INDEX
Explanations
political/religious groups/divisions
The neuron activates on words referring to religious or political “sects” or factions.
New Auto-Interp
Negative Logits
Mathematical
-0.06
Bar
-0.06
<Document
-0.06
homicides
-0.06
urd
-0.06
YouTube
-0.06
overlays
-0.06
homicide
-0.06
poměrně
-0.06
ặng
-0.06
POSITIVE LOGITS
faction
0.08
Faction
0.08
factions
0.07
جذ
0.07
机构
0.07
CACHE
0.07
мн
0.06
openly
0.06
ยว
0.06
considered
0.06
Activations Density 0.004%