List your company

Alignment Task Evaluation on PKU-SafeRLHF w/o trigger

70.09RM Safety AccRankPoison
69.705269.805169.90570.0049Nov 16, 2023

Evaluation results

MethodLinks
RankPoisonSetting=w/o trigger2023.1170.092.0214.26
Random FlipSetting=w/o trigger2023.1170.051.5819.6
BaselineSetting=w/o trigger2023.1169.722.328.49