Disentangled Style Domain for Implicit $z$-Watermark Towards Copyright Protection
Authors: Junqiang Huang, Zhaojun Guo, Ge Luo, Zhenxing Qian, Sheng Li, Xinpeng Zhang
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 5 Experiments 5.1 Experimental Setting Datasets and Models. In this paper, we pre-train the style domain encoder [49] and decoder [50] on MS COCO [51]. We conduct experiments on three open-source benchmark datasets (i.e., Celeb A [52], Pokenmon [53], Dreambooth dataset [5]), 17 Artists (e.g, Van Gogh and Monet) and 10 AI artworks (e.g., Ghost Mix and Cat Lora). |
| Researcher Affiliation | Academia | Junqiang Huang Department of Computer Science Fudan University 23210240188@m.fudan.edu.cn Zhaojun Guo Department of Computer Science Fudan University 22110240087@m.fudan.edu.cn Ge Luo Department of Computer Science Fudan University gluo18@fudan.edu.cn Zhenxing Qian Department of Computer Science Fudan University zxqian@fudan.edu.cn Sheng Li Department of Computer Science Fudan University lisheng@fudan.edu.cn Xinpeng Zhang Department of Computer Science Fudan University zhangxinpeng@fudan.edu.cn |
| Pseudocode | No | The paper describes its methods through textual explanations and mathematical equations, such as in Section 3.2 and 3.3, but it does not include a distinct pseudocode block or an algorithm formally labeled as such. |
| Open Source Code | Yes | The code is available at: https://github.com/Hlufies/ZWatermarking |
| Open Datasets | Yes | Datasets and Models. In this paper, we pre-train the style domain encoder [49] and decoder [50] on MS COCO [51]. We conduct experiments on three open-source benchmark datasets (i.e., Celeb A [52], Pokenmon [53], Dreambooth dataset [5]), 17 Artists (e.g, Van Gogh and Monet) and 10 AI artworks (e.g., Ghost Mix and Cat Lora). |
| Dataset Splits | Yes | We have provided all training and testing details in the appendix (such as data splits, hyperparameters, selection methods, optimizer types, etc.). |
| Hardware Specification | Yes | We provide sufficient information on the computer resources required to reproduce each experiment in the appendix of the paper. |
| Software Dependencies | No | The paper states that ‘all training and testing details’ are provided in the appendix, but it does not explicitly list specific software dependencies with their version numbers in the main text, which is required for reproducibility. |
| Experiment Setup | Yes | We have provided all training and testing details in the appendix (such as data splits, hyperparameters, selection methods, optimizer types, etc.). |