Scholarlink

In recent years, Large Language Models (LLMs) have shown an impressive capability to exchange natural language with a human user. In this paper, we explore their capacity to construct a support and an attack relation from input arguments. To do so, we propose an evaluation method based on the Twelve Angry Men example. Then, we discuss the results as well as the discrepancy in accuracy among the different models. Finally, we propose future works.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Assessing the Robustness of LLMs in Predicting Supports and Attacks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Benoît Alcaraz,
  • Aria Nourbakhsh,
  • Liuwen Yu

摘要

In recent years, Large Language Models (LLMs) have shown an impressive capability to exchange natural language with a human user. In this paper, we explore their capacity to construct a support and an attack relation from input arguments. To do so, we propose an evaluation method based on the Twelve Angry Men example. Then, we discuss the results as well as the discrepancy in accuracy among the different models. Finally, we propose future works.