INDEX

Explanations

video

This neuron fires on rare proper‐name tokens—particularly trademarked brand names and scientific Latin species names.

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

TextField

0.93

TreeNavigator

0.93

 ActionBar

0.91

𝚆

0.88

ым

0.88

MainWindow

0.87

RAchievement

0.86

jár

0.86

FlaskApp

0.86

Buffer

0.85

POSITIVE LOGITS

崛起

0.67

 captures

0.64

ने

0.63

 succès

0.63

 Kond

0.63

 успех

0.61

 rocked

0.59

 konkuren

0.59

 modus

0.58

 بور

0.57

Activations Density 0.001%