KorSET (Direct Request) - Unfair Representation: leaderboard
Metric: Attack success rate (%) of KorSET's Unfair Representation prompts sent as direct requests (no adversarial optimization) to the target model; a GPT-4.1 judge with the authors' safety rubric decides whether each response is harmful; lower is better. Source: arxiv.org. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 2 9B (IT) | 15.52 | #774 |
| 2 | EXAONE 3.5 7.8B Instruct | 24.54 | #850 |
| 3 | Gemma 3 12B (IT) | 28.47 | #655 |
| 4 | Qwen 2.5 7B Instruct | 38.35 | #846 |
| 5 | Llama 3.1 8B Instruct | 41.34 | #1018 |
Interactive version: theaggregate.ai/benchmark?slug=korset-direct-request-unfair-representation · How It Works · Data refreshed daily, snapshot 2026-10-11.