INDEX

Explanations

how to use or change

The neuron fires on short, high-frequency function words—especially articles and simple prepositions/conjunctions (e.g. “un,” “une,” “as,” “como,” “all”).

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

urnya

-1.52

要

-1.43

幸好

-1.43

but

-1.42

ooooo

-1.41

 coexistence

-1.41

'{

-1.38

According

-1.37

oooo

-1.36

also

-1.36

POSITIVE LOGITS

苼

1.62

茀

1.60

痦

1.59

棪

1.57

olding

1.52

沨

1.49

takes

1.48

໊

1.47

荩

1.46

嗩

1.42

Activations Density 0.006%