search

SforAiDl / vformer

A modular PyTorch library for vision transformer models
https://vformer.readthedocs.io/
MIT License
162 stars 22 forks source link

Remove timm as dependency #88

Closed alvanli closed 2 years ago

codecov-commenter commented 2 years ago

Codecov Report

Merging #88 (86c3391) into main (459592a) will not change coverage. The diff coverage is 100.00%.

@@            Coverage Diff            @@
##              main       #88   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           66        66           
  Lines         1977      1972    -5     
=========================================
- Hits          1977      1972    -5
Impacted Files Coverage Δ
vformer/models/classification/swin.py 100.00% <ø> (ø)
vformer/attention/window.py 100.00% <100.00%> (ø)
vformer/encoder/convit.py 100.00% <100.00%> (ø)
vformer/encoder/convvt.py 100.00% <100.00%> (ø)
vformer/encoder/embedding/pos_embedding.py 100.00% <100.00%> (ø)
vformer/encoder/pyramid.py 100.00% <100.00%> (ø)
vformer/encoder/swin.py 100.00% <100.00%> (ø)
vformer/encoder/vanilla.py 100.00% <100.00%> (ø)
vformer/models/classification/convvt.py 100.00% <100.00%> (ø)
vformer/models/classification/pyramid.py 100.00% <100.00%> (ø)
abhi-glitchhg commented 2 years ago

Can we give user some more flexibility by having mode as an input parameter of the class?

Other than that it looks good to merge.

NeelayS commented 2 years ago

Can we give user some more flexibility by having mode as an input parameter of the class?

Yeah, I guess we can have a drop_path_mode parameter in the constructor. What do you think?

NeelayS commented 2 years ago

Looks good, merging!