Vijay Vankadaru

Vijay Vankadaru

MS Data Science, UC Berkeley
Mechanistic interpretability of LLMs and VLMs
Machine Learning Engineer, DASION

Believing Is Seeing: How LLaVA Models Weigh a Contradicting Image

Under reviewICLR 2027

V. Vankadaru, V. DuTell

With Dr. Vasha DuTell. Under review at ICLR 2027.

Cite as: Vankadaru, V., & DuTell, V. Believing Is Seeing: How LLaVA Models Weigh a Contradicting Image. Under review.