Explaining and Tuning Transformer-based LLMs in Arithmetic Tasks with Human Strategies
Machine Learning
Summary
The authors studied how Transformer-based language models learn arithmetic tasks and found that these models learn simpler problems faster, similar to how humans learn. They improved the models' arithmetic accuracy by breaking tasks into smaller parts and using learning strategies that help people solve problems better. Their work shows that these language models might process arithmetic in ways resembling human thinking. They also used tools to explain and verify how the models learn, aiming to build more trust in these AI systems for important uses.
Authors
Luyu Qiu, Jianing Li, Hwanhee Kim, Xiaoyong Wei, Yueyuan Zheng, Janet Hsiao, Lei Chen
Abstract
Transformer-based large language models (LLMs) continue to achieve state-of-the-art performance across various natural language processing tasks. However, their subpar performance on seemingly elementary problems, such as basic arithmetic, raises concerns about model reliability, safety, and ethical deployment. In this study, we demonstrate that the performance of a vanilla Transformer model trained on integer arithmetic tasks can be improved using methods effective for human learners. We begin by decomposing the arithmetic task into well-defined subtasks and conducting loss convergence order analysis together with ablation studies for each subtask. Our findings reveal that LLMs exhibit learning patterns similar to those of human learners, with a faster learning speed for simpler subtasks compared to more complex ones. In addition, we successfully improved the accuracy of LLMs by applying problem-solving strategies and cognitive empowerment methods shown to enhance the performance of human learners. This suggests that transformer-based LLMs may share cognitive processes with human learners in arithmetic. Lastly, we provide a comprehensive demonstration of our method's effectiveness, including significant accuracy improvement experiments, visualization verification, and explanation-based analysis to illuminate the intricacies of LLMs in arithmetic learning. In general, this work explores the potential similarities between transformer-based LLMs and human learners, supported by explainable AI (XAI) verifications, ultimately fostering trust in LLMs for critical and high-stakes applications.