Text-to-Image Generation on bias-prone prompts Gender×Race intersectional
0.124Fairness ScoreBiasNeuron
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| BiasNeuronBase Model=Stable Diffusion 32026.07 | 0.124 | 0.71 | 0.737 | 0.22 | |
| VanillaBase Model=FLUX.1-dev2026.07 | 0.135 | 0.712 | 0.735 | 0.238 | |
| VanillaBase Model=Stable Diffusion 32026.07 | 0.156 | 0.707 | 0.749 | 0.27 | |
| FairImagenBase Model=Stable Diffusion 32026.07 | 0.174 | 0.684 | 0.717 | 0.295 | |
| FairImagenBase Model=FLUX.1-dev2026.07 | 0.21 | 0.705 | 0.728 | 0.347 | |
| BiasNeuronBase Model=FLUX.1-dev2026.07 | 0.222 | 0.707 | 0.738 | 0.363 | |
| WeakGuidanceBase Model=FLUX.1-dev2026.07 | 0.286 | 0.651 | 0.728 | 0.44 | |
| FairDiffusionBase Model=FLUX.1-dev2026.07 | 0.323 | 0.673 | 0.742 | 0.485 | |
| FairDiffusionBase Model=Stable Diffusion 32026.07 | 0.452 | 0.56 | 0.504 | 0.558 | |
| FairFlowBase Model=Stable Diffusion 32026.07 | 0.494 | 0.679 | 0.67 | 0.644 | |
| WeakGuidanceBase Model=Stable Diffusion 32026.07 | 0.508 | 0.688 | 0.739 | 0.669 | |
| FairFlowBase Model=FLUX.1-dev2026.07 | 0.556 | 0.687 | 0.71 | 0.706 |