INDEX

Explanations

code and programming terms

This neuron primarily activates on proper nouns and headings—that is, capitalized named entities like country and organization names or section titles.

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

 ਅਤੇ

0.33

 และ

0.32

 மற்றும்

0.31

 ซึ่ง

0.31

 आणि

0.30

 અને

0.30

and

0.30

 because

0.29

และ

0.29

 ಮತ್ತು

0.29

POSITIVE LOGITS

0.29

the

0.27

an

0.25

any

0.25

 eponymous

0.22

 various

0.22

 diurnal

0.21

 hapless

0.21

 flimsy

0.21

 canonical

0.21

Activations Density 1.030%