INDEX

Explanations

academic and publication details

The neuron is specialized for spotting numeric tokens—especially digits in parentheses or citation‐style numbers like years, volume/page numbers, and identifiers.

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

ارتمان

-1.52

ктором

-1.48

orgung

-1.35

of

-1.34

சிய

-1.32

れました

-1.29

蔼

-1.28

CodeAnalysis

-1.25

 joueurs

-1.24

Tomorrow

-1.24

POSITIVE LOGITS

 encuentras

1.35

who

1.34

 have

1.27

 necesitas

1.24

–

1.18

也不敢

1.17

小姐

1.17

um

1.16

 kutoka

1.16

 consegue

1.16

Activations Density 0.040%