GRAFT: Growing Agglomerative Foundation Models via Continual Teacher Distillation
cs.CV
Submitted: 2026-10-01
Updated: 2026-10-01
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- ARKitScenes: A Diverse Real-World Dataset For 3D Indoor Scene Understanding Using Mobile RGB-D Data
- Distillation Scaling Laws
- Virtual KITTI 2
- SigLino: Efficient Multi-Teacher Distillation for Agglomerative Vision Foundation Models
- Asymmetric Decision-Making in Online Knowledge Distillation:Unifying Consensus and Divergence
- Pathryoshka: Compressing Pathology Foundation Models via Multi-Teacher Knowledge Distillation with Nested Embeddings
- LoRA: Low-Rank Adaptation of Large Language Models
- Fine-Tuning can Distort Pretrained Features and Underperform Out-of-Distribution
- Continual Distillation of Teachers from Different Domains
- DINOv2: Learning Robust Visual Features without Supervision
- PHI-S: Distribution Balancing for Label-Free Multi-Teacher Distillation
- ImageNet-21K Pretraining for the Masses
- Theia: Distilling Diverse Vision Foundation Models for Robot Learning
- DINOv3
- LoFTR: Detector-Free Local Feature Matching with Transformers
- SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features
- Agglomerating Large Vision Encoders via Distillation for VFSS Segmentation
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models