INDEX

Explanations

license urls

The neuron fires strongly on URL‐style tokens (e.g. “http”, “://”, domain/path segments) and other resource locator strings in code headers.

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

 while

-1.46

 numerous

-1.34

 several

-1.29

 both

-1.27

 other

-1.27

会社の

-1.24

two

-1.24

 focusing

-1.19

 focused

-1.16

 three

-1.12

POSITIVE LOGITS

 fidé

1.34

ことも

1.30

ghest

1.29

 euró

1.29

entist

1.25

 discuter

1.24

 gän

1.23

鵙

1.23

 ACTUALLY

1.23

}}$,

1.23

Activations Density 0.056%