"My Answer is C": First-Token Probabilities Do Not Match Text Answers in Instruction-Tuned Language Models.
Xinpeng WangBolei MaChengzhi HuLeon Weber-GenzelPaul RöttgerFrauke KreuterDirk HovyBarbara PlankPublished in: CoRR (2024)
Keyphrases
- language model
- information retrieval
- language modeling
- document retrieval
- question answer pairs
- document level
- n gram
- question answering systems
- probabilistic model
- query expansion
- correct answers
- text retrieval
- speech recognition
- language modelling
- retrieval model
- test collection
- context sensitive
- question answering
- query terms
- multiword
- structured queries
- statistical language models
- document ranking
- ad hoc information retrieval
- text mining
- language models for information retrieval
- document analysis
- vector space model
- pseudo relevance feedback
- keywords
- sentence level
- passage retrieval
- retrieval effectiveness
- query specific
- text documents
- semantic information
- okapi bm
- document collections
- web search