INDEX
Explanations
Numbers in technical contexts
This neuron fires on numeral tokens (i.e. individual numbers or numeric sequences).
New Auto-Interp
Negative Logits
цький
-0.07
yle
-0.06
Над
-0.06
Index
-0.06
�
-0.06
NL
-0.06
χει
-0.06
bast
-0.06
ाट
-0.06
customer
-0.06
POSITIVE LOGITS
Ein
0.07
торгов
0.06
._
0.06
[q
0.06
Puerto
0.06
.colors
0.06
}}">↵
0.06
eruption
0.06
زيز
0.06
0.06
Activations Density 0.073%