INDEX

Explanations

people who or those who

The neuron primarily fires on demonstrative and relative-clause words—tokens like “that,” “those,” “who,” and “there” (along with their accompanying forms of “is/are”).

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

 what

-1.68

if

-1.57

how

-1.55

What

-1.50

There

-1.42

 whats

-1.39

Also

-1.39

How

-1.38

They

-1.38

现

-1.36

POSITIVE LOGITS

of

1.49

 Muchos

1.38

 Septembre

1.32

动了

1.32

fof

1.31

 beso

1.30

 avon

1.30

1.28

ene

1.27

 lage

1.27

Activations Density 0.006%