Cross-Attention Watermarking of Large Language Models

Abstract

A new approach to linguistic watermarking of language models is presented in which information is imperceptibly inserted into the output text while preserving its readability and original meaning. A cross-attention mechanism is used to embed watermarks in the text during inference. Two methods using cross-attention are presented that minimize the effect of watermarking on the performance of a pretrained model. Exploration of different training strategies for optimizing the watermarking and of the challenges and implications of applying this approach in real-world scenarios clarified the tradeoff between watermark robustness and text quality. Watermark selection substantially affects the generated output for high entropy sentences. This proactive watermarking approach has potential application in future model development.

arXiv comment5 pages, 3 figures. Accepted to ICASSP 2024

References in corpus (9)

• Efficient Estimation of Word Representations in Vector Space
• Categorical Reparameterization with Gumbel-Softmax
• Flamingo: a Visual Language Model for Few-Shot Learning
• Can AI-Generated Text be Reliably Detected?
• The Curse of Recursion: Training on Generated Data Makes Models Forget
• Generative Language Models and Automated Influence Operations: Emerging Threats and Potential Mitigations
• A Watermark for Large Language Models
• Undetectable Watermarks for Language Models
• DeepTextMark: A Deep Learning-Driven Text Watermarking Approach for Identifying Large Language Model Generated Text