INDEX

Explanations

links to code examples

The neuron activates on segments of web addresses (e.g. parts of hostnames like “codes & box” for codesandbox, “teepublic”, “facebook”, etc.)

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

any

-1.63

 with

-1.26

and

-1.26

 before

-1.25

on

-1.24

 without

-1.23

 that

-1.23

in

-1.21

or

-1.21

 große

-1.16

POSITIVE LOGITS

com

1.63

org

1.41

particularly

1.38

com

1.36

MÉ

1.32

 exuber

1.26

では

1.23

 innumer

1.22

 accor

1.20

 commer

1.18

Activations Density 0.097%