INDEX

Explanations

doubling and halving

The neuron fires on words that describe multiplicative numerical changes (e.g. “doubles,” “doubled,” “halve,” “halving”).

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

 inklusive

-1.35

is

-1.31

 másik

-1.24

 Министер

-1.20

Gives

-1.19

jogo

-1.16

levant

-1.14

 reflejo

-1.12

艾

-1.12

POSITIVE LOGITS

 consente

1.32

//

1.30

waltung

1.30

是一種

1.26

論

1.25

+='

1.24

 présidenti

1.23

 interessa

1.23

日間

1.22

敷地

1.22

Activations Density 0.039%