INDEX
Negative Logits
are
1.42
are
1.36
ها
1.26
ется
1.11
$,
1.10
;
1.10
ן
0.97
지
0.96
+](=
0.96
هاي
0.95
POSITIVE LOGITS
'
1.41
for
1.35
Clock
1.25
S
1.22
B
1.20
clock
1.18
り
1.16
E
1.13
U
1.10
L
1.09
Activations Density 0.005%
are
are
ها
ется
$,
;
ן
지
+](=
هاي
'
for
Clock
S
B
clock
り
E
U
L