Publications

2026

  1. Multivariate Conformal Prediction using Optimal Transport

    Michal Klein, Louis Béthune, Eugène Ndiaye, Marco Cuturi

    TMLR arXiv

  2. LaCy: What Small Language Models Can and Should Learn into Limited Parametric Memory

    Szilvia Ujváry, Louis Béthune, Pierre Ablin, João Monteiro, Marco Cuturi, Michael Kirchhof

    ICLR Workshop 2026 arXiv

  3. Completed Hyperparameter Transfer across Modules, Width, Depth, Batch and Duration

    Bruno Mlodozeniec, Pierre Ablin, Louis Béthune, Dan Busbridge, Michal Klein, Jason Ramapuram, Marco Cuturi

    ICLR 2026 arXiv

  4. Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretraining

    T. N. Saada, Louis Béthune, Michal Klein, David Grangier, Pierre Ablin

    ICML 2026 arXiv

  5. Optimal Splitting of Language Models from Mixtures to Specialized Domains

    Skyler Seto, Pierre Ablin, Anastasiia Filippova, Jiayuan Ye, Louis Béthune, Angelos Katharopoulos, David Grangier

    ICML 2026 arXiv

  6. Learning unmasking policies for diffusion language models

    Metod Jazbec, Theo X. Olausson, Louis Béthune, Pierre Ablin, Michael Kirchhof, João Monteiro, Vítor Turrisi, Jason Ramapuram, Marco Cuturi

    ICML 2026 Oral arXiv

  7. Scaling Categorical Flow Maps

    Oscar Davis, Anastasiia Filippova, Pierre Ablin, Vítor Turrisi, Amitis Shidani, Marco Cuturi, Louis Béthune

    arXiv arXiv

  8. The Design Space of Tri-Modal Masked Diffusion Models

    Louis Béthune, Vítor Turrisi, Bruno Kacper Mlodozeniec, Pau Rodríguez López, Lokesh Boominathan, Nikhil Bhendawade, Amitis Shidani, Joris Pelemans, Theo X. Olausson, Devon Hjelm, Paul Dixon, João Monteiro, Pierre Ablin, Vishnu Banna, Arno Blaas, Nick Henderson, Kari Noriy, Dan Busbridge, Josh Susskind, Marco Cuturi, Irina Belousova, Luca Zappella, Russ Webb, Jason Ramapuram

    arXiv arXiv

  9. Mix, Don't Tune: Bilingual Pre-Training Outperforms Hyperparameter Search in Data-Constrained Settings

    Paul Jeha, Anastasiia Sedova, Louis Béthune, Skyler Seto, Jes Frellsen, Pierre Ablin, Natalie Schluter

    arXiv arXiv

  10. DynaMiCS: Fine-tuning LLMs with Performance Constraints using Dynamic Mixtures

    Eleonora Gualdoni, Sonia Laguna, Louis Béthune, João Monteiro, Pierre Ablin, Marco Cuturi

    arXiv arXiv

2025

  1. Controlled Generation with Distilled Diffusion Energy Models and Sequential Monte Carlo

    James Thornton, Louis Béthune, Ruixiang Zhang, Arwen Bradley, Preetum Nakkiran, Shuangfei Zhai

    AISTATS 2025 arXiv

  2. Improved learning theory for kernel distribution regression with two-stage sampling

    François Bachoc, Louis Béthune, Alberto González-Sanz, Jean-Michel Loubes

    The Annals of Statistics arXiv

  3. Multimodal autoregressive pre-training of large vision encoders

    Enrico Fini, Mustafa Shukor, Xiujun Li, Philipp Dufter, Michal Klein, David Haldimann, Sai Aitharaju, Victor Guilherme Turrisi da Costa, Louis Béthune, Zhe Gan, Alexander T. Toshev, Marcin Eichner, Moin Nabi, Yinfei Yang, Joshua M. Susskind, Alaaeldin El-Nouby

    CVPR 2025 arXiv

  4. Scaling Laws for Forgetting during Finetuning with Pretraining Data Injection

    Louis Béthune, David Grangier, Dan Busbridge, Eleonora Gualdoni, Marco Cuturi, Pierre Ablin

    ICML 2025 arXiv

  5. Shielded Diffusion: Generating Novel and Diverse Images using Sparse Repellency

    Michael Kirchhof, James Thornton, Louis Béthune, Pierre Ablin, Eugène Ndiaye, Marco Cuturi

    ICML 2025 arXiv

  6. The Geometries of Truth Are Orthogonal Across Tasks

    Waïss Azizian, Michael Kirchhof, Eugène Ndiaye, Louis Béthune, Michal Klein, Pierre Ablin, Marco Cuturi

    ICML Workshop 2025 arXiv

  7. Deep Sturm–Liouville: From Sample-Based to 1D Regularization with Learnable Orthogonal Basis Functions

    David Vigouroux, Joseba Dalmau, Louis Béthune, Victor Boutin

    ICML 2025 arXiv

  8. Scaling Laws for Optimal Data Mixtures

    Mustafa Shukor, Louis Béthune, Dan Busbridge, David Grangier, Enrico Fini, Alaaeldin El-Nouby, Pierre Ablin

    NeurIPS 2025 arXiv

  9. Sample and Map from a Single Convex Potential: Generation using Conjugate Moment Measures

    Nina Vesseron, Louis Béthune, Marco Cuturi

    NeurIPS 2025 arXiv

  10. Follow the Energy, Find the Path: Riemannian Metrics from Energy-Based Models

    Louis Béthune, David Vigouroux, Yilun Du, Rufin VanRullen, Thomas Serre, Victor Boutin

    NeurIPS 2025 arXiv

2024

  1. DP-SGD Without Clipping: The Lipschitz Neural Network Way

    Louis Béthune, Thomas Masséna, Thibaut Boissin, Yannick Prudent, Corentin Friedrich, Franck Mamalet, Aurélien Bellet, Mathieu Serrurier, David Vigouroux

    ICLR 2024 arXiv

  2. Deep learning with Lipschitz constraints

    Louis Béthune

    PhD thesis 2024 Paper

  3. 1-Lipschitz Neural Distance Fields

    Guillaume Coiffier, Louis Béthune

    SGP 2024 Best paper arXiv

  4. Understanding Visual Feature Reliance through the Lens of Complexity

    Thomas Fel, Louis Béthune, Andrew K. Lampinen, Thomas Serre, Katherine Hermann

    NeurIPS 2024 arXiv

  5. Graph-based captioning: Enhancing visual descriptions by interconnecting region captions

    Yu-Guan Hsieh, Cheng-Yu Hsieh, Shih-Ying Yeh, Louis Béthune, Hadi Pouransari, Pavan Kumar Anasosalu Vasu, Chun-Liang Li, Ranjay Krishna, Oncel Tuzel, Marco Cuturi

    arXiv arXiv

2023

  1. Gaussian processes on distributions based on regularized optimal transport

    François Bachoc, Louis Béthune, Alberto González-Sanz, Jean-Michel Loubes

    AISTATS 2023 Paper

  2. CRAFT: Concept Recursive Activation FacTorization for Explainability

    Thomas Fel, Agustin Picard, Louis Béthune, Thibaut Boissin, David Vigouroux, Julien Colin, Rémi Cadène, Thomas Serre

    CVPR 2023 Paper

  3. Robust One-Class Classification with Signed Distance Function using 1-Lipschitz Neural Networks

    Louis Béthune, Paul Novello, Thibaut Boissin, Guillaume Coiffier, Mathieu Serrurier, Quentin Vincenot, Andres Troya-Galvis

    ICML 2023 Paper

  4. A holistic approach to unifying automatic concept extraction and concept importance estimation

    Thomas Fel, Victor Boutin, Louis Béthune, Rémi Cadène, Mazda Moayeri, Léo Andéol, Mathieu Chalvidal, Thomas Serre

    NeurIPS 2023 arXiv

  5. On the explainable properties of 1-Lipschitz Neural Networks: An Optimal Transport Perspective

    Mathieu Serrurier, Franck Mamalet, Thomas Fel, Louis Béthune, Thibaut Boissin

    NeurIPS 2023 arXiv

  6. Taco: Targeted concept erasure prevents non-linear classifiers from detecting protected attributes

    Fanny Jourdan, Louis Béthune, Agustin Picard, Laurent Risser, Nicholas Asher

    arXiv arXiv

2022

  1. Xplique: A deep learning explainability toolbox

    Thomas Fel, Lucas Hervier, David Vigouroux, Antonin Poché, Justin Plakoo, Rémi Cadène, Mathieu Chalvidal, Julien Colin, Thibaut Boissin, Louis Béthune, Agustin Picard, Claire Nicodème, Laurent Gardes, Gregory Flandin, Thomas Serre

    CVPR Workshop 2022 arXiv

  2. Efficient circuit implementation for coined quantum walks on binary trees and application to reinforcement learning

    Thomas Mullor, David Vigouroux, Louis Béthune

    SEC 2022 arXiv

  3. Pay attention to your loss: understanding misconceptions about Lipschitz neural networks

    Louis Béthune, Thibaut Boissin, Mathieu Serrurier, Franck Mamalet, Corentin Friedrich, Alberto González-Sanz

    NeurIPS 2022 Paper

  4. Certifiable Metric One Class Learning with adversarially trained Lipschitz Classifier

    Louis Béthune, Mathieu Serrurier

    NeurIPS Workshop 2022

  5. GAN estimation of Lipschitz optimal transport maps

    Alberto González-Sanz, Lucas de Lara, Louis Béthune, Jean-Michel Loubes

    arXiv arXiv

2021

  1. Predicting the generalization ability of a few-shot classifier

    Myriam Bontonou, Louis Béthune, Vincent Gripon

    Information 2021 Paper

2020

  1. Hierarchical and unsupervised graph representation learning with Loukas's coarsening

    Louis Béthune, Yacouba Kaloga, Pierre Borgnat, Aurélien Garivier, Amaury Habrard

    Algorithms 2020 Paper

  2. Apprentissage de représentations hiérarchiques de graphes avec graph2vec et la réduction de Loukas

    Louis Béthune, Yacouba Kaloga, Pierre Borgnat, Aurélien Garivier, Amaury Habrard

    CAp 2020

  3. Ranking deep learning generalization using label variation in latent geometry graphs

    Carlos Lassance, Louis Béthune, Myriam Bontonou, Mounia Hamidouche, Vincent Gripon

    NeurIPS 2020 Competition arXiv