INDEX

Explanations

sizes, numbers, letters, and Q:

The neuron selectively fires on numeric tokens—especially those in product‐size or measurement tables (e.g. UK/US/EU size numbers and related values).

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

↵↵

-1.49

：……

-1.38

に行ってきました

-1.30

 المسؤول

-1.27

に行きました

-1.25

on

-1.24

 получает

-1.23

 знает

-1.18

 thậm

-1.17

 видел

-1.09

POSITIVE LOGITS

ׇ

1.43

lifies

1.33

 your

1.31

 ilmu

1.30

ان

1.29

like

1.27

как

1.26

೭

1.24

Ɂ

1.21

 Nomor

1.20

Activations Density 0.032%