INDEX

Explanations

start of proper names

The neuron fires on capitalized words that are names or titles—i.e. proper nouns (people’s names, brands, place names, etc.).

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

deutig

-1.30

aktivi

-1.28

Locations

-1.27

 defekt

-1.27

vue

-1.25

}:${

-1.23

❩

-1.23

 Ajouter

-1.23

 LOCK

-1.21

 Location

-1.20

POSITIVE LOGITS

 what

1.51

it

1.35

ஂ

1.34

don

1.27

 that

1.27

 inneren

1.26

how

1.24

 each

1.23

žití

1.22

 This

1.19

Activations Density 0.387%