INDEX
Explanations
commit suicide
It detects prominent named entities and salient topic tokens (titles, product names, and other key content words).
New Auto-Interp
Negative Logits
ம்
1.11
на
1.05
ک
1.04
ة
1.00
م
0.99
ა
0.98
ین
0.97
ী
0.96
м
0.95
ہ
0.93
POSITIVE LOGITS
of
1.09
to
1.04
。
1.04
it
1.03
0.89
(
0.81
a
0.80
you
0.78
we
0.76
y
0.75
Activations Density 0.003%