List your company

Unsafe Instruction Mitigation on Libero Harm (test)

7.8ASRSAFE-Dict
4.72425.48746.2567.013Feb 2, 2026

Evaluation results

MethodLinks
SAFE-DictSetting=SAFE-Dict2026.027.8
Prompt-based SafetySetting=Prompt-based Safety2026.0241.2
Default (no defense)Setting=no defense2026.0284.7