Image Generation On Ffhq 1024 X 1024
评估指标
FID
评测结果
各个模型在此基准测试上的表现结果
比较表格
| 模型名称 | FID |
|---|---|
| alias-free-generative-adversarial-networks | 2.79 |
| diffusion-gan-training-gans-with-diffusion | 2.83 |
| very-deep-vaes-generalize-autoregressive-1 | - |
| efficient-vdvae-less-is-more | - |
| adversarial-latent-autoencoders | 13.09 |
| swagan-a-style-based-wavelet-driven | 4.06 |
| msg-gan-multi-scale-gradients-gan-for-more | 5.8 |
| polarity-sampling-quality-and-diversity | 2.57 |
| a-style-based-generator-architecture-for | 4.4 |
| stylenat-giving-each-head-a-new-perspective | 4.17 |
| alias-free-generative-adversarial-networks | 3.07 |
| analyzing-and-improving-the-image-quality-of | 2.84 |
| styleswin-transformer-based-gan-for-high-1 | 5.07 |
| adversarially-slicing-generative-networks | 1.61 |
| image-generators-with-conditionally | 10.07 |
| magnet-uniform-sampling-from-deep-generative-1 | 2.66 |
| stylegan-xl-scaling-stylegan-to-large-diverse | 2.02 |
| feature-quantization-improves-gan-training | 3.19 |
| improved-transformer-for-high-resolution-gans | 6.37 |
| training-generative-adversarial-networks-with-2 | 3.62 |