INDEX

Explanations

from X to Y

The neuron fires on numeric tokens (i.e. standalone numbers) in the text.

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

 both

-1.64

 from

-1.46

in

-1.34

 bakgrund

-1.21

 because

-1.20

 primarily

-1.16

or

-1.16

for

-1.16

 categorized

-1.15

 there

-1.13

POSITIVE LOGITS

 downright

1.54

 zelfs

1.43

 addirittura

1.42

 finally

1.40

 eventually

1.34

 πιο

1.31

最後

1.27

 более

1.26

 حتی

1.24

 being

1.23

Activations Density 0.046%