Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Safety on AIR-Bench
Loading...
0.66
Average Score
REINFORCE++ (Ours)
0.2336
0.3443
0.455
0.5657
Dec 1, 2025
Average Score
Updated 4d ago
Evaluation Results
Method
Method
Links
Average Score
REINFORCE++ (Ours)
Base Model=DeepSeek-R1...
2025.12
0.66
DeepSeek-R1-Distill-Qwen-7B + SFT (STAR-1)
Base Model=DeepSeek-R1...
2025.12
0.59
REINFORCE++ (Ours)
Base Model=Qwen3-8B, T...
2025.12
0.58
Qwen3-8B + CPO
Base Model=Qwen3-8B, T...
2025.12
0.55
Qwen3-8B + SFT (STAR-1)
Base Model=Qwen3-8B, T...
2025.12
0.51
Qwen3-8B + SFT (R2D-R1)
Base Model=Qwen3-8B, T...
2025.12
0.43
DeepSeek-R1-Distill-Qwen-7B + SFT (R2D-R1)
Base Model=DeepSeek-R1...
2025.12
0.41
DeepSeek-R1-Distill-Qwen-7B + CPO
Base Model=DeepSeek-R1...
2025.12
0.41
Qwen3-8B (thinking)
Base Model=Qwen3-8B, T...
2025.12
0.4
Qwen3-8B + SFT (SafeChain)
Base Model=Qwen3-8B, T...
2025.12
0.29
DeepSeek-R1-Distill-Qwen-7B
Base Model=DeepSeek-R1...
2025.12
0.26
DeepSeek-R1-Distill-Qwen-7B + SFT (SafeChain)
Base Model=DeepSeek-R1...
2025.12
0.25
Feedback
Search any
task
Search any
task