Safety Classification on UnsafeBench (ECE, NLL)
0.061ECELlama
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LlamaPrompt Selection Protocol=Mean2026.05 | 0.061 | 0.607 | |
| InternVLPrompt Selection Protocol=Mean2026.05 | 0.086 | 0.566 | |
| MiniCPMPrompt Selection Protocol=Selected2026.05 | 0.09 | 0.641 | |
| MiniCPMPrompt Selection Protocol=Random2026.05 | 0.097 | 0.678 | |
| Qwen30BPrompt Selection Protocol=Selected2026.05 | 0.099 | 0.629 | |
| InternVLPrompt Selection Protocol=Selected2026.05 | 0.117 | 0.678 | |
| MiniCPMPrompt Selection Protocol=Mean2026.05 | 0.119 | 0.608 | |
| Qwen30BPrompt Selection Protocol=Mean2026.05 | 0.12 | 0.623 | |
| Qwen3-VL 4BPrompt Selection Protocol=Mean2026.05 | 0.123 | 0.689 | |
| Qwen3-VL 8BPrompt Selection Protocol=Mean2026.05 | 0.125 | 0.623 | |
| GemmaPrompt Selection Protocol=Mean2026.05 | 0.129 | 0.654 | |
| LlamaPrompt Selection Protocol=Selected2026.05 | 0.13 | 0.696 | |
| Qwen30BPrompt Selection Protocol=Random2026.05 | 0.192 | 0.898 | |
| InternVLPrompt Selection Protocol=Random2026.05 | 0.195 | 0.857 | |
| Qwen3-VL 4BPrompt Selection Protocol=Selected2026.05 | 0.22 | 1.198 | |
| Qwen3-VL 4BPrompt Selection Protocol=Random2026.05 | 0.229 | 1.463 | |
| Qwen3-VL 8BPrompt Selection Protocol=Random2026.05 | 0.241 | 1.612 | |
| LlamaPrompt Selection Protocol=Random2026.05 | 0.264 | 0.954 | |
| GemmaPrompt Selection Protocol=Random2026.05 | 0.268 | 2.061 | |
| Qwen3-VL 8BPrompt Selection Protocol=Selected2026.05 | 0.366 | 1.267 | |
| GemmaPrompt Selection Protocol=Selected2026.05 | 0.391 | 1.765 |