Negar Foroutan Eghlidi

À propos
Confidentialité
Mentions légales

Graph Chatbot

Publications associées (4)

Discovering Knowledge-Critical Subnetworks in Pretrained Language Models

Negar Foroutan Eghlidi, Antoine Bosselut, Deniz Bayazit, Zeming Chen

Pretrained language models (LMs) encode implicit representations of knowledge in their parameters. However, localizing these representations and disentangling them from each other remains an open problem. In this work, we investigate whether pretrained lan ...

Association for Computational Linguistics2024

Breaking the Language Barrier: Improving Cross-Lingual Reasoning with Structured Self-Attention

Karl Aberer, Negar Foroutan Eghlidi, Mohammadreza Banaei, Antoine Bosselut

In this work, we study whether multilingual language models (MultiLMs) can transfer logical reasoning abilities to other languages when they are fine-tuned for reasoning in a different language. We evaluate the cross-lingual reasoning abilities of MultiLMs ...

Association for Computational Linguistics2023

Stop Pre-Training: Adapt Visual-Language Models to Unseen Languages

Karl Aberer, Rémi Philippe Lebret, Negar Foroutan Eghlidi

Vision-Language Pre-training (VLP) has advanced the performance of many visionlanguage tasks, such as image-text retrieval, visual entailment, and visual reasoning. The pre-training mostly utilizes lexical databases and image queries in English. Previous w ...

Assoc Computational Linguistics-Acl2023

Discovering Language-neutral Sub-networks in Multilingual Language Models

Karl Aberer, Rémi Philippe Lebret, Negar Foroutan Eghlidi, Mohammadreza Banaei, Antoine Bosselut

Multilingual pre-trained language models transfer remarkably well on cross-lingual downstream tasks. However, the extent to which they learn language-neutral representations (i.e., shared representations that encode similar phenomena across languages), and ...

Association for Computational Linguistics2022

Page 1 sur 1