DACT-BERT: Differentiable Adaptive Computation Time for an Efficient BERT Inference

DACT-BERT: Differentiable Adaptive Computation Time for an Efficient BERT Inference Cristobal Eyzaguirre author Felipe del Rio author Vladimir Araujo author Alvaro Soto author 2022-05 text Proceedings of NLP Power! The First Workshop on Efficient Benchmarking in NLP Tatiana Shavrina editor Vladislav Mikhailov editor Valentin Malykh editor Ekaterina Artemova editor Oleg Serikov editor Vitaly Protasov editor Association for Computational Linguistics Dublin, Ireland conference publication eyzaguirre-etal-2022-dact 10.18653/v1/2022.nlppower-1.10 https://aclanthology.org/2022.nlppower-1.10/ 2022-05 93 99