INDEX

Explanations

The neuron fires on section headings and titles—i.e. the capitalized or prominent header text that introduces or labels a block of content.

About the English Essays

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

so

-1.75

 when

-1.71

at

-1.71

in

-1.60

of

-1.56

 which

-1.52

had

-1.50

 their

-1.47

 being

-1.45

 these

-1.43

POSITIVE LOGITS

 revêtement

1.41

 élastique

1.33

 tôt

1.29

 verrou

1.27

 trochu

1.27

 zimní

1.24

 rafraî

1.24

 ไม่

1.24

よりも

1.23

 två

1.22

Activations Density 0.126%