INDEX
Negative Logits
in
-1.84
at
-1.63
so
-1.63
all
-1.62
not
-1.54
Although
-1.53
on
-1.53
two
-1.50
their
-1.48
just
-1.47
POSITIVE LOGITS
différentes
1.65
verschillende
1.65
nieuwe
1.48
refroid
1.42
différents
1.42
adatt
1.39
новую
1.37
dépens
1.37
いろんな
1.37
новых
1.37
Activations Density 0.020%