IndicBERTv2 is fine-tuned on Bodo SampurNER dataset for Fine-grained Named Entity Recognition. It is created using the EaMaTa framework, utilizing the Few-NERD dataset.
Read the paper: SampurNER in AAAI-2026
SampurNER Dataset: datasets/prachuryyaIITG/SampurNER
The tagset of Few-NERD is a fine-grained tagset. The fine to coarse level mapping of the tags are as follows:
- Location : GPE, Body of Water, Island, Mountain, Park, Road/Transit, Other
- Person : Actor, Artist/Author, Athlete, Director, Politician, Scholar, Soldier, Other
- ORG : Company, Education, Government, Media, Political Party, Religion, Sports League, Show Organization, Other
- Building : Airport, Hospital, Hotel, Library, Restaurant, Sports Facility, Theater, Other
- Art : Music, Film, Written Art, Broadcast, Painting, Other
- Product : Airplane, Car, Food, Game, Ship, Software, Train, Weapon, Other
- Event : Attack, Election, Natural Disaster, Protest, Sports Event, Other
- Misc : Astronomy, Award, Biology, Chemistry, Currency, Disease, Educational Degree, God, Language, Law, Living Thing, Medical
Model performance:
Precision: 64.17
Recall: 67.14
F1: 65.62
Training Parameters:
Epochs: 6
Optimizer: AdamW
Learning Rate: 5e-5
Weight Decay: 0.01
Batch Size: 64
Contributors
Prachuryya Kaushik
Prof. Ashish Anand
It is part of the AWED-PIPER ecosystem: Paper | Agent for FgNER | Web App for FgNER | Agent for PII Protection | Web App for PII Protection
Sample Usage
The AWED-FiNER agentic tool can be used to interact with expert models trained using this framework. Below is an example:
pip install smolagents gradio_client
from tool import AWEDFiNERTool
tool = AWEDFiNERTool(
space_id="prachuryyaIITG/AWED-FiNER"
)
result = tool.forward(
text="Jude Bellingham joined Real Madrid in 2023.",
language="English"
)
print(result)
Citation
If you use this model, please cite the following papers:
@inproceedings{kaushik2026sampurner,
title={SampurNER: Fine-Grained Named Entity Recognition Dataset for 22 Indian Languages},
volume={40},
url={https://ojs.aaai.org/index.php/AAAI/article/view/40405},
DOI={10.1609/aaai.v40i37.40405},
number={37},
journal={Proceedings of the AAAI Conference on Artificial Intelligence},
author={Kaushik, Prachuryya and Anand, Ashish},
year={2026},
month={Mar.},
pages={31410-31418}
}
@misc{kaushik2026awedpiperagentswebapplications,
title={AWED-PIPER: Agents, Web Applications & Expert Detectors for Personally Identifiable Information Protection & Fine-grained Named Entity Recognition across 36 languages for 6.6 Billion Speakers},
author={Prachuryya Kaushik and Ashish Anand},
year={2026},
eprint={2601.10161},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2601.10161},
}
@inproceedings{ding-etal-2021-nerd,
title = "Few-{NERD}: A Few-shot Named Entity Recognition Dataset",
author = "Ding, Ning and Xu, Guangwei and Chen, Yulin and Wang, Xiaobin and Han, Xu and Xie, Pengjun and Zheng, Haitao and Liu, Zhiyuan",
booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
month = aug,
year = "2021",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.acl-long.248",
doi = "10.18653/v1/2021.acl-long.248",
pages = "3198--3213",
}