INDEX
Explanations
locations and names of people who have passed away in obituary-type texts
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1520
+0.11
0.5%
1140
+0.11
0.4%
506
+0.10
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
648
+0.11
0.07
609
+0.11
0.06
227
+0.10
0.07
Negative Logits
<bos>
-1.59
ⓧ
-0.62
//});
-0.55
<?
-0.55
///**
-0.54
ORGANISM
-0.54
ernoo
-0.54
WireFormatLite
-0.53
EndProject
-0.53
IsMutable
-0.53
POSITIVE LOGITS
maksi
1.19
kasa
1.18
saar
1.12
véhic
1.11
panik
1.09
Minang
1.06
stockholm
1.05
étrang
1.04
Dica
1.01
confé
1.01
Activations Density 0.537%