INDEX

Explanations

qualifiers and summaries

The neuron fires on general-purpose discourse or evaluative words—those signals of overall judgment or modality (e.g. “overall,” “general,” “majority,” “others,” “really,” “right,” “can,” “will,” “I”).

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

 แต่

-2.28

but

-2.23

凄く

-1.61

但是

-1.58

BUT

-1.52

 tetapi

-1.52

 però

-1.48

 אך

-1.46

 azonban

-1.45

 nhưng

-1.42

POSITIVE LOGITS

it

2.45

 will

1.51

 could

1.27

are

1.17

—

1.16

岁

1.15

 especially

1.14

 that

1.14

as

1.12

 sowohl

1.11

Activations Density 0.072%