AD-KD: Attribution-Driven Knowledge Distillation for Language Model Compression.
Siyue WuHongzhan ChenXiaojun QuanQifan WangRui WangPublished in: CoRR (2023)
Keyphrases
- language model
- language modeling
- n gram
- probabilistic model
- document retrieval
- speech recognition
- information retrieval
- smoothing methods
- retrieval model
- statistical language models
- translation model
- query expansion
- language modelling
- prior knowledge
- document ranking
- context sensitive
- test collection
- mixture model
- vector space model
- ad hoc information retrieval
- pseudo relevance feedback
- relevance model
- query terms
- statistical models
- dependency structure
- document collections
- document length
- statistical language modeling