Mental Imagery That Matters: Curing Latent Collapse in Multimodal Reasoning
cs.CV, cs.AI
Submitted: 2026-11-07
Updated: 2026-11-07
Project page: https://vision.cs.utexas.edu/projects/AnchorSim
Terminology
Sources
- V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning
- Qwen2.5-VL Technical Report
- Revisiting Feature Prediction for Learning Visual Representations from Video
- Emerging Properties in Unified Multimodal Pretraining
- Interleaved Latent Visual Reasoning with Selective Perceptual Modeling
- GRIT: Teaching MLLMs to Think with Images
- Video-R1: Reinforcing Video Reasoning in MLLMs
- BLINK: Multimodal Large Language Models Can See but Not Perceive
- Think before you speak: Training Language Models With Pause Tokens
- ThinkMorph: Emergent Properties in Multimodal Interleaved Chain-of-Thought Reasoning
- Training Large Language Models to Reason in a Continuous Latent Space
- Can MLLMs Reason in Multimodality? EMMA: An Enhanced MultiModal ReAsoning Benchmark
- DeepEyesV2: Toward Agentic Multimodal Model
- Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models
- GPT-4o System Card
- VLM-R$^3$: Region Recognition, Reasoning, and Refinement for Enhanced Multimodal Chain-of-Thought
- Zebra-CoT: A Dataset for Interleaved Vision Language Reasoning
- Latent Visual Reasoning
- LLaVA-OneVision: Easy Visual Task Transfer
- Imagine while Reasoning in Space: Multimodal Visualization-of-Thought
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