Romain Fabre — accepted AI-conference papers

2026

Simultaneous Speech-to-Speech Translation Without Aligned Data

Tom Labiausse, Romain Fabre, Yannick Estève, Alexandre Défossez, Neil Zeghidour

ICML 2026oral

Simultaneous speech translation is the task of translating source speech into a target language in real-time. Given that the dependencies between source and target words are non-monotonic (e.g. the word order can change between German and English), this means learning to jointly align and translate.…

Cited by 0 Source ↗code ↗Scholar ↗

2026

Understanding Data Temporality Impact on Large Language Models Pre-training

Hippolyte Pilchen, Romain Fabre, Franck SIGNE TALLA, Patrick Perez, Edouard Grave

ICML 2026poster

Large language models (LLMs) are typically trained on shuffled corpora, yielding models whose knowledge is frozen at training time and whose temporal grounding remains poorly understood. In this work, we study the impact of pretraining dynamics on the acquisition of time-sensitive factual knowledge,…

Cited by 0 Source ↗Scholar ↗