INDEX
Explanations
emotional responses and interactions in narratives
New Auto-Interp
Negative Logits
asis
-0.16
rael
-0.15
ør
-0.15
antz
-0.15
vale
-0.15
quam
-0.15
infeld
-0.15
entai
-0.15
.maven
-0.14
conde
-0.14
POSITIVE LOGITS
Thing
0.15
.strict
0.14
edImage
0.14
=č↵
0.14
æ
0.14
thing
0.14
Thing
0.13
thing
0.13
è½
0.13
richt
0.13
Activations Density 0.006%