Uploaded February 2025 | Updated November 2026, 3 hours ago
Abstract:
I will talk about the foundations of flow matching, scaling them for large-scale text-to-image pretraining, preference-tuning approaches and techniques such as model distillation that allow to serve these models efficiently at scale. In addition, I will talk a bit about conducting research in a startup context.
Abstract:
I will talk about the foundations of flow matching, scaling them for large-scale text-to-image pretraining, preference-tuning approaches and techniques such as model distillation that allow to serve these models efficiently at scale. In addition, I will talk a bit about conducting research in a startup context.










