🚀 Model Card for gliner_small_news-v2.1
This model is a fine - tuned version of GLiNER. Its aim is to enhance accuracy across a wide range of topics, especially in long - context news entity extraction. As shown in the table below, these fine - tuning efforts improved the zero - shot accuracy of the base GLiNER model by up to 7.5% across 18 benchmark datasets.

The underlying dataset, AskNews - NER - v0, was designed to diversify global perspectives by ensuring country/language/topic/temporal diversity. All the data used for fine - tuning this model was synthetically generated. WizardLM 13B v1.2 was used for translating and summarizing open - web news articles, while Llama3 70b instruct was used for entity extraction. Both the diversification and fine - tuning methods are presented in our paper on ArXiv.
🚀 Quick Start
from gliner import GLiNER
model = GLiNER.from_pretrained("EmergentMethods/gliner_small_news-v2.1")
text = """
The Chihuahua State Public Security Secretariat (SSPE) arrested 35 - year - old Salomón C. T. in Ciudad Juárez, found in possession of a stolen vehicle, a white GMC Yukon, which was reported stolen in the city's streets. The arrest was made by intelligence and police analysis personnel during an investigation in the border city. The arrest is related to a previous detention on February 6, which involved armed men in a private vehicle. The detainee and the vehicle were turned over to the Chihuahua State Attorney General's Office for further investigation into the case.
"""
labels = ["person", "location", "date", "event", "facility", "vehicle", "number", "organization"]
entities = model.predict_entities(text, labels)
for entity in entities:
print(entity["text"], "=>", entity["label"])
Output:
Chihuahua State Public Security Secretariat => organization
SSPE => organization
35 - year - old => number
Salomón C. T. => person
Ciudad Juárez => location
GMC Yukon => vehicle
February 6 => date
Chihuahua State Attorney General's Office => organization
✨ Features
- Enhanced Accuracy: Improves the zero - shot accuracy of the base GLiNER model by up to 7.5% across 18 benchmark datasets, especially in long - context news entity extraction.
- Diversified Data: Uses a synthetic dataset engineered for country/language/topic/temporal diversity.
- Compact and High - Throughput: Can be used for high - throughput production usecases due to its compact size.
📚 Documentation
Model Details
Model Description
The synthetic data for this news fine - tuning was sourced from the AskNews API. We ensured diversity across country, language, topic, and time.
Countries:

Entity types:

Topics:

Property |
Details |
Developed by |
Emergent Methods |
Funded by |
Emergent Methods |
Shared by |
Emergent Methods |
Model Type |
microsoft/deberta |
Language(s) (NLP) |
English (en) (English texts and translations from Spanish (es), Portuguese (pt), German (de), Russian (ru), French (fr), Arabic (ar), Italian (it), Ukrainian (uk), Norwegian (no), Swedish (sv), Danish (da)) |
License |
Apache 2.0 |
Finetuned from model |
GLiNER |
Model Sources [optional]
- Repository: To be added
- Paper: To be added
- Demo: To be added
Uses
Direct Use
As the name implies, this model is designed for generalist entity extraction. Despite being fine - tuned on news data, it improves accuracy across 18 benchmark datasets by up to 7.5%. This indicates that the broad and diversified underlying dataset helps it recognize and extract more entity types. The model is surprisingly compact and can be used for high - throughput production scenarios. Currently, AskNews is using this fine - tuned model for entity extraction in their system.
Bias, Risks, and Limitations
Although the dataset aims to reduce bias and improve diversity, it is still biased towards western languages and countries. This limitation stems from the capabilities of Llama2 for translation and summary generation. Additionally, any bias present in the Llama2 training data will also be in this dataset since Llama2 was used to summarize open - web articles. Moreover, any biases in Llama3 will be reflected in the current dataset as Llama3 was used to extract entities from the summaries.

Training Details
The training dataset is AskNews - NER - v0. Other training details can be found in the companion paper.
Environmental Impact
Citation
BibTeX: To be added
APA: To be added
Model Authors
Elin Törnquist, Emergent Methods elin at emergentmethods.ai
Robert Caulk, Emergent Methods rob at emergentmethods.ai
Model Contact
Elin Törnquist, Emergent Methods elin at emergentmethods.ai
Robert Caulk, Emergent Methods rob at emergentmethods.ai