RepreGuard: Detecting LLM-Generated Text by Revealing Hidden Representation Patterns
Xin Chen, Junchao Wu, Shu Yang, Runzhe Zhan et al.
51.20/100
🫥 Mediocre
Incremental, thin
Content 51.2 · Citation bonus +0.0 · no citation data
💡 This paper proposes RepreGuard, an LLM-generated text detection method leveraging discriminative activation feature patterns from LLM hidden representations, achieving an average AUROC of 94.92% acros
#AI生成文本检测#LLM隐式表示分析#OOD鲁棒性#统计特征提取#检测方法优化#AI-generated text detect#LLM implicit representat#OOD robustness#statistical feature extr#detection method optimiz
Score breakdown
Novelty5.0 / 10
Rigor6.0 / 10
Significance7.0 / 10
Clarity8.0 / 10
Reproducibility7.0 / 10
This tone hasn't been generated yet — roast it again to create it.