INDEX

Explanations

sacred places and religious texts

The neuron fires on proper names, titles, and numbered volume/section identifiers (i.e. named entities and document headings).

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

 কবিত

0.80

ローマ

0.77

 ખ્ર

0.77

葡萄酒

0.75

 கிரேக்க

0.75

 रोमन

0.71

 অভিব্যক্তি

0.70

 तानाशाह

0.70

🍕

0.68

 الأوروب

0.66

POSITIVE LOGITS

 devotees

1.51

 devotee

1.38

 Swami

1.30

 temple

1.25

 devotional

1.23

Swami

1.22

 भक्त

1.17

 Maharaj

1.16

 pilgrimage

1.15

 Hindu

1.13

Activations Density 0.323%