List your company

Commonsense Reasoning on CSQA (TSR and ASR)

82.5Task Success Rate (TSR)STAR
8.1427.44546.7566.055May 27, 2026

Evaluation results

MethodLinks
STARBackbone=Qwen-Plus, Attack Setting=Cooperative2026.0582.56
STARBackbone=Qwen-Plus, Attack Setting=Independent2026.05818.5
ARGUSBackbone=Qwen-Plus, Attack Setting=Cooperative2026.0574.7517.5
STARBackbone=GPT-3.5-Turbo, Attack Setting=Cooperative2026.0574.2512.25
STARBackbone=GPT-3.5-Turbo, Attack Setting=Independent2026.057313.25
ARGUSBackbone=Qwen-Plus, Attack Setting=Independent2026.0571.2521.5
Blind-GuardBackbone=Qwen-Plus, Attack Setting=Independent2026.056924.75
G-SafeguardBackbone=Qwen-Plus, Attack Setting=Cooperative2026.0568.523.5
G-SafeguardBackbone=Qwen-Plus, Attack Setting=Independent2026.0567.7525.25
Blind-GuardBackbone=Qwen-Plus, Attack Setting=Cooperative2026.0567.7525.5
C&IBackbone=Qwen-Plus, Attack Setting=Independent2026.0564.7528.25
STARBackbone=GPT-5-Nano, Attack Setting=Independent2026.056423.5
Blind-GuardBackbone=GPT-3.5-Turbo, Attack Setting=Cooperative2026.0562.7531.25
C&IBackbone=Qwen-Plus, Attack Setting=Cooperative2026.0562.531.25
Blind-GuardBackbone=GPT-3.5-Turbo, Attack Setting=Independent2026.056134
STARBackbone=GPT-5-Nano, Attack Setting=Cooperative2026.0559.528
C&IBackbone=GPT-3.5-Turbo, Attack Setting=Cooperative2026.0558.2534.5
C&IBackbone=GPT-3.5-Turbo, Attack Setting=Independent2026.055440.25
Blind-GuardBackbone=GPT-5-Nano, Attack Setting=Independent2026.0546.7546.25
G-SafeguardBackbone=GPT-5-Nano, Attack Setting=Cooperative2026.0545.2548
ARGUSBackbone=GPT-3.5-Turbo, Attack Setting=Cooperative2026.054252.25
ARGUSBackbone=GPT-3.5-Turbo, Attack Setting=Independent2026.0540.7554.25
G-SafeguardBackbone=GPT-5-Nano, Attack Setting=Independent2026.0534.2560
G-SafeguardBackbone=GPT-3.5-Turbo, Attack Setting=Cooperative2026.0533.7562
ARGUSBackbone=GPT-5-Nano, Attack Setting=Independent2026.0529.7566.5
Blind-GuardBackbone=GPT-5-Nano, Attack Setting=Cooperative2026.0528.7567
G-SafeguardBackbone=GPT-3.5-Turbo, Attack Setting=Independent2026.052669.5
ARGUSBackbone=GPT-5-Nano, Attack Setting=Cooperative2026.0524.571.5
C&IBackbone=GPT-5-Nano, Attack Setting=Independent2026.0523.2574
C&IBackbone=GPT-5-Nano, Attack Setting=Cooperative2026.051188.5