BriefGPT.xyz
Jun, 2023
从算术任务中学习多步推理
Learning Multi-step Reasoning from Arithmetic Task
HTML
PDF
Tianduo Wang, Wei Lu
TL;DR
本研究探究如何将较小的语言模型与多步推理能力相结合,通过在合成数据集MsAT上进行持续的预训练,我们的实验证明了该方法对增强语言模型的数学推理能力的有效性。
Abstract
mathematical reasoning
is regarded as a necessary ability for
language models
(LMs). Recent works demonstrate large LMs' impressive performance in solving math problems. The success is attributed to their Chain-o
→