INDEX

Explanations

static_cast<char>

The neuron activates on code‐style identifiers—especially long, CamelCase or PascalCase names of classes, types, methods, or API symbols.

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

 main

0.56

 ಹಿಂದೆ

0.55

 depende

0.54

ພາ

0.52

ێن

0.52

 stimmen

0.51

 foll

0.50

 parecen

0.50

ერ

0.50

isien

0.50

POSITIVE LOGITS

1.98

<(

1.49

<{

1.49

1.48

<\

1.48

<?>

1.39

::<

1.34

$<\

1.33

\<

1.28

/<

1.27

Activations Density 0.007%