INDEX
Explanations
instances of proper names, particularly Asian names
Chinese surnames
New Auto-Interp
Negative Logits
<eos>
-0.48
7
-0.45
8
-0.44
5
-0.42
3
-0.40
VE
-0.40
4
-0.39
RE
-0.39
-
-0.38
/
-0.37
POSITIVE LOGITS
ainfi
1.00
Houſe
0.98
Zhu
0.98
Zhang
0.97
Verſ
0.95
Jiang
0.94
Zhu
0.93
Monfieur
0.93
Zhao
0.93
increí
0.91
Activations Density 0.064%