Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Harmfulness Evaluation on I-Malicious
Loading...
0
Harmful Rate
NLCf/800 step
-1.36
7.82
17
26.18
May 30, 2026
Harmful Rate
Harmfulness Score
Updated 1d ago
Evaluation Results
Method
Method
Links
Harmful Rate
Harmfulness Score
NLCf/800 step
Steps=800, Loss=NLCf
2026.05
0
0.2
EMD/800 step
Steps=800, Loss=EMD
2026.05
0
0.16
Mistral MoE-XL
Architecture=MoE-XL
2026.05
6
0.005
Qwen-3B MoE-XL
Architecture=MoE-XL
2026.05
13
0.01
Mistral
2026.05
27
3
Qwen-3B
2026.05
34
0.03
Feedback
Search any
task
Search any
task