INDEX

Explanations

names and words from Turkey

The neuron strongly activates on fragments of capitalized words—especially the leading segments of proper nouns.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 informa

-0.87

 lavabo

-0.83

 tene

-0.82

ัก

-0.82

 bomba

-0.79

は

-0.79

ranked

-0.78

 ?>">

-0.78

 poros

-0.77

 tali

-0.77

POSITIVE LOGITS

 Arzt

0.86

 Před

0.86

 pantallas

0.84

方がいい

0.83

 spodnie

0.82

 afhankelijk

0.82

 impresiones

0.81

بيض

0.81

 enfans

0.80

 EventArgs

0.79

Activations Density 0.001%