INDEX

Explanations

to you, essential, sanctions

The neuron is primarily detecting very common English function words (especially short prepositions and articles like “for,” “as,” “to,” “an,” 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

but

-1.75

佼

-1.27

 complementar

-1.23

beten

-1.18

and

-1.17

ссмо

-1.16

ではまた

-1.16

凌天

-1.16

or

-1.15

 Quảng

-1.13

POSITIVE LOGITS

its

1.80

 they

1.73

其

1.58

 mantiene

1.34

 their

1.34

ctions

1.33

 Económica

1.30

alers

1.30

 otorga

1.26

 anunció

1.24

Activations Density 0.462%