arXiv 2408.069042025-06-19cs.CL

Re-TASK：从能力、技能和知识视角重新审视LLM任务

Re-TASK: Revisiting LLM Tasks from Capability, Skill, and Knowledge Perspectives

Huawei Technologies Ltd.（华为技术有限公司）；Nankai University（南开大学）；Xi’an Jiaotong University（西安交通大学）；Shanghai Jiaotong University（上海交通大学）

Zhihu Wang, Shiwan Zhao, Yu Wang, Heyuan Huang, Sitao Xie, Yubo Zhang, Jiaxin Shi, Zhixing Wang, Hongyan Li, Junchi Yan

备注：: ACL 2025 Findings; First three authors contributed equally

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

Re-TASK框架从能力、技能和知识视角重新审视LLM任务，提出学习链范式及提示策略，通过知识注入和技能适应增强任务相关能力，在法律任务上显著提升性能。

摘要：

思维链（Chain-of-Thought, CoT）范式已成为使用大型语言模型（LLMs）解决复杂问题的关键方法。然而，其在特定领域任务中的应用仍具挑战性，因为LLMs往往无法准确分解任务或有效执行子任务。本文介绍了Re-TASK框架，一种新颖的理论模型，借鉴布鲁姆分类法和知识空间理论的原理，从能力、技能和知识视角重新审视LLM任务。虽然CoT提供了以工作流为中心的任务视角，但Re-TASK引入了学习链（Chain-of-Learning, CoL）范式，强调任务对特定能力项的依赖，并进一步将其分解为构成性的知识和技能组件。为解决CoT的失败，我们提出了Re-TASK提示策略，通过针对性的知识注入和技能适应来增强与任务相关的能力。跨多个领域的实验证明了Re-TASK的有效性。特别是，在法律任务上，我们在Yi-1.5-9B上实现了45.00%的提升，在Llama3-Chinese-8B上实现了24.50%的提升。这些结果凸显了Re-TASK在显著提升LLM性能及其在专业领域适用性方面的潜力。我们在https://github.com/Uylee/Re-TASK发布了我们的代码和数据。

英文摘要：

The Chain-of-Thought (CoT) paradigm has become a pivotal method for solving complex problems with large language models (LLMs). However, its application to domain-specific tasks remains challenging, as LLMs often fail to decompose tasks accurately or execute subtasks effectively. This paper introduces the Re-TASK framework, a novel theoretical model that revisits LLM tasks from capability, skill, and knowledge perspectives, drawing on the principles of Bloom's Taxonomy and Knowledge Space Theory. While CoT provides a workflow-centric perspective on tasks, Re-TASK introduces a Chain-of-Learning (CoL) paradigm that highlights task dependencies on specific capability items, further broken down into their constituent knowledge and skill components. To address CoT failures, we propose a Re-TASK prompting strategy, which strengthens task-relevant capabilities through targeted knowledge injection and skill adaptation. Experiments across diverse domains demonstrate the effectiveness of Re-TASK. In particular, we achieve improvements of 45.00% on Yi-1.5-9B and 24.50% on Llama3-Chinese-8B for legal tasks. These results highlight the potential of Re-TASK to significantly enhance LLM performance and its applicability in specialized domains. We release our code and data at https://github.com/Uylee/Re-TASK.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。