INDEX

Explanations

plaintiff, defendant, or appellant

The neuron strongly responds to sequences of uppercase tokens in formal document headings—e.g. legal case captions, party names, and boilerplate titles.

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

 花朵

-1.45

 ressemble

-1.44

Nhưng

-1.38

 écrit

-1.38

 Père

-1.36

 arbe

-1.34

 vack

-1.34

 Turki

-1.33

 Tré

-1.32

 illustre

-1.32

POSITIVE LOGITS

and

2.39

as

1.70

to

1.63

or

1.59

 including

1.56

 sowie

1.51

 with

1.49

has

1.46

 during

1.45

以及

1.39

Activations Density 0.118%