NASH: A Simple Unified Framework of Structured Pruning for Accelerating Encoder-Decoder Language Models.
Jongwoo KoSeungjoon ParkYujin KimSumyeong AhnDu-Seong ChangEuijai AhnSe-Young YunPublished in: EMNLP (Findings) (2023)
Keyphrases
- language model
- unified framework
- probabilistic model
- language modeling
- n gram
- document retrieval
- information retrieval
- speech recognition
- low complexity
- query expansion
- language modelling
- retrieval model
- pseudo relevance feedback
- video codec
- relevance model
- test collection
- bit rate
- document ranking
- statistical language models
- decoding process
- translation model
- ad hoc information retrieval
- context sensitive
- term dependencies
- motion estimation
- noisy channel
- wyner ziv video coding
- okapi bm
- feature selection