How Do Large Language Models Llms Understand The Content They Process