Imagen de Portada

One Token, Two Fates: A Unified Framework via Vision Token Manipulation Against MLLMs Hallucination

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Fa, Zhan, Duan, Yue, Zhang, Jian, Qi, Lei, Shi, Yinghuan
Formato:	Preprint
Publicado:	2026
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

Decomposing and Composing: Towards Efficient Vision-Language Continual Learning via Rank-1 Expert Pool in a Single LoRA
por: Fa, Zhan, et al.
Publicado: (2026)

SETA: Semantic-Aware Token Augmentation for Domain Generalization
por: Guo, Jintao, et al.
Publicado: (2024)

START: A Generalized State Space Model with Saliency-Driven Token-Aware Transformation
por: Guo, Jintao, et al.
Publicado: (2024)

Patch-as-Decodable-Token: Towards Unified Multi-Modal Vision Tasks in MLLMs
por: Su, Yongyi, et al.
Publicado: (2025)

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

Dual-Teacher Ensemble Models with Double-Copy-Paste for 3D Semi-Supervised Medical Image Segmentation
por: Fa, Zhan, et al.
Publicado: (2024)

Divide-and-Conquer for Enhancing Unlabeled Learning, Stability, and Plasticity in Semi-supervised Continual Learning
por: Duan, Yue, et al.
Publicado: (2025)

An Adaptor for Triggering Semi-Supervised Learning to Out-of-Box Serve Deep Image Clustering
por: Duan, Yue, et al.
Publicado: (2025)

Two Causes, Not One: Rethinking Omission and Fabrication Hallucinations in MLLMs
por: Si, Guangzong, et al.
Publicado: (2025)

Token Transforming: A Unified and Training-Free Token Compression Framework for Vision Transformer Acceleration
por: Zeng, Fanhu, et al.
Publicado: (2025)

A Unified Hallucination Mitigation Framework for Large Vision-Language Models
por: Chang, Yue, et al.
Publicado: (2024)

Unified Language-Vision Pretraining in LLM with Dynamic Discrete Visual Tokenization
por: Jin, Yang, et al.
Publicado: (2023)

MergeTok: Unified Continuous and Discrete Visual Tokenization via Token Merging
por: Zhang, Luyuan, et al.
Publicado: (2026)

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

Correspondence as Video: Test-Time Adaption on SAM2 for Reference Segmentation in the Wild
por: Wang, Haoran, et al.
Publicado: (2025)

Learn to Preserve and Diversify: Parameter-Efficient Group with Orthogonal Regularization for Domain Generalization
por: Hu, Jiajun, et al.
Publicado: (2024)

Efficient3D: A Unified Framework for Adaptive and Debiased Token Reduction in 3D MLLMs
por: Lin, Yuhui, et al.
Publicado: (2026)

HalLoc: Token-level Localization of Hallucinations for Vision Language Models
por: Park, Eunkyu, et al.
Publicado: (2025)

QG-VTC: Question-Guided Visual Token Compression in MLLMs for Efficient VQA
por: Li, Shuai, et al.
Publicado: (2025)

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

Image Tokens Matter: Mitigating Hallucination in Discrete Tokenizer-based Large Vision-Language Models via Latent Editing
por: Wang, Weixing, et al.
Publicado: (2025)

Token Bottleneck: One Token to Remember Dynamics
por: Kim, Taekyung, et al.
Publicado: (2025)

Where MLLMs Attend and What They Rely On: Explaining Autoregressive Token Generation
por: Chen, Ruoyu, et al.
Publicado: (2025)

Semantic One-Dimensional Tokenizer for Image Reconstruction and Generation
por: Qu, Yunpeng, et al.
Publicado: (2026)

Mamba-Sea: A Mamba-based Framework with Global-to-Local Sequence Augmentation for Generalizable Medical Image Segmentation
por: Cheng, Zihan, et al.
Publicado: (2025)

AToken: A Unified Tokenizer for Vision
por: Lu, Jiasen, et al.
Publicado: (2025)

Token Warping Helps MLLMs Look from Nearby Viewpoints
por: Lee, Phillip Y., et al.
Publicado: (2026)

SEA: Supervised Embedding Alignment for Token-Level Visual-Textual Integration in MLLMs
por: Yin, Yuanyang, et al.
Publicado: (2024)

MINT: Mitigating Hallucinations in Large Vision-Language Models via Token Reduction
por: Wang, Chao, et al.
Publicado: (2025)

Vision Foundation Models as Generalist Tokenizers for Image Generation
por: Zheng, Anlin, et al.
Publicado: (2026)

Magic-MM-Embedding: Towards Visual-Token-Efficient Universal Multimodal Embedding with MLLMs
por: Li, Qi, et al.
Publicado: (2026)

VisionTrim: Unified Vision Token Compression for Training-Free MLLM Acceleration
por: Yu, Hanxun, et al.
Publicado: (2026)

HALP: Detecting Hallucinations in Vision-Language Models without Generating a Single Token
por: Kogilathota, Sai Akhil, et al.
Publicado: (2026)

Learning Generalizable Models via Disentangling Spurious and Enhancing Potential Correlations
por: Wang, Na, et al.
Publicado: (2024)

Taste More, Taste Better: Diverse Data and Strong Model Boost Semi-Supervised Crowd Counting
por: Yang, Maochen, et al.
Publicado: (2025)

Unified Pix Token And Word Token Generative Language Model
por: Leung, Haun, et al.
Publicado: (2026)

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

The Devil is in the Statistics: Mitigating and Exploiting Statistics Difference for Generalizable Semi-supervised Medical Image Segmentation
por: Qiu, Muyang, et al.
Publicado: (2024)

Unleashing the Power of Intermediate Domains for Mixed Domain Semi-Supervised Medical Image Segmentation
por: Ma, Qinghe, et al.
Publicado: (2025)

Steady Progress Beats Stagnation: Mutual Aid of Foundation and Conventional Models in Mixed Domain Semi-Supervised Medical Image Segmentation
por: Ma, Qinghe, et al.
Publicado: (2025)