[MRG] Include EEGSym

[PierreGtch]

Follow-up on #664

I fixed the channel split
Also, the time and channel dimensions were inverted in the convolutions which I fixed

However I don't understand why in_channels!=filters in the residual blocks. It does not seem possible because of the resudual connections.
@Serpeve could you have a look? In particular if the in_channels, out_channels and kernel_size are consistent with your original implementation

CC @bruAristimunha

[PierreGtch]

@bruAristimunha its not working yet. The model integration test should fail (because of the number of filters or kernel sizes, not sure)

[bruAristimunha]

yeah yeah. I can incorporate a small benchmark with physionet to reproduce the results when you tell me it works

[PierreGtch]

Let's wait for @Serpeve's input to figure out the filters and kernel sizes, and make sure this pytorch re-implementation is consistent with his original tensorflow one before we merge

[codecov]

Codecov Report

❌ Patch coverage is 97.60766% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 87.43%. Comparing base (0b0b876) to head (23a014c).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #697      +/-   ##
==========================================
+ Coverage   87.18%   87.43%   +0.25%     
==========================================
  Files          99      101       +2     
  Lines        8406     8614     +208     
==========================================
+ Hits         7329     7532     +203     
- Misses       1077     1082       +5     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[bruAristimunha]

Thanks @PierreGtch :)

[PierreGtch]

Thanks for the PR @bruAristimunha !!!

FYI @Serpeve your model is in braindecode :D