TaCL: Improving BERT pre-training with token-aware contrastive learning

Y Su, F Liu, Z Meng, T Lan, L Shu, E Shareghi… - arxiv preprint arxiv …, 2021 - arxiv.org
Masked language models (MLMs) such as BERT and RoBERTa have revolutionized the
field of Natural Language Understanding in the past few years. However, existing pre …

Improving word translation via two-stage contrastive learning

Y Li, F Liu, N Collier, A Korhonen, I Vulić - arxiv preprint arxiv:2203.08307, 2022 - arxiv.org
Word translation or bilingual lexicon induction (BLI) is a key cross-lingual task, aiming to
bridge the lexical gap between different languages. In this work, we propose a robust and …

Rewire-then-probe: A contrastive recipe for probing biomedical knowledge of pre-trained language models

Z Meng, F Liu, E Shareghi, Y Su, C Collins… - arxiv preprint arxiv …, 2021 - arxiv.org
Knowledge probing is crucial for understanding the knowledge transfer mechanism behind
the pre-trained language models (PLMs). Despite the growing progress of probing …

Distilling semantic concept embeddings from contrastively fine-tuned language models

N Li, H Kteich, Z Bouraoui, S Schockaert - Proceedings of the 46th …, 2023 - dl.acm.org
Learning vectors that capture the meaning of concepts remains a fundamental challenge.
Somewhat surprisingly, perhaps, pre-trained language models have thus far only enabled …

Labels Need Prompts Too: Mask Matching for Natural Language Understanding Tasks

B Li, W Ye, Q Wang, W Zhao, S Zhang - Proceedings of the AAAI …, 2024 - ojs.aaai.org
Textual label names (descriptions) are typically semantically rich in many natural language
understanding (NLU) tasks. In this paper, we incorporate the prompting methodology, which …

Automatically suggesting diverse example sentences for l2 japanese learners using pre-trained language models

E Benedetti, A Aizawa, F Boudin - … of the 62nd Annual Meeting of the …, 2024 - hal.science
Providing example sentences that are diverse and aligned with learners {'} proficiency levels
is essential for fostering effective language acquisition. This study examines the use of Pre …

Modelling multi-modal cross-interaction for multi-label few-shot image classification based on local feature selection

K Yan, Z Bouraoui, F Wei, C Xu, P Wang… - ACM Transactions on …, 2025 - dl.acm.org
The aim of multi-label few-shot image classification (ML-FSIC) is to assign semantic labels to
images, in settings where only a small number of training examples are available for each …

Reranking Overgenerated Responses for End-to-End Task-Oriented Dialogue Systems

S Hu, I Vulić, F Liu, A Korhonen - arxiv preprint arxiv:2211.03648, 2022 - arxiv.org
End-to-end (E2E) task-oriented dialogue (ToD) systems are prone to fall into the so-called"
likelihood trap", resulting in generated responses which are dull, repetitive, and often …

Modelling Multi-modal Cross-interaction for ML-FSIC Based on Local Feature Selection

K Yan, Z Bouraoui, F Wei, C Xu, P Wang… - arxiv preprint arxiv …, 2024 - arxiv.org
The aim of multi-label few-shot image classification (ML-FSIC) is to assign semantic labels to
images, in settings where only a small number of training examples are available for each …

Distilling Monolingual and Crosslingual Word-in-Context Representations

Y Arase, T Kajiwara - arxiv preprint arxiv:2409.08719, 2024 - arxiv.org
In this study, we propose a method that distils representations of word meaning in context
from a pre-trained masked language model in both monolingual and crosslingual settings …