PaperLLM水印论文 / 马里兰大学水印论文
A Watermark for Large Language Models
An academic paper published by the University of Maryland in 2023 on text watermarking for large language models. The paper proposes partitioning the vocabulary into green and red token lists and detecting AI-generated text by statistically measuring the frequency of green token occurrences.
Timeline (last 90 days)
Oct 12
The paper "A Watermark for Large Language Models" published by the University of Maryland in 2023 proposes dividing the vocabulary into green words and red words, with green words appearing at a significantly higher frequency in watermarked text than the statistical baseline.
Unverified50%