INDEX

Explanations

technical documentation and code snippets

The neuron fires on code‐style or markup tokens—especially uppercase identifiers, tags, and technical names (e.g. class names, HTML tags, license headers).

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

 januar

-0.93

логу

-0.83

 februar

-0.81

LogFactory

-0.80

topRight

-0.80

めんな

-0.77

seille

-0.77

ńskich

-0.75

кто

-0.75

lipo

-0.74

POSITIVE LOGITS

 style

0.98

Estilo

0.94

raises

0.90

gaat

0.82

 postulated

0.81

 stijl

0.80

年も

0.79

 スタイル

0.79

取决

0.78

 styles

0.78

Activations Density 0.000%