projecte-aina / DEBERTA_CIEL

DeBERTav2 finetuned for Named Entity Recognition.

Table of Contents

Model description

The DEBERTA_CIEL is a Multilingual Named Entity Recognition (NER) model fine-tuned from the [deberta-v3-large] model, a DeBERTa base model pre-trained on English language data collected from publicly available corpora and crawlers.

It has been trained with CEIL, Catalan Entity Identification and Linking , a Catalan-language dataset that contains 9 main types and 52 subtypes on all kinds of short texts, with almost 59K documents.

The model is able to detect and classify entities in at least 3 languages: Spanish, Catalan and English, although no formal evaluation has been carried out yet.

Intended uses and limitations

How to use

from transformers import pipeline

pipe = pipeline("ner", model="projecte-aina/DEBERTA_CIEL")
example = "George Smith Patton fué un general del Ejército de los Estados Unidos en Europa durante la Segunda Guerra Mundial. "

ner_entity_results = pipe(example, aggregation_strategy="simple")
print(ner_entity_results)

[{'entity_group': 'person-other', 'score': 0.55470794, 'word': 'George Smith Patton', 'start': 0, 'end': 19}, {'entity_group': 'organization-other', 'score': 0.8435558, 'word': 'Ejército de los Estados Unidos', 'start': 38, 'end': 69}, {'entity_group': 'location-other', 'score': 0.8351533, 'word': 'Europa', 'start': 72, 'end': 79}, {'entity_group': 'event-attack/terrorism/militaryconflict', 'score': 0.8960169, 'word': 'Segunda Guerra Mundial', 'start': 90, 'end': 113}]

Limitations and bias

At the time of submission, no measures have been taken to estimate the bias embedded in the model. However, we are well aware that our models may be biased since the corpora have been collected using crawling techniques on multiple web sources. We intend to conduct research in these areas in the future, and if completed, this model card will be updated.

Training

We used the NERC dataset in Catalan called Catalan Entity Identification and Linking for training and evaluation.

Evaluation

Accuracy was calculated using the development set, and reflects the non-balanced nature of the dataset.

Major types

*: Cultural Work

Subtypes

Additional information

Author

Language Technologies Unit (LangTech) at the Barcelona Supercomputing Center ( [email protected] )

Contact information

For further information, send an email to [email protected]

Copyright

Copyright (c) 2023 Language Technologies Unit (LangTech) at Barcelona Supercomputing Center

Licensing Information

Apache License, Version 2.0

Funding

This work/research has been promoted and financed by the Government of Catalonia through the Aina project .

Citation information

Disclaimer