MaCSC: Towards Multimodal-augmented Pre-trained Language Models via Conceptual Prototypes and Self-balancing Calibration.
Xianwei ZhuangZhichang WangXuxin ChengYuxin XieLiming LiangYuexian ZouPublished in: NAACL-HLT (2024)
Keyphrases
- language model
- pre trained
- language modeling
- training data
- n gram
- document retrieval
- speech recognition
- probabilistic model
- language modelling
- information retrieval
- training examples
- query expansion
- retrieval model
- control signals
- test collection
- statistical language models
- multi modal
- relevance model
- audio visual
- training set
- pairwise
- feature selection
- computer vision