TY - JOUR
AU - Goldstein, Tom
AB - Abstract: 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.
TI - A Watermark for Large Language Models
JF - Computing Research Repository
DO - 10.48550/arxiv.2301.10226
DA - 2023-01-24
UR - https://www.deepdyve.com/lp/arxiv-cornell-university/a-watermark-for-large-language-models-Fwanp7LQFZ
VL - 2023
IS - 2301
DP - DeepDyve
ER -