INDEX

Explanations

postal codes and city names

The neuron is strongly activated by tokens representing numbers or numeric sequences (digits).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

Vivo

0.72

 Основ

0.71

Созда

0.67

 능력

0.67

 навча

0.66

﹀

0.66

MHz

0.65

xyplot

0.65

 ഗവേഷ

0.65

 профе

0.65

POSITIVE LOGITS

 ciudad

0.62

ud

0.59

 inconven

0.57

 doubly

0.56

 laterally

0.55

 haul

0.54

 unincorporated

0.54

Ud

0.53

 commencer

0.53

 hauled

0.53

Activations Density 0.006%