Fetching the paper…

Rethinking Attention-Model Explainability through Faithfulness Violation Test · Around