arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.17366cs.CLcs.AI
收藏

ArgBench:对计算论证任务的LLM基准测试

ArgBench: Benchmarking LLMs on Computational Argumentation Tasks

Leibniz University Hannover(汉诺威莱布尼茨大学);Adobe Research(Adobe研究)

Yamen Ajjour, Carlotta Quensel, Nedim Lipka, Henning Wachsmuth

AI总结:

本文提出ArgBench,首个用于评估LLM在计算论证任务中性能的基准,涵盖33个数据集,评估五种LLM家族在46种任务中的泛化能力,分析少样本示例、推理步骤、模型大小和训练技能对性能的影响。

摘要:

本文提出ArgBench,首个用于评估LLM在计算论证任务中性能的基准,涵盖33个数据集,评估五种LLM家族在46种任务中的泛化能力,分析少样本示例、推理步骤、模型大小和训练技能对性能的影响。

英文摘要:

Argumentation skills are an essential toolkit for large language models (LLMs). These skills are crucial in various use cases, including self-reflection, debating collaboratively for diverse answers, and countering hate speech. In this paper, we create the first benchmark for a standardized evaluation of LLM-based approaches to computational argumentation, encompassing 33 datasets from previous work in unified form. Using the benchmark, we evaluate the generalizability of five LLM families across 46 computational argumentation tasks that cover mining arguments, assessing perspectives, assessing argument quality, reasoning about arguments, and generating arguments. On the benchmark, we conduct an extensive systematic analysis of the contribution of few-shot examples, reasoning steps, model size, and training skills to the performance of LLMs on the computational argumentation tasks in the benchmark.

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。

↑