Cross-Attention Watermarking of Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Baldassini, Folco Bertini, Nguyen, Huy H., Chang, Ching-Chung, Echizen, Isao
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916198178881536
author Baldassini, Folco Bertini
Nguyen, Huy H.
Chang, Ching-Chung
Echizen, Isao
author_facet Baldassini, Folco Bertini
Nguyen, Huy H.
Chang, Ching-Chung
Echizen, Isao
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2401_06829
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Cross-Attention Watermarking of Large Language Models
Baldassini, Folco Bertini
Nguyen, Huy H.
Chang, Ching-Chung
Echizen, Isao
Computation and Language
Artificial Intelligence
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.
title Cross-Attention Watermarking of Large Language Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2401.06829