INDEX
Explanations
The neuron activates on subword tokens making up the artist’s name “Kanye West” (including his short form “Ye”).
New Auto-Interp
Negative Logits
ๆ
-0.06
ılıp
-0.06
madan
-0.06
_ter
-0.06
chút
-0.06
Mayıs
-0.06
Uint
-0.06
forgettable
-0.06
_keyword
-0.06
ibox
-0.06
POSITIVE LOGITS
Kanye
0.13
Ye
0.08
OBJ
0.07
anye
0.07
GTA
0.07
.getNode
0.06
>");↵
0.06
Kob
0.06
EMS
0.06
(sh
0.06
Activations Density 0.001%