Pickle Error

[vikashg]

Describe the bug
I wrote few transforms for handling geometric objects (points and lines). The code was written based on monai 0.8.1, when AddChannel was a valid Transform. However, as I am trying to revive the code with the latest monai 1.3, I get the following error.

[MemorizedFunc(func=<class 'GeometricTransforms_monai_1_3.AugmentBatchJPEG'>, location=./data/cache/joblib)]: Exception while loading results for AugmentBatchJPEG(num_augmentation=5, augmentation_dict={'adjustContrast': 1, 'intensityGaussian': 2, 'rotation': 0}, labels=['bbox'], output_size=(299, 299))
 Traceback (most recent call last):
  File "/usr/local/lib/python3.10/dist-packages/joblib/memory.py", line 568, in _cached_call
    out = self.store_backend.load_item(
  File "/usr/local/lib/python3.10/dist-packages/joblib/_store_backends.py", line 176, in load_item
    item = numpy_pickle.load(f)
  File "/usr/local/lib/python3.10/dist-packages/joblib/numpy_pickle.py", line 648, in load
    obj = _unpickle(fobj)
  File "/usr/local/lib/python3.10/dist-packages/joblib/numpy_pickle.py", line 577, in _unpickle
    obj = unpickler.load()
  File "/usr/lib/python3.10/pickle.py", line 1211, in load
    raise EOFError
EOFError

Instead of using AddChannel, I am using EnsureChannelFirst

Environment
Using monai 0.8.1 with nvcr.io/nvidia/pytorch:21.12-py3 gives correct results.
Using monai 1.3.0 with nvcr.io/nvidia/pytorch:23.12-py3 produces the above error.

Additional context
If someone knows a quick fix or is aware of this problem source, let me know. I will share the gist soon and link it with this issue.

[KumoLiu]

Perhaps useful, looks like a wrong file format than expected.
https://stackoverflow.com/questions/64865854/eoferror-by-using-joblib-load

[vikashg]

Thanks @KumoLiu I will investigate. Curious why it will happen though as I just changed the monai version. But I will look into that link you sent.