| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Refusal benchmark | INNSTEER | Alignment Probability97.81 | 55 | 1mo ago | |
| Llama-3-8B n≈200 | ASR0 | 42 | 2mo ago | ||
| MuSiQue Un | Qwen3-14B | R-Acc91 | 21 | 1mo ago | |
| SaladBench | ICL | Score93.04 | 21 | 4mo ago | |
| Refusal (test) | System Prompt | BAS (Baseline)64.3 | 6 | 18d ago | |
| Financial queries on-topic (test) | Average Appearance Score1 | 2 | 1mo ago | ||
| General queries off-topic (test) | Avg Appearance Score0.91 | 2 | 1mo ago |