Feedback Loops With Language Models Drive In-Context Reward Hacking.
Alexander PanErik JonesMeena JagadeesanJacob SteinhardtPublished in: CoRR (2024)
Keyphrases
- language model
- feedback loops
- context sensitive
- language modeling
- n gram
- language modelling
- probabilistic model
- speech recognition
- document retrieval
- information retrieval
- retrieval model
- statistical language models
- language models for information retrieval
- query terms
- reinforcement learning
- smoothing methods
- query expansion
- relevance model
- test collection
- graphical models
- neural network
- pseudo relevance feedback
- feedback loop
- bayesian networks
- ad hoc information retrieval