Optimized Couplings For Watermarking Large Language Models

Carol Xuan Long, Dor Tsur, Claudio Mayrink Verdun, Hsiang Hsu, Haim H. Permuter, Flavio Calmon

Track: long paper (up to 9 pages)

Keywords: large language model watermarking, information theory

TL;DR: We characterize the trade-off between watermark detection and distortion with shared side information, and optimize the coupling between shared side information and vocabulary partition and the randomization strategy under a worst-case distribution.

Abstract: Large-language models (LLMs) are now able to produce text that is indistinguishable from human-generated content. This has fueled the development of watermarks that imprint a ``signal'' in LLM-generated text with minimal perturbation of an LLM's output. This paper provides an analysis of text watermarking in a one-shot setting. Through the lens of hypothesis testing with side information, we formulate and analyze the fundamental trade-off between watermark detection power and distortion in generated textual quality. We argue that a key component in watermark design is generating a coupling between the side information shared with the watermark detector and a random partition of the LLM vocabulary. Our analysis identifies the optimal coupling and randomization strategy under the worst-case LLM next-token distribution that satisfies a min-entropy constraint. We provide a closed-form expression of the resulting detection rate under the proposed scheme and quantify the cost in a max-min sense. Finally, we numerically compare the proposed scheme with the theoretical optimum.

Presenter: ~Carol_Xuan_Long1

Format: Maybe: the presenting author will attend in person, contingent on other factors that still need to be determined (e.g., visa, funding).

Funding: Yes, the presenting author of this submission falls under ICLR’s funding aims, and funding would significantly impact their ability to attend the workshop in person.

Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.

Submission Number: 34