INDEX

Explanations

year ranges

The neuron fires on multi-digit numeric tokens (e.g. years, version numbers, large numbers).

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

 from

-1.65

of

-1.52

in

-1.34

 aspects

-1.27

–

-1.24

 since

-1.20

for

-1.16

//

-1.16

so

-1.16

除此之外

-1.16

POSITIVE LOGITS

gruss

1.43

fondos

1.36

 Franse

1.33

podr

1.31

schwar

1.23

 muligt

1.19

wurde

1.15

呸

1.15

کفش

1.15

 denna

1.14

Activations Density 0.077%