INDEX

Explanations

pop up, kicks, themes

The neuron is strongly triggered by rare, content-bearing nouns or proper names—i.e. low-frequency, domain-specific tokens.

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

 some

-1.17

if

-1.06

谁能

-1.03

et

-1.01

 keine

-1.00

get

-0.96

can

-0.96

not

-0.95

-0.94

POSITIVE LOGITS

 ekolog

1.27

 obé

1.26

 goû

1.26

 siè

1.25

kyt

1.24

 demokrat

1.23

 balón

1.23

 konflikt

1.23

 triko

1.20

srp

1.20

Activations Density 0.086%