INDEX

Explanations

ANY or ALL

The neuron flags tokens used as inline emphasis or quoting markers (e.g. standalone asterisks or mismatched single-quote marks).

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

reiches

-1.25

小行星

-1.19

至尊

-1.18

kond

-1.16

[])

-1.16

[]):

-1.15

 Ejecutivo

-1.09

gebruik

-1.08

 garagem

-1.06

Organization

-1.05

POSITIVE LOGITS

idance

1.52

 didn

1.39

 zakoń

1.34

しと

1.33

the

1.30

 lname

1.28

しなかった

1.26

 coloque

1.24

 assumes

1.22

中に

1.22

Activations Density 0.168%