模型介绍
内容详情
替代品
模型简介
spaCy的大型韩语处理模型,基于UD Korean Kaist和KLUE数据集训练,支持完整的NLP处理流程,包括分词、词性标注、依存句法分析、命名实体识别等任务。
模型特点
完整的韩语处理流程
提供从分词到命名实体识别的完整NLP处理能力
CPU优化
专门针对CPU环境进行优化,适合资源受限的场景
高质量词向量
包含floret词向量(200000, 300),支持更好的语义理解
多任务处理
同时支持词性标注、依存分析、命名实体识别等多种NLP任务
模型能力
分词
词性标注
依存句法分析
命名实体识别
词形还原
句子边界检测
使用案例
文本处理
韩语文本分析
对韩语新闻、社交媒体内容进行语法分析和结构解析
准确识别句子成分和语法关系
信息提取
从韩语文本中提取命名实体(人名、地名、组织名等)
F1值达到85.19%
语言学研究
韩语语法研究
分析韩语句法结构和词法特征
提供详细的词性标注和依存关系分析
标签:
- spacy
- 标记分类 语言:
- 韩语 许可证:cc-by-sa-4.0 模型索引:
- 名称:ko_core_news_lg
结果:
- 任务:
名称:NER
类型:标记分类
指标:
- 名称:NER精确率 类型:精确率 值:0.8669446273
- 名称:NER召回率 类型:召回率 值:0.837301307
- 名称:NER F值 类型:f_score 值:0.8518651621
- 任务:
名称:TAG
类型:标记分类
指标:
- 名称:TAG (XPOS)准确率 类型:准确率 值:0.8400253175
- 任务:
名称:POS
类型:标记分类
指标:
- 名称:POS (UPOS)准确率 类型:准确率 值:0.9487717077
- 任务:
名称:LEMMA
类型:标记分类
指标:
- 名称:词形还原准确率 类型:准确率 值:0.9009276291
- 任务:
名称:UNLABELED_DEPENDENCIES
类型:标记分类
指标:
- 名称:无标记依存关系评分(UAS) 类型:f_score 值:0.8416620252
- 任务:
名称:LABELED_DEPENDENCIES
类型:标记分类
指标:
- 名称:标记依存关系评分(LAS) 类型:f_score 值:0.8140177338
- 任务:
名称:SENTS
类型:标记分类
指标:
- 名称:句子F值 类型:f_score 值:1.0
- 任务:
名称:NER
类型:标记分类
指标:
详情:https://spacy.io/models/ko#ko_core_news_lg
针对CPU优化的韩语处理流程。组件包括:tok2vec、tagger、morphologizer、parser、lemmatizer(可训练的词形还原器)、senter、ner。
特性 | 描述 |
---|---|
名称 | ko_core_news_lg |
版本 | 3.7.0 |
spaCy | >=3.7.0,<3.8.0 |
默认流程 | tok2vec , tagger , morphologizer , parser , lemmatizer , attribute_ruler , ner |
组件 | tok2vec , tagger , morphologizer , parser , lemmatizer , senter , attribute_ruler , ner |
向量 | floret (200000, 300) |
来源 | UD Korean Kaist v2.8 (Choi, Jinho; Han, Na-Rae; Hwang, Jena; Chun, Jayeol) KLUE v1.1.0 (Sungjoon Park, Jihyung Moon, Sungdong Kim, Won Ik Cho, Jiyoon Han, Jangwon Park, Chisung Song, Junseong Kim, Youngsook Song, Taehwan Oh, Joohong Lee, Juhyun Oh, Sungwon Ryu, Younghoon Jeong, Inkwon Lee, Sangwoo Seo, Dongjun Lee, Hyunwoo Kim, Myeonghwa Lee, Seongbo Jang, Seungwon Do, Sunkyoung Kim, Kyungtae Lim, Jongwon Lee, Kyumin Park, Jamin Shin, Seonghyun Kim, Lucy Park, Alice Oh, Jung-Woo Ha, Kyunghyun Cho) Explosion Vectors (OSCAR 2109 + Wikipedia + OpenSubtitles + WMT News Crawl) (Explosion) |
许可证 | CC BY-SA 4.0 |
作者 | Explosion |
标签方案
查看标签方案(4个组件的2028个标签)
组件 | 标签 |
---|---|
tagger |
_SP , ecs , etm , f , f+f+jcj , f+f+jcs , f+f+jct , f+f+jxt , f+jca , f+jca+jp+ecc , f+jca+jp+ep+ef , f+jca+jxc , f+jca+jxc+jcm , f+jca+jxt , f+jcj , f+jcm , f+jco , f+jcs , f+jct , f+jct+jcm , f+jp+ef , f+jp+ep+ef , f+jp+etm , f+jxc , f+jxt , f+ncn , f+ncn+jcm , f+ncn+jcs , f+ncn+jp+ecc , f+ncn+jxt , f+ncpa+jcm , f+npp+jcs , f+nq , f+xsn , f+xsn+jco , f+xsn+jxt , ii , jca , jca+jcm , jca+jxc , jca+jxt , jcc , jcj , jcm , jco , jcr , jcr+jxc , jcs , jct , jct+jcm , jct+jxt , jp+ecc , jp+ecs , jp+ef , jp+ef+jcr , jp+ef+jcr+jxc , jp+ep+ecs , jp+ep+ef , jp+ep+etm , jp+ep+etn , jp+etm , jp+etn , jp+etn+jco , jp+etn+jxc , jxc , jxc+jca , jxc+jco , jxc+jcs , jxt , mad , mad+jxc , mad+jxt , mag , mag+jca , mag+jcm , mag+jcs , mag+jp+ef+jcr , mag+jxc , mag+jxc+jxc , mag+jxt , mag+xsn , maj , maj+jxc , maj+jxt , mma , mmd , nbn , nbn+jca , nbn+jca+jcj , nbn+jca+jcm , nbn+jca+jp+ef , nbn+jca+jxc , nbn+jca+jxt , nbn+jcc , nbn+jcj , nbn+jcm , nbn+jco , nbn+jcr , nbn+jcs , nbn+jct , nbn+jct+jcm , nbn+jct+jxt , nbn+jp+ecc , nbn+jp+ecs , nbn+jp+ecs+jca , nbn+jp+ecs+jcm , nbn+jp+ecs+jco , nbn+jp+ecs+jxc , nbn+jp+ecs+jxt , nbn+jp+ecx , nbn+jp+ef , nbn+jp+ef+jca , nbn+jp+ef+jco , nbn+jp+ef+jcr , nbn+jp+ef+jcr+jxc , nbn+jp+ef+jcr+jxt , nbn+jp+ef+jp+etm , nbn+jp+ef+jxc , nbn+jp+ef+jxf , nbn+jp+ef+jxt , nbn+jp+ep+ecc , nbn+jp+ep+ecs , nbn+jp+ep+ecs+jxc , nbn+jp+ep+ecx , nbn+jp+ep+ef , nbn+jp+ep+ef+jcr , nbn+jp+ep+ef+jcr+jxc , nbn+jp+ep+ef+jxc , nbn+jp+ep+ef+jxf , nbn+jp+ep+ef+jxt , nbn+jp+ep+ep+etm , nbn+jp+ep+etm , nbn+jp+ep+etn , nbn+jp+ep+etn+jca , nbn+jp+ep+etn+jca+jxc , nbn+jp+ep+etn+jco , nbn+jp+ep+etn+jcs , nbn+jp+ep+etn+jxt , nbn+jp+etm , nbn+jp+etn , nbn+jp+etn+jca , nbn+jp+etn+jca+jxc , nbn+jp+etn+jca+jxt , nbn+jp+etn+jco , nbn+jp+etn+jcs , nbn+jp+etn+jct , nbn+jp+etn+jxc , nbn+jp+etn+jxt , nbn+jxc , nbn+jxc+jca , nbn+jxc+jca+jxc , nbn+jxc+jca+jxt , nbn+jxc+jcc , nbn+jxc+jcm , nbn+jxc+jco , nbn+jxc+jcs , nbn+jxc+jp+ef , nbn+jxc+jxc , nbn+jxc+jxt , nbn+jxt , nbn+nbn , nbn+nbn+jp+ef , nbn+xsm+ecs , nbn+xsm+ef , nbn+xsm+ep+ef , nbn+xsm+ep+ef+jcr , nbn+xsm+etm , nbn+xsn , nbn+xsn+jca , nbn+xsn+jca+jp+ef+jcr , nbn+xsn+jca+jxc , nbn+xsn+jca+jxt , nbn+xsn+jcm , nbn+xsn+jco , nbn+xsn+jcs , nbn+xsn+jct , nbn+xsn+jp+ecc , nbn+xsn+jp+ecs , nbn+xsn+jp+ef , nbn+xsn+jp+ef+jcr , nbn+xsn+jp+ep+ef , nbn+xsn+jxc , nbn+xsn+jxt , nbn+xsv+etm , nbu , nbu+jca , nbu+jca+jxc , nbu+jca+jxt , nbu+jcc , nbu+jcc+jxc , nbu+jcj , nbu+jcm , nbu+jco , nbu+jcs , nbu+jct , nbu+jct+jxc , nbu+jp+ecc , nbu+jp+ecs , nbu+jp+ef , nbu+jp+ef+jcr , nbu+jp+ef+jxc , nbu+jp+ep+ecc , nbu+jp+ep+ecs , nbu+jp+ep+ef , nbu+jp+ep+ef+jcr , nbu+jp+ep+etm , nbu+jp+ep+etn+jco , nbu+jp+etm , nbu+jxc , nbu+jxc+jca , nbu+jxc+jcs , nbu+jxc+jp+ef , nbu+jxc+jp+ep+ef , nbu+jxc+jxt , nbu+jxt , nbu+ncn , nbu+ncn+jca , nbu+ncn+jcm , nbu+xsn , nbu+xsn+jca , nbu+xsn+jca+jxc , nbu+xsn+jca+jxt , nbu+xsn+jcm , nbu+xsn+jco , nbu+xsn+jcs , nbu+xsn+jp+ecs , nbu+xsn+jp+ep+ef , nbu+xsn+jxc , nbu+xsn+jxc+jxt , nbu+xsn+jxt , nbu+xsv+ecc , nbu+xsv+etm , ncn , ncn+f+ncpa+jco , ncn+jca , ncn+jca+jca , ncn+jca+jcc , ncn+jca+jcj , ncn+jca+jcm , ncn+jca+jcs , ncn+jca+jct , ncn+jca+jp+ecc , ncn+jca+jp+ecs , ncn+jca+jp+ef , ncn+jca+jp+ep+ef , ncn+jca+jp+etm , ncn+jca+jp+etn+jxt , ncn+jca+jxc , ncn+jca+jxc+jcc , ncn+jca+jxc+jcm , ncn+jca+jxc+jxc , ncn+jca+jxc+jxt , ncn+jca+jxt , ncn+jcc , ncn+jcc+jxc , ncn+jcj , ncn+jcj+jxt , ncn+jcm , ncn+jco , ncn+jcr , ncn+jcr+jxc , ncn+jcs , ncn+jcs+jxt , ncn+jct , ncn+jct+jcm , ncn+jct+jxc , ncn+jct+jxt , ncn+jcv , ncn+jp+ecc , ncn+jp+ecc+jct , ncn+jp+ecc+jxc , ncn+jp+ecs , ncn+jp+ecs+jcm , ncn+jp+ecs+jco , ncn+jp+ecs+jxc , ncn+jp+ecs+jxt , ncn+jp+ecx , ncn+jp+ef , ncn+jp+ef+jca , ncn+jp+ef+jcm , ncn+jp+ef+jco , ncn+jp+ef+jcr , ncn+jp+ef+jcr+jxc , ncn+jp+ef+jcr+jxt , ncn+jp+ef+jp+etm , ncn+jp+ef+jxc , ncn+jp+ef+jxf , ncn+jp+ef+jxt , ncn+jp+ep+ecc , ncn+jp+ep+ecs , ncn+jp+ep+ecs+jxc , ncn+jp+ep+ecx , ncn+jp+ep+ef , ncn+jp+ep+ef+jcr , ncn+jp+ep+ef+jcr+jxc , ncn+jp+ep+ef+jxc , ncn+jp+ep+ef+jxf , ncn+jp+ep+ef+jxt , ncn+jp+ep+ep+etm , ncn+jp+ep+etm , ncn+jp+ep+etn , ncn+jp+ep+etn+jca , ncn+jp+ep+etn+jca+jxc , ncn+jp+ep+etn+jco , ncn+jp+ep+etn+jcs , ncn+jp+ep+etn+jxt , ncn+jp+etm , ncn+jp+etn , ncn+jp+etn+jca , ncn+jp+etn+jca+jxc , ncn+jp+etn+jca+jxt , ncn+jp+etn+jco , ncn+jp+etn+jcs , ncn+jp+etn+jct , ncn+jp+etn+jxc , ncn+jp+etn+jxt , ncn+jxc , ncn+jxc+jca , ncn+jxc+jca+jxc , ncn+jxc+jca+jxt , ncn+jxc+jcc , ncn+jxc+jcm , ncn+jxc+jco , ncn+jxc+jcs , ncn+jxc+jct+jxt , ncn+jxc+jp+ef , ncn+jxc+jp+ef+jcr , ncn+jxc+jp+ep+ecs , ncn+jxc+jp+ep+ef , ncn+jxc+jp+etm , ncn+jxc+jxc , ncn+jxc+jxt , ncn+jxt , ncn+jxt+jcm , ncn+jxt+jxc , ncn+nbn , ncn+nbn+jca , ncn+nbn+jcm , ncn+nbn+jcs , ncn+nbn+jp+ecc , ncn+nbn+jp+ep+ef , ncn+nbn+jxc , ncn+nbn+jxt , ncn+nbu , ncn+nbu+jca , ncn+nbu+jcm , ncn+nbu+jco , ncn+nbu+jp+ef , ncn+nbu+jxc , ncn+nbu+ncn , ncn+ncn , ncn+ncn+jca , ncn+ncn+jca+jcc , ncn+ncn+jca+jcm , ncn+ncn+jca+jxc , ncn+ncn+jca+jxc+jcm , ncn+ncn+jca+jxc+jxc , ncn+ncn+jca+jxt , ncn+ncn+jcc , ncn+ncn+jcj , ncn+ncn+jcm , ncn+ncn+jco , ncn+ncn+jcr , ncn+ncn+jcs , ncn+ncn+jct , ncn+ncn+jct+jcm , ncn+ncn+jct+jxc , ncn+ncn+jct+jxt , ncn+ncn+jp+ecc , ncn+ncn+jp+ecs , ncn+ncn+jp+ef , ncn+ncn+jp+ef+jcm , ncn+ncn+jp+ef+jcr , ncn+ncn+jp+ef+jcs , ncn+ncn+jp+ep+ecc , ncn+ncn+jp+ep+ecs , ncn+ncn+jp+ep+ef , ncn+ncn+jp+ep+ef+jcr , ncn+ncn+jp+ep+ep+etm , ncn+ncn+jp+ep+etm , ncn+ncn+jp+ep+etn , ncn+ncn+jp+etm , ncn+ncn+jp+etn , ncn+ncn+jp+etn+jca , ncn+ncn+jp+etn+jco , ncn+ncn+jp+etn+jxc , ncn+ncn+jxc , ncn+ncn+jxc+jca , ncn+ncn+jxc+jcc , ncn+ncn+jxc+jcm , ncn+ncn+jxc+jco , ncn+ncn+jxc+jcs , ncn+ncn+jxc+jxc , ncn+ncn+jxt , ncn+ncn+nbn , ncn+ncn+ncn , ncn+ncn+ncn+jca , ncn+ncn+ncn+jca+jcm , ncn+ncn+ncn+jca+jxt , ncn+ncn+ncn+jcj , ncn+ncn+ncn+jcm , ncn+ncn+ncn+jco , ncn+ncn+ncn+jcs , ncn+ncn+ncn+jct+jxt , ncn+ncn+ncn+jp+etn+jxc , ncn+ncn+ncn+jxt , ncn+ncn+ncn+ncn+jca , ncn+ncn+ncn+ncn+jca+jxt , ncn+ncn+ncn+ncn+jco , ncn+ncn+ncn+xsn+jp+etm , ncn+ncn+ncpa , ncn+ncn+ncpa+jca , ncn+ncn+ncpa+jcm , ncn+ncn+ncpa+jco , ncn+ncn+ncpa+jcs , ncn+ncn+ncpa+jxc , ncn+ncn+ncpa+jxt , ncn+ncn+ncpa+ncn , ncn+ncn+ncpa+ncn+jca , ncn+ncn+ncpa+ncn+jcj , ncn+ncn+ncpa+ncn+jcm , ncn+ncn+ncpa+ncn+jxt , ncn+ncn+xsn , ncn+ncn+xsn+jca , ncn+ncn+xsn+jca+jxt , ncn+ncn+xsn+jcj , ncn+ncn+xsn+jcm , ncn+ncn+xsn+jco , ncn+ncn+xsn+jcs , ncn+ncn+xsn+jct , ncn+ncn+xsn+jp+ecs , ncn+ncn+xsn+jp+ep+ef , ncn+ncn+xsn+jp+etm , ncn+ncn+xsn+jxc , ncn+ncn+xsn+jxc+jcs , ncn+ncn+xsn+jxt , ncn+ncn+xsv+ecc , ncn+ncn+xsv+etm , ncn+ncpa , ncn+ncpa+jca , ncn+ncpa+jca+jcm , ncn+ncpa+jca+jxc , ncn+ncpa+jca+jxt , ncn+ncpa+jcc , ncn+ncpa+jcj , ncn+ncpa+jcm , ncn+ncpa+jco , ncn+ncpa+jcr , ncn+ncpa+jcs , ncn+ncpa+jct , ncn+ncpa+jct+jcm , ncn+ncpa+jct+jxt , ncn+ncpa+jp+ecc , ncn+ncpa+jp+ecc+jxc , ncn+ncpa+jp+ecs , ncn+ncpa+jp+ecs+jxc , ncn+ncpa+jp+ef , ncn+ncpa+jp+ef+jcr , ncn+ncpa+jp+ef+jcr+jxc , ncn+ncpa+jp+ep+ef , ncn+ncpa+jp+ep+etm , ncn+ncpa+jp+ep+etn , ncn+ncpa+jp+etm , ncn+ncpa+jxc , ncn+ncpa+jxc+jca+jxc , ncn+ncpa+jxc+jco , ncn+ncpa+jxc+jcs , ncn+ncpa+jxt , ncn+ncpa+nbn+jcs , ncn+ncpa+ncn , ncn+ncpa+ncn+jca , ncn+ncpa+ncn+jca+jcm , ncn+ncpa+ncn+jca+jxc , ncn+ncpa+ncn+jca+jxt , ncn+ncpa+ncn+jcj , ncn+ncpa+ncn+jcm , ncn+ncpa+ncn+jco , ncn+ncpa+ncn+jcs , ncn+ncpa+ncn+jct , ncn+ncpa+ncn+jct+jcm , ncn+ncpa+ncn+jp+ef+jcr , ncn+ncpa+ncn+jp+ep+etm , ncn+ncpa+ncn+jxc , ncn+ncpa+ncn+jxt , ncn+ncpa+ncn+xsn+jcm , ncn+ncpa+ncn+xsn+jxt , ncn+ncpa+ncpa , ncn+ncpa+ncpa+jca , ncn+ncpa+ncpa+jcj , ncn+ncpa+ncpa+jcm , ncn+ncpa+ncpa+jco , ncn+ncpa+ncpa+jcs , ncn+ncpa+ncpa+jp+ep+ef , ncn+ncpa+ncpa+jxt , ncn+ncpa+ncpa+ncn , ncn+ncpa+xsn , ncn+ncpa+xsn+jcm , ncn+ncpa+xsn+jco , ncn+ncpa+xsn+jcs , ncn+ncpa+xsn+jp+ecc , ncn+ncpa+xsn+jp+etm , ncn+ncpa+xsn+jxt , ncn+ncpa+xsv+ecc , ncn+ncpa+xsv+ecs , ncn+ncpa+xsv+ecx , ncn+ncpa+xsv+ecx+px+etm , ncn+ncpa+xsv+ef , ncn+ncpa+xsv+ef+jcm , ncn+ncpa+xsv+ef+jcr , ncn+ncpa+xsv+etm , (截断:完整列表见流程元数据) |
morphologizer |
POS=CCONJ , POS=ADV , POS=SCONJ , POS=DET , POS=NOUN , POS=VERB , POS=ADJ , POS=PUNCT , POS=SPACE , POS=AUX , POS=PRON , POS=PROPN , POS=NUM , POS=INTJ , POS=PART , POS=X , POS=ADP , POS=SYM |
parser |
ROOT , acl , advcl , advmod , amod , appos , aux , case , cc , ccomp , compound , conj , cop , csubj , dep , det , dislocated , fixed , flat , iobj , mark , nmod , nsubj , nummod , obj , obl , punct , xcomp |
ner |
DT , LC , OG , PS , QT , TI |
准确率
类型 | 得分 |
---|---|
TOKEN_ACC |
100.00 |
TOKEN_P |
100.00 |
TOKEN_R |
100.00 |
TOKEN_F |
100.00 |
TAG_ACC |
84.00 |
POS_ACC |
94.88 |
SENTS_P |
100.00 |
SENTS_R |
100.00 |
SENTS_F |
100.00 |
DEP_UAS |
84.17 |
DEP_LAS |
81.40 |
LEMMA_ACC |
90.09 |
ENTS_P |
86.69 |
ENTS_R |
83.73 |
ENTS_F |
85.19 |
Indonesian Roberta Base Posp Tagger
MIT
这是一个基于印尼语RoBERTa模型微调的词性标注模型,在indonlu数据集上训练,用于印尼语文本的词性标注任务。
序列标注
Transformers

其他
I
w11wo
2.2M
7
Bert Base NER
MIT
基于BERT微调的命名实体识别模型,可识别四类实体:地点(LOC)、组织机构(ORG)、人名(PER)和杂项(MISC)
序列标注
英语
B
dslim
1.8M
592
Deid Roberta I2b2
MIT
该模型是基于RoBERTa微调的序列标注模型,用于识别和移除医疗记录中的受保护健康信息(PHI/PII)。
序列标注
Transformers

支持多种语言
D
obi
1.1M
33
Ner English Fast
Flair自带的英文快速4类命名实体识别模型,基于Flair嵌入和LSTM-CRF架构,在CoNLL-03数据集上达到92.92的F1分数。
序列标注
PyTorch
英语
N
flair
978.01k
24
French Camembert Postag Model
基于Camembert-base的法语词性标注模型,使用free-french-treebank数据集训练
序列标注
Transformers

法语
F
gilf
950.03k
9
Xlm Roberta Large Ner Spanish
基于XLM-Roberta-large架构微调的西班牙语命名实体识别模型,在CoNLL-2002数据集上表现优异。
序列标注
Transformers

西班牙语
X
MMG
767.35k
29
Nusabert Ner V1.3
MIT
基于NusaBert-v1.3在印尼语NER任务上微调的命名实体识别模型
序列标注
Transformers

其他
N
cahya
759.09k
3
Ner English Large
Flair框架内置的英文4类大型NER模型,基于文档级XLM-R嵌入和FLERT技术,在CoNLL-03数据集上F1分数达94.36。
序列标注
PyTorch
英语
N
flair
749.04k
44
Punctuate All
MIT
基于xlm-roberta-base微调的多语言标点符号预测模型,支持12种欧洲语言的标点符号自动补全
序列标注
Transformers

P
kredor
728.70k
20
Xlm Roberta Ner Japanese
MIT
基于xlm-roberta-base微调的日语命名实体识别模型
序列标注
Transformers

支持多种语言
X
tsmatz
630.71k
25
精选推荐AI模型
Llama 3 Typhoon V1.5x 8b Instruct
专为泰语设计的80亿参数指令模型,性能媲美GPT-3.5-turbo,优化了应用场景、检索增强生成、受限生成和推理任务
大型语言模型
Transformers

支持多种语言
L
scb10x
3,269
16
Cadet Tiny
Openrail
Cadet-Tiny是一个基于SODA数据集训练的超小型对话模型,专为边缘设备推理设计,体积仅为Cosmo-3B模型的2%左右。
对话系统
Transformers

英语
C
ToddGoldfarb
2,691
6
Roberta Base Chinese Extractive Qa
基于RoBERTa架构的中文抽取式问答模型,适用于从给定文本中提取答案的任务。
问答系统
中文
R
uer
2,694
98
AIbase是一个专注于MCP服务的平台,为AI开发者提供高质量的模型上下文协议服务,助力AI应用开发。
简体中文