arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.14742cs.CL
收藏

探索指令数据缩放对大语言模型的影响:基于真实用例的实证研究

Exploring the Impact of Instruction Data Scaling on Large Language Models: An Empirical Study on Real-World Use Cases

Beike Inc.(贝壳公司)

Yunjie Ji, Yong Deng, Yan Gong, Yiping Peng, Qiang Niu, Lei Zhang, Baochang Ma, Xiangang Li

摘要:

ChatGPT的成功最近吸引了众多的努力来复制它,预防调整策略是取得显着成果的关键因素。指令调整不仅显著提高了模型的性能和泛化能力,而且使模型生成的结果更符合人类语音模式。然而,目前的研究很少研究不同数量的指令数据对模型性能的影响,特别是在现实世界的用例。在本文中,我们探讨了大语言模型的性能的基础上,指令微调跨不同规模的指令数据。实验中构建了由12个主要在线用例组成的评估数据集。以Bloomz-7 B1-mt为基础模型,结果表明:1)仅仅增加指令数据量就能导致开放式生成等任务的持续改进; 2)在数学和代码等任务中,随着数据量的增加,模型性能曲线保持相当平坦。我们进一步分析了这些现象的可能原因,并提出了潜在的未来研究方向,如有效地选择高质量的训练数据,缩放的基础模型和专门用于硬任务的训练方法。我们将发布我们的训练和评估数据集,以及模型检查点。

英文摘要:

The success of ChatGPT has recently attracted numerous efforts to replicate it, with instruction-tuning strategies being a key factor in achieving remarkable results. Instruction-tuning not only significantly enhances the model's performance and generalization but also makes the model's generated results more consistent with human speech patterns. However current research rarely studies the impact of different amounts of instruction data on model performance, especially in the real-world use cases. In this paper we explore the performance of large language models based on instruction tuning across different scales of instruction data. An evaluation dataset consisting of 12 major online use cases is constructed in the experiment. With Bloomz-7B1-mt as the base model, the results show that 1) merely increasing the amount of instruction data leads to continuous improvement in tasks such as open-ended generation, 2) in tasks such as math and code, the model performance curve remains quite flat while increasing data size. We further analyze the possible causes of these phenomena and propose potential future research directions such as effectively selecting high-quality training data, scaling base models and training methods specialized for hard tasks. We will release our training and evaluation datasets, as well as model checkpoints.

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。

↑