INDEX

Explanations

lists of specific nouns and states

The neuron fires strongly on content-bearing “real” words (nouns, verbs, adjectives) and stays silent on common function words.

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

of

-2.00

 instead

-1.68

-1.59

龶

-1.59

ungalows

-1.53

 along

-1.48

-1.41

 stijl

-1.34

 المدن

-1.34

 they

-1.32

POSITIVE LOGITS

Vä

1.55

Ostat

1.50

 sacré

1.47

 Erişim

1.46

ك

1.41

Wszyst

1.40

現在

1.40

Odpo

1.39

Teraz

1.39

 zimowe

1.38

Activations Density 0.014%