BERTastic at SemEval-2023 Task 3: Fine-Tuning Pretrained Multilingual Transformers Does Order Matter?

Tarek Mahmoud, Preslav Nakov


Abstract
The naive approach for fine-tuning pretrained deep learning models on downstream tasks involves feeding them mini-batches of randomly sampled data. In this paper, we propose a more elaborate method for fine-tuning Pretrained Multilingual Transformers (PMTs) on multilingual data. Inspired by the success of curriculum learning approaches, we investigate the significance of fine-tuning PMTs on multilingual data in a sequential fashion language by language. Unlike the curriculum learning paradigm where the model is presented with increasingly complex examples, we do not adopt a notion of “easy” and “hard” samples. Instead, our experiments draw insight from psychological findings on how the human brain processes new information and the persistence of newly learned concepts. We perform our experiments on a challenging news-framing dataset that contains texts in six languages. Our proposed method outperforms the naïve approach by achieving improvements of 2.57\% in terms of F1 score. Even when we supplement the naïve approach with recency fine-tuning, we still achieve an improvement of 1.34\% with a 3.63\%$ convergence speed-up. Moreover, we are the first to observe an interesting pattern in which deep learning models exhibit a human-like primacy-recency effect.
Anthology ID:
2023.semeval-1.7
Volume:
Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023)
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Atul Kr. Ojha, A. Seza Doğruöz, Giovanni Da San Martino, Harish Tayyar Madabushi, Ritesh Kumar, Elisa Sartori
Venue:
SemEval
SIG:
SIGLEX
Publisher:
Association for Computational Linguistics
Note:
Pages:
58–63
Language:
URL:
https://aclanthology.org/2023.semeval-1.7
DOI:
10.18653/v1/2023.semeval-1.7
Bibkey:
Cite (ACL):
Tarek Mahmoud and Preslav Nakov. 2023. BERTastic at SemEval-2023 Task 3: Fine-Tuning Pretrained Multilingual Transformers Does Order Matter?. In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 58–63, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
BERTastic at SemEval-2023 Task 3: Fine-Tuning Pretrained Multilingual Transformers Does Order Matter? (Mahmoud & Nakov, SemEval 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.semeval-1.7.pdf
Video:
 https://aclanthology.org/2023.semeval-1.7.mp4