Multi-Objective Generation Evaluation on HH-RLHF (test)
1.24HypervolumeBone Soup
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Bone SoupBackbone=LLama-2 7B2025.02 | 1.24 | 2.11 | 100 | 11 | 0.33 | 0.07 | |
| Bone Soup2025.02 | 1.16 | 2 | 1 | — | — | — | |
| MODBackbone=LLama-2 7B2025.02 | 1.08 | 1.83 | 100 | 11 | 0.24 | 0.02 | |
| Rewarded SoupsBackbone=LLama-2 7B2025.02 | 1.06 | 1.7 | 84 | 11 | 0.24 | 0.02 | |
| MODPO2025.02 | 0.93 | -0.57 | 1 | — | — | — | |
| Rewards-in-ContextBackbone=LLama-2 7B2025.02 | 0.45 | 1.09 | 85 | 8 | 0.07 | 0.01 |