INDEX

Explanations

technical details and specific entities

This neuron detects MediaWiki namespace bracket syntax (e.g. the “[[:Category:…]]” pattern).

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

describing

-1.23

 inept

-1.18

 composants

-1.17

 facteurs

-1.17

ふわ

-1.16

いつか

-1.15

 actifs

-1.14

 règles

-1.14

摟

-1.13

 besoins

-1.12

POSITIVE LOGITS

at

1.52

as

1.52

 without

1.46

by

1.39

and

1.21

 which

1.20

 ostensibly

1.20

or

1.20

 while

1.20

 that

1.19

Activations Density 0.097%