INDEX

Explanations

role of, effects of, consequences of

The neuron fires on prominent content words (especially nouns) in academic‐style titles or headings—typically the head noun in “X of Y” constructions.

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

as

-2.19

 with

-2.17

by

-1.98

on

-1.80

 without

-1.76

 while

-1.72

in

-1.70

for

-1.69

 like

-1.63

 that

-1.52

POSITIVE LOGITS

 rése

1.46

gâ

1.41

 réellement

1.38

gelopen

1.38

 sappi

1.36

 bluza

1.36

nieuw

1.34

 sûrement

1.33

 remé

1.33

 provincie

1.31

Activations Density 0.263%