Defender Robustness on CHINESE Scam Simulation Victim Models
11.6Success Ratellama-3.1-405b-instruct
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| llama-3.1-405b-instructTOTAL=11192026.01 | 11.6 | 15.5 | 45.2 | 27.6 | |
| qwen-max-latestTOTAL=11202026.01 | 8.7 | 25.6 | 31.7 | 33.9 | |
| grok-4TOTAL=11192026.01 | 4.2 | 33.9 | 35.5 | 26.5 | |
| glm-4.6-thinkingTOTAL=11192026.01 | 3.6 | 42.6 | 25.1 | 28.7 | |
| ERNIE-4.0-8KTOTAL=11202026.01 | 1.8 | 32.1 | 38.8 | 27.2 | |
| SparkDesk-v3.5TOTAL=11192026.01 | 1.7 | 38.7 | 27.8 | 31.8 | |
| deepseek-r1TOTAL=11202026.01 | 1.1 | 45.4 | 24.2 | 29.4 | |
| claude-sonnet-4-5-thinkingTOTAL=11192026.01 | 0.4 | 69.3 | 8.5 | 21.8 |