INDEX
Explanations
numbers, years, and URLs
This neuron detects tokens that are floating-point numeric strings (numbers with a decimal point).
New Auto-Interp
Negative Logits
것이
0.49
시간을
0.49
ри
0.49
ρν
0.49
人材
0.49
ಾರ
0.48
이다
0.47
τού
0.47
ਣ
0.46
ações
0.46
POSITIVE LOGITS
s
0.94
(
0.72
n
0.70
z
0.67
}
0.65
y
0.64
n
0.60
ের
0.60
for
0.59
al
0.59
Activations Density 0.079%