INDEX

Explanations

patent application Ser. No

The neuron detects patent‐style metadata fields—especially serial and application numbers (e.g. “Ser. No.”, “filed”), dates, and continuation identifiers.

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

したり

-1.52

绘画

-1.41

middag

-1.33

讀書

-1.32

 craindre

-1.31

還在

-1.30

 denkt

-1.29

okat

-1.29

 meines

-1.28

 prends

-1.28

POSITIVE LOGITS

 will

1.48

will

1.47

 lejos

1.44

return

1.42

 menggunakan

1.38

ffled

1.34

but

1.32

畲

1.32

 etwas

1.31

чення

1.30

Activations Density 0.308%