INDEX
Explanations
references to citrus fruits and their attributes
New Auto-Interp
Negative Logits
lessly
-0.18
PerPixel
-0.16
TEL
-0.16
avanaugh
-0.15
chest
-0.15
sy
-0.15
line
-0.15
ester
-0.14
ään
-0.14
su
-0.14
POSITIVE LOGITS
-yellow
0.19
-headed
0.17
(es
0.15
ockey
0.15
zaz
0.15
eyed
0.14
apolis
0.14
cas
0.14
vais
0.14
orton
0.14
Activations Density 0.078%