INDEX

Explanations

Rule, sensor, session, link, report, support, party, war, patient, track, volume

The neuron fires on isolated, content-bearing nouns—especially section- or title-level keywords (e.g. “variable,” “project,” “session,” “sensor,” “report,” “support,” “party,” “solution,” “war”)—marking major topical terms.

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

 both

-1.56

 also

-1.24

 even

-1.14

 only

-1.13

 Both

-1.06

 Unlike

-1.05

 Like

-1.03

 Generally

-1.02

 ambos

-1.02

 while

-1.02

POSITIVE LOGITS

 zondag

1.39

 viſ

1.38

ѕ

1.34

 hebben

1.32

 zaterdag

1.27

stolic

1.26

konzert

1.23

 guapos

1.20

бър

1.18

齶

1.16

Activations Density 0.245%