INDEX

Explanations

legal and statistical terms

The neuron fires strongly on content words in formal or official‐style text—especially headings, titles, dates, and key terms in proclamations, policies, and similar documents.

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

 قوات

-1.16

Tentang

-0.92

も多く

-0.92

 luks

-0.88

جغرافيا

-0.87

 nikah

-0.86

 kitab

-0.85

 القوات

-0.85

adaptiveStyles

-0.84

 verfügb

-0.84

POSITIVE LOGITS

Österreich

1.12

mnop

1.00

 inspiradoras

0.99

揾

0.98

 Österreich

0.98

 characterised

0.98

 mūsų

0.97

 regeringen

0.96

sữa

0.94

וּ

0.94

Activations Density 0.596%