Why do universal adversarial attacks work on large language models?: Geometry might be the answer.
Varshini SubhashAnna BialasWeiwei PanFinale Doshi-VelezPublished in: CoRR (2023)
Keyphrases
- language model
- language modeling
- n gram
- document retrieval
- probabilistic model
- language modelling
- test collection
- speech recognition
- retrieval model
- language models for information retrieval
- statistical language models
- context sensitive
- query expansion
- information retrieval
- ad hoc information retrieval
- vector space model
- smoothing methods
- decision trees
- word error rate
- query terms
- question answering
- error rate
- language model for information retrieval