Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2601.12376 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908774145458176 |
|---|---|
| author | Raban, Ofek Fetaya, Ethan Chechik, Gal |
| author_facet | Raban, Ofek Fetaya, Ethan Chechik, Gal |
| contents | Watermarking (WM) is a critical mechanism for detecting and attributing AI-generated content. Current WM methods for Large Language Models (LLMs) are predominantly tailored for autoregressive (AR) models: They rely on tokens being generated sequentially, and embed stable signals within the generated sequence based on the previously sampled text. Diffusion Language Models (DLMs) generate text via non-sequential iterative denoising, which requires significant modification to use WM methods designed for AR models. Recent work proposed to watermark DLMs by inverting the process when needed, but suffers significant computational or memory overhead. We introduce Left-Right Diffusion Watermarking (LR-DWM), a scheme that biases the generated token based on both left and right neighbors, when they are available. LR-DWM incurs minimal runtime and memory overhead, remaining close to the non-watermarked baseline DLM while enabling reliable statistical detection under standard evaluation settings. Our results demonstrate that DLMs can be watermarked efficiently, achieving high detectability with negligible computational and memory overhead. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_12376 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LR-DWM: Efficient Watermarking for Diffusion Language Models Raban, Ofek Fetaya, Ethan Chechik, Gal Computation and Language Watermarking (WM) is a critical mechanism for detecting and attributing AI-generated content. Current WM methods for Large Language Models (LLMs) are predominantly tailored for autoregressive (AR) models: They rely on tokens being generated sequentially, and embed stable signals within the generated sequence based on the previously sampled text. Diffusion Language Models (DLMs) generate text via non-sequential iterative denoising, which requires significant modification to use WM methods designed for AR models. Recent work proposed to watermark DLMs by inverting the process when needed, but suffers significant computational or memory overhead. We introduce Left-Right Diffusion Watermarking (LR-DWM), a scheme that biases the generated token based on both left and right neighbors, when they are available. LR-DWM incurs minimal runtime and memory overhead, remaining close to the non-watermarked baseline DLM while enabling reliable statistical detection under standard evaluation settings. Our results demonstrate that DLMs can be watermarked efficiently, achieving high detectability with negligible computational and memory overhead. |
| title | LR-DWM: Efficient Watermarking for Diffusion Language Models |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.12376 |