INDEX
Explanations
names followed by descriptors
The neuron strongly detects named entities—proper nouns like people, organizations, places, and other capitalized names.
New Auto-Interp
Negative Logits
ប់
0.36
rowave
0.36
Tb
0.35
rol
0.34
ตัวเอง
0.33
<unused13>
0.33
rowned
0.33
rying
0.32
нев
0.32
くだ
0.32
POSITIVE LOGITS
selaku
0.93
،
0.91
，
0.75
،
0.71
,
0.64
ซึ่ง
0.64
၊
0.61
iaitu
0.61
नामक
0.59
®,
0.59
Activations Density 0.085%