INDEX

Explanations

The neuron activates on literal HTML/SVG tag names (e.g. “div”, “g”, “path”, “canvas”, “svg”, “text”).

foreign words, especially recipes

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

-1.87

and

-1.80

 Finally

-1.45

the

-1.34

And

-1.32

 being

-1.28

 Should

-1.26

 coming

-1.25

To

-1.22

 HttpNotFound

-1.22

POSITIVE LOGITS

 kalori

1.45

 varier

1.41

 profiter

1.41

 degust

1.39

Recept

1.38

 kulak

1.36

 bantu

1.36

 instante

1.36

 volant

1.34

 tarifi

1.34

Activations Density 0.003%