INDEX

Explanations

lists and explanations

This neuron fires strongly on isolated proper nouns, technical terms, and standalone numerals—that is, tokens that look like names, headings, or reference‐style numbers.

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

 মোহ

0.81

Projectile

0.73

 Researchers

0.72

 Journalism

0.71

umal

0.71

0.70

 Midlands

0.70

 Satisfaction

0.69

ucie

0.68

umam

0.68

POSITIVE LOGITS

일까지

0.79

 afloat

0.77

 sift

0.75

RAR

0.73

 osim

0.71

irradiated

0.71

 greeting

0.70

्ला

0.70

 outerwear

0.70

𝑪

0.70

Activations Density 0.000%