INDEX

Explanations

Roman and historical periods

This neuron detects references to classical antiquity, especially names and adjectives related to ancient Roman (and to a lesser extent Greek/Latin) contexts.

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

 وعلى

0.72

 situations

0.70

د

0.69

 आहे

0.68

та

0.66

 êtes

0.66

 molestias

0.63

 notifies

0.61

 situaciones

0.61

 frivolous

0.61

POSITIVE LOGITS

 chrét

0.86

 filóso

0.83

 అనంతరం

0.80

时期

0.79

 avevano

0.78

era

0.76

Era

0.75

 আমলে

0.75

 philos

0.72

 hymn

0.71

Activations Density 0.004%