Added Docstrings for Deberta and DebertaV2 [PyTorch] #18610
Conversation
This reverts commit 307185e.
|
The documentation is not available anymore as the PR was closed or merged. |
There was a problem hiding this comment.
Excellent job, @Tegzes ! Thank you very much.
@patrickvonplaten Need your approval to merge this PR (the review should be fairly easy)
|
@patrickvonplaten would like to have your feedback on this |
|
request @LysandreJik for a final review in order to merge |
There was a problem hiding this comment.
No problem for me to merge like this, but it's a bit sad that there are no models on the Hub that can be used for sequence classification with DeBERTa, nor any model that can be used for DeBERTa-v2.
Should we train some with the scripts available in the repo? :)
Up to you @ydshieh!
| _TOKEN_CLASS_EXPECTED_OUTPUT = ( | ||
| "['LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0', 'LABEL_0'," | ||
| " 'LABEL_0', 'LABEL_0']" | ||
| ) |
There was a problem hiding this comment.
It's a bit unfortunate that we have LABEL_0 here. It would be better to have a label map that is defined.
|
Let's merge as it is. The usage of tiny models for doc is not ideal, but we decided to use them so doctest could run. There are several (downstream) models have done this way. We/I can definitely find some time to train (at least a few) models. |
|
Sounds good to me! |
* Added Doctest for Deberta Pytorch * Added path in documentation test file * Added docstrings for DebertaV2 * Revert "Added docstrings for DebertaV2" This reverts commit 307185e. * Added DebertaV2 Docstrings
Adds Doctest for DeBerta and DeBertaV2 [Pytorch version]
Issue: #16292
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
@ydshieh @patrickvonplaten