Foundry: Distilling 3D Foundation Models for the Edge
Letellier, Guillaume, Srivastava, Siddharth, Jurie, Frédéric, Sharma, Gaurav
–arXiv.org Artificial Intelligence
Foundation models pre-trained with self-supervised learning (SSL) on large-scale datasets have become powerful general-purpose feature extractors. However, their immense size and computational cost make them prohibitive for deployment on edge devices such as robots and AR/VR headsets. Existing compression techniques like standard knowledge distillation create efficient 'specialist' models but sacrifice the crucial, downstream-agnostic generality that makes foundation models so valuable. In this paper, we introduce Foundation Model Distillation (FMD), a new paradigm for compressing large SSL models into compact, efficient, and faithful proxies that retain their general-purpose representational power. We present Foundry, the first implementation of FMD for 3D point clouds. Our approach, Foundry, trains a student to learn a compressed set of SuperTokens that reconstruct the teacher's token-level representations, capturing a compact basis of its latent space. A single distilled model maintains strong transferability across diverse downstream tasks-classification, part segmentation, and few-shot scenarios-approaching full foundation-model performance while using significantly fewer tokens and FLOPs, making such models more practical for deployment on resourceconstrained hardware.
arXiv.org Artificial Intelligence
Nov-27-2025
- Country:
- Asia > China (0.04)
- Europe
- Austria (0.04)
- France > Île-de-France
- North America
- Canada > British Columbia
- Vancouver (0.04)
- United States > Washington
- King County > Seattle (0.04)
- Canada > British Columbia
- Genre:
- Research Report (0.49)
- Industry:
- Consumer Products & Services (0.67)
- Leisure & Entertainment (0.67)
- Transportation (0.46)
- Technology: