INDEX
Negative Logits
ك
0.99
وا
0.74
व
0.72
ные
0.68
două
0.64
the
0.61
雎
0.61
는
0.61
am
0.61
ছিল
0.59
POSITIVE LOGITS
’
0.95
-
0.92
5
0.73
’-
0.66
3
0.66
*
0.65
adata
0.62
errors
0.61
4
0.61
}
0.61
Activations Density 0.002%
ك
وا
व
ные
două
the
雎
는
am
ছিল
’
-
5
’-
3
*
adata
errors
4
}