Safety Alignment (Jailbreak Resistance) on JailbreakBench
74JailbreakBench ScoreDSA:LST+:TA
Evaluation Results
| Method | Links | |
|---|---|---|
| DSA:LST+:TAlambda value=02025.05 | 74 | |
| Zephyr LoRA-alignedlambda value=02025.05 | 66 | |
| DSA:LLD:TAlambda value=02025.05 | 55 | |
| Zephyr Base Modellambda value=12025.05 | 15 |