transformers_sklearn wrap the powerful APIs of transformers into three class. They are BERTologyClassifier for classification task, BERTologyNERClassifier for name entity recognition(NER) task and BERTologyRegressor for regression task. Within each class, there are three methods, which are fit for fine-tuning the pre-trained models downloaded from community, score for scoring the performance of fine-tuned model, and predict for predicting the labels of given datasets.
fit and score methods accept two parameters, X and y. The type of X and y could be list, ndarray and DataFrame. predict only need X.