Munawar Hayat

FLoC: Facility Location-Based Efficient Visual Token Compression for Long Video Understanding

DisCo: Reinforcement with Diversity Constraints for Multi-Human Generation

Attention Guided Alignment in Efficient Vision-Language Models

DEVA: Fine-tuning Multimodal Large Language Models for Visual Perception Tasks

Generalized Contrastive Learning for Universal Multimodal Retrieval

EraseDiff: Erasing Data Influence in Diffusion Models

DDIL: Improved Diffusion Distillation with Imitation Learning