Revisiting the Practical Effectiveness of Constituency Parse Extraction from Pre-trained Language Models

4749

Knowledge Triples