INDEX

Explanations

brought to you by / was aimed at

The neuron fires specifically on numeric metadata and engagement‐metric tokens (e.g. content length, view/download counts, percentages) in article headers or footers.

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

to

-1.66

 kazak

-1.36

 vorrei

-1.34

 ſont

-1.31

Of

-1.30

 gustado

-1.28

 Özel

-1.28



-1.27

Привет

-1.26

']>

-1.26

POSITIVE LOGITS

ulously

1.48

 those

1.45

if

1.39

ガ

1.34

gensen

1.30

ższy

1.28

Possui

1.27

espejo

1.23

 estás

1.22

这两个

1.22

Activations Density 0.062%