List your company

LLM Red-teaming on S-GFN-defended Target Model

7.33Unsuccessful Attack Rate (UA)S-GFN
-0.29321.68593.6655.6441May 1, 2026

Evaluation results

MethodLinks
S-GFN2026.057.3375
Jailbreak R1CoT=enabled2026.055.6755
Rainbow Teaming2026.052.3323
ICL2026.05110
GFN2026.050.333
SFT2026.0500
DPO2026.0500
PPO2026.0500
PPO + Curiosity2026.0500