QA-Attack fuses attention-based and removal-based word ranking to locate vulnerable words in question-answering contexts and substitutes synonyms, fooling T5, LongT5, and BERT QA models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Deceiving Question-Answering Models: A Hybrid Word-Level Adversarial Approach
QA-Attack fuses attention-based and removal-based word ranking to locate vulnerable words in question-answering contexts and substitutes synonyms, fooling T5, LongT5, and BERT QA models.