INDEX

Explanations

citations and section markers

The neuron is picking up on non-word formatting and structural punctuation tokens (e.g. asterisks, braces, carets, equals signs) that mark sections, emphasis, or other document layout elements.

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

 have

-1.31

。"

-1.25

of

-1.25

𝓋

-1.22

 compromis

-1.22

by

-1.21

haup

-1.20

 captur

-1.19

in

-1.15

has

-1.13

POSITIVE LOGITS

those

1.61

 those

1.48

 различных

1.44

 estekak

1.39

 Ausführungen

1.38

 modificaciones

1.38

 respetu

1.38

 reducido

1.35

 различ

1.34

 them

1.32

Activations Density 0.137%