INDEX

Explanations

code syntax (`=`, `(`)

This neuron detects standout marketing and product‐related keywords—uncommon promotional or SEO terms in e-commerce/descriptive text.

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

0.62

auto

0.59

 Unternehmen

0.57

ume

0.55

 Telem

0.55

امية

0.54

 persoonlijke

0.54

accus

0.53

($

0.53

POSITIVE LOGITS

 const

0.80

 CONST

0.62

const

0.61

CONST

0.58

 Protected

0.57

 function

0.55

 protected

0.55

var

0.54

 constant

0.54

 Const

0.54

Activations Density 0.020%