INDEX
Explanations
information related to statistical data and analysis
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1741
+0.22
0.7%
1842
+0.13
0.4%
1042
+0.12
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
382
+0.22
0.10
1265
+0.13
0.07
736
+0.12
0.09
Negative Logits
umo
-1.55
jaya
-1.49
bandung
-1.48
parati
-1.44
lele
-1.41
kasa
-1.40
istan
-1.40
maroc
-1.38
ordina
-1.38
ristor
-1.36
POSITIVE LOGITS
but
0.76
which
0.73
or
0.71
whose
0.69
although
0.68
though
0.63
however
0.62
who
0.62
etc
0.61
and
0.61
Activations Density 0.679%