INDEX
Explanations
connections between numerical data and related contextual phrases
New Auto-Interp
Negative Logits
assi
-0.18
asil
-0.16
azzi
-0.15
assy
-0.15
ddit
-0.14
ruž
-0.14
ianne
-0.14
ude
-0.14
رب
-0.14
jac
-0.14
POSITIVE LOGITS
respectively
0.19
ousse
0.16
ouz
0.15
0.15
respective
0.15
arella
0.15
ween
0.15
hof
0.14
Mund
0.14
Dream
0.14
Activations Density 0.043%