INDEX

Explanations

orders designated for publication

The neuron strongly fires on formal court‐opinion boilerplate—particularly uppercase headings and key procedural terms like “SUMMARY ORDER,” “ORDER,” “FILED,” “CITATION,” and other legal‐document markers.

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

 نے

-1.23

 उन्होंने

-1.21

 удалось

-1.12

 hadn

-1.03

蒻

-1.02

 jamón

-1.02

 gewesen

-1.00

Iv

-0.98

æt

-0.97

nachron

-0.97

POSITIVE LOGITS

 will

2.55

or

1.98

is

1.92

 that

1.88

 shall

1.48

 should

1.37

can

1.36

 must

1.35

may

1.31

 automatically

1.29

Activations Density 0.154%