camelbert-ner-arabic

This model is a fine-tuned version of CAMeL-Lab/bert-base-arabic-camelbert-mix on unimelb-nlp/wikiann dataset. It achieves the following results on the evaluation set:

• Loss: 0.2111
• Precision: 0.8884
• Recall: 0.8955
• F1: 0.8919
• Accuracy: 0.9513

Model description

• Base Model: CAMeL-Lab/bert-base-arabic-camelbert-mix
• Task: Named Entity Recognition (NER)
• Language: Arabic
• Training Data: WikiAnn dataset for Arabic

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1892	1.0	1250	0.2003	0.8653	0.8677	0.8665	0.9430
0.123	2.0	2500	0.1912	0.8802	0.8826	0.8814	0.9493
0.0809	3.0	3750	0.1942	0.8928	0.8969	0.8948	0.9539

Usage

from transformers import pipeline

# Load the NER pipeline
nlp = pipeline("ner", model="Tevfik-istanbullu/camelbert-ner-arabic")

# Example text
text = "يعمل محمد في شركة جوجل في دبي"
results = nlp(text)
print(results)

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3