Red-teaming on HarmBench Llama-3-8B (test)
0.98ASRAGENTICRED
Evaluation Results
| Method | Links | |
|---|---|---|
| AGENTICRED2026.01 | 0.98 | |
| AdvReasoning2026.01 | 0.88 | |
| TransferAttack2026.01 | 0.85 | |
| CoP2026.01 | 0.71 | |
| AutoDAN-Turbo2026.01 | 0.62 |
| Method | Links | |
|---|---|---|
| AGENTICRED2026.01 | 0.98 | |
| AdvReasoning2026.01 | 0.88 | |
| TransferAttack2026.01 | 0.85 | |
| CoP2026.01 | 0.71 | |
| AutoDAN-Turbo2026.01 | 0.62 |