INDEX
Explanations
proper nouns related to various individuals
the occurrences of the name "Mark."
New Auto-Interp
Negative Logits
curfew
-0.74
girls
-0.67
crumble
-0.65
sisters
-0.64
clan
-0.62
comedy
-0.61
awakening
-0.61
backbone
-0.61
defer
-0.61
hower
-0.60
POSITIVE LOGITS
Mark
3.84
mark
2.17
Mark
2.17
MARK
1.95
marks
1.80
mark
1.49
Matthew
1.40
Marc
1.31
emark
1.28
marked
1.25
Activations Density 0.009%