Fixed TorchvisionExtractor model_parameters

[a1247418]

Made model_parameters in TorchvisionExtractor a dict instead of a tuple of dict. Previously, extract_cls_token did not work for torchvision models, as line 36 in torch.py would prevent the creation of the extract_cls_token attribute when a tuple was passed as model_parameters from TorchvisionExtractor.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 73.24%. Comparing base (107d317) to head (5913fdb).
Report is 9 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #157   +/-   ##
=======================================
  Coverage   73.24%   73.24%           
=======================================
  Files          30       30           
  Lines        1555     1555           
  Branches      222      222           
=======================================
  Hits         1139     1139           
  Misses        349      349           
  Partials       67       67           

Flag	Coverage Δ
unittests	73.24% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[LukasMut]

@a1247418 LGTM. Could you add a test for one of the ImageNet-trained ViTs (from PyTorch) to this file where the CLS token is extracted, similar to how we do it for DINO? Then we can quickly check whether this workaround fixes the problem that you addressed.