INDEX

Explanations

The neuron detects major capitalized keywords in formal titles and names—especially the distinctive words in the names of laws, acts, institutions, and technical or organizational entities.

broad categories followed by specifics

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

 jornalista

-1.66

 aktive

-1.55

 tesoro

-1.48

 nativo

-1.48

 emblema

-1.45

 corazon

-1.45

 cantora

-1.43

 castelo

-1.41

 obat

-1.40

が

-1.40

POSITIVE LOGITS

and

1.50

For

1.41

 betrek

1.41

 That

1.38

 Since

1.37

 angesehen

1.29

رياض

1.27

もので

1.25

 These

1.20

 постоянно

1.20

Activations Density 0.118%