GLiNER-ID

GLiNER-ID is an Indonesian Named Entity Recognition (NER) model based on GLiNER, fine-tuned from muchad/mdeberta-hybrid-30k.

The model uses the 30k-token hybrid Indonesian–English vocabulary developed through VocabPrune and is designed for flexible, span-based NER in Indonesian text.

The model is a fine-tuned NER checkpoint and can recognize user-defined entity types without requiring a fixed predefined label set.

Model Details

Usage

Install the GLiNER library:

pip install gliner

Load the model and define the entity types you want to extract:

from gliner import GLiNER

model = GLiNER.from_pretrained("muchad/gliner-id")

text = "Penyerang Inter Miami, Lionel Messi, mempersembahkan penghargaan Ballon d'Or 2023 untuk Argentina dan Diego Maradona."

labels = ["Person", "Club", "Award", "Country"]

entities = model.predict_entities(
    text,
    labels,
    threshold=0.5
)

for entity in entities:
    print(entity["text"], "=>", entity["label"])

Resources

• Paper: Engineering, Technology & Applied Science Research
• Paper DOI: 10.48084/etasr.18482
• Base Model: muchad/mdeberta-hybrid-30k
• GLiNER: GitHub – urchade/GLiNER

For the methodology, training data, experimental setup, and detailed evaluation results, please refer to the published paper.

Citation

If you use GLiNER-ID in your research, please cite:

@article{fuadi2026robust,
  author  = {Fuadi, Mukhlish and Wibawa, Adhi Dharma and Sumpeno, Surya},
  title   = {Robust and Efficient Indonesian Span-Based Named Entity
             Recognition via Compact GLiNER: Towards Enhanced
             Retrieval-Augmented Generation},
  journal = {Engineering, Technology and Applied Science Research},
  volume  = {16},
  number  = {3},
  pages   = {36225--36232},
  year    = {2026},
  doi     = {10.48084/etasr.18482}
}