INDEX

Explanations

book editions and publishers

This neuron fires on bibliographic metadata—especially numerical tokens such as publication years, edition numbers, page counts, and similar numeric references.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Publishers

0.86

 Publisher

0.85

 Elsevier

0.83

 publisher

0.81

 Kluwer

0.78

 Verlag

0.77

 Publishing

0.77

 publishers

0.73

 Springer

0.72

eds

0.66

POSITIVE LOGITS

묘

0.57

第二

0.53

 δεύτε

0.53

 tượng

0.53

вторых

0.52

 disegno

0.52

 insinu

0.51

מי

0.51

 השני

0.51

此之外

0.50

Activations Density 0.023%