INDEX

Explanations

iOS/Swift code identifiers

The neuron activates on numeric literal values (especially floating‐point measurements) embedded in code.

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

 $^{\

-1.65

吀

-1.55

 députés

-1.41

甚麼

-1.34

quelize

-1.34

甗

-1.30

の地

-1.27

ベーション

-1.26

 Mexicano

-1.26

$^{

-1.24

POSITIVE LOGITS

 هذه

1.62

 like

1.57

käse

1.47

少ない

1.45

leun

1.42

and

1.40

 THIS

1.39

by

1.35

casse

1.31

 mães

1.30

Activations Density 0.003%