UOR: Universal Backdoor Attacks on Pre-trained Language Models.
Wei DuPeixuan LiHaodong ZhaoTianjie JuGe RenGongshen LiuPublished in: ACL (Findings) (2024)
Keyphrases
- language model
- pre trained
- language modeling
- training data
- probabilistic model
- n gram
- training examples
- document retrieval
- information retrieval
- query expansion
- speech recognition
- smoothing methods
- statistical language models
- retrieval model
- control signals
- test collection
- document ranking
- language modelling
- relevance model
- data sets
- text mining
- small number
- active learning
- learning algorithm
- statistical language modeling
- language models for information retrieval
- spoken term detection