INDEX

Explanations

The neuron strongly activates on proper nouns—especially country and organization names or other capitalized entities.

countries and regions

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

Läs

-1.70

 entusi

-1.64

 ritard

-1.63

ׇ

-1.57

ὴν

-1.55

 esat

-1.53

――――

-1.52

zestaw

-1.51

 gelas

-1.48

Artículos

-1.48

POSITIVE LOGITS

 бампер

2.28

וּ

2.11

ִּ

1.83

ּוֹ

1.76

ﻷ

1.59

фильм

1.59

ְּ

1.56

ֶּ

1.54

 currently

1.35

 сзади

1.35

Activations Density 0.008%