INDEX

Explanations

southern districts

The neuron activates on numeric tokens—particularly integers and decimals—wherever they appear in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

if

-1.71

翃

-1.27

"];

-1.26

 akár

-1.20

"",

-1.15

anthrene

-1.13

如果

-1.11

',

-1.11

dieux

-1.10

"")

-1.09

POSITIVE LOGITS

 until

1.32

 through

1.29

 būti

1.27

НИМА

1.20

 should

1.20

ี้ยง

1.13

ƶ

1.12

佉

1.10

 inexpensive

1.10

xem

1.10

Activations Density 0.015%