INDEX

Explanations

That's [adjective]

This neuron detects subjective evaluative words—particularly sentiment-laden adjectives (like “nice,” “great,” “helpful,” “easy,” or “terrible”) that express an opinion.

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

by

-1.55

 notte

-1.41

 through

-1.39

to

-1.39

it

-1.30

噥

-1.30

via

-1.27

菹

-1.27

 adquirido

-1.26

zembro

-1.23

POSITIVE LOGITS

 ciebie

1.70

 that

1.63

1.52

1.48

あなたが

1.43

 что

1.32

 normalmente

1.31

 trastornos

1.30

 maailman

1.27

 که

1.25

Activations Density 0.011%