Model Card: norbert3-large-ner (Fine-Tuned with WikiANN & norne)

Overview

• Model Name: Kushtrim/norbert3-large-ner
• Model Type: Named Entity Recognition (NER)
• Language: Multilingual with focus on Norwegian (Norsk)
• Fine-Tuned with: WikiANN & norne datasets

Description

The Kushtrim/norbert3-large-ner is a pre-trained BERT (Bidirectional Encoder Representations from Transformers) model that has been fine-tuned on ltg/norbert3-large¹ for Named Entity Recognition (NER) in the Norwegian language (Norsk). This model has been fine-tuned using the WikiANN & norne datasets, which includes annotated named entities from various languages, including Norwegian.

Named Entity Recognition is the task of identifying and classifying named entities in text, such as persons, organizations, locations, dates, and more. This model can be used to extract valuable information from Norwegian text with a focus on NER.

Intended Use

The Kushtrim/norbert3-large-ner model, fine-tuned with the WikiANN & norne datasets, is designed for Named Entity Recognition (NER) applications in Norwegian text. It is particularly well-suited for identifying and classifying various types of named entities within Norwegian language content, including the following categories:

• Persons (PER): Recognizing individuals' names, both at the beginning and within their names.
• Organizations (ORG): Identifying organization names, distinguishing between the beginning and inside of these names.
• Locations (LOC): Recognizing location names, including both the beginning and interior of these names.
• Miscellaneous (MISC): Handling miscellaneous entities or categories within text.

Labels

Source of label information: norne

Usage

from transformers import pipeline, AutoModelForTokenClassification, AutoTokenizer
import pandas as pd

tokenizer = AutoTokenizer.from_pretrained("Kushtrim/norbert3-large-ner", trust_remote_code=True)
model = AutoModelForTokenClassification.from_pretrained("Kushtrim/norbert3-large-ner", trust_remote_code=True)

ner = pipeline("ner", model=model, tokenizer=tokenizer, aggregation_strategy='first')

text = "Sett inn tekst her"

results = ner(text)

pd.DataFrame.from_records(results)

Development and attribution

• Original NorBERT 3 architecture and base weights: Language Technology Group, through ltg/norbert3-large
• Norwegian NER fine-tuning, dataset preparation, evaluation, packaging, and release: Kushtrim Visoka
• Model card and ongoing maintenance: Kushtrim Visoka

Acknowledgements

This work builds on ltg/norbert3-large and the work of its original authors. The upstream architecture and base weights remain the work of their respective creators. This repository documents the task-specific adaptation and release work carried out by Kushtrim Visoka and does not claim authorship of the original model. Dataset creators and tool authors referenced elsewhere in this card should also be cited according to their licenses and documentation.

Citation

If you use this fine-tuned model, please cite this repository and the original upstream model.

@misc{visoka_norbert3_large_ner_2026,
  author       = {Visoka, Kushtrim},
  title        = {NorBERT 3 Large Norwegian NER},
  year         = {2026},
  publisher    = {Hugging Face},
  url          = {https://huggingface.co/Kushtrim/norbert3-large-ner},
  note         = {Fine-tuned derivative of ltg/norbert3-large}
}