Neuro-RIT: Neuron-Guided Instruction Tuning for Robust Retrieval-Augmented Language Model

Jaemin Kim, Jae O Lee, Sumyeong Ahn +1

Retrieval-Augmented Language Models (RALMs) have demonstrated significant potential in knowledge-intensive tasks; however, they remain vulnerable to performance degradation when pr…