Edit model card

BERT-DE-NER

What is it?

This is a German BERT model fine-tuned for named entity recognition.

Base model & training

This model is based on bert-base-german-dbmdz-cased and has been fine-tuned for NER on the training data from GermEval2014.

Model results

The results on the test data from GermEval2014 are (entities only):

Precision Recall F1-Score
0.817 0.842 0.829

How to use

>>> from transformers import pipeline

>>> classifier = pipeline('ner', model="fhswf/bert_de_ner")
>>> classifier('Von der Organisation „medico international“ hieß es, die EU entziehe sich seit vielen Jahren der Verantwortung für die Menschen an ihren Außengrenzen.')

[{'word': 'med', 'score': 0.9996621608734131, 'entity': 'B-ORG', 'index': 6},
 {'word': '##ico', 'score': 0.9995362162590027, 'entity': 'I-ORG', 'index': 7},
 {'word': 'international',
  'score': 0.9996932744979858,
  'entity': 'I-ORG',
  'index': 8},
 {'word': 'eu', 'score': 0.9997008442878723, 'entity': 'B-ORG', 'index': 14}]
Downloads last month
137
Safetensors
Model size
110M params
Tensor type
I64
·
F32
·
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Dataset used to train fhswf/bert_de_ner

Space using fhswf/bert_de_ner 1