Nadir Durrani

Multimodal models struggle with consistency, showing significant judgment discrepancies between text and speech inputs, especially in Arabic contexts.

Manipulating just a few sparse neurons can steer Arabic LLMs to generate dialect-specific outputs without the need for extensive fine-tuning.