Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Dis²Booth: Learning Image Distribution with Disentangled Features for Text-to-Image Diffusion Models
Authors: Guanqi Ding, Chengyu Yang, Shuhui Wang, Xincheng Li, Jinzhe Zhang, Xin Jin, Qingming Huang
AAAI 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments suggest that Dis2Booth can learn the data distribution with higher diversity and complexity and be flexibly applied across various tasks. We have conducted quantitative experiments, qualitative experiments, and downstream classification experiments to validate the effectiveness and superiority of Dis2Booth. |
| Researcher Affiliation | Collaboration | 1University of Chinese Academy of Sciences, Beijing, China 2Beijing Institute of Technology, Beijing, China 3Key Lab of Intell. Info. Process., Inst. of Comput. Tech., CAS, Beijing, China 4Huawei Cloud EI Innovation Lab, China 5Peng Cheng Laboratory, Shenzhen, China |
| Pseudocode | Yes | Algorithm 1: Asynchronous Optimization |
| Open Source Code | Yes | The code can be found at https://github.com/UniBester/Dis2Booth. |
| Open Datasets | Yes | We evaluate our method on Animal Faces (Liu et al. 2019), and Flowers (Nilsback and Zisserman 2008). |
| Dataset Splits | Yes | For the images of the unseen categories, each category s 100 images are divided into Sref and Sreal subsets. Each Sref subset consists of 30 reference images and each Sreal subset consists of 70 images. |
| Hardware Specification | Yes | We conduct experiments on Dis2Booth using Stable Diffusion XL (Podell et al. 2023) as the base model with one NVIDIA A100 GPU, one AMD EPYC 7763 CPU and 100GB Memory. |
| Software Dependencies | No | The paper mentions 'Stable Diffusion XL' as the base model but does not provide specific version numbers for software libraries or dependencies like Python, PyTorch, CUDA, etc., that would be needed to reproduce the experiment. |
| Experiment Setup | Yes | For Dis2Booth, we use λ = 1e 5 and γ = 1e 6 in Eq. 14, α = 0.5 and β = 0.5 in Eq. 4, and k = 5 in Algorithm 1. |