INDEX

Explanations

drawn to

The neuron fires on past‐participle verbs of attraction (like “drawn” or “attracted”) when they occur in “to”-phrases indicating something drawing interest.

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

If

-1.84

 With

-1.81

Since

-1.77

()

-1.74

</i>

-1.70

With

-1.67

 Although

-1.66

 Since

-1.66

 Unless

-1.65

剤

-1.63

POSITIVE LOGITS

 begrij

1.51

 ľudí

1.50

 denkt

1.48

 creciente

1.48

relatively

1.45

でて

1.41

の準備

1.41

 alluring

1.41

 ―――

1.41

不是很

1.40

Activations Density 0.011%