Image de couverture de livre

Flash-Unified: A Training-Free and Task-Aware Acceleration Framework for Native Unified Models

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Ke, Junlong, Wen, Zichen, Yang, Boxue, Yang, Yantai, Liu, Xuyang, Liao, Chenfei, Chen, Zhaorun, Wang, Shaobo, Zhang, Linfeng
Format:	Preprint
Publié:	2026
Sujets:	Computer Vision and Pattern Recognition
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

EvoStreaming: Your Offline Video Model Is a Natively Streaming Assistant
par: Wen, Zichen, et autres
Publié: (2026)

EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models
par: Yang, Yantai, et autres
Publié: (2025)

Accelerating Streaming Video Large Language Models via Hierarchical Token Compression
par: Wang, Yiyu, et autres
Publié: (2025)

Temporal2Seq: A Unified Framework for Temporal Video Understanding Tasks
par: Yang, Min, et autres
Publié: (2024)

Not All Samples Should Be Utilized Equally: Towards Understanding and Improving Dataset Distillation
par: Wang, Shaobo, et autres
Publié: (2024)

AI for Service: Proactive Assistance with AI Glasses
par: Wen, Zichen, et autres
Publié: (2025)

Understanding-in-Generation: Reinforcing Generative Capability of Unified Model via Infusing Understanding into Generation
par: Lyu, Yuanhuiyi, et autres
Publié: (2025)

Prune2Drive: A Plug-and-Play Framework for Accelerating Vision-Language Models in Autonomous Driving
par: Xiong, Minhao, et autres
Publié: (2025)

Efficient Multi-modal Large Language Models via Progressive Consistency Distillation
par: Wen, Zichen, et autres
Publié: (2025)

Token Transforming: A Unified and Training-Free Token Compression Framework for Vision Transformer Acceleration
par: Zeng, Fanhu, et autres
Publié: (2025)

TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models
par: Liu, Zhiheng, et autres
Publié: (2025)

Show-o2: Improved Native Unified Multimodal Models
par: Xie, Jinheng, et autres
Publié: (2025)

Many-for-Many: Unify the Training of Multiple Video and Image Generation and Manipulation Tasks
par: Li, Ruibin, et autres
Publié: (2025)

Unified Batch Normalization: Identifying and Alleviating the Feature Condensation in Batch Normalization and a Unified Framework
par: Wang, Shaobo, et autres
Publié: (2023)

VisionTrim: Unified Vision Token Compression for Training-Free MLLM Acceleration
par: Yu, Hanxun, et autres
Publié: (2026)

Video Compression Commander: Plug-and-Play Inference Acceleration for Video Large Language Models
par: Liu, Xuyang, et autres
Publié: (2025)

Antidote: A Unified Framework for Mitigating LVLM Hallucinations in Counterfactual Presupposition and Object Perception
par: Wu, Yuanchen, et autres
Publié: (2025)

Representation Shift: Unifying Token Compression with FlashAttention
par: Choi, Joonmyung, et autres
Publié: (2025)

One CT Unified Model Training Framework to Rule All Scanning Protocols
par: Xu, Fengzhi, et autres
Publié: (2026)

Shifting AI Efficiency From Model-Centric to Data-Centric Compression
par: Liu, Xuyang, et autres
Publié: (2025)

InteractPro: A Unified Framework for Motion-Aware Image Composition
par: Tao, Weijing, et autres
Publié: (2024)

UniViTAR: Unified Vision Transformer with Native Resolution
par: Qiao, Limeng, et autres
Publié: (2025)

Resolving Task Objective Conflicts in Unified Model via Task-Aware Mixture-of-Experts
par: Zhang, Jiaxing, et autres
Publié: (2025)

Are We Using the Right Benchmark: An Evaluation Framework for Visual Token Compression Methods
par: Liao, Chenfei, et autres
Publié: (2025)

VideoCompressa: Data-Efficient Video Understanding via Joint Temporal Compression and Spatial Reconstruction
par: Wang, Shaobo, et autres
Publié: (2025)

Representation Forcing for Bottleneck-Free Unified Multimodal Models
par: Wang, Yuqing, et autres
Publié: (2026)

UMIT: Unifying Medical Imaging Tasks via Vision-Language Models
par: Yu, Haiyang, et autres
Publié: (2025)

Stop Looking for Important Tokens in Multimodal Language Models: Duplication Matters More
par: Wen, Zichen, et autres
Publié: (2025)

ViCO: A Training Strategy towards Semantic Aware Dynamic High-Resolution
par: Cui, Long, et autres
Publié: (2025)

Training for Identity, Inference for Controllability: A Unified Approach to Tuning-Free Face Personalization
par: Pang, Lianyu, et autres
Publié: (2025)

FreeMotion: A Unified Framework for Number-free Text-to-Motion Synthesis
par: Fan, Ke, et autres
Publié: (2024)

Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching
par: Zheng, Zhixin, et autres
Publié: (2025)

PoseMaster: A Unified 3D Native Framework for Stylized Pose Generation
par: Yan, Hongyu, et autres
Publié: (2025)

PackCache: A Training-Free Acceleration Method for Unified Autoregressive Video Generation via Compact KV-Cache
par: Li, Kunyang, et autres
Publié: (2026)

Filter, Correlate, Compress: Training-Free Token Reduction for MLLM Acceleration
par: Han, Yuhang, et autres
Publié: (2024)

HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformer
par: Cai, Qi, et autres
Publié: (2026)

RoboSVG: A Unified Framework for Interactive SVG Generation with Multi-modal Guidance
par: Wang, Jiuniu, et autres
Publié: (2025)

Differential Vector Erasure: Unified Training-Free Concept Erasure for Flow Matching Models
par: Zhang, Zhiqi, et autres
Publié: (2026)

FlashAR: Efficient Post-Training Acceleration for Autoregressive Image Generation
par: Zhou, Junkang, et autres
Publié: (2026)

FlashMix: Fast Map-Free LiDAR Localization via Feature Mixing and Contrastive-Constrained Accelerated Training
par: Goswami, Raktim Gautam, et autres
Publié: (2024)