INDEX

Explanations

declarative statements following "you" or "the"

The neuron flags user-facing message fragments—especially sentences that begin with “You” (e.g. “You need…,” “You are…,” “You have…”) or similar address forms like “There are…” or “Please…”

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

-1.81

on

-1.79

也

-1.70

 also

-1.64

-1.60

lu

-1.50

ază

-1.39

</sup>

-1.39

now

-1.38

ur

-1.37

POSITIVE LOGITS

1.76

НИМА

1.59

 serão

1.57

wußt

1.52

you

1.44

 réellement

1.41

 zarówno

1.40

Technologie

1.37

先是

1.37

tatuajes

1.35

Activations Density 0.067%