INDEX
Explanations
references to narrative structure and storytelling
New Auto-Interp
Negative Logits
ery
-0.18
oo
-0.16
----</
-0.15
onet
-0.15
TION
-0.15
/bus
-0.15
ick
-0.15
site
-0.14
baar
-0.14
ser
-0.14
POSITIVE LOGITS
arc
0.25
arcs
0.21
_voice
0.18
structure
0.18
device
0.17
thread
0.17
voice
0.17
-driven
0.17
arc
0.17
ative
0.16
Activations Density 0.011%