A Pytorch ModelType for feature extraction and embeddings

[eddyxu]

what makes this experimental - that only resnet is tested?

Yes, it might take a few iterations to improve the API and postprocessing (predict) compatibilities with more models.

[da-liii]

torchvision 0.11.2

>>> import torchvision
>>> torchvision.__version__
'0.11.2'
>>> import torchvision.models.convnext_base
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'torchvision.models.convnext_base'

torchvision 0.12.0

Python 3.8.10 (default, May 19 2021, 11:01:55)
[Clang 10.0.0 ] :: Anaconda, Inc. on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import torchvision
>>> torchvision.__version
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: module 'torchvision' has no attribute '__version'
>>> torchvision.__version__
'0.12.0'
>>> import torchvision.models.convnext_base
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'torchvision.models.convnext_base'
>>>

[da-liii]

Python 3.8.10 (default, May 19 2021, 11:01:55)
[Clang 10.0.0 ] :: Anaconda, Inc. on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import torchvision
>>> torchvision.models.convnext_base
<function convnext_base at 0x7f9b983f5ee0>