University of Indonesia at SemEval-2025 Task 11: Evaluating State-of-the-Art Encoders for Multi-Label Emotion Detection
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912387292987392 |
|---|---|
| author | Hanif, Ikhlasul Akmal Yulianrifat, Eryawan Presma Ongris, Jaycent Gunawan Tjitrahardja, Eduardus Azmi, Muhammad Falensi Naufal, Rahmat Bryan Wicaksono, Alfan Farizki |
| author_facet | Hanif, Ikhlasul Akmal Yulianrifat, Eryawan Presma Ongris, Jaycent Gunawan Tjitrahardja, Eduardus Azmi, Muhammad Falensi Naufal, Rahmat Bryan Wicaksono, Alfan Farizki |
| contents | This paper presents our approach for SemEval 2025 Task 11 Track A, focusing on multilabel emotion classification across 28 languages. We explore two main strategies: fully fine-tuning transformer models and classifier-only training, evaluating different settings such as fine-tuning strategies, model architectures, loss functions, encoders, and classifiers. Our findings suggest that training a classifier on top of prompt-based encoders such as mE5 and BGE yields significantly better results than fully fine-tuning XLMR and mBERT. Our best-performing model on the final leaderboard is an ensemble combining multiple BGE models, where CatBoost serves as the classifier, with different configurations. This ensemble achieves an average F1-macro score of 56.58 across all languages. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_16460 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | University of Indonesia at SemEval-2025 Task 11: Evaluating State-of-the-Art Encoders for Multi-Label Emotion Detection Hanif, Ikhlasul Akmal Yulianrifat, Eryawan Presma Ongris, Jaycent Gunawan Tjitrahardja, Eduardus Azmi, Muhammad Falensi Naufal, Rahmat Bryan Wicaksono, Alfan Farizki Computation and Language Artificial Intelligence I.2.7 This paper presents our approach for SemEval 2025 Task 11 Track A, focusing on multilabel emotion classification across 28 languages. We explore two main strategies: fully fine-tuning transformer models and classifier-only training, evaluating different settings such as fine-tuning strategies, model architectures, loss functions, encoders, and classifiers. Our findings suggest that training a classifier on top of prompt-based encoders such as mE5 and BGE yields significantly better results than fully fine-tuning XLMR and mBERT. Our best-performing model on the final leaderboard is an ensemble combining multiple BGE models, where CatBoost serves as the classifier, with different configurations. This ensemble achieves an average F1-macro score of 56.58 across all languages. |
| title | University of Indonesia at SemEval-2025 Task 11: Evaluating State-of-the-Art Encoders for Multi-Label Emotion Detection |
| topic | Computation and Language Artificial Intelligence I.2.7 |
| url | https://arxiv.org/abs/2505.16460 |