INDEX
Explanations
terms related to legal or formal conduct and procedures
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
204
+0.16
0.7%
528
+0.13
0.5%
169
+0.12
0.5%
Correlated Neurons
Index
P. Corr.
Cos Sim.
204
+0.16
0.02
169
+0.13
0.02
976
+0.12
0.02
Negative Logits
Wat
-0.46
eyes
-0.44
Wat
-0.43
paz
-0.43
Più
-0.43
Spring
-0.43
Jake
-0.42
Buck
-0.42
Jake
-0.42
page
-0.41
POSITIVE LOGITS
conducted
1.18
conduct
1.16
conduct
1.10
Conduct
1.10
CONDUCT
1.09
conducts
1.08
conducted
1.06
Conducted
1.05
conducting
1.02
Conducting
0.99
Activations Density 0.103%