INDEX
Explanations
terms related to families being separated and longing to reunite
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1013
+0.12
0.3%
509
+0.11
0.3%
961
+0.10
0.3%
Correlated Neurons
Index
P. Corr.
Cos Sim.
509
+0.12
0.05
961
+0.11
0.03
1081
+0.10
0.04
Negative Logits
increa
-1.82
disagre
-1.82
encomp
-1.80
suscep
-1.80
unden
-1.78
affor
-1.78
secon
-1.77
depic
-1.77
guarante
-1.77
volunte
-1.75
POSITIVE LOGITS
<bos>
0.95
ProtoMessage
0.79
family
0.72
GraphicsUnit
0.72
home
0.71
ISupport
0.69
oneofs
0.69
family
0.68
prnewswire
0.67
relatives
0.66
Activations Density 0.313%