| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Steering Evaluation Concepts Llama 3.3-70B (test) | Darkness Centric | Granularity1.2331 | 20 | 2mo ago | |
| Gemma 2 2B Steering Evaluation Set | Numerical steering | Granularity1.2961 | 20 | 2mo ago | |
| 10-task steering suite (excluding JailbreakBench) (test) | CLAS | Accuracy88.4 | 16 | 2mo ago | |
| WaterBirds (test) | ViSAE | Worst Group Accuracy98.5 | 15 | 1mo ago | |
| Steering Evaluation Suite Power, Wealth, Corr, TQA Gemma-2-9B-IT (test) | Comm Steer | Power2.61 | 10 | 2mo ago | |
| TRAITFACTORY | ORBIT | Min Per-Attribute Improvement2.67 | 9 | 1mo ago | |
| TONEBANK | ORBIT | Min Improvement2.89 | 9 | 1mo ago |