INDEX

Explanations

dimensions and spacing

This neuron fires on hyphen‐separated class or style fragments that encode layout properties and numeric values (e.g. “nopadding,” “max,” “height,” “1,” or “l–row,” “r–px”).

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

 because

-1.81

 while

-1.74

 first

-1.68

 Because

-1.50

our

-1.49

because

-1.49

 many

-1.42

 from

-1.38

 coupled

-1.37

 ostensibly

-1.37

POSITIVE LOGITS



1.93

 dando

1.72

𐬹

1.59

 Magyar

1.47

 Mereka

1.40

;&

1.39

摁

1.39

☫

1.39

setArguments

1.36

撂

1.34

Activations Density 0.003%