INDEX

Explanations

continents, regions, and directions

The neuron fires on capitalized tokens that are parts of place names or geographic region names (e.g., continents, countries, cities, fictional realms).

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

ख़

-0.92

 storie

-0.91

在

-0.90

を

-0.89

春

-0.89

Iba

-0.88

in

-0.87

 aliments

-0.86

 puta

-0.85

 Pyrenees

-0.84

POSITIVE LOGITS

 Europe

1.41

 Britain

1.31

 America

1.28

OREA

1.08

 Africa

1.07

 Canada

1.05

america

0.98

beria

0.98

 Australia

0.97

bekistan

0.97

Activations Density 0.123%