33 citations · 88 across the 25 of their papers we have counts for
Showing 2020Show all
3 papers · 1 filter
cs.CL2020★ 23 cited
Optimal Subarchitecture Extraction For BERT
Adrian de Wynter, Daniel J. Perry
We extract an optimal subset of architectural parameters for the BERT architecture from Devlin et al. (2018) by applying recent breakthroughs in algorithms for neural architecture…
cs.CL2020★ 1 cited
Mischief: A Simple Black-Box Attack Against Transformer Architectures
Adrian de Wynter
We introduce Mischief, a simple and lightweight method to produce a class of human-readable, realistic adversarial examples for language models. We perform exhaustive experimentati…
cs.LG2020★ 1 cited
An Approximation Algorithm for Optimal Subarchitecture Extraction
Adrian de Wynter
We consider the problem of finding the set of architectural parameters for a chosen deep neural network which is optimal under three metrics: parameter size, inference speed, and e…