INDEX

Explanations

good for you

The neuron fires on evaluative “X is good/bad/better for Y” phrases—i.e. adjective-for-object constructions that say something is beneficial or harmful to something.

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

璈

-1.53

guigu

-1.47

borsa

-1.41

 двумя

-1.34

StateToProps

-1.34

いちご

-1.32

ᥬ

-1.28

 desgra

-1.27

</i>

-1.27

 gewel

-1.26

POSITIVE LOGITS

1.55

can

1.48

1.47

1.43

{~

1.41

_{

1.41

可以

1.40

ner

1.38

ut

1.35

 televisiva

1.35

Activations Density 0.021%