INDEX

Explanations

This neuron detects specialized technical and scientific terms—particularly multi-letter acronyms, formal nomenclature, and domain-specific jargon.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 polyvinyl

0.85

antum

0.72

édération

0.71

megapixel

0.70

uradaki

0.69

áln

0.66

iciencia

0.66

cherichia

0.66

 milliliters

0.66

 kilowatt

0.66

POSITIVE LOGITS

 SMEs

1.32

 РФ

1.30

XRD

1.30

1.29

RPG

1.27

1.26

 WWII

1.26

GHG

1.26

1.25

 СПб

1.24

Activations Density 1.053%