Imagen de Portada

Representation Shift: Unifying Token Compression with FlashAttention

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Choi, Joonmyung, Lee, Sanghyeok, Ko, Byungoh, Kim, Eunseo, Kil, Jihyung, Kim, Hyunwoo J.
Formato:	Preprint
Publicado:	2025
Materias:	Computer Vision and Pattern Recognition
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

DocPrune:Efficient Document Question Answering via Background, Question, and Comprehension-aware Token Pruning
por: Choi, Joonmyung, et al.
Publicado: (2026)

Multi-criteria Token Fusion with One-step-ahead Attention for Efficient Vision Transformers
por: Lee, Sanghyeok, et al.
Publicado: (2024)

EfficientViM: Efficient Vision Mamba with Hidden State Mixer based State Space Duality
por: Lee, Sanghyeok, et al.
Publicado: (2024)

vid-TLDR: Training Free Token merging for Light-weight Video Transformer
por: Choi, Joonmyung, et al.
Publicado: (2024)

TabFlash: Efficient Table Understanding with Progressive Question Conditioning and Token Focusing
por: Kim, Jongha, et al.
Publicado: (2025)

Relevance-aware Multi-context Contrastive Decoding for Retrieval-augmented Visual Question Answering
por: Kim, Jongha, et al.
Publicado: (2026)

Captioning for Text-Video Retrieval via Dual-Group Direct Preference Optimization
por: Lee, Ji Soo, et al.
Publicado: (2025)

MoE-GRPO: Optimizing Mixture-of-Experts via Reinforcement Learning in Vision-Language Models
por: Ko, Dohwan, et al.
Publicado: (2026)

II-MMR: Identifying and Improving Multi-modal Multi-hop Reasoning in Visual Question Answering
por: Kil, Jihyung, et al.
Publicado: (2024)

Robust Multimodal 3D Object Detection via Modality-Agnostic Decoding and Proximity-based Modality Ensemble
por: Cha, Juhan, et al.
Publicado: (2024)

Transferable Model-agnostic Vision-Language Model Adaptation for Efficient Weak-to-Strong Generalization
por: Park, Jihwan, et al.
Publicado: (2025)

Probabilistic Vision-Language Representation for Weakly Supervised Temporal Action Localization
por: Lim, Geuntaek, et al.
Publicado: (2024)

Bidirectional Likelihood Estimation with Multi-Modal Large Language Models for Text-Video Retrieval
por: Ko, Dohwan, et al.
Publicado: (2025)

Prompt Learning via Meta-Regularization
por: Park, Jinyoung, et al.
Publicado: (2024)

Efficient multi-view training for 3D Gaussian Splatting
por: Choi, Minhyuk, et al.
Publicado: (2025)

OTT-Vid: Optimal Transport Temporal Token Compression for Video Large Language Models
por: Kang, Minseok, et al.
Publicado: (2026)

Stochastic Conditional Diffusion Models for Robust Semantic Image Synthesis
por: Ko, Juyeon, et al.
Publicado: (2024)

Why and When Visual Token Pruning Fails? A Study on Relevant Visual Information Shift in MLLMs Decoding
por: Kim, Jiwan, et al.
Publicado: (2026)

Orthogonal Negative Guidance in Attention Feature Space for Text-to-Image Generation
por: Ko, Jungmin, et al.
Publicado: (2026)

Retinal Layer Segmentation in OCT Images With 2.5D Cross-slice Feature Fusion Module for Glaucoma Assessment
por: Kim, Hyunwoo, et al.
Publicado: (2026)

FlashVGGT: Efficient and Scalable Visual Geometry Transformers with Compressed Descriptor Attention
por: Wang, Zipeng, et al.
Publicado: (2025)

Dynamic Full-body Motion Agent with Object Interaction via Blending Pre-trained Modular Controllers
por: Nam, Sanghyeok, et al.
Publicado: (2026)

When Sinks Help or Hurt: Unified Framework for Attention Sink in Large Vision-Language Models
por: Choi, Jiho, et al.
Publicado: (2026)

When Model Knowledge meets Diffusion Model: Diffusion-assisted Data-free Image Synthesis with Alignment of Domain and Class
por: Kim, Yujin, et al.
Publicado: (2025)

GaussianVideo: Efficient Video Representation and Compression by Gaussian Splatting
por: Lee, Inseo, et al.
Publicado: (2025)

CanonCGT: Reference-Based Color Grading via Canonical Pivot Representation
por: Ko, Jinwon, et al.
Publicado: (2026)

UniCompress: Token Compression for Unified Vision-Language Understanding and Generation
por: Wang, Ziyao, et al.
Publicado: (2026)

ADAPT: Attention Driven Adaptive Prompt Scheduling and InTerpolating Orthogonal Complements for Rare Concepts Generation
por: Lee, Kwanyoung, et al.
Publicado: (2026)

Retrieval-Augmented Open-Vocabulary Object Detection
por: Kim, Jooyeon, et al.
Publicado: (2024)

Diffusion-Based sRGB Real Noise Generation via Prompt-Driven Noise Representation Learning
por: Ko, Jaekyun, et al.
Publicado: (2026)

ReCo: Reminder Composition Mitigates Hallucinations in Vision-Language Models
por: Chytas, Sotirios Panagiotis, et al.
Publicado: (2025)

Evaluating Visual Prompts with Eye-Tracking Data for MLLM-Based Human Activity Recognition
por: Choi, Jae Young, et al.
Publicado: (2026)

Learning Equi-angular Representations for Online Continual Learning
por: Seo, Minhyuk, et al.
Publicado: (2024)

MAIR++: Improving Multi-view Attention Inverse Rendering with Implicit Lighting Representation
por: Choi, JunYong, et al.
Publicado: (2024)

Blockwise Flow Matching: Improving Flow Matching Models For Efficient High-Quality Generation
por: Park, Dogyun, et al.
Publicado: (2025)

Diffusion Prior-Based Amortized Variational Inference for Noisy Inverse Problems
por: Lee, Sojin, et al.
Publicado: (2024)

Text-Conditioned Background Generation for Editable Multi-Layer Documents
por: Kang, Taewon, et al.
Publicado: (2025)

GPT-4V(ision) is a Generalist Web Agent, if Grounded
por: Zheng, Boyuan, et al.
Publicado: (2024)

VidChain: Chain-of-Tasks with Metric-based Direct Preference Optimization for Dense Video Captioning
por: Lee, Ji Soo, et al.
Publicado: (2025)

ODPG: Outfitting Diffusion with Pose Guided Condition
por: Lee, Seohyun, et al.
Publicado: (2025)