INDEX

Explanations

title, name, or date separators

The neuron fires on numbered or enumerated headings and other section-number markers (i.e. digits and numerical tokens used in titles and outline numbering).

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

ھی

-1.52

basically

-1.32

また

-1.30

玉ねぎ

-1.27

{?>

-1.26

 however

-1.25

рована

-1.25

Якщо

-1.25

还不错

-1.22

(';

-1.21

POSITIVE LOGITS

 recently

1.46

so

1.31

can

1.27

 podríamos

1.23

报记者

1.20

 sometimes

1.19

new

1.17

 table

1.17

 Sometimes

1.17

 hand

1.16

Activations Density 0.652%