INDEX
Explanations
date and numbers
The neuron detects numeric timestamp‑style tokens (numbers with decimal/float-like values often appearing after dates or time markers).
New Auto-Interp
Negative Logits
was
0.48
失去了
0.48
didn
0.45
year
0.43
appeared
0.42
came
0.42
めの
0.42
diseased
0.42
year
0.41
went
0.41
POSITIVE LOGITS
Ngày
0.52
वाटर
0.51
Usuarios
0.50
Тен
0.50
POSIT
0.49
社会
0.49
Bugünkü
0.49
ํา
0.49
Lotus
0.48
RestorePolicy
0.48
Activations Density 0.002%