Cover Image

LightVLM: Acceleraing Large Multimodal Models with Pyramid Token Merging and KV Cache Compression

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Hu, Lianyu, Shang, Fanhua, Feng, Wei, Wan, Liang
Format:	Preprint
Published:	2025
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

iLLaVA: An Image is Worth Fewer Than 1/3 Input Tokens in Large Multimodal Models
by: Hu, Lianyu, et al.
Published: (2024)

Deep Correlated Prompting for Visual Recognition with Missing Modalities
by: Hu, Lianyu, et al.
Published: (2024)

Pose-Guided Fine-Grained Sign Language Video Generation
by: Shi, Tongkai, et al.
Published: (2024)

Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generation
by: Chen, Jiayu, et al.
Published: (2026)

Forcing-KV: Hybrid KV Cache Compression for Efficient Autoregressive Video Diffusion Models
by: Ji, Yicheng, et al.
Published: (2026)

Completed Feature Disentanglement Learning for Multimodal MRIs Analysis
by: Liu, Tianling, et al.
Published: (2024)

A Survey of Token Compression for Efficient Multimodal Large Language Models
by: Shao, Kele, et al.
Published: (2025)

CalibQuant: 1-Bit KV Cache Quantization for Multimodal LLMs
by: Han, Insu, et al.
Published: (2025)

CorrNet+: Sign Language Recognition and Translation via Spatial-Temporal Correlation
by: Hu, Lianyu, et al.
Published: (2024)

CFCML: A Coarse-to-Fine Crossmodal Learning Framework For Disease Diagnosis Using Multimodal Images and Tabular Data
by: Liu, Tianling, et al.
Published: (2026)

Mixing Importance with Diversity: Joint Optimization for KV Cache Compression in Large Vision-Language Models
by: Liu, Xuyang, et al.
Published: (2025)

Head-Aware KV Cache Compression for Efficient Visual Autoregressive Modeling
by: Qin, Ziran, et al.
Published: (2025)

Beyond Background Shift: Rethinking Instance Replay in Continual Semantic Segmentation
by: Yin, Hongmei, et al.
Published: (2025)

FlashVLM: Text-Guided Visual Token Selection for Large Multimodal Models
by: Cai, Kaitong, et al.
Published: (2025)

Can Visual Input Be Compressed? A Visual Token Compression Benchmark for Large Multimodal Models
by: Peng, Tianfan, et al.
Published: (2025)

Cached Adaptive Token Merging: Dynamic Token Reduction and Redundant Computation Elimination in Diffusion Model
by: Saghatchian, Omid, et al.
Published: (2025)

PDD: Manifold-Prior Diverse Distillation for Medical Anomaly Detection
by: Lu, Xijun, et al.
Published: (2026)

StreamKV: Streaming Video Question-Answering with Segment-based KV Cache Retrieval and Compression
by: Chen, Yilong, et al.
Published: (2025)

LiveVLM: Efficient Online Video Understanding via Streaming-Oriented KV Cache and Retrieval
by: Ning, Zhenyu, et al.
Published: (2025)

LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models
by: Shang, Yuzhang, et al.
Published: (2024)

AirCache: Activating Inter-modal Relevancy KV Cache Compression for Efficient Large Vision-Language Model Inference
by: Huang, Kai, et al.
Published: (2025)

PPE: Positional Preservation Embedding for Token Compression in Multimodal Large Language Models
by: Huang, Mouxiao, et al.
Published: (2025)

LagKV: Lag-Relative Information of the KV Cache Tells Which Tokens Are Important
by: Liang, Manlai, et al.
Published: (2025)

HybridToken-VLM: Hybrid Token Compression for Vision-Language Models
by: Zhang, Jusheng, et al.
Published: (2025)

Revisiting Multimodal KV Cache Compression: A Frequency-Domain-Guided Outlier-KV-Aware Approach
by: Yang, Yaoxin, et al.
Published: (2025)

PackCache: A Training-Free Acceleration Method for Unified Autoregressive Video Generation via Compact KV-Cache
by: Li, Kunyang, et al.
Published: (2026)

InfoMerge: Information-aware Token Compression for Efficient Video Large Language Models
by: Liu, Xinxin, et al.
Published: (2026)

TokenCarve: Information-Preserving Visual Token Compression in Multimodal Large Language Models
by: Tan, Xudong, et al.
Published: (2025)

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

LiteVGGT: Boosting Vanilla VGGT via Geometry-aware Cached Token Merging
by: Shu, Zhijian, et al.
Published: (2025)

XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
by: Su, Zunhai, et al.
Published: (2026)

XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
by: Su, Zunhai, et al.
Published: (2026)

OmniVLM: A Token-Compressed, Sub-Billion-Parameter Vision-Language Model for Efficient On-Device Inference
by: Chen, Wei, et al.
Published: (2024)

LOOK-M: Look-Once Optimization in KV Cache for Efficient Multimodal Long-Context Inference
by: Wan, Zhongwei, et al.
Published: (2024)

Dynamic Spatial-Temporal Aggregation for Skeleton-Aware Sign Language Recognition
by: Hu, Lianyu, et al.
Published: (2024)

Dynamic-VLM: Simple Dynamic Visual Token Compression for VideoLLM
by: Wang, Han, et al.
Published: (2024)

Improving Continuous Sign Language Recognition with Adapted Image Models
by: Hu, Lianyu, et al.
Published: (2024)

Dynamic Pyramid Network for Efficient Multimodal Large Language Model
by: Ai, Hao, et al.
Published: (2025)

STaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models
by: Han, Yuhang, et al.
Published: (2026)

Decouple and Cache: KV Cache Construction for Streaming Video Understanding
by: Pang, Zhanzhong, et al.
Published: (2026)