Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Privacy Expectation Estimation on CONFAIDE Tier 2 (InfoFlow-Expectation)
Loading...
-83.3
Sensitivity
GPT-4 (Drunk language inducement)
-90.224
-43.487
3.25
49.987
Jan 19, 2026
Sensitivity
Updated 3mo ago
Evaluation Results
Method
Method
Links
Sensitivity
GPT-4 (Drunk language inducement)
Variant=Pr.
2026.01
-83.3
GPT-4
Variant=Base
2026.01
-74.9
GPT-4 (Drunk language inducement)
Variant=FT
2026.01
-61.9
GPT-3.5
Variant=Base
2026.01
-56.1
LLaMA3-8B (Drunk language inducement)
Variant=FT
2026.01
-55.5
LLaMA3-8B
Variant=Base
2026.01
-51.8
GPT-3.5 (Drunk language inducement)
Variant=Pr.
2026.01
-50.4
LLaMA3-8B (Drunk language inducement)
Variant=RL
2026.01
-37.3
LLaMA3-8B (Drunk language inducement)
Variant=Pr.
2026.01
-35.6
LLaMA2-7B (Drunk language inducement)
Variant=RL
2026.01
-3.6
LLaMA2-7B
Variant=Base
2026.01
-2.6
Mistral-7B (Drunk language inducement)
Variant=Pr.
2026.01
-1
LLaMA2-7B (Drunk language inducement)
Variant=Pr.
2026.01
-0.4
LLaMA2-7B (Drunk language inducement)
Variant=FT
2026.01
6.8
Mistral-7B
Variant=Base
2026.01
26
Mistral-7B (Drunk language inducement)
Variant=FT
2026.01
66.3
Mistral-7B (Drunk language inducement)
Variant=RL
2026.01
89.8
Feedback
Search any
task
Search any
task