Uncertainty, Not Concealment: A Mechanistic Reappraisal of Sycophancy in Vision-Language Models
cs.CV, cs.LG
Submitted: 2026-11-06
Updated: 2026-11-06
Code: https://github.com/mohamed-eldagla/Uncertainty-Not-Concealment
Terminology
Sources
- Language Models (Mostly) Know What They Know
- Representation Engineering: A Top-Down Approach to AI Transparency
- ContextFocus: Activation Steering for Contextual Faithfulness in Large Language Models
- Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution
- Steering Language Models With Activation Engineering
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