INDEX

Explanations

c. often followed by digits or 'utf-8'

The neuron activates on copyright notices, spotting the “©” or “(c)” marker (and its associated year).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

сил

0.57

levelname

0.57

 iliş

0.56

шот

0.56

stanford

0.56

нением

0.55

 cargar

0.55

莘

0.55

 дре

0.54

шения

0.54

POSITIVE LOGITS

्थ

0.58

 Kwiat

0.56

 Uber

0.55

 Locks

0.54

 Ports

0.53

้ำ

0.52

 キャップ

0.52

 Blacks

0.52

Kow

0.52

 Harbor

0.51

Activations Density 0.001%