Fine-Grained Human Feedback Gives Better Rewards for Language Model Training.
Zeqiu WuYushi HuWeijia ShiNouha DziriAlane SuhrPrithviraj AmmanabroluNoah A. SmithMari OstendorfHannaneh HajishirziPublished in: NeurIPS (2023)
Keyphrases
- fine grained
- language model
- language modeling
- coarse grained
- n gram
- retrieval model
- speech recognition
- probabilistic model
- document retrieval
- query expansion
- information retrieval
- language modelling
- ad hoc information retrieval
- access control
- mixture model
- smoothing methods
- context sensitive
- vector space model
- test collection
- query terms
- relevance model
- user feedback
- relevance feedback
- translation model
- hidden markov models
- xml documents
- high level