INDEX

Explanations

schedule of the

This neuron detects document structural markers, particularly numbered section headings and their titles (e.g., “Schedules 3,” “Description of the Related Art,” numeric labels).

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

 插图

-1.37

vj

-1.32

 tera

-1.32

 tory

-1.30

 nguyen

-1.28



-1.28

"-"

-1.27

 dons

-1.27



-1.27

內容簡介

-1.26

POSITIVE LOGITS

the

1.93

in

1.83

 всех

1.63

 than

1.58

 that

1.57

at

1.46

 usual

1.45

of

1.45

all

1.45

but

1.38

Activations Density 0.062%