Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Neuron Labeling on SAE-TopK Evaluated Neurons

0.98AUC

CSP

0.90720.92610.9450.9639Apr 24, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
0.98
2026.04
0.98
2026.04
0.98
2026.04
0.98
2026.04
0.98
2026.04
0.97
2026.04
0.97
2026.04
0.97
2026.04
0.96
2026.04
0.96
2026.04
0.96
2026.04
0.95
2026.04
0.94
2026.04
0.93
2026.04
0.91