Backdoor text samples show smaller log-probability changes under mask-filling perturbations than clean samples, which enables zero-shot backdoor detection without the poisoned model.
Llama3, The llama 3 herd of models, https://llama.meta.com/, accessed: 2024-7-23 (2024)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Backdoor Samples Detection Based on Perturbation Discrepancy Consistency in Pre-trained Language Models
Backdoor text samples show smaller log-probability changes under mask-filling perturbations than clean samples, which enables zero-shot backdoor detection without the poisoned model.