SynthVLM: High-Efficiency and High-Quality Synthetic Data for Vision Language Models.
Zheng LiuHao LiangXijie HuangWentao XiongQinhan YuLinzhuang SunChong ChenConghui HeBin CuiWentao ZhangPublished in: CoRR (2024)
Keyphrases
- synthetic data
- high efficiency
- language model
- high quality
- language modeling
- n gram
- high accuracy
- document retrieval
- retrieval model
- probabilistic model
- data sets
- test collection
- information retrieval
- speech recognition
- statistical language models
- query expansion
- real world
- context sensitive
- language modelling
- vector space model
- real image data
- language models for information retrieval
- translation model
- relevance model
- query terms
- ad hoc information retrieval
- speech signal
- pseudo relevance feedback
- smoothing methods
- document length
- spoken term detection
- language model for information retrieval