Zobrazeno 1 - 10
of 14 980
pro vyhledávání: '"Tsvetkov, A."'
Autor:
Wen, Bingbing, Yao, Jihan, Feng, Shangbin, Xu, Chenjun, Tsvetkov, Yulia, Howe, Bill, Wang, Lucy Lu
Abstention, the refusal of large language models (LLMs) to provide an answer, is increasingly recognized for its potential to mitigate hallucinations and enhance safety in LLM systems. In this survey, we introduce a framework to examine abstention fr
Externí odkaz:
http://arxiv.org/abs/2407.18418
Autor:
Ahia, Orevaoghene, Kumar, Sachin, Gonen, Hila, Hoffman, Valentin, Limisiewicz, Tomasz, Tsvetkov, Yulia, Smith, Noah A.
In multilingual settings, non-Latin scripts and low-resource languages are usually disadvantaged in terms of language models' utility, efficiency, and cost. Specifically, previous studies have reported multiple modeling biases that the current tokeni
Externí odkaz:
http://arxiv.org/abs/2407.08818
Autor:
Brahman, Faeze, Kumar, Sachin, Balachandran, Vidhisha, Dasigi, Pradeep, Pyatkin, Valentina, Ravichander, Abhilasha, Wiegreffe, Sarah, Dziri, Nouha, Chandu, Khyathi, Hessel, Jack, Tsvetkov, Yulia, Smith, Noah A., Choi, Yejin, Hajishirzi, Hannaneh
Chat-based language models are designed to be helpful, yet they should not comply with every user request. While most existing work primarily focuses on refusal of "unsafe" queries, we posit that the scope of noncompliance should be broadened. We int
Externí odkaz:
http://arxiv.org/abs/2407.12043
Autor:
Park, Chan Young, Li, Shuyue Stella, Jung, Hayoung, Volkova, Svitlana, Mitra, Tanushree, Jurgens, David, Tsvetkov, Yulia
This study introduces ValueScope, a framework leveraging language models to quantify social norms and values within online communities, grounded in social science perspectives on normative structures. We employ ValueScope to dissect and analyze lingu
Externí odkaz:
http://arxiv.org/abs/2407.02472
Autor:
Ahia, Orevaoghene, Aremu, Anuoluwapo, Abagyan, Diana, Gonen, Hila, Adelani, David Ifeoluwa, Abolade, Daud, Smith, Noah A., Tsvetkov, Yulia
Yor\`ub\'a an African language with roughly 47 million speakers encompasses a continuum with several dialects. Recent efforts to develop NLP technologies for African languages have focused on their standard dialects, resulting in disparities for dial
Externí odkaz:
http://arxiv.org/abs/2406.19564
Autor:
Zhang, Yizhuo, Wang, Heng, Feng, Shangbin, Tan, Zhaoxuan, Han, Xiaochuang, He, Tianxing, Tsvetkov, Yulia
Large language models (LLMs) demonstrate great potential for problems with implicit graphical structures, while recent works seek to enhance the graph reasoning capabilities of LLMs through specialized instruction tuning. The resulting 'graph LLMs' a
Externí odkaz:
http://arxiv.org/abs/2406.15992
Autor:
Feng, Shangbin, Sorensen, Taylor, Liu, Yuhan, Fisher, Jillian, Park, Chan Young, Choi, Yejin, Tsvetkov, Yulia
While existing alignment paradigms have been integral in developing large language models (LLMs), LLMs often learn an averaged human preference and struggle to model diverse preferences across cultures, demographics, and communities. We propose Modul
Externí odkaz:
http://arxiv.org/abs/2406.15951
Autor:
Feng, Shangbin, Shi, Weijia, Wang, Yike, Ding, Wenxuan, Ahia, Orevaoghene, Li, Shuyue Stella, Balachandran, Vidhisha, Sitaram, Sunayana, Tsvetkov, Yulia
Multilingual LLMs often have knowledge disparities across languages, with larger gaps in under-resourced languages. Teaching LLMs to abstain in the face of knowledge gaps is thus a promising strategy to mitigate hallucinations in multilingual setting
Externí odkaz:
http://arxiv.org/abs/2406.15948
Autor:
Li, Shuyue Stella, Balachandran, Vidhisha, Feng, Shangbin, Ilgen, Jonathan, Pierson, Emma, Koh, Pang Wei, Tsvetkov, Yulia
In high-stakes domains like clinical reasoning, AI assistants powered by large language models (LLMs) are yet to be reliable and safe. We identify a key obstacle towards reliability: existing LLMs are trained to answer any question, even with incompl
Externí odkaz:
http://arxiv.org/abs/2406.00922
Autor:
Ahuja, Kabir, Balachandran, Vidhisha, Panwar, Madhur, He, Tianxing, Smith, Noah A., Goyal, Navin, Tsvetkov, Yulia
Transformers trained on natural language data have been shown to learn its hierarchical structure and generalize to sentences with unseen syntactic structures without explicitly encoding any structural bias. In this work, we investigate sources of in
Externí odkaz:
http://arxiv.org/abs/2404.16367