PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
June 12, 20241 citationsOpen Access

Watermarking Language Models with Error Correcting Codes

View Full Paper
PCPatrick ChaoOpenAI (United States)EDEdgar DobribanCalifornia University of PennsylvaniaHHHamed HassaniMoscow Institute of Thermal Technology

Key Points

Key points are not available for this paper at this time.

Abstract

Recent progress in large language models enables the creation of realistic machine-generated content. Watermarking is a promising approach to distinguish machine-generated text from human text, embedding statistical signals in the output that are ideally undetectable to humans. We propose a watermarking framework that encodes such signals through an error correcting code. Our method, termed robust binary code (RBC) watermark, introduces no distortion compared to the original probability distribution, and no noticeable degradation in quality. We evaluate our watermark on base and instruction fine-tuned models and find our watermark is robust to edits, deletions, and translations. We provide an information-theoretic perspective on watermarking, a powerful statistical test for detection and for generating p-values, and theoretical guarantees. Our empirical findings suggest our watermark is fast, powerful, and robust, comparing favorably to the state-of-the-art.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Chao et al. (2024) studied this question.

synapsesocial.com/papers/68e651cbb6db6435875e2875https://doi.org/10.48550/arxiv.2406.10281
Ask AI
Helpful
Bookmark
Share
View Full Paper

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Edit Distance Robust Watermarks via Indexing Pseudorandom Codes2024 · 1 citations
  2. 2Multi-Bit Distortion-Free Watermarking for Large Language Models2024 · 1 citations
  3. 3Improving Detection of Watermarked Language Models2025
  4. 4Optimized Couplings for Watermarking Large Language Models2025
  5. 5The Coding Limits of Robust Watermarking for Generative Models2026