INDEX
Explanations
AI assistant safety guidelines
This neuron detects prominent headings, section titles, or otherwise emphasized/topic-signaling words in the text.
key topic nouns in instructional or explanatory content.
New Auto-Interp
Negative Logits
symbolize
0.31
tiež
0.31
सुरुवात
0.29
toning
0.29
ल्प
0.29
centos
0.29
最初に
0.29
will
0.28
dct
0.28
ക്കാണ്
0.27
POSITIVE LOGITS
grabación
0.34
dukkham
0.32
vítimas
0.32
violencia
0.31
violência
0.31
พระเจ้า
0.30
violently
0.30
victimes
0.30
mish
0.29
comercio
0.29
Activations Density 0.081%