INDEX

Explanations

states, limitations, or dependencies

This neuron spots strong emphasis words—particularly high-importance qualifiers (e.g. banned, must, high, long, usually) that signal key constraints or intensity.

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

as

-1.59

of

-1.55

 όμως

-1.32

艸

-1.30

 Krzysz

-1.25



-1.23

ově

-1.23

 mejor

-1.23

ovým

-1.23

 však

-1.20

POSITIVE LOGITS

ikusbot

1.41

it

1.40

The

1.39

 here

1.33

êter

1.32

its

1.27

не

1.26

soort

1.26

鉺

1.24

⇹

1.23

Activations Density 0.098%