Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Prefilling Attack

Benchmarks

Task NameDataset NameSOTA ResultTrend
Jailbreak AttackPrefilling Attack 40 tokens
ASR (%)0
45
Jailbreak AttackPrefilling Attack 20 tokens
ASR0.3
45
Jailbreak AttackPrefilling Attack 10 tokens
ASR70.91
45
Showing 3 of 3 rows