INDEX

Explanations

one or both

The neuron preferentially activates on common discourse and connective words—like just, what, indeed, both, have, been, of—that help structure or qualify a statement.

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

 excès

-1.70

 refroid

-1.69

 déchets

-1.50

 défauts

-1.50

 cérami

-1.50

 compéti

-1.50

 étan

-1.48

 dépenses

-1.47

 pertes

-1.46

 czasu

-1.45

POSITIVE LOGITS

and

2.11

or

2.08

 from

1.91

for

1.90

 which

1.50

as

1.50

at

1.40

in

1.37

„

1.35

is

1.34

Activations Density 0.112%