INDEX

Explanations

fictional work titles

The neuron detects titled headings—especially TV/movie episode or chapter names with their surrounding quotes, numbers, and capitalized words.

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

 saben

-1.48

 nyingi

-1.41

 schreef

-1.39

anhänger

-1.33

 inaweza

-1.32

SetPosition

-1.32

 bepa

-1.31

uties

-1.30

hülle

-1.30

 middag

-1.29

POSITIVE LOGITS

 even

1.59

où

1.46

where

1.41

of

1.41

our

1.38

 gosto

1.34

that

1.34

↵↵↵↵↵

1.32

you

1.30

:"-"`

1.29

Activations Density 0.038%