Text-to-Image Generation on DrawBench (PickScore, ClipScore, HPSv2.1, Aesthetic, ImgRwd, UniRwd)
23.8PickScoreDiffusionNFT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DiffusionNFT#Iter=1.7k, Resolution=512x5122025.09 | 23.8 | 0.293 | 0.331 | 6.01 | 1.49 | 3.49 | |
| FlowGRPO#Iter=4k, Resolution=512x5122025.09 | 23.5 | 0.28 | 0.316 | 5.9 | 1.29 | 3.37 | |
| SD3.5-LResolution=1024x10242025.09 | 22.91 | 0.289 | 0.288 | 5.5 | 0.96 | 3.25 | |
| FLUX.1-Dev2025.09 | 22.84 | 0.295 | 0.274 | 5.71 | 0.96 | 3.27 | |
| Ctrl-Z‡Backbone=Hunyuan-DiT, NFEs=8.792025.06 | 22.62 | — | 29.79 | 5.908 | 1.319 | — | |
| SOP-4Backbone=Hunyuan-DiT, NFEs=9.002025.06 | 22.58 | — | 29.49 | 5.942 | 1.246 | — | |
| FlowGRPO#Iter=>5k, Resolution=512x5122025.09 | 22.51 | 0.293 | 0.274 | 5.32 | 1.06 | 3.18 | |
| Ctrl-Z†Backbone=Hunyuan-DiT, NFEs=2.852025.06 | 22.51 | — | 29.27 | 5.906 | 1.067 | — | |
| SOP-1Backbone=Hunyuan-DiT, NFEs=3.002025.06 | 22.5 | — | 28.77 | 5.902 | 1.102 | — | |
| ResamplingBackbone=Hunyuan-DiT, NFEs=2.002025.06 | 22.49 | — | 28.7 | 5.912 | 0.904 | — | |
| Z-SamplingBackbone=Hunyuan-DiT, NFEs=3.002025.06 | 22.49 | — | 28.6 | 5.997 | 0.899 | — | |
| DDIMBackbone=Hunyuan-DiT, NFEs=1.002025.06 | 22.47 | — | 28.9 | 5.93 | 0.827 | — | |
| SD-XLResolution=1024x10242025.09 | 22.42 | 0.287 | 0.28 | 5.6 | 0.76 | 2.93 | |
| FlowGRPO#Iter=2k, Resolution=512x5122025.09 | 22.41 | 0.29 | 0.28 | 5.32 | 0.95 | 3.15 | |
| SD3.5-M + CFGCFG=true, Resolution=512x5122025.09 | 22.34 | 0.285 | 0.279 | 5.36 | 0.85 | 3.03 | |
| SOP-4Backbone=Stable Diffusion 2.1, NFEs=9.002025.06 | 21.89 | — | 26.72 | 5.469 | 1.008 | — | |
| Ctrl-Z‡Backbone=Stable Diffusion 2.1, NFEs=7.722025.06 | 21.84 | — | 26.73 | 5.501 | 1.025 | — | |
| Ctrl-Z†Backbone=Stable Diffusion 2.1, NFEs=2.772025.06 | 21.68 | — | 26.15 | 5.476 | 0.65 | — | |
| SOP-1Backbone=Stable Diffusion 2.1, NFEs=3.002025.06 | 21.64 | — | 25.78 | 5.426 | 0.637 | — | |
| ResamplingBackbone=Stable Diffusion 2.1, NFEs=2.002025.06 | 21.52 | — | 25.49 | 5.385 | 0.205 | — | |
| Z-SamplingBackbone=Stable Diffusion 2.1, NFEs=3.002025.06 | 21.47 | — | 25.95 | 5.465 | 0.296 | — | |
| DDIMBackbone=Stable Diffusion 2.1, NFEs=1.002025.06 | 21.39 | — | 24.9 | 5.41 | 0.046 | — | |
| SD3.5-MCFG=false, Resolution=512x5122025.09 | 20.51 | 0.237 | 0.204 | 5.13 | -0.58 | 2.02 | |
| SD 3.5 M + DiffusionNFT#Steps=1.7K, NFE_pi_theta_old=40 + 40, NFE_pi_theta=402026.04 | 0.238 | 0.293 | 0.331 | 6.01 | 1.49 | 3.49 | |
| SD 3.5 M + FlowGRPO#Steps=4K, NFE_pi_theta_old=40, NFE_pi_theta=402026.04 | 0.235 | 0.28 | 0.316 | 5.9 | 1.29 | 3.37 | |
| SD 3.5 M + V-GRPO#Steps=580, NFE_pi_theta_old=40 + 6.9, NFE_pi_theta=6.92026.04 | 0.235 | 0.298 | 0.341 | 6.02 | 1.52 | 3.43 | |
| SD 3.5 LResolution=1024×10242026.04 | 0.2291 | 0.289 | 0.288 | 5.5 | 0.96 | 3.25 | |
| SD 3.5 M + FlowGRPO#Steps=>5K, NFE_pi_theta_old=40, NFE_pi_theta=402026.04 | 0.2251 | 0.293 | 0.274 | 5.32 | 1.06 | 3.18 | |
| SD XLResolution=1024×10242026.04 | 0.2242 | 0.287 | 0.28 | 5.6 | 0.76 | 2.93 | |
| SD 3.5 M + FlowGRPO#Steps=2K, NFE_pi_theta_old=40, NFE_pi_theta=402026.04 | 0.2241 | 0.29 | 0.28 | 5.32 | 0.95 | 3.15 | |
| SD 3.5 M + CFGCFG=true2026.04 | 0.2234 | 0.285 | 0.279 | 5.36 | 0.85 | 3.03 | |
| SD 3.5 M (w/o CFG)CFG=false2026.04 | 0.2051 | 0.237 | 0.204 | 5.13 | -0.58 | 2.02 |