INDEX
Explanations
references to different versions of a product or concept
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
1044
+0.14
0.5%
1059
+0.14
0.5%
2011
+0.12
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
1044
+0.14
0.04
1059
+0.14
0.04
2011
+0.12
0.03
Negative Logits
cciale
-0.72
esternos
-0.63
autunno
-0.62
pendente
-0.57
zove
-0.57
reputa
-0.56
Aprile
-0.56
Referencoj
-0.56
fuo
-0.55
anskje
-0.55
POSITIVE LOGITS
version
1.28
version
1.16
versions
1.14
Version
1.12
Version
1.10
VERSION
1.08
Versions
0.96
VERSION
0.96
versions
0.91
版本
0.88
Activations Density 0.091%