Foundry：为边缘设备蒸馏 3D 基础模型
Foundry: Distilling 3D Foundation Models for the Edge
把大型自监督 3D 基础模型蒸馏成紧凑的 SuperTokens 代理模型，学生重建教师 token 级表征，可直接用于点云分类、部件分割等下游任务的轻量化部署，为边缘端 3D 感知提供可复用的压缩范式。
原文依据 trains a student to learn a compressed set of SuperTokens that reconstruct the teacher's token-level representations
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论在大规模数据集上以自监督学习（SSL）预训练的基础模型已成为强大的通用特征提取器。然而，其庞大的规模与计算开销使其难以部署在机器人、AR/VR 头显等边缘设备上。现有的压缩技术（如标准知识蒸馏）虽能产生高效的「专家」模型，却牺牲了基础模型最有价值的、与下游任务无关的通用性。本文提出基础模型蒸馏（Foundation Model Distillation, FMD），一种将大型 SSL 模型压缩为紧凑、高效且忠实的代理模型的新范式，使其保留通用表征能力。我们提出 Foundry，这是 FMD 在 3D 点云上的首个实现。该方法训练学生模型学习一组压缩的 SuperTokens，以重建教师的 token 级表征，从而捕获其潜空间的一个紧凑基。单个蒸馏模型在分类、部件分割和少样本等多种下游任务上保持强迁移性，在显著减少 token 数与 FLOPs 的同时接近完整基础模型的性能，使此类模型更适用于资源受限硬件的部署。
英文原文摘要 1297 字符
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 resource-constrained hardware.