INDEX

Explanations

Simpson Magazine

The neuron fires on formatted reference citations—strings that look like legal or scholarly footnote markers (case names or journal titles followed by volume/report numbers, pages, dates, etc.).

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

 帅气

-2.44

 rése

-2.33

 miniaturka

-2.31

 bluz

-2.17

 pittores

-2.16

 devra

-2.14

€”

-2.13

 própri

-2.05

 zoude

-2.02

秫

-2.02

POSITIVE LOGITS

 There

2.39

for

2.31

If

2.22

on

2.14

2.11

2.09

At

2.03

 ….

2.00

no

1.98

Activations Density 0.000%