diff --git a/data/data_manager.py b/data/data_manager.py index 6e95c08dd2d37a69c2ac167dfb172ced98d2e9b7..787b4570d569161747f52b98594e50c34c8cc755 100644 --- a/data/data_manager.py +++ b/data/data_manager.py @@ -41,8 +41,8 @@ def get_annotated_sentences(dataset, test_instances=10): elif dataset == "Pile-NER-type": return random.sample(pile_interface.get_annotated_sentences(), min(test_instances, len(pile_interface.get_annotated_sentences()))) elif dataset == "FIGER-coarse": - return figer_interface.get_annotated_sentences_coarse() # todo implement limiting output instances internally + return random.sample(figer_interface.get_annotated_sentences_coarse(), min(test_instances, len(figer_interface.get_annotated_sentences_coarse()))) elif dataset == "FIGER-fine": - return figer_interface.get_annotated_sentences_fine() + return random.sample(figer_interface.get_annotated_sentences_fine(), min(test_instances, len(figer_interface.get_annotated_sentences_fine()))) else: raise Exception