INDEX
Explanations
The neuron responds to structural formatting tokens—especially the header‐ID markers and the period after numbered list items.
New Auto-Interp
Negative Logits
sembling
-0.07
Mil
-0.06
ais
-0.06
उल
-0.06
Walker
-0.06
ически
-0.06
optic
-0.06
มต
-0.06
Sed
-0.06
remarks
-0.06
POSITIVE LOGITS
.XML
0.07
Özel
0.07
sell
0.07
'.
0.06
sponsors
0.06
(ListNode
0.06
hypocrisy
0.06
уванні
0.06
قي
0.06
مصر
0.06
Activations Density 0.008%