Ordinal Preference Alignment on OQA
72.13FI Bor.APPA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| APPAModel=llama-3.2-3b2026.04 | 72.13 | 49.9 | 28.5 | 100 | |
| APPAModel=gemma-2-2b2026.04 | 71.41 | 50.3 | 31.5 | 100 | |
| PPO/Bor.Model=gemma-2-2b, Server Agg.=MIN2026.04 | 70.67 | 50 | 31.2 | 100 | |
| APPAModel=Qwen3-0.6B-Instruct2026.04 | 70.37 | 45 | 23 | 100 | |
| PPO/Bor.Model=gemma-2-2b, Server Agg.=AVERAGE2026.04 | 67.87 | 51.8 | 30.9 | 100 | |
| PPO/Bor.Model=llama-3.2-3b, Server Agg.=MIN2026.04 | 67.86 | 43.4 | 29.4 | 100 | |
| PPO/Bor.Model=Qwen3-0.6B-Instruct, Server Agg.=MIN2026.04 | 65.44 | 42.8 | 23.7 | 100 | |
| PPO/Bor.Model=llama-3.2-3b, Server Agg.=AVERAGE2026.04 | 64.99 | 47.5 | 27 | 100 | |
| SFTModel=gemma-2-2b, Server Agg.=-2026.04 | 63.11 | 47.1 | 31.8 | 100 | |
| BASEModel=llama-3.2-3b, Server Agg.=-2026.04 | 62.92 | 36.4 | 23.1 | 91 | |
| SFTModel=llama-3.2-3b, Server Agg.=-2026.04 | 62.84 | 40.2 | 25.3 | 95.8 | |
| PPO/Bor.Model=Qwen3-0.6B-Instruct, Server Agg.=AVERAGE2026.04 | 62.15 | 42.4 | 21 | 100 | |
| SFTModel=Qwen3-0.6B-Instruct, Server Agg.=-2026.04 | 59.06 | 36.6 | 23.9 | 100 | |
| BASEModel=gemma-2-2b, Server Agg.=-2026.04 | 58.98 | 38.3 | 25.5 | 100 | |
| BASEModel=Qwen3-0.6B-Instruct, Server Agg.=-2026.04 | 55.43 | 33.4 | 21.8 | 99.3 |