INDEX

Explanations

initials and punctuation

The neuron is keyed to uppercase single‐letter tokens (initials) and other short capitalized fragments that occur in personal names and header fields.

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

-2.28

 from

-1.67

for

-1.60

our

-1.59

 their

-1.56

as

-1.55

-1.48

by

-1.40

 живет

-1.40

 even

-1.35

POSITIVE LOGITS

 некоторых

1.55

1.51

Belum

1.45

 algunas

1.45

有一些

1.44

 vissa

1.41

 aproxim

1.40

 levou

1.34

 ralenti

1.31

—

1.30

Activations Density 0.008%