Transformers are uninterpretable with myopic methods: a case study with bounded Dyck grammars

Kaiyue Wen, Yuchen Li 0007, Bingbin Liu, Andrej Risteski. Transformers are uninterpretable with myopic methods: a case study with bounded Dyck grammars. In Alice Oh, Tristan Naumann, Amir Globerson, Kate Saenko, Moritz Hardt, Sergey Levine, editors, Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023. 2023. [doi]

Authors

Kaiyue Wen

This author has not been identified. Look up 'Kaiyue Wen' in Google

Yuchen Li 0007

This author has not been identified. Look up 'Yuchen Li 0007' in Google

Bingbin Liu

This author has not been identified. Look up 'Bingbin Liu' in Google

Andrej Risteski

This author has not been identified. Look up 'Andrej Risteski' in Google