INDEX

Explanations

in a specific context or state

This neuron detects key domain-specific content words—particularly technical terms, identifiers, and other semantically rich nouns.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 drau

-1.44

 perspek

-1.44

 décider

-1.42

ᡴ

-1.39

小さく

-1.39

 struktur

-1.37

othes

-1.33

衚

-1.33

 saus

-1.31

 Strukt

-1.31

POSITIVE LOGITS

 with

1.82

}}{\

1.48

 requer

1.30

 that

1.28

and

1.27

1.24

 jusqu

1.22

in

1.22

смотрим

1.22

 which

1.20

Activations Density 0.014%