aajil-labs-sa/arabic-pii-ner

arabic-pii-py — NER models (int8 ONNX)

Two int8-quantized ONNX NER models bundled for apg / arabic-pii-py, a local Arabic/GCC PII gateway. They are redistributed quantizations of existing open models — no new training, same weights. onnxruntime runs them with no PyTorch / Transformers at inference time, which is why apg stays lightweight.

apg uses them as the sole authority for PERSON / ORGANIZATION detection (and LOC → ADDRESS); it has no name/org regex.

Contents

folder	language	entities	from	size
arabic-ner/	Arabic	PERSON, ORGANIZATION, LOCATION, …	hatmimoha/arabic-ner	~105 MB
en-ner/	English	PER, ORG, LOC, MISC	dslim/bert-base-NER	~104 MB

Each folder holds model_quantized.onnx + tokenizer.json + config.json.

Credits & licenses — please cite the originals

Arabic (arabic-ner/)

• Original model: hatmimoha/arabic-ner — Hatim Mohamed.
• Base model: asafaya/bert-base-arabic — Ali Safaya.
• License: not declared upstream. Redistributed here unmodified (only ONNX-exported + quantized) for interoperability; all rights remain with the original authors. If you are an author and would like this copy removed or relicensed, please open an issue on the GitHub repo.

English (en-ner/)

• Original model: dslim/bert-base-NER — David S. Lim — MIT.
• Base: bert-base-cased, fine-tuned on the CoNLL-2003 English NER dataset.

Quantization

Each upstream model was exported to ONNX and dynamically int8-quantized (the standard model_quantized.onnx variant): identical weights, ~4× smaller on disk, fast on CPU via onnxruntime. Nothing was re-trained or fine-tuned.

Integrity — SHA-256 of the .onnx files:

arabic-ner/model_quantized.onnx  8e1c68e5d812835b8d0770e0693764d8d96276971585d77e9b59c97c5217ded9
en-ner/model_quantized.onnx      b324e829f1fad3b897f926d1a1d1372803c6d04546831a3a2ed103652b916adf

Usage

pip install "apg[ner]"

apg auto-downloads these on first use (cached under ~/.cache/huggingface). You can also load them directly with onnxruntime + tokenizers.

Citation

If you use these, please cite the original authors:

@misc{hatmimoha_arabic_ner, title={Arabic Named Entity Recognition Model}, author={Hatim Mohamed}, howpublished={\url{https://huggingface.co/hatmimoha/arabic-ner}}}
@inproceedings{tjong-kim-sang-de-meulder-2003-conll, title={Introduction to the {C}o{NLL}-2003 Shared Task}, author={Tjong Kim Sang, Erik F. and De Meulder, Fien}, booktitle={CoNLL}, year={2003}}