- Added missing extra SEP token in RoBERTa, CamemBERT, and XLMRoBERTA in sentence pair tasks.
0.18.110.18.11
Added
- Added `no_cache` option to `global_args` which disables caching (saving and loading) of features to/from disk.
0.18.100.18.10
Added
- Added Makefile with tests dependency installation, test code, formatter and types. - Added setup.cfg file with Make configuration - Added some tests for the functionality
Changed
- Files linted using flake8 - Files formatted using black - Test tested with pytest - Unused variables deleted
0.18.90.18.9
Fixed
- Fixed bug with importing certain pre-trained models in `MultiLabelClassificationModel` .
0.18.80.18.8
Added
- Added `**kwargs` to the init methods of `ClassificationModel` , `MultiLabelClassificationModel` , `QuestionAnsweringModel` , and `NERModel` . These will be passed to the `from_pretrained()` method of the underlying model class.