12 papers
VisualOverload: Probing Visual Understanding of VLMs in Really Dense Scenes
Paul Gavrikov, Wei Lin, M. Jehanzeb Mirza +6
Is basic visual understanding really solved in state-of-the-art VLMs? We present VisualOverload, a slightly different visual question answering (VQA) benchmark comprising 2,720 que…
PRISMM-Bench: A Benchmark of Peer-Review Grounded Multimodal Inconsistencies
Lukas Selch, Yufang Hou, M. Jehanzeb Mirza +4
Large Multimodal Models (LMMs) are increasingly applied to scientific research, yet it remains unclear whether they can reliably understand and reason over the multimodal complexit…
Probing the effectiveness of World Models for Spatial Reasoning through Test-time Scaling
Saurav Jha, M. Jehanzeb Mirza, Wei Lin +2
Vision-Language Models (VLMs) remain limited in spatial reasoning tasks that require multi-view understanding and embodied perspective shifts. Recent approaches such as MindJourney…
TTRV: Test-Time Reinforcement Learning for Vision Language Models
Akshit Singh, Shyam Marjit, Wei Lin +7
Existing methods for extracting reward signals in Reinforcement Learning typically rely on labeled data and dedicated training splits, a setup that contrasts with how humans learn…
Can Diffusion Models Disentangle? A Theoretical Perspective
Liming Wang, Muhammad Jehanzeb Mirza, Yishu Gong +6
This paper presents a novel theoretical framework for understanding how diffusion models can learn disentangled representations. Within this framework, we establish identifiability…
Overflow Prevention Enhances Long-Context Recurrent LLMs
Assaf Ben-Kish, Itamar Zimerman, M. Jehanzeb Mirza +4
A recent trend in LLMs is developing recurrent sub-quadratic models that improve long-context processing efficiency. We investigate leading large long-context models, focusing on h…