INDEX

Explanations

likely foreign words

This neuron detects interface action labels—e.g. buttons or links for sharing, navigation, or social-media actions.

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

if

-1.80

 丁目

-1.49

Eins

-1.34

 merke

-1.33

geh

-1.32

тиви

-1.32

wendig

-1.30

 kursi

-1.29

 because

-1.28

If

-1.28

POSITIVE LOGITS

认同

1.67

 probablemente

1.53

 bilgiler

1.52

1.38

 도시

1.38

mathrm

1.34

 매우

1.34

 발생

1.32

會有

1.30

 문서

1.28

Activations Density 0.020%