INDEX

Explanations

things that follow

The neuron fires on occurrences of the root “follow” (in any inflected form such as follow, follows, followed).

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

of

-1.88

geſ

-1.57

ong

-1.51

hline

-1.48

'_

-1.45

re

-1.41

dvi

-1.41

氿

-1.41

силання

-1.39

ik

-1.38

POSITIVE LOGITS

How

1.71

 verksamhet

1.63

 pensé

1.59

 pensó

1.55

 Instead

1.54

狁

1.52

 vänner

1.51

 What

1.51

風格

1.50

するので

1.48

Activations Density 0.020%