INDEX

Explanations

people, relationships, audience, organization

The neuron selectively activates on open-class “content” words—i.e. nouns, verbs, adjectives/adverbs, and quantifiers—rather than on function words or punctuation.

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

at

-1.84

in

-1.64

for

-1.64

as

-1.57

on

-1.55

 with

-1.54

 when

-1.54

 epitom

-1.52

 after

-1.52

so

-1.48

POSITIVE LOGITS

傒

1.65

 .........

1.64

 balkon

1.64

 indiví

1.64



1.63

ఽ

1.62

鵙

1.62

<0xF4>

1.59

 ..........

1.58

 pérd

1.57

Activations Density 3.217%