INDEX

Explanations

questions and dialogue

The neuron detects direct reader addresses in questions—especially “you” combined with question or ask phrases.

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

鼇

-1.39

-1.38

 hilfreich

-1.27

っています

-1.27

 gustado

-1.20

 sympathique

-1.19

đầm

-1.16

ﮩ

-1.14

 sukses

-1.14

 enacted

-1.14

POSITIVE LOGITS

but

1.92

but

1.44

us

1.41

But

1.37

 мето

1.28

 लेकिन

1.27

 except

1.25

 국가

1.25

 ولی

1.24

vér

1.23

Activations Density 0.025%