessing like this:***

nnUNetv2_apply_postprocessing -i OUTPUT_FOLDER -o OUTPUT_FOLDER_PP -pp_pkl_file C:/Users/anonymous/local_projects/MATTO/nnunetv2/nnUNet_results\Dataset101_MATTO-GTV-PET\nnUNetTrainer__nnUNetResEncUNetLPlans__3d_fullres\crossval_results_folds_0_1_2_3_4\postprocessing.pkl -np 8 -plans_json C:/Users/anonymous/local_projects/MATTO/nnunetv2/nnUNet_results\Dataset101_MATTO-GTV-PET\nnUNetTrainer__nnUNetResEncUNetLPlans__3d_fullres\crossval_results_folds_0_1_2_3_4\plans.json

But when I try to run the inference with the residual model I get the following error:

(matto) PS C:\Users\anonymous\local_projects\MATTO\nnunetv2> nnUNetv2_predict -d Dataset101_MATTO-GTV-PET -i .\nnUNet_raw\Dataset101_MATTO-GTV-PET\imagesTs\ -o .\inference\Dataset101_MATTO-GTV-PET\ResNet\no_postprocessing -f  0 1 2 3 4 -tr nnUNetTrainer -c 3d_fullres -p nnUNetResEncUNetLPlans

#######################################################################
Please cite the following paper when using nnU-Net:
Isensee, F., Jaeger, P. F., Kohl, S. A., Petersen, J., & Maier-Hein, K. H. (2021). nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature methods, 18(2), 203-211.
#######################################################################

Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "C:\Users\anonymous\AppData\Local\anaconda3\envs\matto\Scripts\nnUNetv2_predict.exe\__main__.py", line 7, in <module>
  File "C:\Users\anonymous\AppData\Local\anaconda3\envs\matto\Lib\site-packages\nnunetv2\inference\predict_from_raw_data.py", line 859, in predict_entry_point
    predictor.initialize_from_trained_model_folder(
  File "C:\Users\anonymous\AppData\Local\anaconda3\envs\matto\Lib\site-packages\nnunetv2\inference\predict_from_raw_data.py", line 84, in initialize_from_trained_model_folder
    checkpoint = torch.load(join(model_training_output_dir, f'fold_{f}', checkpoint_name),
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\anonymous\AppData\Local\anaconda3\envs\matto\Lib\site-packages\torch\serialization.py", line 1025, in load
    return _load(opened_zipfile,
           ^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\anonymous\AppData\Local\anaconda3\envs\matto\Lib\site-packages\torch\serialization.py", line 1446, in _load
    result = unpickler.load()
             ^^^^^^^^^^^^^^^^
  File "C:\Users\anonymous\AppData\Local\anaconda3\envs\matto\Lib\site-packages\torch\serialization.py", line 1439, in find_class
    return super().find_class(mod_name, name)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ModuleNotFoundError: No module named 'numpy._core'

Why is this and how can I fix it? Thanks a lot!