Aitana-ClearLangDetection-R-1.0
Table of Contents
Model Description
This model is fine-tuned from BSC-LT/mRoBERTa for the task of clear language classification in Spanish texts.
It predicts among three categories of linguistic clarity:
- TXT: Original text
- FAC: Facilitated text
- LF: Easy-to-read text
Training Details
Training Data
The dataset consists of Spanish texts annotated with clarity levels:
- Training set: 9,299 instances
- Test set: 3,723 instances
- Extra test set: 465 instances (texts from non-contiguous categories not seen during training, used to evaluate generalization)
The original dataset is available in
gplsi/discriminative_clearsim_es.
Training Hyperparameters
- learning_rate: 2e-5
- num_train_epochs: 2
- per_device_train_batch_size: 8
- per_device_eval_batch_size: 8
- overwrite_output_dir: true
- logging_strategy: steps
- logging_steps: 10
- seed: 852
- fp16: true
Evaluation
Combined test set (4,188 instances)
Confusion Matrix
| Pred FAC | Pred LF | Pred TXT |
|---|
| True FAC | 1373 | 15 | 8 |
| True LF | 29 | 1367 | 0 |
| True TXT | 16 | 1 | 1379 |
| Class | Precision | Recall | F1-score | Support |
|---|
| FAC | 0.9683 | 0.9835 | 0.9758 | 1396 |
| LF | 0.9884 | 0.9792 | 0.9838 | 1396 |
| TXT | 0.9942 | 0.9878 | 0.9910 | 1396 |
- Accuracy: 0.9835
- Macro Avg F1: 0.9836
Test set (3,723 instances)
Confusion Matrix
| Pred FAC | Pred LF | Pred TXT |
|---|
| True FAC | 1220 | 13 | 8 |
| True LF | 28 | 1213 | 0 |
| True TXT | 13 | 1 | 1227 |
| Class | Precision | Recall | F1-score | Support |
|---|
| FAC | 0.9675 | 0.9831 | 0.9752 | 1241 |
| LF | 0.9886 | 0.9774 | 0.9830 | 1241 |
| TXT | 0.9935 | 0.9887 | 0.9911 | 1241 |
- Accuracy: 0.9831
- Macro Avg F1: 0.9831
Extra test set (465 instances)
Confusion Matrix
| Pred FAC | Pred LF | Pred TXT |
|---|
| True FAC | 153 | 2 | 0 |
| True LF | 1 | 154 | 0 |
| True TXT | 3 | 0 | 152 |
| Class | Precision | Recall | F1-score | Support |
|---|
| FAC | 0.9745 | 0.9871 | 0.9808 | 155 |
| LF | 0.9872 | 0.9936 | 0.9903 | 155 |
| TXT | 1.0000 | 0.9806 | 0.9902 | 155 |
- Accuracy: 0.9871
- Macro Avg F1: 0.9871
Technical Specifications
Hardware and Software
For training, we used custom code developed to fine-tuned model using Transformers library.
Compute Infrastructure
This model was trained on NVIDIA DGX systems equipped with A100 GPUs, which enabled efficient large-scale training. For this model, we used one A100 GPU.
Additional Information
Author
The model has been developed by the
Language and Information Systems Group (GPLSI) and the
Centro de Inteligencia Digital (CENID), both part of the
University of Alicante (UA), as part of their ongoing research in
Natural Language Processing (NLP).
Funding
This work is funded by the Ministerio para la Transformación Digital y de la Función Pública, co-financed by the EU – NextGenerationEU, within the framework of the project Desarrollo de Modelos ALIA.
Acknowledgments
We would like to express our gratitude to all individuals and institutions that have contributed to the development of this work.
Special thanks to:
We also acknowledge the financial, technical, and scientific support of the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project Desarrollo de Modelos ALIA, whose contribution has been essential to the completion of this research.
License
Disclaimer
This model has been developed and fine-tuned specifically for classification task. The authors are not responsible for potential errors, misinterpretations, or inappropriate use of the model beyond its intended purpose.
Reference
If you use this model in your research or work, please cite it as follows:
1@misc{gplsi-aitama-clear-r-1.0,
2 author = {Sepúlveda-Torres, Robiert and Martínez-Murillo, Iván and Bonora, Mar and Consuegra-Ayala, Juan Pablo and Galeano, Santiago and Miró Maestre, María and and Grande, Eduardo and Canal-Esteve, Miquel and Estevanell-Valladares, Ernesto L. and Yáñez-Romero, Fabio and Gutierrez, Yoan and Abreu Salas, José Ignacio and Lloret, Elena and Montoyo, Andrés and Muñoz-Guillena and Palomar, Manuel},
3 title = {Aitana-ClearLangDetection-R-1.0: Fine-tuned model for clear language classification (TXT, FAC, LF)},
4 year = {2025},
5 institution = {Language and Information Systems Group (GPLSI) and Centro de Inteligencia Digital (CENID), University of Alicante (UA)},
6 howpublished = {\url{https://huggingface.co/gplsi/Aitana-ClearLangDetection-R-1.0}},
7 note = {Accessed: 2025-10-03}
8}
Copyright © 2026 Language and Information Systems Group (GPLSI) and Centro de Inteligencia Digital (CENID),
University of Alicante (UA).
Distributed under the Apache License 2.0.
