OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocation.
Qidong HuangXiaoyi DongPan ZhangBin WangConghui HeJiaqi WangDahua LinWeiming ZhangNenghai YuPublished in: CoRR (2023)
Keyphrases
- multi modal
- language model
- language modeling
- document retrieval
- information retrieval
- n gram
- probabilistic model
- retrieval model
- multi modality
- query expansion
- language modelling
- test collection
- speech recognition
- audio visual
- cross modal
- smoothing methods
- relevance model
- video search
- translation model
- statistical language models
- image annotation
- high dimensional
- metadata