Zobrazeno 1 - 10
of 38
pro vyhledávání: '"Choenni, Rochelle"'
Stereotypical bias encoded in language models (LMs) poses a threat to safe language technology, yet our understanding of how bias manifests in the parameters of LMs remains incomplete. We introduce local contrastive editing that enables the localizat
Externí odkaz:
http://arxiv.org/abs/2410.17739
Autor:
Choenni, Rochelle, Shutova, Ekaterina
Improving the alignment of Large Language Models (LLMs) with respect to the cultural values that they encode has become an increasingly important topic. In this work, we study whether we can exploit existing knowledge about cultural values at inferen
Externí odkaz:
http://arxiv.org/abs/2408.16482
While multilingual language models (MLMs) have been trained on 100+ languages, they are typically only evaluated across a handful of them due to a lack of available test data in most languages. This is particularly problematic when assessing MLM's po
Externí odkaz:
http://arxiv.org/abs/2406.14267
Texts written in different languages reflect different culturally-dependent beliefs of their writers. Thus, we expect multilingual LMs (MLMs), that are jointly trained on a concatenation of text in multiple languages, to encode different cultural val
Externí odkaz:
http://arxiv.org/abs/2405.12744
Metaphors in natural language are a reflection of fundamental cognitive processes such as analogical reasoning and categorisation, and are deeply rooted in everyday communication. Metaphor understanding is therefore an essential task for large langua
Externí odkaz:
http://arxiv.org/abs/2403.11810
Recent work has proposed explicitly inducing language-wise modularity in multilingual LMs via sparse fine-tuning (SFT) on per-language subnetworks as a means of better guiding cross-lingual sharing. In this work, we investigate (1) the degree to whic
Externí odkaz:
http://arxiv.org/abs/2311.08273
Autor:
Stevenson, Claire E., ter Veen, Mathilde, Choenni, Rochelle, van der Maas, Han L. J., Shutova, Ekaterina
Analogy-making lies at the heart of human cognition. Adults solve analogies such as \textit{Horse belongs to stable like chicken belongs to ...?} by mapping relations (\textit{kept in}) and answering \textit{chicken coop}. In contrast, children often
Externí odkaz:
http://arxiv.org/abs/2310.20384
Autor:
Starace, Giulio, Papakostas, Konstantinos, Choenni, Rochelle, Panagiotopoulos, Apostolos, Rosati, Matteo, Leidinger, Alina, Shutova, Ekaterina
Large Language Models (LLMs) exhibit impressive performance on a range of NLP tasks, due to the general-purpose linguistic knowledge acquired during pretraining. Existing model interpretability research (Tenney et al., 2019) suggests that a linguisti
Externí odkaz:
http://arxiv.org/abs/2310.18696
Multilingual large language models (MLLMs) are jointly trained on data from many different languages such that representation of individual languages can benefit from other languages' data. Impressive performance on zero-shot cross-lingual transfer s
Externí odkaz:
http://arxiv.org/abs/2305.13286
Large multilingual language models typically share their parameters across all languages, which enables cross-lingual task transfer, but learning can also be hindered when training updates from different languages are in conflict. In this paper, we p
Externí odkaz:
http://arxiv.org/abs/2211.00106