INDEX

Explanations

connecting words and phrases

This neuron spots high-frequency English function words—especially clause-linking conjunctions and discourse markers (e.g. as, if, when, since, the, there).

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

 stol

-1.40

しました

-1.38

ństwa

-1.20

gefügt

-1.20

%),

-1.19

がって

-1.17

 satisfe

-1.14

法则

-1.14

возможно

-1.14

чек

-1.14

POSITIVE LOGITS

and

1.94

 that

1.81

to

1.75

⤞

1.46

 läsa

1.37

𐄁

1.33

的一种

1.30

さてさて

1.30

 permitiendo

1.27

 Svizzera

1.26

Activations Density 0.031%