INDEX

Explanations

class and object declarations

The neuron fires on numeric literals (integer constants) in the code.

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

粧

-0.93

uesta

-0.77

 '/':

-0.74

みに

-0.73

thermal

-0.72

companion

-0.72

=="")

-0.71

sólo

-0.71

-0.70

POSITIVE LOGITS

 class

1.35

class

1.24

 object

0.98

 const

0.93

object

0.90

ventud

0.90

 open

0.89

const

0.87

 deschis

0.83

data

0.82

Activations Density 0.007%