PanGu-Σ: Towards Trillion Parameter Language Model with Sparse Heterogeneous Computing.
Xiaozhe RenPingyi ZhouXinfan MengXinjing HuangYadao WangWeichao WangPengfei LiXiaoda ZhangAlexander PodolskiyGrigory ArshinovAndrey BoutIrina PiontkovskayaJiansheng WeiXin JiangTeng SuQun LiuJun YaoPublished in: CoRR (2023)
Keyphrases
- language model
- heterogeneous computing
- language modeling
- compute intensive
- language modelling
- probabilistic model
- document retrieval
- test collection
- dirichlet prior
- speech recognition
- information retrieval
- retrieval model
- n gram
- context sensitive
- query expansion
- high dimensional
- grid computing
- mixture model
- ad hoc information retrieval
- parallel computing
- statistical language models
- smoothing methods
- language model for information retrieval
- high performance computing
- pseudo relevance feedback
- translation model
- relevance model
- computer architecture
- query terms
- graphical models
- data analysis
- word clouds
- bayesian networks