User Study dataset
Benchmarks
TaskDatasetSOTA resultTrendResults
User Study Dataset (100 identity images: 50 synthetic StyleGAN3, 50 real Unsplash) 1.0 (test)
0.333Baseline Selection Rate
6
User study dataset HH-RLHF and PKU-SafeRLHF prompts (test)
3.93Quality Score
4
User Study Dataset fork, gunshot, flower (test)
4.03Magnified Effect
4
User Study Dataset Caricature Synthesis
0.647I-CLIP
3
User Study dataset
90.8Preference Rate
3
User-study dataset SI tasks (Group 1)
14Choice Count
2
User-study dataset ST tasks (Group 1)
16Choice Count
2
User-study dataset CI tasks (Group 1)
14Choice Count
2
User-study dataset CT tasks (Group 1)
13Choice Count
2
User Study dataset real vs. synthesized images
65.5Accuracy
2