Gholamreza Haffari

Even state-of-the-art MLLMs fail to accurately reconstruct academic documents, revealing a critical gap in machine understanding of scientific knowledge.

EASy achieves superior performance-efficiency trade-offs by intelligently coordinating heterogeneous executors based on their capabilities and costs, reshaping agentic system design.

Unearthing two distinct planning competencies in LLMs reveals that scaling up models enhances operational reasoning but leaves structural enumeration largely unchanged.

Long video generation fails not just because of limited context length, but because of *how* that context is allocated – and ReCA's hierarchical approach shows a way to fix it.

Forget costly training or reward models: MATO unlocks personalized LLM alignment by optimizing objective weights *during* generation, offering unprecedented control and adaptability.

Despite advances in vision-language models, reasoning across sparse, multi-view observations remains surprisingly unsolved, with current models barely outperforming random guessing on a new benchmark.

Fine-tuning LALMs on just the right layers, guided by layer-wise analysis, unlocks better paralinguistic understanding than naively fine-tuning everything.