INDEX

Explanations

excited greetings and exclamations

This neuron is detecting specific named entities or unique capitalized terms (e.g., product names, proper nouns) within the 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

PROXY

0.53

 уроке

0.49

Em

0.49

 Pentec

0.48

QSO

0.47

 ಕರೆ

0.47

Ze

0.46

 twig

0.46

 деко

0.46

 گئی۔

0.46

POSITIVE LOGITS

同士

0.59

ites

0.59

 बड़ी

0.58

以前

0.56

 pesky

0.54

rijven

0.53

versible

0.52

ियत

0.52

ビング

0.51

ற்ற

0.50

Activations Density 0.007%