Cover Image

The Unseen Bias: How Norm Discrepancy in Pre-Norm MLLMs Leads to Visual Information Loss

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Li, Bozhou, Xue, Xinda, Yang, Sihan, Shi, Yang, Chen, Xinlong, Guan, Yushuo, Zhang, Yuanxing, Zhang, Wentao
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

GRAN-TED: Generating Robust, Aligned, and Nuanced Text Embedding for Diffusion Models
by: Li, Bozhou, et al.
Published: (2025)

Semantic Routing: Exploring Multi-Layer LLM Feature Weighting for Diffusion Transformers
by: Li, Bozhou, et al.
Published: (2026)

VidBridge-R1: Bridging QA and Captioning for RL-based Video Understanding Models with Intermediate Proxy Tasks
by: Chen, Xinlong, et al.
Published: (2025)

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

ID-Align: RoPE-Conscious Position Remapping for Dynamic High-Resolution Adaptation in Vision-Language Models
by: Li, Bozhou, et al.
Published: (2025)

Efficiency in Focus: LayerNorm as a Catalyst for Fine-tuning Medical Visual Language Pre-trained Models
by: Chen, Jiawei, et al.
Published: (2024)

Mavors: Multi-granularity Video Representation for Multimodal Large Language Model
by: Shi, Yang, et al.
Published: (2025)

NormAUG: Normalization-guided Augmentation for Domain Generalization
by: Qi, Lei, et al.
Published: (2023)

Artifact-Bench: Evaluating MLLMs on Detecting and Assessing the Artifacts of AI-Generated Videos
by: Tang, Yuqi, et al.
Published: (2026)

LatentOmni: Rethinking Omni-Modal Understanding via Unified Audio-Visual Latent Reasoning
by: Dai, Yifan, et al.
Published: (2026)

InfoNorm: Mutual Information Shaping of Normals for Sparse-View Reconstruction
by: Wang, Xulong, et al.
Published: (2024)

Growing Visual Generative Capacity for Pre-Trained MLLMs
by: Wang, Hanyu, et al.
Published: (2025)

AVoCaDO: An Audiovisual Video Captioner Driven by Temporal Orchestration
by: Chen, Xinlong, et al.
Published: (2025)

Small-Large Collaboration: Training-efficient Concept Personalization for Large VLM using a Meta Personalized Small VLM
by: Yang, Sihan, et al.
Published: (2025)

MME-VideoOCR: Evaluating OCR-Based Capabilities of Multimodal LLMs in Video Scenarios
by: Shi, Yang, et al.
Published: (2025)

Parametric $ρ$-Norm Scaling Calibration
by: Zhang, Siyuan, et al.
Published: (2024)

Quaternion Nuclear Norm minus Frobenius Norm Minimization for color image reconstruction
by: Guo, Yu, et al.
Published: (2024)

VisualQuest: A Benchmark for Abstract Visual Reasoning in MLLMs
by: Xiao, Kelaiti, et al.
Published: (2025)

A Multistage Extraction Pipeline for Long Scanned Financial Documents: An Empirical Study in Industrial KYC Workflows
by: Han, Yuxuan, et al.
Published: (2026)

Are Bigger Encoders Always Better in Vision Large Models?
by: Li, Bozhou, et al.
Published: (2024)

CodePercept: Code-Grounded Visual STEM Perception for MLLMs
by: Guan, Tongkun, et al.
Published: (2026)

Matrix Completion Via Reweighted Logarithmic Norm Minimization
by: Wang, Zhijie, et al.
Published: (2025)

EgoNormia: Benchmarking Physical Social Norm Understanding
by: Rezaei, MohammadHossein, et al.
Published: (2025)

LongAV-Compass: Towards Unified Evaluation of Minute-Scale Audio-Visual Generation Across T2AV, I2AV, and V2AV
by: Liu, Tengfei, et al.
Published: (2026)

CrystaL: Spontaneous Emergence of Visual Latents in MLLMs
by: Zhang, Yang, et al.
Published: (2026)

IF-Bench: Benchmarking and Enhancing MLLMs for Infrared Images with Generative Visual Prompting
by: Zhang, Tao, et al.
Published: (2025)

RICO: Improving Accuracy and Completeness in Image Recaptioning via Visual Reconstruction
by: Wang, Yuchi, et al.
Published: (2025)

DiffusionAD: Norm-guided One-step Denoising Diffusion for Anomaly Detection
by: Zhang, Hui, et al.
Published: (2023)

Explore How to Inject Beneficial Noise in MLLMs
by: Zhu, Ruishu, et al.
Published: (2025)

Moment-Video: Diagnosing Temporal Fidelity of Video MLLMs on Momentary Visual Events
by: Liu, Xiaolin, et al.
Published: (2026)

Low-Rank Tensor Recovery via Variational Schatten-p Quasi-Norm and Jacobian Regularization
by: Cheng, Zhengyun, et al.
Published: (2025)

Beyond the Last Layer: Multi-Layer Representation Fusion for Visual Tokenization
by: Zhu, Xuanyu, et al.
Published: (2026)

DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States
by: Zhang, Bozhou, et al.
Published: (2024)

ViC-Bench: Benchmarking Visual-Interleaved Chain-of-Thought Capability in MLLMs with Free-Style Intermediate State Representations
by: Wu, Xuecheng, et al.
Published: (2025)

EVE: Verifiable Self-Evolution of MLLMs via Executable Visual Transformations
by: Heng, Yongrui, et al.
Published: (2026)

Unseen Visual Anomaly Generation
by: Sun, Han, et al.
Published: (2024)

Mitigating the Human-Robot Domain Discrepancy in Visual Pre-training for Robotic Manipulation
by: Zhou, Jiaming, et al.
Published: (2024)

Imagine the Unseen: Occluded Pedestrian Detection via Adversarial Feature Completion
by: Zhang, Shanshan, et al.
Published: (2024)

Visual Jigsaw Post-Training Improves MLLMs
by: Wu, Penghao, et al.
Published: (2025)

Bridging Synthetic and Real Worlds for Pre-training Scene Text Detectors
by: Guan, Tongkun, et al.
Published: (2023)