Preference Prediction on HPSv3 (test)
74AccuracyHPSv3
Evaluation results
| Method | Links | |
|---|---|---|
| HPSv3Base Model=Qwen2.5-VL-7B, Category=Fine-Tuned2026.05 | 74 | |
| AutoRubric-T2I (on PickScore)Base Model=Gemini-3-Flash, Judging Protocol=Pointwise2026.05 | 66.8 | |
| AutoRubric-T2I (on HPSv3)Base Model=Qwen3-VL-8B, Judging Protocol=Pointwise2026.05 | 63.9 | |
| AutoRubric-T2I (on PickScore)Base Model=Qwen3-VL-8B, Judging Protocol=Pointwise2026.05 | 61.5 | |
| Qwen3-VL-8BJudging Protocol=Pairwise2026.05 | 60.4 |