INDEX

Explanations

introduction

The neuron activates on front‐matter and editorial headings—labels such as “Introduction,” “Foreword,” “Edited by,” etc.

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

is

-1.70

电视剧

-1.52

 její

-1.49

 verkligen

-1.38

ーズン

-1.31

୮

-1.31

twimg

-1.30

どう

-1.30

genomen

-1.30

AdvEx

-1.28

POSITIVE LOGITS

 Procu

1.31

lectable

1.24

foto

1.21

so

1.21

邙

1.20

鄒

1.17

濤

1.16

媵

1.15

spective

1.14

 Cámara

1.14

Activations Density 0.080%