Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Probing on 113 binary tasks (val)
Loading...
94
Probe AUC
Raw Layer Output
84.64
87.07
89.5
91.93
Feb 6, 2026
Probe AUC
Probe AUC 95% CI
Updated 4d ago
Evaluation Results
Method
Method
Links
Probe AUC
Probe AUC 95% CI
Raw Layer Output
Base Model=Llama8B, Pr...
2026.02
94
-
Raw MLP Neuron
Base Model=Llama8B, Pr...
2026.02
94
-
GLP
Base Model=Llama8B, Pr...
2026.02
94
-
Raw MLP Neuron
Base Model=Llama1B, Pr...
2026.02
93
-
Raw Layer Output
Base Model=Llama1B, Pr...
2026.02
92
-
GLP
Base Model=Llama1B, Pr...
2026.02
92
-
SAE
Base Model=Llama8B, Pr...
2026.02
90
-
SAE
Base Model=Llama1B, Pr...
2026.02
85
-
Feedback
Search any
task
Search any
task