The Truth is in There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction

Pratyusha Sharma, Jordan T. Ash, Dipendra Misra

Transformer-based Large Language Models (LLMs) have become a fixture in modern machine learning. Correspondingly, significant resources are allocated towards research that aims to…