WaterBench: Towards Holistic Evaluation of Watermarks for Large Language Models

November 13, 2023 ยท Entered Twilight ยท ๐Ÿ› Annual Meeting of the Association for Computational Linguistics

๐Ÿ’ค TWILIGHT: Eternal Rest
Repo abandoned since publication

Repo contents: .gitignore, LICENSE, README.md, config, convert_llama_weights_to_hf.py, data, detect.py, detect_human.py, eval.py, generate.py, llama_flash_attn_monkey_patch.py, metrics.py, pictures, plot_image_and_tables.ipynb, pred.py, process, requirements.txt, shell, watermark

Authors Shangqing Tu, Yuliang Sun, Yushi Bai, Jifan Yu, Lei Hou, Juanzi Li arXiv ID 2311.07138 Category cs.CL: Computation & Language Cross-listed cs.AI Citations 22 Venue Annual Meeting of the Association for Computational Linguistics Repository https://github.com/THU-KEG/WaterBench โญ 30 Last Checked 1 month ago
Abstract
To mitigate the potential misuse of large language models (LLMs), recent research has developed watermarking algorithms, which restrict the generation process to leave an invisible trace for watermark detection. Due to the two-stage nature of the task, most studies evaluate the generation and detection separately, thereby presenting a challenge in unbiased, thorough, and applicable evaluations. In this paper, we introduce WaterBench, the first comprehensive benchmark for LLM watermarks, in which we design three crucial factors: (1) For benchmarking procedure, to ensure an apples-to-apples comparison, we first adjust each watermarking method's hyper-parameter to reach the same watermarking strength, then jointly evaluate their generation and detection performance. (2) For task selection, we diversify the input and output length to form a five-category taxonomy, covering $9$ tasks. (3) For evaluation metric, we adopt the GPT4-Judge for automatically evaluating the decline of instruction-following abilities after watermarking. We evaluate $4$ open-source watermarks on $2$ LLMs under $2$ watermarking strengths and observe the common struggles for current methods on maintaining the generation quality. The code and data are available at https://github.com/THU-KEG/WaterBench.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Computation & Language

๐ŸŒ… ๐ŸŒ… Old Age

Attention Is All You Need

Ashish Vaswani, Noam Shazeer, ... (+6 more)

cs.CL ๐Ÿ› NeurIPS ๐Ÿ“š 166.0K cites 8 years ago