INDEX

Explanations

underscore variable names

This neuron detects promotional feature‐and‐offer phrases—action‐oriented marketing language like “This apartment features…,” “offers accommodation…,” “Download the free…,” or “provides….”

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

 intertw

0.88

 slurry

0.86

 bể

0.86

)}]{

0.84

 tetragonal

0.84

ным

0.83



0.83

इकिल

0.81

್ಣ

0.80

 railroads

0.80

POSITIVE LOGITS

FX

0.88

IB

0.82

AK

0.79

eni

0.74

AU

0.73

FN

0.72

0.71

й

0.71

FL

0.71

Activations Density 0.001%