Text-to-Image Generation on Out-of-Domain Evaluation Set
36.25CLIP ScoreFLUX.1-dev + UnifiedReward-Flex
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FLUX.1-dev + UnifiedReward-FlexBackbone=FLUX.1-dev, Optimization=GRPO, Reward Model=UnifiedReward-Flex, Sampling steps=30, CFG scale=3.52026.02 | 36.25 | 23.42 | 3.31 | 6.56 | |
| FLUX.1-dev + UnifiedReward-ThinkBackbone=FLUX.1-dev, Optimization=GRPO, Reward Model=UnifiedReward-Think, Sampling steps=30, CFG scale=3.52026.02 | 35.85 | 23.38 | 3.27 | 6.53 | |
| FLUX.1-dev + UnifiedRewardBackbone=FLUX.1-dev, Optimization=GRPO, Reward Model=UnifiedReward, Sampling steps=30, CFG scale=3.52026.02 | 34.43 | 23.31 | 3.19 | 6.44 | |
| FLUX.1-devSampling steps=30, CFG scale=3.52026.02 | 34.4 | 22.7 | 3.07 | 6.13 | |
| FLUX.1-dev + HPSv3Backbone=FLUX.1-dev, Optimization=GRPO, Reward Model=HPSv3, Sampling steps=30, CFG scale=3.52026.02 | 34.12 | 23.26 | 3.14 | 6.37 | |
| FLUX.1-dev + PickScoreBackbone=FLUX.1-dev, Optimization=GRPO, Reward Model=PickScore, Sampling steps=30, CFG scale=3.52026.02 | 33.61 | 23.78 | 3.12 | 6.42 | |
| FLUX.1-dev + HPSv2Backbone=FLUX.1-dev, Optimization=GRPO, Reward Model=HPSv2, Sampling steps=30, CFG scale=3.52026.02 | 33.35 | 23.12 | 3.1 | 6.23 |