INDEX

Explanations

patent or prior

The neuron selectively activates on the capitalized article “The” when it appears as the leading word of formal patent-section statements.

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

↵

-2.03

of

-1.91

 then

-1.72

</h1>

-1.70

ma

-1.69

卺

-1.68

and

-1.66

-1.58

ní

-1.58

 kön

-1.53

POSITIVE LOGITS

 själ

1.90

 Regardless

1.79

 Both

1.75

 Although

1.73

dersfield

1.67

"-"

1.67

malıdır

1.63

 soiled

1.62

 Despite

1.59

"/"

1.55

Activations Density 0.094%