| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Sycophancy | iPASa | Mean Improvement51 | 9 | 16d ago | |
| Open-ended questions | Single-Agent | Win Rate68.9 | 6 | 2mo ago | |
| Conflict resolution questions | Single-Agent | Win Rate62.6 | 6 | 2mo ago | |
| Multi-Challenge | Alignment Score0.5652 | 6 | 3mo ago | ||
| Socratic Mind | MAH-DPO | Accuracy71.6 | 5 | 1d ago | |
| Web Implementations (final) | Scalable Interactive Oversight | Alignment Score65.6 | 4 | 3mo ago | |
| TruthfulQA | Relaxed FPO (R̃_FPO) | A Wins188 | 3 | 27d ago | |
| HH-RLHF (test) | SFT + TTL | Reward Model Score65.4 | 2 | 23d ago |