Combining Transformers with Natural Language Explanations

Federico Ruggeri, Marco Lippi, Paolo Torroni

Many NLP applications require models to be interpretable. However, many successful neural architectures, including transformers, still lack effective interpretation methods. A poss…