Named　entity　recognition　(NER)　is　a　basic　technology　of　Natural　Language　Processing　(NLP).　It　is　mainly　used　to　identify　entities　and　entity　types.　Compared　with　traditional　entity　recognition,　fine-grained　e　ntity　recognition　can　provide　more　precise　semantics.　In　order　to　improve　the　effect　of　fine-grained　C　hinese　N　ER,　w　e　p　ropose　a　m　odel　based　on　RoBERTa-WWM-BiLSTM-CRF　and　compare　it　with　other　high-quality　models.　The　experimental　results　show　that　this　model　has　better　effect　on　the　CLUENER2020　dataset　of　fine-grained　Chinese　NER.　©　2021　IEEE.