INDEX

Explanations

title or section

This neuron detects headline-style separators and dividers—single-character punctuation such as dashes, pipes, and em-dashes.

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

 njego

-1.54

펴

-1.48

.}(

-1.41

𝒈

-1.34

}}}{

-1.31

 непосред

-1.30

;%%

-1.30

鎘

-1.30

ものである

-1.29

 επίσης

-1.28

POSITIVE LOGITS

 their

1.54

1.52

an

1.51

1.50

any

1.42

 spesielt

1.38

 many

1.33

 then

1.30

 transmissão

1.28

 until

1.27

Activations Density 0.133%