ROSE Doesn't Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding.
Qihuang ZhongLiang DingJuhua LiuBo DuDacheng TaoPublished in: CoRR (2024)
Keyphrases
- hidden markov models
- language model
- language modeling
- speech recognition
- probabilistic model
- n gram
- information retrieval
- language modelling
- test collection
- document retrieval
- smoothing methods
- query expansion
- vector space model
- ad hoc information retrieval
- feature selection
- learning algorithm
- multimedia
- query terms
- context sensitive
- retrieval model
- pseudo relevance feedback
- document ranking
- statistical language models
- language models for information retrieval
- language model for information retrieval
- relevance model
- translation model
- word error rate
- co occurrence