INDEX
Explanations
The neuron is triggered by occurrences of “.com” in domain names (i.e. web addresses).
New Auto-Interp
Negative Logits
Macro
-0.06
885
-0.06
inski
-0.06
afirm
-0.06
weighting
-0.06
getCode
-0.06
lateinit
-0.06
308
-0.06
attitudes
-0.06
šet
-0.06
POSITIVE LOGITS
,)↵
0.06
"), ↵
0.06
Beam
0.06
_into
0.06
ece
0.06
Faith
0.06
обл
0.06
信用
0.06
(fin
0.06
raith
0.06
Activations Density 0.005%