Language Models Coupled with Metacognition Can Outperform Reasoning Models

Vedant Khandelwal, Francesca Rossi, Keerthiram Murugesan +4

Large language models (LLMs) excel in speed and adaptability across various reasoning tasks, but they often struggle when strict logic or constraint enforcement is required. In con…