INDEX

Explanations

names and company information

The neuron fires on the capitalized name parts in heading lines—i.e. proper‐name tokens (personal surnames or organizational name segments) in article titles.

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

how

-1.24

why

-1.14

 then

-1.13

 laikā

-1.13

-1.09

 that

-1.05

 what

-1.02

istiwa

-1.01

bedaan

-1.00

POSITIVE LOGITS

tifiable

1.26

ΙΚ

1.25

irical

1.23

⿴

1.22

IGEN

1.22

IRMED

1.21

hluk

1.20

ATIONS

1.20

★★★★☆

1.19

鶩

1.18

Activations Density 0.268%