Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Guardrail on Prompt-side and Response-side benchmarks Aggregate
Loading...
87.8
F1 Score
Stage3-DPO
82.6
83.95
85.3
86.65
Jul 7, 2026
F1 Score
Updated 17d ago
Evaluation Results
Method
Method
Links
F1 Score
Stage3-DPO
Params=4B
2026.07
87.8
Stage2-SFT
Params=4B
2026.07
86.4
Stage1-SFT-v4
Params=4B
2026.07
86
Qwen3Guard-8B-Gen
Params=8B
2026.07
85.8
Stage1-SFT-v1
Params=4B
2026.07
85.6
Stage1-SFT-v0
Params=4B
2026.07
85.1
Qwen3Guard-4B-Gen
Params=4B
2026.07
84.8
YuFeng-XGuard-Reason-8B
Params=8B
2026.07
84.5
Qwen3Guard-0.8B-Gen
Params=0.8B
2026.07
84.2
Stage1-SFT-v3
Params=4B
2026.07
84
YuFeng-XGuard-Reason-0.6B
Params=0.6B
2026.07
83.4
Stage1-SFT-v2
Params=4B
2026.07
82.8
Feedback
Search any
task
Search any
task