Can I understand what I create? Self-Knowledge Evaluation of Large Language Models

TL;DR

This paper introduces a self-knowledge evaluation framework for large language models, assessing their ability to understand and respond to self-generated questions, revealing significant gaps and potential for improvement.

Contribution

The paper presents a novel, easy-to-implement self-knowledge evaluation method for LLMs, highlighting their limitations and potential enhancement strategies.

Findings

Models show significant gaps in self-knowledge ability.

Misalignment with human attention may cause these gaps.

Fine-tuning on self-generated math tasks improves performance.

Abstract

Large language models (LLMs) have achieved remarkable progress in linguistic tasks, necessitating robust evaluation frameworks to understand their capabilities and limitations. Inspired by Feynman's principle of understanding through creation, we introduce a self-knowledge evaluation framework that is easy to implement, evaluating models on their ability to comprehend and respond to self-generated questions. Our findings, based on testing multiple models across diverse tasks, reveal significant gaps in the model's self-knowledge ability. Further analysis indicates these gaps may be due to misalignment with human attention mechanisms. Additionally, fine-tuning on self-generated math task may enhance the model's math performance, highlighting the potential of the framework for efficient and insightful model evaluation and may also contribute to the improvement of LLMs.