INDEX

Explanations

criminal charges and offenses

The main thing this neuron does is find words naming or describing criminal offenses.

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

 élar

-1.73

 kampen

-1.70

 that

-1.64

 rafraî

-1.63

 každý

-1.63

íso

-1.63

 něj

-1.60

 réserver

-1.59

蓠

-1.55

 diminuer

-1.55

POSITIVE LOGITS

and

2.53

on

2.14

at

2.03

2.00

for

1.98

in

1.98

if

1.73

as

1.69

 then

1.65

1.56

Activations Density 0.032%