AdaMoE: Token-Adaptive Routing with Null Experts for Mixture-of-Experts Language Models.
Zihao ZengYibo MiaoHongcheng GaoHao ZhangZhijie DengPublished in: CoRR (2024)
Keyphrases
- language model
- language modeling
- mixture model
- expert finding
- document retrieval
- n gram
- information retrieval
- statistical language models
- probabilistic model
- test collection
- speech recognition
- language modelling
- context sensitive
- query expansion
- relevant documents
- retrieval model
- language model for information retrieval
- pseudo relevance feedback
- vector space model
- hidden markov models