INDEX
Explanations
mentions of a particular country along with associated locations
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
411
+0.16
0.6%
849
+0.15
0.5%
874
+0.15
0.5%
Correlated Neurons
Index
P. Corr.
Cos Sim.
849
+0.16
0.03
1516
+0.15
0.02
492
+0.15
0.02
Negative Logits
conlle
-0.50
Minang
-0.44
tagHelperRunner
-0.44
huahua
-0.44
<bos>
-0.44
semblait
-0.44
<?
-0.42
strictEqual
-0.41
estuvieron
-0.41
Descubre
-0.41
POSITIVE LOGITS
Yemen
1.26
Yemen
1.19
emeni
0.91
Aden
0.64
Hieronymus
0.60
Darío
0.58
Mlle
0.57
himo
0.57
Whence
0.56
kasama
0.56
Activations Density 0.062%