erythropygia/bert-turkish-entityrecognition

This model detects name-entity on Turkish texts. It tags the words Person, Location, Organization.

FYI: You can replace the folder redirect with "erythropygia/bert-turkish-entityrecognition" in the appropriate fields.

The model usage is as follows:

import torch
from transformers import AutoModelForTokenClassification, AutoTokenizer, AutoConfig, pipeline

config_ner = AutoConfig.from_pretrained("erythropygia/bert-turkish-entityrecognition")
model_ner = AutoModelForTokenClassification.from_pretrained("erythropygia/bert-turkish-entityrecognition",config=config_ner)
tokenizer_ner= AutoTokenizer.from_pretrained("erythropygia/bert-turkish-entityrecognition")

device = torch.device('cpu') #or 'cuda'

ner_pipeline = pipeline(
    task='ner',
    model=model_ner,
    tokenizer=tokenizer_ner,
    framework='pt',
    #device=device
)

input = "Bugün Fenerbahçe maçına gideceğim. Maçtan sonra Mehmet Yıldırım'la Bostancıda buluşacağım"

ner_result = ner_pipeline(input)
print(ner_result)