INDEX
Explanations
references to various types of meat dishes and their qualities
New Auto-Interp
Negative Logits
pleted
-0.17
ez
-0.16
ssel
-0.16
esktop
-0.15
udad
-0.15
erator
-0.15
ingo
-0.15
cker
-0.15
td
-0.14
gist
-0.14
POSITIVE LOGITS
balls
0.39
ball
0.31
packing
0.28
y
0.27
lo
0.26
ier
0.26
BALL
0.25
less
0.22
IER
0.21
locker
0.21
Activations Density 0.011%