Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Evaluation on GCG
Loading...
95.96
Safety Score
REFLECTOR (+GDPO)
38.5936
53.4868
68.38
83.2732
May 20, 2026
Safety Score
Updated 13d ago
Evaluation Results
Method
Method
Links
Safety Score
REFLECTOR (+GDPO)
Backbone=Qwen-2.5-7B-I...
2026.05
95.96
REFLECTOR (+GDPO)
Backbone=Llama-3.1-8B-...
2026.05
94.23
REFLECTOR (+SFT)
Backbone=Llama-3.1-8B-...
2026.05
90.96
REFLECTOR (SFT)
Backbone=Qwen-2.5-7B-I...
2026.05
88.69
STAIR
Backbone=Qwen-2.5-7B-I...
2026.05
87.69
STAIR
Backbone=Llama-3.1-8B-...
2026.05
86.15
Shallow-Align
Backbone=Qwen-2.5-7B-I...
2026.05
83.5
Shallow-Align
Backbone=Llama-3.1-8B-...
2026.05
83.4
DPO
Backbone=Llama-3.1-8B-...
2026.05
82.31
Self-Critique
Backbone=Llama-3.1-8B-...
2026.05
81.15
SFT
Backbone=Llama-3.1-8B-...
2026.05
79.8
Original
Backbone=Llama-3.1-8B-...
2026.05
78.4
DPO
Backbone=Qwen-2.5-7B-I...
2026.05
51.92
Self-Critique
Backbone=Qwen-2.5-7B-I...
2026.05
46.8
SFT
Backbone=Qwen-2.5-7B-I...
2026.05
42.31
Original
Backbone=Qwen-2.5-7B-I...
2026.05
40.8
Feedback
Search any
task
Search any
task