Cover Image

HiDrop: Hierarchical Vision Token Reduction in MLLMs via Late Injection, Concave Pyramid Pruning, and Early Exit

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wu, Hao, Fan, Yingqi, Dai, Jinyang, Tong, Junlong, Ma, Yunpu, Shen, Xiaoyu
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

StreamingThinker: Large Language Models Can Think While Reading
by: Tong, Junlong, et al.
Published: (2025)

SkipGPT: Dynamic Layer Pruning Reinvented with Token Awareness and Module Decoupling
by: Zhao, Anhao, et al.
Published: (2025)

UTPTrack: Towards Simple and Unified Token Pruning for Visual Tracking
by: Wu, Hao, et al.
Published: (2026)

What Do Visual Tokens Really Encode? Uncovering Sparsity and Redundancy in Multimodal Large Language Models
by: Fan, Yingqi, et al.
Published: (2026)

HiPrune: Hierarchical Attention for Efficient Token Pruning in Vision-Language Models
by: Liu, Jizhihui, et al.
Published: (2025)

PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction
by: Xing, Long, et al.
Published: (2024)

The Few Govern the Many:Unveiling Few-Layer Dominance for Time Series Models
by: Qiu, Xin, et al.
Published: (2025)

Think-as-You-See: Streaming Chain-of-Thought Reasoning for Large Vision-Language Models
by: Zhang, Jialiang, et al.
Published: (2026)

Rethinking the Role of LLMs in Time Series Forecasting
by: Qiu, Xin, et al.
Published: (2026)

Decoupling KL and Trajectories: A Unified Perspective for SFT, DAgger, Offline RL, and OPD in LLM Distillation
by: Zhao, Anhao, et al.
Published: (2026)

ViCA: Efficient Multimodal LLMs with Vision-Only Cross-Attention
by: Liu, Wenjie, et al.
Published: (2026)

EvoPrune: Early-Stage Visual Token Pruning for Efficient MLLMs
by: Chen, Yuhao, et al.
Published: (2026)

LLM as Effective Streaming Processor: Bridging Streaming-Batch Mismatches with Group Position Encoding
by: Tong, Junlong, et al.
Published: (2025)

$\mathcal{V}isi\mathcal{P}runer$: Decoding Discontinuous Cross-Modal Dynamics for Efficient Multimodal LLMs
by: Fan, Yingqi, et al.
Published: (2025)

HiPP-Prune: Hierarchical Preference-Conditioned Structured Pruning for Vision-Language Models
by: Bai, Lincen, et al.
Published: (2026)

Dynamic Vocabulary Pruning in Early-Exit LLMs
by: Vincenti, Jort, et al.
Published: (2024)

HiRED: Attention-Guided Token Dropping for Efficient Inference of High-Resolution Vision-Language Models
by: Arif, Kazi Hasan Ibn, et al.
Published: (2024)

Pyramid Token Pruning for High-Resolution Large Vision-Language Models via Region, Token, and Instruction-Guided Importance
by: Liang, Yuxuan, et al.
Published: (2025)

GridPrune: From "Where to Look" to "What to Select" in Visual Token Pruning for MLLMs
by: Duan, Yuxiang, et al.
Published: (2025)

Early Exit Is a Natural Capability in Transformer-based Models: An Empirical Study on Early Exit without Joint Optimization
by: Shan, Weiqiao, et al.
Published: (2024)

IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs
by: Tan, Yifan, et al.
Published: (2026)

TinyDrop: Tiny Model Guided Token Dropping for Vision Transformers
by: Wang, Guoxin, et al.
Published: (2025)

HiSpatial: Taming Hierarchical 3D Spatial Understanding in Vision-Language Models
by: Liang, Huizhi, et al.
Published: (2026)

HERO: Rethinking Visual Token Early Dropping in High-Resolution Large Vision-Language Models
by: Li, Xu, et al.
Published: (2025)

TIDE: Token-Informed Depth Execution for Per-Token Early Exit in LLM Inference
by: Jaber, Jaber, et al.
Published: (2026)

Beyond Attention or Similarity: Maximizing Conditional Diversity for Token Pruning in MLLMs
by: Zhang, Qizhe, et al.
Published: (2025)

Does Visual Token Pruning Improve Calibration? An Empirical Study on Confidence in MLLMs
by: Tan, Kaizhen
Published: (2026)

Grounding-Aware Token Pruning: Recovering from Drastic Performance Drops in Visual Grounding Caused by Pruning
by: Chien, Tzu-Chun, et al.
Published: (2025)

Video Patch Pruning: Efficient Video Instance Segmentation via Early Token Reduction
by: Glandorf, Patrick, et al.
Published: (2026)

Context Guided Transformer Entropy Modeling for Video Compression
by: Tong, Junlong, et al.
Published: (2025)

HiP-AD: Hierarchical and Multi-Granularity Planning with Deformable Attention for Autonomous Driving in a Single Decoder
by: Tang, Yingqi, et al.
Published: (2025)

HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling
by: Liu, Xianjie, et al.
Published: (2025)

Attention Debiasing for Token Pruning in Vision Language Models
by: Zhao, Kai, et al.
Published: (2025)

Variation-aware Vision Token Dropping for Faster Large Vision-Language Models
by: Chen, Junjie, et al.
Published: (2025)

Amortized-Precision Quantization for Early-Exit Vision Transformers
by: Fang, Rui, et al.
Published: (2026)

Efficient Token Compression for Vision Transformer with Spatial Information Preserved
by: Mao, Junzhu, et al.
Published: (2025)

HiLa: Hierarchical Vision-Language Collaboration for Cancer Survival Prediction
by: Cui, Jiaqi, et al.
Published: (2025)

HiTVideo: Hierarchical Tokenizers for Enhancing Text-to-Video Generation with Autoregressive Large Language Models
by: Zhou, Ziqin, et al.
Published: (2025)

TARS: MinMax Token-Adaptive Preference Strategy for Hallucination Reduction in MLLMs
by: Zhang, Kejia, et al.
Published: (2025)

FREE: Fast and Robust Vision Language Models with Early Exits
by: Bajpai, Divya Jyoti, et al.
Published: (2025)