Saved in:
Bibliographic Details
Main Authors: Mao, Minjia, Wei, Dongjun, Chen, Zeyu, Fang, Xiao, Chau, Michael
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2405.14604
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913683631767552
author Mao, Minjia
Wei, Dongjun
Chen, Zeyu
Fang, Xiao
Chau, Michael
author_facet Mao, Minjia
Wei, Dongjun
Chen, Zeyu
Fang, Xiao
Chau, Michael
contents Recent advancements in large language models (LLMs) have highlighted the risk of misusing them, raising the need for accurate detection of LLM-generated content. In response, a viable solution is to inject imperceptible identifiers into LLMs, known as watermarks. Our research extends the existing watermarking methods by proposing the novel Sampling One Then Accepting (STA-1) method. STA-1 is an unbiased watermark that preserves the original token distribution in expectation and has a lower risk of producing unsatisfactory outputs in low-entropy scenarios compared to existing unbiased watermarks. In watermark detection, STA-1 does not require prompts or a white-box LLM, provides statistical guarantees, demonstrates high efficiency in detection time, and remains robust against various watermarking attacks. Experimental results on low-entropy and high-entropy datasets demonstrate that STA-1 achieves the above properties simultaneously, making it a desirable solution for watermarking LLMs. Implementation codes for this study are available online.
format Preprint
id arxiv_https___arxiv_org_abs_2405_14604
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Watermarking Low-entropy Generation for Large Language Models: An Unbiased and Low-risk Method
Mao, Minjia
Wei, Dongjun
Chen, Zeyu
Fang, Xiao
Chau, Michael
Computation and Language
Recent advancements in large language models (LLMs) have highlighted the risk of misusing them, raising the need for accurate detection of LLM-generated content. In response, a viable solution is to inject imperceptible identifiers into LLMs, known as watermarks. Our research extends the existing watermarking methods by proposing the novel Sampling One Then Accepting (STA-1) method. STA-1 is an unbiased watermark that preserves the original token distribution in expectation and has a lower risk of producing unsatisfactory outputs in low-entropy scenarios compared to existing unbiased watermarks. In watermark detection, STA-1 does not require prompts or a white-box LLM, provides statistical guarantees, demonstrates high efficiency in detection time, and remains robust against various watermarking attacks. Experimental results on low-entropy and high-entropy datasets demonstrate that STA-1 achieves the above properties simultaneously, making it a desirable solution for watermarking LLMs. Implementation codes for this study are available online.
title Watermarking Low-entropy Generation for Large Language Models: An Unbiased and Low-risk Method
topic Computation and Language
url https://arxiv.org/abs/2405.14604