Sayeed Shafayet Chowdhury
1 indexed paper
Recent (6 mo)
1With code
0Influential cites
0Benchmarked
0Publications per year
126
Top categories
Vision×1AI×1
Frequent co-authors
Research Timeline
2026
Detect Before You Leap: Mirage Detection in Vision-Language Models
The paper introduces Text-Conditioned Layer-wise Internal Alignment (TC-LIA), a model-agnostic method that significantly improves the detection of 'mirage'—when Vision-Language Models confidently answer questions without visual evidence—achieving high accuracy and drastically reducing mirage rates.
Highlighted terms show continued research focus across papers