Hugging Face
Models
Datasets
Spaces
Posts
Docs
Solutions
Pricing
Log In
Sign Up
Datasets:
PlanTL-GOB-ES
/
cantemist-ner
like
7
Follow
Plan de Tecnologías del Lenguaje - Gobierno de España
58
Tasks:
Token Classification
Modalities:
Text
Sub-tasks:
named-entity-recognition
Languages:
Spanish
Size:
10K - 100K
Tags:
biomedical
clinical
spanish
Libraries:
Datasets
Croissant
License:
cc-by-4.0
Dataset card
Viewer
Files
Files and versions
Community
refs/convert/parquet
cantemist-ner
/
CantemistNer
/
train
/
0000.parquet
Commit History
Update parquet files
980b893
parquet-converter
commited on
Aug 17, 2023