Reward Maximization on SHP
0.53Win RateAISP
Evaluation Results
| Method | Links | |
|---|---|---|
| AISPBase Model=Gemma3 4B, Reward Model=UltraRM2025.10 | 0.53 | |
| AISPBase Model=Gemma3 4B, Reward Model=Eurus2025.10 | 0.527 | |
| AISPBase Model=Llama3 8B, Reward Model=UltraRM2025.10 | 0.513 | |
| AISPBase Model=Llama3 8B, Reward Model=Eurus2025.10 | 0.47 | |
| BoN (top-p)Base Model=Llama3 8B, Reward Model=Eurus2025.10 | 0.453 | |
| BoN (top-p)Base Model=Llama3 8B, Reward Model=UltraRM2025.10 | 0.42 | |
| BoN (top-p)Base Model=Gemma3 4B, Reward Model=UltraRM2025.10 | 0.413 | |
| BoN (top-p)Base Model=Gemma3 4B, Reward Model=Eurus2025.10 | 0.39 | |
| AISPBase Model=Vicuna 7B, Reward Model=Eurus2025.10 | 0.36 | |
| AISPBase Model=Vicuna 7B, Reward Model=UltraRM2025.10 | 0.353 | |
| BoN (top-p)Base Model=Vicuna 7B, Reward Model=UltraRM2025.10 | 0.343 | |
| BoN (top-p)Base Model=Vicuna 7B, Reward Model=Eurus2025.10 | 0.28 |