INDEX

Explanations

bending positions

the main thing this neuron does is detect descriptions of a person or object bending or leaning over (e.g., “bending,” “leaning over,” “stooping”).

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

 damaligen

-1.65

-1.40

 them

-1.40

 käytet

-1.33

 then

-1.28

 vasto

-1.27

niejszych

-1.24

 所以

-1.20

 vähän

-1.20

</i>

-1.20

POSITIVE LOGITS

፨

1.22

odeficiency

1.21

HTH

1.17

 една

1.15

 завжди

1.14

LLYWOOD

1.13

нева

1.12

磊

1.12

íferos

1.10

 unattractive

1.09

Activations Density 0.017%