Downloads · 30 days
10
29% of all-time downloads
Werea-TR-NER
Werea-co/Werea-TR-NER
Werea-TR-NER is a token classification model from Werea-co. Use it when you need labels on individual words, such as names. The card lists the license as mit.
Kişi (PER), yer (LOC), kurum (ORG) tanıma. Gerçek, insan-etiketli WikiANN-tr verisiyle eğitildi ve aynı benchmark'ın test setinde dürüstçe ölçüldü.
All-time downloads
34
Public
Parameters
110M
880 MB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors440 MB · 100%
At a glance
- Task
- Token Classification
- License
- mit
- Model type
- bert
- Access
- Public
- Created
- Aug 21, 2026
- Updated
- Aug 21, 2026
- SHA
- be268be7
Base models
Datasets
Model card
From the Hugging Face model README
Werea-TR-NER — Türkçe Varlık Tanıma
Kişi (PER), yer (LOC), kurum (ORG) tanıma. Gerçek, insan-etiketli WikiANN-tr verisiyle eğitildi ve aynı benchmark'ın test setinde dürüstçe ölçüldü.
Benchmark (WikiANN-tr test — insan etiketli)
| Metrik | Skor |
|---|---|
| Entity-F1 (genel) | 91.7% |
| PER (kişi) | 94.2% |
| LOC (yer) | 91.4% |
| ORG (kurum) | 89.2% |
Precision 91.7% · Recall 91.6%. dbmdz/bert-base-turkish-cased tabanlı, 3 epoch.
Kullanım
from transformers import pipeline
ner=pipeline("token-classification",model="Werea-co/Werea-TR-NER",aggregation_strategy="simple")
print(ner("Ahmet Yılmaz İstanbul'da Werea şirketinde çalışıyor."))
💜 werea.co
- Task
- Token Classification
- Type
- bert
- License
- mit
- Languages
- tr
- Created
- Aug 21, 2026
- Updated
- Aug 21, 2026