Semantic Consistency of Vision Tokens: A Vision-Centric Perspective on Multimodal Large Language Models
cs.CV
Submitted: 2026-11-11
Updated: 2026-11-11
Code: https://github.com/djukicn/sgt
Terminology
Sources
- Qwen Technical Report
- Qwen2.5-VL Technical Report
- Language Models are Few-Shot Learners
- What Do Visual Tokens Really Encode? Uncovering Sparsity and Redundancy in Multimodal Large Language Models
- The Llama 3 Herd of Models
- From CLIP to DINO: Visual Encoders Shout in Multi-modal Large Language Models
- Gemma 3 Technical Report
- Gemma: Open Models Based on Gemini Research and Technology
- Qwen2.5 Technical Report
- DINOv3
- OpenAI GPT-5 System Card
- LLaMA: Open and Efficient Foundation Language Models
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