INDEX
Explanations
statements describing halves or parts of something
elements of contrast or duality in descriptions
New Auto-Interp
Negative Logits
iev
-0.79
airo
-0.76
ilus
-0.76
802
-0.68
2500
-0.68
scrib
-0.66
ahu
-0.65
iam
-0.65
eve
-0.65
arcity
-0.65
POSITIVE LOGITS
part
1.68
partly
1.50
Part
1.45
Part
1.44
halves
1.42
part
1.37
partially
1.36
half
1.36
half
1.35
PART
1.31
Activations Density 0.155%