INDEX

Explanations

Keycloak account, PDDL script, Python application

The neuron is highly activated by trademarked names and proper nouns referring to brands, products, or companies.

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

 abreast

0.64

 lofty

0.64

 etiqu

0.64

 albumen

0.61

 duodenum

0.61

 leer

0.59

 holotype

0.59

 natives

0.59

 sleek

0.58

 Hodge

0.58

POSITIVE LOGITS

®

0.91

®.

0.77

®,

0.75

Ⓡ

0.72

czek

0.63

ův

0.62

áns

0.61

κολ

0.61

ᾷ

0.61

uesday

0.60

Activations Density 0.783%