INDEX
Explanations
This neuron detects named entities—tokens that are capitalized proper nouns such as chemical names, brand names, locations, or other specialized terms.
New Auto-Interp
Negative Logits
셜
-0.07
Alban
-0.06
PY
-0.06
примен
-0.06
Dimensions
-0.06
будинку
-0.06
ап
-0.06
********************************************************************************
-0.06
Phones
-0.06
involved
-0.06
POSITIVE LOGITS
object
0.06
История
0.06
'}↵↵
0.06
جستارهای
0.06
lust
0.06
наук
0.06
สมเด
0.06
/article
0.05
cairo
0.05
}`}
0.05
Activations Density 0.167%