Delta-LLaVA: Base-then-Specialize Alignment for Token-Efficient Vision-Language Models.

Mohamad Zamini, Diksha Shukla

Browse the full WACV paper archive.