Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Truthfulness on TruthfulQA (Utility)
Loading...
51.6
Utility
AdvGRPO (ST)
47.96
48.905
49.85
50.795
Jun 8, 2026
Utility
Updated 1mo ago
Evaluation Results
Method
Method
Links
Utility
AdvGRPO (ST)
Attacker=Qwen2.5-7B
2026.06
51.6
Base
2026.06
51.5
AdvGRPO (ST)
Attacker=Qwen2.5-14B
2026.06
51.2
AdvGame
Attacker=Qwen2.5-7B
2026.06
48.7
Self-RedTeam
Attacker=Qwen2.5-7B
2026.06
48.1
Feedback
Search any
task
Search any
task