INDEX

Explanations

will followed by a verb

The neuron spotlights numeric expressions (counts, percentages, and other data values) 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

ätigung

-1.40

*"

-1.29

"?

-1.27

¼

-1.24

fferty

-1.23

THEN

-1.19

~"

-1.18

quit

-1.17

TOO

-1.16

(“

-1.16

POSITIVE LOGITS

 will

1.76

 this

1.53

one

1.47

 belki

1.41

 scattata

1.39

 Often

1.35

今も

1.27

may

1.25

or

1.23

 innumerable

1.19

Activations Density 0.065%