Does the VGGT Family Need All Its Layers?
cs.CV
Submitted: 2026-09-29
Updated: 2026-09-29
Terminology
Sources
- Qwen Technical Report
- DeepSeek LLM: Scaling Open-Source Language Models with Longtermism
- Quantized Visual Geometry Grounded Transformer
- What Matters in Transformers? Not All Attention is Needed
- Mistral 7B
- Rethinking Layer Redundancy: Calibration Matters More Than Search in LLM Depth Pruning
- FastVGGT: Training-Free Acceleration of Visual Geometry Transformer
- DINOv3
- MV-DUSt3R+: Single-Stage Scene Reconstruction from Sparse Views In 2 Seconds
- Llama 2: Open Foundation and Fine-Tuned Chat Models
- Entropy Reveals Block Importance in Masked Self-Supervised Vision Transformers
- LaCo: Large Language Model Pruning via Layer Collapse
- RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer
- Ghosted Layers: Unconstrained Activation Alignment for Recovering Layer-Pruned LLMs
- MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion
- DeepViT: Towards Deeper Vision Transformer
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