INDEX

Explanations

citations and publication years

The neuron detects numeric tokens—especially numbers in citations, dates, page counts, and other reference notations.

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

ætter

-2.03

one

-1.66

 infatti

-1.51

 like

-1.47

 need

-1.42

if

-1.41

四年

-1.41

 either

-1.39

 first

-1.38

 require

-1.38

POSITIVE LOGITS

 แต่

1.44

1.41

 deveria

1.41

のですが

1.38

”

1.38

↵

1.37

 ۱۱

1.32

攥

1.31

but

1.30

 bästa

1.30

Activations Density 0.037%