INDEX
Explanations
texts related to crime detection, surveillance, and privacy issues
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
2034
+0.22
0.7%
1535
+0.15
0.5%
752
+0.14
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
382
+0.22
0.11
310
+0.15
0.07
610
+0.14
0.08
Negative Logits
sappi
-1.18
uncin
-1.09
jetta
-1.08
viciss
-1.07
husqvarna
-1.05
lamborghini
-1.04
gsx
-0.99
logitech
-0.99
camry
-0.98
isuzu
-0.98
POSITIVE LOGITS
Now
0.72
Even
0.65
})();
0.64
Now
0.63
So
0.62
And
0.60
Hence
0.58
This
0.58
new
0.58
My
0.58
Activations Density 0.551%