metadata

language:
  - ca
license: apache-2.0
tags:
  - catalan
  - named entity recognition
  - ner
  - CaText
  - Catalan Textual Corpus
datasets:
  - projecte-aina/ancora-ca-ner
metrics:
  - f1
model-index:
  - name: roberta-base-ca-v2-cased-ner
    results:
      - task:
          type: token-classification
        dataset:
          type: projecte-aina/ancora-ca-ner
          name: Ancora-ca-NER
        metrics:
          - name: F1
            type: f1
            value: 0.8929
widget:
  - text: Em dic Lluïsa i visc a Santa Maria del Camí.
  - text: L'Aina, la Berta i la Norma són molt amigues.
  - text: El Martí llegeix el Cavall Fort.

Catalan BERTa-v2 (roberta-base-ca-v2) finetuned for Named Entity Recognition.

Model Description
Intended Uses and Limitations
How to Use
Training
- Training Data
- Training Procedure
Evaluation
- Variable and Metrics
- Evaluation Results
Licensing Information
Citation Information
Funding
Contributions

Model description

The roberta-base-ca-v2-cased-ner is a Named Entity Recognition (NER) model for the Catalan language fine-tuned from the roberta-base-ca-v2 model, a RoBERTa base model pre-trained on a medium-size corpus collected from publicly available corpora and crawlers (check the roberta-base-ca-v2 model card for more details).

Intended Uses and Limitations

roberta-base-ca-v2-cased-ner model can be used to recognize Named Entities in the provided text. The model is limited by its training dataset and may not generalize well for all use cases.

How to Use

Here is how to use this model:

from transformers import pipeline
from pprint import pprint

nlp = pipeline("ner", model="projecte-aina/roberta-base-ca-v2-cased-ner")
example = "Em dic Lluïsa i visc a Santa Maria del Camí."

ner_results = nlp(example)
pprint(ner_results)

Training

Training data

We used the NER dataset in Catalan called Ancora-ca-NER for training and evaluation.

Training Procedure

The model was trained with a batch size of 16 and a learning rate of 5e-5 for 5 epochs. We then selected the best checkpoint using the downstream task metric in the corresponding development set and then evaluated it on the test set.

Evaluation

Variable and Metrics

This model was finetuned maximizing F1 score.

Evaluation results

We evaluated the roberta-base-ca-v2-cased-ner on the Ancora-ca-ner test set against standard multilingual and monolingual baselines:

Model	Ancora-ca-ner (F1)
roberta-base-ca-v2-cased-ner	89.29
roberta-base-ca-cased-ner	89.76
mBERT	86.87
XLM-RoBERTa	86.31

For more details, check the fine-tuning and evaluation scripts in the official GitHub repository.

Licensing Information

Apache License, Version 2.0

Citation Information

If you use any of these resources (datasets or models) in your work, please cite our latest paper:

@inproceedings{armengol-estape-etal-2021-multilingual,
    title = "Are Multilingual Models the Best Choice for Moderately Under-resourced Languages? {A} Comprehensive Assessment for {C}atalan",
    author = "Armengol-Estap{\'e}, Jordi  and
      Carrino, Casimiro Pio  and
      Rodriguez-Penagos, Carlos  and
      de Gibert Bonet, Ona  and
      Armentano-Oller, Carme  and
      Gonzalez-Agirre, Aitor  and
      Melero, Maite  and
      Villegas, Marta",
    booktitle = "Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021",
    month = aug,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.findings-acl.437",
    doi = "10.18653/v1/2021.findings-acl.437",
    pages = "4933--4946",
}

Funding

This work was funded by the Departament de la Vicepresidència i de Polítiques Digitals i Territori de la Generalitat de Catalunya within the framework of Projecte AINA.

projecte-aina
/

roberta-base-ca-v2-cased-ner