Automatic and Universal Prompt Injection Attacks against Large Language Models.
Xiaogeng LiuZhiyuan YuYizhe ZhangNing ZhangChaowei XiaoPublished in: CoRR (2024)
Keyphrases
- language model
- language modeling
- language modelling
- n gram
- injection attacks
- document retrieval
- context sensitive
- information retrieval
- query expansion
- retrieval model
- probabilistic model
- speech recognition
- statistical language models
- test collection
- smoothing methods
- statistical language modeling
- language models for information retrieval
- relevance model
- document ranking
- query terms
- prediction accuracy
- database