INDEX

Explanations

links present website details

This neuron flags user‐interface and navigational cues—imperative words and labels (e.g. “please,” “click,” “select,” “contact”) used to guide actions on a web form or page.

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

on

-1.54

 both

-1.53

in

-1.38

 errone

-1.34

at

-1.30

 those

-1.29

 from

-1.27

 nine

-1.25

any

-1.24

 various

-1.20

POSITIVE LOGITS

 этой

1.30

küche

1.19

 нашей

1.13

 criterios

1.10

 gleiche

1.10

 أكثر

1.09

 homicidio

1.09

 alcuni

1.08

 černé

1.07

 наших

1.05

Activations Density 0.050%