paper-with-me

홈 › Papers

A Watermark for Large Language Models

2023-01-24 · John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, Tom Goldstein

Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens. We propose a watermarking framework for proprietary language models. The watermark can be embedded with negligible impact on text quality, and can be detected using an efficient open-source algorithm without access to the language model API or parameters. The watermark works by selecting a randomized set of "green" tokens before a word is generated, and then softly promoting use of green tokens during sampling. We propose a statistical test for detecting the watermark with interpretable p-values, and derive an information-theoretic framework for analyzing the sensitivity of the watermark. We test the watermark using a multi-billion parameter model from the Open Pretrained Transformer (OPT) family, and discuss robustness and security.

📄 PDF Abstract BibTeX arXiv:2301.10226

Code (8)

jwkirchenbauer/lm-watermarking 공식 구현 pytorch
BrianPulfer/LMWatermark pytorch
Xieyangxinyu/Unbiased-Watermark-via-Maximal-Coupling pytorch
chengez/adversarial-paraphrasing pytorch
eva-giboulot/watermax pytorch
facebookresearch/three_bricks pytorch
fyyfu/semantic-watermark pytorch
huggingface/text-generation-inference pytorch

Tasks

Language ModelingLanguage Modelling

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Attention 설명 없음
Test 설명 없음
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Adam 설명 없음
Position-Wise Feed-Forward Layer 설명 없음
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.

Similar Papers 제목 키워드 기반

Can Watermarks Survive Translation? On the Cross-lingual Consistency of Text Watermark for Large Language Models

2024-02-21 · Zhiwei He, Binglin Zhou, Hongkun Hao, Aiwei Liu 외

Text watermarking technology aims to tag and identify content produced by large language models (LLMs) to prevent misuse. In this study, we introduce the concept of cross-lingual consistency in text watermarking, which a…

TAG

Baselines for Identifying Watermarked Large Language Models

2023-05-29 · Leonard Tang, Gavin Uberti, Tom Shlomi

We consider the emerging problem of identifying the presence and use of watermarking schemes in widely used, publicly hosted, closed source large language models (LLMs). We introduce a suite of baseline algorithms for id…

Multi-Bit Distortion-Free Watermarking for Large Language Models

2024-02-26 · Massieh Kordi Boroujeny, Ya Jiang, Kai Zeng, Brian Mark

Methods for watermarking large language models have been proposed that distinguish AI-generated text from human-generated text by slightly altering the model output distribution, but they also distort the quality of the …

Decoder

Watermarking LLMs with Weight Quantization

2023-10-17 · Linyang Li, Botian Jiang, Pengyu Wang, Ke Ren 외

Abuse of large language models reveals high risks as large language models are being deployed at an astonishing speed. It is important to protect the model weights to avoid malicious usage that violates licenses of open-…

Language ModelingLanguage ModellingLarge Language ModelQuantization

RTLMarker: Protecting LLM-Generated RTL Copyright via a Hardware Watermarking Framework

2025-01-05 · Kun Wang, Kaiyan Chang, Mengdi Wang, Xinqi Zou 외

Recent advances of large language models in the field of Verilog generation have raised several ethical and security concerns, such as code copyright protection and dissemination of malicious code. Researchers have emplo…