List your company

Safety Defense on I-BeaverTails

12.58ASRSPARD
9.905227.960146.01564.0699May 27, 2026

Evaluation results

MethodLinks
SPARDBackbone=Qwen-2.5-7B-Instruct2026.0512.581.36
LisaBackbone=Qwen-2.5-7B-Instruct2026.0535.642
Qwen-2.5-7B-InstructBackbone=Qwen-2.5-7B-Instruct2026.0536.061.97
SafeGradBackbone=Qwen-2.5-7B-Instruct2026.0541.722.22
SafeInstrBackbone=Qwen-2.5-7B-Instruct2026.0566.253.18
PTSTBackbone=Qwen-2.5-7B-Instruct2026.0570.233.29
SFTBackbone=Qwen-2.5-7B-Instruct2026.0579.453.78