INDEX

Explanations

clickable components and links

The neuron strongly activates on programming–style identifiers and property names (e.g. camelCase or PascalCase tokens like “userInteractionEnabled,” “clickable,” “tabIndex,” “selectableItemBackground,” etc.).

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

-1.43

of

-1.09

 に関する

-1.05

щается

-1.02

 する

-1.00

 Pág

-0.99

roja

-0.98

in

-0.96

 podľa

-0.96

PageState

-0.96

POSITIVE LOGITS

 mouse

1.00

ıkları

1.00

嘉宾

0.99

Bibliografie

0.97

 británica

0.95

他们

0.94

 BIBLE

0.94

 onClick

0.93

ôts

0.91

 prisoners

0.90

Activations Density 0.001%