INDEX

Explanations

royalty and honors

The neuron detects formal honorific titles (e.g. “Her,” “Majesty,” “Royal,” “Highness,” “H.E.,” “H.M.”) that precede distinguished names.

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

-2.16

);

-1.93

and

-1.91

 following

-1.80

↵

-1.70

is

-1.64

vací

-1.62

 vigorously

-1.58

</strong>

-1.56

 received

-1.56

POSITIVE LOGITS

 youre

1.63

tics

1.62

͆

1.59

 Highness

1.56

czne

1.55

licitations

1.54

белый

1.52

lijke

1.52

abinieri

1.52

arschu

1.51

Activations Density 0.002%