Avatar

Benjamin Muller

AI Researcher

                         

I am a Postdoctoral researcher at Meta AI in the FAIR lab based in NYC.

I completed my PhD in 2022 at Sorbonne Université and INRIA Paris in the Almanach research team.

AI technologies are changing the way we learn, communicate and connect with each other. I am interested in scaling them to the largest number of languages. This involves designing better multimodal and multilingual models, adaptation techniques, and evaluation benchmarks.

I’m also a former mentor and vigorous supporter of the Fatima Fellowship, a program dedicated to breaking down barriers in AI research and welcoming students from various backgrounds and origins.

Publications

Updated List here 💡

(2024). SPIRIT-LM: Interleaved Spoken and Written Language Model.

PDF Project

(2023). Evaluating and Modeling Attribution for Cross-Lingual Question Answering. EMNLP.

PDF Dataset

(2023). The Gender-GAP Pipeline: A Gender-Aware Polyglot Pipeline for Gender Characterisation in 55 Languages. WMT.

PDF Dataset Project

(2022). Cross-Lingual GENQA: A Language-Agnostic Generative Question Answering Approach for Open-Domain Question Answering. AACL.

PDF

(2021). When Being Unseen from mBERT is just the Beginning: Handling New Languages With Multilingual Language Models. NAACL.

PDF

Teaching

2019-2022 - Lecturer and Main Instructor at ENSAE Paris for the Machine Learning for Natural Language Processing course.

2018-2019 - Teaching assistant for Master students in Initiation to Research in Statistics at University Paris-Descartes

2014-2016 - Teaching assistant for Bachelor students in Mathematics, Prépa ECS at Lycée Ipécom, Paris

Mentoring

2022-2023 Mentor for the Fatima Fellowship program. Mentoring three talented students in analysing cultural biases in multilingual language models leading to a publication at EMNLP Findings 2023.

2018-2022 - Supervising MSc students’ research projects. Focusing on implementation, evaluation and analysis of static Word Embedding techniques and Language Modeling.

2019 - Supervising internship on Domain Adaptation for Non-Canonical Data (6 months)