uitnlp/ViANLI

ViANLI（越南语对抗性自然语言推理）是针对越南语NLI的第一个对抗性基准数据集，设计用于评估模型在面对复杂数学现象时的鲁棒性。该数据集通过人类与机器循环的方法，以及多轮对抗性生成和双重人类-机器验证构建而成。ViANLI包含超过10,000个高质量的前提-假设对，跨越13个不同的领域，源自越南新闻文章。每个对被标记为蕴含、矛盾或中立，符合标准NLI框架。ViANLI (Vietnamese Adversarial Natural Language Inference) is the first adversarial benchmark dataset for Vietnamese NLI, designed to evaluate model robustness against complex linguistic phenomena. The dataset was constructed using a human-and-machine-in-the-loop approach with multi-round adversarial generation and dual human–machine verification. ViANLI contains over 10,000 high-quality premise–hypothesis pairs across 13 diverse domains from Vietnamese news articles. Each pair is labeled as entailment, contradiction, or neutral, following the standard NLI framework.