基于sbintuitions/modernbert-ja-30m,针对轻小说文本的NER任务优化,尤其是人物名(含片假名/汉字/假名混写、敬称、昵称等)。参数量极小,准确度够用。


标签

  • PERSON - 人名
  • LOCATION - 地点
  • ORGANIZATION - 组织
  • MISC - 物品

Downloads last month
22
Safetensors
Model size
36.8M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for 3tic/Orion-NER-30M-v1

Finetuned
(6)
this model

Collection including 3tic/Orion-NER-30M-v1