INDEX

Explanations

names ending in 'er' or 'ed'

The neuron fires on forum post author nicknames—i.e. it detects tokens that are parts of user handles or display names.

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

to

-2.13

 thermomix

-1.42

ornata

-1.34

 todavía

-1.33

 trico

-1.31

Aé

-1.29

 conster

-1.29

ikyuu

-1.26

 Brasileiro

-1.25

 />);

-1.25

POSITIVE LOGITS

 also

1.57

 those

1.44

Sch

1.40

alties

1.38

ggiato

1.36

 поскольку

1.32

aneity

1.32

ZIONI

1.32

 your

1.30

 restauración

1.29

Activations Density 0.107%