Chaque session du séminaire est divisée en deux présentations scientifiques de 40 minutes chacune : 30 minutes d’exposé et 10 minutes de questions. Tony Silveti-Falls & Erwan Allys animeront la session de mai 2025 !
Inscriptions gratuites mais obligatoires, dans la limite des places disponibles. Un buffet sera servi à l'issue du séminaire.
Résumé
In this talk, I discuss optimization methods that leverage the linear minimization oracle (LMO) over a norm-ball and their application to training huge neural networks. We propose a new stochastic family of algorithms that uses the LMO to adapt to the geometry of the problem and, perhaps surprisingly, show that they can be applied to unconstrained problems. The resulting update rule unifies several existing optimization methods under a single framework. Furthermore, we propose an explicit choice of norm for deep architectures, which, as a side benefit, leads to the transferability of hyperparameters across model sizes. Experimentally, we demonstrate significant speedups on nanoGPT training without any reliance on Adam. The proposed method is memory-efficient, requiring only one set of model weights and one set of gradients, which can be stored in half-precision.
Résumé
TBA