technical terms and foreign words

The neuron fires strongly on uncommon, domain-specific keywords or proper-noun–style tokens (i.e. rare content words that identify a topic).

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

-1.35

-1.32

-1.30

—

-1.30

-1.27

-1.26

-1.22

you

-1.20

POSITIVE LOGITS

 sokak

1.47

 bluz

1.44

Válasz

1.34

Hozzá

1.31

 parfüm

1.27

 みた

1.23

fruta

1.22

 miniaturka

1.20

Για

1.17

 Rumuni

1.16

Activations Density 0.027%