INDEX

Explanations

one or two

The neuron specializes in detecting small-quantity expressions—especially low cardinal numbers and brief quantifier phrases like “two,” “a couple,” “a few,” or “once or twice.”

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

龊

-1.33

任何

-1.13

 miniatur

-1.12

 bougie

-1.09

 standar

-1.09

无数

-1.08

 offensi

-1.06

ᐟ

-1.05

tttt

-1.05

 personnels

-1.05

POSITIVE LOGITS

but

1.37

few

1.27

few

1.26

Quelques

1.23

(<

1.20

 wenigen

1.14

(~

1.13

of

1.13

几个

1.12

čas

1.12

Activations Density 0.068%