INDEX
Explanations
references to the name "Hijab"
the occurrences of the letters "ij" in context
New Auto-Interp
Negative Logits
projected
-0.72
Aliens
-0.65
flush
-0.65
tears
-0.64
Ducks
-0.64
ponies
-0.62
butterflies
-0.62
Metall
-0.62
chests
-0.62
steal
-0.61
POSITIVE LOGITS
ij
4.29
ijn
2.02
IJ
1.90
ijk
1.79
ija
1.71
iji
1.64
ih
1.33
ieu
1.33
iy
1.32
aj
1.17
Activations Density 0.018%