INDEX
Explanations
short entity descriptions in news articles
New Auto-Interp
Neuron Alignment
Index
Value
% of L₁
453
+0.20
0.6%
1150
+0.15
0.5%
845
+0.12
0.4%
Correlated Neurons
Index
P. Corr.
Cos Sim.
453
+0.20
0.05
16
+0.15
0.04
1150
+0.12
0.01
Negative Logits
milano
-1.29
fta
-1.24
doman
-1.23
napoli
-1.20
wien
-1.20
marseille
-1.19
ftu
-1.18
matel
-1.17
bordeaux
-1.17
frankfurt
-1.17
POSITIVE LOGITS
StatefulWidget
0.55
Viitteet
0.55
bosis
0.54
estekak
0.53
StatelessWidget
0.53
above
0.53
Einzelnachweise
0.52
دریافتشده
0.52
aforementioned
0.51
Palmar
0.50
Activations Density 0.198%