从零到英雄：探究符号任务在指导调整中的作用（英文版）

英文标题：From Zero to Hero: Examining the Power of Symbolic Tasks in Instruction Tuning中文摘要：本研究提出一种采用符号任务来增强指令调整的简单而有效方法，在多个基准测试中验证该方法可以提高语言模型的零 - shot 表推理能力，并证明符号任务不会影响语言模型的普适性。英文摘要：Fine-tuning language models on tasks with instructions has demonstratedpotential in facilitating zero-shot generalization to unseen tasks. In thispaper, we introduce a straightforward yet effective method for enhancinginstruction tuning by employing symbolic tasks. Compared to crowdsourced humantasks or model-generated