Explaining and Tuning Transformer-based LLMs in Arithmetic Tasks with Human Strategies

2026-07-19

AI Summary

This work addresses the poor performance of Transformer-based large language models (LLMs) on basic arithmetic tasks, which hinders their reliable deployment in high-stakes scenarios. The authors propose a novel training framework that integrates human-inspired learning strategies with cognitive enhancement mechanisms. By incorporating task decomposition, loss convergence ordering, and transfer of cognitive strategies, the framework optimizes model training while leveraging explainable AI (XAI) techniques for visual analysis. This study presents the first systematic integration of human arithmetic learning principles into LLM training, significantly improving integer arithmetic accuracy and revealing striking parallels between model and human learning patterns. These findings provide empirical support for enhancing model trustworthiness through cognitively grounded approaches.

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.

Problem

• arithmetic tasks
• large language models
• model reliability
• Transformer-based LLMs
• human learning strategies

Innovation

• human-inspired learning
• arithmetic reasoning
• explainable AI
• Transformer-based LLMs
• cognitive strategies