Can grounding the NLI task into a different modality and leveraging existing pre-trained models through zero-shot capabilities improve efficiency and reduce computational demands while maintaining or enhancing task performance?

2024

Santeer, Ayman