| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Arena-Hard | TAG-INSTRUCT | Score22.1 | 46 | 3mo ago | |
| Arena-Hard v2 | Score85.9 | 37 | 3mo ago | ||
| IFBench | Multi-DPOP | Accuracy84.1 | 30 | 22d ago | |
| WildBench | GenRM-R-Align-14B | Score92.6 | 19 | 3mo ago | |
| WebArena random 100-sample | SiRA | Success Rate23 | 3 | 12d ago | |
| MMLU style (test) | Student (Self-Distillation) | Accuracy51.06 | 3 | 3mo ago |