INDEX

Explanations

By offering context after a phrase

The neuron is attuned to document metadata lines—spiking on by‐lines, author names, dates, headers, etc.

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

in

-1.56

},

-1.44

ing

-1.42

-1.40

-1.34

$,

-1.33

 Later

-1.28

no

-1.27

if

-1.27

-1.22

POSITIVE LOGITS

Already

1.54

oló

1.50

 pérd

1.49

ޏ

1.44

ISTAS

1.44

 już

1.39

ρίες

1.37

 already

1.37

CHREIB

1.35

៖

1.34

Activations Density 0.068%