Research map / Image generation / Fast diffusion sampling
Exploring Guided Sampling of Conditional GANs
Yifei Zhang, Mengfei Xia, Yujun Shen, Jiapeng Zhu, Ceyuan Yang, Kecheng Zheng, Lianghua Huang, Yu Liu, Fan Cheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add training-free latent-space guidance | Aurora · ImageNet 64x64 | FID: 8.87 -> 6.06 | Improved |
| train with an additional empty condition (learning-based variant) | Aurora · ImageNet 64x64 | FID: 6.06 -> 4.37 | Improved |
| enlarge opposite condition set from 10 to 1000 (ablation) | training-free GANdance · ImageNet 64x64 | FID: 7.16 -> 6.92 at lambda 1.1 | No clear effect |
| use layer-wise decreasing guidance scale (ablation) | Aurora + GANdance · ImageNet 64x64 | FID: 4.72 -> 4.37 | Improved |
| apply strong guidance to layers above 16x16 resolution (ablation) | Aurora + GANdance · ImageNet 64x64 | FID | Hurt |
| combine guidance with truncation trick (ablation) | Aurora + GANdance · ImageNet 64x64 | FID: 4.48 -> 4.07 | Improved |
| add GANdance to StyleGAN2 | StyleGAN2 · ImageNet 64x64 | FID: 21.32 -> 17.80 | Improved |
| add GANdance to BigGAN by finetuning | BigGAN · ImageNet 128x128 | FID: 10.76 -> 9.07 | Improved |
| raise probability of resetting condition to the additional condition (ablation) | learning-based GANdance · ImageNet | FID | Hurt |
Closest papers by meaning
- Diffusion Models Beat GANs on Image Synthesis (NeurIPS 2021 · similarity 0.92)
- Rectifying GAN Gradient Supervision Via Score Functions (TPAMI 2026 · similarity 0.91)
- Distilling Diffusion Models into Conditional GANs (ECCV 2024 · similarity 0.90)
- Revisiting Diffusion Models: From Generative Pre-training to One-Step Generation (ICML 2025 · similarity 0.90)
- Progressive Distillation for Fast Sampling of Diffusion Models (ICLR 2022 · similarity 0.90)
- Consistency-GAN: Training GANs with Consistency Model (AAAI 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
