INDEX

Explanations

data class or terms of

The neuron fires on license‐header boilerplate in source files—especially copyright notices, license names, years, and “terms” wording.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

POK

-0.76

Vortex

-0.73

lamine

-0.71

 caudal

-0.71

 HYPER

-0.70

foobar

-0.70

Brin

-0.69

lex

-0.68

vd

-0.68

 Vortex

-0.68

POSITIVE LOGITS

Sch

0.71

 kann

0.70

throated

0.70

приєм

0.69

 Math

0.68

Cal

0.67

ériale

0.65

context

0.65

 Random

0.65

at

0.64

Activations Density 0.058%