Human feedback evaluation consistency on RichHF-18K
40.4Pearson-rUnifiedReward_Q(8B)
Evaluation results
| Method | Links | |
|---|---|---|
| UnifiedReward_Q(8B)Model=UnifiedReward_Q(8B)2025.06 | 40.4 | |
| UnifiedReward_L(7B)Model=UnifiedReward_L(7B)2025.06 | 39.9 | |
| Gemini-2.5-ProModel=Gemini-2.5-Pro2025.06 | 39.7 | |
| Qwen3-VL(8B)Model=Qwen3-VL(8B)2025.06 | 38.9 | |
| Minos(8B)Model=Minos(8B)2025.06 | 36 | |
| LLaVA-Critic(72B)Model=LLaVA-Critic(72B)2025.06 | 33 | |
| GPT-4oModel=GPT-4o2025.06 | 31.1 | |
| LLaVA-OV(72B)Model=LLaVA-OV(72B)2025.06 | 27.2 | |
| LLaVA-Critic(7B)Model=LLaVA-Critic(7B)2025.06 | 18.4 | |
| Prometheus-V(7B)Model=Prometheus-V(7B)2025.06 | 8.19 | |
| LLaVA-OV(7B)Model=LLaVA-OV(7B)2025.06 | 5.85 |