The paper investigates why large language models sometimes hallucinate when asked about facts in a language different from the one in which the facts were learned. By training small Transformer models on synthetic multilingual datasets, the authors show that the degree of correlation between facts and their learning language (informativeness) and the ease of language identification (extractability) determine whether models develop unified or separate representations across languages. Unified representations enable cross‑lingual fact transfer, while separate representations do not. The study proposes a unifying perspective on cross‑lingual transfer and suggests training methods to promote representational unification.
By Carter Blum, Katja Filippova, Ann Yuan, Asma Ghandeharioun, Julian Zimmert, Fred Zhang, Jessica Hoffmann, Tal Linzen, Martin Wattenberg, Lucas Dixon, Mor Geva
arXiv:2311. 17633v2 Announce Type: replace-cross Abstract: Transformers have dominated empirical machine learning models of natural language processing.
By Tong Xiao, Jingbo Zhu
We’ve trained a large-scale unsupervised language model which generates coherent paragraphs of text, achieves state-of-the-art performance on many language modeling benchmarks, and performs rudimentary reading comprehension, machine translation, question answering, and summarization—all without task-specific training.
arXiv:2503. 05500v3 Announce Type: replace-cross Abstract: General-purpose multilingual vector representations, used in retrieval, regression and classification, are traditionally obtained from bidirectional encoder models.
By Nicolas Boizard, Hippolyte Gisserot-Boukhlef, Duarte M. Alves, Andr\'e Martins, Ayoub Hammal, Caio Corro, C\'eline Hudelot, Emmanuel Malherbe, Etienne Malaboeuf, Fanny Jourdan, Gabriel Hautreux, Jo\~ao Alves, Kevin El Haddad, Manuel Faysse, Maxime Peyrard, Nuno M. Guerreiro, Patrick Fernandes, Ricardo Rei, Pierre Colombo
Multilingual Large Language Models (LLMs) traditionally rely on a single vocabulary shared by all supported languages, which can lead to uneven compression across them. Moreover, their large embedding...
The paper proposes a modular tokenizer framework for multilingual large language models, allowing the creation of language‑specific subtokenizers that match monolingual compression quality. It introduces a pretraining strategy that samples these subtokenizers to limit predictions to relevant vocabularies, enabling efficient training and inference. This approach reduces memory usage and speeds up inference without compromising performance.
By Franck Signe, Hippolyte Pilchen, Fran\c{c}ois Yvon, \'Edouard Grave