INDEX

Explanations

purpose and intentionality

The main thing this neuron does is detect words about purpose or intentionality (e.g., “purpose,” “intentional,” “purposeful”).

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

the

-2.25

粁

-1.75

 telles

-1.73

supports

-1.67

 отношениях

-1.61

 encabez

-1.57

 désactiv

-1.55

 achet

-1.55

 unſ

-1.52

鉿

-1.50

POSITIVE LOGITS

1.34

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.33

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.32

and

1.30

 each

1.28

 does

1.25

之路

1.25

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.24

 octubre

1.24

↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.23

Activations Density 0.009%