INDEX

Explanations

events that precede others

This neuron activates on words that express temporal precedence (e.g. forms of “predate,” “precede,” “antedate,” 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

 forty

-1.45

↵↵

-1.45

 Púb

-1.41

 considerada

-1.33

if

-1.33

но

-1.29

is

-1.29

+=

-1.27

 considered

-1.27

вое

-1.24

POSITIVE LOGITS

 than

1.75

 diki

1.55

 plastiques

1.51

artiges

1.47

 بسیار

1.46

mittags

1.45

Quand

1.43

遢

1.43

 balles

1.42

 almofada

1.41

Activations Density 0.038%