用三色词表分区法给大模型输出加水印，检测准且不伤可读性。
HATS: High-Accuracy Triple-Set Watermarking for Large Language Models
- 每步将词汇表分为绿黄红三区，采样仅限绿黄区。
- 在固定误报率下检测准确率高，文本质量不受影响。
- 适合需要溯源生成内容的场景，如防伪与版权保护。
大语言模型生成内容的滥用可通过水印技术加以遏制，该技术在输出中嵌入隐式信号。本文提出一种三重集合水印方法，每个解码步骤将词汇表划分为绿、黄、红三组，比例固定，并限制采样仅在绿区和黄区进行。检测时重复相同划分，计算绿区富集与红区耗竭统计量，转换为单侧z分数，通过Fisher方法聚合p值以判断文本是否含水印。我们在Llama 2 7B上实现了生成、检测与测试，评估了真阳性率、假阳性率及文本质量。结果表明，三分区方案在固定假阳性率下实现高检测精度，同时保持良好可读性。
原文摘要 · Abstract (English)
Misuse of LLM-generated text can be curbed by watermarking techniques that embed implicit signals into the output. We propose a watermark that partitions the vocabulary at each decoding step into three sets (Green/Yellow/Red) with fixed ratios and restricts sampling to the Green and Yellow sets. At detection time, we replay the same partitions, compute Green-enrichment and Red-depletion statistics, convert them to one-sided z-scores, and aggregate their p-values via Fisher's method to decide whether a passage is watermarked. We implement generation, detection, and testing on Llama 2 7B, and evaluate true-positive rate, false-positive rate, and text quality. Results show that the triple-partition scheme achieves high detection accuracy at fixed FPR while preserving readability.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。