INDEX

Explanations

Indian places and languages

The neuron strongly activates on Hindi (and related Indic) function words and verb inflections, effectively identifying Hindi/Urdu text.

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

trying

0.59

にして

0.56

XC

0.56

manner

0.55

Button

0.55

Const

0.54

ترك

0.54

YC

0.54

炒

0.54

Competition

0.53

POSITIVE LOGITS

 है

0.68

är

0.67

 आहे

0.65

 हैं

0.63

 ambayo

0.61

 represented

0.60

 lousy

0.60

 हूं

0.60

 damn

0.59

yra

0.58

Activations Density 0.006%