INDEX

Explanations

The neuron is picking out imperative instruction steps—action verbs like “Tap,” “Click,” “Hold,” and similar UI/navigation commands.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

;-)

0.66

܀

0.63

:-)

0.62



0.60

ικη

0.59

 попада

0.58

 नामक

0.56

امة

0.56

。「

0.55

 curviliné

0.54

POSITIVE LOGITS

<strong>

1.98

<b>

1.81

mathbf

1.22

1.20

 mathbf

1.12

boldsymbol

0.89

//

0.87

<!--

0.86

**

0.79

<code>

0.79

Activations Density 0.202%