Transformer-based language models significantly advanced the state-of-the-art in many
linguistic tasks. As this revolution continues, the ability to explain model predictions has …

Variants of the BERT architecture specialised for producing full-sentence representations
often achieve better performance on downstream tasks than sentence embeddings …

While recent years have witnessed the emergence of various explainable methods in
machine learning, to what degree the explanations really represent the reasoning process …

Transformer-based language models significantly advanced the state-of-the-art in many
linguistic tasks. As this revolution continues, the ability to explain model predictions has …