INDEX

Explanations

UI element selection and actions

The neuron flags UI‐interaction and procedural instruction terms—especially names of controls (e.g. button, menu, select) and action prompts like “click” or step labels.

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

woch

-1.31

 Können

-1.30

 famí

-1.26

rehension

-1.21

 these

-1.19

 tych

-1.16

llten

-1.11

 teils

-1.11

 esos

-1.10

 Dingen

-1.10

POSITIVE LOGITS

 your

1.80

you

1.67

就可以了

1.30

 Tourisme

1.28

 then

1.27

 musíte

1.15

if

1.14

 persoonlijke

1.13

場合には

1.13

就可以

1.12

Activations Density 0.089%