INDEX
Explanations
punctuation and "and"
This neuron fires on short function words and auxiliaries (like “if,” “make,” “It,” “doesn’t,” “have,” “This”) used to introduce, qualify, or structure instructions and explanatory statements.
New Auto-Interp
Negative Logits
.outputs
-0.07
Голов
-0.07
propia
-0.07
Border
-0.06
системы
-0.06
mamız
-0.06
@Inject
-0.06
gz
-0.06
Theater
-0.06
498
-0.06
POSITIVE LOGITS
oft
0.07
Quaternion
0.06
?"↵↵
0.06
_docs
0.06
jackets
0.06
_san
0.06
ủ
0.06
bist
0.06
0.06
этот
0.06
Activations Density 0.010%