arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.17396cs.CLcs.AIcs.NE
收藏

GPT-4在算法问题上的基准测试:提示策略的系统性评估

Benchmarking GPT-4 on Algorithmic Problems: A Systematic Evaluation of Prompting Strategies

University of Padova(帕多瓦大学)

Flavio Petruzzellis, Alberto Testolin, Alessandro Sperduti

AI总结:

本研究针对GPT-4开展算法任务系统性基准测试,对比其与GPT-3.5、Neural Data Router的性能,证实先进提示技术可提升GPT-4在需系统性泛化的任务中的准确率,树立强基线。

摘要:

大语言模型(LLMs)凭借其在海量文本语料上习得的知识可迁移至各类下游任务、且仅需极少(若有)微调步骤的特性,已彻底变革了自然语言处理领域。但与此同时,已有多项研究反复表明LLMs缺乏系统性泛化能力,即无法将习得的统计规律外推至训练分布之外的场景。本研究针对目前最先进的LLMs之一GPT-4,在三类可通过两个参数控制问题难度的算法任务上开展了系统性基准测试。我们将GPT-4的性能与其前代模型GPT-3.5,以及近期被提出用于解决同类任务的Transformer编码器架构变体Neural Data Router(神经数据路由器)进行了对比。研究发现,运用先进的提示技术能让GPT-4在所有任务上达到更优的准确率,证明了当前最先进的LLMs即便在需要系统性泛化能力的挑战性任务中,也能构成极具竞争力的基线。

英文摘要:

Large Language Models (LLMs) have revolutionized the field of Natural Language Processing thanks to their ability to reuse knowledge acquired on massive text corpora on a wide variety of downstream tasks, with minimal (if any) tuning steps. At the same time, it has been repeatedly shown that LLMs lack systematic generalization, which allows to extrapolate the learned statistical regularities outside the training distribution. In this work, we offer a systematic benchmarking of GPT-4, one of the most advanced LLMs available, on three algorithmic tasks characterized by the possibility to control the problem difficulty with two parameters. We compare the performance of GPT-4 with that of its predecessor (GPT-3.5) and with a variant of the Transformer-Encoder architecture recently introduced to solve similar tasks, the Neural Data Router. We find that the deployment of advanced prompting techniques allows GPT-4 to reach superior accuracy on all tasks, demonstrating that state-of-the-art LLMs constitute a very strong baseline also in challenging tasks that require systematic generalization.

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。

↑