INDEX
Explanations
the frequency of the word "junk."
New Auto-Interp
Negative Logits
çĽĹ
-0.16
zer
-0.16
̧
-0.15
DonaldTrump
-0.15
تÙĩ
-0.14
lich
-0.14
unbind
-0.14
ence
-0.14
:animated
-0.14
Bender
-0.14
POSITIVE LOGITS
uj
0.16
arn
0.15
oodle
0.15
aji
0.15
Cable
0.14
ette
0.14
pies
0.14
ाब
0.13
Converted
0.13
ensi
0.13
Activations Density 0.003%