INDEX

Explanations

finite mathematical properties

This neuron fires on technical mathematical adjectives that specify properties of objects (e.g. “closed,” “connected,” “finite,” “real,” “oriented,” “hyperbolic”).

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

to

-1.94

</b>

-1.88

not

-1.72

-1.64

-1.59

 instead

-1.51

to

-1.51

ge

-1.50

 gata

-1.44

for

-1.40

POSITIVE LOGITS

 къде

2.11

 combatir

1.77

 этого

1.66

 hierbas

1.64

 изпол

1.64

 която

1.59

értel

1.59

 защото

1.59

 daje

1.57

 sorpresas

1.56

Activations Density 0.113%