INDEX

Explanations

acts for subjects followed by descriptions

The neuron fires on prominent title or section‐heading words (especially the first few words of headings).

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

 także

-1.23

inį

-1.19

 également

-1.16

גם

-1.16

and

-1.16

 וגם

-1.14

 also

-1.11

愀

-1.11

謫

-1.10

 wers

-1.09

POSITIVE LOGITS

is

2.38

are

1.50

has

1.44

はとても

1.21

 påvir

1.18

 давно

1.15

 αποτε

1.15

Often

1.13

 often

1.13

OLUTION

1.13

Activations Density 0.746%