Louis Béthune

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

The data-quality illusion: Rethinking Classifier-based Quality Filtering for LLM Pretraining

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

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

Scaling Laws for Optimal Data Mixtures

Multivariate Conformal Prediction using Optimal Transport

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

Scaling Laws for Forgetting during Finetuning with Pretraining Data Injection

Sparse Repellency for Shielded Generation in Text-to-Image Diffusion Models

Shielded Diffusion: Generating Novel and Diverse Images using Sparse Repellency