INDEX
Explanations
occurrences of the prefix "re-" indicating repetition or restoration
New Auto-Interp
Negative Logits
z
-0.23
b
-0.21
p
-0.20
v
-0.19
j
-0.19
lek
-0.18
un
-0.18
ak
-0.18
f
-0.17
w
-0.16
POSITIVE LOGITS
semb
0.20
ductive
0.18
straints
0.17
ationship
0.17
re
0.17
ection
0.16
duct
0.16
inton
0.16
erald
0.16
pute
0.15
Activations Density 0.022%