Shallow to Deep: Aligning Token Pruning with Stage-wise Roles in LVLMs

arXiv:2609.25635 · cs.CV · Submitted 2026-09-22 · Read on arXiv

cs.CV

Submitted: 2026-09-22

Updated: 2026-09-22

Code: https://github.com/Twilight03/STD

Terminology

• Pruning — 100× in this paper · explained in 4 episodes
• Benchmark — 18× in this paper · explained in 2 episodes
• Aggregation — 16× in this paper · explained in 1 episode
• Redundancy — 13× in this paper · explained in 2 episodes
• High-Frequency Spectral Analysis — 9× in this paper
• Gaussian-Smoothed Attention — 6× in this paper
• Accuracy — 5× in this paper · explained in 1 episode
• Ablation — 3× in this paper · explained in 1 episode
• Generalization — 3× in this paper · explained in 13 episodes
• Hallucination — 3× in this paper · explained in 3 episodes
• Robustness — 3× in this paper · explained in 5 episodes
• Alignment — 2× in this paper · explained in 1 episode
• Instruction Tuning — 2× in this paper · explained in 2 episodes
• Semantic Understanding — 2× in this paper · explained in 2 episodes
• Transformer — 2× in this paper · explained in 2 episodes
• alpha — 2× in this paper · explained in 1 episode
• Ablation Studies — 1× in this paper · explained in 1 episode
• Ablation study — 1× in this paper · explained in 1 episode
• Adaptability — 1× in this paper · explained in 1 episode
• Adaptive Strategy — 1× in this paper · explained in 1 episode
• Attention Mechanism — 1× in this paper · explained in 3 episodes
• Core — 1× in this paper · explained in 2 episodes
• Cosine Similarity — 1× in this paper · explained in 3 episodes
• F1 Score — 1× in this paper · explained in 3 episodes
• Few-Shot Learning — 1× in this paper · explained in 4 episodes
• Interpretability — 1× in this paper · explained in 15 episodes
• Multimodal Reasoning — 1× in this paper · explained in 5 episodes
• Semantic Similarity — 1× in this paper · explained in 2 episodes

Sources

• Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond
• Qwen3-VL Technical Report
• EvoPrune: Early-Stage Visual Token Pruning for Efficient MLLMs
• MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models
• iLLaVA: An Image is Worth Fewer Than 1/3 Input Tokens in Large Multimodal Models
• Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution
• AdaptInfer: Adaptive Token Pruning for Vision-Language Model Inference via Dynamical Text Guidance
• InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models

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