AI Sandbagging: Language Models can Strategically Underperform on Evaluations.
Teun van der WeijFelix HofstätterOllie JaffeSamuel F. BrownFrancis Rhys WardPublished in: CoRR (2024)
Keyphrases
- term dependencies
- language model
- language modeling
- document retrieval
- n gram
- speech recognition
- retrieval model
- language modelling
- probabilistic model
- test collection
- relevance model
- information retrieval
- ad hoc information retrieval
- query terms
- query expansion
- context sensitive
- document ranking
- pseudo relevance feedback
- retrieval effectiveness
- passage retrieval
- statistical language models
- machine learning
- query specific
- smoothing methods
- feature selection
- document length
- okapi bm