WAPITI: A Watermark for Finetuned Open-Source LLMs
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917859361292288 |
|---|---|
| author | Chen, Lingjie Qiu, Ruizhong Yuan, Siyu Liu, Zhining Wei, Tianxin Yoo, Hyunsik Zeng, Zhichen Yang, Deqing Tong, Hanghang |
| author_facet | Chen, Lingjie Qiu, Ruizhong Yuan, Siyu Liu, Zhining Wei, Tianxin Yoo, Hyunsik Zeng, Zhichen Yang, Deqing Tong, Hanghang |
| contents | Watermarking of large language models (LLMs) generation embeds an imperceptible statistical pattern within texts, making it algorithmically detectable. Watermarking is a promising method for addressing potential harm and biases from LLMs, as it enables traceability, accountability, and detection of manipulated content, helping to mitigate unintended consequences. However, for open-source models, watermarking faces two major challenges: (i) incompatibility with fine-tuned models, and (ii) vulnerability to fine-tuning attacks. In this work, we propose WAPITI, a new method that transfers watermarking from base models to fine-tuned models through parameter integration. To the best of our knowledge, we propose the first watermark for fine-tuned open-source LLMs that preserves their fine-tuned capabilities. Furthermore, our approach offers an effective defense against fine-tuning attacks. We test our method on various model architectures and watermarking strategies. Results demonstrate that our method can successfully inject watermarks and is highly compatible with fine-tuned models. Additionally, we offer an in-depth analysis of how parameter editing influences the watermark strength and overall capabilities of the resulting models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_06467 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | WAPITI: A Watermark for Finetuned Open-Source LLMs Chen, Lingjie Qiu, Ruizhong Yuan, Siyu Liu, Zhining Wei, Tianxin Yoo, Hyunsik Zeng, Zhichen Yang, Deqing Tong, Hanghang Cryptography and Security Watermarking of large language models (LLMs) generation embeds an imperceptible statistical pattern within texts, making it algorithmically detectable. Watermarking is a promising method for addressing potential harm and biases from LLMs, as it enables traceability, accountability, and detection of manipulated content, helping to mitigate unintended consequences. However, for open-source models, watermarking faces two major challenges: (i) incompatibility with fine-tuned models, and (ii) vulnerability to fine-tuning attacks. In this work, we propose WAPITI, a new method that transfers watermarking from base models to fine-tuned models through parameter integration. To the best of our knowledge, we propose the first watermark for fine-tuned open-source LLMs that preserves their fine-tuned capabilities. Furthermore, our approach offers an effective defense against fine-tuning attacks. We test our method on various model architectures and watermarking strategies. Results demonstrate that our method can successfully inject watermarks and is highly compatible with fine-tuned models. Additionally, we offer an in-depth analysis of how parameter editing influences the watermark strength and overall capabilities of the resulting models. |
| title | WAPITI: A Watermark for Finetuned Open-Source LLMs |
| topic | Cryptography and Security |
| url | https://arxiv.org/abs/2410.06467 |