INDEX

Explanations

or

The neuron activates on the key domain-specific content words—nouns and verbs carrying the main legal or technical substance—in formal statutes and technical descriptions.

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

 just

-1.50

 like

-1.45

 will

-1.27

 near

-1.23

 even

-1.22

 very

-1.17

czego

-1.12

 prêtres

-1.11

 every

-1.06

 around

-1.06

POSITIVE LOGITS

无论是

1.37

ಆ

1.36

 combinatie

1.32

歟

1.31

Either

1.30

utveck

1.28

either

1.27

or

1.26

 Atención

1.24

смарт

1.20

Activations Density 0.031%