INDEX
Explanations
This neuron detects instances of the C++ scope resolution operator (::) in qualified names.
New Auto-Interp
Negative Logits
rical
-0.07
_SOCKET
-0.06
_seek
-0.06
keyboards
-0.06
sl
-0.06
Species
-0.06
_da
-0.06
LOAT
-0.06
dere
-0.06
orting
-0.06
POSITIVE LOGITS
suggesting
0.06
confidence
0.06
并不
0.06
既
0.06
(elem
0.06
(!_
0.06
:)
0.06
reconstructed
0.06
().__
0.06
recibir
0.06
Activations Density 0.004%