Text-to-image generation on GenEval (ImgReward, GenEval, NIQE, CLIP)
1.5214ImgRewardRTS
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RTSBase Model=Qwen-Image, #Params=20B, NFEs=1000, Runtime (s)=260.42026.05 | 1.5214 | 0.8634 | 3.9147 | 0.3061 | |
| RTSBase Model=Flux-1.dev, #Params=12B, NFEs=1000, Runtime (s)=217.72026.05 | 1.4703 | 0.7535 | 3.8506 | 0.2964 | |
| RTSBase Model=SD v3, #Params=2B, NFEs=560, Runtime (s)=38.72026.05 | 1.4636 | 0.7438 | 4.0734 | 0.2936 | |
| FKBase Model=Qwen-Image, #Params=20B, NFEs=1000, Runtime (s)=292.32026.05 | 1.4471 | 0.8109 | 3.9238 | 0.2974 | |
| ZOBase Model=SD v3, #Params=2B, NFEs=560, Runtime (s)=35.92026.05 | 1.3842 | 0.7016 | 4.1742 | 0.2896 | |
| FKBase Model=Flux-1.dev, #Params=12B, NFEs=1000, Runtime (s)=239.12026.05 | 1.3679 | 0.7206 | 3.9587 | 0.2889 | |
| FKBase Model=SD v3, #Params=2B, NFEs=560, Runtime (s)=44.22026.05 | 1.3465 | 0.7126 | 4.181 | 0.2904 | |
| ZOBase Model=Flux-1.dev, #Params=12B, NFEs=1000, Runtime (s)=200.42026.05 | 1.3355 | 0.7168 | 3.9855 | 0.2791 | |
| BoNBase Model=Flux-1.dev, #Params=12B, NFEs=1000, Runtime (s)=196.52026.05 | 1.3192 | 0.7101 | 4.0454 | 0.2878 | |
| BoNBase Model=SD v3, #Params=2B, NFEs=560, Runtime (s)=34.12026.05 | 1.3176 | 0.7054 | 4.1951 | 0.2867 | |
| TTS Free (Qwen-Image)Base Model=Qwen-Image, #Params=20B, NFEs=50, Runtime (s)=11.2s2026.05 | 1.2131 | 0.7692 | 4.0349 | 0.2935 | |
| TTS Free (Flux-1.dev)Base Model=Flux-1.dev, #Params=12B, NFEs=50, Runtime (s)=9.82026.05 | 0.9159 | 0.6516 | 4.217 | 0.2727 | |
| TTS Free (SD v3)Base Model=SD v3, #Params=2B, NFEs=28, Runtime (s)=1.72026.05 | 0.8814 | 0.6165 | 4.3724 | 0.2631 | |
| TTS Free (SD v2.1)Base Model=SD v2.1, #Params=0.86B, NFEs=28, Runtime (s)=1.52026.05 | 0.1779 | 0.3307 | 4.4351 | 0.2701 |