INDEX

Explanations

Wikipedia pages and edit history

This neuron strongly activates on Wikipedia-style metadata tokens (timestamps, revision numbers, years, UI labels like Read/Edit/View, and similar numeric or structural elements).

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

 olen

-0.92

 groupe

-0.84

 sana

-0.84

SIN

-0.79

 اجرای

-0.79

 goreng

-0.79

¯¯

-0.79

💸

-0.78

 mixte

-0.78

 ayam

-0.78

POSITIVE LOGITS

Wiki

1.09

 Wiki

1.08

 wiki

1.07

 edit

0.93

edit

0.91

wiki

0.90

 Wikipedia

0.86

wikipedia

0.85

 Wikimedia

0.85

 zurück

0.81

Activations Density 0.003%