INDEX
Explanations
This neuron detects mentions of family lineage, specifically phrases stating someone is the son or daughter of a named parent.
New Auto-Interp
Negative Logits
Frog
-0.07
,、
-0.07
xn
-0.07
bunlar
-0.06
withStyles
-0.06
album
-0.06
Laurie
-0.06
자동
-0.06
tudo
-0.06
geliyor
-0.06
POSITIVE LOGITS
อเร
0.08
coupon
0.07
三
0.06
酒
0.06
managers
0.06
_pixels
0.06
_kernel
0.06
ij
0.06
89
0.06
oppress
0.06
Activations Density 0.005%