Mdeberta V3 Base Turkish Ner

🚀 Turkish Named Entity Recognition (NER) Model

This is a fine - tuned Turkish Named Entity Recognition (NER) model based on the multilingual "microsoft/mDeBERTa - v3 - base", offering high - accuracy NER capabilities for Turkish text.

🚀 Quick Start

This model is the fine - tuned version of "microsoft/mDeBERTa - v3 - base" (a multilingual version of DeBERTa V3) using a reviewed version of the well - known Turkish NER dataset (https://github.com/stefan - it/turkish - bert/files/4558187/nerdata.txt).

🔧 Technical Details

Fine - tuning parameters:

task = "ner"
model_checkpoint = "microsoft/mdeberta - v3 - base"
batch_size = 8 
label_list = ['O', 'B - PER', 'I - PER', 'B - ORG', 'I - ORG', 'B - LOC', 'I - LOC']
max_length = 512 
learning_rate = 2e - 5 
num_train_epochs = 2 
weight_decay = 0.01 

💻 Usage Examples

Basic Usage

model = AutoModelForTokenClassification.from_pretrained("akdeniz27/mDeBERTa - v3 - base - turkish - ner")
tokenizer = AutoTokenizer.from_pretrained("akdeniz27/mDeBERTa - v3 - base - turkish - ner")
ner = pipeline('ner', model = model, tokenizer = tokenizer, aggregation_strategy="simple")
ner("<your text here>")

Pls refer "https://huggingface.co/transformers/_modules/transformers/pipelines/token_classification.html" for entity grouping with aggregation_strategy parameter.

📚 Documentation

Reference test results:

• f1: 0.95
• precision: 0.94
• recall: 0.96

📄 License

This project is licensed under the MIT License.