AutoDistil: Few-shot Task-agnostic Neural Architecture Search for Distilling Large Language Models.
Dongkuan XuSubhabrata MukherjeeXiaodong LiuDebadeepta DeyWenhui WangXiang ZhangAhmed Hassan AwadallahJianfeng GaoPublished in: CoRR (2022)
Keyphrases
- language model
- neural architecture
- language modeling
- probabilistic model
- n gram
- query specific
- document retrieval
- query expansion
- document ranking
- speech recognition
- language modelling
- information retrieval
- retrieval model
- relevance model
- statistical language models
- test collection
- search space
- neural network
- activation function
- language models for information retrieval
- feed forward
- user queries
- smoothing methods