INDEX

Explanations

court citations

This neuron detects legal case‐citation reporter abbreviations and edition markers (e.g. “U.S.”, “Ct.”, “Ed.”, “2d”).

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

for

-1.75

 skolen

-1.64

 أيضا

-1.58

and

-1.58

 voltak

-1.55

iapkan

-1.49

 knop

-1.48

 також

-1.48

 toont

-1.47

역사

-1.45

POSITIVE LOGITS

 where

1.97

ية

1.95

him

1.65

1.56

行った

1.55

will

1.55

 "..\..\

1.54

的血

1.52

 Может

1.52

www

1.52

Activations Density 0.036%