Entropogram-guided Visual Realignment Decoding: Capturing the Rhythm of Multimodal Reasoning
cs.CV, cs.MM
Submitted: 2026-11-04
Updated: 2026-11-04
Code: https://github.com/jden4524/SelfRetrospectiveDecoding
Project page: https://mlrm-lead.github.io
Terminology
Sources
- Qwen3-VL Technical Report
- InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models
- Mind What Matters for Reasoning: Aligning Cross-Modal Attention via Selective Probability Mass Concentration
- VLMEvalKit: An Open-Source Toolkit for Evaluating Large Multi-Modality Models
- Mitigating Hallucination in Large Vision-Language Models via Adaptive Attention Calibration
- Interleaved-Modal Chain-of-Thought
- The Llama 3 Herd of Models
- HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Models
- Think Twice Before You Write -- an Entropy-based Decoding Strategy to Enhance LLM Reasoning
- VisOnlyQA: Large Vision Language Models Still Struggle with Visual Perception of Geometric Information
- Evaluating Object Hallucination in Large Vision-Language Models
- HulluEdit: Single-Pass Evidence-Consistent Subspace Editing for Mitigating Hallucinations in Large Vision-Language Models
- Mitigating Hallucination in Large Multi-Modal Models via Robust Instruction Tuning
- A Survey on Hallucination in Large Vision-Language Models
- Think Just Enough: Sequence-Level Entropy as a Confidence Signal for LLM Reasoning
- RL makes MLLMs see better than SFT
- Latent Chain-of-Thought for Visual Reasoning
- Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs