INDEX

Explanations

spatial separation and scale

The neuron lights up on words and phrases that signal global or spatial distance crossings—e.g. “world,” “across,” “other side,” “borders,” “continents.”

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

is

-1.34

но

-1.26



-1.25



-1.25

чя

-1.22

cæ

-1.18

 journal

-1.16

 antenn

-1.13

向上

-1.13

…’

-1.12

POSITIVE LOGITS

 strerror

1.38

but

1.35

uasion

1.34

("")

1.33

 belki

1.27

preise

1.26

 diez

1.26

却不

1.24

 yapa

1.23

 müm

1.22

Activations Density 0.036%