INDEX
Negative Logits
on
1.38
and
1.30
на
1.16
as
1.13
ü
1.13
are
1.09
about
1.01
for
0.98
т
0.98
but
0.95
POSITIVE LOGITS
V
1.10
N
1.00
OF
0.97
of
0.96
i
0.96
l
0.95
P
0.93
iyev
0.92
EN
0.92
ac
0.91
Activations Density 0.004%
on
and
на
as
ü
are
about
for
т
but
V
N
OF
of
i
l
P
iyev
EN
ac