Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Concept Recovery on REFUSAL
Loading...
0.5669
Mean MCC
Linear Probe
0.408508
0.449629
0.49075
0.531871
Feb 14, 2025
Mean MCC
Updated 1mo ago
Evaluation Results
Method
Method
Links
Mean MCC
Linear Probe
Evaluation Protocol=Re...
2025.02
0.5669
SSAE
Evaluation Protocol=Re...
2025.02
0.4781
ReLU-SAE
Evaluation Protocol=Re...
2025.02
0.4554
JumpReLU SAE
Evaluation Protocol=Re...
2025.02
0.4381
TopK-SAE
Evaluation Protocol=Re...
2025.02
0.4276
PythiaSAE
Evaluation Protocol=Re...
2025.02
0.4146
Feedback
Search any
task
Search any
task