INDEX
Explanations
act like, quadruped, first aid
The neuron strongly responds to tokens containing the substring "act" (e.g., act, actuated, actuation, etc.).
New Auto-Interp
Negative Logits
opee
0.47
ارق
0.46
హిట్
0.44
подробно
0.43
ạ
0.42
ார்கள்
0.42
ナソニック
0.42
consequatur
0.42
้ง
0.41
ड़
0.41
POSITIVE LOGITS
l
0.53
wear
0.52
together
0.51
it
0.50
il
0.49
where
0.48
ഒന്ന
0.47
ranging
0.46
north
0.46
institution
0.46
Activations Density 0.000%