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

Kaiyue Wen Yuchen Li Bingbin Liu Andrej Risteski

Published in: NeurIPS (2023)

Keyphrases

significant improvement
benchmark datasets
genetic algorithm
decision trees
case study
machine learning methods
computationally expensive
learning algorithm
computer vision
similarity measure
bayesian networks